{
  "best_metric": 0.9025216698646545,
  "best_model_checkpoint": "outputs/checkpoint-7650",
  "epoch": 4.998087312506292,
  "eval_steps": 50,
  "global_step": 9310,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0005368947350760042,
      "grad_norm": 1.5859375,
      "learning_rate": 5e-06,
      "loss": 1.5327,
      "step": 1
    },
    {
      "epoch": 0.0010737894701520084,
      "grad_norm": 1.9375,
      "learning_rate": 1e-05,
      "loss": 1.4627,
      "step": 2
    },
    {
      "epoch": 0.0016106842052280126,
      "grad_norm": 2.0,
      "learning_rate": 1.5e-05,
      "loss": 1.5503,
      "step": 3
    },
    {
      "epoch": 0.0021475789403040168,
      "grad_norm": 1.6875,
      "learning_rate": 2e-05,
      "loss": 1.4298,
      "step": 4
    },
    {
      "epoch": 0.0026844736753800207,
      "grad_norm": 1.546875,
      "learning_rate": 2.5e-05,
      "loss": 1.302,
      "step": 5
    },
    {
      "epoch": 0.003221368410456025,
      "grad_norm": 1.6640625,
      "learning_rate": 3e-05,
      "loss": 1.6321,
      "step": 6
    },
    {
      "epoch": 0.003758263145532029,
      "grad_norm": 1.6328125,
      "learning_rate": 3.5e-05,
      "loss": 1.4398,
      "step": 7
    },
    {
      "epoch": 0.0042951578806080335,
      "grad_norm": 1.46875,
      "learning_rate": 4e-05,
      "loss": 1.4147,
      "step": 8
    },
    {
      "epoch": 0.0048320526156840375,
      "grad_norm": 1.484375,
      "learning_rate": 4.5e-05,
      "loss": 1.3094,
      "step": 9
    },
    {
      "epoch": 0.0053689473507600415,
      "grad_norm": 1.40625,
      "learning_rate": 5e-05,
      "loss": 1.5951,
      "step": 10
    },
    {
      "epoch": 0.005905842085836045,
      "grad_norm": 1.3046875,
      "learning_rate": 4.99999985735917e-05,
      "loss": 1.6913,
      "step": 11
    },
    {
      "epoch": 0.00644273682091205,
      "grad_norm": 1.4140625,
      "learning_rate": 4.999999429436697e-05,
      "loss": 1.3411,
      "step": 12
    },
    {
      "epoch": 0.006979631555988054,
      "grad_norm": 1.1796875,
      "learning_rate": 4.9999987162326275e-05,
      "loss": 1.225,
      "step": 13
    },
    {
      "epoch": 0.007516526291064058,
      "grad_norm": 1.484375,
      "learning_rate": 4.9999977177470465e-05,
      "loss": 1.1572,
      "step": 14
    },
    {
      "epoch": 0.008053421026140062,
      "grad_norm": 1.1875,
      "learning_rate": 4.999996433980065e-05,
      "loss": 1.4555,
      "step": 15
    },
    {
      "epoch": 0.008590315761216067,
      "grad_norm": 1.1953125,
      "learning_rate": 4.999994864931831e-05,
      "loss": 1.3826,
      "step": 16
    },
    {
      "epoch": 0.00912721049629207,
      "grad_norm": 1.0625,
      "learning_rate": 4.999993010602522e-05,
      "loss": 1.4102,
      "step": 17
    },
    {
      "epoch": 0.009664105231368075,
      "grad_norm": 1.03125,
      "learning_rate": 4.999990870992352e-05,
      "loss": 1.0747,
      "step": 18
    },
    {
      "epoch": 0.01020099996644408,
      "grad_norm": 1.0859375,
      "learning_rate": 4.9999884461015623e-05,
      "loss": 1.3182,
      "step": 19
    },
    {
      "epoch": 0.010737894701520083,
      "grad_norm": 1.09375,
      "learning_rate": 4.999985735930432e-05,
      "loss": 1.3525,
      "step": 20
    },
    {
      "epoch": 0.011274789436596088,
      "grad_norm": 1.0546875,
      "learning_rate": 4.999982740479269e-05,
      "loss": 1.2065,
      "step": 21
    },
    {
      "epoch": 0.01181168417167209,
      "grad_norm": 0.96875,
      "learning_rate": 4.9999794597484165e-05,
      "loss": 1.1728,
      "step": 22
    },
    {
      "epoch": 0.012348578906748096,
      "grad_norm": 0.96875,
      "learning_rate": 4.999975893738247e-05,
      "loss": 1.213,
      "step": 23
    },
    {
      "epoch": 0.0128854736418241,
      "grad_norm": 1.0703125,
      "learning_rate": 4.999972042449169e-05,
      "loss": 1.1957,
      "step": 24
    },
    {
      "epoch": 0.013422368376900104,
      "grad_norm": 2.0625,
      "learning_rate": 4.9999679058816204e-05,
      "loss": 1.4011,
      "step": 25
    },
    {
      "epoch": 0.013959263111976108,
      "grad_norm": 0.9921875,
      "learning_rate": 4.9999634840360755e-05,
      "loss": 1.0791,
      "step": 26
    },
    {
      "epoch": 0.014496157847052112,
      "grad_norm": 0.91796875,
      "learning_rate": 4.999958776913037e-05,
      "loss": 0.921,
      "step": 27
    },
    {
      "epoch": 0.015033052582128116,
      "grad_norm": 0.9140625,
      "learning_rate": 4.9999537845130426e-05,
      "loss": 1.2306,
      "step": 28
    },
    {
      "epoch": 0.015569947317204121,
      "grad_norm": 1.09375,
      "learning_rate": 4.9999485068366625e-05,
      "loss": 1.304,
      "step": 29
    },
    {
      "epoch": 0.016106842052280124,
      "grad_norm": 1.25,
      "learning_rate": 4.999942943884498e-05,
      "loss": 1.34,
      "step": 30
    },
    {
      "epoch": 0.016643736787356127,
      "grad_norm": 0.94921875,
      "learning_rate": 4.999937095657184e-05,
      "loss": 1.2341,
      "step": 31
    },
    {
      "epoch": 0.017180631522432134,
      "grad_norm": 1.09375,
      "learning_rate": 4.9999309621553894e-05,
      "loss": 1.1547,
      "step": 32
    },
    {
      "epoch": 0.017717526257508137,
      "grad_norm": 0.921875,
      "learning_rate": 4.999924543379812e-05,
      "loss": 1.1039,
      "step": 33
    },
    {
      "epoch": 0.01825442099258414,
      "grad_norm": 1.0625,
      "learning_rate": 4.9999178393311855e-05,
      "loss": 1.1743,
      "step": 34
    },
    {
      "epoch": 0.018791315727660147,
      "grad_norm": 0.9453125,
      "learning_rate": 4.999910850010275e-05,
      "loss": 1.2444,
      "step": 35
    },
    {
      "epoch": 0.01932821046273615,
      "grad_norm": 0.96875,
      "learning_rate": 4.999903575417877e-05,
      "loss": 1.2285,
      "step": 36
    },
    {
      "epoch": 0.019865105197812153,
      "grad_norm": 0.84765625,
      "learning_rate": 4.9998960155548227e-05,
      "loss": 0.9883,
      "step": 37
    },
    {
      "epoch": 0.02040199993288816,
      "grad_norm": 0.91796875,
      "learning_rate": 4.9998881704219745e-05,
      "loss": 0.9327,
      "step": 38
    },
    {
      "epoch": 0.020938894667964163,
      "grad_norm": 0.96875,
      "learning_rate": 4.999880040020227e-05,
      "loss": 1.1627,
      "step": 39
    },
    {
      "epoch": 0.021475789403040166,
      "grad_norm": 1.09375,
      "learning_rate": 4.9998716243505096e-05,
      "loss": 1.1357,
      "step": 40
    },
    {
      "epoch": 0.02201268413811617,
      "grad_norm": 1.265625,
      "learning_rate": 4.999862923413781e-05,
      "loss": 1.0989,
      "step": 41
    },
    {
      "epoch": 0.022549578873192175,
      "grad_norm": 1.0078125,
      "learning_rate": 4.999853937211034e-05,
      "loss": 1.2711,
      "step": 42
    },
    {
      "epoch": 0.02308647360826818,
      "grad_norm": 0.88671875,
      "learning_rate": 4.999844665743295e-05,
      "loss": 1.3272,
      "step": 43
    },
    {
      "epoch": 0.02362336834334418,
      "grad_norm": 0.86328125,
      "learning_rate": 4.9998351090116226e-05,
      "loss": 1.114,
      "step": 44
    },
    {
      "epoch": 0.02416026307842019,
      "grad_norm": 0.83984375,
      "learning_rate": 4.999825267017105e-05,
      "loss": 1.1632,
      "step": 45
    },
    {
      "epoch": 0.02469715781349619,
      "grad_norm": 0.86328125,
      "learning_rate": 4.9998151397608674e-05,
      "loss": 1.2763,
      "step": 46
    },
    {
      "epoch": 0.025234052548572194,
      "grad_norm": 0.96484375,
      "learning_rate": 4.999804727244065e-05,
      "loss": 1.223,
      "step": 47
    },
    {
      "epoch": 0.0257709472836482,
      "grad_norm": 0.98046875,
      "learning_rate": 4.999794029467886e-05,
      "loss": 1.1651,
      "step": 48
    },
    {
      "epoch": 0.026307842018724204,
      "grad_norm": 0.9140625,
      "learning_rate": 4.99978304643355e-05,
      "loss": 1.1746,
      "step": 49
    },
    {
      "epoch": 0.026844736753800207,
      "grad_norm": 0.98046875,
      "learning_rate": 4.9997717781423114e-05,
      "loss": 1.3132,
      "step": 50
    },
    {
      "epoch": 0.026844736753800207,
      "eval_loss": 1.0921833515167236,
      "eval_model_preparation_time": 0.0202,
      "eval_runtime": 47.907,
      "eval_samples_per_second": 20.081,
      "eval_steps_per_second": 20.081,
      "step": 50
    },
    {
      "epoch": 0.027381631488876214,
      "grad_norm": 0.90234375,
      "learning_rate": 4.999760224595456e-05,
      "loss": 1.1041,
      "step": 51
    },
    {
      "epoch": 0.027918526223952217,
      "grad_norm": 0.87890625,
      "learning_rate": 4.999748385794302e-05,
      "loss": 1.1916,
      "step": 52
    },
    {
      "epoch": 0.02845542095902822,
      "grad_norm": 1.0546875,
      "learning_rate": 4.999736261740201e-05,
      "loss": 1.0385,
      "step": 53
    },
    {
      "epoch": 0.028992315694104223,
      "grad_norm": 1.2109375,
      "learning_rate": 4.999723852434535e-05,
      "loss": 1.293,
      "step": 54
    },
    {
      "epoch": 0.02952921042918023,
      "grad_norm": 1.0234375,
      "learning_rate": 4.9997111578787214e-05,
      "loss": 1.4005,
      "step": 55
    },
    {
      "epoch": 0.030066105164256233,
      "grad_norm": 0.87109375,
      "learning_rate": 4.999698178074209e-05,
      "loss": 1.0834,
      "step": 56
    },
    {
      "epoch": 0.030602999899332236,
      "grad_norm": 1.0,
      "learning_rate": 4.999684913022478e-05,
      "loss": 1.0948,
      "step": 57
    },
    {
      "epoch": 0.031139894634408243,
      "grad_norm": 0.8125,
      "learning_rate": 4.9996713627250426e-05,
      "loss": 1.1725,
      "step": 58
    },
    {
      "epoch": 0.031676789369484246,
      "grad_norm": 1.1015625,
      "learning_rate": 4.9996575271834475e-05,
      "loss": 1.3369,
      "step": 59
    },
    {
      "epoch": 0.03221368410456025,
      "grad_norm": 0.9140625,
      "learning_rate": 4.999643406399275e-05,
      "loss": 1.0575,
      "step": 60
    },
    {
      "epoch": 0.03275057883963625,
      "grad_norm": 0.84375,
      "learning_rate": 4.999629000374133e-05,
      "loss": 1.0591,
      "step": 61
    },
    {
      "epoch": 0.033287473574712255,
      "grad_norm": 0.86328125,
      "learning_rate": 4.9996143091096684e-05,
      "loss": 1.2379,
      "step": 62
    },
    {
      "epoch": 0.033824368309788265,
      "grad_norm": 0.890625,
      "learning_rate": 4.999599332607554e-05,
      "loss": 1.1148,
      "step": 63
    },
    {
      "epoch": 0.03436126304486427,
      "grad_norm": 0.984375,
      "learning_rate": 4.999584070869502e-05,
      "loss": 1.0229,
      "step": 64
    },
    {
      "epoch": 0.03489815777994027,
      "grad_norm": 0.7890625,
      "learning_rate": 4.999568523897254e-05,
      "loss": 1.1811,
      "step": 65
    },
    {
      "epoch": 0.035435052515016274,
      "grad_norm": 1.0859375,
      "learning_rate": 4.999552691692581e-05,
      "loss": 0.9079,
      "step": 66
    },
    {
      "epoch": 0.03597194725009228,
      "grad_norm": 0.97265625,
      "learning_rate": 4.999536574257292e-05,
      "loss": 1.1585,
      "step": 67
    },
    {
      "epoch": 0.03650884198516828,
      "grad_norm": 0.91796875,
      "learning_rate": 4.999520171593226e-05,
      "loss": 1.2089,
      "step": 68
    },
    {
      "epoch": 0.03704573672024429,
      "grad_norm": 0.94921875,
      "learning_rate": 4.999503483702255e-05,
      "loss": 1.0255,
      "step": 69
    },
    {
      "epoch": 0.037582631455320294,
      "grad_norm": 0.92578125,
      "learning_rate": 4.999486510586282e-05,
      "loss": 1.0401,
      "step": 70
    },
    {
      "epoch": 0.0381195261903963,
      "grad_norm": 1.0390625,
      "learning_rate": 4.9994692522472455e-05,
      "loss": 1.2341,
      "step": 71
    },
    {
      "epoch": 0.0386564209254723,
      "grad_norm": 1.0546875,
      "learning_rate": 4.999451708687114e-05,
      "loss": 1.0865,
      "step": 72
    },
    {
      "epoch": 0.0391933156605483,
      "grad_norm": 1.6484375,
      "learning_rate": 4.9994338799078896e-05,
      "loss": 1.3078,
      "step": 73
    },
    {
      "epoch": 0.039730210395624306,
      "grad_norm": 0.9609375,
      "learning_rate": 4.999415765911606e-05,
      "loss": 1.0737,
      "step": 74
    },
    {
      "epoch": 0.04026710513070031,
      "grad_norm": 0.83203125,
      "learning_rate": 4.999397366700332e-05,
      "loss": 1.1142,
      "step": 75
    },
    {
      "epoch": 0.04080399986577632,
      "grad_norm": 0.79296875,
      "learning_rate": 4.9993786822761656e-05,
      "loss": 0.9994,
      "step": 76
    },
    {
      "epoch": 0.04134089460085232,
      "grad_norm": 0.96875,
      "learning_rate": 4.99935971264124e-05,
      "loss": 1.227,
      "step": 77
    },
    {
      "epoch": 0.041877789335928325,
      "grad_norm": 1.09375,
      "learning_rate": 4.999340457797718e-05,
      "loss": 1.2586,
      "step": 78
    },
    {
      "epoch": 0.04241468407100433,
      "grad_norm": 0.83203125,
      "learning_rate": 4.999320917747799e-05,
      "loss": 0.9687,
      "step": 79
    },
    {
      "epoch": 0.04295157880608033,
      "grad_norm": 1.125,
      "learning_rate": 4.999301092493712e-05,
      "loss": 1.115,
      "step": 80
    },
    {
      "epoch": 0.043488473541156335,
      "grad_norm": 1.0,
      "learning_rate": 4.99928098203772e-05,
      "loss": 0.9307,
      "step": 81
    },
    {
      "epoch": 0.04402536827623234,
      "grad_norm": 0.75390625,
      "learning_rate": 4.999260586382116e-05,
      "loss": 0.8806,
      "step": 82
    },
    {
      "epoch": 0.04456226301130835,
      "grad_norm": 0.9140625,
      "learning_rate": 4.999239905529229e-05,
      "loss": 0.947,
      "step": 83
    },
    {
      "epoch": 0.04509915774638435,
      "grad_norm": 0.796875,
      "learning_rate": 4.999218939481418e-05,
      "loss": 1.0248,
      "step": 84
    },
    {
      "epoch": 0.045636052481460354,
      "grad_norm": 1.40625,
      "learning_rate": 4.999197688241076e-05,
      "loss": 1.0739,
      "step": 85
    },
    {
      "epoch": 0.04617294721653636,
      "grad_norm": 0.9375,
      "learning_rate": 4.999176151810629e-05,
      "loss": 1.2311,
      "step": 86
    },
    {
      "epoch": 0.04670984195161236,
      "grad_norm": 0.83984375,
      "learning_rate": 4.999154330192532e-05,
      "loss": 1.0813,
      "step": 87
    },
    {
      "epoch": 0.04724673668668836,
      "grad_norm": 0.90234375,
      "learning_rate": 4.9991322233892784e-05,
      "loss": 1.0484,
      "step": 88
    },
    {
      "epoch": 0.04778363142176437,
      "grad_norm": 0.81640625,
      "learning_rate": 4.999109831403388e-05,
      "loss": 1.2286,
      "step": 89
    },
    {
      "epoch": 0.04832052615684038,
      "grad_norm": 0.95703125,
      "learning_rate": 4.999087154237418e-05,
      "loss": 1.3113,
      "step": 90
    },
    {
      "epoch": 0.04885742089191638,
      "grad_norm": 0.9375,
      "learning_rate": 4.999064191893955e-05,
      "loss": 1.0101,
      "step": 91
    },
    {
      "epoch": 0.04939431562699238,
      "grad_norm": 0.80859375,
      "learning_rate": 4.999040944375619e-05,
      "loss": 0.9629,
      "step": 92
    },
    {
      "epoch": 0.049931210362068386,
      "grad_norm": 0.83984375,
      "learning_rate": 4.999017411685064e-05,
      "loss": 1.4695,
      "step": 93
    },
    {
      "epoch": 0.05046810509714439,
      "grad_norm": 0.8046875,
      "learning_rate": 4.998993593824975e-05,
      "loss": 1.0943,
      "step": 94
    },
    {
      "epoch": 0.05100499983222039,
      "grad_norm": 0.76953125,
      "learning_rate": 4.998969490798069e-05,
      "loss": 1.0775,
      "step": 95
    },
    {
      "epoch": 0.0515418945672964,
      "grad_norm": 0.8203125,
      "learning_rate": 4.9989451026070975e-05,
      "loss": 1.0083,
      "step": 96
    },
    {
      "epoch": 0.052078789302372405,
      "grad_norm": 1.0234375,
      "learning_rate": 4.998920429254844e-05,
      "loss": 1.3772,
      "step": 97
    },
    {
      "epoch": 0.05261568403744841,
      "grad_norm": 0.9921875,
      "learning_rate": 4.9988954707441226e-05,
      "loss": 0.9465,
      "step": 98
    },
    {
      "epoch": 0.05315257877252441,
      "grad_norm": 0.74609375,
      "learning_rate": 4.998870227077782e-05,
      "loss": 1.1598,
      "step": 99
    },
    {
      "epoch": 0.053689473507600415,
      "grad_norm": 1.15625,
      "learning_rate": 4.9988446982587035e-05,
      "loss": 1.3205,
      "step": 100
    },
    {
      "epoch": 0.053689473507600415,
      "eval_loss": 1.0550328493118286,
      "eval_runtime": 48.3738,
      "eval_samples_per_second": 19.887,
      "eval_steps_per_second": 19.887,
      "step": 100
    },
    {
      "epoch": 0.05422636824267642,
      "grad_norm": 0.9453125,
      "learning_rate": 4.998818884289799e-05,
      "loss": 1.2547,
      "step": 101
    },
    {
      "epoch": 0.05476326297775243,
      "grad_norm": 1.4921875,
      "learning_rate": 4.998792785174014e-05,
      "loss": 1.2115,
      "step": 102
    },
    {
      "epoch": 0.05530015771282843,
      "grad_norm": 0.83203125,
      "learning_rate": 4.998766400914329e-05,
      "loss": 1.2368,
      "step": 103
    },
    {
      "epoch": 0.055837052447904434,
      "grad_norm": 0.78125,
      "learning_rate": 4.998739731513753e-05,
      "loss": 1.0902,
      "step": 104
    },
    {
      "epoch": 0.05637394718298044,
      "grad_norm": 0.7265625,
      "learning_rate": 4.9987127769753295e-05,
      "loss": 1.0537,
      "step": 105
    },
    {
      "epoch": 0.05691084191805644,
      "grad_norm": 1.0390625,
      "learning_rate": 4.998685537302135e-05,
      "loss": 0.9671,
      "step": 106
    },
    {
      "epoch": 0.05744773665313244,
      "grad_norm": 0.83203125,
      "learning_rate": 4.998658012497276e-05,
      "loss": 1.0743,
      "step": 107
    },
    {
      "epoch": 0.057984631388208446,
      "grad_norm": 0.7890625,
      "learning_rate": 4.998630202563896e-05,
      "loss": 1.1052,
      "step": 108
    },
    {
      "epoch": 0.058521526123284456,
      "grad_norm": 0.83984375,
      "learning_rate": 4.998602107505167e-05,
      "loss": 0.8357,
      "step": 109
    },
    {
      "epoch": 0.05905842085836046,
      "grad_norm": 1.0,
      "learning_rate": 4.998573727324295e-05,
      "loss": 1.2236,
      "step": 110
    },
    {
      "epoch": 0.05959531559343646,
      "grad_norm": 0.8671875,
      "learning_rate": 4.9985450620245196e-05,
      "loss": 1.1851,
      "step": 111
    },
    {
      "epoch": 0.060132210328512466,
      "grad_norm": 0.796875,
      "learning_rate": 4.998516111609111e-05,
      "loss": 0.9234,
      "step": 112
    },
    {
      "epoch": 0.06066910506358847,
      "grad_norm": 0.87109375,
      "learning_rate": 4.9984868760813717e-05,
      "loss": 1.1588,
      "step": 113
    },
    {
      "epoch": 0.06120599979866447,
      "grad_norm": 0.80078125,
      "learning_rate": 4.9984573554446404e-05,
      "loss": 1.0522,
      "step": 114
    },
    {
      "epoch": 0.06174289453374048,
      "grad_norm": 0.875,
      "learning_rate": 4.998427549702284e-05,
      "loss": 1.0305,
      "step": 115
    },
    {
      "epoch": 0.062279789268816485,
      "grad_norm": 0.84375,
      "learning_rate": 4.998397458857704e-05,
      "loss": 1.0935,
      "step": 116
    },
    {
      "epoch": 0.06281668400389248,
      "grad_norm": 0.7890625,
      "learning_rate": 4.9983670829143346e-05,
      "loss": 1.1522,
      "step": 117
    },
    {
      "epoch": 0.06335357873896849,
      "grad_norm": 0.921875,
      "learning_rate": 4.998336421875641e-05,
      "loss": 1.4326,
      "step": 118
    },
    {
      "epoch": 0.0638904734740445,
      "grad_norm": 0.77734375,
      "learning_rate": 4.9983054757451244e-05,
      "loss": 1.0302,
      "step": 119
    },
    {
      "epoch": 0.0644273682091205,
      "grad_norm": 0.76953125,
      "learning_rate": 4.998274244526313e-05,
      "loss": 1.2382,
      "step": 120
    },
    {
      "epoch": 0.06496426294419651,
      "grad_norm": 0.85546875,
      "learning_rate": 4.998242728222773e-05,
      "loss": 1.0985,
      "step": 121
    },
    {
      "epoch": 0.0655011576792725,
      "grad_norm": 0.84375,
      "learning_rate": 4.9982109268380995e-05,
      "loss": 1.2041,
      "step": 122
    },
    {
      "epoch": 0.06603805241434851,
      "grad_norm": 0.8515625,
      "learning_rate": 4.9981788403759224e-05,
      "loss": 1.0621,
      "step": 123
    },
    {
      "epoch": 0.06657494714942451,
      "grad_norm": 0.8515625,
      "learning_rate": 4.998146468839903e-05,
      "loss": 1.3032,
      "step": 124
    },
    {
      "epoch": 0.06711184188450052,
      "grad_norm": 0.8046875,
      "learning_rate": 4.998113812233735e-05,
      "loss": 0.9213,
      "step": 125
    },
    {
      "epoch": 0.06764873661957653,
      "grad_norm": 0.88671875,
      "learning_rate": 4.9980808705611435e-05,
      "loss": 1.0401,
      "step": 126
    },
    {
      "epoch": 0.06818563135465253,
      "grad_norm": 0.90234375,
      "learning_rate": 4.998047643825891e-05,
      "loss": 1.2438,
      "step": 127
    },
    {
      "epoch": 0.06872252608972854,
      "grad_norm": 0.96875,
      "learning_rate": 4.998014132031766e-05,
      "loss": 0.9991,
      "step": 128
    },
    {
      "epoch": 0.06925942082480453,
      "grad_norm": 0.97265625,
      "learning_rate": 4.9979803351825936e-05,
      "loss": 1.0276,
      "step": 129
    },
    {
      "epoch": 0.06979631555988054,
      "grad_norm": 0.7734375,
      "learning_rate": 4.997946253282231e-05,
      "loss": 1.194,
      "step": 130
    },
    {
      "epoch": 0.07033321029495654,
      "grad_norm": 0.89453125,
      "learning_rate": 4.997911886334567e-05,
      "loss": 1.1206,
      "step": 131
    },
    {
      "epoch": 0.07087010503003255,
      "grad_norm": 0.78125,
      "learning_rate": 4.9978772343435234e-05,
      "loss": 0.9971,
      "step": 132
    },
    {
      "epoch": 0.07140699976510856,
      "grad_norm": 0.890625,
      "learning_rate": 4.997842297313054e-05,
      "loss": 1.1272,
      "step": 133
    },
    {
      "epoch": 0.07194389450018455,
      "grad_norm": 0.78515625,
      "learning_rate": 4.997807075247146e-05,
      "loss": 0.9863,
      "step": 134
    },
    {
      "epoch": 0.07248078923526056,
      "grad_norm": 0.8046875,
      "learning_rate": 4.997771568149818e-05,
      "loss": 1.2149,
      "step": 135
    },
    {
      "epoch": 0.07301768397033656,
      "grad_norm": 0.875,
      "learning_rate": 4.997735776025124e-05,
      "loss": 1.1231,
      "step": 136
    },
    {
      "epoch": 0.07355457870541257,
      "grad_norm": 0.76953125,
      "learning_rate": 4.997699698877145e-05,
      "loss": 1.0775,
      "step": 137
    },
    {
      "epoch": 0.07409147344048858,
      "grad_norm": 0.80078125,
      "learning_rate": 4.99766333671e-05,
      "loss": 0.9318,
      "step": 138
    },
    {
      "epoch": 0.07462836817556458,
      "grad_norm": 0.890625,
      "learning_rate": 4.9976266895278384e-05,
      "loss": 1.0069,
      "step": 139
    },
    {
      "epoch": 0.07516526291064059,
      "grad_norm": 0.85546875,
      "learning_rate": 4.997589757334842e-05,
      "loss": 0.9013,
      "step": 140
    },
    {
      "epoch": 0.07570215764571658,
      "grad_norm": 0.953125,
      "learning_rate": 4.9975525401352235e-05,
      "loss": 1.0932,
      "step": 141
    },
    {
      "epoch": 0.0762390523807926,
      "grad_norm": 0.921875,
      "learning_rate": 4.997515037933232e-05,
      "loss": 1.1049,
      "step": 142
    },
    {
      "epoch": 0.07677594711586859,
      "grad_norm": 0.80859375,
      "learning_rate": 4.997477250733146e-05,
      "loss": 1.2303,
      "step": 143
    },
    {
      "epoch": 0.0773128418509446,
      "grad_norm": 0.875,
      "learning_rate": 4.997439178539278e-05,
      "loss": 1.0978,
      "step": 144
    },
    {
      "epoch": 0.07784973658602061,
      "grad_norm": 0.8828125,
      "learning_rate": 4.997400821355972e-05,
      "loss": 1.2369,
      "step": 145
    },
    {
      "epoch": 0.0783866313210966,
      "grad_norm": 0.85546875,
      "learning_rate": 4.9973621791876055e-05,
      "loss": 1.2848,
      "step": 146
    },
    {
      "epoch": 0.07892352605617262,
      "grad_norm": 0.84375,
      "learning_rate": 4.997323252038588e-05,
      "loss": 1.1895,
      "step": 147
    },
    {
      "epoch": 0.07946042079124861,
      "grad_norm": 0.84375,
      "learning_rate": 4.99728403991336e-05,
      "loss": 1.024,
      "step": 148
    },
    {
      "epoch": 0.07999731552632462,
      "grad_norm": 0.7421875,
      "learning_rate": 4.9972445428163994e-05,
      "loss": 0.8837,
      "step": 149
    },
    {
      "epoch": 0.08053421026140062,
      "grad_norm": 0.91796875,
      "learning_rate": 4.99720476075221e-05,
      "loss": 1.1442,
      "step": 150
    },
    {
      "epoch": 0.08053421026140062,
      "eval_loss": 1.036392331123352,
      "eval_runtime": 48.3992,
      "eval_samples_per_second": 19.876,
      "eval_steps_per_second": 19.876,
      "step": 150
    },
    {
      "epoch": 0.08107110499647663,
      "grad_norm": 0.859375,
      "learning_rate": 4.9971646937253334e-05,
      "loss": 1.2851,
      "step": 151
    },
    {
      "epoch": 0.08160799973155264,
      "grad_norm": 0.88671875,
      "learning_rate": 4.9971243417403414e-05,
      "loss": 1.287,
      "step": 152
    },
    {
      "epoch": 0.08214489446662863,
      "grad_norm": 0.90625,
      "learning_rate": 4.997083704801838e-05,
      "loss": 1.2777,
      "step": 153
    },
    {
      "epoch": 0.08268178920170464,
      "grad_norm": 0.8828125,
      "learning_rate": 4.997042782914462e-05,
      "loss": 1.2524,
      "step": 154
    },
    {
      "epoch": 0.08321868393678064,
      "grad_norm": 0.69921875,
      "learning_rate": 4.997001576082881e-05,
      "loss": 1.0275,
      "step": 155
    },
    {
      "epoch": 0.08375557867185665,
      "grad_norm": 0.84375,
      "learning_rate": 4.996960084311798e-05,
      "loss": 1.2872,
      "step": 156
    },
    {
      "epoch": 0.08429247340693265,
      "grad_norm": 0.9453125,
      "learning_rate": 4.996918307605949e-05,
      "loss": 1.1124,
      "step": 157
    },
    {
      "epoch": 0.08482936814200866,
      "grad_norm": 0.91015625,
      "learning_rate": 4.9968762459700994e-05,
      "loss": 1.0677,
      "step": 158
    },
    {
      "epoch": 0.08536626287708467,
      "grad_norm": 0.8671875,
      "learning_rate": 4.9968338994090496e-05,
      "loss": 1.0741,
      "step": 159
    },
    {
      "epoch": 0.08590315761216066,
      "grad_norm": 0.8125,
      "learning_rate": 4.9967912679276316e-05,
      "loss": 1.1287,
      "step": 160
    },
    {
      "epoch": 0.08644005234723667,
      "grad_norm": 0.81640625,
      "learning_rate": 4.996748351530712e-05,
      "loss": 0.9927,
      "step": 161
    },
    {
      "epoch": 0.08697694708231267,
      "grad_norm": 0.9453125,
      "learning_rate": 4.996705150223186e-05,
      "loss": 1.1585,
      "step": 162
    },
    {
      "epoch": 0.08751384181738868,
      "grad_norm": 0.87890625,
      "learning_rate": 4.996661664009984e-05,
      "loss": 1.3801,
      "step": 163
    },
    {
      "epoch": 0.08805073655246468,
      "grad_norm": 0.75390625,
      "learning_rate": 4.996617892896069e-05,
      "loss": 1.0954,
      "step": 164
    },
    {
      "epoch": 0.08858763128754069,
      "grad_norm": 0.84375,
      "learning_rate": 4.996573836886435e-05,
      "loss": 1.1691,
      "step": 165
    },
    {
      "epoch": 0.0891245260226167,
      "grad_norm": 0.78515625,
      "learning_rate": 4.9965294959861095e-05,
      "loss": 1.0086,
      "step": 166
    },
    {
      "epoch": 0.08966142075769269,
      "grad_norm": 0.82421875,
      "learning_rate": 4.9964848702001524e-05,
      "loss": 1.0528,
      "step": 167
    },
    {
      "epoch": 0.0901983154927687,
      "grad_norm": 0.82421875,
      "learning_rate": 4.996439959533656e-05,
      "loss": 1.1207,
      "step": 168
    },
    {
      "epoch": 0.0907352102278447,
      "grad_norm": 0.70703125,
      "learning_rate": 4.996394763991746e-05,
      "loss": 0.85,
      "step": 169
    },
    {
      "epoch": 0.09127210496292071,
      "grad_norm": 0.83203125,
      "learning_rate": 4.9963492835795797e-05,
      "loss": 1.1433,
      "step": 170
    },
    {
      "epoch": 0.09180899969799672,
      "grad_norm": 0.828125,
      "learning_rate": 4.9963035183023455e-05,
      "loss": 1.0846,
      "step": 171
    },
    {
      "epoch": 0.09234589443307271,
      "grad_norm": 0.9453125,
      "learning_rate": 4.9962574681652675e-05,
      "loss": 1.3203,
      "step": 172
    },
    {
      "epoch": 0.09288278916814872,
      "grad_norm": 0.921875,
      "learning_rate": 4.996211133173599e-05,
      "loss": 1.076,
      "step": 173
    },
    {
      "epoch": 0.09341968390322472,
      "grad_norm": 0.8125,
      "learning_rate": 4.996164513332628e-05,
      "loss": 1.1099,
      "step": 174
    },
    {
      "epoch": 0.09395657863830073,
      "grad_norm": 0.84765625,
      "learning_rate": 4.996117608647676e-05,
      "loss": 1.1499,
      "step": 175
    },
    {
      "epoch": 0.09449347337337673,
      "grad_norm": 0.75390625,
      "learning_rate": 4.9960704191240926e-05,
      "loss": 1.0821,
      "step": 176
    },
    {
      "epoch": 0.09503036810845274,
      "grad_norm": 0.9453125,
      "learning_rate": 4.996022944767265e-05,
      "loss": 1.0568,
      "step": 177
    },
    {
      "epoch": 0.09556726284352875,
      "grad_norm": 0.875,
      "learning_rate": 4.99597518558261e-05,
      "loss": 1.0568,
      "step": 178
    },
    {
      "epoch": 0.09610415757860474,
      "grad_norm": 0.69140625,
      "learning_rate": 4.995927141575577e-05,
      "loss": 1.1308,
      "step": 179
    },
    {
      "epoch": 0.09664105231368075,
      "grad_norm": 0.86328125,
      "learning_rate": 4.995878812751649e-05,
      "loss": 0.9887,
      "step": 180
    },
    {
      "epoch": 0.09717794704875675,
      "grad_norm": 0.90234375,
      "learning_rate": 4.9958301991163405e-05,
      "loss": 1.1791,
      "step": 181
    },
    {
      "epoch": 0.09771484178383276,
      "grad_norm": 0.76953125,
      "learning_rate": 4.995781300675199e-05,
      "loss": 1.0797,
      "step": 182
    },
    {
      "epoch": 0.09825173651890876,
      "grad_norm": 0.71875,
      "learning_rate": 4.9957321174338043e-05,
      "loss": 1.0162,
      "step": 183
    },
    {
      "epoch": 0.09878863125398477,
      "grad_norm": 1.0,
      "learning_rate": 4.99568264939777e-05,
      "loss": 1.0634,
      "step": 184
    },
    {
      "epoch": 0.09932552598906078,
      "grad_norm": 0.7734375,
      "learning_rate": 4.9956328965727394e-05,
      "loss": 1.1382,
      "step": 185
    },
    {
      "epoch": 0.09986242072413677,
      "grad_norm": 0.71875,
      "learning_rate": 4.995582858964392e-05,
      "loss": 0.9861,
      "step": 186
    },
    {
      "epoch": 0.10039931545921278,
      "grad_norm": 0.80078125,
      "learning_rate": 4.995532536578435e-05,
      "loss": 1.0947,
      "step": 187
    },
    {
      "epoch": 0.10093621019428878,
      "grad_norm": 1.109375,
      "learning_rate": 4.9954819294206124e-05,
      "loss": 1.2332,
      "step": 188
    },
    {
      "epoch": 0.10147310492936479,
      "grad_norm": 0.90234375,
      "learning_rate": 4.9954310374966996e-05,
      "loss": 1.1089,
      "step": 189
    },
    {
      "epoch": 0.10200999966444078,
      "grad_norm": 0.875,
      "learning_rate": 4.9953798608125025e-05,
      "loss": 1.0227,
      "step": 190
    },
    {
      "epoch": 0.1025468943995168,
      "grad_norm": 1.421875,
      "learning_rate": 4.995328399373862e-05,
      "loss": 0.8639,
      "step": 191
    },
    {
      "epoch": 0.1030837891345928,
      "grad_norm": 0.72265625,
      "learning_rate": 4.995276653186651e-05,
      "loss": 0.9628,
      "step": 192
    },
    {
      "epoch": 0.1036206838696688,
      "grad_norm": 1.0,
      "learning_rate": 4.995224622256773e-05,
      "loss": 1.2933,
      "step": 193
    },
    {
      "epoch": 0.10415757860474481,
      "grad_norm": 0.859375,
      "learning_rate": 4.9951723065901665e-05,
      "loss": 1.0998,
      "step": 194
    },
    {
      "epoch": 0.1046944733398208,
      "grad_norm": 0.7578125,
      "learning_rate": 4.995119706192801e-05,
      "loss": 1.146,
      "step": 195
    },
    {
      "epoch": 0.10523136807489682,
      "grad_norm": 0.8203125,
      "learning_rate": 4.995066821070679e-05,
      "loss": 0.9429,
      "step": 196
    },
    {
      "epoch": 0.10576826280997281,
      "grad_norm": 0.78515625,
      "learning_rate": 4.9950136512298354e-05,
      "loss": 1.1212,
      "step": 197
    },
    {
      "epoch": 0.10630515754504882,
      "grad_norm": 0.70703125,
      "learning_rate": 4.994960196676337e-05,
      "loss": 1.1948,
      "step": 198
    },
    {
      "epoch": 0.10684205228012483,
      "grad_norm": 0.97265625,
      "learning_rate": 4.994906457416284e-05,
      "loss": 1.218,
      "step": 199
    },
    {
      "epoch": 0.10737894701520083,
      "grad_norm": 0.74609375,
      "learning_rate": 4.994852433455809e-05,
      "loss": 0.9155,
      "step": 200
    },
    {
      "epoch": 0.10737894701520083,
      "eval_loss": 1.021553874015808,
      "eval_runtime": 45.6574,
      "eval_samples_per_second": 21.07,
      "eval_steps_per_second": 21.07,
      "step": 200
    },
    {
      "epoch": 0.10791584175027684,
      "grad_norm": 1.0625,
      "learning_rate": 4.9947981248010764e-05,
      "loss": 1.1516,
      "step": 201
    },
    {
      "epoch": 0.10845273648535284,
      "grad_norm": 0.796875,
      "learning_rate": 4.9947435314582844e-05,
      "loss": 1.075,
      "step": 202
    },
    {
      "epoch": 0.10898963122042885,
      "grad_norm": 0.77734375,
      "learning_rate": 4.994688653433661e-05,
      "loss": 1.1804,
      "step": 203
    },
    {
      "epoch": 0.10952652595550486,
      "grad_norm": 0.7109375,
      "learning_rate": 4.99463349073347e-05,
      "loss": 0.9413,
      "step": 204
    },
    {
      "epoch": 0.11006342069058085,
      "grad_norm": 0.75,
      "learning_rate": 4.9945780433640066e-05,
      "loss": 1.1992,
      "step": 205
    },
    {
      "epoch": 0.11060031542565686,
      "grad_norm": 0.703125,
      "learning_rate": 4.9945223113315966e-05,
      "loss": 1.004,
      "step": 206
    },
    {
      "epoch": 0.11113721016073286,
      "grad_norm": 0.61328125,
      "learning_rate": 4.9944662946425995e-05,
      "loss": 0.8692,
      "step": 207
    },
    {
      "epoch": 0.11167410489580887,
      "grad_norm": 0.828125,
      "learning_rate": 4.994409993303409e-05,
      "loss": 1.3249,
      "step": 208
    },
    {
      "epoch": 0.11221099963088486,
      "grad_norm": 0.875,
      "learning_rate": 4.994353407320449e-05,
      "loss": 0.9644,
      "step": 209
    },
    {
      "epoch": 0.11274789436596087,
      "grad_norm": 0.9296875,
      "learning_rate": 4.994296536700177e-05,
      "loss": 1.1006,
      "step": 210
    },
    {
      "epoch": 0.11328478910103688,
      "grad_norm": 0.7578125,
      "learning_rate": 4.994239381449082e-05,
      "loss": 1.0991,
      "step": 211
    },
    {
      "epoch": 0.11382168383611288,
      "grad_norm": 0.87109375,
      "learning_rate": 4.994181941573687e-05,
      "loss": 1.1392,
      "step": 212
    },
    {
      "epoch": 0.11435857857118889,
      "grad_norm": 0.68359375,
      "learning_rate": 4.9941242170805465e-05,
      "loss": 0.9001,
      "step": 213
    },
    {
      "epoch": 0.11489547330626489,
      "grad_norm": 0.90625,
      "learning_rate": 4.994066207976247e-05,
      "loss": 1.1169,
      "step": 214
    },
    {
      "epoch": 0.1154323680413409,
      "grad_norm": 0.65234375,
      "learning_rate": 4.9940079142674076e-05,
      "loss": 1.0101,
      "step": 215
    },
    {
      "epoch": 0.11596926277641689,
      "grad_norm": 0.9453125,
      "learning_rate": 4.993949335960683e-05,
      "loss": 1.182,
      "step": 216
    },
    {
      "epoch": 0.1165061575114929,
      "grad_norm": 0.7109375,
      "learning_rate": 4.993890473062754e-05,
      "loss": 1.0099,
      "step": 217
    },
    {
      "epoch": 0.11704305224656891,
      "grad_norm": 0.7265625,
      "learning_rate": 4.9938313255803406e-05,
      "loss": 0.9841,
      "step": 218
    },
    {
      "epoch": 0.11757994698164491,
      "grad_norm": 0.91015625,
      "learning_rate": 4.9937718935201905e-05,
      "loss": 1.3115,
      "step": 219
    },
    {
      "epoch": 0.11811684171672092,
      "grad_norm": 0.7578125,
      "learning_rate": 4.993712176889086e-05,
      "loss": 1.0432,
      "step": 220
    },
    {
      "epoch": 0.11865373645179692,
      "grad_norm": 0.92578125,
      "learning_rate": 4.9936521756938426e-05,
      "loss": 1.0891,
      "step": 221
    },
    {
      "epoch": 0.11919063118687293,
      "grad_norm": 0.7265625,
      "learning_rate": 4.993591889941306e-05,
      "loss": 0.9407,
      "step": 222
    },
    {
      "epoch": 0.11972752592194892,
      "grad_norm": 0.75390625,
      "learning_rate": 4.993531319638356e-05,
      "loss": 1.0568,
      "step": 223
    },
    {
      "epoch": 0.12026442065702493,
      "grad_norm": 0.921875,
      "learning_rate": 4.993470464791904e-05,
      "loss": 1.2601,
      "step": 224
    },
    {
      "epoch": 0.12080131539210094,
      "grad_norm": 0.87890625,
      "learning_rate": 4.993409325408895e-05,
      "loss": 1.1111,
      "step": 225
    },
    {
      "epoch": 0.12133821012717694,
      "grad_norm": 0.80078125,
      "learning_rate": 4.9933479014963055e-05,
      "loss": 1.1189,
      "step": 226
    },
    {
      "epoch": 0.12187510486225295,
      "grad_norm": 0.88671875,
      "learning_rate": 4.9932861930611454e-05,
      "loss": 0.8981,
      "step": 227
    },
    {
      "epoch": 0.12241199959732894,
      "grad_norm": 0.671875,
      "learning_rate": 4.9932242001104556e-05,
      "loss": 0.8803,
      "step": 228
    },
    {
      "epoch": 0.12294889433240495,
      "grad_norm": 0.6640625,
      "learning_rate": 4.99316192265131e-05,
      "loss": 0.8567,
      "step": 229
    },
    {
      "epoch": 0.12348578906748096,
      "grad_norm": 0.73828125,
      "learning_rate": 4.9930993606908154e-05,
      "loss": 0.9962,
      "step": 230
    },
    {
      "epoch": 0.12402268380255696,
      "grad_norm": 0.76953125,
      "learning_rate": 4.9930365142361115e-05,
      "loss": 1.0204,
      "step": 231
    },
    {
      "epoch": 0.12455957853763297,
      "grad_norm": 0.8046875,
      "learning_rate": 4.99297338329437e-05,
      "loss": 1.2177,
      "step": 232
    },
    {
      "epoch": 0.12509647327270898,
      "grad_norm": 1.0,
      "learning_rate": 4.992909967872794e-05,
      "loss": 1.0827,
      "step": 233
    },
    {
      "epoch": 0.12563336800778496,
      "grad_norm": 0.79296875,
      "learning_rate": 4.992846267978621e-05,
      "loss": 1.137,
      "step": 234
    },
    {
      "epoch": 0.12617026274286097,
      "grad_norm": 0.8515625,
      "learning_rate": 4.992782283619118e-05,
      "loss": 0.9979,
      "step": 235
    },
    {
      "epoch": 0.12670715747793698,
      "grad_norm": 0.80859375,
      "learning_rate": 4.99271801480159e-05,
      "loss": 0.9574,
      "step": 236
    },
    {
      "epoch": 0.127244052213013,
      "grad_norm": 0.71875,
      "learning_rate": 4.9926534615333675e-05,
      "loss": 1.2623,
      "step": 237
    },
    {
      "epoch": 0.127780946948089,
      "grad_norm": 0.7734375,
      "learning_rate": 4.992588623821819e-05,
      "loss": 0.9104,
      "step": 238
    },
    {
      "epoch": 0.12831784168316498,
      "grad_norm": 0.93359375,
      "learning_rate": 4.992523501674342e-05,
      "loss": 1.0249,
      "step": 239
    },
    {
      "epoch": 0.128854736418241,
      "grad_norm": 0.734375,
      "learning_rate": 4.992458095098368e-05,
      "loss": 0.9328,
      "step": 240
    },
    {
      "epoch": 0.129391631153317,
      "grad_norm": 0.75390625,
      "learning_rate": 4.992392404101361e-05,
      "loss": 1.0394,
      "step": 241
    },
    {
      "epoch": 0.12992852588839301,
      "grad_norm": 0.8046875,
      "learning_rate": 4.9923264286908164e-05,
      "loss": 1.0472,
      "step": 242
    },
    {
      "epoch": 0.130465420623469,
      "grad_norm": 1.046875,
      "learning_rate": 4.992260168874265e-05,
      "loss": 1.0126,
      "step": 243
    },
    {
      "epoch": 0.131002315358545,
      "grad_norm": 0.83984375,
      "learning_rate": 4.9921936246592656e-05,
      "loss": 0.9135,
      "step": 244
    },
    {
      "epoch": 0.13153921009362102,
      "grad_norm": 0.796875,
      "learning_rate": 4.9921267960534124e-05,
      "loss": 1.1838,
      "step": 245
    },
    {
      "epoch": 0.13207610482869703,
      "grad_norm": 0.7734375,
      "learning_rate": 4.992059683064332e-05,
      "loss": 1.0228,
      "step": 246
    },
    {
      "epoch": 0.13261299956377304,
      "grad_norm": 0.703125,
      "learning_rate": 4.991992285699682e-05,
      "loss": 0.9633,
      "step": 247
    },
    {
      "epoch": 0.13314989429884902,
      "grad_norm": 0.765625,
      "learning_rate": 4.991924603967154e-05,
      "loss": 1.0594,
      "step": 248
    },
    {
      "epoch": 0.13368678903392503,
      "grad_norm": 0.90625,
      "learning_rate": 4.9918566378744704e-05,
      "loss": 1.3159,
      "step": 249
    },
    {
      "epoch": 0.13422368376900104,
      "grad_norm": 1.0625,
      "learning_rate": 4.991788387429388e-05,
      "loss": 1.2449,
      "step": 250
    },
    {
      "epoch": 0.13422368376900104,
      "eval_loss": 1.0114037990570068,
      "eval_runtime": 45.6876,
      "eval_samples_per_second": 21.056,
      "eval_steps_per_second": 21.056,
      "step": 250
    },
    {
      "epoch": 0.13476057850407705,
      "grad_norm": 0.74609375,
      "learning_rate": 4.9917198526396945e-05,
      "loss": 1.0977,
      "step": 251
    },
    {
      "epoch": 0.13529747323915306,
      "grad_norm": 0.734375,
      "learning_rate": 4.991651033513212e-05,
      "loss": 0.9752,
      "step": 252
    },
    {
      "epoch": 0.13583436797422904,
      "grad_norm": 0.87890625,
      "learning_rate": 4.99158193005779e-05,
      "loss": 1.1368,
      "step": 253
    },
    {
      "epoch": 0.13637126270930505,
      "grad_norm": 0.71875,
      "learning_rate": 4.9915125422813187e-05,
      "loss": 1.019,
      "step": 254
    },
    {
      "epoch": 0.13690815744438106,
      "grad_norm": 0.88671875,
      "learning_rate": 4.991442870191713e-05,
      "loss": 1.133,
      "step": 255
    },
    {
      "epoch": 0.13744505217945707,
      "grad_norm": 0.8125,
      "learning_rate": 4.991372913796924e-05,
      "loss": 1.2847,
      "step": 256
    },
    {
      "epoch": 0.13798194691453308,
      "grad_norm": 0.76171875,
      "learning_rate": 4.991302673104935e-05,
      "loss": 0.9674,
      "step": 257
    },
    {
      "epoch": 0.13851884164960906,
      "grad_norm": 0.8515625,
      "learning_rate": 4.991232148123761e-05,
      "loss": 1.2452,
      "step": 258
    },
    {
      "epoch": 0.13905573638468507,
      "grad_norm": 0.75390625,
      "learning_rate": 4.99116133886145e-05,
      "loss": 0.926,
      "step": 259
    },
    {
      "epoch": 0.13959263111976108,
      "grad_norm": 0.89453125,
      "learning_rate": 4.9910902453260824e-05,
      "loss": 1.0061,
      "step": 260
    },
    {
      "epoch": 0.1401295258548371,
      "grad_norm": 0.9296875,
      "learning_rate": 4.991018867525771e-05,
      "loss": 1.2372,
      "step": 261
    },
    {
      "epoch": 0.14066642058991308,
      "grad_norm": 0.6953125,
      "learning_rate": 4.99094720546866e-05,
      "loss": 0.9288,
      "step": 262
    },
    {
      "epoch": 0.1412033153249891,
      "grad_norm": 0.7734375,
      "learning_rate": 4.990875259162928e-05,
      "loss": 1.1232,
      "step": 263
    },
    {
      "epoch": 0.1417402100600651,
      "grad_norm": 0.83984375,
      "learning_rate": 4.990803028616785e-05,
      "loss": 1.0133,
      "step": 264
    },
    {
      "epoch": 0.1422771047951411,
      "grad_norm": 0.859375,
      "learning_rate": 4.990730513838472e-05,
      "loss": 1.1617,
      "step": 265
    },
    {
      "epoch": 0.14281399953021712,
      "grad_norm": 1.0390625,
      "learning_rate": 4.990657714836266e-05,
      "loss": 1.2293,
      "step": 266
    },
    {
      "epoch": 0.1433508942652931,
      "grad_norm": 0.83984375,
      "learning_rate": 4.990584631618472e-05,
      "loss": 1.2229,
      "step": 267
    },
    {
      "epoch": 0.1438877890003691,
      "grad_norm": 0.71875,
      "learning_rate": 4.990511264193431e-05,
      "loss": 1.0121,
      "step": 268
    },
    {
      "epoch": 0.14442468373544512,
      "grad_norm": 0.75,
      "learning_rate": 4.990437612569515e-05,
      "loss": 0.9549,
      "step": 269
    },
    {
      "epoch": 0.14496157847052113,
      "grad_norm": 0.9765625,
      "learning_rate": 4.9903636767551285e-05,
      "loss": 1.0386,
      "step": 270
    },
    {
      "epoch": 0.14549847320559714,
      "grad_norm": 1.0,
      "learning_rate": 4.990289456758709e-05,
      "loss": 1.0869,
      "step": 271
    },
    {
      "epoch": 0.14603536794067312,
      "grad_norm": 0.703125,
      "learning_rate": 4.9902149525887255e-05,
      "loss": 1.0721,
      "step": 272
    },
    {
      "epoch": 0.14657226267574913,
      "grad_norm": 0.68359375,
      "learning_rate": 4.990140164253679e-05,
      "loss": 0.9355,
      "step": 273
    },
    {
      "epoch": 0.14710915741082514,
      "grad_norm": 0.8671875,
      "learning_rate": 4.990065091762106e-05,
      "loss": 1.0889,
      "step": 274
    },
    {
      "epoch": 0.14764605214590115,
      "grad_norm": 0.73828125,
      "learning_rate": 4.989989735122571e-05,
      "loss": 0.8915,
      "step": 275
    },
    {
      "epoch": 0.14818294688097716,
      "grad_norm": 0.734375,
      "learning_rate": 4.989914094343675e-05,
      "loss": 1.046,
      "step": 276
    },
    {
      "epoch": 0.14871984161605314,
      "grad_norm": 0.828125,
      "learning_rate": 4.989838169434048e-05,
      "loss": 1.0452,
      "step": 277
    },
    {
      "epoch": 0.14925673635112915,
      "grad_norm": 1.0546875,
      "learning_rate": 4.9897619604023545e-05,
      "loss": 1.2057,
      "step": 278
    },
    {
      "epoch": 0.14979363108620516,
      "grad_norm": 0.93359375,
      "learning_rate": 4.989685467257291e-05,
      "loss": 1.243,
      "step": 279
    },
    {
      "epoch": 0.15033052582128117,
      "grad_norm": 0.8125,
      "learning_rate": 4.9896086900075865e-05,
      "loss": 1.1677,
      "step": 280
    },
    {
      "epoch": 0.15086742055635716,
      "grad_norm": 0.96484375,
      "learning_rate": 4.989531628662002e-05,
      "loss": 1.1926,
      "step": 281
    },
    {
      "epoch": 0.15140431529143317,
      "grad_norm": 0.9453125,
      "learning_rate": 4.989454283229331e-05,
      "loss": 0.9606,
      "step": 282
    },
    {
      "epoch": 0.15194121002650918,
      "grad_norm": 0.7578125,
      "learning_rate": 4.9893766537184006e-05,
      "loss": 1.0276,
      "step": 283
    },
    {
      "epoch": 0.1524781047615852,
      "grad_norm": 0.75390625,
      "learning_rate": 4.9892987401380686e-05,
      "loss": 1.1926,
      "step": 284
    },
    {
      "epoch": 0.1530149994966612,
      "grad_norm": 0.97265625,
      "learning_rate": 4.989220542497226e-05,
      "loss": 0.974,
      "step": 285
    },
    {
      "epoch": 0.15355189423173718,
      "grad_norm": 0.8125,
      "learning_rate": 4.989142060804796e-05,
      "loss": 1.0926,
      "step": 286
    },
    {
      "epoch": 0.1540887889668132,
      "grad_norm": 1.0078125,
      "learning_rate": 4.989063295069733e-05,
      "loss": 1.2385,
      "step": 287
    },
    {
      "epoch": 0.1546256837018892,
      "grad_norm": 0.8671875,
      "learning_rate": 4.988984245301028e-05,
      "loss": 1.1706,
      "step": 288
    },
    {
      "epoch": 0.1551625784369652,
      "grad_norm": 0.765625,
      "learning_rate": 4.9889049115077005e-05,
      "loss": 0.9743,
      "step": 289
    },
    {
      "epoch": 0.15569947317204122,
      "grad_norm": 0.8046875,
      "learning_rate": 4.988825293698802e-05,
      "loss": 1.1596,
      "step": 290
    },
    {
      "epoch": 0.1562363679071172,
      "grad_norm": 0.8046875,
      "learning_rate": 4.988745391883421e-05,
      "loss": 1.126,
      "step": 291
    },
    {
      "epoch": 0.1567732626421932,
      "grad_norm": 0.6328125,
      "learning_rate": 4.988665206070671e-05,
      "loss": 1.0426,
      "step": 292
    },
    {
      "epoch": 0.15731015737726922,
      "grad_norm": 0.77734375,
      "learning_rate": 4.988584736269706e-05,
      "loss": 1.0523,
      "step": 293
    },
    {
      "epoch": 0.15784705211234523,
      "grad_norm": 0.74609375,
      "learning_rate": 4.988503982489707e-05,
      "loss": 0.923,
      "step": 294
    },
    {
      "epoch": 0.15838394684742121,
      "grad_norm": 0.85546875,
      "learning_rate": 4.988422944739889e-05,
      "loss": 1.1883,
      "step": 295
    },
    {
      "epoch": 0.15892084158249722,
      "grad_norm": 0.76171875,
      "learning_rate": 4.988341623029499e-05,
      "loss": 1.0239,
      "step": 296
    },
    {
      "epoch": 0.15945773631757323,
      "grad_norm": 1.171875,
      "learning_rate": 4.9882600173678186e-05,
      "loss": 0.9064,
      "step": 297
    },
    {
      "epoch": 0.15999463105264924,
      "grad_norm": 0.73828125,
      "learning_rate": 4.9881781277641586e-05,
      "loss": 0.9026,
      "step": 298
    },
    {
      "epoch": 0.16053152578772525,
      "grad_norm": 0.78515625,
      "learning_rate": 4.9880959542278645e-05,
      "loss": 1.0779,
      "step": 299
    },
    {
      "epoch": 0.16106842052280124,
      "grad_norm": 0.74609375,
      "learning_rate": 4.9880134967683124e-05,
      "loss": 0.7856,
      "step": 300
    },
    {
      "epoch": 0.16106842052280124,
      "eval_loss": 1.0021421909332275,
      "eval_runtime": 45.8583,
      "eval_samples_per_second": 20.978,
      "eval_steps_per_second": 20.978,
      "step": 300
    },
    {
      "epoch": 0.16160531525787725,
      "grad_norm": 0.796875,
      "learning_rate": 4.987930755394912e-05,
      "loss": 0.9734,
      "step": 301
    },
    {
      "epoch": 0.16214220999295326,
      "grad_norm": 0.890625,
      "learning_rate": 4.987847730117106e-05,
      "loss": 1.0082,
      "step": 302
    },
    {
      "epoch": 0.16267910472802927,
      "grad_norm": 0.80078125,
      "learning_rate": 4.987764420944367e-05,
      "loss": 0.9955,
      "step": 303
    },
    {
      "epoch": 0.16321599946310528,
      "grad_norm": 0.8359375,
      "learning_rate": 4.987680827886203e-05,
      "loss": 1.1873,
      "step": 304
    },
    {
      "epoch": 0.16375289419818126,
      "grad_norm": 0.94140625,
      "learning_rate": 4.9875969509521526e-05,
      "loss": 0.9822,
      "step": 305
    },
    {
      "epoch": 0.16428978893325727,
      "grad_norm": 0.73046875,
      "learning_rate": 4.987512790151787e-05,
      "loss": 0.9305,
      "step": 306
    },
    {
      "epoch": 0.16482668366833328,
      "grad_norm": 0.8671875,
      "learning_rate": 4.98742834549471e-05,
      "loss": 1.1409,
      "step": 307
    },
    {
      "epoch": 0.1653635784034093,
      "grad_norm": 0.765625,
      "learning_rate": 4.987343616990559e-05,
      "loss": 1.0024,
      "step": 308
    },
    {
      "epoch": 0.1659004731384853,
      "grad_norm": 0.83203125,
      "learning_rate": 4.987258604649001e-05,
      "loss": 1.0753,
      "step": 309
    },
    {
      "epoch": 0.16643736787356128,
      "grad_norm": 0.8515625,
      "learning_rate": 4.987173308479738e-05,
      "loss": 1.0717,
      "step": 310
    },
    {
      "epoch": 0.1669742626086373,
      "grad_norm": 0.76953125,
      "learning_rate": 4.987087728492503e-05,
      "loss": 1.2467,
      "step": 311
    },
    {
      "epoch": 0.1675111573437133,
      "grad_norm": 0.67578125,
      "learning_rate": 4.987001864697062e-05,
      "loss": 0.8469,
      "step": 312
    },
    {
      "epoch": 0.1680480520787893,
      "grad_norm": 0.87109375,
      "learning_rate": 4.986915717103212e-05,
      "loss": 1.234,
      "step": 313
    },
    {
      "epoch": 0.1685849468138653,
      "grad_norm": 0.98046875,
      "learning_rate": 4.986829285720785e-05,
      "loss": 1.0514,
      "step": 314
    },
    {
      "epoch": 0.1691218415489413,
      "grad_norm": 0.71875,
      "learning_rate": 4.9867425705596434e-05,
      "loss": 1.0181,
      "step": 315
    },
    {
      "epoch": 0.16965873628401731,
      "grad_norm": 0.8515625,
      "learning_rate": 4.986655571629682e-05,
      "loss": 1.0346,
      "step": 316
    },
    {
      "epoch": 0.17019563101909332,
      "grad_norm": 0.87890625,
      "learning_rate": 4.9865682889408295e-05,
      "loss": 1.1612,
      "step": 317
    },
    {
      "epoch": 0.17073252575416933,
      "grad_norm": 0.8671875,
      "learning_rate": 4.9864807225030454e-05,
      "loss": 0.8934,
      "step": 318
    },
    {
      "epoch": 0.17126942048924532,
      "grad_norm": 0.6640625,
      "learning_rate": 4.986392872326322e-05,
      "loss": 0.8901,
      "step": 319
    },
    {
      "epoch": 0.17180631522432133,
      "grad_norm": 0.7578125,
      "learning_rate": 4.9863047384206835e-05,
      "loss": 1.072,
      "step": 320
    },
    {
      "epoch": 0.17234320995939734,
      "grad_norm": 0.7265625,
      "learning_rate": 4.9862163207961887e-05,
      "loss": 1.1071,
      "step": 321
    },
    {
      "epoch": 0.17288010469447335,
      "grad_norm": 0.734375,
      "learning_rate": 4.9861276194629256e-05,
      "loss": 1.0346,
      "step": 322
    },
    {
      "epoch": 0.17341699942954936,
      "grad_norm": 0.8203125,
      "learning_rate": 4.9860386344310175e-05,
      "loss": 1.0508,
      "step": 323
    },
    {
      "epoch": 0.17395389416462534,
      "grad_norm": 0.67578125,
      "learning_rate": 4.9859493657106185e-05,
      "loss": 0.9795,
      "step": 324
    },
    {
      "epoch": 0.17449078889970135,
      "grad_norm": 0.77734375,
      "learning_rate": 4.985859813311914e-05,
      "loss": 1.096,
      "step": 325
    },
    {
      "epoch": 0.17502768363477736,
      "grad_norm": 0.875,
      "learning_rate": 4.985769977245124e-05,
      "loss": 1.2532,
      "step": 326
    },
    {
      "epoch": 0.17556457836985337,
      "grad_norm": 0.81640625,
      "learning_rate": 4.9856798575205e-05,
      "loss": 1.2384,
      "step": 327
    },
    {
      "epoch": 0.17610147310492935,
      "grad_norm": 0.94140625,
      "learning_rate": 4.985589454148326e-05,
      "loss": 1.1486,
      "step": 328
    },
    {
      "epoch": 0.17663836784000536,
      "grad_norm": 0.70703125,
      "learning_rate": 4.9854987671389174e-05,
      "loss": 0.9248,
      "step": 329
    },
    {
      "epoch": 0.17717526257508137,
      "grad_norm": 0.80859375,
      "learning_rate": 4.9854077965026234e-05,
      "loss": 0.9731,
      "step": 330
    },
    {
      "epoch": 0.17771215731015738,
      "grad_norm": 0.69921875,
      "learning_rate": 4.985316542249825e-05,
      "loss": 0.8214,
      "step": 331
    },
    {
      "epoch": 0.1782490520452334,
      "grad_norm": 0.7890625,
      "learning_rate": 4.985225004390934e-05,
      "loss": 0.8996,
      "step": 332
    },
    {
      "epoch": 0.17878594678030937,
      "grad_norm": 0.98046875,
      "learning_rate": 4.985133182936399e-05,
      "loss": 1.2789,
      "step": 333
    },
    {
      "epoch": 0.17932284151538538,
      "grad_norm": 0.80078125,
      "learning_rate": 4.985041077896695e-05,
      "loss": 1.1322,
      "step": 334
    },
    {
      "epoch": 0.1798597362504614,
      "grad_norm": 0.79296875,
      "learning_rate": 4.984948689282333e-05,
      "loss": 1.0258,
      "step": 335
    },
    {
      "epoch": 0.1803966309855374,
      "grad_norm": 0.7265625,
      "learning_rate": 4.984856017103857e-05,
      "loss": 0.8328,
      "step": 336
    },
    {
      "epoch": 0.18093352572061341,
      "grad_norm": 0.76171875,
      "learning_rate": 4.984763061371841e-05,
      "loss": 1.2591,
      "step": 337
    },
    {
      "epoch": 0.1814704204556894,
      "grad_norm": 0.78515625,
      "learning_rate": 4.9846698220968934e-05,
      "loss": 1.0533,
      "step": 338
    },
    {
      "epoch": 0.1820073151907654,
      "grad_norm": 0.98828125,
      "learning_rate": 4.9845762992896526e-05,
      "loss": 1.076,
      "step": 339
    },
    {
      "epoch": 0.18254420992584142,
      "grad_norm": 0.84375,
      "learning_rate": 4.984482492960791e-05,
      "loss": 0.9798,
      "step": 340
    },
    {
      "epoch": 0.18308110466091743,
      "grad_norm": 0.83203125,
      "learning_rate": 4.984388403121015e-05,
      "loss": 1.0483,
      "step": 341
    },
    {
      "epoch": 0.18361799939599344,
      "grad_norm": 0.78125,
      "learning_rate": 4.984294029781059e-05,
      "loss": 1.2484,
      "step": 342
    },
    {
      "epoch": 0.18415489413106942,
      "grad_norm": 1.125,
      "learning_rate": 4.9841993729516933e-05,
      "loss": 1.3144,
      "step": 343
    },
    {
      "epoch": 0.18469178886614543,
      "grad_norm": 0.66796875,
      "learning_rate": 4.9841044326437194e-05,
      "loss": 1.0227,
      "step": 344
    },
    {
      "epoch": 0.18522868360122144,
      "grad_norm": 0.7421875,
      "learning_rate": 4.984009208867971e-05,
      "loss": 0.9629,
      "step": 345
    },
    {
      "epoch": 0.18576557833629745,
      "grad_norm": 0.86328125,
      "learning_rate": 4.9839137016353147e-05,
      "loss": 1.0943,
      "step": 346
    },
    {
      "epoch": 0.18630247307137343,
      "grad_norm": 0.90234375,
      "learning_rate": 4.983817910956648e-05,
      "loss": 1.1398,
      "step": 347
    },
    {
      "epoch": 0.18683936780644944,
      "grad_norm": 0.90234375,
      "learning_rate": 4.983721836842903e-05,
      "loss": 0.9662,
      "step": 348
    },
    {
      "epoch": 0.18737626254152545,
      "grad_norm": 0.7265625,
      "learning_rate": 4.983625479305043e-05,
      "loss": 0.9173,
      "step": 349
    },
    {
      "epoch": 0.18791315727660146,
      "grad_norm": 0.95703125,
      "learning_rate": 4.9835288383540626e-05,
      "loss": 1.1645,
      "step": 350
    },
    {
      "epoch": 0.18791315727660146,
      "eval_loss": 0.9941990971565247,
      "eval_runtime": 45.8692,
      "eval_samples_per_second": 20.973,
      "eval_steps_per_second": 20.973,
      "step": 350
    },
    {
      "epoch": 0.18845005201167747,
      "grad_norm": 0.85546875,
      "learning_rate": 4.983431914000991e-05,
      "loss": 1.0875,
      "step": 351
    },
    {
      "epoch": 0.18898694674675345,
      "grad_norm": 0.78125,
      "learning_rate": 4.983334706256888e-05,
      "loss": 1.0765,
      "step": 352
    },
    {
      "epoch": 0.18952384148182946,
      "grad_norm": 0.80078125,
      "learning_rate": 4.983237215132846e-05,
      "loss": 1.2291,
      "step": 353
    },
    {
      "epoch": 0.19006073621690547,
      "grad_norm": 0.78515625,
      "learning_rate": 4.98313944063999e-05,
      "loss": 0.8445,
      "step": 354
    },
    {
      "epoch": 0.19059763095198148,
      "grad_norm": 0.9296875,
      "learning_rate": 4.983041382789477e-05,
      "loss": 1.2183,
      "step": 355
    },
    {
      "epoch": 0.1911345256870575,
      "grad_norm": 0.796875,
      "learning_rate": 4.9829430415924974e-05,
      "loss": 1.1945,
      "step": 356
    },
    {
      "epoch": 0.19167142042213348,
      "grad_norm": 0.80859375,
      "learning_rate": 4.982844417060273e-05,
      "loss": 0.9253,
      "step": 357
    },
    {
      "epoch": 0.1922083151572095,
      "grad_norm": 0.7734375,
      "learning_rate": 4.982745509204058e-05,
      "loss": 1.0803,
      "step": 358
    },
    {
      "epoch": 0.1927452098922855,
      "grad_norm": 0.8515625,
      "learning_rate": 4.982646318035139e-05,
      "loss": 0.9815,
      "step": 359
    },
    {
      "epoch": 0.1932821046273615,
      "grad_norm": 0.7890625,
      "learning_rate": 4.982546843564834e-05,
      "loss": 0.8256,
      "step": 360
    },
    {
      "epoch": 0.1938189993624375,
      "grad_norm": 1.2265625,
      "learning_rate": 4.982447085804496e-05,
      "loss": 1.0912,
      "step": 361
    },
    {
      "epoch": 0.1943558940975135,
      "grad_norm": 0.8828125,
      "learning_rate": 4.982347044765508e-05,
      "loss": 1.1494,
      "step": 362
    },
    {
      "epoch": 0.1948927888325895,
      "grad_norm": 0.71875,
      "learning_rate": 4.982246720459286e-05,
      "loss": 1.1104,
      "step": 363
    },
    {
      "epoch": 0.19542968356766552,
      "grad_norm": 0.73828125,
      "learning_rate": 4.982146112897277e-05,
      "loss": 1.0217,
      "step": 364
    },
    {
      "epoch": 0.19596657830274153,
      "grad_norm": 0.734375,
      "learning_rate": 4.982045222090964e-05,
      "loss": 0.9799,
      "step": 365
    },
    {
      "epoch": 0.1965034730378175,
      "grad_norm": 0.984375,
      "learning_rate": 4.9819440480518574e-05,
      "loss": 1.0773,
      "step": 366
    },
    {
      "epoch": 0.19704036777289352,
      "grad_norm": 0.69921875,
      "learning_rate": 4.9818425907915045e-05,
      "loss": 0.9679,
      "step": 367
    },
    {
      "epoch": 0.19757726250796953,
      "grad_norm": 0.76953125,
      "learning_rate": 4.981740850321481e-05,
      "loss": 0.9586,
      "step": 368
    },
    {
      "epoch": 0.19811415724304554,
      "grad_norm": 0.6875,
      "learning_rate": 4.9816388266533994e-05,
      "loss": 0.9833,
      "step": 369
    },
    {
      "epoch": 0.19865105197812155,
      "grad_norm": 0.9921875,
      "learning_rate": 4.9815365197988986e-05,
      "loss": 1.0994,
      "step": 370
    },
    {
      "epoch": 0.19918794671319753,
      "grad_norm": 0.8359375,
      "learning_rate": 4.9814339297696555e-05,
      "loss": 1.0632,
      "step": 371
    },
    {
      "epoch": 0.19972484144827354,
      "grad_norm": 0.8671875,
      "learning_rate": 4.981331056577376e-05,
      "loss": 0.9068,
      "step": 372
    },
    {
      "epoch": 0.20026173618334955,
      "grad_norm": 0.7265625,
      "learning_rate": 4.9812279002338e-05,
      "loss": 1.0384,
      "step": 373
    },
    {
      "epoch": 0.20079863091842556,
      "grad_norm": 1.0,
      "learning_rate": 4.981124460750698e-05,
      "loss": 1.4422,
      "step": 374
    },
    {
      "epoch": 0.20133552565350157,
      "grad_norm": 0.78515625,
      "learning_rate": 4.981020738139874e-05,
      "loss": 0.8491,
      "step": 375
    },
    {
      "epoch": 0.20187242038857756,
      "grad_norm": 0.78125,
      "learning_rate": 4.9809167324131645e-05,
      "loss": 0.9107,
      "step": 376
    },
    {
      "epoch": 0.20240931512365357,
      "grad_norm": 0.9296875,
      "learning_rate": 4.980812443582438e-05,
      "loss": 1.1261,
      "step": 377
    },
    {
      "epoch": 0.20294620985872958,
      "grad_norm": 0.63671875,
      "learning_rate": 4.980707871659593e-05,
      "loss": 0.9291,
      "step": 378
    },
    {
      "epoch": 0.20348310459380559,
      "grad_norm": 0.8125,
      "learning_rate": 4.980603016656567e-05,
      "loss": 1.0956,
      "step": 379
    },
    {
      "epoch": 0.20401999932888157,
      "grad_norm": 0.80859375,
      "learning_rate": 4.9804978785853196e-05,
      "loss": 1.1436,
      "step": 380
    },
    {
      "epoch": 0.20455689406395758,
      "grad_norm": 0.85546875,
      "learning_rate": 4.980392457457853e-05,
      "loss": 1.1608,
      "step": 381
    },
    {
      "epoch": 0.2050937887990336,
      "grad_norm": 0.703125,
      "learning_rate": 4.980286753286195e-05,
      "loss": 1.1159,
      "step": 382
    },
    {
      "epoch": 0.2056306835341096,
      "grad_norm": 0.85546875,
      "learning_rate": 4.980180766082408e-05,
      "loss": 1.1763,
      "step": 383
    },
    {
      "epoch": 0.2061675782691856,
      "grad_norm": 0.73828125,
      "learning_rate": 4.9800744958585864e-05,
      "loss": 1.0288,
      "step": 384
    },
    {
      "epoch": 0.2067044730042616,
      "grad_norm": 0.78515625,
      "learning_rate": 4.979967942626858e-05,
      "loss": 1.1797,
      "step": 385
    },
    {
      "epoch": 0.2072413677393376,
      "grad_norm": 0.7578125,
      "learning_rate": 4.9798611063993805e-05,
      "loss": 0.7979,
      "step": 386
    },
    {
      "epoch": 0.2077782624744136,
      "grad_norm": 0.72265625,
      "learning_rate": 4.9797539871883456e-05,
      "loss": 1.0582,
      "step": 387
    },
    {
      "epoch": 0.20831515720948962,
      "grad_norm": 0.76953125,
      "learning_rate": 4.979646585005978e-05,
      "loss": 1.1325,
      "step": 388
    },
    {
      "epoch": 0.20885205194456563,
      "grad_norm": 0.9140625,
      "learning_rate": 4.979538899864532e-05,
      "loss": 1.0,
      "step": 389
    },
    {
      "epoch": 0.2093889466796416,
      "grad_norm": 0.6875,
      "learning_rate": 4.979430931776298e-05,
      "loss": 0.8141,
      "step": 390
    },
    {
      "epoch": 0.20992584141471762,
      "grad_norm": 0.7109375,
      "learning_rate": 4.9793226807535944e-05,
      "loss": 0.8747,
      "step": 391
    },
    {
      "epoch": 0.21046273614979363,
      "grad_norm": 0.86328125,
      "learning_rate": 4.9792141468087746e-05,
      "loss": 0.9623,
      "step": 392
    },
    {
      "epoch": 0.21099963088486964,
      "grad_norm": 0.78515625,
      "learning_rate": 4.979105329954224e-05,
      "loss": 0.8694,
      "step": 393
    },
    {
      "epoch": 0.21153652561994563,
      "grad_norm": 0.7734375,
      "learning_rate": 4.97899623020236e-05,
      "loss": 0.8215,
      "step": 394
    },
    {
      "epoch": 0.21207342035502164,
      "grad_norm": 0.98046875,
      "learning_rate": 4.9788868475656325e-05,
      "loss": 1.1688,
      "step": 395
    },
    {
      "epoch": 0.21261031509009765,
      "grad_norm": 0.83984375,
      "learning_rate": 4.978777182056523e-05,
      "loss": 1.0416,
      "step": 396
    },
    {
      "epoch": 0.21314720982517366,
      "grad_norm": 0.671875,
      "learning_rate": 4.978667233687546e-05,
      "loss": 1.0144,
      "step": 397
    },
    {
      "epoch": 0.21368410456024967,
      "grad_norm": 0.78515625,
      "learning_rate": 4.9785570024712475e-05,
      "loss": 1.0425,
      "step": 398
    },
    {
      "epoch": 0.21422099929532565,
      "grad_norm": 0.875,
      "learning_rate": 4.9784464884202067e-05,
      "loss": 1.1703,
      "step": 399
    },
    {
      "epoch": 0.21475789403040166,
      "grad_norm": 0.73828125,
      "learning_rate": 4.9783356915470344e-05,
      "loss": 1.168,
      "step": 400
    },
    {
      "epoch": 0.21475789403040166,
      "eval_loss": 0.9871569871902466,
      "eval_runtime": 45.8833,
      "eval_samples_per_second": 20.966,
      "eval_steps_per_second": 20.966,
      "step": 400
    },
    {
      "epoch": 0.21529478876547767,
      "grad_norm": 0.75,
      "learning_rate": 4.978224611864375e-05,
      "loss": 0.9708,
      "step": 401
    },
    {
      "epoch": 0.21583168350055368,
      "grad_norm": 0.76953125,
      "learning_rate": 4.9781132493849025e-05,
      "loss": 1.0356,
      "step": 402
    },
    {
      "epoch": 0.2163685782356297,
      "grad_norm": 0.74609375,
      "learning_rate": 4.9780016041213254e-05,
      "loss": 1.0487,
      "step": 403
    },
    {
      "epoch": 0.21690547297070567,
      "grad_norm": 0.875,
      "learning_rate": 4.977889676086383e-05,
      "loss": 1.0294,
      "step": 404
    },
    {
      "epoch": 0.21744236770578168,
      "grad_norm": 0.94140625,
      "learning_rate": 4.977777465292851e-05,
      "loss": 1.3656,
      "step": 405
    },
    {
      "epoch": 0.2179792624408577,
      "grad_norm": 0.8046875,
      "learning_rate": 4.97766497175353e-05,
      "loss": 1.1777,
      "step": 406
    },
    {
      "epoch": 0.2185161571759337,
      "grad_norm": 0.86328125,
      "learning_rate": 4.977552195481259e-05,
      "loss": 0.9866,
      "step": 407
    },
    {
      "epoch": 0.2190530519110097,
      "grad_norm": 0.76171875,
      "learning_rate": 4.977439136488907e-05,
      "loss": 1.0517,
      "step": 408
    },
    {
      "epoch": 0.2195899466460857,
      "grad_norm": 1.015625,
      "learning_rate": 4.977325794789375e-05,
      "loss": 1.1073,
      "step": 409
    },
    {
      "epoch": 0.2201268413811617,
      "grad_norm": 0.8671875,
      "learning_rate": 4.977212170395598e-05,
      "loss": 0.8281,
      "step": 410
    },
    {
      "epoch": 0.2206637361162377,
      "grad_norm": 0.88671875,
      "learning_rate": 4.97709826332054e-05,
      "loss": 1.1745,
      "step": 411
    },
    {
      "epoch": 0.22120063085131372,
      "grad_norm": 0.87890625,
      "learning_rate": 4.9769840735772e-05,
      "loss": 1.0383,
      "step": 412
    },
    {
      "epoch": 0.2217375255863897,
      "grad_norm": 0.6953125,
      "learning_rate": 4.976869601178609e-05,
      "loss": 1.0757,
      "step": 413
    },
    {
      "epoch": 0.22227442032146572,
      "grad_norm": 0.81640625,
      "learning_rate": 4.9767548461378296e-05,
      "loss": 1.0685,
      "step": 414
    },
    {
      "epoch": 0.22281131505654173,
      "grad_norm": 0.83203125,
      "learning_rate": 4.976639808467957e-05,
      "loss": 1.0155,
      "step": 415
    },
    {
      "epoch": 0.22334820979161774,
      "grad_norm": 0.79296875,
      "learning_rate": 4.976524488182118e-05,
      "loss": 1.0609,
      "step": 416
    },
    {
      "epoch": 0.22388510452669375,
      "grad_norm": 0.78515625,
      "learning_rate": 4.976408885293472e-05,
      "loss": 1.1626,
      "step": 417
    },
    {
      "epoch": 0.22442199926176973,
      "grad_norm": 0.6640625,
      "learning_rate": 4.976292999815211e-05,
      "loss": 0.9083,
      "step": 418
    },
    {
      "epoch": 0.22495889399684574,
      "grad_norm": 0.80078125,
      "learning_rate": 4.9761768317605596e-05,
      "loss": 1.0945,
      "step": 419
    },
    {
      "epoch": 0.22549578873192175,
      "grad_norm": 0.859375,
      "learning_rate": 4.976060381142773e-05,
      "loss": 0.8575,
      "step": 420
    },
    {
      "epoch": 0.22603268346699776,
      "grad_norm": 0.71484375,
      "learning_rate": 4.9759436479751406e-05,
      "loss": 1.0251,
      "step": 421
    },
    {
      "epoch": 0.22656957820207377,
      "grad_norm": 0.75390625,
      "learning_rate": 4.975826632270982e-05,
      "loss": 0.7913,
      "step": 422
    },
    {
      "epoch": 0.22710647293714975,
      "grad_norm": 0.76953125,
      "learning_rate": 4.975709334043651e-05,
      "loss": 0.9925,
      "step": 423
    },
    {
      "epoch": 0.22764336767222576,
      "grad_norm": 0.90625,
      "learning_rate": 4.975591753306533e-05,
      "loss": 1.0412,
      "step": 424
    },
    {
      "epoch": 0.22818026240730177,
      "grad_norm": 0.76171875,
      "learning_rate": 4.975473890073045e-05,
      "loss": 1.0033,
      "step": 425
    },
    {
      "epoch": 0.22871715714237778,
      "grad_norm": 0.80078125,
      "learning_rate": 4.975355744356637e-05,
      "loss": 0.9722,
      "step": 426
    },
    {
      "epoch": 0.2292540518774538,
      "grad_norm": 0.734375,
      "learning_rate": 4.97523731617079e-05,
      "loss": 0.8067,
      "step": 427
    },
    {
      "epoch": 0.22979094661252977,
      "grad_norm": 1.0,
      "learning_rate": 4.975118605529019e-05,
      "loss": 1.1088,
      "step": 428
    },
    {
      "epoch": 0.23032784134760578,
      "grad_norm": 0.6953125,
      "learning_rate": 4.974999612444871e-05,
      "loss": 0.8336,
      "step": 429
    },
    {
      "epoch": 0.2308647360826818,
      "grad_norm": 0.94921875,
      "learning_rate": 4.974880336931923e-05,
      "loss": 1.0894,
      "step": 430
    },
    {
      "epoch": 0.2314016308177578,
      "grad_norm": 1.0234375,
      "learning_rate": 4.974760779003787e-05,
      "loss": 1.2324,
      "step": 431
    },
    {
      "epoch": 0.23193852555283379,
      "grad_norm": 0.796875,
      "learning_rate": 4.974640938674107e-05,
      "loss": 1.2026,
      "step": 432
    },
    {
      "epoch": 0.2324754202879098,
      "grad_norm": 0.87890625,
      "learning_rate": 4.974520815956555e-05,
      "loss": 1.0034,
      "step": 433
    },
    {
      "epoch": 0.2330123150229858,
      "grad_norm": 0.76171875,
      "learning_rate": 4.974400410864842e-05,
      "loss": 0.9869,
      "step": 434
    },
    {
      "epoch": 0.23354920975806182,
      "grad_norm": 0.671875,
      "learning_rate": 4.974279723412706e-05,
      "loss": 0.9489,
      "step": 435
    },
    {
      "epoch": 0.23408610449313783,
      "grad_norm": 0.765625,
      "learning_rate": 4.9741587536139204e-05,
      "loss": 1.1474,
      "step": 436
    },
    {
      "epoch": 0.2346229992282138,
      "grad_norm": 0.640625,
      "learning_rate": 4.974037501482287e-05,
      "loss": 0.7989,
      "step": 437
    },
    {
      "epoch": 0.23515989396328982,
      "grad_norm": 0.73046875,
      "learning_rate": 4.973915967031644e-05,
      "loss": 0.9636,
      "step": 438
    },
    {
      "epoch": 0.23569678869836583,
      "grad_norm": 1.1953125,
      "learning_rate": 4.973794150275859e-05,
      "loss": 1.1533,
      "step": 439
    },
    {
      "epoch": 0.23623368343344184,
      "grad_norm": 0.7109375,
      "learning_rate": 4.9736720512288334e-05,
      "loss": 0.9366,
      "step": 440
    },
    {
      "epoch": 0.23677057816851785,
      "grad_norm": 1.2421875,
      "learning_rate": 4.973549669904501e-05,
      "loss": 1.1356,
      "step": 441
    },
    {
      "epoch": 0.23730747290359383,
      "grad_norm": 0.890625,
      "learning_rate": 4.973427006316826e-05,
      "loss": 1.1003,
      "step": 442
    },
    {
      "epoch": 0.23784436763866984,
      "grad_norm": 1.0234375,
      "learning_rate": 4.973304060479807e-05,
      "loss": 0.9756,
      "step": 443
    },
    {
      "epoch": 0.23838126237374585,
      "grad_norm": 0.76953125,
      "learning_rate": 4.9731808324074717e-05,
      "loss": 0.936,
      "step": 444
    },
    {
      "epoch": 0.23891815710882186,
      "grad_norm": 0.69140625,
      "learning_rate": 4.973057322113883e-05,
      "loss": 1.0434,
      "step": 445
    },
    {
      "epoch": 0.23945505184389784,
      "grad_norm": 0.8359375,
      "learning_rate": 4.972933529613135e-05,
      "loss": 0.9196,
      "step": 446
    },
    {
      "epoch": 0.23999194657897385,
      "grad_norm": 0.7578125,
      "learning_rate": 4.972809454919355e-05,
      "loss": 0.9158,
      "step": 447
    },
    {
      "epoch": 0.24052884131404986,
      "grad_norm": 0.890625,
      "learning_rate": 4.9726850980467e-05,
      "loss": 1.2073,
      "step": 448
    },
    {
      "epoch": 0.24106573604912587,
      "grad_norm": 0.7265625,
      "learning_rate": 4.972560459009361e-05,
      "loss": 1.0106,
      "step": 449
    },
    {
      "epoch": 0.24160263078420188,
      "grad_norm": 0.98828125,
      "learning_rate": 4.972435537821562e-05,
      "loss": 1.0991,
      "step": 450
    },
    {
      "epoch": 0.24160263078420188,
      "eval_loss": 0.9799502491950989,
      "eval_runtime": 46.5737,
      "eval_samples_per_second": 20.655,
      "eval_steps_per_second": 20.655,
      "step": 450
    },
    {
      "epoch": 0.24213952551927787,
      "grad_norm": 0.78515625,
      "learning_rate": 4.972310334497557e-05,
      "loss": 1.1096,
      "step": 451
    },
    {
      "epoch": 0.24267642025435388,
      "grad_norm": 0.84375,
      "learning_rate": 4.972184849051633e-05,
      "loss": 0.9439,
      "step": 452
    },
    {
      "epoch": 0.24321331498942989,
      "grad_norm": 0.69921875,
      "learning_rate": 4.972059081498111e-05,
      "loss": 1.169,
      "step": 453
    },
    {
      "epoch": 0.2437502097245059,
      "grad_norm": 0.671875,
      "learning_rate": 4.971933031851341e-05,
      "loss": 0.8649,
      "step": 454
    },
    {
      "epoch": 0.2442871044595819,
      "grad_norm": 0.90234375,
      "learning_rate": 4.971806700125708e-05,
      "loss": 0.9772,
      "step": 455
    },
    {
      "epoch": 0.2448239991946579,
      "grad_norm": 1.0390625,
      "learning_rate": 4.971680086335627e-05,
      "loss": 1.1061,
      "step": 456
    },
    {
      "epoch": 0.2453608939297339,
      "grad_norm": 0.68359375,
      "learning_rate": 4.971553190495547e-05,
      "loss": 0.8425,
      "step": 457
    },
    {
      "epoch": 0.2458977886648099,
      "grad_norm": 0.83203125,
      "learning_rate": 4.971426012619949e-05,
      "loss": 1.1449,
      "step": 458
    },
    {
      "epoch": 0.24643468339988592,
      "grad_norm": 0.8671875,
      "learning_rate": 4.9712985527233444e-05,
      "loss": 1.1962,
      "step": 459
    },
    {
      "epoch": 0.24697157813496193,
      "grad_norm": 0.83984375,
      "learning_rate": 4.971170810820279e-05,
      "loss": 1.0111,
      "step": 460
    },
    {
      "epoch": 0.2475084728700379,
      "grad_norm": 0.76953125,
      "learning_rate": 4.971042786925328e-05,
      "loss": 1.0539,
      "step": 461
    },
    {
      "epoch": 0.24804536760511392,
      "grad_norm": 0.73828125,
      "learning_rate": 4.9709144810531026e-05,
      "loss": 0.9385,
      "step": 462
    },
    {
      "epoch": 0.24858226234018993,
      "grad_norm": 0.83984375,
      "learning_rate": 4.9707858932182436e-05,
      "loss": 1.1144,
      "step": 463
    },
    {
      "epoch": 0.24911915707526594,
      "grad_norm": 0.6796875,
      "learning_rate": 4.970657023435424e-05,
      "loss": 0.8469,
      "step": 464
    },
    {
      "epoch": 0.24965605181034192,
      "grad_norm": 0.71875,
      "learning_rate": 4.9705278717193496e-05,
      "loss": 0.9331,
      "step": 465
    },
    {
      "epoch": 0.25019294654541796,
      "grad_norm": 0.8359375,
      "learning_rate": 4.970398438084758e-05,
      "loss": 0.9805,
      "step": 466
    },
    {
      "epoch": 0.25072984128049397,
      "grad_norm": 0.72265625,
      "learning_rate": 4.97026872254642e-05,
      "loss": 1.0235,
      "step": 467
    },
    {
      "epoch": 0.2512667360155699,
      "grad_norm": 0.78515625,
      "learning_rate": 4.9701387251191364e-05,
      "loss": 1.0945,
      "step": 468
    },
    {
      "epoch": 0.25180363075064593,
      "grad_norm": 0.578125,
      "learning_rate": 4.970008445817744e-05,
      "loss": 0.7344,
      "step": 469
    },
    {
      "epoch": 0.25234052548572194,
      "grad_norm": 0.78125,
      "learning_rate": 4.969877884657107e-05,
      "loss": 0.921,
      "step": 470
    },
    {
      "epoch": 0.25287742022079795,
      "grad_norm": 0.8203125,
      "learning_rate": 4.969747041652125e-05,
      "loss": 0.815,
      "step": 471
    },
    {
      "epoch": 0.25341431495587396,
      "grad_norm": 0.73046875,
      "learning_rate": 4.969615916817728e-05,
      "loss": 1.0928,
      "step": 472
    },
    {
      "epoch": 0.25395120969095,
      "grad_norm": 0.765625,
      "learning_rate": 4.9694845101688806e-05,
      "loss": 0.9634,
      "step": 473
    },
    {
      "epoch": 0.254488104426026,
      "grad_norm": 0.90625,
      "learning_rate": 4.969352821720577e-05,
      "loss": 1.0081,
      "step": 474
    },
    {
      "epoch": 0.255024999161102,
      "grad_norm": 0.765625,
      "learning_rate": 4.9692208514878444e-05,
      "loss": 0.9553,
      "step": 475
    },
    {
      "epoch": 0.255561893896178,
      "grad_norm": 0.765625,
      "learning_rate": 4.969088599485743e-05,
      "loss": 1.0275,
      "step": 476
    },
    {
      "epoch": 0.25609878863125396,
      "grad_norm": 0.671875,
      "learning_rate": 4.9689560657293633e-05,
      "loss": 0.9847,
      "step": 477
    },
    {
      "epoch": 0.25663568336632997,
      "grad_norm": 0.8671875,
      "learning_rate": 4.96882325023383e-05,
      "loss": 1.0905,
      "step": 478
    },
    {
      "epoch": 0.257172578101406,
      "grad_norm": 0.78515625,
      "learning_rate": 4.968690153014298e-05,
      "loss": 0.9663,
      "step": 479
    },
    {
      "epoch": 0.257709472836482,
      "grad_norm": 0.7890625,
      "learning_rate": 4.968556774085957e-05,
      "loss": 1.132,
      "step": 480
    },
    {
      "epoch": 0.258246367571558,
      "grad_norm": 0.6796875,
      "learning_rate": 4.9684231134640264e-05,
      "loss": 0.9002,
      "step": 481
    },
    {
      "epoch": 0.258783262306634,
      "grad_norm": 0.71484375,
      "learning_rate": 4.968289171163758e-05,
      "loss": 1.0553,
      "step": 482
    },
    {
      "epoch": 0.25932015704171,
      "grad_norm": 0.87109375,
      "learning_rate": 4.9681549472004376e-05,
      "loss": 1.1524,
      "step": 483
    },
    {
      "epoch": 0.25985705177678603,
      "grad_norm": 0.828125,
      "learning_rate": 4.9680204415893804e-05,
      "loss": 1.0497,
      "step": 484
    },
    {
      "epoch": 0.26039394651186204,
      "grad_norm": 0.6328125,
      "learning_rate": 4.967885654345936e-05,
      "loss": 0.7001,
      "step": 485
    },
    {
      "epoch": 0.260930841246938,
      "grad_norm": 0.84375,
      "learning_rate": 4.967750585485484e-05,
      "loss": 1.1598,
      "step": 486
    },
    {
      "epoch": 0.261467735982014,
      "grad_norm": 0.86328125,
      "learning_rate": 4.967615235023441e-05,
      "loss": 1.0899,
      "step": 487
    },
    {
      "epoch": 0.26200463071709,
      "grad_norm": 0.80859375,
      "learning_rate": 4.967479602975248e-05,
      "loss": 1.1266,
      "step": 488
    },
    {
      "epoch": 0.262541525452166,
      "grad_norm": 0.7578125,
      "learning_rate": 4.967343689356385e-05,
      "loss": 1.0635,
      "step": 489
    },
    {
      "epoch": 0.26307842018724203,
      "grad_norm": 0.828125,
      "learning_rate": 4.967207494182361e-05,
      "loss": 1.083,
      "step": 490
    },
    {
      "epoch": 0.26361531492231804,
      "grad_norm": 0.72265625,
      "learning_rate": 4.967071017468716e-05,
      "loss": 1.0513,
      "step": 491
    },
    {
      "epoch": 0.26415220965739405,
      "grad_norm": 0.79296875,
      "learning_rate": 4.966934259231026e-05,
      "loss": 0.9036,
      "step": 492
    },
    {
      "epoch": 0.26468910439247006,
      "grad_norm": 0.6640625,
      "learning_rate": 4.9667972194848955e-05,
      "loss": 1.0816,
      "step": 493
    },
    {
      "epoch": 0.2652259991275461,
      "grad_norm": 1.0078125,
      "learning_rate": 4.9666598982459635e-05,
      "loss": 1.1239,
      "step": 494
    },
    {
      "epoch": 0.2657628938626221,
      "grad_norm": 0.85546875,
      "learning_rate": 4.9665222955298984e-05,
      "loss": 1.0615,
      "step": 495
    },
    {
      "epoch": 0.26629978859769804,
      "grad_norm": 0.75390625,
      "learning_rate": 4.9663844113524035e-05,
      "loss": 0.833,
      "step": 496
    },
    {
      "epoch": 0.26683668333277405,
      "grad_norm": 0.76171875,
      "learning_rate": 4.966246245729213e-05,
      "loss": 0.8344,
      "step": 497
    },
    {
      "epoch": 0.26737357806785006,
      "grad_norm": 0.74609375,
      "learning_rate": 4.966107798676095e-05,
      "loss": 1.17,
      "step": 498
    },
    {
      "epoch": 0.26791047280292607,
      "grad_norm": 0.78125,
      "learning_rate": 4.9659690702088447e-05,
      "loss": 0.8675,
      "step": 499
    },
    {
      "epoch": 0.2684473675380021,
      "grad_norm": 0.859375,
      "learning_rate": 4.965830060343295e-05,
      "loss": 0.9243,
      "step": 500
    },
    {
      "epoch": 0.2684473675380021,
      "eval_loss": 0.9750539660453796,
      "eval_runtime": 46.2834,
      "eval_samples_per_second": 20.785,
      "eval_steps_per_second": 20.785,
      "step": 500
    },
    {
      "epoch": 0.2689842622730781,
      "grad_norm": 0.74609375,
      "learning_rate": 4.965690769095308e-05,
      "loss": 0.8827,
      "step": 501
    },
    {
      "epoch": 0.2695211570081541,
      "grad_norm": 0.80859375,
      "learning_rate": 4.9655511964807785e-05,
      "loss": 1.1728,
      "step": 502
    },
    {
      "epoch": 0.2700580517432301,
      "grad_norm": 0.84375,
      "learning_rate": 4.965411342515635e-05,
      "loss": 1.0448,
      "step": 503
    },
    {
      "epoch": 0.2705949464783061,
      "grad_norm": 0.74609375,
      "learning_rate": 4.965271207215835e-05,
      "loss": 0.959,
      "step": 504
    },
    {
      "epoch": 0.2711318412133821,
      "grad_norm": 0.7265625,
      "learning_rate": 4.965130790597369e-05,
      "loss": 1.2904,
      "step": 505
    },
    {
      "epoch": 0.2716687359484581,
      "grad_norm": 0.8515625,
      "learning_rate": 4.964990092676263e-05,
      "loss": 0.9568,
      "step": 506
    },
    {
      "epoch": 0.2722056306835341,
      "grad_norm": 0.74609375,
      "learning_rate": 4.964849113468569e-05,
      "loss": 1.0171,
      "step": 507
    },
    {
      "epoch": 0.2727425254186101,
      "grad_norm": 1.125,
      "learning_rate": 4.964707852990378e-05,
      "loss": 1.2031,
      "step": 508
    },
    {
      "epoch": 0.2732794201536861,
      "grad_norm": 0.94140625,
      "learning_rate": 4.9645663112578075e-05,
      "loss": 0.9805,
      "step": 509
    },
    {
      "epoch": 0.2738163148887621,
      "grad_norm": 0.82421875,
      "learning_rate": 4.964424488287009e-05,
      "loss": 1.0737,
      "step": 510
    },
    {
      "epoch": 0.27435320962383813,
      "grad_norm": 0.64453125,
      "learning_rate": 4.9642823840941684e-05,
      "loss": 1.0363,
      "step": 511
    },
    {
      "epoch": 0.27489010435891414,
      "grad_norm": 0.92578125,
      "learning_rate": 4.9641399986955e-05,
      "loss": 1.2236,
      "step": 512
    },
    {
      "epoch": 0.27542699909399015,
      "grad_norm": 0.64453125,
      "learning_rate": 4.9639973321072506e-05,
      "loss": 0.8331,
      "step": 513
    },
    {
      "epoch": 0.27596389382906616,
      "grad_norm": 0.72265625,
      "learning_rate": 4.963854384345702e-05,
      "loss": 0.9885,
      "step": 514
    },
    {
      "epoch": 0.2765007885641421,
      "grad_norm": 0.94140625,
      "learning_rate": 4.9637111554271665e-05,
      "loss": 0.9867,
      "step": 515
    },
    {
      "epoch": 0.27703768329921813,
      "grad_norm": 0.71484375,
      "learning_rate": 4.963567645367988e-05,
      "loss": 0.974,
      "step": 516
    },
    {
      "epoch": 0.27757457803429414,
      "grad_norm": 0.7421875,
      "learning_rate": 4.9634238541845415e-05,
      "loss": 1.044,
      "step": 517
    },
    {
      "epoch": 0.27811147276937015,
      "grad_norm": 1.0,
      "learning_rate": 4.9632797818932374e-05,
      "loss": 1.099,
      "step": 518
    },
    {
      "epoch": 0.27864836750444616,
      "grad_norm": 0.859375,
      "learning_rate": 4.9631354285105145e-05,
      "loss": 1.0237,
      "step": 519
    },
    {
      "epoch": 0.27918526223952217,
      "grad_norm": 0.84765625,
      "learning_rate": 4.962990794052847e-05,
      "loss": 1.0559,
      "step": 520
    },
    {
      "epoch": 0.2797221569745982,
      "grad_norm": 0.74609375,
      "learning_rate": 4.962845878536738e-05,
      "loss": 0.9844,
      "step": 521
    },
    {
      "epoch": 0.2802590517096742,
      "grad_norm": 0.71484375,
      "learning_rate": 4.962700681978725e-05,
      "loss": 0.931,
      "step": 522
    },
    {
      "epoch": 0.2807959464447502,
      "grad_norm": 0.81640625,
      "learning_rate": 4.962555204395377e-05,
      "loss": 0.9192,
      "step": 523
    },
    {
      "epoch": 0.28133284117982615,
      "grad_norm": 0.71484375,
      "learning_rate": 4.9624094458032946e-05,
      "loss": 1.1457,
      "step": 524
    },
    {
      "epoch": 0.28186973591490216,
      "grad_norm": 0.83984375,
      "learning_rate": 4.96226340621911e-05,
      "loss": 1.0615,
      "step": 525
    },
    {
      "epoch": 0.2824066306499782,
      "grad_norm": 0.72265625,
      "learning_rate": 4.962117085659489e-05,
      "loss": 0.9441,
      "step": 526
    },
    {
      "epoch": 0.2829435253850542,
      "grad_norm": 0.77734375,
      "learning_rate": 4.961970484141128e-05,
      "loss": 1.1009,
      "step": 527
    },
    {
      "epoch": 0.2834804201201302,
      "grad_norm": 0.671875,
      "learning_rate": 4.9618236016807564e-05,
      "loss": 0.9566,
      "step": 528
    },
    {
      "epoch": 0.2840173148552062,
      "grad_norm": 0.8203125,
      "learning_rate": 4.961676438295135e-05,
      "loss": 0.8927,
      "step": 529
    },
    {
      "epoch": 0.2845542095902822,
      "grad_norm": 0.8203125,
      "learning_rate": 4.9615289940010584e-05,
      "loss": 1.1116,
      "step": 530
    },
    {
      "epoch": 0.2850911043253582,
      "grad_norm": 0.671875,
      "learning_rate": 4.96138126881535e-05,
      "loss": 1.004,
      "step": 531
    },
    {
      "epoch": 0.28562799906043423,
      "grad_norm": 0.7578125,
      "learning_rate": 4.9612332627548686e-05,
      "loss": 1.0446,
      "step": 532
    },
    {
      "epoch": 0.28616489379551024,
      "grad_norm": 0.71484375,
      "learning_rate": 4.961084975836503e-05,
      "loss": 0.9992,
      "step": 533
    },
    {
      "epoch": 0.2867017885305862,
      "grad_norm": 0.64453125,
      "learning_rate": 4.9609364080771735e-05,
      "loss": 0.8808,
      "step": 534
    },
    {
      "epoch": 0.2872386832656622,
      "grad_norm": 0.84375,
      "learning_rate": 4.960787559493836e-05,
      "loss": 1.1595,
      "step": 535
    },
    {
      "epoch": 0.2877755780007382,
      "grad_norm": 0.890625,
      "learning_rate": 4.960638430103473e-05,
      "loss": 1.056,
      "step": 536
    },
    {
      "epoch": 0.28831247273581423,
      "grad_norm": 0.74609375,
      "learning_rate": 4.960489019923105e-05,
      "loss": 0.9335,
      "step": 537
    },
    {
      "epoch": 0.28884936747089024,
      "grad_norm": 0.63671875,
      "learning_rate": 4.96033932896978e-05,
      "loss": 0.8555,
      "step": 538
    },
    {
      "epoch": 0.28938626220596625,
      "grad_norm": 1.1328125,
      "learning_rate": 4.9601893572605796e-05,
      "loss": 1.1849,
      "step": 539
    },
    {
      "epoch": 0.28992315694104226,
      "grad_norm": 0.76171875,
      "learning_rate": 4.960039104812618e-05,
      "loss": 1.1725,
      "step": 540
    },
    {
      "epoch": 0.29046005167611827,
      "grad_norm": 0.77734375,
      "learning_rate": 4.9598885716430404e-05,
      "loss": 0.8767,
      "step": 541
    },
    {
      "epoch": 0.2909969464111943,
      "grad_norm": 0.84765625,
      "learning_rate": 4.959737757769025e-05,
      "loss": 1.11,
      "step": 542
    },
    {
      "epoch": 0.29153384114627023,
      "grad_norm": 0.82421875,
      "learning_rate": 4.9595866632077816e-05,
      "loss": 1.0985,
      "step": 543
    },
    {
      "epoch": 0.29207073588134624,
      "grad_norm": 0.96875,
      "learning_rate": 4.959435287976551e-05,
      "loss": 1.2694,
      "step": 544
    },
    {
      "epoch": 0.29260763061642225,
      "grad_norm": 0.796875,
      "learning_rate": 4.959283632092608e-05,
      "loss": 1.0265,
      "step": 545
    },
    {
      "epoch": 0.29314452535149826,
      "grad_norm": 0.80078125,
      "learning_rate": 4.9591316955732595e-05,
      "loss": 1.0899,
      "step": 546
    },
    {
      "epoch": 0.2936814200865743,
      "grad_norm": 0.82421875,
      "learning_rate": 4.9589794784358406e-05,
      "loss": 1.1073,
      "step": 547
    },
    {
      "epoch": 0.2942183148216503,
      "grad_norm": 0.6875,
      "learning_rate": 4.9588269806977236e-05,
      "loss": 0.8764,
      "step": 548
    },
    {
      "epoch": 0.2947552095567263,
      "grad_norm": 0.6328125,
      "learning_rate": 4.958674202376309e-05,
      "loss": 1.0164,
      "step": 549
    },
    {
      "epoch": 0.2952921042918023,
      "grad_norm": 0.98828125,
      "learning_rate": 4.958521143489032e-05,
      "loss": 1.3314,
      "step": 550
    },
    {
      "epoch": 0.2952921042918023,
      "eval_loss": 0.9689125418663025,
      "eval_runtime": 46.1941,
      "eval_samples_per_second": 20.825,
      "eval_steps_per_second": 20.825,
      "step": 550
    },
    {
      "epoch": 0.2958289990268783,
      "grad_norm": 0.73046875,
      "learning_rate": 4.958367804053357e-05,
      "loss": 0.9652,
      "step": 551
    },
    {
      "epoch": 0.2963658937619543,
      "grad_norm": 0.76171875,
      "learning_rate": 4.9582141840867835e-05,
      "loss": 0.9144,
      "step": 552
    },
    {
      "epoch": 0.2969027884970303,
      "grad_norm": 0.71875,
      "learning_rate": 4.9580602836068414e-05,
      "loss": 0.7835,
      "step": 553
    },
    {
      "epoch": 0.2974396832321063,
      "grad_norm": 0.67578125,
      "learning_rate": 4.957906102631091e-05,
      "loss": 0.8645,
      "step": 554
    },
    {
      "epoch": 0.2979765779671823,
      "grad_norm": 0.70703125,
      "learning_rate": 4.957751641177128e-05,
      "loss": 0.9647,
      "step": 555
    },
    {
      "epoch": 0.2985134727022583,
      "grad_norm": 0.79296875,
      "learning_rate": 4.9575968992625775e-05,
      "loss": 1.0369,
      "step": 556
    },
    {
      "epoch": 0.2990503674373343,
      "grad_norm": 0.8984375,
      "learning_rate": 4.957441876905098e-05,
      "loss": 1.3041,
      "step": 557
    },
    {
      "epoch": 0.29958726217241033,
      "grad_norm": 0.734375,
      "learning_rate": 4.957286574122379e-05,
      "loss": 1.2297,
      "step": 558
    },
    {
      "epoch": 0.30012415690748634,
      "grad_norm": 0.77734375,
      "learning_rate": 4.957130990932144e-05,
      "loss": 1.1134,
      "step": 559
    },
    {
      "epoch": 0.30066105164256235,
      "grad_norm": 0.703125,
      "learning_rate": 4.9569751273521454e-05,
      "loss": 0.9296,
      "step": 560
    },
    {
      "epoch": 0.30119794637763836,
      "grad_norm": 0.79296875,
      "learning_rate": 4.956818983400169e-05,
      "loss": 0.9917,
      "step": 561
    },
    {
      "epoch": 0.3017348411127143,
      "grad_norm": 0.6640625,
      "learning_rate": 4.956662559094034e-05,
      "loss": 0.9292,
      "step": 562
    },
    {
      "epoch": 0.3022717358477903,
      "grad_norm": 0.69140625,
      "learning_rate": 4.9565058544515905e-05,
      "loss": 0.8894,
      "step": 563
    },
    {
      "epoch": 0.30280863058286633,
      "grad_norm": 0.7421875,
      "learning_rate": 4.9563488694907186e-05,
      "loss": 0.9931,
      "step": 564
    },
    {
      "epoch": 0.30334552531794234,
      "grad_norm": 0.75,
      "learning_rate": 4.9561916042293345e-05,
      "loss": 1.0937,
      "step": 565
    },
    {
      "epoch": 0.30388242005301835,
      "grad_norm": 0.72265625,
      "learning_rate": 4.9560340586853825e-05,
      "loss": 0.9625,
      "step": 566
    },
    {
      "epoch": 0.30441931478809436,
      "grad_norm": 0.8046875,
      "learning_rate": 4.955876232876842e-05,
      "loss": 0.9373,
      "step": 567
    },
    {
      "epoch": 0.3049562095231704,
      "grad_norm": 0.6953125,
      "learning_rate": 4.9557181268217227e-05,
      "loss": 0.949,
      "step": 568
    },
    {
      "epoch": 0.3054931042582464,
      "grad_norm": 0.7109375,
      "learning_rate": 4.9555597405380647e-05,
      "loss": 0.9483,
      "step": 569
    },
    {
      "epoch": 0.3060299989933224,
      "grad_norm": 0.80078125,
      "learning_rate": 4.9554010740439435e-05,
      "loss": 0.9059,
      "step": 570
    },
    {
      "epoch": 0.30656689372839835,
      "grad_norm": 0.6953125,
      "learning_rate": 4.955242127357465e-05,
      "loss": 1.0488,
      "step": 571
    },
    {
      "epoch": 0.30710378846347436,
      "grad_norm": 0.7578125,
      "learning_rate": 4.955082900496766e-05,
      "loss": 0.9612,
      "step": 572
    },
    {
      "epoch": 0.30764068319855037,
      "grad_norm": 0.90234375,
      "learning_rate": 4.954923393480017e-05,
      "loss": 1.2172,
      "step": 573
    },
    {
      "epoch": 0.3081775779336264,
      "grad_norm": 0.7421875,
      "learning_rate": 4.9547636063254196e-05,
      "loss": 1.2283,
      "step": 574
    },
    {
      "epoch": 0.3087144726687024,
      "grad_norm": 0.9296875,
      "learning_rate": 4.954603539051208e-05,
      "loss": 0.9913,
      "step": 575
    },
    {
      "epoch": 0.3092513674037784,
      "grad_norm": 0.82421875,
      "learning_rate": 4.954443191675648e-05,
      "loss": 1.0052,
      "step": 576
    },
    {
      "epoch": 0.3097882621388544,
      "grad_norm": 0.62890625,
      "learning_rate": 4.954282564217036e-05,
      "loss": 0.665,
      "step": 577
    },
    {
      "epoch": 0.3103251568739304,
      "grad_norm": 0.74609375,
      "learning_rate": 4.954121656693703e-05,
      "loss": 0.9044,
      "step": 578
    },
    {
      "epoch": 0.31086205160900643,
      "grad_norm": 0.92578125,
      "learning_rate": 4.95396046912401e-05,
      "loss": 1.3292,
      "step": 579
    },
    {
      "epoch": 0.31139894634408244,
      "grad_norm": 1.0078125,
      "learning_rate": 4.9537990015263505e-05,
      "loss": 0.9049,
      "step": 580
    },
    {
      "epoch": 0.3119358410791584,
      "grad_norm": 0.6875,
      "learning_rate": 4.95363725391915e-05,
      "loss": 0.9415,
      "step": 581
    },
    {
      "epoch": 0.3124727358142344,
      "grad_norm": 0.9609375,
      "learning_rate": 4.953475226320866e-05,
      "loss": 0.9518,
      "step": 582
    },
    {
      "epoch": 0.3130096305493104,
      "grad_norm": 0.66015625,
      "learning_rate": 4.9533129187499885e-05,
      "loss": 0.8416,
      "step": 583
    },
    {
      "epoch": 0.3135465252843864,
      "grad_norm": 0.7265625,
      "learning_rate": 4.9531503312250375e-05,
      "loss": 1.0242,
      "step": 584
    },
    {
      "epoch": 0.31408342001946243,
      "grad_norm": 0.828125,
      "learning_rate": 4.9529874637645675e-05,
      "loss": 1.0651,
      "step": 585
    },
    {
      "epoch": 0.31462031475453844,
      "grad_norm": 1.046875,
      "learning_rate": 4.952824316387163e-05,
      "loss": 0.9305,
      "step": 586
    },
    {
      "epoch": 0.31515720948961445,
      "grad_norm": 0.92578125,
      "learning_rate": 4.9526608891114415e-05,
      "loss": 0.919,
      "step": 587
    },
    {
      "epoch": 0.31569410422469046,
      "grad_norm": 0.71484375,
      "learning_rate": 4.952497181956053e-05,
      "loss": 1.0465,
      "step": 588
    },
    {
      "epoch": 0.3162309989597665,
      "grad_norm": 0.69921875,
      "learning_rate": 4.952333194939677e-05,
      "loss": 0.9995,
      "step": 589
    },
    {
      "epoch": 0.31676789369484243,
      "grad_norm": 0.85546875,
      "learning_rate": 4.952168928081027e-05,
      "loss": 1.1415,
      "step": 590
    },
    {
      "epoch": 0.31730478842991844,
      "grad_norm": 0.73828125,
      "learning_rate": 4.952004381398848e-05,
      "loss": 1.0067,
      "step": 591
    },
    {
      "epoch": 0.31784168316499445,
      "grad_norm": 0.59765625,
      "learning_rate": 4.951839554911917e-05,
      "loss": 0.9907,
      "step": 592
    },
    {
      "epoch": 0.31837857790007046,
      "grad_norm": 0.8359375,
      "learning_rate": 4.951674448639043e-05,
      "loss": 1.0772,
      "step": 593
    },
    {
      "epoch": 0.31891547263514647,
      "grad_norm": 0.72265625,
      "learning_rate": 4.951509062599066e-05,
      "loss": 0.9815,
      "step": 594
    },
    {
      "epoch": 0.3194523673702225,
      "grad_norm": 0.7734375,
      "learning_rate": 4.95134339681086e-05,
      "loss": 1.0787,
      "step": 595
    },
    {
      "epoch": 0.3199892621052985,
      "grad_norm": 0.72265625,
      "learning_rate": 4.951177451293328e-05,
      "loss": 0.9766,
      "step": 596
    },
    {
      "epoch": 0.3205261568403745,
      "grad_norm": 0.56640625,
      "learning_rate": 4.951011226065407e-05,
      "loss": 0.6743,
      "step": 597
    },
    {
      "epoch": 0.3210630515754505,
      "grad_norm": 0.765625,
      "learning_rate": 4.950844721146066e-05,
      "loss": 0.8134,
      "step": 598
    },
    {
      "epoch": 0.3215999463105265,
      "grad_norm": 0.796875,
      "learning_rate": 4.9506779365543046e-05,
      "loss": 1.1082,
      "step": 599
    },
    {
      "epoch": 0.3221368410456025,
      "grad_norm": 0.7421875,
      "learning_rate": 4.950510872309155e-05,
      "loss": 1.0203,
      "step": 600
    },
    {
      "epoch": 0.3221368410456025,
      "eval_loss": 0.9650207757949829,
      "eval_runtime": 45.8281,
      "eval_samples_per_second": 20.991,
      "eval_steps_per_second": 20.991,
      "step": 600
    },
    {
      "epoch": 0.3226737357806785,
      "grad_norm": 0.84765625,
      "learning_rate": 4.950343528429683e-05,
      "loss": 1.0154,
      "step": 601
    },
    {
      "epoch": 0.3232106305157545,
      "grad_norm": 0.80078125,
      "learning_rate": 4.950175904934982e-05,
      "loss": 1.0035,
      "step": 602
    },
    {
      "epoch": 0.3237475252508305,
      "grad_norm": 0.8125,
      "learning_rate": 4.950008001844182e-05,
      "loss": 0.9076,
      "step": 603
    },
    {
      "epoch": 0.3242844199859065,
      "grad_norm": 1.671875,
      "learning_rate": 4.949839819176442e-05,
      "loss": 1.0813,
      "step": 604
    },
    {
      "epoch": 0.3248213147209825,
      "grad_norm": 0.703125,
      "learning_rate": 4.949671356950953e-05,
      "loss": 0.8759,
      "step": 605
    },
    {
      "epoch": 0.32535820945605853,
      "grad_norm": 0.859375,
      "learning_rate": 4.949502615186941e-05,
      "loss": 0.9611,
      "step": 606
    },
    {
      "epoch": 0.32589510419113454,
      "grad_norm": 0.79296875,
      "learning_rate": 4.94933359390366e-05,
      "loss": 0.8841,
      "step": 607
    },
    {
      "epoch": 0.32643199892621055,
      "grad_norm": 0.8359375,
      "learning_rate": 4.949164293120397e-05,
      "loss": 1.0533,
      "step": 608
    },
    {
      "epoch": 0.3269688936612865,
      "grad_norm": 0.69921875,
      "learning_rate": 4.948994712856472e-05,
      "loss": 1.0198,
      "step": 609
    },
    {
      "epoch": 0.3275057883963625,
      "grad_norm": 0.74609375,
      "learning_rate": 4.948824853131236e-05,
      "loss": 0.9614,
      "step": 610
    },
    {
      "epoch": 0.32804268313143853,
      "grad_norm": 0.75,
      "learning_rate": 4.948654713964073e-05,
      "loss": 1.0106,
      "step": 611
    },
    {
      "epoch": 0.32857957786651454,
      "grad_norm": 0.875,
      "learning_rate": 4.948484295374397e-05,
      "loss": 1.1489,
      "step": 612
    },
    {
      "epoch": 0.32911647260159055,
      "grad_norm": 0.75,
      "learning_rate": 4.9483135973816564e-05,
      "loss": 0.9575,
      "step": 613
    },
    {
      "epoch": 0.32965336733666656,
      "grad_norm": 0.6640625,
      "learning_rate": 4.948142620005328e-05,
      "loss": 0.9872,
      "step": 614
    },
    {
      "epoch": 0.33019026207174257,
      "grad_norm": 0.6796875,
      "learning_rate": 4.9479713632649236e-05,
      "loss": 0.8048,
      "step": 615
    },
    {
      "epoch": 0.3307271568068186,
      "grad_norm": 0.75390625,
      "learning_rate": 4.947799827179986e-05,
      "loss": 0.8788,
      "step": 616
    },
    {
      "epoch": 0.3312640515418946,
      "grad_norm": 0.6875,
      "learning_rate": 4.9476280117700884e-05,
      "loss": 0.9128,
      "step": 617
    },
    {
      "epoch": 0.3318009462769706,
      "grad_norm": 0.69140625,
      "learning_rate": 4.9474559170548387e-05,
      "loss": 0.9896,
      "step": 618
    },
    {
      "epoch": 0.33233784101204655,
      "grad_norm": 0.7421875,
      "learning_rate": 4.947283543053874e-05,
      "loss": 1.0966,
      "step": 619
    },
    {
      "epoch": 0.33287473574712256,
      "grad_norm": 0.71484375,
      "learning_rate": 4.947110889786864e-05,
      "loss": 1.0515,
      "step": 620
    },
    {
      "epoch": 0.3334116304821986,
      "grad_norm": 0.75,
      "learning_rate": 4.946937957273513e-05,
      "loss": 0.9433,
      "step": 621
    },
    {
      "epoch": 0.3339485252172746,
      "grad_norm": 1.3359375,
      "learning_rate": 4.946764745533552e-05,
      "loss": 0.9242,
      "step": 622
    },
    {
      "epoch": 0.3344854199523506,
      "grad_norm": 0.65625,
      "learning_rate": 4.946591254586748e-05,
      "loss": 0.9455,
      "step": 623
    },
    {
      "epoch": 0.3350223146874266,
      "grad_norm": 0.80859375,
      "learning_rate": 4.9464174844528984e-05,
      "loss": 0.8518,
      "step": 624
    },
    {
      "epoch": 0.3355592094225026,
      "grad_norm": 0.859375,
      "learning_rate": 4.946243435151832e-05,
      "loss": 0.9709,
      "step": 625
    },
    {
      "epoch": 0.3360961041575786,
      "grad_norm": 0.7890625,
      "learning_rate": 4.946069106703411e-05,
      "loss": 0.9883,
      "step": 626
    },
    {
      "epoch": 0.33663299889265463,
      "grad_norm": 0.76953125,
      "learning_rate": 4.945894499127528e-05,
      "loss": 1.1203,
      "step": 627
    },
    {
      "epoch": 0.3371698936277306,
      "grad_norm": 0.80859375,
      "learning_rate": 4.9457196124441073e-05,
      "loss": 1.0605,
      "step": 628
    },
    {
      "epoch": 0.3377067883628066,
      "grad_norm": 0.69140625,
      "learning_rate": 4.9455444466731066e-05,
      "loss": 1.022,
      "step": 629
    },
    {
      "epoch": 0.3382436830978826,
      "grad_norm": 0.75390625,
      "learning_rate": 4.9453690018345144e-05,
      "loss": 0.8821,
      "step": 630
    },
    {
      "epoch": 0.3387805778329586,
      "grad_norm": 0.63671875,
      "learning_rate": 4.9451932779483514e-05,
      "loss": 0.6977,
      "step": 631
    },
    {
      "epoch": 0.33931747256803463,
      "grad_norm": 2.40625,
      "learning_rate": 4.9450172750346684e-05,
      "loss": 0.965,
      "step": 632
    },
    {
      "epoch": 0.33985436730311064,
      "grad_norm": 0.90625,
      "learning_rate": 4.944840993113551e-05,
      "loss": 1.0392,
      "step": 633
    },
    {
      "epoch": 0.34039126203818665,
      "grad_norm": 0.7421875,
      "learning_rate": 4.944664432205115e-05,
      "loss": 0.8958,
      "step": 634
    },
    {
      "epoch": 0.34092815677326266,
      "grad_norm": 0.76953125,
      "learning_rate": 4.944487592329509e-05,
      "loss": 0.9605,
      "step": 635
    },
    {
      "epoch": 0.34146505150833867,
      "grad_norm": 0.8046875,
      "learning_rate": 4.944310473506911e-05,
      "loss": 0.9886,
      "step": 636
    },
    {
      "epoch": 0.3420019462434146,
      "grad_norm": 0.734375,
      "learning_rate": 4.944133075757533e-05,
      "loss": 1.1848,
      "step": 637
    },
    {
      "epoch": 0.34253884097849063,
      "grad_norm": 0.82421875,
      "learning_rate": 4.9439553991016187e-05,
      "loss": 0.8984,
      "step": 638
    },
    {
      "epoch": 0.34307573571356664,
      "grad_norm": 0.82421875,
      "learning_rate": 4.943777443559444e-05,
      "loss": 0.9367,
      "step": 639
    },
    {
      "epoch": 0.34361263044864265,
      "grad_norm": 0.92578125,
      "learning_rate": 4.943599209151314e-05,
      "loss": 1.0085,
      "step": 640
    },
    {
      "epoch": 0.34414952518371866,
      "grad_norm": 0.921875,
      "learning_rate": 4.9434206958975695e-05,
      "loss": 1.0865,
      "step": 641
    },
    {
      "epoch": 0.3446864199187947,
      "grad_norm": 0.82421875,
      "learning_rate": 4.9432419038185794e-05,
      "loss": 1.0724,
      "step": 642
    },
    {
      "epoch": 0.3452233146538707,
      "grad_norm": 0.8828125,
      "learning_rate": 4.9430628329347475e-05,
      "loss": 1.0542,
      "step": 643
    },
    {
      "epoch": 0.3457602093889467,
      "grad_norm": 0.66796875,
      "learning_rate": 4.942883483266507e-05,
      "loss": 1.0099,
      "step": 644
    },
    {
      "epoch": 0.3462971041240227,
      "grad_norm": 0.72265625,
      "learning_rate": 4.942703854834324e-05,
      "loss": 0.9407,
      "step": 645
    },
    {
      "epoch": 0.3468339988590987,
      "grad_norm": 0.6875,
      "learning_rate": 4.942523947658698e-05,
      "loss": 0.8455,
      "step": 646
    },
    {
      "epoch": 0.34737089359417467,
      "grad_norm": 0.8515625,
      "learning_rate": 4.9423437617601564e-05,
      "loss": 1.0387,
      "step": 647
    },
    {
      "epoch": 0.3479077883292507,
      "grad_norm": 0.88671875,
      "learning_rate": 4.942163297159263e-05,
      "loss": 0.8568,
      "step": 648
    },
    {
      "epoch": 0.3484446830643267,
      "grad_norm": 0.71875,
      "learning_rate": 4.941982553876609e-05,
      "loss": 1.0762,
      "step": 649
    },
    {
      "epoch": 0.3489815777994027,
      "grad_norm": 0.76171875,
      "learning_rate": 4.9418015319328204e-05,
      "loss": 0.9849,
      "step": 650
    },
    {
      "epoch": 0.3489815777994027,
      "eval_loss": 0.9629884362220764,
      "eval_runtime": 48.9415,
      "eval_samples_per_second": 19.656,
      "eval_steps_per_second": 19.656,
      "step": 650
    },
    {
      "epoch": 0.3495184725344787,
      "grad_norm": 0.6875,
      "learning_rate": 4.941620231348555e-05,
      "loss": 1.0022,
      "step": 651
    },
    {
      "epoch": 0.3500553672695547,
      "grad_norm": 0.67578125,
      "learning_rate": 4.9414386521445e-05,
      "loss": 0.8935,
      "step": 652
    },
    {
      "epoch": 0.35059226200463073,
      "grad_norm": 0.67578125,
      "learning_rate": 4.9412567943413755e-05,
      "loss": 1.0854,
      "step": 653
    },
    {
      "epoch": 0.35112915673970674,
      "grad_norm": 0.7421875,
      "learning_rate": 4.941074657959937e-05,
      "loss": 1.2573,
      "step": 654
    },
    {
      "epoch": 0.35166605147478275,
      "grad_norm": 0.78515625,
      "learning_rate": 4.940892243020964e-05,
      "loss": 0.7695,
      "step": 655
    },
    {
      "epoch": 0.3522029462098587,
      "grad_norm": 0.6796875,
      "learning_rate": 4.940709549545276e-05,
      "loss": 0.9791,
      "step": 656
    },
    {
      "epoch": 0.3527398409449347,
      "grad_norm": 0.82421875,
      "learning_rate": 4.940526577553719e-05,
      "loss": 0.8833,
      "step": 657
    },
    {
      "epoch": 0.3532767356800107,
      "grad_norm": 0.72265625,
      "learning_rate": 4.940343327067172e-05,
      "loss": 0.9296,
      "step": 658
    },
    {
      "epoch": 0.35381363041508673,
      "grad_norm": 0.62109375,
      "learning_rate": 4.9401597981065486e-05,
      "loss": 0.8494,
      "step": 659
    },
    {
      "epoch": 0.35435052515016274,
      "grad_norm": 0.82421875,
      "learning_rate": 4.939975990692789e-05,
      "loss": 0.9775,
      "step": 660
    },
    {
      "epoch": 0.35488741988523875,
      "grad_norm": 0.67578125,
      "learning_rate": 4.939791904846869e-05,
      "loss": 0.8377,
      "step": 661
    },
    {
      "epoch": 0.35542431462031476,
      "grad_norm": 0.640625,
      "learning_rate": 4.939607540589795e-05,
      "loss": 0.8226,
      "step": 662
    },
    {
      "epoch": 0.3559612093553908,
      "grad_norm": 0.7890625,
      "learning_rate": 4.939422897942606e-05,
      "loss": 1.0506,
      "step": 663
    },
    {
      "epoch": 0.3564981040904668,
      "grad_norm": 0.79296875,
      "learning_rate": 4.9392379769263716e-05,
      "loss": 0.927,
      "step": 664
    },
    {
      "epoch": 0.3570349988255428,
      "grad_norm": 0.734375,
      "learning_rate": 4.939052777562193e-05,
      "loss": 0.9213,
      "step": 665
    },
    {
      "epoch": 0.35757189356061875,
      "grad_norm": 0.6171875,
      "learning_rate": 4.9388672998712046e-05,
      "loss": 0.7729,
      "step": 666
    },
    {
      "epoch": 0.35810878829569476,
      "grad_norm": 0.73046875,
      "learning_rate": 4.938681543874572e-05,
      "loss": 1.0298,
      "step": 667
    },
    {
      "epoch": 0.35864568303077077,
      "grad_norm": 0.60546875,
      "learning_rate": 4.938495509593492e-05,
      "loss": 0.8244,
      "step": 668
    },
    {
      "epoch": 0.3591825777658468,
      "grad_norm": 0.7734375,
      "learning_rate": 4.938309197049192e-05,
      "loss": 0.9015,
      "step": 669
    },
    {
      "epoch": 0.3597194725009228,
      "grad_norm": 0.8046875,
      "learning_rate": 4.938122606262936e-05,
      "loss": 0.9868,
      "step": 670
    },
    {
      "epoch": 0.3602563672359988,
      "grad_norm": 0.6796875,
      "learning_rate": 4.937935737256012e-05,
      "loss": 0.9103,
      "step": 671
    },
    {
      "epoch": 0.3607932619710748,
      "grad_norm": 0.7734375,
      "learning_rate": 4.9377485900497476e-05,
      "loss": 0.9998,
      "step": 672
    },
    {
      "epoch": 0.3613301567061508,
      "grad_norm": 0.72265625,
      "learning_rate": 4.9375611646654976e-05,
      "loss": 0.8885,
      "step": 673
    },
    {
      "epoch": 0.36186705144122683,
      "grad_norm": 0.77734375,
      "learning_rate": 4.937373461124649e-05,
      "loss": 0.9337,
      "step": 674
    },
    {
      "epoch": 0.3624039461763028,
      "grad_norm": 0.75,
      "learning_rate": 4.937185479448623e-05,
      "loss": 1.016,
      "step": 675
    },
    {
      "epoch": 0.3629408409113788,
      "grad_norm": 0.75,
      "learning_rate": 4.9369972196588676e-05,
      "loss": 1.1903,
      "step": 676
    },
    {
      "epoch": 0.3634777356464548,
      "grad_norm": 1.0546875,
      "learning_rate": 4.9368086817768685e-05,
      "loss": 1.165,
      "step": 677
    },
    {
      "epoch": 0.3640146303815308,
      "grad_norm": 0.921875,
      "learning_rate": 4.936619865824138e-05,
      "loss": 1.0212,
      "step": 678
    },
    {
      "epoch": 0.3645515251166068,
      "grad_norm": 0.796875,
      "learning_rate": 4.9364307718222245e-05,
      "loss": 0.9909,
      "step": 679
    },
    {
      "epoch": 0.36508841985168283,
      "grad_norm": 0.64453125,
      "learning_rate": 4.936241399792705e-05,
      "loss": 1.0334,
      "step": 680
    },
    {
      "epoch": 0.36562531458675884,
      "grad_norm": 0.828125,
      "learning_rate": 4.936051749757189e-05,
      "loss": 1.1268,
      "step": 681
    },
    {
      "epoch": 0.36616220932183485,
      "grad_norm": 0.67578125,
      "learning_rate": 4.935861821737318e-05,
      "loss": 0.9732,
      "step": 682
    },
    {
      "epoch": 0.36669910405691086,
      "grad_norm": 0.79296875,
      "learning_rate": 4.935671615754765e-05,
      "loss": 1.0614,
      "step": 683
    },
    {
      "epoch": 0.3672359987919869,
      "grad_norm": 0.83203125,
      "learning_rate": 4.9354811318312367e-05,
      "loss": 1.142,
      "step": 684
    },
    {
      "epoch": 0.3677728935270628,
      "grad_norm": 0.73828125,
      "learning_rate": 4.935290369988468e-05,
      "loss": 1.1117,
      "step": 685
    },
    {
      "epoch": 0.36830978826213884,
      "grad_norm": 0.64453125,
      "learning_rate": 4.935099330248227e-05,
      "loss": 0.9285,
      "step": 686
    },
    {
      "epoch": 0.36884668299721485,
      "grad_norm": 0.92578125,
      "learning_rate": 4.934908012632315e-05,
      "loss": 1.1574,
      "step": 687
    },
    {
      "epoch": 0.36938357773229086,
      "grad_norm": 0.75,
      "learning_rate": 4.934716417162563e-05,
      "loss": 0.979,
      "step": 688
    },
    {
      "epoch": 0.36992047246736687,
      "grad_norm": 0.7421875,
      "learning_rate": 4.9345245438608344e-05,
      "loss": 1.1321,
      "step": 689
    },
    {
      "epoch": 0.3704573672024429,
      "grad_norm": 0.8515625,
      "learning_rate": 4.934332392749025e-05,
      "loss": 0.9974,
      "step": 690
    },
    {
      "epoch": 0.3709942619375189,
      "grad_norm": 0.81640625,
      "learning_rate": 4.93413996384906e-05,
      "loss": 1.122,
      "step": 691
    },
    {
      "epoch": 0.3715311566725949,
      "grad_norm": 0.80078125,
      "learning_rate": 4.933947257182901e-05,
      "loss": 1.0571,
      "step": 692
    },
    {
      "epoch": 0.3720680514076709,
      "grad_norm": 0.71875,
      "learning_rate": 4.9337542727725356e-05,
      "loss": 0.9182,
      "step": 693
    },
    {
      "epoch": 0.37260494614274686,
      "grad_norm": 0.82421875,
      "learning_rate": 4.9335610106399864e-05,
      "loss": 1.0838,
      "step": 694
    },
    {
      "epoch": 0.3731418408778229,
      "grad_norm": 1.2265625,
      "learning_rate": 4.933367470807308e-05,
      "loss": 1.2538,
      "step": 695
    },
    {
      "epoch": 0.3736787356128989,
      "grad_norm": 0.76953125,
      "learning_rate": 4.933173653296585e-05,
      "loss": 0.9813,
      "step": 696
    },
    {
      "epoch": 0.3742156303479749,
      "grad_norm": 0.74609375,
      "learning_rate": 4.932979558129934e-05,
      "loss": 0.946,
      "step": 697
    },
    {
      "epoch": 0.3747525250830509,
      "grad_norm": 0.75,
      "learning_rate": 4.932785185329505e-05,
      "loss": 0.8854,
      "step": 698
    },
    {
      "epoch": 0.3752894198181269,
      "grad_norm": 0.75,
      "learning_rate": 4.9325905349174765e-05,
      "loss": 1.1353,
      "step": 699
    },
    {
      "epoch": 0.3758263145532029,
      "grad_norm": 0.81640625,
      "learning_rate": 4.932395606916062e-05,
      "loss": 1.1588,
      "step": 700
    },
    {
      "epoch": 0.3758263145532029,
      "eval_loss": 0.9581480026245117,
      "eval_runtime": 46.2527,
      "eval_samples_per_second": 20.799,
      "eval_steps_per_second": 20.799,
      "step": 700
    },
    {
      "epoch": 0.37636320928827893,
      "grad_norm": 0.734375,
      "learning_rate": 4.932200401347506e-05,
      "loss": 0.8196,
      "step": 701
    },
    {
      "epoch": 0.37690010402335494,
      "grad_norm": 0.87109375,
      "learning_rate": 4.932004918234082e-05,
      "loss": 1.1573,
      "step": 702
    },
    {
      "epoch": 0.37743699875843095,
      "grad_norm": 0.77734375,
      "learning_rate": 4.931809157598099e-05,
      "loss": 0.8634,
      "step": 703
    },
    {
      "epoch": 0.3779738934935069,
      "grad_norm": 0.69140625,
      "learning_rate": 4.931613119461893e-05,
      "loss": 0.8944,
      "step": 704
    },
    {
      "epoch": 0.3785107882285829,
      "grad_norm": 0.78515625,
      "learning_rate": 4.931416803847837e-05,
      "loss": 1.0816,
      "step": 705
    },
    {
      "epoch": 0.3790476829636589,
      "grad_norm": 0.99609375,
      "learning_rate": 4.931220210778332e-05,
      "loss": 1.1423,
      "step": 706
    },
    {
      "epoch": 0.37958457769873494,
      "grad_norm": 0.80078125,
      "learning_rate": 4.9310233402758133e-05,
      "loss": 1.0044,
      "step": 707
    },
    {
      "epoch": 0.38012147243381095,
      "grad_norm": 0.66796875,
      "learning_rate": 4.930826192362744e-05,
      "loss": 1.0846,
      "step": 708
    },
    {
      "epoch": 0.38065836716888696,
      "grad_norm": 0.69140625,
      "learning_rate": 4.930628767061622e-05,
      "loss": 0.9084,
      "step": 709
    },
    {
      "epoch": 0.38119526190396297,
      "grad_norm": 0.94921875,
      "learning_rate": 4.930431064394977e-05,
      "loss": 1.0305,
      "step": 710
    },
    {
      "epoch": 0.381732156639039,
      "grad_norm": 0.69921875,
      "learning_rate": 4.9302330843853686e-05,
      "loss": 0.8617,
      "step": 711
    },
    {
      "epoch": 0.382269051374115,
      "grad_norm": 0.82421875,
      "learning_rate": 4.930034827055388e-05,
      "loss": 0.992,
      "step": 712
    },
    {
      "epoch": 0.38280594610919094,
      "grad_norm": 0.734375,
      "learning_rate": 4.9298362924276604e-05,
      "loss": 1.0499,
      "step": 713
    },
    {
      "epoch": 0.38334284084426695,
      "grad_norm": 0.75,
      "learning_rate": 4.92963748052484e-05,
      "loss": 1.0629,
      "step": 714
    },
    {
      "epoch": 0.38387973557934296,
      "grad_norm": 0.7421875,
      "learning_rate": 4.929438391369615e-05,
      "loss": 1.0508,
      "step": 715
    },
    {
      "epoch": 0.384416630314419,
      "grad_norm": 0.6953125,
      "learning_rate": 4.929239024984702e-05,
      "loss": 1.0549,
      "step": 716
    },
    {
      "epoch": 0.384953525049495,
      "grad_norm": 0.796875,
      "learning_rate": 4.9290393813928534e-05,
      "loss": 0.7663,
      "step": 717
    },
    {
      "epoch": 0.385490419784571,
      "grad_norm": 0.7734375,
      "learning_rate": 4.9288394606168494e-05,
      "loss": 0.9906,
      "step": 718
    },
    {
      "epoch": 0.386027314519647,
      "grad_norm": 0.62890625,
      "learning_rate": 4.928639262679505e-05,
      "loss": 0.9696,
      "step": 719
    },
    {
      "epoch": 0.386564209254723,
      "grad_norm": 0.6328125,
      "learning_rate": 4.928438787603664e-05,
      "loss": 0.8313,
      "step": 720
    },
    {
      "epoch": 0.387101103989799,
      "grad_norm": 0.8984375,
      "learning_rate": 4.928238035412204e-05,
      "loss": 1.0274,
      "step": 721
    },
    {
      "epoch": 0.387637998724875,
      "grad_norm": 0.7109375,
      "learning_rate": 4.928037006128032e-05,
      "loss": 0.9403,
      "step": 722
    },
    {
      "epoch": 0.388174893459951,
      "grad_norm": 0.8046875,
      "learning_rate": 4.9278356997740904e-05,
      "loss": 0.9096,
      "step": 723
    },
    {
      "epoch": 0.388711788195027,
      "grad_norm": 0.640625,
      "learning_rate": 4.927634116373349e-05,
      "loss": 0.876,
      "step": 724
    },
    {
      "epoch": 0.389248682930103,
      "grad_norm": 0.953125,
      "learning_rate": 4.927432255948811e-05,
      "loss": 0.9584,
      "step": 725
    },
    {
      "epoch": 0.389785577665179,
      "grad_norm": 0.70703125,
      "learning_rate": 4.9272301185235116e-05,
      "loss": 0.8188,
      "step": 726
    },
    {
      "epoch": 0.390322472400255,
      "grad_norm": 0.83203125,
      "learning_rate": 4.927027704120518e-05,
      "loss": 1.1977,
      "step": 727
    },
    {
      "epoch": 0.39085936713533104,
      "grad_norm": 0.77734375,
      "learning_rate": 4.9268250127629265e-05,
      "loss": 0.9294,
      "step": 728
    },
    {
      "epoch": 0.39139626187040705,
      "grad_norm": 0.73046875,
      "learning_rate": 4.926622044473869e-05,
      "loss": 0.8519,
      "step": 729
    },
    {
      "epoch": 0.39193315660548306,
      "grad_norm": 0.69140625,
      "learning_rate": 4.926418799276504e-05,
      "loss": 0.9133,
      "step": 730
    },
    {
      "epoch": 0.39247005134055907,
      "grad_norm": 0.8203125,
      "learning_rate": 4.926215277194027e-05,
      "loss": 0.8231,
      "step": 731
    },
    {
      "epoch": 0.393006946075635,
      "grad_norm": 0.8203125,
      "learning_rate": 4.926011478249661e-05,
      "loss": 1.044,
      "step": 732
    },
    {
      "epoch": 0.39354384081071103,
      "grad_norm": 0.80859375,
      "learning_rate": 4.925807402466663e-05,
      "loss": 0.9314,
      "step": 733
    },
    {
      "epoch": 0.39408073554578704,
      "grad_norm": 0.78125,
      "learning_rate": 4.925603049868319e-05,
      "loss": 1.1221,
      "step": 734
    },
    {
      "epoch": 0.39461763028086305,
      "grad_norm": 0.76953125,
      "learning_rate": 4.92539842047795e-05,
      "loss": 1.0373,
      "step": 735
    },
    {
      "epoch": 0.39515452501593906,
      "grad_norm": 0.80859375,
      "learning_rate": 4.925193514318906e-05,
      "loss": 0.9045,
      "step": 736
    },
    {
      "epoch": 0.39569141975101507,
      "grad_norm": 0.875,
      "learning_rate": 4.924988331414569e-05,
      "loss": 1.0327,
      "step": 737
    },
    {
      "epoch": 0.3962283144860911,
      "grad_norm": 0.65234375,
      "learning_rate": 4.924782871788354e-05,
      "loss": 0.8908,
      "step": 738
    },
    {
      "epoch": 0.3967652092211671,
      "grad_norm": 0.953125,
      "learning_rate": 4.924577135463705e-05,
      "loss": 1.4129,
      "step": 739
    },
    {
      "epoch": 0.3973021039562431,
      "grad_norm": 0.68359375,
      "learning_rate": 4.924371122464101e-05,
      "loss": 0.9474,
      "step": 740
    },
    {
      "epoch": 0.39783899869131906,
      "grad_norm": 0.7265625,
      "learning_rate": 4.92416483281305e-05,
      "loss": 1.0218,
      "step": 741
    },
    {
      "epoch": 0.39837589342639507,
      "grad_norm": 0.72265625,
      "learning_rate": 4.923958266534091e-05,
      "loss": 0.7931,
      "step": 742
    },
    {
      "epoch": 0.3989127881614711,
      "grad_norm": 0.703125,
      "learning_rate": 4.923751423650797e-05,
      "loss": 0.9955,
      "step": 743
    },
    {
      "epoch": 0.3994496828965471,
      "grad_norm": 0.70703125,
      "learning_rate": 4.923544304186771e-05,
      "loss": 0.9051,
      "step": 744
    },
    {
      "epoch": 0.3999865776316231,
      "grad_norm": 0.73046875,
      "learning_rate": 4.923336908165649e-05,
      "loss": 0.8443,
      "step": 745
    },
    {
      "epoch": 0.4005234723666991,
      "grad_norm": 0.81640625,
      "learning_rate": 4.923129235611096e-05,
      "loss": 0.9819,
      "step": 746
    },
    {
      "epoch": 0.4010603671017751,
      "grad_norm": 0.921875,
      "learning_rate": 4.922921286546812e-05,
      "loss": 1.1488,
      "step": 747
    },
    {
      "epoch": 0.4015972618368511,
      "grad_norm": 0.69140625,
      "learning_rate": 4.922713060996524e-05,
      "loss": 0.8344,
      "step": 748
    },
    {
      "epoch": 0.40213415657192714,
      "grad_norm": 0.703125,
      "learning_rate": 4.922504558983995e-05,
      "loss": 0.9631,
      "step": 749
    },
    {
      "epoch": 0.40267105130700315,
      "grad_norm": 0.6953125,
      "learning_rate": 4.922295780533017e-05,
      "loss": 0.8723,
      "step": 750
    },
    {
      "epoch": 0.40267105130700315,
      "eval_loss": 0.9560365676879883,
      "eval_runtime": 48.5323,
      "eval_samples_per_second": 19.822,
      "eval_steps_per_second": 19.822,
      "step": 750
    },
    {
      "epoch": 0.4032079460420791,
      "grad_norm": 0.76171875,
      "learning_rate": 4.922086725667415e-05,
      "loss": 0.9373,
      "step": 751
    },
    {
      "epoch": 0.4037448407771551,
      "grad_norm": 0.77734375,
      "learning_rate": 4.921877394411045e-05,
      "loss": 1.0295,
      "step": 752
    },
    {
      "epoch": 0.4042817355122311,
      "grad_norm": 0.8671875,
      "learning_rate": 4.921667786787792e-05,
      "loss": 0.8979,
      "step": 753
    },
    {
      "epoch": 0.40481863024730713,
      "grad_norm": 0.94921875,
      "learning_rate": 4.9214579028215776e-05,
      "loss": 1.1372,
      "step": 754
    },
    {
      "epoch": 0.40535552498238314,
      "grad_norm": 0.81640625,
      "learning_rate": 4.9212477425363514e-05,
      "loss": 1.2777,
      "step": 755
    },
    {
      "epoch": 0.40589241971745915,
      "grad_norm": 0.6875,
      "learning_rate": 4.921037305956095e-05,
      "loss": 0.935,
      "step": 756
    },
    {
      "epoch": 0.40642931445253516,
      "grad_norm": 0.671875,
      "learning_rate": 4.920826593104822e-05,
      "loss": 0.8783,
      "step": 757
    },
    {
      "epoch": 0.40696620918761117,
      "grad_norm": 0.69921875,
      "learning_rate": 4.920615604006578e-05,
      "loss": 0.8403,
      "step": 758
    },
    {
      "epoch": 0.4075031039226872,
      "grad_norm": 0.8046875,
      "learning_rate": 4.9204043386854377e-05,
      "loss": 1.0082,
      "step": 759
    },
    {
      "epoch": 0.40803999865776314,
      "grad_norm": 0.703125,
      "learning_rate": 4.920192797165511e-05,
      "loss": 0.8792,
      "step": 760
    },
    {
      "epoch": 0.40857689339283915,
      "grad_norm": 0.87109375,
      "learning_rate": 4.919980979470937e-05,
      "loss": 1.0941,
      "step": 761
    },
    {
      "epoch": 0.40911378812791516,
      "grad_norm": 0.765625,
      "learning_rate": 4.919768885625887e-05,
      "loss": 1.1464,
      "step": 762
    },
    {
      "epoch": 0.40965068286299117,
      "grad_norm": 0.65625,
      "learning_rate": 4.919556515654562e-05,
      "loss": 0.8907,
      "step": 763
    },
    {
      "epoch": 0.4101875775980672,
      "grad_norm": 0.859375,
      "learning_rate": 4.9193438695811985e-05,
      "loss": 0.9733,
      "step": 764
    },
    {
      "epoch": 0.4107244723331432,
      "grad_norm": 0.8046875,
      "learning_rate": 4.919130947430061e-05,
      "loss": 1.0208,
      "step": 765
    },
    {
      "epoch": 0.4112613670682192,
      "grad_norm": 0.7890625,
      "learning_rate": 4.9189177492254455e-05,
      "loss": 1.1111,
      "step": 766
    },
    {
      "epoch": 0.4117982618032952,
      "grad_norm": 0.73046875,
      "learning_rate": 4.918704274991683e-05,
      "loss": 0.7626,
      "step": 767
    },
    {
      "epoch": 0.4123351565383712,
      "grad_norm": 0.76171875,
      "learning_rate": 4.9184905247531316e-05,
      "loss": 1.104,
      "step": 768
    },
    {
      "epoch": 0.4128720512734472,
      "grad_norm": 0.7578125,
      "learning_rate": 4.918276498534184e-05,
      "loss": 0.9929,
      "step": 769
    },
    {
      "epoch": 0.4134089460085232,
      "grad_norm": 0.9609375,
      "learning_rate": 4.918062196359263e-05,
      "loss": 1.2401,
      "step": 770
    },
    {
      "epoch": 0.4139458407435992,
      "grad_norm": 0.640625,
      "learning_rate": 4.917847618252822e-05,
      "loss": 1.0889,
      "step": 771
    },
    {
      "epoch": 0.4144827354786752,
      "grad_norm": 0.78515625,
      "learning_rate": 4.917632764239349e-05,
      "loss": 0.8042,
      "step": 772
    },
    {
      "epoch": 0.4150196302137512,
      "grad_norm": 0.88671875,
      "learning_rate": 4.917417634343361e-05,
      "loss": 1.1121,
      "step": 773
    },
    {
      "epoch": 0.4155565249488272,
      "grad_norm": 0.765625,
      "learning_rate": 4.9172022285894074e-05,
      "loss": 1.0736,
      "step": 774
    },
    {
      "epoch": 0.41609341968390323,
      "grad_norm": 0.7421875,
      "learning_rate": 4.916986547002067e-05,
      "loss": 1.145,
      "step": 775
    },
    {
      "epoch": 0.41663031441897924,
      "grad_norm": 0.796875,
      "learning_rate": 4.9167705896059527e-05,
      "loss": 1.0813,
      "step": 776
    },
    {
      "epoch": 0.41716720915405525,
      "grad_norm": 0.85546875,
      "learning_rate": 4.916554356425709e-05,
      "loss": 0.9823,
      "step": 777
    },
    {
      "epoch": 0.41770410388913126,
      "grad_norm": 0.765625,
      "learning_rate": 4.91633784748601e-05,
      "loss": 1.1357,
      "step": 778
    },
    {
      "epoch": 0.4182409986242072,
      "grad_norm": 0.671875,
      "learning_rate": 4.9161210628115615e-05,
      "loss": 0.7951,
      "step": 779
    },
    {
      "epoch": 0.4187778933592832,
      "grad_norm": 0.66015625,
      "learning_rate": 4.915904002427103e-05,
      "loss": 0.8144,
      "step": 780
    },
    {
      "epoch": 0.41931478809435924,
      "grad_norm": 0.703125,
      "learning_rate": 4.915686666357402e-05,
      "loss": 0.9838,
      "step": 781
    },
    {
      "epoch": 0.41985168282943525,
      "grad_norm": 0.83984375,
      "learning_rate": 4.9154690546272606e-05,
      "loss": 0.9819,
      "step": 782
    },
    {
      "epoch": 0.42038857756451126,
      "grad_norm": 0.8203125,
      "learning_rate": 4.915251167261511e-05,
      "loss": 1.1217,
      "step": 783
    },
    {
      "epoch": 0.42092547229958727,
      "grad_norm": 0.66015625,
      "learning_rate": 4.9150330042850155e-05,
      "loss": 0.8995,
      "step": 784
    },
    {
      "epoch": 0.4214623670346633,
      "grad_norm": 0.7421875,
      "learning_rate": 4.914814565722671e-05,
      "loss": 0.9941,
      "step": 785
    },
    {
      "epoch": 0.4219992617697393,
      "grad_norm": 0.75,
      "learning_rate": 4.9145958515994025e-05,
      "loss": 0.9457,
      "step": 786
    },
    {
      "epoch": 0.4225361565048153,
      "grad_norm": 0.8671875,
      "learning_rate": 4.914376861940171e-05,
      "loss": 1.0138,
      "step": 787
    },
    {
      "epoch": 0.42307305123989125,
      "grad_norm": 0.578125,
      "learning_rate": 4.914157596769962e-05,
      "loss": 0.7708,
      "step": 788
    },
    {
      "epoch": 0.42360994597496726,
      "grad_norm": 0.859375,
      "learning_rate": 4.9139380561137994e-05,
      "loss": 1.121,
      "step": 789
    },
    {
      "epoch": 0.42414684071004327,
      "grad_norm": 0.6328125,
      "learning_rate": 4.9137182399967343e-05,
      "loss": 0.8208,
      "step": 790
    },
    {
      "epoch": 0.4246837354451193,
      "grad_norm": 0.73828125,
      "learning_rate": 4.91349814844385e-05,
      "loss": 1.0937,
      "step": 791
    },
    {
      "epoch": 0.4252206301801953,
      "grad_norm": 0.77734375,
      "learning_rate": 4.9132777814802634e-05,
      "loss": 0.9219,
      "step": 792
    },
    {
      "epoch": 0.4257575249152713,
      "grad_norm": 0.921875,
      "learning_rate": 4.91305713913112e-05,
      "loss": 1.0178,
      "step": 793
    },
    {
      "epoch": 0.4262944196503473,
      "grad_norm": 0.7265625,
      "learning_rate": 4.9128362214215986e-05,
      "loss": 1.0026,
      "step": 794
    },
    {
      "epoch": 0.4268313143854233,
      "grad_norm": 0.67578125,
      "learning_rate": 4.912615028376907e-05,
      "loss": 1.0,
      "step": 795
    },
    {
      "epoch": 0.42736820912049933,
      "grad_norm": 0.96875,
      "learning_rate": 4.912393560022288e-05,
      "loss": 0.9567,
      "step": 796
    },
    {
      "epoch": 0.42790510385557534,
      "grad_norm": 0.87890625,
      "learning_rate": 4.912171816383014e-05,
      "loss": 1.189,
      "step": 797
    },
    {
      "epoch": 0.4284419985906513,
      "grad_norm": 0.62890625,
      "learning_rate": 4.911949797484388e-05,
      "loss": 0.8717,
      "step": 798
    },
    {
      "epoch": 0.4289788933257273,
      "grad_norm": 0.74609375,
      "learning_rate": 4.911727503351744e-05,
      "loss": 1.0839,
      "step": 799
    },
    {
      "epoch": 0.4295157880608033,
      "grad_norm": 0.64453125,
      "learning_rate": 4.9115049340104505e-05,
      "loss": 0.8312,
      "step": 800
    },
    {
      "epoch": 0.4295157880608033,
      "eval_loss": 0.9527614712715149,
      "eval_runtime": 46.4794,
      "eval_samples_per_second": 20.697,
      "eval_steps_per_second": 20.697,
      "step": 800
    },
    {
      "epoch": 0.4300526827958793,
      "grad_norm": 0.80078125,
      "learning_rate": 4.9112820894859046e-05,
      "loss": 0.815,
      "step": 801
    },
    {
      "epoch": 0.43058957753095534,
      "grad_norm": 0.78125,
      "learning_rate": 4.911058969803536e-05,
      "loss": 1.0427,
      "step": 802
    },
    {
      "epoch": 0.43112647226603135,
      "grad_norm": 0.7890625,
      "learning_rate": 4.9108355749888056e-05,
      "loss": 0.8677,
      "step": 803
    },
    {
      "epoch": 0.43166336700110736,
      "grad_norm": 0.69921875,
      "learning_rate": 4.910611905067205e-05,
      "loss": 1.048,
      "step": 804
    },
    {
      "epoch": 0.43220026173618337,
      "grad_norm": 0.78515625,
      "learning_rate": 4.910387960064259e-05,
      "loss": 0.8238,
      "step": 805
    },
    {
      "epoch": 0.4327371564712594,
      "grad_norm": 0.82421875,
      "learning_rate": 4.91016374000552e-05,
      "loss": 0.8858,
      "step": 806
    },
    {
      "epoch": 0.43327405120633533,
      "grad_norm": 0.73828125,
      "learning_rate": 4.9099392449165774e-05,
      "loss": 1.0776,
      "step": 807
    },
    {
      "epoch": 0.43381094594141134,
      "grad_norm": 0.65234375,
      "learning_rate": 4.909714474823047e-05,
      "loss": 0.7519,
      "step": 808
    },
    {
      "epoch": 0.43434784067648735,
      "grad_norm": 0.7265625,
      "learning_rate": 4.909489429750579e-05,
      "loss": 0.6753,
      "step": 809
    },
    {
      "epoch": 0.43488473541156336,
      "grad_norm": 1.0,
      "learning_rate": 4.909264109724853e-05,
      "loss": 1.1741,
      "step": 810
    },
    {
      "epoch": 0.43542163014663937,
      "grad_norm": 0.91796875,
      "learning_rate": 4.909038514771581e-05,
      "loss": 1.0277,
      "step": 811
    },
    {
      "epoch": 0.4359585248817154,
      "grad_norm": 0.7265625,
      "learning_rate": 4.9088126449165065e-05,
      "loss": 0.9186,
      "step": 812
    },
    {
      "epoch": 0.4364954196167914,
      "grad_norm": 0.7109375,
      "learning_rate": 4.9085865001854035e-05,
      "loss": 1.105,
      "step": 813
    },
    {
      "epoch": 0.4370323143518674,
      "grad_norm": 0.67578125,
      "learning_rate": 4.90836008060408e-05,
      "loss": 0.7978,
      "step": 814
    },
    {
      "epoch": 0.4375692090869434,
      "grad_norm": 0.8125,
      "learning_rate": 4.90813338619837e-05,
      "loss": 1.0604,
      "step": 815
    },
    {
      "epoch": 0.4381061038220194,
      "grad_norm": 0.7734375,
      "learning_rate": 4.907906416994146e-05,
      "loss": 1.0758,
      "step": 816
    },
    {
      "epoch": 0.4386429985570954,
      "grad_norm": 0.703125,
      "learning_rate": 4.907679173017305e-05,
      "loss": 1.1113,
      "step": 817
    },
    {
      "epoch": 0.4391798932921714,
      "grad_norm": 0.765625,
      "learning_rate": 4.9074516542937795e-05,
      "loss": 0.9405,
      "step": 818
    },
    {
      "epoch": 0.4397167880272474,
      "grad_norm": 0.73046875,
      "learning_rate": 4.907223860849533e-05,
      "loss": 0.9341,
      "step": 819
    },
    {
      "epoch": 0.4402536827623234,
      "grad_norm": 0.68359375,
      "learning_rate": 4.9069957927105586e-05,
      "loss": 0.9659,
      "step": 820
    },
    {
      "epoch": 0.4407905774973994,
      "grad_norm": 0.74609375,
      "learning_rate": 4.9067674499028824e-05,
      "loss": 0.9065,
      "step": 821
    },
    {
      "epoch": 0.4413274722324754,
      "grad_norm": 0.8046875,
      "learning_rate": 4.906538832452561e-05,
      "loss": 0.9159,
      "step": 822
    },
    {
      "epoch": 0.44186436696755144,
      "grad_norm": 0.7265625,
      "learning_rate": 4.906309940385682e-05,
      "loss": 0.7843,
      "step": 823
    },
    {
      "epoch": 0.44240126170262745,
      "grad_norm": 0.89453125,
      "learning_rate": 4.9060807737283656e-05,
      "loss": 0.9702,
      "step": 824
    },
    {
      "epoch": 0.44293815643770346,
      "grad_norm": 0.94921875,
      "learning_rate": 4.905851332506762e-05,
      "loss": 1.0678,
      "step": 825
    },
    {
      "epoch": 0.4434750511727794,
      "grad_norm": 0.65625,
      "learning_rate": 4.905621616747054e-05,
      "loss": 1.0175,
      "step": 826
    },
    {
      "epoch": 0.4440119459078554,
      "grad_norm": 0.91015625,
      "learning_rate": 4.905391626475455e-05,
      "loss": 1.1155,
      "step": 827
    },
    {
      "epoch": 0.44454884064293143,
      "grad_norm": 0.6796875,
      "learning_rate": 4.905161361718209e-05,
      "loss": 1.0649,
      "step": 828
    },
    {
      "epoch": 0.44508573537800744,
      "grad_norm": 0.8671875,
      "learning_rate": 4.9049308225015936e-05,
      "loss": 0.9817,
      "step": 829
    },
    {
      "epoch": 0.44562263011308345,
      "grad_norm": 0.66796875,
      "learning_rate": 4.9047000088519144e-05,
      "loss": 0.9786,
      "step": 830
    },
    {
      "epoch": 0.44615952484815946,
      "grad_norm": 0.6953125,
      "learning_rate": 4.904468920795512e-05,
      "loss": 0.8776,
      "step": 831
    },
    {
      "epoch": 0.44669641958323547,
      "grad_norm": 0.89453125,
      "learning_rate": 4.9042375583587555e-05,
      "loss": 0.9766,
      "step": 832
    },
    {
      "epoch": 0.4472333143183115,
      "grad_norm": 0.59765625,
      "learning_rate": 4.904005921568045e-05,
      "loss": 0.8547,
      "step": 833
    },
    {
      "epoch": 0.4477702090533875,
      "grad_norm": 0.76171875,
      "learning_rate": 4.9037740104498166e-05,
      "loss": 0.8818,
      "step": 834
    },
    {
      "epoch": 0.4483071037884635,
      "grad_norm": 0.8359375,
      "learning_rate": 4.9035418250305314e-05,
      "loss": 1.1426,
      "step": 835
    },
    {
      "epoch": 0.44884399852353946,
      "grad_norm": 0.84375,
      "learning_rate": 4.903309365336686e-05,
      "loss": 1.0365,
      "step": 836
    },
    {
      "epoch": 0.44938089325861547,
      "grad_norm": 0.8046875,
      "learning_rate": 4.903076631394806e-05,
      "loss": 0.9775,
      "step": 837
    },
    {
      "epoch": 0.4499177879936915,
      "grad_norm": 1.40625,
      "learning_rate": 4.90284362323145e-05,
      "loss": 1.3363,
      "step": 838
    },
    {
      "epoch": 0.4504546827287675,
      "grad_norm": 0.74609375,
      "learning_rate": 4.9026103408732065e-05,
      "loss": 1.0259,
      "step": 839
    },
    {
      "epoch": 0.4509915774638435,
      "grad_norm": 0.67578125,
      "learning_rate": 4.902376784346697e-05,
      "loss": 0.9859,
      "step": 840
    },
    {
      "epoch": 0.4515284721989195,
      "grad_norm": 0.6953125,
      "learning_rate": 4.902142953678573e-05,
      "loss": 0.9236,
      "step": 841
    },
    {
      "epoch": 0.4520653669339955,
      "grad_norm": 0.78125,
      "learning_rate": 4.901908848895517e-05,
      "loss": 0.9455,
      "step": 842
    },
    {
      "epoch": 0.4526022616690715,
      "grad_norm": 0.84765625,
      "learning_rate": 4.901674470024244e-05,
      "loss": 1.111,
      "step": 843
    },
    {
      "epoch": 0.45313915640414754,
      "grad_norm": 0.83203125,
      "learning_rate": 4.901439817091499e-05,
      "loss": 0.869,
      "step": 844
    },
    {
      "epoch": 0.4536760511392235,
      "grad_norm": 0.80078125,
      "learning_rate": 4.9012048901240585e-05,
      "loss": 0.8917,
      "step": 845
    },
    {
      "epoch": 0.4542129458742995,
      "grad_norm": 0.75390625,
      "learning_rate": 4.9009696891487325e-05,
      "loss": 1.0983,
      "step": 846
    },
    {
      "epoch": 0.4547498406093755,
      "grad_norm": 0.8203125,
      "learning_rate": 4.900734214192358e-05,
      "loss": 1.0704,
      "step": 847
    },
    {
      "epoch": 0.4552867353444515,
      "grad_norm": 0.8828125,
      "learning_rate": 4.9004984652818076e-05,
      "loss": 0.898,
      "step": 848
    },
    {
      "epoch": 0.45582363007952753,
      "grad_norm": 0.875,
      "learning_rate": 4.900262442443981e-05,
      "loss": 1.1245,
      "step": 849
    },
    {
      "epoch": 0.45636052481460354,
      "grad_norm": 0.58984375,
      "learning_rate": 4.900026145705815e-05,
      "loss": 0.7363,
      "step": 850
    },
    {
      "epoch": 0.45636052481460354,
      "eval_loss": 0.9519462585449219,
      "eval_runtime": 46.3396,
      "eval_samples_per_second": 20.76,
      "eval_steps_per_second": 20.76,
      "step": 850
    },
    {
      "epoch": 0.45689741954967955,
      "grad_norm": 0.7421875,
      "learning_rate": 4.8997895750942704e-05,
      "loss": 1.0622,
      "step": 851
    },
    {
      "epoch": 0.45743431428475556,
      "grad_norm": 0.8828125,
      "learning_rate": 4.899552730636345e-05,
      "loss": 0.9928,
      "step": 852
    },
    {
      "epoch": 0.45797120901983157,
      "grad_norm": 0.75390625,
      "learning_rate": 4.899315612359065e-05,
      "loss": 0.8667,
      "step": 853
    },
    {
      "epoch": 0.4585081037549076,
      "grad_norm": 0.74609375,
      "learning_rate": 4.899078220289489e-05,
      "loss": 1.0888,
      "step": 854
    },
    {
      "epoch": 0.45904499848998354,
      "grad_norm": 0.84765625,
      "learning_rate": 4.8988405544547054e-05,
      "loss": 1.2631,
      "step": 855
    },
    {
      "epoch": 0.45958189322505955,
      "grad_norm": 0.97265625,
      "learning_rate": 4.898602614881836e-05,
      "loss": 1.1243,
      "step": 856
    },
    {
      "epoch": 0.46011878796013556,
      "grad_norm": 0.7734375,
      "learning_rate": 4.898364401598034e-05,
      "loss": 1.0798,
      "step": 857
    },
    {
      "epoch": 0.46065568269521157,
      "grad_norm": 0.73828125,
      "learning_rate": 4.898125914630479e-05,
      "loss": 1.0191,
      "step": 858
    },
    {
      "epoch": 0.4611925774302876,
      "grad_norm": 0.7578125,
      "learning_rate": 4.897887154006388e-05,
      "loss": 0.9533,
      "step": 859
    },
    {
      "epoch": 0.4617294721653636,
      "grad_norm": 0.640625,
      "learning_rate": 4.897648119753006e-05,
      "loss": 0.9253,
      "step": 860
    },
    {
      "epoch": 0.4622663669004396,
      "grad_norm": 2.328125,
      "learning_rate": 4.8974088118976094e-05,
      "loss": 0.9924,
      "step": 861
    },
    {
      "epoch": 0.4628032616355156,
      "grad_norm": 0.72265625,
      "learning_rate": 4.897169230467506e-05,
      "loss": 0.9716,
      "step": 862
    },
    {
      "epoch": 0.4633401563705916,
      "grad_norm": 0.64453125,
      "learning_rate": 4.896929375490037e-05,
      "loss": 0.9661,
      "step": 863
    },
    {
      "epoch": 0.46387705110566757,
      "grad_norm": 0.78125,
      "learning_rate": 4.896689246992572e-05,
      "loss": 0.9774,
      "step": 864
    },
    {
      "epoch": 0.4644139458407436,
      "grad_norm": 0.8046875,
      "learning_rate": 4.896448845002511e-05,
      "loss": 0.8279,
      "step": 865
    },
    {
      "epoch": 0.4649508405758196,
      "grad_norm": 0.6484375,
      "learning_rate": 4.8962081695472886e-05,
      "loss": 0.8951,
      "step": 866
    },
    {
      "epoch": 0.4654877353108956,
      "grad_norm": 0.65234375,
      "learning_rate": 4.895967220654368e-05,
      "loss": 0.8701,
      "step": 867
    },
    {
      "epoch": 0.4660246300459716,
      "grad_norm": 0.71875,
      "learning_rate": 4.895725998351246e-05,
      "loss": 0.887,
      "step": 868
    },
    {
      "epoch": 0.4665615247810476,
      "grad_norm": 1.5546875,
      "learning_rate": 4.8954845026654484e-05,
      "loss": 1.0683,
      "step": 869
    },
    {
      "epoch": 0.46709841951612363,
      "grad_norm": 0.76953125,
      "learning_rate": 4.8952427336245324e-05,
      "loss": 1.0042,
      "step": 870
    },
    {
      "epoch": 0.46763531425119964,
      "grad_norm": 0.703125,
      "learning_rate": 4.8950006912560867e-05,
      "loss": 0.9935,
      "step": 871
    },
    {
      "epoch": 0.46817220898627565,
      "grad_norm": 0.77734375,
      "learning_rate": 4.894758375587733e-05,
      "loss": 0.8753,
      "step": 872
    },
    {
      "epoch": 0.4687091037213516,
      "grad_norm": 0.79296875,
      "learning_rate": 4.89451578664712e-05,
      "loss": 0.9927,
      "step": 873
    },
    {
      "epoch": 0.4692459984564276,
      "grad_norm": 0.68359375,
      "learning_rate": 4.894272924461932e-05,
      "loss": 0.9619,
      "step": 874
    },
    {
      "epoch": 0.4697828931915036,
      "grad_norm": 0.703125,
      "learning_rate": 4.894029789059883e-05,
      "loss": 1.0445,
      "step": 875
    },
    {
      "epoch": 0.47031978792657964,
      "grad_norm": 0.71484375,
      "learning_rate": 4.8937863804687165e-05,
      "loss": 1.0827,
      "step": 876
    },
    {
      "epoch": 0.47085668266165565,
      "grad_norm": 0.70703125,
      "learning_rate": 4.893542698716209e-05,
      "loss": 0.9637,
      "step": 877
    },
    {
      "epoch": 0.47139357739673166,
      "grad_norm": 0.64453125,
      "learning_rate": 4.893298743830168e-05,
      "loss": 0.8322,
      "step": 878
    },
    {
      "epoch": 0.47193047213180767,
      "grad_norm": 0.78515625,
      "learning_rate": 4.8930545158384324e-05,
      "loss": 1.0453,
      "step": 879
    },
    {
      "epoch": 0.4724673668668837,
      "grad_norm": 0.8203125,
      "learning_rate": 4.89281001476887e-05,
      "loss": 0.9922,
      "step": 880
    },
    {
      "epoch": 0.4730042616019597,
      "grad_norm": 0.609375,
      "learning_rate": 4.892565240649383e-05,
      "loss": 0.863,
      "step": 881
    },
    {
      "epoch": 0.4735411563370357,
      "grad_norm": 0.78515625,
      "learning_rate": 4.892320193507902e-05,
      "loss": 0.8975,
      "step": 882
    },
    {
      "epoch": 0.47407805107211165,
      "grad_norm": 0.7265625,
      "learning_rate": 4.892074873372392e-05,
      "loss": 0.859,
      "step": 883
    },
    {
      "epoch": 0.47461494580718766,
      "grad_norm": 0.765625,
      "learning_rate": 4.8918292802708445e-05,
      "loss": 1.1057,
      "step": 884
    },
    {
      "epoch": 0.47515184054226367,
      "grad_norm": 0.80078125,
      "learning_rate": 4.891583414231287e-05,
      "loss": 0.9268,
      "step": 885
    },
    {
      "epoch": 0.4756887352773397,
      "grad_norm": 0.66796875,
      "learning_rate": 4.891337275281774e-05,
      "loss": 1.0443,
      "step": 886
    },
    {
      "epoch": 0.4762256300124157,
      "grad_norm": 0.69140625,
      "learning_rate": 4.8910908634503946e-05,
      "loss": 0.8716,
      "step": 887
    },
    {
      "epoch": 0.4767625247474917,
      "grad_norm": 0.68359375,
      "learning_rate": 4.890844178765267e-05,
      "loss": 0.7626,
      "step": 888
    },
    {
      "epoch": 0.4772994194825677,
      "grad_norm": 0.74609375,
      "learning_rate": 4.89059722125454e-05,
      "loss": 0.9247,
      "step": 889
    },
    {
      "epoch": 0.4778363142176437,
      "grad_norm": 0.84765625,
      "learning_rate": 4.8903499909463966e-05,
      "loss": 0.9973,
      "step": 890
    },
    {
      "epoch": 0.47837320895271973,
      "grad_norm": 0.82421875,
      "learning_rate": 4.890102487869048e-05,
      "loss": 0.8405,
      "step": 891
    },
    {
      "epoch": 0.4789101036877957,
      "grad_norm": 1.265625,
      "learning_rate": 4.889854712050737e-05,
      "loss": 1.079,
      "step": 892
    },
    {
      "epoch": 0.4794469984228717,
      "grad_norm": 1.0078125,
      "learning_rate": 4.889606663519739e-05,
      "loss": 1.0182,
      "step": 893
    },
    {
      "epoch": 0.4799838931579477,
      "grad_norm": 1.3203125,
      "learning_rate": 4.8893583423043574e-05,
      "loss": 0.9582,
      "step": 894
    },
    {
      "epoch": 0.4805207878930237,
      "grad_norm": 1.0625,
      "learning_rate": 4.8891097484329315e-05,
      "loss": 0.942,
      "step": 895
    },
    {
      "epoch": 0.4810576826280997,
      "grad_norm": 0.796875,
      "learning_rate": 4.888860881933826e-05,
      "loss": 0.9177,
      "step": 896
    },
    {
      "epoch": 0.48159457736317574,
      "grad_norm": 0.68359375,
      "learning_rate": 4.8886117428354434e-05,
      "loss": 0.983,
      "step": 897
    },
    {
      "epoch": 0.48213147209825175,
      "grad_norm": 0.8203125,
      "learning_rate": 4.888362331166211e-05,
      "loss": 0.9985,
      "step": 898
    },
    {
      "epoch": 0.48266836683332776,
      "grad_norm": 0.73828125,
      "learning_rate": 4.88811264695459e-05,
      "loss": 0.9199,
      "step": 899
    },
    {
      "epoch": 0.48320526156840377,
      "grad_norm": 0.828125,
      "learning_rate": 4.887862690229073e-05,
      "loss": 0.9292,
      "step": 900
    },
    {
      "epoch": 0.48320526156840377,
      "eval_loss": 0.949388325214386,
      "eval_runtime": 45.824,
      "eval_samples_per_second": 20.993,
      "eval_steps_per_second": 20.993,
      "step": 900
    },
    {
      "epoch": 0.4837421563034798,
      "grad_norm": 0.8046875,
      "learning_rate": 4.887612461018183e-05,
      "loss": 0.914,
      "step": 901
    },
    {
      "epoch": 0.48427905103855573,
      "grad_norm": 0.66796875,
      "learning_rate": 4.887361959350475e-05,
      "loss": 0.937,
      "step": 902
    },
    {
      "epoch": 0.48481594577363174,
      "grad_norm": 0.64453125,
      "learning_rate": 4.887111185254534e-05,
      "loss": 0.9092,
      "step": 903
    },
    {
      "epoch": 0.48535284050870775,
      "grad_norm": 0.69921875,
      "learning_rate": 4.8868601387589765e-05,
      "loss": 0.805,
      "step": 904
    },
    {
      "epoch": 0.48588973524378376,
      "grad_norm": 0.65625,
      "learning_rate": 4.88660881989245e-05,
      "loss": 0.8396,
      "step": 905
    },
    {
      "epoch": 0.48642662997885977,
      "grad_norm": 0.6875,
      "learning_rate": 4.8863572286836324e-05,
      "loss": 0.9667,
      "step": 906
    },
    {
      "epoch": 0.4869635247139358,
      "grad_norm": 0.69140625,
      "learning_rate": 4.8861053651612347e-05,
      "loss": 0.8923,
      "step": 907
    },
    {
      "epoch": 0.4875004194490118,
      "grad_norm": 0.71875,
      "learning_rate": 4.885853229353998e-05,
      "loss": 0.7539,
      "step": 908
    },
    {
      "epoch": 0.4880373141840878,
      "grad_norm": 0.6796875,
      "learning_rate": 4.8856008212906925e-05,
      "loss": 0.7695,
      "step": 909
    },
    {
      "epoch": 0.4885742089191638,
      "grad_norm": 0.7109375,
      "learning_rate": 4.885348141000122e-05,
      "loss": 1.1176,
      "step": 910
    },
    {
      "epoch": 0.48911110365423976,
      "grad_norm": 0.82421875,
      "learning_rate": 4.8850951885111215e-05,
      "loss": 0.9238,
      "step": 911
    },
    {
      "epoch": 0.4896479983893158,
      "grad_norm": 0.65234375,
      "learning_rate": 4.8848419638525545e-05,
      "loss": 0.8153,
      "step": 912
    },
    {
      "epoch": 0.4901848931243918,
      "grad_norm": 0.73828125,
      "learning_rate": 4.8845884670533184e-05,
      "loss": 1.0782,
      "step": 913
    },
    {
      "epoch": 0.4907217878594678,
      "grad_norm": 0.69140625,
      "learning_rate": 4.884334698142339e-05,
      "loss": 0.8182,
      "step": 914
    },
    {
      "epoch": 0.4912586825945438,
      "grad_norm": 0.65625,
      "learning_rate": 4.884080657148576e-05,
      "loss": 0.9822,
      "step": 915
    },
    {
      "epoch": 0.4917955773296198,
      "grad_norm": 0.72265625,
      "learning_rate": 4.8838263441010186e-05,
      "loss": 0.8813,
      "step": 916
    },
    {
      "epoch": 0.4923324720646958,
      "grad_norm": 0.6796875,
      "learning_rate": 4.8835717590286857e-05,
      "loss": 0.8641,
      "step": 917
    },
    {
      "epoch": 0.49286936679977184,
      "grad_norm": 0.74609375,
      "learning_rate": 4.88331690196063e-05,
      "loss": 1.0673,
      "step": 918
    },
    {
      "epoch": 0.49340626153484785,
      "grad_norm": 0.74609375,
      "learning_rate": 4.883061772925933e-05,
      "loss": 0.812,
      "step": 919
    },
    {
      "epoch": 0.49394315626992386,
      "grad_norm": 1.15625,
      "learning_rate": 4.88280637195371e-05,
      "loss": 1.0478,
      "step": 920
    },
    {
      "epoch": 0.4944800510049998,
      "grad_norm": 0.81640625,
      "learning_rate": 4.882550699073103e-05,
      "loss": 0.9996,
      "step": 921
    },
    {
      "epoch": 0.4950169457400758,
      "grad_norm": 0.94921875,
      "learning_rate": 4.882294754313289e-05,
      "loss": 1.2174,
      "step": 922
    },
    {
      "epoch": 0.49555384047515183,
      "grad_norm": 0.796875,
      "learning_rate": 4.882038537703474e-05,
      "loss": 1.1008,
      "step": 923
    },
    {
      "epoch": 0.49609073521022784,
      "grad_norm": 0.84375,
      "learning_rate": 4.881782049272896e-05,
      "loss": 1.1056,
      "step": 924
    },
    {
      "epoch": 0.49662762994530385,
      "grad_norm": 0.78125,
      "learning_rate": 4.881525289050824e-05,
      "loss": 1.0885,
      "step": 925
    },
    {
      "epoch": 0.49716452468037986,
      "grad_norm": 0.81640625,
      "learning_rate": 4.8812682570665556e-05,
      "loss": 0.8931,
      "step": 926
    },
    {
      "epoch": 0.49770141941545587,
      "grad_norm": 0.828125,
      "learning_rate": 4.881010953349423e-05,
      "loss": 0.9753,
      "step": 927
    },
    {
      "epoch": 0.4982383141505319,
      "grad_norm": 0.66015625,
      "learning_rate": 4.880753377928788e-05,
      "loss": 0.9947,
      "step": 928
    },
    {
      "epoch": 0.4987752088856079,
      "grad_norm": 0.80859375,
      "learning_rate": 4.8804955308340425e-05,
      "loss": 0.9936,
      "step": 929
    },
    {
      "epoch": 0.49931210362068384,
      "grad_norm": 0.796875,
      "learning_rate": 4.880237412094611e-05,
      "loss": 1.0456,
      "step": 930
    },
    {
      "epoch": 0.49984899835575985,
      "grad_norm": 0.76953125,
      "learning_rate": 4.879979021739947e-05,
      "loss": 0.8974,
      "step": 931
    },
    {
      "epoch": 0.5003858930908359,
      "grad_norm": 0.78515625,
      "learning_rate": 4.879720359799537e-05,
      "loss": 0.9897,
      "step": 932
    },
    {
      "epoch": 0.5009227878259119,
      "grad_norm": 0.8828125,
      "learning_rate": 4.879461426302897e-05,
      "loss": 1.0921,
      "step": 933
    },
    {
      "epoch": 0.5014596825609879,
      "grad_norm": 0.6875,
      "learning_rate": 4.879202221279575e-05,
      "loss": 0.9196,
      "step": 934
    },
    {
      "epoch": 0.5019965772960638,
      "grad_norm": 0.765625,
      "learning_rate": 4.8789427447591486e-05,
      "loss": 1.1765,
      "step": 935
    },
    {
      "epoch": 0.5025334720311398,
      "grad_norm": 0.76171875,
      "learning_rate": 4.878682996771229e-05,
      "loss": 0.9488,
      "step": 936
    },
    {
      "epoch": 0.5030703667662159,
      "grad_norm": 0.734375,
      "learning_rate": 4.8784229773454555e-05,
      "loss": 0.8722,
      "step": 937
    },
    {
      "epoch": 0.5036072615012919,
      "grad_norm": 1.1015625,
      "learning_rate": 4.8781626865115005e-05,
      "loss": 1.0332,
      "step": 938
    },
    {
      "epoch": 0.5041441562363679,
      "grad_norm": 0.69140625,
      "learning_rate": 4.877902124299065e-05,
      "loss": 0.8815,
      "step": 939
    },
    {
      "epoch": 0.5046810509714439,
      "grad_norm": 0.625,
      "learning_rate": 4.877641290737884e-05,
      "loss": 0.9193,
      "step": 940
    },
    {
      "epoch": 0.5052179457065199,
      "grad_norm": 1.4609375,
      "learning_rate": 4.877380185857722e-05,
      "loss": 1.0847,
      "step": 941
    },
    {
      "epoch": 0.5057548404415959,
      "grad_norm": 0.73828125,
      "learning_rate": 4.877118809688372e-05,
      "loss": 0.8385,
      "step": 942
    },
    {
      "epoch": 0.5062917351766719,
      "grad_norm": 0.7734375,
      "learning_rate": 4.876857162259664e-05,
      "loss": 1.125,
      "step": 943
    },
    {
      "epoch": 0.5068286299117479,
      "grad_norm": 0.734375,
      "learning_rate": 4.8765952436014515e-05,
      "loss": 0.9923,
      "step": 944
    },
    {
      "epoch": 0.5073655246468239,
      "grad_norm": 0.72265625,
      "learning_rate": 4.8763330537436255e-05,
      "loss": 0.8513,
      "step": 945
    },
    {
      "epoch": 0.5079024193819,
      "grad_norm": 1.03125,
      "learning_rate": 4.876070592716105e-05,
      "loss": 1.0621,
      "step": 946
    },
    {
      "epoch": 0.508439314116976,
      "grad_norm": 0.8125,
      "learning_rate": 4.875807860548838e-05,
      "loss": 1.0328,
      "step": 947
    },
    {
      "epoch": 0.508976208852052,
      "grad_norm": 0.64453125,
      "learning_rate": 4.875544857271808e-05,
      "loss": 0.9213,
      "step": 948
    },
    {
      "epoch": 0.509513103587128,
      "grad_norm": 0.67578125,
      "learning_rate": 4.875281582915026e-05,
      "loss": 0.883,
      "step": 949
    },
    {
      "epoch": 0.510049998322204,
      "grad_norm": 0.65234375,
      "learning_rate": 4.8750180375085344e-05,
      "loss": 0.9326,
      "step": 950
    },
    {
      "epoch": 0.510049998322204,
      "eval_loss": 0.9448692202568054,
      "eval_runtime": 46.4895,
      "eval_samples_per_second": 20.693,
      "eval_steps_per_second": 20.693,
      "step": 950
    },
    {
      "epoch": 0.51058689305728,
      "grad_norm": 0.80078125,
      "learning_rate": 4.874754221082407e-05,
      "loss": 1.0214,
      "step": 951
    },
    {
      "epoch": 0.511123787792356,
      "grad_norm": 0.85546875,
      "learning_rate": 4.874490133666749e-05,
      "loss": 0.8779,
      "step": 952
    },
    {
      "epoch": 0.511660682527432,
      "grad_norm": 0.8203125,
      "learning_rate": 4.874225775291697e-05,
      "loss": 1.1105,
      "step": 953
    },
    {
      "epoch": 0.5121975772625079,
      "grad_norm": 0.9296875,
      "learning_rate": 4.873961145987417e-05,
      "loss": 1.1597,
      "step": 954
    },
    {
      "epoch": 0.5127344719975839,
      "grad_norm": 0.84375,
      "learning_rate": 4.873696245784106e-05,
      "loss": 0.9641,
      "step": 955
    },
    {
      "epoch": 0.5132713667326599,
      "grad_norm": 0.75,
      "learning_rate": 4.8734310747119935e-05,
      "loss": 1.0023,
      "step": 956
    },
    {
      "epoch": 0.513808261467736,
      "grad_norm": 0.68359375,
      "learning_rate": 4.873165632801338e-05,
      "loss": 0.9229,
      "step": 957
    },
    {
      "epoch": 0.514345156202812,
      "grad_norm": 0.6796875,
      "learning_rate": 4.87289992008243e-05,
      "loss": 0.8132,
      "step": 958
    },
    {
      "epoch": 0.514882050937888,
      "grad_norm": 0.7421875,
      "learning_rate": 4.872633936585591e-05,
      "loss": 1.0589,
      "step": 959
    },
    {
      "epoch": 0.515418945672964,
      "grad_norm": 0.66796875,
      "learning_rate": 4.872367682341173e-05,
      "loss": 0.9771,
      "step": 960
    },
    {
      "epoch": 0.51595584040804,
      "grad_norm": 0.7265625,
      "learning_rate": 4.872101157379558e-05,
      "loss": 1.1707,
      "step": 961
    },
    {
      "epoch": 0.516492735143116,
      "grad_norm": 0.7109375,
      "learning_rate": 4.871834361731162e-05,
      "loss": 0.9431,
      "step": 962
    },
    {
      "epoch": 0.517029629878192,
      "grad_norm": 0.71875,
      "learning_rate": 4.871567295426427e-05,
      "loss": 0.894,
      "step": 963
    },
    {
      "epoch": 0.517566524613268,
      "grad_norm": 0.8359375,
      "learning_rate": 4.8712999584958314e-05,
      "loss": 1.1675,
      "step": 964
    },
    {
      "epoch": 0.518103419348344,
      "grad_norm": 0.6484375,
      "learning_rate": 4.87103235096988e-05,
      "loss": 0.8444,
      "step": 965
    },
    {
      "epoch": 0.51864031408342,
      "grad_norm": 0.6640625,
      "learning_rate": 4.87076447287911e-05,
      "loss": 0.7976,
      "step": 966
    },
    {
      "epoch": 0.519177208818496,
      "grad_norm": 0.68359375,
      "learning_rate": 4.870496324254091e-05,
      "loss": 0.9452,
      "step": 967
    },
    {
      "epoch": 0.5197141035535721,
      "grad_norm": 0.7578125,
      "learning_rate": 4.870227905125422e-05,
      "loss": 0.9365,
      "step": 968
    },
    {
      "epoch": 0.5202509982886481,
      "grad_norm": 0.71875,
      "learning_rate": 4.8699592155237316e-05,
      "loss": 0.9319,
      "step": 969
    },
    {
      "epoch": 0.5207878930237241,
      "grad_norm": 0.796875,
      "learning_rate": 4.869690255479682e-05,
      "loss": 0.9302,
      "step": 970
    },
    {
      "epoch": 0.5213247877588001,
      "grad_norm": 0.63671875,
      "learning_rate": 4.869421025023965e-05,
      "loss": 0.8786,
      "step": 971
    },
    {
      "epoch": 0.521861682493876,
      "grad_norm": 1.203125,
      "learning_rate": 4.8691515241873023e-05,
      "loss": 0.8714,
      "step": 972
    },
    {
      "epoch": 0.522398577228952,
      "grad_norm": 0.6328125,
      "learning_rate": 4.8688817530004473e-05,
      "loss": 0.9017,
      "step": 973
    },
    {
      "epoch": 0.522935471964028,
      "grad_norm": 0.94140625,
      "learning_rate": 4.868611711494186e-05,
      "loss": 1.1617,
      "step": 974
    },
    {
      "epoch": 0.523472366699104,
      "grad_norm": 0.71484375,
      "learning_rate": 4.868341399699331e-05,
      "loss": 0.9164,
      "step": 975
    },
    {
      "epoch": 0.52400926143418,
      "grad_norm": 0.71875,
      "learning_rate": 4.8680708176467305e-05,
      "loss": 0.9218,
      "step": 976
    },
    {
      "epoch": 0.524546156169256,
      "grad_norm": 0.890625,
      "learning_rate": 4.86779996536726e-05,
      "loss": 1.0642,
      "step": 977
    },
    {
      "epoch": 0.525083050904332,
      "grad_norm": 0.6328125,
      "learning_rate": 4.867528842891828e-05,
      "loss": 0.9238,
      "step": 978
    },
    {
      "epoch": 0.5256199456394081,
      "grad_norm": 0.65625,
      "learning_rate": 4.867257450251373e-05,
      "loss": 0.9341,
      "step": 979
    },
    {
      "epoch": 0.5261568403744841,
      "grad_norm": 1.0078125,
      "learning_rate": 4.866985787476863e-05,
      "loss": 1.1306,
      "step": 980
    },
    {
      "epoch": 0.5266937351095601,
      "grad_norm": 0.77734375,
      "learning_rate": 4.866713854599301e-05,
      "loss": 0.9636,
      "step": 981
    },
    {
      "epoch": 0.5272306298446361,
      "grad_norm": 0.7890625,
      "learning_rate": 4.866441651649715e-05,
      "loss": 1.0581,
      "step": 982
    },
    {
      "epoch": 0.5277675245797121,
      "grad_norm": 0.96875,
      "learning_rate": 4.866169178659168e-05,
      "loss": 0.9646,
      "step": 983
    },
    {
      "epoch": 0.5283044193147881,
      "grad_norm": 0.7421875,
      "learning_rate": 4.865896435658752e-05,
      "loss": 0.9056,
      "step": 984
    },
    {
      "epoch": 0.5288413140498641,
      "grad_norm": 0.80078125,
      "learning_rate": 4.865623422679593e-05,
      "loss": 1.0613,
      "step": 985
    },
    {
      "epoch": 0.5293782087849401,
      "grad_norm": 0.78125,
      "learning_rate": 4.865350139752841e-05,
      "loss": 0.8856,
      "step": 986
    },
    {
      "epoch": 0.5299151035200161,
      "grad_norm": 0.84765625,
      "learning_rate": 4.865076586909685e-05,
      "loss": 1.0529,
      "step": 987
    },
    {
      "epoch": 0.5304519982550921,
      "grad_norm": 0.6875,
      "learning_rate": 4.8648027641813384e-05,
      "loss": 1.0098,
      "step": 988
    },
    {
      "epoch": 0.5309888929901682,
      "grad_norm": 0.77734375,
      "learning_rate": 4.864528671599049e-05,
      "loss": 1.0291,
      "step": 989
    },
    {
      "epoch": 0.5315257877252442,
      "grad_norm": 0.65625,
      "learning_rate": 4.864254309194093e-05,
      "loss": 0.7555,
      "step": 990
    },
    {
      "epoch": 0.5320626824603201,
      "grad_norm": 0.71875,
      "learning_rate": 4.86397967699778e-05,
      "loss": 1.0176,
      "step": 991
    },
    {
      "epoch": 0.5325995771953961,
      "grad_norm": 0.66015625,
      "learning_rate": 4.863704775041449e-05,
      "loss": 0.9002,
      "step": 992
    },
    {
      "epoch": 0.5331364719304721,
      "grad_norm": 0.6640625,
      "learning_rate": 4.863429603356469e-05,
      "loss": 0.9122,
      "step": 993
    },
    {
      "epoch": 0.5336733666655481,
      "grad_norm": 0.765625,
      "learning_rate": 4.8631541619742405e-05,
      "loss": 0.9334,
      "step": 994
    },
    {
      "epoch": 0.5342102614006241,
      "grad_norm": 0.71875,
      "learning_rate": 4.862878450926195e-05,
      "loss": 0.7841,
      "step": 995
    },
    {
      "epoch": 0.5347471561357001,
      "grad_norm": 0.70703125,
      "learning_rate": 4.8626024702437956e-05,
      "loss": 0.958,
      "step": 996
    },
    {
      "epoch": 0.5352840508707761,
      "grad_norm": 0.76171875,
      "learning_rate": 4.862326219958533e-05,
      "loss": 0.9678,
      "step": 997
    },
    {
      "epoch": 0.5358209456058521,
      "grad_norm": 0.71484375,
      "learning_rate": 4.862049700101934e-05,
      "loss": 0.894,
      "step": 998
    },
    {
      "epoch": 0.5363578403409281,
      "grad_norm": 0.5625,
      "learning_rate": 4.86177291070555e-05,
      "loss": 0.771,
      "step": 999
    },
    {
      "epoch": 0.5368947350760042,
      "grad_norm": 0.71484375,
      "learning_rate": 4.861495851800968e-05,
      "loss": 0.8994,
      "step": 1000
    },
    {
      "epoch": 0.5368947350760042,
      "eval_loss": 0.942378580570221,
      "eval_runtime": 46.24,
      "eval_samples_per_second": 20.804,
      "eval_steps_per_second": 20.804,
      "step": 1000
    },
    {
      "epoch": 0.5374316298110802,
      "grad_norm": 0.7109375,
      "learning_rate": 4.861218523419804e-05,
      "loss": 0.9404,
      "step": 1001
    },
    {
      "epoch": 0.5379685245461562,
      "grad_norm": 0.66015625,
      "learning_rate": 4.860940925593703e-05,
      "loss": 0.7577,
      "step": 1002
    },
    {
      "epoch": 0.5385054192812322,
      "grad_norm": 0.7265625,
      "learning_rate": 4.860663058354344e-05,
      "loss": 0.9183,
      "step": 1003
    },
    {
      "epoch": 0.5390423140163082,
      "grad_norm": 0.96875,
      "learning_rate": 4.860384921733434e-05,
      "loss": 1.0715,
      "step": 1004
    },
    {
      "epoch": 0.5395792087513842,
      "grad_norm": 0.6484375,
      "learning_rate": 4.8601065157627135e-05,
      "loss": 0.904,
      "step": 1005
    },
    {
      "epoch": 0.5401161034864602,
      "grad_norm": 0.8984375,
      "learning_rate": 4.8598278404739507e-05,
      "loss": 1.0282,
      "step": 1006
    },
    {
      "epoch": 0.5406529982215362,
      "grad_norm": 0.671875,
      "learning_rate": 4.8595488958989466e-05,
      "loss": 0.8043,
      "step": 1007
    },
    {
      "epoch": 0.5411898929566122,
      "grad_norm": 0.9609375,
      "learning_rate": 4.859269682069533e-05,
      "loss": 0.8917,
      "step": 1008
    },
    {
      "epoch": 0.5417267876916882,
      "grad_norm": 0.828125,
      "learning_rate": 4.85899019901757e-05,
      "loss": 1.1733,
      "step": 1009
    },
    {
      "epoch": 0.5422636824267641,
      "grad_norm": 1.015625,
      "learning_rate": 4.858710446774951e-05,
      "loss": 0.9036,
      "step": 1010
    },
    {
      "epoch": 0.5428005771618402,
      "grad_norm": 0.71484375,
      "learning_rate": 4.858430425373599e-05,
      "loss": 0.9303,
      "step": 1011
    },
    {
      "epoch": 0.5433374718969162,
      "grad_norm": 0.7265625,
      "learning_rate": 4.8581501348454696e-05,
      "loss": 0.924,
      "step": 1012
    },
    {
      "epoch": 0.5438743666319922,
      "grad_norm": 0.73828125,
      "learning_rate": 4.857869575222545e-05,
      "loss": 0.8699,
      "step": 1013
    },
    {
      "epoch": 0.5444112613670682,
      "grad_norm": 0.6171875,
      "learning_rate": 4.857588746536843e-05,
      "loss": 0.8638,
      "step": 1014
    },
    {
      "epoch": 0.5449481561021442,
      "grad_norm": 0.75,
      "learning_rate": 4.857307648820408e-05,
      "loss": 0.9615,
      "step": 1015
    },
    {
      "epoch": 0.5454850508372202,
      "grad_norm": 0.80078125,
      "learning_rate": 4.857026282105318e-05,
      "loss": 0.9276,
      "step": 1016
    },
    {
      "epoch": 0.5460219455722962,
      "grad_norm": 0.9375,
      "learning_rate": 4.856744646423679e-05,
      "loss": 1.3419,
      "step": 1017
    },
    {
      "epoch": 0.5465588403073722,
      "grad_norm": 0.75390625,
      "learning_rate": 4.8564627418076306e-05,
      "loss": 0.997,
      "step": 1018
    },
    {
      "epoch": 0.5470957350424482,
      "grad_norm": 0.78515625,
      "learning_rate": 4.856180568289341e-05,
      "loss": 1.0873,
      "step": 1019
    },
    {
      "epoch": 0.5476326297775242,
      "grad_norm": 0.67578125,
      "learning_rate": 4.85589812590101e-05,
      "loss": 0.9318,
      "step": 1020
    },
    {
      "epoch": 0.5481695245126003,
      "grad_norm": 0.8671875,
      "learning_rate": 4.855615414674868e-05,
      "loss": 0.9326,
      "step": 1021
    },
    {
      "epoch": 0.5487064192476763,
      "grad_norm": 0.96875,
      "learning_rate": 4.855332434643175e-05,
      "loss": 1.1083,
      "step": 1022
    },
    {
      "epoch": 0.5492433139827523,
      "grad_norm": 0.7890625,
      "learning_rate": 4.8550491858382244e-05,
      "loss": 1.0586,
      "step": 1023
    },
    {
      "epoch": 0.5497802087178283,
      "grad_norm": 0.640625,
      "learning_rate": 4.854765668292337e-05,
      "loss": 0.8821,
      "step": 1024
    },
    {
      "epoch": 0.5503171034529043,
      "grad_norm": 0.671875,
      "learning_rate": 4.8544818820378656e-05,
      "loss": 1.0563,
      "step": 1025
    },
    {
      "epoch": 0.5508539981879803,
      "grad_norm": 0.82421875,
      "learning_rate": 4.8541978271071946e-05,
      "loss": 0.8077,
      "step": 1026
    },
    {
      "epoch": 0.5513908929230563,
      "grad_norm": 0.77734375,
      "learning_rate": 4.8539135035327386e-05,
      "loss": 1.1503,
      "step": 1027
    },
    {
      "epoch": 0.5519277876581323,
      "grad_norm": 0.91796875,
      "learning_rate": 4.8536289113469416e-05,
      "loss": 0.9103,
      "step": 1028
    },
    {
      "epoch": 0.5524646823932082,
      "grad_norm": 0.65625,
      "learning_rate": 4.853344050582279e-05,
      "loss": 0.9062,
      "step": 1029
    },
    {
      "epoch": 0.5530015771282842,
      "grad_norm": 0.8828125,
      "learning_rate": 4.853058921271259e-05,
      "loss": 1.0392,
      "step": 1030
    },
    {
      "epoch": 0.5535384718633602,
      "grad_norm": 0.77734375,
      "learning_rate": 4.8527735234464156e-05,
      "loss": 1.0447,
      "step": 1031
    },
    {
      "epoch": 0.5540753665984363,
      "grad_norm": 0.78125,
      "learning_rate": 4.8524878571403175e-05,
      "loss": 0.7923,
      "step": 1032
    },
    {
      "epoch": 0.5546122613335123,
      "grad_norm": 0.796875,
      "learning_rate": 4.852201922385564e-05,
      "loss": 0.9418,
      "step": 1033
    },
    {
      "epoch": 0.5551491560685883,
      "grad_norm": 0.78515625,
      "learning_rate": 4.851915719214782e-05,
      "loss": 0.8726,
      "step": 1034
    },
    {
      "epoch": 0.5556860508036643,
      "grad_norm": 1.109375,
      "learning_rate": 4.851629247660633e-05,
      "loss": 1.1544,
      "step": 1035
    },
    {
      "epoch": 0.5562229455387403,
      "grad_norm": 0.8984375,
      "learning_rate": 4.851342507755805e-05,
      "loss": 1.1065,
      "step": 1036
    },
    {
      "epoch": 0.5567598402738163,
      "grad_norm": 0.7265625,
      "learning_rate": 4.8510554995330205e-05,
      "loss": 1.0341,
      "step": 1037
    },
    {
      "epoch": 0.5572967350088923,
      "grad_norm": 0.69921875,
      "learning_rate": 4.850768223025028e-05,
      "loss": 0.9686,
      "step": 1038
    },
    {
      "epoch": 0.5578336297439683,
      "grad_norm": 0.87109375,
      "learning_rate": 4.850480678264613e-05,
      "loss": 1.014,
      "step": 1039
    },
    {
      "epoch": 0.5583705244790443,
      "grad_norm": 0.7109375,
      "learning_rate": 4.8501928652845854e-05,
      "loss": 0.9053,
      "step": 1040
    },
    {
      "epoch": 0.5589074192141203,
      "grad_norm": 0.875,
      "learning_rate": 4.849904784117789e-05,
      "loss": 1.1054,
      "step": 1041
    },
    {
      "epoch": 0.5594443139491964,
      "grad_norm": 0.6796875,
      "learning_rate": 4.849616434797098e-05,
      "loss": 0.861,
      "step": 1042
    },
    {
      "epoch": 0.5599812086842724,
      "grad_norm": 0.83984375,
      "learning_rate": 4.8493278173554156e-05,
      "loss": 0.9483,
      "step": 1043
    },
    {
      "epoch": 0.5605181034193484,
      "grad_norm": 0.7421875,
      "learning_rate": 4.8490389318256776e-05,
      "loss": 0.7822,
      "step": 1044
    },
    {
      "epoch": 0.5610549981544244,
      "grad_norm": 0.79296875,
      "learning_rate": 4.8487497782408506e-05,
      "loss": 1.0347,
      "step": 1045
    },
    {
      "epoch": 0.5615918928895004,
      "grad_norm": 0.58984375,
      "learning_rate": 4.848460356633928e-05,
      "loss": 0.9687,
      "step": 1046
    },
    {
      "epoch": 0.5621287876245764,
      "grad_norm": 0.73828125,
      "learning_rate": 4.848170667037938e-05,
      "loss": 1.0925,
      "step": 1047
    },
    {
      "epoch": 0.5626656823596523,
      "grad_norm": 0.8515625,
      "learning_rate": 4.847880709485938e-05,
      "loss": 0.9552,
      "step": 1048
    },
    {
      "epoch": 0.5632025770947283,
      "grad_norm": 0.69921875,
      "learning_rate": 4.847590484011015e-05,
      "loss": 1.0899,
      "step": 1049
    },
    {
      "epoch": 0.5637394718298043,
      "grad_norm": 0.89453125,
      "learning_rate": 4.847299990646289e-05,
      "loss": 1.0149,
      "step": 1050
    },
    {
      "epoch": 0.5637394718298043,
      "eval_loss": 0.9401158690452576,
      "eval_runtime": 46.0879,
      "eval_samples_per_second": 20.873,
      "eval_steps_per_second": 20.873,
      "step": 1050
    },
    {
      "epoch": 0.5642763665648803,
      "grad_norm": 0.82421875,
      "learning_rate": 4.8470092294249075e-05,
      "loss": 0.9582,
      "step": 1051
    },
    {
      "epoch": 0.5648132612999563,
      "grad_norm": 0.83984375,
      "learning_rate": 4.84671820038005e-05,
      "loss": 0.7745,
      "step": 1052
    },
    {
      "epoch": 0.5653501560350324,
      "grad_norm": 0.73828125,
      "learning_rate": 4.8464269035449275e-05,
      "loss": 1.0579,
      "step": 1053
    },
    {
      "epoch": 0.5658870507701084,
      "grad_norm": 0.59375,
      "learning_rate": 4.84613533895278e-05,
      "loss": 0.893,
      "step": 1054
    },
    {
      "epoch": 0.5664239455051844,
      "grad_norm": 1.0625,
      "learning_rate": 4.845843506636879e-05,
      "loss": 1.0196,
      "step": 1055
    },
    {
      "epoch": 0.5669608402402604,
      "grad_norm": 0.73828125,
      "learning_rate": 4.845551406630526e-05,
      "loss": 0.968,
      "step": 1056
    },
    {
      "epoch": 0.5674977349753364,
      "grad_norm": 0.6328125,
      "learning_rate": 4.845259038967054e-05,
      "loss": 0.9198,
      "step": 1057
    },
    {
      "epoch": 0.5680346297104124,
      "grad_norm": 0.70703125,
      "learning_rate": 4.844966403679825e-05,
      "loss": 0.9292,
      "step": 1058
    },
    {
      "epoch": 0.5685715244454884,
      "grad_norm": 0.703125,
      "learning_rate": 4.8446735008022324e-05,
      "loss": 0.9598,
      "step": 1059
    },
    {
      "epoch": 0.5691084191805644,
      "grad_norm": 0.875,
      "learning_rate": 4.844380330367701e-05,
      "loss": 1.2437,
      "step": 1060
    },
    {
      "epoch": 0.5696453139156404,
      "grad_norm": 0.7265625,
      "learning_rate": 4.844086892409685e-05,
      "loss": 0.8978,
      "step": 1061
    },
    {
      "epoch": 0.5701822086507164,
      "grad_norm": 1.5546875,
      "learning_rate": 4.8437931869616686e-05,
      "loss": 0.9245,
      "step": 1062
    },
    {
      "epoch": 0.5707191033857925,
      "grad_norm": 0.64453125,
      "learning_rate": 4.8434992140571675e-05,
      "loss": 0.8795,
      "step": 1063
    },
    {
      "epoch": 0.5712559981208685,
      "grad_norm": 0.86328125,
      "learning_rate": 4.843204973729729e-05,
      "loss": 1.1411,
      "step": 1064
    },
    {
      "epoch": 0.5717928928559445,
      "grad_norm": 0.76171875,
      "learning_rate": 4.8429104660129284e-05,
      "loss": 1.0165,
      "step": 1065
    },
    {
      "epoch": 0.5723297875910205,
      "grad_norm": 0.77734375,
      "learning_rate": 4.842615690940373e-05,
      "loss": 0.9937,
      "step": 1066
    },
    {
      "epoch": 0.5728666823260964,
      "grad_norm": 0.67578125,
      "learning_rate": 4.8423206485456995e-05,
      "loss": 0.9006,
      "step": 1067
    },
    {
      "epoch": 0.5734035770611724,
      "grad_norm": 0.953125,
      "learning_rate": 4.842025338862578e-05,
      "loss": 1.0157,
      "step": 1068
    },
    {
      "epoch": 0.5739404717962484,
      "grad_norm": 0.828125,
      "learning_rate": 4.841729761924706e-05,
      "loss": 0.8855,
      "step": 1069
    },
    {
      "epoch": 0.5744773665313244,
      "grad_norm": 0.9296875,
      "learning_rate": 4.841433917765812e-05,
      "loss": 0.9902,
      "step": 1070
    },
    {
      "epoch": 0.5750142612664004,
      "grad_norm": 0.78125,
      "learning_rate": 4.841137806419657e-05,
      "loss": 0.9989,
      "step": 1071
    },
    {
      "epoch": 0.5755511560014764,
      "grad_norm": 0.85546875,
      "learning_rate": 4.840841427920029e-05,
      "loss": 1.1161,
      "step": 1072
    },
    {
      "epoch": 0.5760880507365524,
      "grad_norm": 0.703125,
      "learning_rate": 4.84054478230075e-05,
      "loss": 0.852,
      "step": 1073
    },
    {
      "epoch": 0.5766249454716285,
      "grad_norm": 0.7578125,
      "learning_rate": 4.840247869595671e-05,
      "loss": 1.0496,
      "step": 1074
    },
    {
      "epoch": 0.5771618402067045,
      "grad_norm": 0.8125,
      "learning_rate": 4.839950689838674e-05,
      "loss": 1.0162,
      "step": 1075
    },
    {
      "epoch": 0.5776987349417805,
      "grad_norm": 0.8046875,
      "learning_rate": 4.8396532430636684e-05,
      "loss": 0.757,
      "step": 1076
    },
    {
      "epoch": 0.5782356296768565,
      "grad_norm": 0.82421875,
      "learning_rate": 4.8393555293046e-05,
      "loss": 0.9283,
      "step": 1077
    },
    {
      "epoch": 0.5787725244119325,
      "grad_norm": 0.7421875,
      "learning_rate": 4.83905754859544e-05,
      "loss": 1.0616,
      "step": 1078
    },
    {
      "epoch": 0.5793094191470085,
      "grad_norm": 0.82421875,
      "learning_rate": 4.8387593009701914e-05,
      "loss": 1.1749,
      "step": 1079
    },
    {
      "epoch": 0.5798463138820845,
      "grad_norm": 0.84765625,
      "learning_rate": 4.838460786462888e-05,
      "loss": 1.1078,
      "step": 1080
    },
    {
      "epoch": 0.5803832086171605,
      "grad_norm": 0.76171875,
      "learning_rate": 4.838162005107596e-05,
      "loss": 0.7911,
      "step": 1081
    },
    {
      "epoch": 0.5809201033522365,
      "grad_norm": 0.609375,
      "learning_rate": 4.837862956938408e-05,
      "loss": 0.8446,
      "step": 1082
    },
    {
      "epoch": 0.5814569980873125,
      "grad_norm": 0.6875,
      "learning_rate": 4.837563641989451e-05,
      "loss": 1.1396,
      "step": 1083
    },
    {
      "epoch": 0.5819938928223886,
      "grad_norm": 0.73046875,
      "learning_rate": 4.837264060294878e-05,
      "loss": 1.0937,
      "step": 1084
    },
    {
      "epoch": 0.5825307875574646,
      "grad_norm": 1.0546875,
      "learning_rate": 4.836964211888878e-05,
      "loss": 1.1845,
      "step": 1085
    },
    {
      "epoch": 0.5830676822925405,
      "grad_norm": 0.6015625,
      "learning_rate": 4.836664096805665e-05,
      "loss": 0.9594,
      "step": 1086
    },
    {
      "epoch": 0.5836045770276165,
      "grad_norm": 0.7265625,
      "learning_rate": 4.8363637150794884e-05,
      "loss": 1.0269,
      "step": 1087
    },
    {
      "epoch": 0.5841414717626925,
      "grad_norm": 0.71484375,
      "learning_rate": 4.8360630667446236e-05,
      "loss": 0.9155,
      "step": 1088
    },
    {
      "epoch": 0.5846783664977685,
      "grad_norm": 0.70703125,
      "learning_rate": 4.835762151835379e-05,
      "loss": 0.8554,
      "step": 1089
    },
    {
      "epoch": 0.5852152612328445,
      "grad_norm": 0.74609375,
      "learning_rate": 4.835460970386093e-05,
      "loss": 0.856,
      "step": 1090
    },
    {
      "epoch": 0.5857521559679205,
      "grad_norm": 0.7265625,
      "learning_rate": 4.835159522431134e-05,
      "loss": 0.8173,
      "step": 1091
    },
    {
      "epoch": 0.5862890507029965,
      "grad_norm": 0.66796875,
      "learning_rate": 4.8348578080049004e-05,
      "loss": 1.0669,
      "step": 1092
    },
    {
      "epoch": 0.5868259454380725,
      "grad_norm": 0.75390625,
      "learning_rate": 4.8345558271418235e-05,
      "loss": 0.987,
      "step": 1093
    },
    {
      "epoch": 0.5873628401731485,
      "grad_norm": 0.68359375,
      "learning_rate": 4.834253579876361e-05,
      "loss": 0.8267,
      "step": 1094
    },
    {
      "epoch": 0.5878997349082246,
      "grad_norm": 1.0390625,
      "learning_rate": 4.8339510662430046e-05,
      "loss": 0.9502,
      "step": 1095
    },
    {
      "epoch": 0.5884366296433006,
      "grad_norm": 0.671875,
      "learning_rate": 4.833648286276275e-05,
      "loss": 0.9575,
      "step": 1096
    },
    {
      "epoch": 0.5889735243783766,
      "grad_norm": 0.81640625,
      "learning_rate": 4.8333452400107217e-05,
      "loss": 1.1301,
      "step": 1097
    },
    {
      "epoch": 0.5895104191134526,
      "grad_norm": 0.83203125,
      "learning_rate": 4.833041927480927e-05,
      "loss": 1.1176,
      "step": 1098
    },
    {
      "epoch": 0.5900473138485286,
      "grad_norm": 0.73046875,
      "learning_rate": 4.832738348721503e-05,
      "loss": 0.8473,
      "step": 1099
    },
    {
      "epoch": 0.5905842085836046,
      "grad_norm": 0.76953125,
      "learning_rate": 4.832434503767092e-05,
      "loss": 0.9467,
      "step": 1100
    },
    {
      "epoch": 0.5905842085836046,
      "eval_loss": 0.9381348490715027,
      "eval_runtime": 46.4802,
      "eval_samples_per_second": 20.697,
      "eval_steps_per_second": 20.697,
      "step": 1100
    },
    {
      "epoch": 0.5911211033186806,
      "grad_norm": 0.6640625,
      "learning_rate": 4.8321303926523665e-05,
      "loss": 1.0117,
      "step": 1101
    },
    {
      "epoch": 0.5916579980537566,
      "grad_norm": 0.75,
      "learning_rate": 4.8318260154120286e-05,
      "loss": 1.2182,
      "step": 1102
    },
    {
      "epoch": 0.5921948927888326,
      "grad_norm": 0.765625,
      "learning_rate": 4.831521372080812e-05,
      "loss": 1.0126,
      "step": 1103
    },
    {
      "epoch": 0.5927317875239086,
      "grad_norm": 0.70703125,
      "learning_rate": 4.831216462693481e-05,
      "loss": 0.8073,
      "step": 1104
    },
    {
      "epoch": 0.5932686822589845,
      "grad_norm": 0.91796875,
      "learning_rate": 4.8309112872848296e-05,
      "loss": 0.9932,
      "step": 1105
    },
    {
      "epoch": 0.5938055769940606,
      "grad_norm": 0.8515625,
      "learning_rate": 4.830605845889681e-05,
      "loss": 1.0643,
      "step": 1106
    },
    {
      "epoch": 0.5943424717291366,
      "grad_norm": 0.5859375,
      "learning_rate": 4.83030013854289e-05,
      "loss": 0.8294,
      "step": 1107
    },
    {
      "epoch": 0.5948793664642126,
      "grad_norm": 0.703125,
      "learning_rate": 4.829994165279343e-05,
      "loss": 0.8283,
      "step": 1108
    },
    {
      "epoch": 0.5954162611992886,
      "grad_norm": 0.63671875,
      "learning_rate": 4.829687926133955e-05,
      "loss": 0.9631,
      "step": 1109
    },
    {
      "epoch": 0.5959531559343646,
      "grad_norm": 0.6875,
      "learning_rate": 4.829381421141671e-05,
      "loss": 0.8741,
      "step": 1110
    },
    {
      "epoch": 0.5964900506694406,
      "grad_norm": 0.83203125,
      "learning_rate": 4.8290746503374687e-05,
      "loss": 1.0303,
      "step": 1111
    },
    {
      "epoch": 0.5970269454045166,
      "grad_norm": 0.76171875,
      "learning_rate": 4.828767613756352e-05,
      "loss": 0.9702,
      "step": 1112
    },
    {
      "epoch": 0.5975638401395926,
      "grad_norm": 0.84375,
      "learning_rate": 4.82846031143336e-05,
      "loss": 0.9661,
      "step": 1113
    },
    {
      "epoch": 0.5981007348746686,
      "grad_norm": 0.859375,
      "learning_rate": 4.828152743403559e-05,
      "loss": 1.0958,
      "step": 1114
    },
    {
      "epoch": 0.5986376296097446,
      "grad_norm": 0.82421875,
      "learning_rate": 4.8278449097020464e-05,
      "loss": 1.0832,
      "step": 1115
    },
    {
      "epoch": 0.5991745243448207,
      "grad_norm": 0.7421875,
      "learning_rate": 4.827536810363949e-05,
      "loss": 1.1014,
      "step": 1116
    },
    {
      "epoch": 0.5997114190798967,
      "grad_norm": 0.65234375,
      "learning_rate": 4.827228445424426e-05,
      "loss": 0.9986,
      "step": 1117
    },
    {
      "epoch": 0.6002483138149727,
      "grad_norm": 0.765625,
      "learning_rate": 4.8269198149186656e-05,
      "loss": 1.0819,
      "step": 1118
    },
    {
      "epoch": 0.6007852085500487,
      "grad_norm": 0.6875,
      "learning_rate": 4.826610918881886e-05,
      "loss": 0.9433,
      "step": 1119
    },
    {
      "epoch": 0.6013221032851247,
      "grad_norm": 0.6484375,
      "learning_rate": 4.826301757349337e-05,
      "loss": 1.0211,
      "step": 1120
    },
    {
      "epoch": 0.6018589980202007,
      "grad_norm": 0.6875,
      "learning_rate": 4.825992330356297e-05,
      "loss": 0.9769,
      "step": 1121
    },
    {
      "epoch": 0.6023958927552767,
      "grad_norm": 0.6484375,
      "learning_rate": 4.825682637938075e-05,
      "loss": 0.8121,
      "step": 1122
    },
    {
      "epoch": 0.6029327874903526,
      "grad_norm": 0.984375,
      "learning_rate": 4.825372680130013e-05,
      "loss": 1.0143,
      "step": 1123
    },
    {
      "epoch": 0.6034696822254286,
      "grad_norm": 0.765625,
      "learning_rate": 4.825062456967479e-05,
      "loss": 1.0155,
      "step": 1124
    },
    {
      "epoch": 0.6040065769605046,
      "grad_norm": 0.7734375,
      "learning_rate": 4.824751968485874e-05,
      "loss": 0.937,
      "step": 1125
    },
    {
      "epoch": 0.6045434716955806,
      "grad_norm": 0.6953125,
      "learning_rate": 4.8244412147206284e-05,
      "loss": 0.928,
      "step": 1126
    },
    {
      "epoch": 0.6050803664306567,
      "grad_norm": 0.65625,
      "learning_rate": 4.8241301957072046e-05,
      "loss": 0.8057,
      "step": 1127
    },
    {
      "epoch": 0.6056172611657327,
      "grad_norm": 0.83984375,
      "learning_rate": 4.8238189114810916e-05,
      "loss": 0.8886,
      "step": 1128
    },
    {
      "epoch": 0.6061541559008087,
      "grad_norm": 0.66015625,
      "learning_rate": 4.823507362077813e-05,
      "loss": 0.7433,
      "step": 1129
    },
    {
      "epoch": 0.6066910506358847,
      "grad_norm": 0.76953125,
      "learning_rate": 4.823195547532919e-05,
      "loss": 0.9903,
      "step": 1130
    },
    {
      "epoch": 0.6072279453709607,
      "grad_norm": 0.5859375,
      "learning_rate": 4.822883467881992e-05,
      "loss": 0.9196,
      "step": 1131
    },
    {
      "epoch": 0.6077648401060367,
      "grad_norm": 0.82421875,
      "learning_rate": 4.822571123160645e-05,
      "loss": 1.0092,
      "step": 1132
    },
    {
      "epoch": 0.6083017348411127,
      "grad_norm": 0.71484375,
      "learning_rate": 4.8222585134045194e-05,
      "loss": 0.9917,
      "step": 1133
    },
    {
      "epoch": 0.6088386295761887,
      "grad_norm": 0.921875,
      "learning_rate": 4.821945638649289e-05,
      "loss": 0.942,
      "step": 1134
    },
    {
      "epoch": 0.6093755243112647,
      "grad_norm": 0.8359375,
      "learning_rate": 4.821632498930656e-05,
      "loss": 0.9521,
      "step": 1135
    },
    {
      "epoch": 0.6099124190463407,
      "grad_norm": 0.95703125,
      "learning_rate": 4.8213190942843535e-05,
      "loss": 1.0247,
      "step": 1136
    },
    {
      "epoch": 0.6104493137814168,
      "grad_norm": 0.671875,
      "learning_rate": 4.821005424746145e-05,
      "loss": 0.8676,
      "step": 1137
    },
    {
      "epoch": 0.6109862085164928,
      "grad_norm": 0.8203125,
      "learning_rate": 4.820691490351824e-05,
      "loss": 1.0077,
      "step": 1138
    },
    {
      "epoch": 0.6115231032515688,
      "grad_norm": 0.78515625,
      "learning_rate": 4.820377291137216e-05,
      "loss": 0.8987,
      "step": 1139
    },
    {
      "epoch": 0.6120599979866448,
      "grad_norm": 0.93359375,
      "learning_rate": 4.820062827138173e-05,
      "loss": 1.058,
      "step": 1140
    },
    {
      "epoch": 0.6125968927217208,
      "grad_norm": 0.71875,
      "learning_rate": 4.819748098390581e-05,
      "loss": 1.0804,
      "step": 1141
    },
    {
      "epoch": 0.6131337874567967,
      "grad_norm": 0.5859375,
      "learning_rate": 4.819433104930352e-05,
      "loss": 0.7626,
      "step": 1142
    },
    {
      "epoch": 0.6136706821918727,
      "grad_norm": 0.7421875,
      "learning_rate": 4.8191178467934346e-05,
      "loss": 1.0532,
      "step": 1143
    },
    {
      "epoch": 0.6142075769269487,
      "grad_norm": 0.82421875,
      "learning_rate": 4.818802324015801e-05,
      "loss": 0.9695,
      "step": 1144
    },
    {
      "epoch": 0.6147444716620247,
      "grad_norm": 0.69921875,
      "learning_rate": 4.8184865366334564e-05,
      "loss": 0.8658,
      "step": 1145
    },
    {
      "epoch": 0.6152813663971007,
      "grad_norm": 0.765625,
      "learning_rate": 4.818170484682437e-05,
      "loss": 0.826,
      "step": 1146
    },
    {
      "epoch": 0.6158182611321767,
      "grad_norm": 0.609375,
      "learning_rate": 4.8178541681988074e-05,
      "loss": 0.7409,
      "step": 1147
    },
    {
      "epoch": 0.6163551558672528,
      "grad_norm": 0.66796875,
      "learning_rate": 4.817537587218665e-05,
      "loss": 0.9064,
      "step": 1148
    },
    {
      "epoch": 0.6168920506023288,
      "grad_norm": 0.7734375,
      "learning_rate": 4.817220741778135e-05,
      "loss": 0.8841,
      "step": 1149
    },
    {
      "epoch": 0.6174289453374048,
      "grad_norm": 0.71875,
      "learning_rate": 4.816903631913372e-05,
      "loss": 1.0278,
      "step": 1150
    },
    {
      "epoch": 0.6174289453374048,
      "eval_loss": 0.9374992251396179,
      "eval_runtime": 46.3897,
      "eval_samples_per_second": 20.737,
      "eval_steps_per_second": 20.737,
      "step": 1150
    },
    {
      "epoch": 0.6179658400724808,
      "grad_norm": 0.9140625,
      "learning_rate": 4.816586257660564e-05,
      "loss": 0.963,
      "step": 1151
    },
    {
      "epoch": 0.6185027348075568,
      "grad_norm": 0.90234375,
      "learning_rate": 4.816268619055926e-05,
      "loss": 1.2144,
      "step": 1152
    },
    {
      "epoch": 0.6190396295426328,
      "grad_norm": 0.734375,
      "learning_rate": 4.815950716135706e-05,
      "loss": 1.0242,
      "step": 1153
    },
    {
      "epoch": 0.6195765242777088,
      "grad_norm": 0.75390625,
      "learning_rate": 4.81563254893618e-05,
      "loss": 0.9588,
      "step": 1154
    },
    {
      "epoch": 0.6201134190127848,
      "grad_norm": 0.8046875,
      "learning_rate": 4.815314117493655e-05,
      "loss": 0.6064,
      "step": 1155
    },
    {
      "epoch": 0.6206503137478608,
      "grad_norm": 0.8828125,
      "learning_rate": 4.814995421844468e-05,
      "loss": 1.1407,
      "step": 1156
    },
    {
      "epoch": 0.6211872084829368,
      "grad_norm": 0.8046875,
      "learning_rate": 4.814676462024988e-05,
      "loss": 0.9002,
      "step": 1157
    },
    {
      "epoch": 0.6217241032180129,
      "grad_norm": 0.73046875,
      "learning_rate": 4.814357238071608e-05,
      "loss": 0.9564,
      "step": 1158
    },
    {
      "epoch": 0.6222609979530889,
      "grad_norm": 0.64453125,
      "learning_rate": 4.8140377500207604e-05,
      "loss": 0.693,
      "step": 1159
    },
    {
      "epoch": 0.6227978926881649,
      "grad_norm": 0.9140625,
      "learning_rate": 4.8137179979088995e-05,
      "loss": 0.941,
      "step": 1160
    },
    {
      "epoch": 0.6233347874232408,
      "grad_norm": 0.73828125,
      "learning_rate": 4.813397981772514e-05,
      "loss": 0.9724,
      "step": 1161
    },
    {
      "epoch": 0.6238716821583168,
      "grad_norm": 0.734375,
      "learning_rate": 4.813077701648122e-05,
      "loss": 0.961,
      "step": 1162
    },
    {
      "epoch": 0.6244085768933928,
      "grad_norm": 0.62890625,
      "learning_rate": 4.812757157572272e-05,
      "loss": 0.815,
      "step": 1163
    },
    {
      "epoch": 0.6249454716284688,
      "grad_norm": 0.72265625,
      "learning_rate": 4.812436349581542e-05,
      "loss": 0.9533,
      "step": 1164
    },
    {
      "epoch": 0.6254823663635448,
      "grad_norm": 0.74609375,
      "learning_rate": 4.812115277712539e-05,
      "loss": 1.0351,
      "step": 1165
    },
    {
      "epoch": 0.6260192610986208,
      "grad_norm": 0.76953125,
      "learning_rate": 4.811793942001902e-05,
      "loss": 0.9545,
      "step": 1166
    },
    {
      "epoch": 0.6265561558336968,
      "grad_norm": 0.60546875,
      "learning_rate": 4.8114723424863003e-05,
      "loss": 0.8702,
      "step": 1167
    },
    {
      "epoch": 0.6270930505687728,
      "grad_norm": 0.80078125,
      "learning_rate": 4.811150479202432e-05,
      "loss": 1.1377,
      "step": 1168
    },
    {
      "epoch": 0.6276299453038489,
      "grad_norm": 0.76953125,
      "learning_rate": 4.8108283521870255e-05,
      "loss": 0.9175,
      "step": 1169
    },
    {
      "epoch": 0.6281668400389249,
      "grad_norm": 0.7578125,
      "learning_rate": 4.81050596147684e-05,
      "loss": 0.8968,
      "step": 1170
    },
    {
      "epoch": 0.6287037347740009,
      "grad_norm": 0.74609375,
      "learning_rate": 4.8101833071086635e-05,
      "loss": 1.1216,
      "step": 1171
    },
    {
      "epoch": 0.6292406295090769,
      "grad_norm": 0.859375,
      "learning_rate": 4.809860389119316e-05,
      "loss": 0.8454,
      "step": 1172
    },
    {
      "epoch": 0.6297775242441529,
      "grad_norm": 0.7734375,
      "learning_rate": 4.809537207545646e-05,
      "loss": 1.0855,
      "step": 1173
    },
    {
      "epoch": 0.6303144189792289,
      "grad_norm": 0.71875,
      "learning_rate": 4.8092137624245326e-05,
      "loss": 0.8737,
      "step": 1174
    },
    {
      "epoch": 0.6308513137143049,
      "grad_norm": 0.625,
      "learning_rate": 4.808890053792885e-05,
      "loss": 0.9678,
      "step": 1175
    },
    {
      "epoch": 0.6313882084493809,
      "grad_norm": 0.8125,
      "learning_rate": 4.808566081687643e-05,
      "loss": 0.9752,
      "step": 1176
    },
    {
      "epoch": 0.6319251031844569,
      "grad_norm": 0.80859375,
      "learning_rate": 4.808241846145775e-05,
      "loss": 1.0641,
      "step": 1177
    },
    {
      "epoch": 0.632461997919533,
      "grad_norm": 0.6640625,
      "learning_rate": 4.807917347204281e-05,
      "loss": 0.9785,
      "step": 1178
    },
    {
      "epoch": 0.632998892654609,
      "grad_norm": 0.7265625,
      "learning_rate": 4.80759258490019e-05,
      "loss": 0.9183,
      "step": 1179
    },
    {
      "epoch": 0.6335357873896849,
      "grad_norm": 0.703125,
      "learning_rate": 4.807267559270563e-05,
      "loss": 0.9138,
      "step": 1180
    },
    {
      "epoch": 0.6340726821247609,
      "grad_norm": 0.7265625,
      "learning_rate": 4.806942270352487e-05,
      "loss": 0.8897,
      "step": 1181
    },
    {
      "epoch": 0.6346095768598369,
      "grad_norm": 0.703125,
      "learning_rate": 4.8066167181830835e-05,
      "loss": 1.0239,
      "step": 1182
    },
    {
      "epoch": 0.6351464715949129,
      "grad_norm": 0.75,
      "learning_rate": 4.806290902799501e-05,
      "loss": 0.9348,
      "step": 1183
    },
    {
      "epoch": 0.6356833663299889,
      "grad_norm": 0.73046875,
      "learning_rate": 4.80596482423892e-05,
      "loss": 0.786,
      "step": 1184
    },
    {
      "epoch": 0.6362202610650649,
      "grad_norm": 0.609375,
      "learning_rate": 4.8056384825385495e-05,
      "loss": 0.7737,
      "step": 1185
    },
    {
      "epoch": 0.6367571558001409,
      "grad_norm": 0.66015625,
      "learning_rate": 4.80531187773563e-05,
      "loss": 0.8627,
      "step": 1186
    },
    {
      "epoch": 0.6372940505352169,
      "grad_norm": 0.66796875,
      "learning_rate": 4.804985009867431e-05,
      "loss": 0.9286,
      "step": 1187
    },
    {
      "epoch": 0.6378309452702929,
      "grad_norm": 0.6875,
      "learning_rate": 4.8046578789712515e-05,
      "loss": 0.9034,
      "step": 1188
    },
    {
      "epoch": 0.638367840005369,
      "grad_norm": 0.64453125,
      "learning_rate": 4.804330485084422e-05,
      "loss": 1.03,
      "step": 1189
    },
    {
      "epoch": 0.638904734740445,
      "grad_norm": 0.6875,
      "learning_rate": 4.8040028282443025e-05,
      "loss": 0.9326,
      "step": 1190
    },
    {
      "epoch": 0.639441629475521,
      "grad_norm": 0.82421875,
      "learning_rate": 4.8036749084882826e-05,
      "loss": 0.9014,
      "step": 1191
    },
    {
      "epoch": 0.639978524210597,
      "grad_norm": 0.6328125,
      "learning_rate": 4.8033467258537817e-05,
      "loss": 0.8208,
      "step": 1192
    },
    {
      "epoch": 0.640515418945673,
      "grad_norm": 0.734375,
      "learning_rate": 4.80301828037825e-05,
      "loss": 0.9107,
      "step": 1193
    },
    {
      "epoch": 0.641052313680749,
      "grad_norm": 0.7265625,
      "learning_rate": 4.802689572099167e-05,
      "loss": 1.1873,
      "step": 1194
    },
    {
      "epoch": 0.641589208415825,
      "grad_norm": 0.65234375,
      "learning_rate": 4.8023606010540426e-05,
      "loss": 1.0099,
      "step": 1195
    },
    {
      "epoch": 0.642126103150901,
      "grad_norm": 0.88671875,
      "learning_rate": 4.802031367280416e-05,
      "loss": 0.9254,
      "step": 1196
    },
    {
      "epoch": 0.642662997885977,
      "grad_norm": 0.7890625,
      "learning_rate": 4.801701870815859e-05,
      "loss": 1.2033,
      "step": 1197
    },
    {
      "epoch": 0.643199892621053,
      "grad_norm": 0.7421875,
      "learning_rate": 4.801372111697968e-05,
      "loss": 0.9532,
      "step": 1198
    },
    {
      "epoch": 0.6437367873561289,
      "grad_norm": 0.703125,
      "learning_rate": 4.801042089964376e-05,
      "loss": 0.8776,
      "step": 1199
    },
    {
      "epoch": 0.644273682091205,
      "grad_norm": 0.828125,
      "learning_rate": 4.800711805652741e-05,
      "loss": 0.9975,
      "step": 1200
    },
    {
      "epoch": 0.644273682091205,
      "eval_loss": 0.9368125200271606,
      "eval_runtime": 46.1689,
      "eval_samples_per_second": 20.837,
      "eval_steps_per_second": 20.837,
      "step": 1200
    },
    {
      "epoch": 0.644810576826281,
      "grad_norm": 0.859375,
      "learning_rate": 4.800381258800752e-05,
      "loss": 0.9638,
      "step": 1201
    },
    {
      "epoch": 0.645347471561357,
      "grad_norm": 0.7578125,
      "learning_rate": 4.8000504494461304e-05,
      "loss": 0.9991,
      "step": 1202
    },
    {
      "epoch": 0.645884366296433,
      "grad_norm": 0.64453125,
      "learning_rate": 4.799719377626624e-05,
      "loss": 0.8162,
      "step": 1203
    },
    {
      "epoch": 0.646421261031509,
      "grad_norm": 0.703125,
      "learning_rate": 4.799388043380013e-05,
      "loss": 0.9042,
      "step": 1204
    },
    {
      "epoch": 0.646958155766585,
      "grad_norm": 0.81640625,
      "learning_rate": 4.799056446744107e-05,
      "loss": 1.1001,
      "step": 1205
    },
    {
      "epoch": 0.647495050501661,
      "grad_norm": 0.765625,
      "learning_rate": 4.798724587756745e-05,
      "loss": 0.8964,
      "step": 1206
    },
    {
      "epoch": 0.648031945236737,
      "grad_norm": 0.73046875,
      "learning_rate": 4.798392466455798e-05,
      "loss": 0.8578,
      "step": 1207
    },
    {
      "epoch": 0.648568839971813,
      "grad_norm": 0.7265625,
      "learning_rate": 4.7980600828791625e-05,
      "loss": 0.9347,
      "step": 1208
    },
    {
      "epoch": 0.649105734706889,
      "grad_norm": 0.66796875,
      "learning_rate": 4.797727437064769e-05,
      "loss": 0.9061,
      "step": 1209
    },
    {
      "epoch": 0.649642629441965,
      "grad_norm": 0.78125,
      "learning_rate": 4.7973945290505766e-05,
      "loss": 1.0133,
      "step": 1210
    },
    {
      "epoch": 0.6501795241770411,
      "grad_norm": 0.6953125,
      "learning_rate": 4.797061358874575e-05,
      "loss": 1.0449,
      "step": 1211
    },
    {
      "epoch": 0.6507164189121171,
      "grad_norm": 0.67578125,
      "learning_rate": 4.796727926574781e-05,
      "loss": 0.803,
      "step": 1212
    },
    {
      "epoch": 0.6512533136471931,
      "grad_norm": 0.8046875,
      "learning_rate": 4.796394232189246e-05,
      "loss": 0.8309,
      "step": 1213
    },
    {
      "epoch": 0.6517902083822691,
      "grad_norm": 0.8046875,
      "learning_rate": 4.796060275756047e-05,
      "loss": 0.9637,
      "step": 1214
    },
    {
      "epoch": 0.6523271031173451,
      "grad_norm": 0.69921875,
      "learning_rate": 4.7957260573132944e-05,
      "loss": 0.8338,
      "step": 1215
    },
    {
      "epoch": 0.6528639978524211,
      "grad_norm": 0.73828125,
      "learning_rate": 4.795391576899125e-05,
      "loss": 1.0785,
      "step": 1216
    },
    {
      "epoch": 0.6534008925874971,
      "grad_norm": 0.91015625,
      "learning_rate": 4.795056834551708e-05,
      "loss": 1.2755,
      "step": 1217
    },
    {
      "epoch": 0.653937787322573,
      "grad_norm": 0.8828125,
      "learning_rate": 4.7947218303092414e-05,
      "loss": 1.132,
      "step": 1218
    },
    {
      "epoch": 0.654474682057649,
      "grad_norm": 0.7265625,
      "learning_rate": 4.794386564209953e-05,
      "loss": 0.9399,
      "step": 1219
    },
    {
      "epoch": 0.655011576792725,
      "grad_norm": 0.67578125,
      "learning_rate": 4.794051036292102e-05,
      "loss": 0.9379,
      "step": 1220
    },
    {
      "epoch": 0.655548471527801,
      "grad_norm": 0.88671875,
      "learning_rate": 4.793715246593976e-05,
      "loss": 1.0229,
      "step": 1221
    },
    {
      "epoch": 0.6560853662628771,
      "grad_norm": 0.83203125,
      "learning_rate": 4.793379195153894e-05,
      "loss": 0.8744,
      "step": 1222
    },
    {
      "epoch": 0.6566222609979531,
      "grad_norm": 0.7578125,
      "learning_rate": 4.793042882010202e-05,
      "loss": 1.0583,
      "step": 1223
    },
    {
      "epoch": 0.6571591557330291,
      "grad_norm": 0.87109375,
      "learning_rate": 4.7927063072012765e-05,
      "loss": 1.09,
      "step": 1224
    },
    {
      "epoch": 0.6576960504681051,
      "grad_norm": 0.7734375,
      "learning_rate": 4.7923694707655275e-05,
      "loss": 0.8834,
      "step": 1225
    },
    {
      "epoch": 0.6582329452031811,
      "grad_norm": 0.76171875,
      "learning_rate": 4.7920323727413917e-05,
      "loss": 0.9068,
      "step": 1226
    },
    {
      "epoch": 0.6587698399382571,
      "grad_norm": 0.734375,
      "learning_rate": 4.791695013167336e-05,
      "loss": 1.0008,
      "step": 1227
    },
    {
      "epoch": 0.6593067346733331,
      "grad_norm": 0.62890625,
      "learning_rate": 4.791357392081857e-05,
      "loss": 0.8071,
      "step": 1228
    },
    {
      "epoch": 0.6598436294084091,
      "grad_norm": 0.69140625,
      "learning_rate": 4.791019509523482e-05,
      "loss": 0.8902,
      "step": 1229
    },
    {
      "epoch": 0.6603805241434851,
      "grad_norm": 0.734375,
      "learning_rate": 4.790681365530768e-05,
      "loss": 0.9749,
      "step": 1230
    },
    {
      "epoch": 0.6609174188785611,
      "grad_norm": 0.6796875,
      "learning_rate": 4.7903429601423004e-05,
      "loss": 0.8686,
      "step": 1231
    },
    {
      "epoch": 0.6614543136136372,
      "grad_norm": 0.6953125,
      "learning_rate": 4.790004293396697e-05,
      "loss": 1.219,
      "step": 1232
    },
    {
      "epoch": 0.6619912083487132,
      "grad_norm": 0.86328125,
      "learning_rate": 4.789665365332602e-05,
      "loss": 1.2762,
      "step": 1233
    },
    {
      "epoch": 0.6625281030837892,
      "grad_norm": 0.79296875,
      "learning_rate": 4.789326175988693e-05,
      "loss": 1.0241,
      "step": 1234
    },
    {
      "epoch": 0.6630649978188652,
      "grad_norm": 0.6484375,
      "learning_rate": 4.7889867254036755e-05,
      "loss": 0.8839,
      "step": 1235
    },
    {
      "epoch": 0.6636018925539412,
      "grad_norm": 0.74609375,
      "learning_rate": 4.788647013616285e-05,
      "loss": 0.9357,
      "step": 1236
    },
    {
      "epoch": 0.6641387872890171,
      "grad_norm": 0.6875,
      "learning_rate": 4.7883070406652875e-05,
      "loss": 0.7911,
      "step": 1237
    },
    {
      "epoch": 0.6646756820240931,
      "grad_norm": 0.75390625,
      "learning_rate": 4.7879668065894766e-05,
      "loss": 0.8811,
      "step": 1238
    },
    {
      "epoch": 0.6652125767591691,
      "grad_norm": 0.80078125,
      "learning_rate": 4.7876263114276786e-05,
      "loss": 1.2227,
      "step": 1239
    },
    {
      "epoch": 0.6657494714942451,
      "grad_norm": 0.7109375,
      "learning_rate": 4.787285555218748e-05,
      "loss": 0.8563,
      "step": 1240
    },
    {
      "epoch": 0.6662863662293211,
      "grad_norm": 0.73828125,
      "learning_rate": 4.7869445380015697e-05,
      "loss": 1.0592,
      "step": 1241
    },
    {
      "epoch": 0.6668232609643971,
      "grad_norm": 0.77734375,
      "learning_rate": 4.786603259815058e-05,
      "loss": 0.9877,
      "step": 1242
    },
    {
      "epoch": 0.6673601556994732,
      "grad_norm": 0.7734375,
      "learning_rate": 4.7862617206981565e-05,
      "loss": 1.1321,
      "step": 1243
    },
    {
      "epoch": 0.6678970504345492,
      "grad_norm": 0.80078125,
      "learning_rate": 4.785919920689838e-05,
      "loss": 0.8493,
      "step": 1244
    },
    {
      "epoch": 0.6684339451696252,
      "grad_norm": 0.9921875,
      "learning_rate": 4.78557785982911e-05,
      "loss": 1.1701,
      "step": 1245
    },
    {
      "epoch": 0.6689708399047012,
      "grad_norm": 0.74609375,
      "learning_rate": 4.785235538155003e-05,
      "loss": 1.1437,
      "step": 1246
    },
    {
      "epoch": 0.6695077346397772,
      "grad_norm": 1.0390625,
      "learning_rate": 4.784892955706581e-05,
      "loss": 1.0322,
      "step": 1247
    },
    {
      "epoch": 0.6700446293748532,
      "grad_norm": 0.76171875,
      "learning_rate": 4.7845501125229367e-05,
      "loss": 0.9838,
      "step": 1248
    },
    {
      "epoch": 0.6705815241099292,
      "grad_norm": 0.6640625,
      "learning_rate": 4.784207008643193e-05,
      "loss": 0.9461,
      "step": 1249
    },
    {
      "epoch": 0.6711184188450052,
      "grad_norm": 0.77734375,
      "learning_rate": 4.783863644106502e-05,
      "loss": 1.0143,
      "step": 1250
    },
    {
      "epoch": 0.6711184188450052,
      "eval_loss": 0.9342340230941772,
      "eval_runtime": 46.1501,
      "eval_samples_per_second": 20.845,
      "eval_steps_per_second": 20.845,
      "step": 1250
    },
    {
      "epoch": 0.6716553135800812,
      "grad_norm": 0.9375,
      "learning_rate": 4.783520018952048e-05,
      "loss": 1.2811,
      "step": 1251
    },
    {
      "epoch": 0.6721922083151572,
      "grad_norm": 0.68359375,
      "learning_rate": 4.78317613321904e-05,
      "loss": 0.9107,
      "step": 1252
    },
    {
      "epoch": 0.6727291030502333,
      "grad_norm": 0.734375,
      "learning_rate": 4.7828319869467214e-05,
      "loss": 1.0983,
      "step": 1253
    },
    {
      "epoch": 0.6732659977853093,
      "grad_norm": 0.69140625,
      "learning_rate": 4.7824875801743636e-05,
      "loss": 0.8884,
      "step": 1254
    },
    {
      "epoch": 0.6738028925203853,
      "grad_norm": 0.734375,
      "learning_rate": 4.782142912941268e-05,
      "loss": 0.9643,
      "step": 1255
    },
    {
      "epoch": 0.6743397872554612,
      "grad_norm": 0.71875,
      "learning_rate": 4.781797985286763e-05,
      "loss": 0.9157,
      "step": 1256
    },
    {
      "epoch": 0.6748766819905372,
      "grad_norm": 0.640625,
      "learning_rate": 4.781452797250213e-05,
      "loss": 0.884,
      "step": 1257
    },
    {
      "epoch": 0.6754135767256132,
      "grad_norm": 0.83203125,
      "learning_rate": 4.7811073488710065e-05,
      "loss": 0.8141,
      "step": 1258
    },
    {
      "epoch": 0.6759504714606892,
      "grad_norm": 0.7109375,
      "learning_rate": 4.7807616401885634e-05,
      "loss": 0.7466,
      "step": 1259
    },
    {
      "epoch": 0.6764873661957652,
      "grad_norm": 0.625,
      "learning_rate": 4.780415671242334e-05,
      "loss": 0.8875,
      "step": 1260
    },
    {
      "epoch": 0.6770242609308412,
      "grad_norm": 0.60546875,
      "learning_rate": 4.7800694420717966e-05,
      "loss": 0.9136,
      "step": 1261
    },
    {
      "epoch": 0.6775611556659172,
      "grad_norm": 0.66015625,
      "learning_rate": 4.779722952716462e-05,
      "loss": 0.7917,
      "step": 1262
    },
    {
      "epoch": 0.6780980504009932,
      "grad_norm": 0.75390625,
      "learning_rate": 4.7793762032158676e-05,
      "loss": 1.0076,
      "step": 1263
    },
    {
      "epoch": 0.6786349451360693,
      "grad_norm": 0.57421875,
      "learning_rate": 4.779029193609582e-05,
      "loss": 0.8647,
      "step": 1264
    },
    {
      "epoch": 0.6791718398711453,
      "grad_norm": 0.6875,
      "learning_rate": 4.7786819239372046e-05,
      "loss": 0.8765,
      "step": 1265
    },
    {
      "epoch": 0.6797087346062213,
      "grad_norm": 0.6875,
      "learning_rate": 4.778334394238362e-05,
      "loss": 1.0059,
      "step": 1266
    },
    {
      "epoch": 0.6802456293412973,
      "grad_norm": 0.578125,
      "learning_rate": 4.777986604552712e-05,
      "loss": 0.9935,
      "step": 1267
    },
    {
      "epoch": 0.6807825240763733,
      "grad_norm": 0.671875,
      "learning_rate": 4.777638554919943e-05,
      "loss": 0.8443,
      "step": 1268
    },
    {
      "epoch": 0.6813194188114493,
      "grad_norm": 0.7421875,
      "learning_rate": 4.777290245379771e-05,
      "loss": 1.0868,
      "step": 1269
    },
    {
      "epoch": 0.6818563135465253,
      "grad_norm": 0.64453125,
      "learning_rate": 4.776941675971941e-05,
      "loss": 0.7599,
      "step": 1270
    },
    {
      "epoch": 0.6823932082816013,
      "grad_norm": 0.84375,
      "learning_rate": 4.776592846736232e-05,
      "loss": 1.1578,
      "step": 1271
    },
    {
      "epoch": 0.6829301030166773,
      "grad_norm": 0.76171875,
      "learning_rate": 4.776243757712448e-05,
      "loss": 0.928,
      "step": 1272
    },
    {
      "epoch": 0.6834669977517533,
      "grad_norm": 0.75390625,
      "learning_rate": 4.775894408940425e-05,
      "loss": 0.9463,
      "step": 1273
    },
    {
      "epoch": 0.6840038924868292,
      "grad_norm": 0.640625,
      "learning_rate": 4.775544800460028e-05,
      "loss": 0.8065,
      "step": 1274
    },
    {
      "epoch": 0.6845407872219053,
      "grad_norm": 0.73046875,
      "learning_rate": 4.775194932311152e-05,
      "loss": 0.9241,
      "step": 1275
    },
    {
      "epoch": 0.6850776819569813,
      "grad_norm": 0.8125,
      "learning_rate": 4.774844804533721e-05,
      "loss": 1.1012,
      "step": 1276
    },
    {
      "epoch": 0.6856145766920573,
      "grad_norm": 0.77734375,
      "learning_rate": 4.774494417167689e-05,
      "loss": 1.1509,
      "step": 1277
    },
    {
      "epoch": 0.6861514714271333,
      "grad_norm": 0.6484375,
      "learning_rate": 4.77414377025304e-05,
      "loss": 0.7877,
      "step": 1278
    },
    {
      "epoch": 0.6866883661622093,
      "grad_norm": 0.84375,
      "learning_rate": 4.773792863829788e-05,
      "loss": 0.9988,
      "step": 1279
    },
    {
      "epoch": 0.6872252608972853,
      "grad_norm": 0.828125,
      "learning_rate": 4.7734416979379736e-05,
      "loss": 0.9574,
      "step": 1280
    },
    {
      "epoch": 0.6877621556323613,
      "grad_norm": 0.64453125,
      "learning_rate": 4.773090272617672e-05,
      "loss": 0.9265,
      "step": 1281
    },
    {
      "epoch": 0.6882990503674373,
      "grad_norm": 0.71484375,
      "learning_rate": 4.772738587908982e-05,
      "loss": 0.821,
      "step": 1282
    },
    {
      "epoch": 0.6888359451025133,
      "grad_norm": 0.7578125,
      "learning_rate": 4.7723866438520394e-05,
      "loss": 0.8526,
      "step": 1283
    },
    {
      "epoch": 0.6893728398375893,
      "grad_norm": 0.62109375,
      "learning_rate": 4.7720344404870024e-05,
      "loss": 0.8551,
      "step": 1284
    },
    {
      "epoch": 0.6899097345726654,
      "grad_norm": 0.6796875,
      "learning_rate": 4.7716819778540625e-05,
      "loss": 1.032,
      "step": 1285
    },
    {
      "epoch": 0.6904466293077414,
      "grad_norm": 0.78515625,
      "learning_rate": 4.771329255993441e-05,
      "loss": 0.8922,
      "step": 1286
    },
    {
      "epoch": 0.6909835240428174,
      "grad_norm": 0.703125,
      "learning_rate": 4.770976274945387e-05,
      "loss": 1.0449,
      "step": 1287
    },
    {
      "epoch": 0.6915204187778934,
      "grad_norm": 0.85546875,
      "learning_rate": 4.77062303475018e-05,
      "loss": 0.8638,
      "step": 1288
    },
    {
      "epoch": 0.6920573135129694,
      "grad_norm": 0.6875,
      "learning_rate": 4.7702695354481296e-05,
      "loss": 0.8569,
      "step": 1289
    },
    {
      "epoch": 0.6925942082480454,
      "grad_norm": 0.7734375,
      "learning_rate": 4.7699157770795755e-05,
      "loss": 1.0324,
      "step": 1290
    },
    {
      "epoch": 0.6931311029831214,
      "grad_norm": 0.65625,
      "learning_rate": 4.769561759684884e-05,
      "loss": 0.8717,
      "step": 1291
    },
    {
      "epoch": 0.6936679977181974,
      "grad_norm": 0.80859375,
      "learning_rate": 4.7692074833044556e-05,
      "loss": 0.9943,
      "step": 1292
    },
    {
      "epoch": 0.6942048924532733,
      "grad_norm": 0.9375,
      "learning_rate": 4.7688529479787144e-05,
      "loss": 0.8082,
      "step": 1293
    },
    {
      "epoch": 0.6947417871883493,
      "grad_norm": 0.7578125,
      "learning_rate": 4.76849815374812e-05,
      "loss": 0.8062,
      "step": 1294
    },
    {
      "epoch": 0.6952786819234253,
      "grad_norm": 0.76171875,
      "learning_rate": 4.768143100653158e-05,
      "loss": 1.2008,
      "step": 1295
    },
    {
      "epoch": 0.6958155766585014,
      "grad_norm": 0.88671875,
      "learning_rate": 4.7677877887343445e-05,
      "loss": 0.9233,
      "step": 1296
    },
    {
      "epoch": 0.6963524713935774,
      "grad_norm": 0.68359375,
      "learning_rate": 4.767432218032225e-05,
      "loss": 0.9462,
      "step": 1297
    },
    {
      "epoch": 0.6968893661286534,
      "grad_norm": 0.7578125,
      "learning_rate": 4.767076388587375e-05,
      "loss": 0.9398,
      "step": 1298
    },
    {
      "epoch": 0.6974262608637294,
      "grad_norm": 0.73828125,
      "learning_rate": 4.766720300440398e-05,
      "loss": 0.9359,
      "step": 1299
    },
    {
      "epoch": 0.6979631555988054,
      "grad_norm": 0.69921875,
      "learning_rate": 4.7663639536319296e-05,
      "loss": 1.1696,
      "step": 1300
    },
    {
      "epoch": 0.6979631555988054,
      "eval_loss": 0.9328204393386841,
      "eval_runtime": 46.1153,
      "eval_samples_per_second": 20.861,
      "eval_steps_per_second": 20.861,
      "step": 1300
    },
    {
      "epoch": 0.6985000503338814,
      "grad_norm": 0.80859375,
      "learning_rate": 4.766007348202633e-05,
      "loss": 1.034,
      "step": 1301
    },
    {
      "epoch": 0.6990369450689574,
      "grad_norm": 0.71875,
      "learning_rate": 4.765650484193202e-05,
      "loss": 0.8461,
      "step": 1302
    },
    {
      "epoch": 0.6995738398040334,
      "grad_norm": 0.74609375,
      "learning_rate": 4.765293361644357e-05,
      "loss": 1.0729,
      "step": 1303
    },
    {
      "epoch": 0.7001107345391094,
      "grad_norm": 0.6328125,
      "learning_rate": 4.764935980596853e-05,
      "loss": 0.7903,
      "step": 1304
    },
    {
      "epoch": 0.7006476292741854,
      "grad_norm": 1.40625,
      "learning_rate": 4.76457834109147e-05,
      "loss": 0.8027,
      "step": 1305
    },
    {
      "epoch": 0.7011845240092615,
      "grad_norm": 0.75,
      "learning_rate": 4.7642204431690206e-05,
      "loss": 0.8505,
      "step": 1306
    },
    {
      "epoch": 0.7017214187443375,
      "grad_norm": 0.73828125,
      "learning_rate": 4.763862286870343e-05,
      "loss": 0.943,
      "step": 1307
    },
    {
      "epoch": 0.7022583134794135,
      "grad_norm": 0.7109375,
      "learning_rate": 4.7635038722363105e-05,
      "loss": 1.0102,
      "step": 1308
    },
    {
      "epoch": 0.7027952082144895,
      "grad_norm": 0.734375,
      "learning_rate": 4.763145199307821e-05,
      "loss": 0.9867,
      "step": 1309
    },
    {
      "epoch": 0.7033321029495655,
      "grad_norm": 0.6953125,
      "learning_rate": 4.7627862681258037e-05,
      "loss": 1.0336,
      "step": 1310
    },
    {
      "epoch": 0.7038689976846415,
      "grad_norm": 0.69921875,
      "learning_rate": 4.7624270787312166e-05,
      "loss": 0.8188,
      "step": 1311
    },
    {
      "epoch": 0.7044058924197174,
      "grad_norm": 0.90625,
      "learning_rate": 4.762067631165049e-05,
      "loss": 0.8758,
      "step": 1312
    },
    {
      "epoch": 0.7049427871547934,
      "grad_norm": 0.8359375,
      "learning_rate": 4.761707925468318e-05,
      "loss": 1.4241,
      "step": 1313
    },
    {
      "epoch": 0.7054796818898694,
      "grad_norm": 0.8671875,
      "learning_rate": 4.7613479616820705e-05,
      "loss": 1.0613,
      "step": 1314
    },
    {
      "epoch": 0.7060165766249454,
      "grad_norm": 0.7578125,
      "learning_rate": 4.7609877398473826e-05,
      "loss": 0.8751,
      "step": 1315
    },
    {
      "epoch": 0.7065534713600214,
      "grad_norm": 0.6796875,
      "learning_rate": 4.7606272600053605e-05,
      "loss": 1.0087,
      "step": 1316
    },
    {
      "epoch": 0.7070903660950975,
      "grad_norm": 0.8046875,
      "learning_rate": 4.7602665221971396e-05,
      "loss": 1.0294,
      "step": 1317
    },
    {
      "epoch": 0.7076272608301735,
      "grad_norm": 0.65625,
      "learning_rate": 4.7599055264638856e-05,
      "loss": 0.8163,
      "step": 1318
    },
    {
      "epoch": 0.7081641555652495,
      "grad_norm": 0.73046875,
      "learning_rate": 4.75954427284679e-05,
      "loss": 0.7875,
      "step": 1319
    },
    {
      "epoch": 0.7087010503003255,
      "grad_norm": 0.63671875,
      "learning_rate": 4.7591827613870785e-05,
      "loss": 0.7542,
      "step": 1320
    },
    {
      "epoch": 0.7092379450354015,
      "grad_norm": 0.71484375,
      "learning_rate": 4.758820992126005e-05,
      "loss": 0.9595,
      "step": 1321
    },
    {
      "epoch": 0.7097748397704775,
      "grad_norm": 0.80078125,
      "learning_rate": 4.758458965104849e-05,
      "loss": 1.2754,
      "step": 1322
    },
    {
      "epoch": 0.7103117345055535,
      "grad_norm": 0.6640625,
      "learning_rate": 4.758096680364925e-05,
      "loss": 0.9228,
      "step": 1323
    },
    {
      "epoch": 0.7108486292406295,
      "grad_norm": 0.77734375,
      "learning_rate": 4.757734137947573e-05,
      "loss": 1.001,
      "step": 1324
    },
    {
      "epoch": 0.7113855239757055,
      "grad_norm": 0.78515625,
      "learning_rate": 4.757371337894164e-05,
      "loss": 0.9495,
      "step": 1325
    },
    {
      "epoch": 0.7119224187107815,
      "grad_norm": 0.7265625,
      "learning_rate": 4.757008280246098e-05,
      "loss": 1.0422,
      "step": 1326
    },
    {
      "epoch": 0.7124593134458576,
      "grad_norm": 0.7421875,
      "learning_rate": 4.756644965044804e-05,
      "loss": 1.0319,
      "step": 1327
    },
    {
      "epoch": 0.7129962081809336,
      "grad_norm": 0.7421875,
      "learning_rate": 4.756281392331742e-05,
      "loss": 1.2311,
      "step": 1328
    },
    {
      "epoch": 0.7135331029160096,
      "grad_norm": 0.76171875,
      "learning_rate": 4.7559175621484e-05,
      "loss": 0.9433,
      "step": 1329
    },
    {
      "epoch": 0.7140699976510856,
      "grad_norm": 0.7109375,
      "learning_rate": 4.755553474536294e-05,
      "loss": 0.938,
      "step": 1330
    },
    {
      "epoch": 0.7146068923861615,
      "grad_norm": 0.73828125,
      "learning_rate": 4.755189129536973e-05,
      "loss": 0.8623,
      "step": 1331
    },
    {
      "epoch": 0.7151437871212375,
      "grad_norm": 0.66015625,
      "learning_rate": 4.754824527192013e-05,
      "loss": 0.9582,
      "step": 1332
    },
    {
      "epoch": 0.7156806818563135,
      "grad_norm": 0.6796875,
      "learning_rate": 4.754459667543019e-05,
      "loss": 1.0101,
      "step": 1333
    },
    {
      "epoch": 0.7162175765913895,
      "grad_norm": 0.7890625,
      "learning_rate": 4.754094550631627e-05,
      "loss": 0.8064,
      "step": 1334
    },
    {
      "epoch": 0.7167544713264655,
      "grad_norm": 0.67578125,
      "learning_rate": 4.7537291764995006e-05,
      "loss": 0.8832,
      "step": 1335
    },
    {
      "epoch": 0.7172913660615415,
      "grad_norm": 0.640625,
      "learning_rate": 4.753363545188334e-05,
      "loss": 0.9058,
      "step": 1336
    },
    {
      "epoch": 0.7178282607966175,
      "grad_norm": 0.72265625,
      "learning_rate": 4.7529976567398506e-05,
      "loss": 1.0384,
      "step": 1337
    },
    {
      "epoch": 0.7183651555316936,
      "grad_norm": 0.78515625,
      "learning_rate": 4.752631511195802e-05,
      "loss": 1.0729,
      "step": 1338
    },
    {
      "epoch": 0.7189020502667696,
      "grad_norm": 0.6484375,
      "learning_rate": 4.752265108597971e-05,
      "loss": 0.7538,
      "step": 1339
    },
    {
      "epoch": 0.7194389450018456,
      "grad_norm": 0.74609375,
      "learning_rate": 4.751898448988169e-05,
      "loss": 0.9167,
      "step": 1340
    },
    {
      "epoch": 0.7199758397369216,
      "grad_norm": 0.81640625,
      "learning_rate": 4.751531532408236e-05,
      "loss": 0.9067,
      "step": 1341
    },
    {
      "epoch": 0.7205127344719976,
      "grad_norm": 0.828125,
      "learning_rate": 4.751164358900041e-05,
      "loss": 0.9941,
      "step": 1342
    },
    {
      "epoch": 0.7210496292070736,
      "grad_norm": 0.66796875,
      "learning_rate": 4.7507969285054845e-05,
      "loss": 0.8699,
      "step": 1343
    },
    {
      "epoch": 0.7215865239421496,
      "grad_norm": 0.6640625,
      "learning_rate": 4.750429241266494e-05,
      "loss": 0.9196,
      "step": 1344
    },
    {
      "epoch": 0.7221234186772256,
      "grad_norm": 0.67578125,
      "learning_rate": 4.750061297225028e-05,
      "loss": 0.896,
      "step": 1345
    },
    {
      "epoch": 0.7226603134123016,
      "grad_norm": 0.8515625,
      "learning_rate": 4.749693096423073e-05,
      "loss": 0.7731,
      "step": 1346
    },
    {
      "epoch": 0.7231972081473776,
      "grad_norm": 0.82421875,
      "learning_rate": 4.749324638902646e-05,
      "loss": 0.8657,
      "step": 1347
    },
    {
      "epoch": 0.7237341028824537,
      "grad_norm": 0.77734375,
      "learning_rate": 4.748955924705792e-05,
      "loss": 1.0607,
      "step": 1348
    },
    {
      "epoch": 0.7242709976175297,
      "grad_norm": 0.7734375,
      "learning_rate": 4.748586953874587e-05,
      "loss": 0.9679,
      "step": 1349
    },
    {
      "epoch": 0.7248078923526056,
      "grad_norm": 0.7265625,
      "learning_rate": 4.7482177264511335e-05,
      "loss": 0.8223,
      "step": 1350
    },
    {
      "epoch": 0.7248078923526056,
      "eval_loss": 0.9316281676292419,
      "eval_runtime": 45.924,
      "eval_samples_per_second": 20.948,
      "eval_steps_per_second": 20.948,
      "step": 1350
    },
    {
      "epoch": 0.7253447870876816,
      "grad_norm": 1.3515625,
      "learning_rate": 4.747848242477566e-05,
      "loss": 1.1274,
      "step": 1351
    },
    {
      "epoch": 0.7258816818227576,
      "grad_norm": 0.69140625,
      "learning_rate": 4.747478501996048e-05,
      "loss": 0.7566,
      "step": 1352
    },
    {
      "epoch": 0.7264185765578336,
      "grad_norm": 0.59765625,
      "learning_rate": 4.747108505048771e-05,
      "loss": 0.7889,
      "step": 1353
    },
    {
      "epoch": 0.7269554712929096,
      "grad_norm": 0.83984375,
      "learning_rate": 4.7467382516779554e-05,
      "loss": 0.8987,
      "step": 1354
    },
    {
      "epoch": 0.7274923660279856,
      "grad_norm": 0.71484375,
      "learning_rate": 4.746367741925854e-05,
      "loss": 0.9739,
      "step": 1355
    },
    {
      "epoch": 0.7280292607630616,
      "grad_norm": 0.78125,
      "learning_rate": 4.7459969758347444e-05,
      "loss": 0.8851,
      "step": 1356
    },
    {
      "epoch": 0.7285661554981376,
      "grad_norm": 0.66015625,
      "learning_rate": 4.745625953446936e-05,
      "loss": 0.9751,
      "step": 1357
    },
    {
      "epoch": 0.7291030502332136,
      "grad_norm": 0.90234375,
      "learning_rate": 4.745254674804769e-05,
      "loss": 0.8199,
      "step": 1358
    },
    {
      "epoch": 0.7296399449682897,
      "grad_norm": 0.80078125,
      "learning_rate": 4.74488313995061e-05,
      "loss": 1.0326,
      "step": 1359
    },
    {
      "epoch": 0.7301768397033657,
      "grad_norm": 0.76171875,
      "learning_rate": 4.7445113489268544e-05,
      "loss": 0.9174,
      "step": 1360
    },
    {
      "epoch": 0.7307137344384417,
      "grad_norm": 0.70703125,
      "learning_rate": 4.74413930177593e-05,
      "loss": 0.8355,
      "step": 1361
    },
    {
      "epoch": 0.7312506291735177,
      "grad_norm": 0.94921875,
      "learning_rate": 4.743766998540291e-05,
      "loss": 1.1939,
      "step": 1362
    },
    {
      "epoch": 0.7317875239085937,
      "grad_norm": 0.75,
      "learning_rate": 4.743394439262423e-05,
      "loss": 0.725,
      "step": 1363
    },
    {
      "epoch": 0.7323244186436697,
      "grad_norm": 0.70703125,
      "learning_rate": 4.743021623984839e-05,
      "loss": 1.0349,
      "step": 1364
    },
    {
      "epoch": 0.7328613133787457,
      "grad_norm": 0.8671875,
      "learning_rate": 4.742648552750082e-05,
      "loss": 0.8164,
      "step": 1365
    },
    {
      "epoch": 0.7333982081138217,
      "grad_norm": 0.65625,
      "learning_rate": 4.742275225600725e-05,
      "loss": 0.8941,
      "step": 1366
    },
    {
      "epoch": 0.7339351028488977,
      "grad_norm": 0.71875,
      "learning_rate": 4.741901642579368e-05,
      "loss": 0.9404,
      "step": 1367
    },
    {
      "epoch": 0.7344719975839737,
      "grad_norm": 0.73046875,
      "learning_rate": 4.7415278037286424e-05,
      "loss": 1.0523,
      "step": 1368
    },
    {
      "epoch": 0.7350088923190496,
      "grad_norm": 0.703125,
      "learning_rate": 4.741153709091208e-05,
      "loss": 1.0108,
      "step": 1369
    },
    {
      "epoch": 0.7355457870541257,
      "grad_norm": 0.94921875,
      "learning_rate": 4.740779358709753e-05,
      "loss": 1.1026,
      "step": 1370
    },
    {
      "epoch": 0.7360826817892017,
      "grad_norm": 0.66796875,
      "learning_rate": 4.740404752626996e-05,
      "loss": 0.7894,
      "step": 1371
    },
    {
      "epoch": 0.7366195765242777,
      "grad_norm": 0.7734375,
      "learning_rate": 4.740029890885684e-05,
      "loss": 1.0289,
      "step": 1372
    },
    {
      "epoch": 0.7371564712593537,
      "grad_norm": 0.80859375,
      "learning_rate": 4.7396547735285945e-05,
      "loss": 1.0261,
      "step": 1373
    },
    {
      "epoch": 0.7376933659944297,
      "grad_norm": 0.65625,
      "learning_rate": 4.7392794005985326e-05,
      "loss": 0.8073,
      "step": 1374
    },
    {
      "epoch": 0.7382302607295057,
      "grad_norm": 0.84375,
      "learning_rate": 4.7389037721383325e-05,
      "loss": 0.8464,
      "step": 1375
    },
    {
      "epoch": 0.7387671554645817,
      "grad_norm": 0.671875,
      "learning_rate": 4.738527888190858e-05,
      "loss": 0.9254,
      "step": 1376
    },
    {
      "epoch": 0.7393040501996577,
      "grad_norm": 0.66015625,
      "learning_rate": 4.738151748799003e-05,
      "loss": 0.8076,
      "step": 1377
    },
    {
      "epoch": 0.7398409449347337,
      "grad_norm": 0.69921875,
      "learning_rate": 4.73777535400569e-05,
      "loss": 0.9767,
      "step": 1378
    },
    {
      "epoch": 0.7403778396698097,
      "grad_norm": 0.59765625,
      "learning_rate": 4.73739870385387e-05,
      "loss": 0.9295,
      "step": 1379
    },
    {
      "epoch": 0.7409147344048858,
      "grad_norm": 0.8046875,
      "learning_rate": 4.737021798386524e-05,
      "loss": 0.938,
      "step": 1380
    },
    {
      "epoch": 0.7414516291399618,
      "grad_norm": 0.66015625,
      "learning_rate": 4.73664463764666e-05,
      "loss": 0.9509,
      "step": 1381
    },
    {
      "epoch": 0.7419885238750378,
      "grad_norm": 0.59375,
      "learning_rate": 4.736267221677318e-05,
      "loss": 0.8211,
      "step": 1382
    },
    {
      "epoch": 0.7425254186101138,
      "grad_norm": 0.6328125,
      "learning_rate": 4.735889550521567e-05,
      "loss": 0.9719,
      "step": 1383
    },
    {
      "epoch": 0.7430623133451898,
      "grad_norm": 0.828125,
      "learning_rate": 4.735511624222503e-05,
      "loss": 0.9421,
      "step": 1384
    },
    {
      "epoch": 0.7435992080802658,
      "grad_norm": 0.6640625,
      "learning_rate": 4.735133442823252e-05,
      "loss": 0.801,
      "step": 1385
    },
    {
      "epoch": 0.7441361028153418,
      "grad_norm": 0.8359375,
      "learning_rate": 4.7347550063669695e-05,
      "loss": 1.1049,
      "step": 1386
    },
    {
      "epoch": 0.7446729975504178,
      "grad_norm": 0.671875,
      "learning_rate": 4.7343763148968394e-05,
      "loss": 0.876,
      "step": 1387
    },
    {
      "epoch": 0.7452098922854937,
      "grad_norm": 0.76171875,
      "learning_rate": 4.733997368456076e-05,
      "loss": 0.9858,
      "step": 1388
    },
    {
      "epoch": 0.7457467870205697,
      "grad_norm": 0.62109375,
      "learning_rate": 4.733618167087922e-05,
      "loss": 0.8927,
      "step": 1389
    },
    {
      "epoch": 0.7462836817556457,
      "grad_norm": 0.67578125,
      "learning_rate": 4.733238710835648e-05,
      "loss": 1.0067,
      "step": 1390
    },
    {
      "epoch": 0.7468205764907218,
      "grad_norm": 0.7421875,
      "learning_rate": 4.7328589997425554e-05,
      "loss": 0.9566,
      "step": 1391
    },
    {
      "epoch": 0.7473574712257978,
      "grad_norm": 0.68359375,
      "learning_rate": 4.7324790338519744e-05,
      "loss": 0.8136,
      "step": 1392
    },
    {
      "epoch": 0.7478943659608738,
      "grad_norm": 0.75,
      "learning_rate": 4.7320988132072633e-05,
      "loss": 0.887,
      "step": 1393
    },
    {
      "epoch": 0.7484312606959498,
      "grad_norm": 0.671875,
      "learning_rate": 4.731718337851811e-05,
      "loss": 0.8988,
      "step": 1394
    },
    {
      "epoch": 0.7489681554310258,
      "grad_norm": 0.76171875,
      "learning_rate": 4.7313376078290335e-05,
      "loss": 1.0616,
      "step": 1395
    },
    {
      "epoch": 0.7495050501661018,
      "grad_norm": 0.7421875,
      "learning_rate": 4.730956623182377e-05,
      "loss": 1.0494,
      "step": 1396
    },
    {
      "epoch": 0.7500419449011778,
      "grad_norm": 0.59765625,
      "learning_rate": 4.7305753839553174e-05,
      "loss": 0.802,
      "step": 1397
    },
    {
      "epoch": 0.7505788396362538,
      "grad_norm": 0.75,
      "learning_rate": 4.730193890191359e-05,
      "loss": 0.9595,
      "step": 1398
    },
    {
      "epoch": 0.7511157343713298,
      "grad_norm": 0.67578125,
      "learning_rate": 4.7298121419340335e-05,
      "loss": 0.9304,
      "step": 1399
    },
    {
      "epoch": 0.7516526291064058,
      "grad_norm": 0.70703125,
      "learning_rate": 4.729430139226905e-05,
      "loss": 0.9384,
      "step": 1400
    },
    {
      "epoch": 0.7516526291064058,
      "eval_loss": 0.9312136173248291,
      "eval_runtime": 46.4614,
      "eval_samples_per_second": 20.705,
      "eval_steps_per_second": 20.705,
      "step": 1400
    },
    {
      "epoch": 0.7521895238414819,
      "grad_norm": 0.66015625,
      "learning_rate": 4.7290478821135644e-05,
      "loss": 0.8603,
      "step": 1401
    },
    {
      "epoch": 0.7527264185765579,
      "grad_norm": 0.6875,
      "learning_rate": 4.728665370637631e-05,
      "loss": 0.871,
      "step": 1402
    },
    {
      "epoch": 0.7532633133116339,
      "grad_norm": 0.7734375,
      "learning_rate": 4.7282826048427555e-05,
      "loss": 1.0149,
      "step": 1403
    },
    {
      "epoch": 0.7538002080467099,
      "grad_norm": 0.734375,
      "learning_rate": 4.727899584772616e-05,
      "loss": 0.722,
      "step": 1404
    },
    {
      "epoch": 0.7543371027817859,
      "grad_norm": 0.828125,
      "learning_rate": 4.72751631047092e-05,
      "loss": 0.9323,
      "step": 1405
    },
    {
      "epoch": 0.7548739975168619,
      "grad_norm": 0.66796875,
      "learning_rate": 4.7271327819814035e-05,
      "loss": 0.9867,
      "step": 1406
    },
    {
      "epoch": 0.7554108922519378,
      "grad_norm": 0.70703125,
      "learning_rate": 4.726748999347832e-05,
      "loss": 0.7637,
      "step": 1407
    },
    {
      "epoch": 0.7559477869870138,
      "grad_norm": 0.6484375,
      "learning_rate": 4.726364962614e-05,
      "loss": 1.0353,
      "step": 1408
    },
    {
      "epoch": 0.7564846817220898,
      "grad_norm": 0.7578125,
      "learning_rate": 4.725980671823732e-05,
      "loss": 0.9195,
      "step": 1409
    },
    {
      "epoch": 0.7570215764571658,
      "grad_norm": 0.94921875,
      "learning_rate": 4.725596127020879e-05,
      "loss": 1.3082,
      "step": 1410
    },
    {
      "epoch": 0.7575584711922418,
      "grad_norm": 0.7890625,
      "learning_rate": 4.725211328249323e-05,
      "loss": 1.0194,
      "step": 1411
    },
    {
      "epoch": 0.7580953659273179,
      "grad_norm": 0.78125,
      "learning_rate": 4.724826275552975e-05,
      "loss": 0.9548,
      "step": 1412
    },
    {
      "epoch": 0.7586322606623939,
      "grad_norm": 0.76171875,
      "learning_rate": 4.724440968975773e-05,
      "loss": 1.0079,
      "step": 1413
    },
    {
      "epoch": 0.7591691553974699,
      "grad_norm": 0.7734375,
      "learning_rate": 4.724055408561687e-05,
      "loss": 1.1499,
      "step": 1414
    },
    {
      "epoch": 0.7597060501325459,
      "grad_norm": 0.69140625,
      "learning_rate": 4.723669594354713e-05,
      "loss": 0.826,
      "step": 1415
    },
    {
      "epoch": 0.7602429448676219,
      "grad_norm": 0.73828125,
      "learning_rate": 4.723283526398877e-05,
      "loss": 1.047,
      "step": 1416
    },
    {
      "epoch": 0.7607798396026979,
      "grad_norm": 0.65625,
      "learning_rate": 4.722897204738237e-05,
      "loss": 1.0241,
      "step": 1417
    },
    {
      "epoch": 0.7613167343377739,
      "grad_norm": 0.7109375,
      "learning_rate": 4.7225106294168735e-05,
      "loss": 0.8048,
      "step": 1418
    },
    {
      "epoch": 0.7618536290728499,
      "grad_norm": 0.84375,
      "learning_rate": 4.722123800478902e-05,
      "loss": 0.8397,
      "step": 1419
    },
    {
      "epoch": 0.7623905238079259,
      "grad_norm": 0.6796875,
      "learning_rate": 4.721736717968464e-05,
      "loss": 0.854,
      "step": 1420
    },
    {
      "epoch": 0.762927418543002,
      "grad_norm": 0.69921875,
      "learning_rate": 4.7213493819297306e-05,
      "loss": 0.9567,
      "step": 1421
    },
    {
      "epoch": 0.763464313278078,
      "grad_norm": 0.765625,
      "learning_rate": 4.720961792406901e-05,
      "loss": 1.0173,
      "step": 1422
    },
    {
      "epoch": 0.764001208013154,
      "grad_norm": 1.0625,
      "learning_rate": 4.720573949444205e-05,
      "loss": 0.9527,
      "step": 1423
    },
    {
      "epoch": 0.76453810274823,
      "grad_norm": 0.6953125,
      "learning_rate": 4.7201858530859e-05,
      "loss": 0.9913,
      "step": 1424
    },
    {
      "epoch": 0.7650749974833059,
      "grad_norm": 1.09375,
      "learning_rate": 4.719797503376273e-05,
      "loss": 0.9947,
      "step": 1425
    },
    {
      "epoch": 0.7656118922183819,
      "grad_norm": 0.80859375,
      "learning_rate": 4.71940890035964e-05,
      "loss": 0.9944,
      "step": 1426
    },
    {
      "epoch": 0.7661487869534579,
      "grad_norm": 0.76953125,
      "learning_rate": 4.7190200440803434e-05,
      "loss": 0.9377,
      "step": 1427
    },
    {
      "epoch": 0.7666856816885339,
      "grad_norm": 0.7734375,
      "learning_rate": 4.71863093458276e-05,
      "loss": 1.0204,
      "step": 1428
    },
    {
      "epoch": 0.7672225764236099,
      "grad_norm": 0.6171875,
      "learning_rate": 4.7182415719112885e-05,
      "loss": 0.9268,
      "step": 1429
    },
    {
      "epoch": 0.7677594711586859,
      "grad_norm": 0.6640625,
      "learning_rate": 4.717851956110363e-05,
      "loss": 0.7247,
      "step": 1430
    },
    {
      "epoch": 0.7682963658937619,
      "grad_norm": 0.87890625,
      "learning_rate": 4.717462087224442e-05,
      "loss": 0.9339,
      "step": 1431
    },
    {
      "epoch": 0.768833260628838,
      "grad_norm": 0.67578125,
      "learning_rate": 4.717071965298015e-05,
      "loss": 0.8147,
      "step": 1432
    },
    {
      "epoch": 0.769370155363914,
      "grad_norm": 0.7578125,
      "learning_rate": 4.7166815903756004e-05,
      "loss": 0.842,
      "step": 1433
    },
    {
      "epoch": 0.76990705009899,
      "grad_norm": 0.71484375,
      "learning_rate": 4.716290962501744e-05,
      "loss": 0.8458,
      "step": 1434
    },
    {
      "epoch": 0.770443944834066,
      "grad_norm": 0.91015625,
      "learning_rate": 4.7159000817210205e-05,
      "loss": 0.9644,
      "step": 1435
    },
    {
      "epoch": 0.770980839569142,
      "grad_norm": 0.69921875,
      "learning_rate": 4.715508948078037e-05,
      "loss": 0.9705,
      "step": 1436
    },
    {
      "epoch": 0.771517734304218,
      "grad_norm": 0.671875,
      "learning_rate": 4.715117561617425e-05,
      "loss": 0.8341,
      "step": 1437
    },
    {
      "epoch": 0.772054629039294,
      "grad_norm": 0.80859375,
      "learning_rate": 4.714725922383847e-05,
      "loss": 0.8103,
      "step": 1438
    },
    {
      "epoch": 0.77259152377437,
      "grad_norm": 0.8125,
      "learning_rate": 4.714334030421994e-05,
      "loss": 0.9949,
      "step": 1439
    },
    {
      "epoch": 0.773128418509446,
      "grad_norm": 0.828125,
      "learning_rate": 4.713941885776586e-05,
      "loss": 0.8243,
      "step": 1440
    },
    {
      "epoch": 0.773665313244522,
      "grad_norm": 0.97265625,
      "learning_rate": 4.713549488492371e-05,
      "loss": 1.1497,
      "step": 1441
    },
    {
      "epoch": 0.774202207979598,
      "grad_norm": 0.97265625,
      "learning_rate": 4.713156838614128e-05,
      "loss": 0.8016,
      "step": 1442
    },
    {
      "epoch": 0.774739102714674,
      "grad_norm": 0.71484375,
      "learning_rate": 4.7127639361866616e-05,
      "loss": 0.9364,
      "step": 1443
    },
    {
      "epoch": 0.77527599744975,
      "grad_norm": 0.8125,
      "learning_rate": 4.712370781254808e-05,
      "loss": 1.0553,
      "step": 1444
    },
    {
      "epoch": 0.775812892184826,
      "grad_norm": 0.62109375,
      "learning_rate": 4.711977373863431e-05,
      "loss": 1.0205,
      "step": 1445
    },
    {
      "epoch": 0.776349786919902,
      "grad_norm": 0.734375,
      "learning_rate": 4.7115837140574227e-05,
      "loss": 0.9484,
      "step": 1446
    },
    {
      "epoch": 0.776886681654978,
      "grad_norm": 1.0234375,
      "learning_rate": 4.711189801881706e-05,
      "loss": 1.2984,
      "step": 1447
    },
    {
      "epoch": 0.777423576390054,
      "grad_norm": 0.74609375,
      "learning_rate": 4.710795637381229e-05,
      "loss": 0.8438,
      "step": 1448
    },
    {
      "epoch": 0.77796047112513,
      "grad_norm": 0.73828125,
      "learning_rate": 4.710401220600974e-05,
      "loss": 0.9522,
      "step": 1449
    },
    {
      "epoch": 0.778497365860206,
      "grad_norm": 0.73046875,
      "learning_rate": 4.710006551585946e-05,
      "loss": 0.9466,
      "step": 1450
    },
    {
      "epoch": 0.778497365860206,
      "eval_loss": 0.9296481609344482,
      "eval_runtime": 46.0144,
      "eval_samples_per_second": 20.906,
      "eval_steps_per_second": 20.906,
      "step": 1450
    },
    {
      "epoch": 0.779034260595282,
      "grad_norm": 0.76953125,
      "learning_rate": 4.709611630381184e-05,
      "loss": 1.1011,
      "step": 1451
    },
    {
      "epoch": 0.779571155330358,
      "grad_norm": 0.7421875,
      "learning_rate": 4.709216457031752e-05,
      "loss": 1.0408,
      "step": 1452
    },
    {
      "epoch": 0.780108050065434,
      "grad_norm": 0.83984375,
      "learning_rate": 4.7088210315827444e-05,
      "loss": 0.8894,
      "step": 1453
    },
    {
      "epoch": 0.78064494480051,
      "grad_norm": 0.75,
      "learning_rate": 4.708425354079286e-05,
      "loss": 1.0962,
      "step": 1454
    },
    {
      "epoch": 0.7811818395355861,
      "grad_norm": 0.9296875,
      "learning_rate": 4.708029424566526e-05,
      "loss": 1.088,
      "step": 1455
    },
    {
      "epoch": 0.7817187342706621,
      "grad_norm": 0.77734375,
      "learning_rate": 4.7076332430896464e-05,
      "loss": 0.989,
      "step": 1456
    },
    {
      "epoch": 0.7822556290057381,
      "grad_norm": 1.03125,
      "learning_rate": 4.707236809693856e-05,
      "loss": 1.1858,
      "step": 1457
    },
    {
      "epoch": 0.7827925237408141,
      "grad_norm": 0.5625,
      "learning_rate": 4.706840124424394e-05,
      "loss": 0.8159,
      "step": 1458
    },
    {
      "epoch": 0.7833294184758901,
      "grad_norm": 0.7265625,
      "learning_rate": 4.706443187326526e-05,
      "loss": 0.8357,
      "step": 1459
    },
    {
      "epoch": 0.7838663132109661,
      "grad_norm": 0.9609375,
      "learning_rate": 4.706045998445548e-05,
      "loss": 1.0639,
      "step": 1460
    },
    {
      "epoch": 0.7844032079460421,
      "grad_norm": 0.8046875,
      "learning_rate": 4.7056485578267846e-05,
      "loss": 1.1264,
      "step": 1461
    },
    {
      "epoch": 0.7849401026811181,
      "grad_norm": 0.7265625,
      "learning_rate": 4.705250865515588e-05,
      "loss": 1.141,
      "step": 1462
    },
    {
      "epoch": 0.785476997416194,
      "grad_norm": 0.81640625,
      "learning_rate": 4.70485292155734e-05,
      "loss": 1.1089,
      "step": 1463
    },
    {
      "epoch": 0.78601389215127,
      "grad_norm": 0.78515625,
      "learning_rate": 4.7044547259974525e-05,
      "loss": 1.002,
      "step": 1464
    },
    {
      "epoch": 0.786550786886346,
      "grad_norm": 0.6875,
      "learning_rate": 4.7040562788813636e-05,
      "loss": 1.015,
      "step": 1465
    },
    {
      "epoch": 0.7870876816214221,
      "grad_norm": 0.85546875,
      "learning_rate": 4.703657580254541e-05,
      "loss": 0.9949,
      "step": 1466
    },
    {
      "epoch": 0.7876245763564981,
      "grad_norm": 0.765625,
      "learning_rate": 4.70325863016248e-05,
      "loss": 1.0667,
      "step": 1467
    },
    {
      "epoch": 0.7881614710915741,
      "grad_norm": 0.81640625,
      "learning_rate": 4.7028594286507095e-05,
      "loss": 0.9384,
      "step": 1468
    },
    {
      "epoch": 0.7886983658266501,
      "grad_norm": 0.75390625,
      "learning_rate": 4.7024599757647804e-05,
      "loss": 1.0406,
      "step": 1469
    },
    {
      "epoch": 0.7892352605617261,
      "grad_norm": 0.6953125,
      "learning_rate": 4.702060271550276e-05,
      "loss": 0.8912,
      "step": 1470
    },
    {
      "epoch": 0.7897721552968021,
      "grad_norm": 1.09375,
      "learning_rate": 4.701660316052807e-05,
      "loss": 1.0978,
      "step": 1471
    },
    {
      "epoch": 0.7903090500318781,
      "grad_norm": 0.86328125,
      "learning_rate": 4.701260109318015e-05,
      "loss": 0.9843,
      "step": 1472
    },
    {
      "epoch": 0.7908459447669541,
      "grad_norm": 0.87109375,
      "learning_rate": 4.7008596513915684e-05,
      "loss": 0.9987,
      "step": 1473
    },
    {
      "epoch": 0.7913828395020301,
      "grad_norm": 0.64453125,
      "learning_rate": 4.7004589423191623e-05,
      "loss": 0.8369,
      "step": 1474
    },
    {
      "epoch": 0.7919197342371062,
      "grad_norm": 0.75,
      "learning_rate": 4.700057982146526e-05,
      "loss": 1.1304,
      "step": 1475
    },
    {
      "epoch": 0.7924566289721822,
      "grad_norm": 0.62109375,
      "learning_rate": 4.699656770919412e-05,
      "loss": 0.8327,
      "step": 1476
    },
    {
      "epoch": 0.7929935237072582,
      "grad_norm": 0.70703125,
      "learning_rate": 4.6992553086836035e-05,
      "loss": 0.9318,
      "step": 1477
    },
    {
      "epoch": 0.7935304184423342,
      "grad_norm": 0.7265625,
      "learning_rate": 4.698853595484913e-05,
      "loss": 0.977,
      "step": 1478
    },
    {
      "epoch": 0.7940673131774102,
      "grad_norm": 0.703125,
      "learning_rate": 4.6984516313691816e-05,
      "loss": 0.8479,
      "step": 1479
    },
    {
      "epoch": 0.7946042079124862,
      "grad_norm": 0.69921875,
      "learning_rate": 4.698049416382278e-05,
      "loss": 1.0746,
      "step": 1480
    },
    {
      "epoch": 0.7951411026475622,
      "grad_norm": 0.671875,
      "learning_rate": 4.697646950570099e-05,
      "loss": 0.8582,
      "step": 1481
    },
    {
      "epoch": 0.7956779973826381,
      "grad_norm": 0.95703125,
      "learning_rate": 4.697244233978573e-05,
      "loss": 0.9033,
      "step": 1482
    },
    {
      "epoch": 0.7962148921177141,
      "grad_norm": 0.703125,
      "learning_rate": 4.696841266653653e-05,
      "loss": 0.9216,
      "step": 1483
    },
    {
      "epoch": 0.7967517868527901,
      "grad_norm": 0.73046875,
      "learning_rate": 4.696438048641324e-05,
      "loss": 1.0413,
      "step": 1484
    },
    {
      "epoch": 0.7972886815878661,
      "grad_norm": 0.640625,
      "learning_rate": 4.6960345799875995e-05,
      "loss": 1.1256,
      "step": 1485
    },
    {
      "epoch": 0.7978255763229422,
      "grad_norm": 0.69921875,
      "learning_rate": 4.6956308607385166e-05,
      "loss": 0.9003,
      "step": 1486
    },
    {
      "epoch": 0.7983624710580182,
      "grad_norm": 0.6953125,
      "learning_rate": 4.695226890940149e-05,
      "loss": 0.9033,
      "step": 1487
    },
    {
      "epoch": 0.7988993657930942,
      "grad_norm": 0.6953125,
      "learning_rate": 4.6948226706385915e-05,
      "loss": 1.1069,
      "step": 1488
    },
    {
      "epoch": 0.7994362605281702,
      "grad_norm": 0.65625,
      "learning_rate": 4.694418199879973e-05,
      "loss": 0.895,
      "step": 1489
    },
    {
      "epoch": 0.7999731552632462,
      "grad_norm": 0.66015625,
      "learning_rate": 4.694013478710447e-05,
      "loss": 0.8756,
      "step": 1490
    },
    {
      "epoch": 0.8005100499983222,
      "grad_norm": 0.56640625,
      "learning_rate": 4.693608507176199e-05,
      "loss": 0.8967,
      "step": 1491
    },
    {
      "epoch": 0.8010469447333982,
      "grad_norm": 0.73828125,
      "learning_rate": 4.6932032853234406e-05,
      "loss": 0.973,
      "step": 1492
    },
    {
      "epoch": 0.8015838394684742,
      "grad_norm": 0.76171875,
      "learning_rate": 4.692797813198412e-05,
      "loss": 0.8524,
      "step": 1493
    },
    {
      "epoch": 0.8021207342035502,
      "grad_norm": 0.88671875,
      "learning_rate": 4.692392090847383e-05,
      "loss": 1.0615,
      "step": 1494
    },
    {
      "epoch": 0.8026576289386262,
      "grad_norm": 0.64453125,
      "learning_rate": 4.6919861183166536e-05,
      "loss": 0.7777,
      "step": 1495
    },
    {
      "epoch": 0.8031945236737023,
      "grad_norm": 0.94921875,
      "learning_rate": 4.691579895652548e-05,
      "loss": 1.1297,
      "step": 1496
    },
    {
      "epoch": 0.8037314184087783,
      "grad_norm": 0.703125,
      "learning_rate": 4.6911734229014215e-05,
      "loss": 0.8149,
      "step": 1497
    },
    {
      "epoch": 0.8042683131438543,
      "grad_norm": 0.70703125,
      "learning_rate": 4.690766700109659e-05,
      "loss": 0.966,
      "step": 1498
    },
    {
      "epoch": 0.8048052078789303,
      "grad_norm": 0.94921875,
      "learning_rate": 4.690359727323672e-05,
      "loss": 0.9832,
      "step": 1499
    },
    {
      "epoch": 0.8053421026140063,
      "grad_norm": 0.65625,
      "learning_rate": 4.6899525045899024e-05,
      "loss": 0.8791,
      "step": 1500
    },
    {
      "epoch": 0.8053421026140063,
      "eval_loss": 0.928687572479248,
      "eval_runtime": 45.9461,
      "eval_samples_per_second": 20.938,
      "eval_steps_per_second": 20.938,
      "step": 1500
    },
    {
      "epoch": 0.8058789973490822,
      "grad_norm": 0.74609375,
      "learning_rate": 4.6895450319548175e-05,
      "loss": 0.9375,
      "step": 1501
    },
    {
      "epoch": 0.8064158920841582,
      "grad_norm": 0.73828125,
      "learning_rate": 4.6891373094649165e-05,
      "loss": 0.9404,
      "step": 1502
    },
    {
      "epoch": 0.8069527868192342,
      "grad_norm": 0.71875,
      "learning_rate": 4.688729337166725e-05,
      "loss": 0.8475,
      "step": 1503
    },
    {
      "epoch": 0.8074896815543102,
      "grad_norm": 0.6015625,
      "learning_rate": 4.688321115106798e-05,
      "loss": 0.8304,
      "step": 1504
    },
    {
      "epoch": 0.8080265762893862,
      "grad_norm": 0.7421875,
      "learning_rate": 4.68791264333172e-05,
      "loss": 1.1578,
      "step": 1505
    },
    {
      "epoch": 0.8085634710244622,
      "grad_norm": 0.59765625,
      "learning_rate": 4.687503921888101e-05,
      "loss": 0.7527,
      "step": 1506
    },
    {
      "epoch": 0.8091003657595383,
      "grad_norm": 0.734375,
      "learning_rate": 4.6870949508225826e-05,
      "loss": 1.0027,
      "step": 1507
    },
    {
      "epoch": 0.8096372604946143,
      "grad_norm": 0.71875,
      "learning_rate": 4.686685730181832e-05,
      "loss": 0.8635,
      "step": 1508
    },
    {
      "epoch": 0.8101741552296903,
      "grad_norm": 0.7734375,
      "learning_rate": 4.6862762600125476e-05,
      "loss": 0.9531,
      "step": 1509
    },
    {
      "epoch": 0.8107110499647663,
      "grad_norm": 0.6875,
      "learning_rate": 4.685866540361456e-05,
      "loss": 0.8614,
      "step": 1510
    },
    {
      "epoch": 0.8112479446998423,
      "grad_norm": 0.6328125,
      "learning_rate": 4.685456571275309e-05,
      "loss": 0.8557,
      "step": 1511
    },
    {
      "epoch": 0.8117848394349183,
      "grad_norm": 0.79296875,
      "learning_rate": 4.685046352800891e-05,
      "loss": 0.941,
      "step": 1512
    },
    {
      "epoch": 0.8123217341699943,
      "grad_norm": 0.6796875,
      "learning_rate": 4.6846358849850136e-05,
      "loss": 0.8133,
      "step": 1513
    },
    {
      "epoch": 0.8128586289050703,
      "grad_norm": 0.671875,
      "learning_rate": 4.6842251678745156e-05,
      "loss": 0.8278,
      "step": 1514
    },
    {
      "epoch": 0.8133955236401463,
      "grad_norm": 0.71875,
      "learning_rate": 4.683814201516264e-05,
      "loss": 1.0006,
      "step": 1515
    },
    {
      "epoch": 0.8139324183752223,
      "grad_norm": 0.65625,
      "learning_rate": 4.683402985957157e-05,
      "loss": 0.7521,
      "step": 1516
    },
    {
      "epoch": 0.8144693131102984,
      "grad_norm": 0.7578125,
      "learning_rate": 4.682991521244118e-05,
      "loss": 0.9986,
      "step": 1517
    },
    {
      "epoch": 0.8150062078453744,
      "grad_norm": 0.76171875,
      "learning_rate": 4.682579807424101e-05,
      "loss": 0.9481,
      "step": 1518
    },
    {
      "epoch": 0.8155431025804504,
      "grad_norm": 0.91015625,
      "learning_rate": 4.6821678445440884e-05,
      "loss": 0.9817,
      "step": 1519
    },
    {
      "epoch": 0.8160799973155263,
      "grad_norm": 0.67578125,
      "learning_rate": 4.68175563265109e-05,
      "loss": 0.8421,
      "step": 1520
    },
    {
      "epoch": 0.8166168920506023,
      "grad_norm": 0.734375,
      "learning_rate": 4.681343171792144e-05,
      "loss": 0.7721,
      "step": 1521
    },
    {
      "epoch": 0.8171537867856783,
      "grad_norm": 0.7578125,
      "learning_rate": 4.6809304620143175e-05,
      "loss": 1.0361,
      "step": 1522
    },
    {
      "epoch": 0.8176906815207543,
      "grad_norm": 0.6796875,
      "learning_rate": 4.680517503364706e-05,
      "loss": 0.947,
      "step": 1523
    },
    {
      "epoch": 0.8182275762558303,
      "grad_norm": 1.0859375,
      "learning_rate": 4.680104295890433e-05,
      "loss": 0.9522,
      "step": 1524
    },
    {
      "epoch": 0.8187644709909063,
      "grad_norm": 0.68359375,
      "learning_rate": 4.679690839638652e-05,
      "loss": 1.1382,
      "step": 1525
    },
    {
      "epoch": 0.8193013657259823,
      "grad_norm": 0.6484375,
      "learning_rate": 4.6792771346565425e-05,
      "loss": 0.7148,
      "step": 1526
    },
    {
      "epoch": 0.8198382604610583,
      "grad_norm": 0.92578125,
      "learning_rate": 4.678863180991313e-05,
      "loss": 1.0069,
      "step": 1527
    },
    {
      "epoch": 0.8203751551961344,
      "grad_norm": 0.81640625,
      "learning_rate": 4.678448978690201e-05,
      "loss": 0.8415,
      "step": 1528
    },
    {
      "epoch": 0.8209120499312104,
      "grad_norm": 0.73828125,
      "learning_rate": 4.678034527800474e-05,
      "loss": 1.0156,
      "step": 1529
    },
    {
      "epoch": 0.8214489446662864,
      "grad_norm": 0.90234375,
      "learning_rate": 4.677619828369424e-05,
      "loss": 1.2303,
      "step": 1530
    },
    {
      "epoch": 0.8219858394013624,
      "grad_norm": 0.8359375,
      "learning_rate": 4.677204880444375e-05,
      "loss": 1.1202,
      "step": 1531
    },
    {
      "epoch": 0.8225227341364384,
      "grad_norm": 0.63671875,
      "learning_rate": 4.676789684072676e-05,
      "loss": 0.9635,
      "step": 1532
    },
    {
      "epoch": 0.8230596288715144,
      "grad_norm": 0.8359375,
      "learning_rate": 4.6763742393017076e-05,
      "loss": 1.0879,
      "step": 1533
    },
    {
      "epoch": 0.8235965236065904,
      "grad_norm": 0.73828125,
      "learning_rate": 4.6759585461788774e-05,
      "loss": 0.8324,
      "step": 1534
    },
    {
      "epoch": 0.8241334183416664,
      "grad_norm": 0.6875,
      "learning_rate": 4.6755426047516205e-05,
      "loss": 1.0856,
      "step": 1535
    },
    {
      "epoch": 0.8246703130767424,
      "grad_norm": 0.68359375,
      "learning_rate": 4.675126415067401e-05,
      "loss": 0.967,
      "step": 1536
    },
    {
      "epoch": 0.8252072078118184,
      "grad_norm": 0.63671875,
      "learning_rate": 4.674709977173712e-05,
      "loss": 0.7723,
      "step": 1537
    },
    {
      "epoch": 0.8257441025468945,
      "grad_norm": 0.73046875,
      "learning_rate": 4.674293291118075e-05,
      "loss": 0.8389,
      "step": 1538
    },
    {
      "epoch": 0.8262809972819704,
      "grad_norm": 0.859375,
      "learning_rate": 4.6738763569480375e-05,
      "loss": 1.1869,
      "step": 1539
    },
    {
      "epoch": 0.8268178920170464,
      "grad_norm": 0.6796875,
      "learning_rate": 4.673459174711178e-05,
      "loss": 0.9137,
      "step": 1540
    },
    {
      "epoch": 0.8273547867521224,
      "grad_norm": 0.82421875,
      "learning_rate": 4.673041744455102e-05,
      "loss": 1.1577,
      "step": 1541
    },
    {
      "epoch": 0.8278916814871984,
      "grad_norm": 0.703125,
      "learning_rate": 4.672624066227444e-05,
      "loss": 0.8871,
      "step": 1542
    },
    {
      "epoch": 0.8284285762222744,
      "grad_norm": 0.6328125,
      "learning_rate": 4.6722061400758654e-05,
      "loss": 0.7516,
      "step": 1543
    },
    {
      "epoch": 0.8289654709573504,
      "grad_norm": 0.6171875,
      "learning_rate": 4.671787966048057e-05,
      "loss": 0.763,
      "step": 1544
    },
    {
      "epoch": 0.8295023656924264,
      "grad_norm": 0.78125,
      "learning_rate": 4.6713695441917396e-05,
      "loss": 0.9525,
      "step": 1545
    },
    {
      "epoch": 0.8300392604275024,
      "grad_norm": 0.70703125,
      "learning_rate": 4.6709508745546575e-05,
      "loss": 0.9975,
      "step": 1546
    },
    {
      "epoch": 0.8305761551625784,
      "grad_norm": 0.765625,
      "learning_rate": 4.670531957184588e-05,
      "loss": 0.7822,
      "step": 1547
    },
    {
      "epoch": 0.8311130498976544,
      "grad_norm": 0.8671875,
      "learning_rate": 4.6701127921293355e-05,
      "loss": 1.0512,
      "step": 1548
    },
    {
      "epoch": 0.8316499446327305,
      "grad_norm": 0.7578125,
      "learning_rate": 4.669693379436731e-05,
      "loss": 0.8699,
      "step": 1549
    },
    {
      "epoch": 0.8321868393678065,
      "grad_norm": 0.76171875,
      "learning_rate": 4.6692737191546346e-05,
      "loss": 1.1393,
      "step": 1550
    },
    {
      "epoch": 0.8321868393678065,
      "eval_loss": 0.9287117719650269,
      "eval_runtime": 46.0676,
      "eval_samples_per_second": 20.882,
      "eval_steps_per_second": 20.882,
      "step": 1550
    },
    {
      "epoch": 0.8327237341028825,
      "grad_norm": 0.7109375,
      "learning_rate": 4.6688538113309355e-05,
      "loss": 0.7913,
      "step": 1551
    },
    {
      "epoch": 0.8332606288379585,
      "grad_norm": 0.6796875,
      "learning_rate": 4.6684336560135497e-05,
      "loss": 0.8743,
      "step": 1552
    },
    {
      "epoch": 0.8337975235730345,
      "grad_norm": 0.66796875,
      "learning_rate": 4.668013253250423e-05,
      "loss": 0.913,
      "step": 1553
    },
    {
      "epoch": 0.8343344183081105,
      "grad_norm": 0.64453125,
      "learning_rate": 4.667592603089529e-05,
      "loss": 0.7931,
      "step": 1554
    },
    {
      "epoch": 0.8348713130431865,
      "grad_norm": 0.71875,
      "learning_rate": 4.6671717055788675e-05,
      "loss": 0.9818,
      "step": 1555
    },
    {
      "epoch": 0.8354082077782625,
      "grad_norm": 0.9921875,
      "learning_rate": 4.666750560766471e-05,
      "loss": 0.995,
      "step": 1556
    },
    {
      "epoch": 0.8359451025133385,
      "grad_norm": 0.74609375,
      "learning_rate": 4.666329168700394e-05,
      "loss": 0.8833,
      "step": 1557
    },
    {
      "epoch": 0.8364819972484144,
      "grad_norm": 0.765625,
      "learning_rate": 4.665907529428726e-05,
      "loss": 0.7022,
      "step": 1558
    },
    {
      "epoch": 0.8370188919834904,
      "grad_norm": 0.7109375,
      "learning_rate": 4.6654856429995794e-05,
      "loss": 0.8494,
      "step": 1559
    },
    {
      "epoch": 0.8375557867185665,
      "grad_norm": 0.64453125,
      "learning_rate": 4.665063509461097e-05,
      "loss": 0.8747,
      "step": 1560
    },
    {
      "epoch": 0.8380926814536425,
      "grad_norm": 0.8515625,
      "learning_rate": 4.66464112886145e-05,
      "loss": 1.0861,
      "step": 1561
    },
    {
      "epoch": 0.8386295761887185,
      "grad_norm": 0.60546875,
      "learning_rate": 4.6642185012488374e-05,
      "loss": 0.8035,
      "step": 1562
    },
    {
      "epoch": 0.8391664709237945,
      "grad_norm": 0.75,
      "learning_rate": 4.6637956266714855e-05,
      "loss": 0.9639,
      "step": 1563
    },
    {
      "epoch": 0.8397033656588705,
      "grad_norm": 0.69140625,
      "learning_rate": 4.663372505177651e-05,
      "loss": 0.9753,
      "step": 1564
    },
    {
      "epoch": 0.8402402603939465,
      "grad_norm": 0.7421875,
      "learning_rate": 4.662949136815617e-05,
      "loss": 0.9643,
      "step": 1565
    },
    {
      "epoch": 0.8407771551290225,
      "grad_norm": 0.77734375,
      "learning_rate": 4.662525521633695e-05,
      "loss": 1.007,
      "step": 1566
    },
    {
      "epoch": 0.8413140498640985,
      "grad_norm": 0.76171875,
      "learning_rate": 4.662101659680224e-05,
      "loss": 1.0901,
      "step": 1567
    },
    {
      "epoch": 0.8418509445991745,
      "grad_norm": 0.734375,
      "learning_rate": 4.6616775510035735e-05,
      "loss": 1.0253,
      "step": 1568
    },
    {
      "epoch": 0.8423878393342505,
      "grad_norm": 0.73046875,
      "learning_rate": 4.661253195652139e-05,
      "loss": 0.8747,
      "step": 1569
    },
    {
      "epoch": 0.8429247340693266,
      "grad_norm": 0.6953125,
      "learning_rate": 4.6608285936743445e-05,
      "loss": 0.8479,
      "step": 1570
    },
    {
      "epoch": 0.8434616288044026,
      "grad_norm": 0.70703125,
      "learning_rate": 4.660403745118642e-05,
      "loss": 0.9357,
      "step": 1571
    },
    {
      "epoch": 0.8439985235394786,
      "grad_norm": 0.828125,
      "learning_rate": 4.659978650033514e-05,
      "loss": 0.7952,
      "step": 1572
    },
    {
      "epoch": 0.8445354182745546,
      "grad_norm": 0.76171875,
      "learning_rate": 4.659553308467468e-05,
      "loss": 1.1169,
      "step": 1573
    },
    {
      "epoch": 0.8450723130096306,
      "grad_norm": 0.828125,
      "learning_rate": 4.659127720469041e-05,
      "loss": 0.9751,
      "step": 1574
    },
    {
      "epoch": 0.8456092077447066,
      "grad_norm": 1.1640625,
      "learning_rate": 4.658701886086798e-05,
      "loss": 0.8086,
      "step": 1575
    },
    {
      "epoch": 0.8461461024797825,
      "grad_norm": 0.59765625,
      "learning_rate": 4.658275805369331e-05,
      "loss": 0.8188,
      "step": 1576
    },
    {
      "epoch": 0.8466829972148585,
      "grad_norm": 0.73828125,
      "learning_rate": 4.6578494783652635e-05,
      "loss": 0.8419,
      "step": 1577
    },
    {
      "epoch": 0.8472198919499345,
      "grad_norm": 0.6015625,
      "learning_rate": 4.6574229051232433e-05,
      "loss": 0.8852,
      "step": 1578
    },
    {
      "epoch": 0.8477567866850105,
      "grad_norm": 0.75,
      "learning_rate": 4.656996085691948e-05,
      "loss": 0.9037,
      "step": 1579
    },
    {
      "epoch": 0.8482936814200865,
      "grad_norm": 0.72265625,
      "learning_rate": 4.656569020120083e-05,
      "loss": 0.9597,
      "step": 1580
    },
    {
      "epoch": 0.8488305761551626,
      "grad_norm": 0.7890625,
      "learning_rate": 4.656141708456382e-05,
      "loss": 1.1091,
      "step": 1581
    },
    {
      "epoch": 0.8493674708902386,
      "grad_norm": 0.7421875,
      "learning_rate": 4.655714150749607e-05,
      "loss": 0.9231,
      "step": 1582
    },
    {
      "epoch": 0.8499043656253146,
      "grad_norm": 0.71484375,
      "learning_rate": 4.655286347048547e-05,
      "loss": 1.011,
      "step": 1583
    },
    {
      "epoch": 0.8504412603603906,
      "grad_norm": 0.640625,
      "learning_rate": 4.6548582974020216e-05,
      "loss": 0.9487,
      "step": 1584
    },
    {
      "epoch": 0.8509781550954666,
      "grad_norm": 0.64453125,
      "learning_rate": 4.654430001858874e-05,
      "loss": 0.7564,
      "step": 1585
    },
    {
      "epoch": 0.8515150498305426,
      "grad_norm": 0.62109375,
      "learning_rate": 4.65400146046798e-05,
      "loss": 0.7795,
      "step": 1586
    },
    {
      "epoch": 0.8520519445656186,
      "grad_norm": 0.87109375,
      "learning_rate": 4.653572673278241e-05,
      "loss": 0.8144,
      "step": 1587
    },
    {
      "epoch": 0.8525888393006946,
      "grad_norm": 0.7109375,
      "learning_rate": 4.653143640338588e-05,
      "loss": 0.8644,
      "step": 1588
    },
    {
      "epoch": 0.8531257340357706,
      "grad_norm": 0.90234375,
      "learning_rate": 4.6527143616979776e-05,
      "loss": 1.0435,
      "step": 1589
    },
    {
      "epoch": 0.8536626287708466,
      "grad_norm": 0.70703125,
      "learning_rate": 4.652284837405397e-05,
      "loss": 0.8945,
      "step": 1590
    },
    {
      "epoch": 0.8541995235059227,
      "grad_norm": 0.83203125,
      "learning_rate": 4.65185506750986e-05,
      "loss": 0.7452,
      "step": 1591
    },
    {
      "epoch": 0.8547364182409987,
      "grad_norm": 0.7109375,
      "learning_rate": 4.6514250520604074e-05,
      "loss": 1.0434,
      "step": 1592
    },
    {
      "epoch": 0.8552733129760747,
      "grad_norm": 0.76171875,
      "learning_rate": 4.6509947911061125e-05,
      "loss": 1.001,
      "step": 1593
    },
    {
      "epoch": 0.8558102077111507,
      "grad_norm": 0.75,
      "learning_rate": 4.6505642846960704e-05,
      "loss": 1.0016,
      "step": 1594
    },
    {
      "epoch": 0.8563471024462266,
      "grad_norm": 0.98046875,
      "learning_rate": 4.650133532879409e-05,
      "loss": 1.067,
      "step": 1595
    },
    {
      "epoch": 0.8568839971813026,
      "grad_norm": 0.61328125,
      "learning_rate": 4.6497025357052835e-05,
      "loss": 0.8001,
      "step": 1596
    },
    {
      "epoch": 0.8574208919163786,
      "grad_norm": 0.86328125,
      "learning_rate": 4.6492712932228736e-05,
      "loss": 0.9735,
      "step": 1597
    },
    {
      "epoch": 0.8579577866514546,
      "grad_norm": 0.71875,
      "learning_rate": 4.6488398054813916e-05,
      "loss": 0.8591,
      "step": 1598
    },
    {
      "epoch": 0.8584946813865306,
      "grad_norm": 0.640625,
      "learning_rate": 4.648408072530074e-05,
      "loss": 1.0515,
      "step": 1599
    },
    {
      "epoch": 0.8590315761216066,
      "grad_norm": 0.703125,
      "learning_rate": 4.647976094418189e-05,
      "loss": 0.9288,
      "step": 1600
    },
    {
      "epoch": 0.8590315761216066,
      "eval_loss": 0.927656888961792,
      "eval_runtime": 46.3378,
      "eval_samples_per_second": 20.761,
      "eval_steps_per_second": 20.761,
      "step": 1600
    },
    {
      "epoch": 0.8595684708566826,
      "grad_norm": 0.64453125,
      "learning_rate": 4.647543871195029e-05,
      "loss": 0.9534,
      "step": 1601
    },
    {
      "epoch": 0.8601053655917587,
      "grad_norm": 0.69921875,
      "learning_rate": 4.6471114029099174e-05,
      "loss": 0.9108,
      "step": 1602
    },
    {
      "epoch": 0.8606422603268347,
      "grad_norm": 0.765625,
      "learning_rate": 4.646678689612203e-05,
      "loss": 0.8877,
      "step": 1603
    },
    {
      "epoch": 0.8611791550619107,
      "grad_norm": 0.8125,
      "learning_rate": 4.646245731351265e-05,
      "loss": 1.1448,
      "step": 1604
    },
    {
      "epoch": 0.8617160497969867,
      "grad_norm": 0.78125,
      "learning_rate": 4.64581252817651e-05,
      "loss": 0.913,
      "step": 1605
    },
    {
      "epoch": 0.8622529445320627,
      "grad_norm": 0.7265625,
      "learning_rate": 4.6453790801373695e-05,
      "loss": 0.9422,
      "step": 1606
    },
    {
      "epoch": 0.8627898392671387,
      "grad_norm": 0.7578125,
      "learning_rate": 4.6449453872833084e-05,
      "loss": 0.982,
      "step": 1607
    },
    {
      "epoch": 0.8633267340022147,
      "grad_norm": 0.76953125,
      "learning_rate": 4.6445114496638136e-05,
      "loss": 0.8221,
      "step": 1608
    },
    {
      "epoch": 0.8638636287372907,
      "grad_norm": 0.7109375,
      "learning_rate": 4.644077267328405e-05,
      "loss": 1.0404,
      "step": 1609
    },
    {
      "epoch": 0.8644005234723667,
      "grad_norm": 0.67578125,
      "learning_rate": 4.643642840326627e-05,
      "loss": 0.7026,
      "step": 1610
    },
    {
      "epoch": 0.8649374182074427,
      "grad_norm": 0.8203125,
      "learning_rate": 4.6432081687080554e-05,
      "loss": 1.0056,
      "step": 1611
    },
    {
      "epoch": 0.8654743129425188,
      "grad_norm": 0.8203125,
      "learning_rate": 4.642773252522289e-05,
      "loss": 0.7586,
      "step": 1612
    },
    {
      "epoch": 0.8660112076775948,
      "grad_norm": 0.84765625,
      "learning_rate": 4.642338091818959e-05,
      "loss": 0.8835,
      "step": 1613
    },
    {
      "epoch": 0.8665481024126707,
      "grad_norm": 0.6484375,
      "learning_rate": 4.641902686647722e-05,
      "loss": 0.9294,
      "step": 1614
    },
    {
      "epoch": 0.8670849971477467,
      "grad_norm": 0.66015625,
      "learning_rate": 4.641467037058263e-05,
      "loss": 0.9119,
      "step": 1615
    },
    {
      "epoch": 0.8676218918828227,
      "grad_norm": 1.25,
      "learning_rate": 4.6410311431002966e-05,
      "loss": 1.1248,
      "step": 1616
    },
    {
      "epoch": 0.8681587866178987,
      "grad_norm": 0.6640625,
      "learning_rate": 4.6405950048235625e-05,
      "loss": 1.0666,
      "step": 1617
    },
    {
      "epoch": 0.8686956813529747,
      "grad_norm": 0.81640625,
      "learning_rate": 4.640158622277829e-05,
      "loss": 1.2896,
      "step": 1618
    },
    {
      "epoch": 0.8692325760880507,
      "grad_norm": 0.6640625,
      "learning_rate": 4.639721995512895e-05,
      "loss": 1.175,
      "step": 1619
    },
    {
      "epoch": 0.8697694708231267,
      "grad_norm": 0.7890625,
      "learning_rate": 4.6392851245785834e-05,
      "loss": 1.0522,
      "step": 1620
    },
    {
      "epoch": 0.8703063655582027,
      "grad_norm": 0.69140625,
      "learning_rate": 4.638848009524747e-05,
      "loss": 0.8324,
      "step": 1621
    },
    {
      "epoch": 0.8708432602932787,
      "grad_norm": 0.70703125,
      "learning_rate": 4.638410650401267e-05,
      "loss": 0.9249,
      "step": 1622
    },
    {
      "epoch": 0.8713801550283548,
      "grad_norm": 0.72265625,
      "learning_rate": 4.6379730472580506e-05,
      "loss": 0.831,
      "step": 1623
    },
    {
      "epoch": 0.8719170497634308,
      "grad_norm": 0.625,
      "learning_rate": 4.637535200145034e-05,
      "loss": 0.8674,
      "step": 1624
    },
    {
      "epoch": 0.8724539444985068,
      "grad_norm": 0.83984375,
      "learning_rate": 4.637097109112182e-05,
      "loss": 1.0633,
      "step": 1625
    },
    {
      "epoch": 0.8729908392335828,
      "grad_norm": 0.6171875,
      "learning_rate": 4.636658774209485e-05,
      "loss": 1.041,
      "step": 1626
    },
    {
      "epoch": 0.8735277339686588,
      "grad_norm": 0.62109375,
      "learning_rate": 4.636220195486964e-05,
      "loss": 0.9218,
      "step": 1627
    },
    {
      "epoch": 0.8740646287037348,
      "grad_norm": 0.64453125,
      "learning_rate": 4.6357813729946655e-05,
      "loss": 0.9361,
      "step": 1628
    },
    {
      "epoch": 0.8746015234388108,
      "grad_norm": 0.8359375,
      "learning_rate": 4.635342306782664e-05,
      "loss": 1.0006,
      "step": 1629
    },
    {
      "epoch": 0.8751384181738868,
      "grad_norm": 0.78125,
      "learning_rate": 4.6349029969010644e-05,
      "loss": 0.7847,
      "step": 1630
    },
    {
      "epoch": 0.8756753129089628,
      "grad_norm": 0.76953125,
      "learning_rate": 4.634463443399995e-05,
      "loss": 1.0114,
      "step": 1631
    },
    {
      "epoch": 0.8762122076440388,
      "grad_norm": 0.703125,
      "learning_rate": 4.634023646329618e-05,
      "loss": 0.9952,
      "step": 1632
    },
    {
      "epoch": 0.8767491023791147,
      "grad_norm": 0.72265625,
      "learning_rate": 4.633583605740117e-05,
      "loss": 0.8418,
      "step": 1633
    },
    {
      "epoch": 0.8772859971141908,
      "grad_norm": 0.70703125,
      "learning_rate": 4.633143321681707e-05,
      "loss": 0.9023,
      "step": 1634
    },
    {
      "epoch": 0.8778228918492668,
      "grad_norm": 0.7734375,
      "learning_rate": 4.6327027942046286e-05,
      "loss": 0.9766,
      "step": 1635
    },
    {
      "epoch": 0.8783597865843428,
      "grad_norm": 0.7734375,
      "learning_rate": 4.632262023359154e-05,
      "loss": 1.0,
      "step": 1636
    },
    {
      "epoch": 0.8788966813194188,
      "grad_norm": 0.73828125,
      "learning_rate": 4.63182100919558e-05,
      "loss": 1.0702,
      "step": 1637
    },
    {
      "epoch": 0.8794335760544948,
      "grad_norm": 0.66015625,
      "learning_rate": 4.631379751764231e-05,
      "loss": 1.0212,
      "step": 1638
    },
    {
      "epoch": 0.8799704707895708,
      "grad_norm": 0.6796875,
      "learning_rate": 4.630938251115461e-05,
      "loss": 0.8073,
      "step": 1639
    },
    {
      "epoch": 0.8805073655246468,
      "grad_norm": 1.515625,
      "learning_rate": 4.6304965072996495e-05,
      "loss": 0.9944,
      "step": 1640
    },
    {
      "epoch": 0.8810442602597228,
      "grad_norm": 0.9375,
      "learning_rate": 4.6300545203672065e-05,
      "loss": 0.9397,
      "step": 1641
    },
    {
      "epoch": 0.8815811549947988,
      "grad_norm": 0.68359375,
      "learning_rate": 4.629612290368568e-05,
      "loss": 0.9176,
      "step": 1642
    },
    {
      "epoch": 0.8821180497298748,
      "grad_norm": 0.71484375,
      "learning_rate": 4.629169817354198e-05,
      "loss": 0.8976,
      "step": 1643
    },
    {
      "epoch": 0.8826549444649509,
      "grad_norm": 0.6640625,
      "learning_rate": 4.628727101374587e-05,
      "loss": 0.9841,
      "step": 1644
    },
    {
      "epoch": 0.8831918392000269,
      "grad_norm": 0.73046875,
      "learning_rate": 4.6282841424802557e-05,
      "loss": 0.9706,
      "step": 1645
    },
    {
      "epoch": 0.8837287339351029,
      "grad_norm": 1.3515625,
      "learning_rate": 4.6278409407217525e-05,
      "loss": 0.8614,
      "step": 1646
    },
    {
      "epoch": 0.8842656286701789,
      "grad_norm": 0.89453125,
      "learning_rate": 4.62739749614965e-05,
      "loss": 0.9592,
      "step": 1647
    },
    {
      "epoch": 0.8848025234052549,
      "grad_norm": 0.734375,
      "learning_rate": 4.626953808814553e-05,
      "loss": 0.9304,
      "step": 1648
    },
    {
      "epoch": 0.8853394181403309,
      "grad_norm": 0.796875,
      "learning_rate": 4.626509878767089e-05,
      "loss": 1.0085,
      "step": 1649
    },
    {
      "epoch": 0.8858763128754069,
      "grad_norm": 0.75,
      "learning_rate": 4.6260657060579185e-05,
      "loss": 0.9249,
      "step": 1650
    },
    {
      "epoch": 0.8858763128754069,
      "eval_loss": 0.9260422587394714,
      "eval_runtime": 45.8059,
      "eval_samples_per_second": 21.002,
      "eval_steps_per_second": 21.002,
      "step": 1650
    },
    {
      "epoch": 0.8864132076104829,
      "grad_norm": 0.71484375,
      "learning_rate": 4.6256212907377274e-05,
      "loss": 0.961,
      "step": 1651
    },
    {
      "epoch": 0.8869501023455588,
      "grad_norm": 0.6796875,
      "learning_rate": 4.625176632857227e-05,
      "loss": 0.8949,
      "step": 1652
    },
    {
      "epoch": 0.8874869970806348,
      "grad_norm": 0.58203125,
      "learning_rate": 4.6247317324671605e-05,
      "loss": 0.8514,
      "step": 1653
    },
    {
      "epoch": 0.8880238918157108,
      "grad_norm": 0.765625,
      "learning_rate": 4.624286589618295e-05,
      "loss": 0.8828,
      "step": 1654
    },
    {
      "epoch": 0.8885607865507869,
      "grad_norm": 0.765625,
      "learning_rate": 4.6238412043614276e-05,
      "loss": 0.9396,
      "step": 1655
    },
    {
      "epoch": 0.8890976812858629,
      "grad_norm": 0.66015625,
      "learning_rate": 4.6233955767473834e-05,
      "loss": 0.9512,
      "step": 1656
    },
    {
      "epoch": 0.8896345760209389,
      "grad_norm": 0.7265625,
      "learning_rate": 4.622949706827013e-05,
      "loss": 1.0584,
      "step": 1657
    },
    {
      "epoch": 0.8901714707560149,
      "grad_norm": 0.70703125,
      "learning_rate": 4.622503594651196e-05,
      "loss": 0.8989,
      "step": 1658
    },
    {
      "epoch": 0.8907083654910909,
      "grad_norm": 0.76953125,
      "learning_rate": 4.622057240270839e-05,
      "loss": 1.1702,
      "step": 1659
    },
    {
      "epoch": 0.8912452602261669,
      "grad_norm": 0.73046875,
      "learning_rate": 4.621610643736878e-05,
      "loss": 0.9822,
      "step": 1660
    },
    {
      "epoch": 0.8917821549612429,
      "grad_norm": 0.68359375,
      "learning_rate": 4.6211638051002734e-05,
      "loss": 1.1196,
      "step": 1661
    },
    {
      "epoch": 0.8923190496963189,
      "grad_norm": 0.7890625,
      "learning_rate": 4.620716724412018e-05,
      "loss": 0.8581,
      "step": 1662
    },
    {
      "epoch": 0.8928559444313949,
      "grad_norm": 0.62890625,
      "learning_rate": 4.620269401723126e-05,
      "loss": 0.8999,
      "step": 1663
    },
    {
      "epoch": 0.8933928391664709,
      "grad_norm": 0.7109375,
      "learning_rate": 4.6198218370846444e-05,
      "loss": 0.9555,
      "step": 1664
    },
    {
      "epoch": 0.893929733901547,
      "grad_norm": 0.7109375,
      "learning_rate": 4.619374030547647e-05,
      "loss": 0.9766,
      "step": 1665
    },
    {
      "epoch": 0.894466628636623,
      "grad_norm": 0.8359375,
      "learning_rate": 4.618925982163232e-05,
      "loss": 1.0356,
      "step": 1666
    },
    {
      "epoch": 0.895003523371699,
      "grad_norm": 0.9296875,
      "learning_rate": 4.618477691982529e-05,
      "loss": 0.9794,
      "step": 1667
    },
    {
      "epoch": 0.895540418106775,
      "grad_norm": 0.7734375,
      "learning_rate": 4.618029160056693e-05,
      "loss": 1.2178,
      "step": 1668
    },
    {
      "epoch": 0.896077312841851,
      "grad_norm": 0.79296875,
      "learning_rate": 4.617580386436906e-05,
      "loss": 0.958,
      "step": 1669
    },
    {
      "epoch": 0.896614207576927,
      "grad_norm": 0.73046875,
      "learning_rate": 4.617131371174381e-05,
      "loss": 0.8052,
      "step": 1670
    },
    {
      "epoch": 0.8971511023120029,
      "grad_norm": 0.7734375,
      "learning_rate": 4.6166821143203555e-05,
      "loss": 1.1036,
      "step": 1671
    },
    {
      "epoch": 0.8976879970470789,
      "grad_norm": 0.76953125,
      "learning_rate": 4.6162326159260945e-05,
      "loss": 1.0919,
      "step": 1672
    },
    {
      "epoch": 0.8982248917821549,
      "grad_norm": 0.9765625,
      "learning_rate": 4.615782876042892e-05,
      "loss": 1.0033,
      "step": 1673
    },
    {
      "epoch": 0.8987617865172309,
      "grad_norm": 0.64453125,
      "learning_rate": 4.6153328947220696e-05,
      "loss": 0.8411,
      "step": 1674
    },
    {
      "epoch": 0.8992986812523069,
      "grad_norm": 0.84375,
      "learning_rate": 4.614882672014975e-05,
      "loss": 0.8503,
      "step": 1675
    },
    {
      "epoch": 0.899835575987383,
      "grad_norm": 0.84765625,
      "learning_rate": 4.6144322079729854e-05,
      "loss": 0.9053,
      "step": 1676
    },
    {
      "epoch": 0.900372470722459,
      "grad_norm": 0.7109375,
      "learning_rate": 4.6139815026475034e-05,
      "loss": 1.1372,
      "step": 1677
    },
    {
      "epoch": 0.900909365457535,
      "grad_norm": 0.6015625,
      "learning_rate": 4.61353055608996e-05,
      "loss": 0.7013,
      "step": 1678
    },
    {
      "epoch": 0.901446260192611,
      "grad_norm": 0.82421875,
      "learning_rate": 4.6130793683518146e-05,
      "loss": 0.8741,
      "step": 1679
    },
    {
      "epoch": 0.901983154927687,
      "grad_norm": 0.859375,
      "learning_rate": 4.612627939484554e-05,
      "loss": 1.1555,
      "step": 1680
    },
    {
      "epoch": 0.902520049662763,
      "grad_norm": 0.6875,
      "learning_rate": 4.612176269539691e-05,
      "loss": 0.9153,
      "step": 1681
    },
    {
      "epoch": 0.903056944397839,
      "grad_norm": 0.66015625,
      "learning_rate": 4.6117243585687664e-05,
      "loss": 1.0942,
      "step": 1682
    },
    {
      "epoch": 0.903593839132915,
      "grad_norm": 0.8515625,
      "learning_rate": 4.61127220662335e-05,
      "loss": 0.9599,
      "step": 1683
    },
    {
      "epoch": 0.904130733867991,
      "grad_norm": 0.80078125,
      "learning_rate": 4.610819813755038e-05,
      "loss": 0.9821,
      "step": 1684
    },
    {
      "epoch": 0.904667628603067,
      "grad_norm": 0.77734375,
      "learning_rate": 4.610367180015454e-05,
      "loss": 0.9118,
      "step": 1685
    },
    {
      "epoch": 0.905204523338143,
      "grad_norm": 0.75390625,
      "learning_rate": 4.6099143054562486e-05,
      "loss": 1.0199,
      "step": 1686
    },
    {
      "epoch": 0.9057414180732191,
      "grad_norm": 0.921875,
      "learning_rate": 4.609461190129101e-05,
      "loss": 1.2331,
      "step": 1687
    },
    {
      "epoch": 0.9062783128082951,
      "grad_norm": 0.62890625,
      "learning_rate": 4.609007834085718e-05,
      "loss": 0.7347,
      "step": 1688
    },
    {
      "epoch": 0.9068152075433711,
      "grad_norm": 0.796875,
      "learning_rate": 4.6085542373778326e-05,
      "loss": 0.8719,
      "step": 1689
    },
    {
      "epoch": 0.907352102278447,
      "grad_norm": 0.859375,
      "learning_rate": 4.608100400057206e-05,
      "loss": 0.942,
      "step": 1690
    },
    {
      "epoch": 0.907888997013523,
      "grad_norm": 0.76953125,
      "learning_rate": 4.607646322175626e-05,
      "loss": 0.8428,
      "step": 1691
    },
    {
      "epoch": 0.908425891748599,
      "grad_norm": 0.80859375,
      "learning_rate": 4.60719200378491e-05,
      "loss": 0.7774,
      "step": 1692
    },
    {
      "epoch": 0.908962786483675,
      "grad_norm": 0.62109375,
      "learning_rate": 4.606737444936902e-05,
      "loss": 0.7665,
      "step": 1693
    },
    {
      "epoch": 0.909499681218751,
      "grad_norm": 0.8359375,
      "learning_rate": 4.6062826456834704e-05,
      "loss": 0.855,
      "step": 1694
    },
    {
      "epoch": 0.910036575953827,
      "grad_norm": 0.6796875,
      "learning_rate": 4.605827606076516e-05,
      "loss": 0.9617,
      "step": 1695
    },
    {
      "epoch": 0.910573470688903,
      "grad_norm": 0.77734375,
      "learning_rate": 4.605372326167963e-05,
      "loss": 1.0468,
      "step": 1696
    },
    {
      "epoch": 0.911110365423979,
      "grad_norm": 0.67578125,
      "learning_rate": 4.604916806009766e-05,
      "loss": 0.8403,
      "step": 1697
    },
    {
      "epoch": 0.9116472601590551,
      "grad_norm": 0.66796875,
      "learning_rate": 4.604461045653904e-05,
      "loss": 0.8779,
      "step": 1698
    },
    {
      "epoch": 0.9121841548941311,
      "grad_norm": 0.9296875,
      "learning_rate": 4.604005045152385e-05,
      "loss": 0.8258,
      "step": 1699
    },
    {
      "epoch": 0.9127210496292071,
      "grad_norm": 0.7265625,
      "learning_rate": 4.603548804557247e-05,
      "loss": 1.0053,
      "step": 1700
    },
    {
      "epoch": 0.9127210496292071,
      "eval_loss": 0.9246048331260681,
      "eval_runtime": 46.856,
      "eval_samples_per_second": 20.531,
      "eval_steps_per_second": 20.531,
      "step": 1700
    },
    {
      "epoch": 0.9132579443642831,
      "grad_norm": 0.75,
      "learning_rate": 4.60309232392055e-05,
      "loss": 1.033,
      "step": 1701
    },
    {
      "epoch": 0.9137948390993591,
      "grad_norm": 0.72265625,
      "learning_rate": 4.6026356032943866e-05,
      "loss": 0.8617,
      "step": 1702
    },
    {
      "epoch": 0.9143317338344351,
      "grad_norm": 0.7109375,
      "learning_rate": 4.6021786427308724e-05,
      "loss": 1.1039,
      "step": 1703
    },
    {
      "epoch": 0.9148686285695111,
      "grad_norm": 0.67578125,
      "learning_rate": 4.601721442282152e-05,
      "loss": 0.9135,
      "step": 1704
    },
    {
      "epoch": 0.9154055233045871,
      "grad_norm": 0.62890625,
      "learning_rate": 4.601264002000401e-05,
      "loss": 1.0555,
      "step": 1705
    },
    {
      "epoch": 0.9159424180396631,
      "grad_norm": 0.7265625,
      "learning_rate": 4.600806321937815e-05,
      "loss": 0.9806,
      "step": 1706
    },
    {
      "epoch": 0.9164793127747392,
      "grad_norm": 0.80078125,
      "learning_rate": 4.6003484021466245e-05,
      "loss": 0.9285,
      "step": 1707
    },
    {
      "epoch": 0.9170162075098152,
      "grad_norm": 0.6640625,
      "learning_rate": 4.599890242679082e-05,
      "loss": 0.9098,
      "step": 1708
    },
    {
      "epoch": 0.9175531022448911,
      "grad_norm": 0.67578125,
      "learning_rate": 4.59943184358747e-05,
      "loss": 0.9752,
      "step": 1709
    },
    {
      "epoch": 0.9180899969799671,
      "grad_norm": 0.78125,
      "learning_rate": 4.598973204924097e-05,
      "loss": 0.9775,
      "step": 1710
    },
    {
      "epoch": 0.9186268917150431,
      "grad_norm": 0.7265625,
      "learning_rate": 4.598514326741301e-05,
      "loss": 0.8949,
      "step": 1711
    },
    {
      "epoch": 0.9191637864501191,
      "grad_norm": 0.86328125,
      "learning_rate": 4.598055209091444e-05,
      "loss": 0.9869,
      "step": 1712
    },
    {
      "epoch": 0.9197006811851951,
      "grad_norm": 0.6484375,
      "learning_rate": 4.597595852026918e-05,
      "loss": 0.8093,
      "step": 1713
    },
    {
      "epoch": 0.9202375759202711,
      "grad_norm": 0.77734375,
      "learning_rate": 4.597136255600142e-05,
      "loss": 0.9374,
      "step": 1714
    },
    {
      "epoch": 0.9207744706553471,
      "grad_norm": 0.7578125,
      "learning_rate": 4.5966764198635606e-05,
      "loss": 1.1599,
      "step": 1715
    },
    {
      "epoch": 0.9213113653904231,
      "grad_norm": 0.7265625,
      "learning_rate": 4.596216344869647e-05,
      "loss": 0.9477,
      "step": 1716
    },
    {
      "epoch": 0.9218482601254991,
      "grad_norm": 0.984375,
      "learning_rate": 4.595756030670902e-05,
      "loss": 1.0141,
      "step": 1717
    },
    {
      "epoch": 0.9223851548605752,
      "grad_norm": 0.76953125,
      "learning_rate": 4.595295477319854e-05,
      "loss": 1.0733,
      "step": 1718
    },
    {
      "epoch": 0.9229220495956512,
      "grad_norm": 0.82421875,
      "learning_rate": 4.594834684869058e-05,
      "loss": 1.1318,
      "step": 1719
    },
    {
      "epoch": 0.9234589443307272,
      "grad_norm": 0.69140625,
      "learning_rate": 4.5943736533710944e-05,
      "loss": 1.0442,
      "step": 1720
    },
    {
      "epoch": 0.9239958390658032,
      "grad_norm": 0.671875,
      "learning_rate": 4.5939123828785735e-05,
      "loss": 1.0249,
      "step": 1721
    },
    {
      "epoch": 0.9245327338008792,
      "grad_norm": 0.8203125,
      "learning_rate": 4.593450873444133e-05,
      "loss": 1.0056,
      "step": 1722
    },
    {
      "epoch": 0.9250696285359552,
      "grad_norm": 0.7109375,
      "learning_rate": 4.592989125120437e-05,
      "loss": 0.8301,
      "step": 1723
    },
    {
      "epoch": 0.9256065232710312,
      "grad_norm": 0.76953125,
      "learning_rate": 4.592527137960175e-05,
      "loss": 1.1348,
      "step": 1724
    },
    {
      "epoch": 0.9261434180061072,
      "grad_norm": 0.8203125,
      "learning_rate": 4.592064912016068e-05,
      "loss": 0.8146,
      "step": 1725
    },
    {
      "epoch": 0.9266803127411832,
      "grad_norm": 0.66015625,
      "learning_rate": 4.59160244734086e-05,
      "loss": 0.845,
      "step": 1726
    },
    {
      "epoch": 0.9272172074762591,
      "grad_norm": 0.87109375,
      "learning_rate": 4.591139743987325e-05,
      "loss": 1.0645,
      "step": 1727
    },
    {
      "epoch": 0.9277541022113351,
      "grad_norm": 0.80078125,
      "learning_rate": 4.5906768020082634e-05,
      "loss": 0.9206,
      "step": 1728
    },
    {
      "epoch": 0.9282909969464112,
      "grad_norm": 0.7421875,
      "learning_rate": 4.590213621456502e-05,
      "loss": 1.0432,
      "step": 1729
    },
    {
      "epoch": 0.9288278916814872,
      "grad_norm": 0.86328125,
      "learning_rate": 4.589750202384896e-05,
      "loss": 0.8983,
      "step": 1730
    },
    {
      "epoch": 0.9293647864165632,
      "grad_norm": 0.7578125,
      "learning_rate": 4.5892865448463274e-05,
      "loss": 0.8719,
      "step": 1731
    },
    {
      "epoch": 0.9299016811516392,
      "grad_norm": 0.69140625,
      "learning_rate": 4.588822648893706e-05,
      "loss": 1.1086,
      "step": 1732
    },
    {
      "epoch": 0.9304385758867152,
      "grad_norm": 0.59375,
      "learning_rate": 4.588358514579967e-05,
      "loss": 0.7372,
      "step": 1733
    },
    {
      "epoch": 0.9309754706217912,
      "grad_norm": 0.6875,
      "learning_rate": 4.587894141958074e-05,
      "loss": 0.9003,
      "step": 1734
    },
    {
      "epoch": 0.9315123653568672,
      "grad_norm": 0.7578125,
      "learning_rate": 4.587429531081019e-05,
      "loss": 0.9172,
      "step": 1735
    },
    {
      "epoch": 0.9320492600919432,
      "grad_norm": 0.73828125,
      "learning_rate": 4.5869646820018185e-05,
      "loss": 0.9648,
      "step": 1736
    },
    {
      "epoch": 0.9325861548270192,
      "grad_norm": 0.79296875,
      "learning_rate": 4.58649959477352e-05,
      "loss": 0.8828,
      "step": 1737
    },
    {
      "epoch": 0.9331230495620952,
      "grad_norm": 0.80859375,
      "learning_rate": 4.586034269449193e-05,
      "loss": 0.9438,
      "step": 1738
    },
    {
      "epoch": 0.9336599442971713,
      "grad_norm": 0.64453125,
      "learning_rate": 4.585568706081939e-05,
      "loss": 0.9909,
      "step": 1739
    },
    {
      "epoch": 0.9341968390322473,
      "grad_norm": 0.734375,
      "learning_rate": 4.585102904724884e-05,
      "loss": 0.8296,
      "step": 1740
    },
    {
      "epoch": 0.9347337337673233,
      "grad_norm": 0.78125,
      "learning_rate": 4.5846368654311825e-05,
      "loss": 0.8979,
      "step": 1741
    },
    {
      "epoch": 0.9352706285023993,
      "grad_norm": 0.64453125,
      "learning_rate": 4.5841705882540134e-05,
      "loss": 0.9257,
      "step": 1742
    },
    {
      "epoch": 0.9358075232374753,
      "grad_norm": 0.69140625,
      "learning_rate": 4.583704073246588e-05,
      "loss": 0.8802,
      "step": 1743
    },
    {
      "epoch": 0.9363444179725513,
      "grad_norm": 0.921875,
      "learning_rate": 4.583237320462138e-05,
      "loss": 1.0676,
      "step": 1744
    },
    {
      "epoch": 0.9368813127076273,
      "grad_norm": 0.78515625,
      "learning_rate": 4.5827703299539293e-05,
      "loss": 0.9623,
      "step": 1745
    },
    {
      "epoch": 0.9374182074427032,
      "grad_norm": 0.67578125,
      "learning_rate": 4.5823031017752485e-05,
      "loss": 0.9597,
      "step": 1746
    },
    {
      "epoch": 0.9379551021777792,
      "grad_norm": 0.7734375,
      "learning_rate": 4.5818356359794146e-05,
      "loss": 1.0,
      "step": 1747
    },
    {
      "epoch": 0.9384919969128552,
      "grad_norm": 0.5546875,
      "learning_rate": 4.58136793261977e-05,
      "loss": 0.8805,
      "step": 1748
    },
    {
      "epoch": 0.9390288916479312,
      "grad_norm": 0.71875,
      "learning_rate": 4.580899991749686e-05,
      "loss": 0.9251,
      "step": 1749
    },
    {
      "epoch": 0.9395657863830073,
      "grad_norm": 0.60546875,
      "learning_rate": 4.58043181342256e-05,
      "loss": 0.7675,
      "step": 1750
    },
    {
      "epoch": 0.9395657863830073,
      "eval_loss": 0.9235332012176514,
      "eval_runtime": 45.9473,
      "eval_samples_per_second": 20.937,
      "eval_steps_per_second": 20.937,
      "step": 1750
    },
    {
      "epoch": 0.9401026811180833,
      "grad_norm": 0.9140625,
      "learning_rate": 4.579963397691818e-05,
      "loss": 1.0145,
      "step": 1751
    },
    {
      "epoch": 0.9406395758531593,
      "grad_norm": 0.6796875,
      "learning_rate": 4.579494744610912e-05,
      "loss": 0.9521,
      "step": 1752
    },
    {
      "epoch": 0.9411764705882353,
      "grad_norm": 0.66796875,
      "learning_rate": 4.57902585423332e-05,
      "loss": 0.7413,
      "step": 1753
    },
    {
      "epoch": 0.9417133653233113,
      "grad_norm": 0.66015625,
      "learning_rate": 4.5785567266125494e-05,
      "loss": 0.8167,
      "step": 1754
    },
    {
      "epoch": 0.9422502600583873,
      "grad_norm": 0.6640625,
      "learning_rate": 4.578087361802135e-05,
      "loss": 0.8401,
      "step": 1755
    },
    {
      "epoch": 0.9427871547934633,
      "grad_norm": 1.0625,
      "learning_rate": 4.577617759855634e-05,
      "loss": 1.123,
      "step": 1756
    },
    {
      "epoch": 0.9433240495285393,
      "grad_norm": 0.73046875,
      "learning_rate": 4.5771479208266363e-05,
      "loss": 1.0633,
      "step": 1757
    },
    {
      "epoch": 0.9438609442636153,
      "grad_norm": 1.0234375,
      "learning_rate": 4.5766778447687565e-05,
      "loss": 0.9038,
      "step": 1758
    },
    {
      "epoch": 0.9443978389986913,
      "grad_norm": 0.68359375,
      "learning_rate": 4.576207531735635e-05,
      "loss": 0.9805,
      "step": 1759
    },
    {
      "epoch": 0.9449347337337674,
      "grad_norm": 0.73046875,
      "learning_rate": 4.5757369817809415e-05,
      "loss": 0.9875,
      "step": 1760
    },
    {
      "epoch": 0.9454716284688434,
      "grad_norm": 0.75390625,
      "learning_rate": 4.575266194958371e-05,
      "loss": 0.95,
      "step": 1761
    },
    {
      "epoch": 0.9460085232039194,
      "grad_norm": 0.65625,
      "learning_rate": 4.574795171321646e-05,
      "loss": 0.7474,
      "step": 1762
    },
    {
      "epoch": 0.9465454179389954,
      "grad_norm": 0.6953125,
      "learning_rate": 4.574323910924517e-05,
      "loss": 0.8633,
      "step": 1763
    },
    {
      "epoch": 0.9470823126740714,
      "grad_norm": 0.7578125,
      "learning_rate": 4.5738524138207616e-05,
      "loss": 0.936,
      "step": 1764
    },
    {
      "epoch": 0.9476192074091473,
      "grad_norm": 0.6640625,
      "learning_rate": 4.5733806800641816e-05,
      "loss": 0.9208,
      "step": 1765
    },
    {
      "epoch": 0.9481561021442233,
      "grad_norm": 0.7265625,
      "learning_rate": 4.572908709708609e-05,
      "loss": 0.8539,
      "step": 1766
    },
    {
      "epoch": 0.9486929968792993,
      "grad_norm": 0.6875,
      "learning_rate": 4.5724365028079017e-05,
      "loss": 0.8122,
      "step": 1767
    },
    {
      "epoch": 0.9492298916143753,
      "grad_norm": 0.93359375,
      "learning_rate": 4.571964059415944e-05,
      "loss": 0.9129,
      "step": 1768
    },
    {
      "epoch": 0.9497667863494513,
      "grad_norm": 0.82421875,
      "learning_rate": 4.571491379586647e-05,
      "loss": 0.7864,
      "step": 1769
    },
    {
      "epoch": 0.9503036810845273,
      "grad_norm": 0.72265625,
      "learning_rate": 4.571018463373952e-05,
      "loss": 0.9741,
      "step": 1770
    },
    {
      "epoch": 0.9508405758196034,
      "grad_norm": 0.67578125,
      "learning_rate": 4.570545310831821e-05,
      "loss": 0.8631,
      "step": 1771
    },
    {
      "epoch": 0.9513774705546794,
      "grad_norm": 0.66796875,
      "learning_rate": 4.5700719220142494e-05,
      "loss": 0.8401,
      "step": 1772
    },
    {
      "epoch": 0.9519143652897554,
      "grad_norm": 0.7734375,
      "learning_rate": 4.5695982969752566e-05,
      "loss": 0.7391,
      "step": 1773
    },
    {
      "epoch": 0.9524512600248314,
      "grad_norm": 0.6484375,
      "learning_rate": 4.569124435768889e-05,
      "loss": 0.9252,
      "step": 1774
    },
    {
      "epoch": 0.9529881547599074,
      "grad_norm": 0.59375,
      "learning_rate": 4.5686503384492194e-05,
      "loss": 0.8185,
      "step": 1775
    },
    {
      "epoch": 0.9535250494949834,
      "grad_norm": 0.67578125,
      "learning_rate": 4.5681760050703485e-05,
      "loss": 0.7863,
      "step": 1776
    },
    {
      "epoch": 0.9540619442300594,
      "grad_norm": 0.87109375,
      "learning_rate": 4.567701435686404e-05,
      "loss": 0.8573,
      "step": 1777
    },
    {
      "epoch": 0.9545988389651354,
      "grad_norm": 0.765625,
      "learning_rate": 4.567226630351542e-05,
      "loss": 1.1293,
      "step": 1778
    },
    {
      "epoch": 0.9551357337002114,
      "grad_norm": 0.7109375,
      "learning_rate": 4.566751589119941e-05,
      "loss": 0.9446,
      "step": 1779
    },
    {
      "epoch": 0.9556726284352874,
      "grad_norm": 1.3828125,
      "learning_rate": 4.5662763120458105e-05,
      "loss": 0.9117,
      "step": 1780
    },
    {
      "epoch": 0.9562095231703635,
      "grad_norm": 0.71484375,
      "learning_rate": 4.565800799183385e-05,
      "loss": 1.1525,
      "step": 1781
    },
    {
      "epoch": 0.9567464179054395,
      "grad_norm": 0.67578125,
      "learning_rate": 4.5653250505869274e-05,
      "loss": 1.1099,
      "step": 1782
    },
    {
      "epoch": 0.9572833126405155,
      "grad_norm": 0.73046875,
      "learning_rate": 4.5648490663107266e-05,
      "loss": 1.0494,
      "step": 1783
    },
    {
      "epoch": 0.9578202073755914,
      "grad_norm": 0.703125,
      "learning_rate": 4.5643728464090986e-05,
      "loss": 0.8424,
      "step": 1784
    },
    {
      "epoch": 0.9583571021106674,
      "grad_norm": 0.63671875,
      "learning_rate": 4.5638963909363845e-05,
      "loss": 0.8713,
      "step": 1785
    },
    {
      "epoch": 0.9588939968457434,
      "grad_norm": 0.78125,
      "learning_rate": 4.563419699946956e-05,
      "loss": 0.7595,
      "step": 1786
    },
    {
      "epoch": 0.9594308915808194,
      "grad_norm": 0.65625,
      "learning_rate": 4.562942773495208e-05,
      "loss": 0.8329,
      "step": 1787
    },
    {
      "epoch": 0.9599677863158954,
      "grad_norm": 0.8515625,
      "learning_rate": 4.5624656116355645e-05,
      "loss": 1.1087,
      "step": 1788
    },
    {
      "epoch": 0.9605046810509714,
      "grad_norm": 0.89453125,
      "learning_rate": 4.5619882144224765e-05,
      "loss": 1.0542,
      "step": 1789
    },
    {
      "epoch": 0.9610415757860474,
      "grad_norm": 0.69140625,
      "learning_rate": 4.5615105819104194e-05,
      "loss": 0.8634,
      "step": 1790
    },
    {
      "epoch": 0.9615784705211234,
      "grad_norm": 0.7890625,
      "learning_rate": 4.561032714153898e-05,
      "loss": 1.0115,
      "step": 1791
    },
    {
      "epoch": 0.9621153652561995,
      "grad_norm": 0.765625,
      "learning_rate": 4.560554611207444e-05,
      "loss": 0.9457,
      "step": 1792
    },
    {
      "epoch": 0.9626522599912755,
      "grad_norm": 0.64453125,
      "learning_rate": 4.560076273125613e-05,
      "loss": 0.8585,
      "step": 1793
    },
    {
      "epoch": 0.9631891547263515,
      "grad_norm": 0.8046875,
      "learning_rate": 4.5595976999629906e-05,
      "loss": 1.007,
      "step": 1794
    },
    {
      "epoch": 0.9637260494614275,
      "grad_norm": 0.75390625,
      "learning_rate": 4.559118891774188e-05,
      "loss": 1.0299,
      "step": 1795
    },
    {
      "epoch": 0.9642629441965035,
      "grad_norm": 0.76171875,
      "learning_rate": 4.5586398486138435e-05,
      "loss": 1.0001,
      "step": 1796
    },
    {
      "epoch": 0.9647998389315795,
      "grad_norm": 0.8984375,
      "learning_rate": 4.5581605705366224e-05,
      "loss": 1.1626,
      "step": 1797
    },
    {
      "epoch": 0.9653367336666555,
      "grad_norm": 0.72265625,
      "learning_rate": 4.5576810575972146e-05,
      "loss": 0.8722,
      "step": 1798
    },
    {
      "epoch": 0.9658736284017315,
      "grad_norm": 0.85546875,
      "learning_rate": 4.55720130985034e-05,
      "loss": 0.9075,
      "step": 1799
    },
    {
      "epoch": 0.9664105231368075,
      "grad_norm": 0.77734375,
      "learning_rate": 4.5567213273507435e-05,
      "loss": 1.0448,
      "step": 1800
    },
    {
      "epoch": 0.9664105231368075,
      "eval_loss": 0.9234041571617126,
      "eval_runtime": 46.0021,
      "eval_samples_per_second": 20.912,
      "eval_steps_per_second": 20.912,
      "step": 1800
    },
    {
      "epoch": 0.9669474178718835,
      "grad_norm": 0.76953125,
      "learning_rate": 4.5562411101531975e-05,
      "loss": 0.8338,
      "step": 1801
    },
    {
      "epoch": 0.9674843126069596,
      "grad_norm": 0.703125,
      "learning_rate": 4.555760658312501e-05,
      "loss": 0.8609,
      "step": 1802
    },
    {
      "epoch": 0.9680212073420355,
      "grad_norm": 0.80859375,
      "learning_rate": 4.555279971883478e-05,
      "loss": 0.8369,
      "step": 1803
    },
    {
      "epoch": 0.9685581020771115,
      "grad_norm": 0.76171875,
      "learning_rate": 4.554799050920984e-05,
      "loss": 0.871,
      "step": 1804
    },
    {
      "epoch": 0.9690949968121875,
      "grad_norm": 0.71875,
      "learning_rate": 4.554317895479895e-05,
      "loss": 1.0161,
      "step": 1805
    },
    {
      "epoch": 0.9696318915472635,
      "grad_norm": 0.625,
      "learning_rate": 4.553836505615119e-05,
      "loss": 0.8452,
      "step": 1806
    },
    {
      "epoch": 0.9701687862823395,
      "grad_norm": 0.93359375,
      "learning_rate": 4.553354881381587e-05,
      "loss": 1.1926,
      "step": 1807
    },
    {
      "epoch": 0.9707056810174155,
      "grad_norm": 0.7421875,
      "learning_rate": 4.5528730228342605e-05,
      "loss": 0.8598,
      "step": 1808
    },
    {
      "epoch": 0.9712425757524915,
      "grad_norm": 1.0234375,
      "learning_rate": 4.552390930028124e-05,
      "loss": 1.0521,
      "step": 1809
    },
    {
      "epoch": 0.9717794704875675,
      "grad_norm": 0.734375,
      "learning_rate": 4.551908603018191e-05,
      "loss": 1.0746,
      "step": 1810
    },
    {
      "epoch": 0.9723163652226435,
      "grad_norm": 1.0546875,
      "learning_rate": 4.551426041859501e-05,
      "loss": 1.1368,
      "step": 1811
    },
    {
      "epoch": 0.9728532599577195,
      "grad_norm": 0.65234375,
      "learning_rate": 4.550943246607121e-05,
      "loss": 0.942,
      "step": 1812
    },
    {
      "epoch": 0.9733901546927956,
      "grad_norm": 0.65234375,
      "learning_rate": 4.550460217316143e-05,
      "loss": 0.8976,
      "step": 1813
    },
    {
      "epoch": 0.9739270494278716,
      "grad_norm": 0.59375,
      "learning_rate": 4.549976954041687e-05,
      "loss": 0.7012,
      "step": 1814
    },
    {
      "epoch": 0.9744639441629476,
      "grad_norm": 0.74609375,
      "learning_rate": 4.5494934568389e-05,
      "loss": 0.7313,
      "step": 1815
    },
    {
      "epoch": 0.9750008388980236,
      "grad_norm": 0.70703125,
      "learning_rate": 4.549009725762955e-05,
      "loss": 0.8401,
      "step": 1816
    },
    {
      "epoch": 0.9755377336330996,
      "grad_norm": 0.67578125,
      "learning_rate": 4.548525760869051e-05,
      "loss": 0.9733,
      "step": 1817
    },
    {
      "epoch": 0.9760746283681756,
      "grad_norm": 0.6484375,
      "learning_rate": 4.5480415622124165e-05,
      "loss": 0.898,
      "step": 1818
    },
    {
      "epoch": 0.9766115231032516,
      "grad_norm": 0.75,
      "learning_rate": 4.547557129848302e-05,
      "loss": 0.7641,
      "step": 1819
    },
    {
      "epoch": 0.9771484178383276,
      "grad_norm": 0.6875,
      "learning_rate": 4.547072463831989e-05,
      "loss": 0.8549,
      "step": 1820
    },
    {
      "epoch": 0.9776853125734036,
      "grad_norm": 0.765625,
      "learning_rate": 4.546587564218785e-05,
      "loss": 0.9291,
      "step": 1821
    },
    {
      "epoch": 0.9782222073084795,
      "grad_norm": 0.58203125,
      "learning_rate": 4.546102431064021e-05,
      "loss": 0.9247,
      "step": 1822
    },
    {
      "epoch": 0.9787591020435555,
      "grad_norm": 0.78515625,
      "learning_rate": 4.545617064423058e-05,
      "loss": 0.8991,
      "step": 1823
    },
    {
      "epoch": 0.9792959967786316,
      "grad_norm": 0.7265625,
      "learning_rate": 4.545131464351282e-05,
      "loss": 1.0273,
      "step": 1824
    },
    {
      "epoch": 0.9798328915137076,
      "grad_norm": 0.91015625,
      "learning_rate": 4.544645630904106e-05,
      "loss": 0.7825,
      "step": 1825
    },
    {
      "epoch": 0.9803697862487836,
      "grad_norm": 0.671875,
      "learning_rate": 4.544159564136972e-05,
      "loss": 1.179,
      "step": 1826
    },
    {
      "epoch": 0.9809066809838596,
      "grad_norm": 0.80078125,
      "learning_rate": 4.543673264105343e-05,
      "loss": 0.8172,
      "step": 1827
    },
    {
      "epoch": 0.9814435757189356,
      "grad_norm": 0.6484375,
      "learning_rate": 4.5431867308647135e-05,
      "loss": 0.8477,
      "step": 1828
    },
    {
      "epoch": 0.9819804704540116,
      "grad_norm": 0.671875,
      "learning_rate": 4.5426999644706044e-05,
      "loss": 0.9885,
      "step": 1829
    },
    {
      "epoch": 0.9825173651890876,
      "grad_norm": 0.6328125,
      "learning_rate": 4.54221296497856e-05,
      "loss": 0.9193,
      "step": 1830
    },
    {
      "epoch": 0.9830542599241636,
      "grad_norm": 0.6796875,
      "learning_rate": 4.541725732444153e-05,
      "loss": 0.9058,
      "step": 1831
    },
    {
      "epoch": 0.9835911546592396,
      "grad_norm": 0.61328125,
      "learning_rate": 4.541238266922984e-05,
      "loss": 0.8006,
      "step": 1832
    },
    {
      "epoch": 0.9841280493943156,
      "grad_norm": 0.66796875,
      "learning_rate": 4.540750568470679e-05,
      "loss": 0.9181,
      "step": 1833
    },
    {
      "epoch": 0.9846649441293917,
      "grad_norm": 0.75,
      "learning_rate": 4.54026263714289e-05,
      "loss": 0.9358,
      "step": 1834
    },
    {
      "epoch": 0.9852018388644677,
      "grad_norm": 0.67578125,
      "learning_rate": 4.5397744729952964e-05,
      "loss": 0.8277,
      "step": 1835
    },
    {
      "epoch": 0.9857387335995437,
      "grad_norm": 0.765625,
      "learning_rate": 4.539286076083603e-05,
      "loss": 0.8649,
      "step": 1836
    },
    {
      "epoch": 0.9862756283346197,
      "grad_norm": 0.76171875,
      "learning_rate": 4.538797446463543e-05,
      "loss": 1.0171,
      "step": 1837
    },
    {
      "epoch": 0.9868125230696957,
      "grad_norm": 0.640625,
      "learning_rate": 4.538308584190876e-05,
      "loss": 0.9048,
      "step": 1838
    },
    {
      "epoch": 0.9873494178047717,
      "grad_norm": 0.796875,
      "learning_rate": 4.537819489321386e-05,
      "loss": 0.9797,
      "step": 1839
    },
    {
      "epoch": 0.9878863125398477,
      "grad_norm": 0.66015625,
      "learning_rate": 4.5373301619108854e-05,
      "loss": 0.7941,
      "step": 1840
    },
    {
      "epoch": 0.9884232072749236,
      "grad_norm": 0.76953125,
      "learning_rate": 4.536840602015212e-05,
      "loss": 0.8707,
      "step": 1841
    },
    {
      "epoch": 0.9889601020099996,
      "grad_norm": 0.7578125,
      "learning_rate": 4.536350809690232e-05,
      "loss": 1.0523,
      "step": 1842
    },
    {
      "epoch": 0.9894969967450756,
      "grad_norm": 0.62109375,
      "learning_rate": 4.5358607849918364e-05,
      "loss": 0.7851,
      "step": 1843
    },
    {
      "epoch": 0.9900338914801516,
      "grad_norm": 0.6875,
      "learning_rate": 4.535370527975943e-05,
      "loss": 0.9351,
      "step": 1844
    },
    {
      "epoch": 0.9905707862152276,
      "grad_norm": 0.66015625,
      "learning_rate": 4.5348800386984954e-05,
      "loss": 0.7522,
      "step": 1845
    },
    {
      "epoch": 0.9911076809503037,
      "grad_norm": 0.8828125,
      "learning_rate": 4.5343893172154675e-05,
      "loss": 1.1531,
      "step": 1846
    },
    {
      "epoch": 0.9916445756853797,
      "grad_norm": 0.6953125,
      "learning_rate": 4.533898363582854e-05,
      "loss": 0.9912,
      "step": 1847
    },
    {
      "epoch": 0.9921814704204557,
      "grad_norm": 0.80078125,
      "learning_rate": 4.5334071778566804e-05,
      "loss": 0.9189,
      "step": 1848
    },
    {
      "epoch": 0.9927183651555317,
      "grad_norm": 0.93359375,
      "learning_rate": 4.5329157600929966e-05,
      "loss": 1.0364,
      "step": 1849
    },
    {
      "epoch": 0.9932552598906077,
      "grad_norm": 0.63671875,
      "learning_rate": 4.5324241103478804e-05,
      "loss": 0.8382,
      "step": 1850
    },
    {
      "epoch": 0.9932552598906077,
      "eval_loss": 0.9214339256286621,
      "eval_runtime": 45.6257,
      "eval_samples_per_second": 21.085,
      "eval_steps_per_second": 21.085,
      "step": 1850
    },
    {
      "epoch": 0.9937921546256837,
      "grad_norm": 0.69140625,
      "learning_rate": 4.531932228677433e-05,
      "loss": 1.0326,
      "step": 1851
    },
    {
      "epoch": 0.9943290493607597,
      "grad_norm": 0.73828125,
      "learning_rate": 4.5314401151377876e-05,
      "loss": 0.9046,
      "step": 1852
    },
    {
      "epoch": 0.9948659440958357,
      "grad_norm": 0.6328125,
      "learning_rate": 4.530947769785099e-05,
      "loss": 0.9507,
      "step": 1853
    },
    {
      "epoch": 0.9954028388309117,
      "grad_norm": 0.7890625,
      "learning_rate": 4.530455192675549e-05,
      "loss": 1.0793,
      "step": 1854
    },
    {
      "epoch": 0.9959397335659878,
      "grad_norm": 0.8125,
      "learning_rate": 4.529962383865348e-05,
      "loss": 1.0272,
      "step": 1855
    },
    {
      "epoch": 0.9964766283010638,
      "grad_norm": 0.78125,
      "learning_rate": 4.529469343410732e-05,
      "loss": 1.0442,
      "step": 1856
    },
    {
      "epoch": 0.9970135230361398,
      "grad_norm": 0.734375,
      "learning_rate": 4.528976071367963e-05,
      "loss": 0.8845,
      "step": 1857
    },
    {
      "epoch": 0.9975504177712158,
      "grad_norm": 0.6875,
      "learning_rate": 4.528482567793329e-05,
      "loss": 0.9361,
      "step": 1858
    },
    {
      "epoch": 0.9980873125062918,
      "grad_norm": 0.71484375,
      "learning_rate": 4.5279888327431455e-05,
      "loss": 0.9168,
      "step": 1859
    },
    {
      "epoch": 0.9986242072413677,
      "grad_norm": 0.69921875,
      "learning_rate": 4.527494866273753e-05,
      "loss": 0.8991,
      "step": 1860
    },
    {
      "epoch": 0.9991611019764437,
      "grad_norm": 0.73828125,
      "learning_rate": 4.527000668441521e-05,
      "loss": 0.8749,
      "step": 1861
    },
    {
      "epoch": 0.9996979967115197,
      "grad_norm": 0.69921875,
      "learning_rate": 4.526506239302842e-05,
      "loss": 0.8481,
      "step": 1862
    },
    {
      "epoch": 1.000536894735076,
      "grad_norm": 0.80859375,
      "learning_rate": 4.526011578914139e-05,
      "loss": 1.45,
      "step": 1863
    },
    {
      "epoch": 1.001073789470152,
      "grad_norm": 0.68359375,
      "learning_rate": 4.525516687331857e-05,
      "loss": 0.8831,
      "step": 1864
    },
    {
      "epoch": 1.001610684205228,
      "grad_norm": 0.65234375,
      "learning_rate": 4.5250215646124684e-05,
      "loss": 0.9929,
      "step": 1865
    },
    {
      "epoch": 1.002147578940304,
      "grad_norm": 0.78515625,
      "learning_rate": 4.524526210812475e-05,
      "loss": 0.9428,
      "step": 1866
    },
    {
      "epoch": 1.00268447367538,
      "grad_norm": 0.83203125,
      "learning_rate": 4.524030625988404e-05,
      "loss": 0.9506,
      "step": 1867
    },
    {
      "epoch": 1.003221368410456,
      "grad_norm": 0.69921875,
      "learning_rate": 4.5235348101968044e-05,
      "loss": 1.1667,
      "step": 1868
    },
    {
      "epoch": 1.003758263145532,
      "grad_norm": 0.68359375,
      "learning_rate": 4.523038763494258e-05,
      "loss": 0.9324,
      "step": 1869
    },
    {
      "epoch": 1.004295157880608,
      "grad_norm": 0.734375,
      "learning_rate": 4.522542485937369e-05,
      "loss": 0.8907,
      "step": 1870
    },
    {
      "epoch": 1.004832052615684,
      "grad_norm": 0.6953125,
      "learning_rate": 4.5220459775827685e-05,
      "loss": 0.9064,
      "step": 1871
    },
    {
      "epoch": 1.00536894735076,
      "grad_norm": 0.69921875,
      "learning_rate": 4.521549238487115e-05,
      "loss": 0.8501,
      "step": 1872
    },
    {
      "epoch": 1.005905842085836,
      "grad_norm": 0.69140625,
      "learning_rate": 4.521052268707092e-05,
      "loss": 0.8544,
      "step": 1873
    },
    {
      "epoch": 1.0064427368209121,
      "grad_norm": 0.6484375,
      "learning_rate": 4.520555068299411e-05,
      "loss": 0.8082,
      "step": 1874
    },
    {
      "epoch": 1.006979631555988,
      "grad_norm": 0.56640625,
      "learning_rate": 4.5200576373208093e-05,
      "loss": 0.6351,
      "step": 1875
    },
    {
      "epoch": 1.0075165262910641,
      "grad_norm": 0.73828125,
      "learning_rate": 4.5195599758280485e-05,
      "loss": 1.1155,
      "step": 1876
    },
    {
      "epoch": 1.00805342102614,
      "grad_norm": 0.703125,
      "learning_rate": 4.519062083877918e-05,
      "loss": 0.9415,
      "step": 1877
    },
    {
      "epoch": 1.0085903157612162,
      "grad_norm": 0.64453125,
      "learning_rate": 4.5185639615272356e-05,
      "loss": 0.7706,
      "step": 1878
    },
    {
      "epoch": 1.009127210496292,
      "grad_norm": 0.75390625,
      "learning_rate": 4.518065608832842e-05,
      "loss": 0.8954,
      "step": 1879
    },
    {
      "epoch": 1.0096641052313682,
      "grad_norm": 0.80859375,
      "learning_rate": 4.517567025851605e-05,
      "loss": 0.985,
      "step": 1880
    },
    {
      "epoch": 1.010200999966444,
      "grad_norm": 0.58203125,
      "learning_rate": 4.5170682126404206e-05,
      "loss": 0.8717,
      "step": 1881
    },
    {
      "epoch": 1.01073789470152,
      "grad_norm": 0.73046875,
      "learning_rate": 4.5165691692562085e-05,
      "loss": 0.8408,
      "step": 1882
    },
    {
      "epoch": 1.011274789436596,
      "grad_norm": 0.78125,
      "learning_rate": 4.5160698957559164e-05,
      "loss": 0.9268,
      "step": 1883
    },
    {
      "epoch": 1.011811684171672,
      "grad_norm": 0.63671875,
      "learning_rate": 4.515570392196518e-05,
      "loss": 0.9071,
      "step": 1884
    },
    {
      "epoch": 1.0123485789067481,
      "grad_norm": 0.87109375,
      "learning_rate": 4.515070658635013e-05,
      "loss": 1.1588,
      "step": 1885
    },
    {
      "epoch": 1.012885473641824,
      "grad_norm": 0.6953125,
      "learning_rate": 4.5145706951284265e-05,
      "loss": 0.9855,
      "step": 1886
    },
    {
      "epoch": 1.0134223683769001,
      "grad_norm": 0.6484375,
      "learning_rate": 4.5140705017338114e-05,
      "loss": 0.8237,
      "step": 1887
    },
    {
      "epoch": 1.013959263111976,
      "grad_norm": 0.8203125,
      "learning_rate": 4.513570078508246e-05,
      "loss": 0.9666,
      "step": 1888
    },
    {
      "epoch": 1.0144961578470522,
      "grad_norm": 0.76953125,
      "learning_rate": 4.5130694255088344e-05,
      "loss": 0.911,
      "step": 1889
    },
    {
      "epoch": 1.015033052582128,
      "grad_norm": 0.58984375,
      "learning_rate": 4.512568542792709e-05,
      "loss": 0.7198,
      "step": 1890
    },
    {
      "epoch": 1.0155699473172042,
      "grad_norm": 0.89453125,
      "learning_rate": 4.5120674304170255e-05,
      "loss": 0.9902,
      "step": 1891
    },
    {
      "epoch": 1.01610684205228,
      "grad_norm": 0.73828125,
      "learning_rate": 4.5115660884389665e-05,
      "loss": 1.0889,
      "step": 1892
    },
    {
      "epoch": 1.0166437367873562,
      "grad_norm": 0.71484375,
      "learning_rate": 4.511064516915743e-05,
      "loss": 1.0408,
      "step": 1893
    },
    {
      "epoch": 1.017180631522432,
      "grad_norm": 0.71875,
      "learning_rate": 4.510562715904589e-05,
      "loss": 0.8262,
      "step": 1894
    },
    {
      "epoch": 1.0177175262575082,
      "grad_norm": 0.63671875,
      "learning_rate": 4.510060685462769e-05,
      "loss": 0.8045,
      "step": 1895
    },
    {
      "epoch": 1.0182544209925841,
      "grad_norm": 0.734375,
      "learning_rate": 4.5095584256475686e-05,
      "loss": 0.9082,
      "step": 1896
    },
    {
      "epoch": 1.0187913157276602,
      "grad_norm": 0.890625,
      "learning_rate": 4.509055936516303e-05,
      "loss": 0.9124,
      "step": 1897
    },
    {
      "epoch": 1.0193282104627361,
      "grad_norm": 0.78515625,
      "learning_rate": 4.508553218126312e-05,
      "loss": 0.9423,
      "step": 1898
    },
    {
      "epoch": 1.0198651051978123,
      "grad_norm": 0.76953125,
      "learning_rate": 4.508050270534963e-05,
      "loss": 1.0588,
      "step": 1899
    },
    {
      "epoch": 1.0204019999328882,
      "grad_norm": 0.7421875,
      "learning_rate": 4.507547093799648e-05,
      "loss": 0.7418,
      "step": 1900
    },
    {
      "epoch": 1.0204019999328882,
      "eval_loss": 0.9220516085624695,
      "eval_runtime": 46.2004,
      "eval_samples_per_second": 20.822,
      "eval_steps_per_second": 20.822,
      "step": 1900
    },
    {
      "epoch": 1.020938894667964,
      "grad_norm": 0.73046875,
      "learning_rate": 4.5070436879777865e-05,
      "loss": 0.6648,
      "step": 1901
    },
    {
      "epoch": 1.0214757894030402,
      "grad_norm": 0.71484375,
      "learning_rate": 4.506540053126823e-05,
      "loss": 0.8462,
      "step": 1902
    },
    {
      "epoch": 1.022012684138116,
      "grad_norm": 0.6640625,
      "learning_rate": 4.506036189304228e-05,
      "loss": 0.7025,
      "step": 1903
    },
    {
      "epoch": 1.0225495788731922,
      "grad_norm": 0.75390625,
      "learning_rate": 4.5055320965675e-05,
      "loss": 0.8031,
      "step": 1904
    },
    {
      "epoch": 1.023086473608268,
      "grad_norm": 0.640625,
      "learning_rate": 4.505027774974162e-05,
      "loss": 0.7238,
      "step": 1905
    },
    {
      "epoch": 1.0236233683433442,
      "grad_norm": 0.76171875,
      "learning_rate": 4.504523224581762e-05,
      "loss": 0.8339,
      "step": 1906
    },
    {
      "epoch": 1.0241602630784201,
      "grad_norm": 0.6953125,
      "learning_rate": 4.5040184454478786e-05,
      "loss": 0.812,
      "step": 1907
    },
    {
      "epoch": 1.0246971578134962,
      "grad_norm": 1.0234375,
      "learning_rate": 4.50351343763011e-05,
      "loss": 0.9015,
      "step": 1908
    },
    {
      "epoch": 1.0252340525485721,
      "grad_norm": 0.7578125,
      "learning_rate": 4.503008201186087e-05,
      "loss": 0.8438,
      "step": 1909
    },
    {
      "epoch": 1.0257709472836483,
      "grad_norm": 0.734375,
      "learning_rate": 4.502502736173462e-05,
      "loss": 1.1629,
      "step": 1910
    },
    {
      "epoch": 1.0263078420187242,
      "grad_norm": 0.62109375,
      "learning_rate": 4.5019970426499145e-05,
      "loss": 0.8084,
      "step": 1911
    },
    {
      "epoch": 1.0268447367538003,
      "grad_norm": 0.671875,
      "learning_rate": 4.5014911206731515e-05,
      "loss": 0.8435,
      "step": 1912
    },
    {
      "epoch": 1.0273816314888762,
      "grad_norm": 0.67578125,
      "learning_rate": 4.500984970300905e-05,
      "loss": 0.9267,
      "step": 1913
    },
    {
      "epoch": 1.0279185262239523,
      "grad_norm": 0.859375,
      "learning_rate": 4.500478591590933e-05,
      "loss": 0.8635,
      "step": 1914
    },
    {
      "epoch": 1.0284554209590282,
      "grad_norm": 0.7578125,
      "learning_rate": 4.49997198460102e-05,
      "loss": 0.9659,
      "step": 1915
    },
    {
      "epoch": 1.0289923156941043,
      "grad_norm": 0.7109375,
      "learning_rate": 4.499465149388975e-05,
      "loss": 0.9466,
      "step": 1916
    },
    {
      "epoch": 1.0295292104291802,
      "grad_norm": 0.62109375,
      "learning_rate": 4.4989580860126354e-05,
      "loss": 0.9998,
      "step": 1917
    },
    {
      "epoch": 1.0300661051642563,
      "grad_norm": 0.6640625,
      "learning_rate": 4.4984507945298644e-05,
      "loss": 0.8697,
      "step": 1918
    },
    {
      "epoch": 1.0306029998993322,
      "grad_norm": 0.671875,
      "learning_rate": 4.497943274998549e-05,
      "loss": 0.8562,
      "step": 1919
    },
    {
      "epoch": 1.0311398946344081,
      "grad_norm": 0.7421875,
      "learning_rate": 4.4974355274766034e-05,
      "loss": 0.8562,
      "step": 1920
    },
    {
      "epoch": 1.0316767893694843,
      "grad_norm": 0.6875,
      "learning_rate": 4.496927552021969e-05,
      "loss": 0.8777,
      "step": 1921
    },
    {
      "epoch": 1.0322136841045602,
      "grad_norm": 0.765625,
      "learning_rate": 4.496419348692612e-05,
      "loss": 0.8184,
      "step": 1922
    },
    {
      "epoch": 1.0327505788396363,
      "grad_norm": 0.76953125,
      "learning_rate": 4.495910917546525e-05,
      "loss": 0.9376,
      "step": 1923
    },
    {
      "epoch": 1.0332874735747122,
      "grad_norm": 0.88671875,
      "learning_rate": 4.495402258641726e-05,
      "loss": 0.8625,
      "step": 1924
    },
    {
      "epoch": 1.0338243683097883,
      "grad_norm": 0.703125,
      "learning_rate": 4.494893372036259e-05,
      "loss": 0.8018,
      "step": 1925
    },
    {
      "epoch": 1.0343612630448642,
      "grad_norm": 0.76171875,
      "learning_rate": 4.4943842577881957e-05,
      "loss": 0.8103,
      "step": 1926
    },
    {
      "epoch": 1.0348981577799403,
      "grad_norm": 1.0,
      "learning_rate": 4.493874915955631e-05,
      "loss": 0.7037,
      "step": 1927
    },
    {
      "epoch": 1.0354350525150162,
      "grad_norm": 0.66796875,
      "learning_rate": 4.493365346596688e-05,
      "loss": 0.7914,
      "step": 1928
    },
    {
      "epoch": 1.0359719472500923,
      "grad_norm": 0.71875,
      "learning_rate": 4.492855549769516e-05,
      "loss": 0.8135,
      "step": 1929
    },
    {
      "epoch": 1.0365088419851682,
      "grad_norm": 0.6796875,
      "learning_rate": 4.492345525532288e-05,
      "loss": 0.7385,
      "step": 1930
    },
    {
      "epoch": 1.0370457367202444,
      "grad_norm": 0.65234375,
      "learning_rate": 4.4918352739432043e-05,
      "loss": 0.8067,
      "step": 1931
    },
    {
      "epoch": 1.0375826314553203,
      "grad_norm": 0.62890625,
      "learning_rate": 4.491324795060491e-05,
      "loss": 0.8124,
      "step": 1932
    },
    {
      "epoch": 1.0381195261903964,
      "grad_norm": 0.63671875,
      "learning_rate": 4.490814088942401e-05,
      "loss": 0.7838,
      "step": 1933
    },
    {
      "epoch": 1.0386564209254723,
      "grad_norm": 0.703125,
      "learning_rate": 4.4903031556472117e-05,
      "loss": 0.7945,
      "step": 1934
    },
    {
      "epoch": 1.0391933156605484,
      "grad_norm": 0.80078125,
      "learning_rate": 4.489791995233227e-05,
      "loss": 1.0002,
      "step": 1935
    },
    {
      "epoch": 1.0397302103956243,
      "grad_norm": 0.67578125,
      "learning_rate": 4.4892806077587765e-05,
      "loss": 0.8675,
      "step": 1936
    },
    {
      "epoch": 1.0402671051307002,
      "grad_norm": 0.703125,
      "learning_rate": 4.4887689932822176e-05,
      "loss": 0.9476,
      "step": 1937
    },
    {
      "epoch": 1.0408039998657763,
      "grad_norm": 0.6171875,
      "learning_rate": 4.48825715186193e-05,
      "loss": 0.81,
      "step": 1938
    },
    {
      "epoch": 1.0413408946008522,
      "grad_norm": 0.6953125,
      "learning_rate": 4.4877450835563215e-05,
      "loss": 0.8288,
      "step": 1939
    },
    {
      "epoch": 1.0418777893359283,
      "grad_norm": 0.84765625,
      "learning_rate": 4.4872327884238274e-05,
      "loss": 1.0518,
      "step": 1940
    },
    {
      "epoch": 1.0424146840710042,
      "grad_norm": 0.69140625,
      "learning_rate": 4.486720266522906e-05,
      "loss": 0.7991,
      "step": 1941
    },
    {
      "epoch": 1.0429515788060804,
      "grad_norm": 0.73828125,
      "learning_rate": 4.486207517912042e-05,
      "loss": 0.9036,
      "step": 1942
    },
    {
      "epoch": 1.0434884735411563,
      "grad_norm": 0.76953125,
      "learning_rate": 4.4856945426497464e-05,
      "loss": 0.9259,
      "step": 1943
    },
    {
      "epoch": 1.0440253682762324,
      "grad_norm": 0.68359375,
      "learning_rate": 4.4851813407945574e-05,
      "loss": 0.8469,
      "step": 1944
    },
    {
      "epoch": 1.0445622630113083,
      "grad_norm": 0.67578125,
      "learning_rate": 4.4846679124050375e-05,
      "loss": 0.9435,
      "step": 1945
    },
    {
      "epoch": 1.0450991577463844,
      "grad_norm": 0.69921875,
      "learning_rate": 4.484154257539774e-05,
      "loss": 0.8936,
      "step": 1946
    },
    {
      "epoch": 1.0456360524814603,
      "grad_norm": 0.578125,
      "learning_rate": 4.4836403762573826e-05,
      "loss": 0.779,
      "step": 1947
    },
    {
      "epoch": 1.0461729472165364,
      "grad_norm": 0.7109375,
      "learning_rate": 4.483126268616504e-05,
      "loss": 1.0763,
      "step": 1948
    },
    {
      "epoch": 1.0467098419516123,
      "grad_norm": 0.6640625,
      "learning_rate": 4.4826119346758036e-05,
      "loss": 1.0075,
      "step": 1949
    },
    {
      "epoch": 1.0472467366866884,
      "grad_norm": 0.73046875,
      "learning_rate": 4.482097374493974e-05,
      "loss": 0.8152,
      "step": 1950
    },
    {
      "epoch": 1.0472467366866884,
      "eval_loss": 0.9233726263046265,
      "eval_runtime": 45.8573,
      "eval_samples_per_second": 20.978,
      "eval_steps_per_second": 20.978,
      "step": 1950
    },
    {
      "epoch": 1.0477836314217643,
      "grad_norm": 0.87890625,
      "learning_rate": 4.481582588129732e-05,
      "loss": 0.9823,
      "step": 1951
    },
    {
      "epoch": 1.0483205261568405,
      "grad_norm": 0.6953125,
      "learning_rate": 4.481067575641822e-05,
      "loss": 0.9266,
      "step": 1952
    },
    {
      "epoch": 1.0488574208919164,
      "grad_norm": 0.71484375,
      "learning_rate": 4.480552337089015e-05,
      "loss": 0.7821,
      "step": 1953
    },
    {
      "epoch": 1.0493943156269925,
      "grad_norm": 0.64453125,
      "learning_rate": 4.480036872530103e-05,
      "loss": 0.9949,
      "step": 1954
    },
    {
      "epoch": 1.0499312103620684,
      "grad_norm": 0.76171875,
      "learning_rate": 4.479521182023909e-05,
      "loss": 1.0427,
      "step": 1955
    },
    {
      "epoch": 1.0504681050971443,
      "grad_norm": 0.86328125,
      "learning_rate": 4.47900526562928e-05,
      "loss": 0.8266,
      "step": 1956
    },
    {
      "epoch": 1.0510049998322204,
      "grad_norm": 0.703125,
      "learning_rate": 4.478489123405088e-05,
      "loss": 0.9953,
      "step": 1957
    },
    {
      "epoch": 1.0515418945672963,
      "grad_norm": 0.93359375,
      "learning_rate": 4.477972755410231e-05,
      "loss": 0.8112,
      "step": 1958
    },
    {
      "epoch": 1.0520787893023724,
      "grad_norm": 0.64453125,
      "learning_rate": 4.477456161703634e-05,
      "loss": 0.7595,
      "step": 1959
    },
    {
      "epoch": 1.0526156840374483,
      "grad_norm": 0.65625,
      "learning_rate": 4.476939342344246e-05,
      "loss": 0.9713,
      "step": 1960
    },
    {
      "epoch": 1.0531525787725244,
      "grad_norm": 0.6875,
      "learning_rate": 4.476422297391044e-05,
      "loss": 0.864,
      "step": 1961
    },
    {
      "epoch": 1.0536894735076003,
      "grad_norm": 0.578125,
      "learning_rate": 4.475905026903028e-05,
      "loss": 0.7046,
      "step": 1962
    },
    {
      "epoch": 1.0542263682426765,
      "grad_norm": 0.68359375,
      "learning_rate": 4.4753875309392266e-05,
      "loss": 1.0367,
      "step": 1963
    },
    {
      "epoch": 1.0547632629777524,
      "grad_norm": 0.66015625,
      "learning_rate": 4.4748698095586904e-05,
      "loss": 0.8485,
      "step": 1964
    },
    {
      "epoch": 1.0553001577128285,
      "grad_norm": 0.625,
      "learning_rate": 4.474351862820499e-05,
      "loss": 0.7381,
      "step": 1965
    },
    {
      "epoch": 1.0558370524479044,
      "grad_norm": 0.80859375,
      "learning_rate": 4.4738336907837584e-05,
      "loss": 1.0235,
      "step": 1966
    },
    {
      "epoch": 1.0563739471829805,
      "grad_norm": 0.6875,
      "learning_rate": 4.473315293507596e-05,
      "loss": 0.886,
      "step": 1967
    },
    {
      "epoch": 1.0569108419180564,
      "grad_norm": 0.7109375,
      "learning_rate": 4.472796671051169e-05,
      "loss": 0.9203,
      "step": 1968
    },
    {
      "epoch": 1.0574477366531325,
      "grad_norm": 0.671875,
      "learning_rate": 4.472277823473658e-05,
      "loss": 0.8866,
      "step": 1969
    },
    {
      "epoch": 1.0579846313882084,
      "grad_norm": 0.68359375,
      "learning_rate": 4.471758750834271e-05,
      "loss": 0.8197,
      "step": 1970
    },
    {
      "epoch": 1.0585215261232845,
      "grad_norm": 0.71484375,
      "learning_rate": 4.47123945319224e-05,
      "loss": 0.805,
      "step": 1971
    },
    {
      "epoch": 1.0590584208583604,
      "grad_norm": 0.5859375,
      "learning_rate": 4.470719930606823e-05,
      "loss": 0.6962,
      "step": 1972
    },
    {
      "epoch": 1.0595953155934366,
      "grad_norm": 0.69140625,
      "learning_rate": 4.470200183137305e-05,
      "loss": 0.9446,
      "step": 1973
    },
    {
      "epoch": 1.0601322103285125,
      "grad_norm": 0.796875,
      "learning_rate": 4.469680210842996e-05,
      "loss": 0.7789,
      "step": 1974
    },
    {
      "epoch": 1.0606691050635884,
      "grad_norm": 0.65625,
      "learning_rate": 4.4691600137832316e-05,
      "loss": 0.8901,
      "step": 1975
    },
    {
      "epoch": 1.0612059997986645,
      "grad_norm": 0.62109375,
      "learning_rate": 4.4686395920173706e-05,
      "loss": 0.8904,
      "step": 1976
    },
    {
      "epoch": 1.0617428945337404,
      "grad_norm": 0.703125,
      "learning_rate": 4.4681189456048026e-05,
      "loss": 0.9188,
      "step": 1977
    },
    {
      "epoch": 1.0622797892688165,
      "grad_norm": 0.69921875,
      "learning_rate": 4.4675980746049384e-05,
      "loss": 1.0461,
      "step": 1978
    },
    {
      "epoch": 1.0628166840038924,
      "grad_norm": 0.76953125,
      "learning_rate": 4.467076979077216e-05,
      "loss": 1.0348,
      "step": 1979
    },
    {
      "epoch": 1.0633535787389685,
      "grad_norm": 0.61328125,
      "learning_rate": 4.4665556590811e-05,
      "loss": 0.9061,
      "step": 1980
    },
    {
      "epoch": 1.0638904734740444,
      "grad_norm": 0.71875,
      "learning_rate": 4.466034114676077e-05,
      "loss": 0.7636,
      "step": 1981
    },
    {
      "epoch": 1.0644273682091205,
      "grad_norm": 0.6640625,
      "learning_rate": 4.4655123459216664e-05,
      "loss": 0.9381,
      "step": 1982
    },
    {
      "epoch": 1.0649642629441964,
      "grad_norm": 0.60546875,
      "learning_rate": 4.464990352877404e-05,
      "loss": 0.7239,
      "step": 1983
    },
    {
      "epoch": 1.0655011576792726,
      "grad_norm": 0.7421875,
      "learning_rate": 4.4644681356028586e-05,
      "loss": 0.9538,
      "step": 1984
    },
    {
      "epoch": 1.0660380524143485,
      "grad_norm": 0.71484375,
      "learning_rate": 4.463945694157621e-05,
      "loss": 0.9682,
      "step": 1985
    },
    {
      "epoch": 1.0665749471494246,
      "grad_norm": 0.765625,
      "learning_rate": 4.463423028601308e-05,
      "loss": 0.8489,
      "step": 1986
    },
    {
      "epoch": 1.0671118418845005,
      "grad_norm": 0.69921875,
      "learning_rate": 4.462900138993562e-05,
      "loss": 0.9274,
      "step": 1987
    },
    {
      "epoch": 1.0676487366195766,
      "grad_norm": 0.765625,
      "learning_rate": 4.462377025394053e-05,
      "loss": 0.848,
      "step": 1988
    },
    {
      "epoch": 1.0681856313546525,
      "grad_norm": 0.77734375,
      "learning_rate": 4.461853687862474e-05,
      "loss": 1.1715,
      "step": 1989
    },
    {
      "epoch": 1.0687225260897286,
      "grad_norm": 0.69140625,
      "learning_rate": 4.461330126458544e-05,
      "loss": 0.898,
      "step": 1990
    },
    {
      "epoch": 1.0692594208248045,
      "grad_norm": 1.0859375,
      "learning_rate": 4.4608063412420074e-05,
      "loss": 0.8517,
      "step": 1991
    },
    {
      "epoch": 1.0697963155598806,
      "grad_norm": 0.671875,
      "learning_rate": 4.460282332272637e-05,
      "loss": 0.7949,
      "step": 1992
    },
    {
      "epoch": 1.0703332102949565,
      "grad_norm": 0.87890625,
      "learning_rate": 4.4597580996102276e-05,
      "loss": 1.0467,
      "step": 1993
    },
    {
      "epoch": 1.0708701050300324,
      "grad_norm": 0.7578125,
      "learning_rate": 4.4592336433146e-05,
      "loss": 1.0597,
      "step": 1994
    },
    {
      "epoch": 1.0714069997651086,
      "grad_norm": 0.66015625,
      "learning_rate": 4.458708963445602e-05,
      "loss": 0.8162,
      "step": 1995
    },
    {
      "epoch": 1.0719438945001845,
      "grad_norm": 0.7421875,
      "learning_rate": 4.4581840600631066e-05,
      "loss": 0.7769,
      "step": 1996
    },
    {
      "epoch": 1.0724807892352606,
      "grad_norm": 0.5703125,
      "learning_rate": 4.457658933227011e-05,
      "loss": 0.8472,
      "step": 1997
    },
    {
      "epoch": 1.0730176839703365,
      "grad_norm": 0.859375,
      "learning_rate": 4.4571335829972396e-05,
      "loss": 0.8712,
      "step": 1998
    },
    {
      "epoch": 1.0735545787054126,
      "grad_norm": 0.69921875,
      "learning_rate": 4.456608009433741e-05,
      "loss": 0.9897,
      "step": 1999
    },
    {
      "epoch": 1.0740914734404885,
      "grad_norm": 0.71484375,
      "learning_rate": 4.4560822125964915e-05,
      "loss": 0.888,
      "step": 2000
    },
    {
      "epoch": 1.0740914734404885,
      "eval_loss": 0.9224899411201477,
      "eval_runtime": 46.5225,
      "eval_samples_per_second": 20.678,
      "eval_steps_per_second": 20.678,
      "step": 2000
    },
    {
      "epoch": 1.0746283681755646,
      "grad_norm": 0.66015625,
      "learning_rate": 4.455556192545489e-05,
      "loss": 0.7956,
      "step": 2001
    },
    {
      "epoch": 1.0751652629106405,
      "grad_norm": 0.7890625,
      "learning_rate": 4.455029949340759e-05,
      "loss": 1.0419,
      "step": 2002
    },
    {
      "epoch": 1.0757021576457166,
      "grad_norm": 0.7734375,
      "learning_rate": 4.454503483042354e-05,
      "loss": 1.0803,
      "step": 2003
    },
    {
      "epoch": 1.0762390523807925,
      "grad_norm": 0.6953125,
      "learning_rate": 4.4539767937103496e-05,
      "loss": 0.7567,
      "step": 2004
    },
    {
      "epoch": 1.0767759471158687,
      "grad_norm": 0.625,
      "learning_rate": 4.453449881404849e-05,
      "loss": 0.829,
      "step": 2005
    },
    {
      "epoch": 1.0773128418509446,
      "grad_norm": 0.59765625,
      "learning_rate": 4.4529227461859766e-05,
      "loss": 0.8505,
      "step": 2006
    },
    {
      "epoch": 1.0778497365860207,
      "grad_norm": 0.85546875,
      "learning_rate": 4.4523953881138885e-05,
      "loss": 0.8171,
      "step": 2007
    },
    {
      "epoch": 1.0783866313210966,
      "grad_norm": 0.640625,
      "learning_rate": 4.451867807248761e-05,
      "loss": 0.8978,
      "step": 2008
    },
    {
      "epoch": 1.0789235260561727,
      "grad_norm": 0.609375,
      "learning_rate": 4.451340003650799e-05,
      "loss": 0.8383,
      "step": 2009
    },
    {
      "epoch": 1.0794604207912486,
      "grad_norm": 0.7109375,
      "learning_rate": 4.45081197738023e-05,
      "loss": 0.8685,
      "step": 2010
    },
    {
      "epoch": 1.0799973155263247,
      "grad_norm": 0.69921875,
      "learning_rate": 4.45028372849731e-05,
      "loss": 0.8838,
      "step": 2011
    },
    {
      "epoch": 1.0805342102614006,
      "grad_norm": 0.76953125,
      "learning_rate": 4.449755257062317e-05,
      "loss": 0.7649,
      "step": 2012
    },
    {
      "epoch": 1.0810711049964765,
      "grad_norm": 0.671875,
      "learning_rate": 4.449226563135559e-05,
      "loss": 0.9033,
      "step": 2013
    },
    {
      "epoch": 1.0816079997315526,
      "grad_norm": 0.765625,
      "learning_rate": 4.448697646777364e-05,
      "loss": 1.1965,
      "step": 2014
    },
    {
      "epoch": 1.0821448944666285,
      "grad_norm": 0.71875,
      "learning_rate": 4.44816850804809e-05,
      "loss": 0.7399,
      "step": 2015
    },
    {
      "epoch": 1.0826817892017047,
      "grad_norm": 0.76171875,
      "learning_rate": 4.447639147008117e-05,
      "loss": 0.8932,
      "step": 2016
    },
    {
      "epoch": 1.0832186839367806,
      "grad_norm": 0.67578125,
      "learning_rate": 4.447109563717853e-05,
      "loss": 0.8015,
      "step": 2017
    },
    {
      "epoch": 1.0837555786718567,
      "grad_norm": 0.76953125,
      "learning_rate": 4.44657975823773e-05,
      "loss": 0.801,
      "step": 2018
    },
    {
      "epoch": 1.0842924734069326,
      "grad_norm": 0.85546875,
      "learning_rate": 4.446049730628204e-05,
      "loss": 0.9231,
      "step": 2019
    },
    {
      "epoch": 1.0848293681420087,
      "grad_norm": 0.69140625,
      "learning_rate": 4.44551948094976e-05,
      "loss": 0.816,
      "step": 2020
    },
    {
      "epoch": 1.0853662628770846,
      "grad_norm": 0.70703125,
      "learning_rate": 4.4449890092629045e-05,
      "loss": 0.8183,
      "step": 2021
    },
    {
      "epoch": 1.0859031576121607,
      "grad_norm": 0.73828125,
      "learning_rate": 4.4444583156281724e-05,
      "loss": 0.9396,
      "step": 2022
    },
    {
      "epoch": 1.0864400523472366,
      "grad_norm": 0.76171875,
      "learning_rate": 4.443927400106122e-05,
      "loss": 0.9589,
      "step": 2023
    },
    {
      "epoch": 1.0869769470823127,
      "grad_norm": 0.65234375,
      "learning_rate": 4.443396262757337e-05,
      "loss": 1.0206,
      "step": 2024
    },
    {
      "epoch": 1.0875138418173886,
      "grad_norm": 0.73828125,
      "learning_rate": 4.442864903642428e-05,
      "loss": 0.9336,
      "step": 2025
    },
    {
      "epoch": 1.0880507365524648,
      "grad_norm": 0.67578125,
      "learning_rate": 4.442333322822028e-05,
      "loss": 0.956,
      "step": 2026
    },
    {
      "epoch": 1.0885876312875407,
      "grad_norm": 0.70703125,
      "learning_rate": 4.4418015203567986e-05,
      "loss": 0.9395,
      "step": 2027
    },
    {
      "epoch": 1.0891245260226168,
      "grad_norm": 0.7265625,
      "learning_rate": 4.441269496307425e-05,
      "loss": 0.9294,
      "step": 2028
    },
    {
      "epoch": 1.0896614207576927,
      "grad_norm": 0.7734375,
      "learning_rate": 4.440737250734618e-05,
      "loss": 0.957,
      "step": 2029
    },
    {
      "epoch": 1.0901983154927688,
      "grad_norm": 0.71875,
      "learning_rate": 4.440204783699112e-05,
      "loss": 0.9635,
      "step": 2030
    },
    {
      "epoch": 1.0907352102278447,
      "grad_norm": 1.3046875,
      "learning_rate": 4.4396720952616705e-05,
      "loss": 0.9689,
      "step": 2031
    },
    {
      "epoch": 1.0912721049629206,
      "grad_norm": 0.7890625,
      "learning_rate": 4.4391391854830785e-05,
      "loss": 0.834,
      "step": 2032
    },
    {
      "epoch": 1.0918089996979967,
      "grad_norm": 0.73828125,
      "learning_rate": 4.4386060544241493e-05,
      "loss": 1.1758,
      "step": 2033
    },
    {
      "epoch": 1.0923458944330726,
      "grad_norm": 0.7421875,
      "learning_rate": 4.438072702145718e-05,
      "loss": 0.9533,
      "step": 2034
    },
    {
      "epoch": 1.0928827891681487,
      "grad_norm": 0.7890625,
      "learning_rate": 4.437539128708647e-05,
      "loss": 0.7748,
      "step": 2035
    },
    {
      "epoch": 1.0934196839032246,
      "grad_norm": 0.72265625,
      "learning_rate": 4.437005334173826e-05,
      "loss": 0.9824,
      "step": 2036
    },
    {
      "epoch": 1.0939565786383008,
      "grad_norm": 0.73046875,
      "learning_rate": 4.436471318602164e-05,
      "loss": 0.9566,
      "step": 2037
    },
    {
      "epoch": 1.0944934733733767,
      "grad_norm": 0.75390625,
      "learning_rate": 4.435937082054603e-05,
      "loss": 0.9278,
      "step": 2038
    },
    {
      "epoch": 1.0950303681084528,
      "grad_norm": 0.74609375,
      "learning_rate": 4.4354026245921034e-05,
      "loss": 0.8977,
      "step": 2039
    },
    {
      "epoch": 1.0955672628435287,
      "grad_norm": 0.97265625,
      "learning_rate": 4.4348679462756556e-05,
      "loss": 0.9137,
      "step": 2040
    },
    {
      "epoch": 1.0961041575786048,
      "grad_norm": 0.8515625,
      "learning_rate": 4.434333047166271e-05,
      "loss": 0.9121,
      "step": 2041
    },
    {
      "epoch": 1.0966410523136807,
      "grad_norm": 0.65234375,
      "learning_rate": 4.43379792732499e-05,
      "loss": 0.931,
      "step": 2042
    },
    {
      "epoch": 1.0971779470487568,
      "grad_norm": 0.68359375,
      "learning_rate": 4.4332625868128755e-05,
      "loss": 0.8208,
      "step": 2043
    },
    {
      "epoch": 1.0977148417838327,
      "grad_norm": 0.671875,
      "learning_rate": 4.4327270256910164e-05,
      "loss": 0.9443,
      "step": 2044
    },
    {
      "epoch": 1.0982517365189088,
      "grad_norm": 0.7421875,
      "learning_rate": 4.4321912440205285e-05,
      "loss": 0.9097,
      "step": 2045
    },
    {
      "epoch": 1.0987886312539847,
      "grad_norm": 0.72265625,
      "learning_rate": 4.43165524186255e-05,
      "loss": 0.9636,
      "step": 2046
    },
    {
      "epoch": 1.0993255259890609,
      "grad_norm": 0.7734375,
      "learning_rate": 4.431119019278246e-05,
      "loss": 1.0085,
      "step": 2047
    },
    {
      "epoch": 1.0998624207241368,
      "grad_norm": 0.640625,
      "learning_rate": 4.430582576328807e-05,
      "loss": 0.7751,
      "step": 2048
    },
    {
      "epoch": 1.1003993154592129,
      "grad_norm": 0.6875,
      "learning_rate": 4.4300459130754465e-05,
      "loss": 0.7163,
      "step": 2049
    },
    {
      "epoch": 1.1009362101942888,
      "grad_norm": 0.6328125,
      "learning_rate": 4.429509029579405e-05,
      "loss": 0.762,
      "step": 2050
    },
    {
      "epoch": 1.1009362101942888,
      "eval_loss": 0.9212960004806519,
      "eval_runtime": 46.7543,
      "eval_samples_per_second": 20.576,
      "eval_steps_per_second": 20.576,
      "step": 2050
    },
    {
      "epoch": 1.1014731049293647,
      "grad_norm": 0.67578125,
      "learning_rate": 4.428971925901949e-05,
      "loss": 0.9842,
      "step": 2051
    },
    {
      "epoch": 1.1020099996644408,
      "grad_norm": 0.6953125,
      "learning_rate": 4.4284346021043664e-05,
      "loss": 0.8665,
      "step": 2052
    },
    {
      "epoch": 1.1025468943995167,
      "grad_norm": 0.60546875,
      "learning_rate": 4.427897058247975e-05,
      "loss": 0.6928,
      "step": 2053
    },
    {
      "epoch": 1.1030837891345928,
      "grad_norm": 0.9609375,
      "learning_rate": 4.427359294394115e-05,
      "loss": 0.8716,
      "step": 2054
    },
    {
      "epoch": 1.1036206838696687,
      "grad_norm": 0.703125,
      "learning_rate": 4.42682131060415e-05,
      "loss": 1.015,
      "step": 2055
    },
    {
      "epoch": 1.1041575786047448,
      "grad_norm": 0.65234375,
      "learning_rate": 4.426283106939474e-05,
      "loss": 0.8309,
      "step": 2056
    },
    {
      "epoch": 1.1046944733398207,
      "grad_norm": 0.69921875,
      "learning_rate": 4.4257446834615e-05,
      "loss": 0.8908,
      "step": 2057
    },
    {
      "epoch": 1.1052313680748969,
      "grad_norm": 0.80078125,
      "learning_rate": 4.4252060402316696e-05,
      "loss": 1.2144,
      "step": 2058
    },
    {
      "epoch": 1.1057682628099728,
      "grad_norm": 0.6640625,
      "learning_rate": 4.4246671773114495e-05,
      "loss": 0.7307,
      "step": 2059
    },
    {
      "epoch": 1.1063051575450489,
      "grad_norm": 0.79296875,
      "learning_rate": 4.424128094762331e-05,
      "loss": 1.0203,
      "step": 2060
    },
    {
      "epoch": 1.1068420522801248,
      "grad_norm": 0.76953125,
      "learning_rate": 4.42358879264583e-05,
      "loss": 0.9436,
      "step": 2061
    },
    {
      "epoch": 1.107378947015201,
      "grad_norm": 0.73046875,
      "learning_rate": 4.4230492710234864e-05,
      "loss": 0.8664,
      "step": 2062
    },
    {
      "epoch": 1.1079158417502768,
      "grad_norm": 0.734375,
      "learning_rate": 4.422509529956868e-05,
      "loss": 0.8325,
      "step": 2063
    },
    {
      "epoch": 1.108452736485353,
      "grad_norm": 0.765625,
      "learning_rate": 4.4219695695075656e-05,
      "loss": 0.9997,
      "step": 2064
    },
    {
      "epoch": 1.1089896312204288,
      "grad_norm": 0.65625,
      "learning_rate": 4.421429389737196e-05,
      "loss": 0.8273,
      "step": 2065
    },
    {
      "epoch": 1.109526525955505,
      "grad_norm": 0.6796875,
      "learning_rate": 4.4208889907074e-05,
      "loss": 0.7933,
      "step": 2066
    },
    {
      "epoch": 1.1100634206905808,
      "grad_norm": 0.671875,
      "learning_rate": 4.420348372479844e-05,
      "loss": 0.8302,
      "step": 2067
    },
    {
      "epoch": 1.110600315425657,
      "grad_norm": 0.60546875,
      "learning_rate": 4.419807535116219e-05,
      "loss": 0.8421,
      "step": 2068
    },
    {
      "epoch": 1.1111372101607329,
      "grad_norm": 0.640625,
      "learning_rate": 4.4192664786782426e-05,
      "loss": 0.8186,
      "step": 2069
    },
    {
      "epoch": 1.1116741048958088,
      "grad_norm": 0.75,
      "learning_rate": 4.418725203227655e-05,
      "loss": 1.0375,
      "step": 2070
    },
    {
      "epoch": 1.1122109996308849,
      "grad_norm": 0.703125,
      "learning_rate": 4.418183708826223e-05,
      "loss": 0.7499,
      "step": 2071
    },
    {
      "epoch": 1.1127478943659608,
      "grad_norm": 0.74609375,
      "learning_rate": 4.417641995535739e-05,
      "loss": 1.141,
      "step": 2072
    },
    {
      "epoch": 1.113284789101037,
      "grad_norm": 0.7109375,
      "learning_rate": 4.417100063418016e-05,
      "loss": 0.779,
      "step": 2073
    },
    {
      "epoch": 1.1138216838361128,
      "grad_norm": 0.64453125,
      "learning_rate": 4.4165579125348996e-05,
      "loss": 0.862,
      "step": 2074
    },
    {
      "epoch": 1.114358578571189,
      "grad_norm": 0.9375,
      "learning_rate": 4.416015542948253e-05,
      "loss": 0.9814,
      "step": 2075
    },
    {
      "epoch": 1.1148954733062648,
      "grad_norm": 0.73046875,
      "learning_rate": 4.4154729547199695e-05,
      "loss": 0.996,
      "step": 2076
    },
    {
      "epoch": 1.115432368041341,
      "grad_norm": 0.7421875,
      "learning_rate": 4.4149301479119643e-05,
      "loss": 1.0847,
      "step": 2077
    },
    {
      "epoch": 1.1159692627764168,
      "grad_norm": 0.73046875,
      "learning_rate": 4.414387122586178e-05,
      "loss": 0.891,
      "step": 2078
    },
    {
      "epoch": 1.116506157511493,
      "grad_norm": 0.65234375,
      "learning_rate": 4.413843878804578e-05,
      "loss": 0.8898,
      "step": 2079
    },
    {
      "epoch": 1.1170430522465689,
      "grad_norm": 0.74609375,
      "learning_rate": 4.413300416629154e-05,
      "loss": 1.0265,
      "step": 2080
    },
    {
      "epoch": 1.117579946981645,
      "grad_norm": 0.765625,
      "learning_rate": 4.4127567361219224e-05,
      "loss": 0.9772,
      "step": 2081
    },
    {
      "epoch": 1.1181168417167209,
      "grad_norm": 0.7890625,
      "learning_rate": 4.412212837344924e-05,
      "loss": 0.9532,
      "step": 2082
    },
    {
      "epoch": 1.118653736451797,
      "grad_norm": 0.74609375,
      "learning_rate": 4.411668720360225e-05,
      "loss": 0.9018,
      "step": 2083
    },
    {
      "epoch": 1.119190631186873,
      "grad_norm": 0.78125,
      "learning_rate": 4.411124385229916e-05,
      "loss": 0.8392,
      "step": 2084
    },
    {
      "epoch": 1.119727525921949,
      "grad_norm": 0.83984375,
      "learning_rate": 4.4105798320161115e-05,
      "loss": 0.9466,
      "step": 2085
    },
    {
      "epoch": 1.120264420657025,
      "grad_norm": 0.64453125,
      "learning_rate": 4.410035060780953e-05,
      "loss": 0.8188,
      "step": 2086
    },
    {
      "epoch": 1.120801315392101,
      "grad_norm": 0.74609375,
      "learning_rate": 4.4094900715866064e-05,
      "loss": 0.8948,
      "step": 2087
    },
    {
      "epoch": 1.121338210127177,
      "grad_norm": 0.65625,
      "learning_rate": 4.4089448644952593e-05,
      "loss": 0.9638,
      "step": 2088
    },
    {
      "epoch": 1.1218751048622528,
      "grad_norm": 0.6484375,
      "learning_rate": 4.408399439569129e-05,
      "loss": 0.9848,
      "step": 2089
    },
    {
      "epoch": 1.122411999597329,
      "grad_norm": 0.6640625,
      "learning_rate": 4.407853796870455e-05,
      "loss": 0.9337,
      "step": 2090
    },
    {
      "epoch": 1.1229488943324049,
      "grad_norm": 0.6953125,
      "learning_rate": 4.407307936461501e-05,
      "loss": 0.8408,
      "step": 2091
    },
    {
      "epoch": 1.123485789067481,
      "grad_norm": 0.73046875,
      "learning_rate": 4.4067618584045576e-05,
      "loss": 0.8009,
      "step": 2092
    },
    {
      "epoch": 1.1240226838025569,
      "grad_norm": 0.62109375,
      "learning_rate": 4.406215562761939e-05,
      "loss": 0.8012,
      "step": 2093
    },
    {
      "epoch": 1.124559578537633,
      "grad_norm": 0.60546875,
      "learning_rate": 4.4056690495959855e-05,
      "loss": 0.6881,
      "step": 2094
    },
    {
      "epoch": 1.125096473272709,
      "grad_norm": 0.703125,
      "learning_rate": 4.405122318969058e-05,
      "loss": 0.6926,
      "step": 2095
    },
    {
      "epoch": 1.125633368007785,
      "grad_norm": 0.71875,
      "learning_rate": 4.404575370943549e-05,
      "loss": 0.9392,
      "step": 2096
    },
    {
      "epoch": 1.126170262742861,
      "grad_norm": 0.62890625,
      "learning_rate": 4.40402820558187e-05,
      "loss": 0.7936,
      "step": 2097
    },
    {
      "epoch": 1.126707157477937,
      "grad_norm": 0.59765625,
      "learning_rate": 4.403480822946461e-05,
      "loss": 0.7749,
      "step": 2098
    },
    {
      "epoch": 1.127244052213013,
      "grad_norm": 0.72265625,
      "learning_rate": 4.402933223099784e-05,
      "loss": 1.0051,
      "step": 2099
    },
    {
      "epoch": 1.127780946948089,
      "grad_norm": 0.6328125,
      "learning_rate": 4.402385406104327e-05,
      "loss": 0.8621,
      "step": 2100
    },
    {
      "epoch": 1.127780946948089,
      "eval_loss": 0.9210036396980286,
      "eval_runtime": 47.3848,
      "eval_samples_per_second": 20.302,
      "eval_steps_per_second": 20.302,
      "step": 2100
    },
    {
      "epoch": 1.128317841683165,
      "grad_norm": 0.72265625,
      "learning_rate": 4.401837372022604e-05,
      "loss": 0.8931,
      "step": 2101
    },
    {
      "epoch": 1.128854736418241,
      "grad_norm": 0.62109375,
      "learning_rate": 4.401289120917152e-05,
      "loss": 0.7536,
      "step": 2102
    },
    {
      "epoch": 1.129391631153317,
      "grad_norm": 0.765625,
      "learning_rate": 4.400740652850534e-05,
      "loss": 1.0879,
      "step": 2103
    },
    {
      "epoch": 1.129928525888393,
      "grad_norm": 0.70703125,
      "learning_rate": 4.400191967885335e-05,
      "loss": 1.0175,
      "step": 2104
    },
    {
      "epoch": 1.130465420623469,
      "grad_norm": 0.765625,
      "learning_rate": 4.3996430660841695e-05,
      "loss": 0.8159,
      "step": 2105
    },
    {
      "epoch": 1.1310023153585451,
      "grad_norm": 0.921875,
      "learning_rate": 4.399093947509674e-05,
      "loss": 0.8421,
      "step": 2106
    },
    {
      "epoch": 1.131539210093621,
      "grad_norm": 0.75390625,
      "learning_rate": 4.3985446122245076e-05,
      "loss": 0.9083,
      "step": 2107
    },
    {
      "epoch": 1.132076104828697,
      "grad_norm": 0.64453125,
      "learning_rate": 4.3979950602913585e-05,
      "loss": 0.8521,
      "step": 2108
    },
    {
      "epoch": 1.132612999563773,
      "grad_norm": 0.73046875,
      "learning_rate": 4.3974452917729366e-05,
      "loss": 0.86,
      "step": 2109
    },
    {
      "epoch": 1.133149894298849,
      "grad_norm": 0.66015625,
      "learning_rate": 4.3968953067319777e-05,
      "loss": 1.0138,
      "step": 2110
    },
    {
      "epoch": 1.133686789033925,
      "grad_norm": 0.74609375,
      "learning_rate": 4.396345105231242e-05,
      "loss": 0.9405,
      "step": 2111
    },
    {
      "epoch": 1.134223683769001,
      "grad_norm": 0.64453125,
      "learning_rate": 4.3957946873335144e-05,
      "loss": 0.7999,
      "step": 2112
    },
    {
      "epoch": 1.134760578504077,
      "grad_norm": 0.76953125,
      "learning_rate": 4.3952440531016046e-05,
      "loss": 0.7411,
      "step": 2113
    },
    {
      "epoch": 1.135297473239153,
      "grad_norm": 0.71484375,
      "learning_rate": 4.394693202598348e-05,
      "loss": 1.024,
      "step": 2114
    },
    {
      "epoch": 1.135834367974229,
      "grad_norm": 0.609375,
      "learning_rate": 4.3941421358866016e-05,
      "loss": 0.8786,
      "step": 2115
    },
    {
      "epoch": 1.136371262709305,
      "grad_norm": 0.76171875,
      "learning_rate": 4.393590853029251e-05,
      "loss": 0.9797,
      "step": 2116
    },
    {
      "epoch": 1.1369081574443811,
      "grad_norm": 0.71875,
      "learning_rate": 4.393039354089203e-05,
      "loss": 0.7747,
      "step": 2117
    },
    {
      "epoch": 1.137445052179457,
      "grad_norm": 0.5546875,
      "learning_rate": 4.3924876391293915e-05,
      "loss": 0.7416,
      "step": 2118
    },
    {
      "epoch": 1.1379819469145331,
      "grad_norm": 0.65625,
      "learning_rate": 4.391935708212774e-05,
      "loss": 0.9829,
      "step": 2119
    },
    {
      "epoch": 1.138518841649609,
      "grad_norm": 0.8515625,
      "learning_rate": 4.391383561402333e-05,
      "loss": 0.8999,
      "step": 2120
    },
    {
      "epoch": 1.1390557363846852,
      "grad_norm": 0.71875,
      "learning_rate": 4.390831198761074e-05,
      "loss": 0.936,
      "step": 2121
    },
    {
      "epoch": 1.139592631119761,
      "grad_norm": 0.6953125,
      "learning_rate": 4.390278620352031e-05,
      "loss": 0.8469,
      "step": 2122
    },
    {
      "epoch": 1.1401295258548372,
      "grad_norm": 0.74609375,
      "learning_rate": 4.389725826238259e-05,
      "loss": 1.0217,
      "step": 2123
    },
    {
      "epoch": 1.140666420589913,
      "grad_norm": 0.6796875,
      "learning_rate": 4.3891728164828385e-05,
      "loss": 0.8167,
      "step": 2124
    },
    {
      "epoch": 1.1412033153249892,
      "grad_norm": 0.75,
      "learning_rate": 4.3886195911488746e-05,
      "loss": 0.9418,
      "step": 2125
    },
    {
      "epoch": 1.141740210060065,
      "grad_norm": 0.69921875,
      "learning_rate": 4.3880661502994986e-05,
      "loss": 0.8184,
      "step": 2126
    },
    {
      "epoch": 1.142277104795141,
      "grad_norm": 0.75390625,
      "learning_rate": 4.387512493997864e-05,
      "loss": 0.9536,
      "step": 2127
    },
    {
      "epoch": 1.1428139995302171,
      "grad_norm": 0.68359375,
      "learning_rate": 4.386958622307151e-05,
      "loss": 0.8705,
      "step": 2128
    },
    {
      "epoch": 1.143350894265293,
      "grad_norm": 0.734375,
      "learning_rate": 4.386404535290563e-05,
      "loss": 1.0281,
      "step": 2129
    },
    {
      "epoch": 1.1438877890003691,
      "grad_norm": 0.703125,
      "learning_rate": 4.385850233011327e-05,
      "loss": 0.942,
      "step": 2130
    },
    {
      "epoch": 1.144424683735445,
      "grad_norm": 0.70703125,
      "learning_rate": 4.385295715532698e-05,
      "loss": 0.922,
      "step": 2131
    },
    {
      "epoch": 1.1449615784705212,
      "grad_norm": 0.80859375,
      "learning_rate": 4.384740982917952e-05,
      "loss": 0.9384,
      "step": 2132
    },
    {
      "epoch": 1.145498473205597,
      "grad_norm": 0.66015625,
      "learning_rate": 4.384186035230392e-05,
      "loss": 1.0251,
      "step": 2133
    },
    {
      "epoch": 1.1460353679406732,
      "grad_norm": 0.79296875,
      "learning_rate": 4.383630872533344e-05,
      "loss": 1.1454,
      "step": 2134
    },
    {
      "epoch": 1.146572262675749,
      "grad_norm": 0.98828125,
      "learning_rate": 4.383075494890159e-05,
      "loss": 1.096,
      "step": 2135
    },
    {
      "epoch": 1.1471091574108252,
      "grad_norm": 0.828125,
      "learning_rate": 4.3825199023642125e-05,
      "loss": 0.8909,
      "step": 2136
    },
    {
      "epoch": 1.147646052145901,
      "grad_norm": 0.65625,
      "learning_rate": 4.381964095018906e-05,
      "loss": 0.8381,
      "step": 2137
    },
    {
      "epoch": 1.1481829468809772,
      "grad_norm": 0.71484375,
      "learning_rate": 4.3814080729176624e-05,
      "loss": 0.8422,
      "step": 2138
    },
    {
      "epoch": 1.1487198416160531,
      "grad_norm": 0.6640625,
      "learning_rate": 4.380851836123931e-05,
      "loss": 0.8447,
      "step": 2139
    },
    {
      "epoch": 1.1492567363511292,
      "grad_norm": 0.83984375,
      "learning_rate": 4.3802953847011865e-05,
      "loss": 0.9898,
      "step": 2140
    },
    {
      "epoch": 1.1497936310862051,
      "grad_norm": 0.671875,
      "learning_rate": 4.379738718712927e-05,
      "loss": 0.9763,
      "step": 2141
    },
    {
      "epoch": 1.1503305258212813,
      "grad_norm": 0.65234375,
      "learning_rate": 4.379181838222675e-05,
      "loss": 0.919,
      "step": 2142
    },
    {
      "epoch": 1.1508674205563572,
      "grad_norm": 0.6640625,
      "learning_rate": 4.378624743293976e-05,
      "loss": 0.7736,
      "step": 2143
    },
    {
      "epoch": 1.1514043152914333,
      "grad_norm": 0.640625,
      "learning_rate": 4.378067433990404e-05,
      "loss": 0.9296,
      "step": 2144
    },
    {
      "epoch": 1.1519412100265092,
      "grad_norm": 0.8125,
      "learning_rate": 4.377509910375554e-05,
      "loss": 0.8183,
      "step": 2145
    },
    {
      "epoch": 1.152478104761585,
      "grad_norm": 0.70703125,
      "learning_rate": 4.376952172513046e-05,
      "loss": 0.9463,
      "step": 2146
    },
    {
      "epoch": 1.1530149994966612,
      "grad_norm": 0.63671875,
      "learning_rate": 4.3763942204665255e-05,
      "loss": 0.8834,
      "step": 2147
    },
    {
      "epoch": 1.153551894231737,
      "grad_norm": 0.7890625,
      "learning_rate": 4.3758360542996625e-05,
      "loss": 0.9232,
      "step": 2148
    },
    {
      "epoch": 1.1540887889668132,
      "grad_norm": 1.0390625,
      "learning_rate": 4.375277674076149e-05,
      "loss": 1.0551,
      "step": 2149
    },
    {
      "epoch": 1.1546256837018891,
      "grad_norm": 0.89453125,
      "learning_rate": 4.374719079859705e-05,
      "loss": 0.9048,
      "step": 2150
    },
    {
      "epoch": 1.1546256837018891,
      "eval_loss": 0.9199674129486084,
      "eval_runtime": 46.1504,
      "eval_samples_per_second": 20.845,
      "eval_steps_per_second": 20.845,
      "step": 2150
    },
    {
      "epoch": 1.1551625784369652,
      "grad_norm": 0.7890625,
      "learning_rate": 4.374160271714073e-05,
      "loss": 1.0633,
      "step": 2151
    },
    {
      "epoch": 1.1556994731720411,
      "grad_norm": 0.66015625,
      "learning_rate": 4.3736012497030194e-05,
      "loss": 0.8048,
      "step": 2152
    },
    {
      "epoch": 1.1562363679071173,
      "grad_norm": 1.078125,
      "learning_rate": 4.373042013890335e-05,
      "loss": 1.0484,
      "step": 2153
    },
    {
      "epoch": 1.1567732626421932,
      "grad_norm": 0.74609375,
      "learning_rate": 4.372482564339838e-05,
      "loss": 0.8295,
      "step": 2154
    },
    {
      "epoch": 1.1573101573772693,
      "grad_norm": 0.7109375,
      "learning_rate": 4.3719229011153675e-05,
      "loss": 0.8643,
      "step": 2155
    },
    {
      "epoch": 1.1578470521123452,
      "grad_norm": 0.82421875,
      "learning_rate": 4.3713630242807866e-05,
      "loss": 1.0369,
      "step": 2156
    },
    {
      "epoch": 1.1583839468474213,
      "grad_norm": 0.69140625,
      "learning_rate": 4.3708029338999867e-05,
      "loss": 0.8695,
      "step": 2157
    },
    {
      "epoch": 1.1589208415824972,
      "grad_norm": 0.88671875,
      "learning_rate": 4.37024263003688e-05,
      "loss": 0.8015,
      "step": 2158
    },
    {
      "epoch": 1.1594577363175733,
      "grad_norm": 0.5546875,
      "learning_rate": 4.369682112755405e-05,
      "loss": 0.8047,
      "step": 2159
    },
    {
      "epoch": 1.1599946310526492,
      "grad_norm": 0.76171875,
      "learning_rate": 4.369121382119523e-05,
      "loss": 0.7613,
      "step": 2160
    },
    {
      "epoch": 1.1605315257877253,
      "grad_norm": 0.59765625,
      "learning_rate": 4.36856043819322e-05,
      "loss": 0.72,
      "step": 2161
    },
    {
      "epoch": 1.1610684205228012,
      "grad_norm": 0.7109375,
      "learning_rate": 4.367999281040509e-05,
      "loss": 1.0082,
      "step": 2162
    },
    {
      "epoch": 1.1616053152578774,
      "grad_norm": 0.671875,
      "learning_rate": 4.3674379107254225e-05,
      "loss": 0.9587,
      "step": 2163
    },
    {
      "epoch": 1.1621422099929533,
      "grad_norm": 0.64453125,
      "learning_rate": 4.3668763273120214e-05,
      "loss": 1.1291,
      "step": 2164
    },
    {
      "epoch": 1.1626791047280292,
      "grad_norm": 0.68359375,
      "learning_rate": 4.366314530864391e-05,
      "loss": 0.9716,
      "step": 2165
    },
    {
      "epoch": 1.1632159994631053,
      "grad_norm": 0.83984375,
      "learning_rate": 4.365752521446635e-05,
      "loss": 1.017,
      "step": 2166
    },
    {
      "epoch": 1.1637528941981812,
      "grad_norm": 0.8515625,
      "learning_rate": 4.36519029912289e-05,
      "loss": 0.7967,
      "step": 2167
    },
    {
      "epoch": 1.1642897889332573,
      "grad_norm": 0.78515625,
      "learning_rate": 4.364627863957311e-05,
      "loss": 0.8033,
      "step": 2168
    },
    {
      "epoch": 1.1648266836683332,
      "grad_norm": 0.7578125,
      "learning_rate": 4.364065216014079e-05,
      "loss": 0.9961,
      "step": 2169
    },
    {
      "epoch": 1.1653635784034093,
      "grad_norm": 0.69921875,
      "learning_rate": 4.363502355357399e-05,
      "loss": 0.9891,
      "step": 2170
    },
    {
      "epoch": 1.1659004731384852,
      "grad_norm": 0.9140625,
      "learning_rate": 4.362939282051501e-05,
      "loss": 0.9474,
      "step": 2171
    },
    {
      "epoch": 1.1664373678735613,
      "grad_norm": 0.765625,
      "learning_rate": 4.362375996160639e-05,
      "loss": 0.9568,
      "step": 2172
    },
    {
      "epoch": 1.1669742626086372,
      "grad_norm": 0.71484375,
      "learning_rate": 4.3618124977490916e-05,
      "loss": 0.8744,
      "step": 2173
    },
    {
      "epoch": 1.1675111573437134,
      "grad_norm": 0.6796875,
      "learning_rate": 4.361248786881159e-05,
      "loss": 0.9549,
      "step": 2174
    },
    {
      "epoch": 1.1680480520787893,
      "grad_norm": 0.70703125,
      "learning_rate": 4.3606848636211693e-05,
      "loss": 0.79,
      "step": 2175
    },
    {
      "epoch": 1.1685849468138654,
      "grad_norm": 0.65625,
      "learning_rate": 4.3601207280334733e-05,
      "loss": 0.9945,
      "step": 2176
    },
    {
      "epoch": 1.1691218415489413,
      "grad_norm": 0.6875,
      "learning_rate": 4.359556380182446e-05,
      "loss": 0.8591,
      "step": 2177
    },
    {
      "epoch": 1.1696587362840174,
      "grad_norm": 0.65625,
      "learning_rate": 4.3589918201324854e-05,
      "loss": 0.807,
      "step": 2178
    },
    {
      "epoch": 1.1701956310190933,
      "grad_norm": 0.5859375,
      "learning_rate": 4.358427047948016e-05,
      "loss": 0.807,
      "step": 2179
    },
    {
      "epoch": 1.1707325257541694,
      "grad_norm": 0.6875,
      "learning_rate": 4.357862063693486e-05,
      "loss": 0.8825,
      "step": 2180
    },
    {
      "epoch": 1.1712694204892453,
      "grad_norm": 0.69140625,
      "learning_rate": 4.357296867433366e-05,
      "loss": 0.9605,
      "step": 2181
    },
    {
      "epoch": 1.1718063152243214,
      "grad_norm": 0.671875,
      "learning_rate": 4.356731459232153e-05,
      "loss": 0.7999,
      "step": 2182
    },
    {
      "epoch": 1.1723432099593973,
      "grad_norm": 0.8828125,
      "learning_rate": 4.356165839154367e-05,
      "loss": 0.9473,
      "step": 2183
    },
    {
      "epoch": 1.1728801046944732,
      "grad_norm": 0.70703125,
      "learning_rate": 4.355600007264552e-05,
      "loss": 0.9375,
      "step": 2184
    },
    {
      "epoch": 1.1734169994295494,
      "grad_norm": 0.62109375,
      "learning_rate": 4.3550339636272775e-05,
      "loss": 0.8811,
      "step": 2185
    },
    {
      "epoch": 1.1739538941646253,
      "grad_norm": 0.6171875,
      "learning_rate": 4.354467708307135e-05,
      "loss": 0.8012,
      "step": 2186
    },
    {
      "epoch": 1.1744907888997014,
      "grad_norm": 0.8359375,
      "learning_rate": 4.353901241368742e-05,
      "loss": 1.0285,
      "step": 2187
    },
    {
      "epoch": 1.1750276836347773,
      "grad_norm": 0.93359375,
      "learning_rate": 4.35333456287674e-05,
      "loss": 1.191,
      "step": 2188
    },
    {
      "epoch": 1.1755645783698534,
      "grad_norm": 0.83984375,
      "learning_rate": 4.352767672895793e-05,
      "loss": 0.9894,
      "step": 2189
    },
    {
      "epoch": 1.1761014731049293,
      "grad_norm": 0.703125,
      "learning_rate": 4.352200571490592e-05,
      "loss": 0.8701,
      "step": 2190
    },
    {
      "epoch": 1.1766383678400054,
      "grad_norm": 0.69921875,
      "learning_rate": 4.35163325872585e-05,
      "loss": 0.9078,
      "step": 2191
    },
    {
      "epoch": 1.1771752625750813,
      "grad_norm": 0.66796875,
      "learning_rate": 4.351065734666303e-05,
      "loss": 0.8073,
      "step": 2192
    },
    {
      "epoch": 1.1777121573101574,
      "grad_norm": 0.70703125,
      "learning_rate": 4.350497999376715e-05,
      "loss": 0.8385,
      "step": 2193
    },
    {
      "epoch": 1.1782490520452333,
      "grad_norm": 0.7578125,
      "learning_rate": 4.34993005292187e-05,
      "loss": 0.9369,
      "step": 2194
    },
    {
      "epoch": 1.1787859467803095,
      "grad_norm": 0.6875,
      "learning_rate": 4.3493618953665785e-05,
      "loss": 0.8755,
      "step": 2195
    },
    {
      "epoch": 1.1793228415153854,
      "grad_norm": 0.8515625,
      "learning_rate": 4.348793526775674e-05,
      "loss": 0.7655,
      "step": 2196
    },
    {
      "epoch": 1.1798597362504615,
      "grad_norm": 0.7265625,
      "learning_rate": 4.348224947214016e-05,
      "loss": 1.0039,
      "step": 2197
    },
    {
      "epoch": 1.1803966309855374,
      "grad_norm": 0.6875,
      "learning_rate": 4.347656156746485e-05,
      "loss": 0.7767,
      "step": 2198
    },
    {
      "epoch": 1.1809335257206135,
      "grad_norm": 0.80859375,
      "learning_rate": 4.347087155437987e-05,
      "loss": 1.0435,
      "step": 2199
    },
    {
      "epoch": 1.1814704204556894,
      "grad_norm": 0.65625,
      "learning_rate": 4.3465179433534543e-05,
      "loss": 0.7952,
      "step": 2200
    },
    {
      "epoch": 1.1814704204556894,
      "eval_loss": 0.9191593527793884,
      "eval_runtime": 46.0425,
      "eval_samples_per_second": 20.894,
      "eval_steps_per_second": 20.894,
      "step": 2200
    },
    {
      "epoch": 1.1820073151907655,
      "grad_norm": 0.70703125,
      "learning_rate": 4.34594852055784e-05,
      "loss": 0.8467,
      "step": 2201
    },
    {
      "epoch": 1.1825442099258414,
      "grad_norm": 0.7109375,
      "learning_rate": 4.345378887116122e-05,
      "loss": 0.8265,
      "step": 2202
    },
    {
      "epoch": 1.1830811046609173,
      "grad_norm": 0.69140625,
      "learning_rate": 4.344809043093303e-05,
      "loss": 1.019,
      "step": 2203
    },
    {
      "epoch": 1.1836179993959934,
      "grad_norm": 0.734375,
      "learning_rate": 4.34423898855441e-05,
      "loss": 0.7523,
      "step": 2204
    },
    {
      "epoch": 1.1841548941310693,
      "grad_norm": 0.8125,
      "learning_rate": 4.3436687235644927e-05,
      "loss": 0.9662,
      "step": 2205
    },
    {
      "epoch": 1.1846917888661455,
      "grad_norm": 0.78125,
      "learning_rate": 4.343098248188626e-05,
      "loss": 1.0414,
      "step": 2206
    },
    {
      "epoch": 1.1852286836012214,
      "grad_norm": 0.625,
      "learning_rate": 4.342527562491908e-05,
      "loss": 0.9376,
      "step": 2207
    },
    {
      "epoch": 1.1857655783362975,
      "grad_norm": 0.69921875,
      "learning_rate": 4.341956666539462e-05,
      "loss": 0.9927,
      "step": 2208
    },
    {
      "epoch": 1.1863024730713734,
      "grad_norm": 0.70703125,
      "learning_rate": 4.341385560396434e-05,
      "loss": 0.993,
      "step": 2209
    },
    {
      "epoch": 1.1868393678064495,
      "grad_norm": 0.77734375,
      "learning_rate": 4.340814244127993e-05,
      "loss": 1.0194,
      "step": 2210
    },
    {
      "epoch": 1.1873762625415254,
      "grad_norm": 0.51953125,
      "learning_rate": 4.3402427177993366e-05,
      "loss": 0.6954,
      "step": 2211
    },
    {
      "epoch": 1.1879131572766015,
      "grad_norm": 0.68359375,
      "learning_rate": 4.33967098147568e-05,
      "loss": 0.8069,
      "step": 2212
    },
    {
      "epoch": 1.1884500520116774,
      "grad_norm": 0.80078125,
      "learning_rate": 4.339099035222267e-05,
      "loss": 0.9309,
      "step": 2213
    },
    {
      "epoch": 1.1889869467467535,
      "grad_norm": 0.59375,
      "learning_rate": 4.3385268791043634e-05,
      "loss": 0.7615,
      "step": 2214
    },
    {
      "epoch": 1.1895238414818294,
      "grad_norm": 0.72265625,
      "learning_rate": 4.3379545131872605e-05,
      "loss": 0.8976,
      "step": 2215
    },
    {
      "epoch": 1.1900607362169056,
      "grad_norm": 0.671875,
      "learning_rate": 4.337381937536272e-05,
      "loss": 0.8906,
      "step": 2216
    },
    {
      "epoch": 1.1905976309519815,
      "grad_norm": 0.80859375,
      "learning_rate": 4.3368091522167356e-05,
      "loss": 1.0466,
      "step": 2217
    },
    {
      "epoch": 1.1911345256870576,
      "grad_norm": 0.7734375,
      "learning_rate": 4.3362361572940136e-05,
      "loss": 0.948,
      "step": 2218
    },
    {
      "epoch": 1.1916714204221335,
      "grad_norm": 0.77734375,
      "learning_rate": 4.335662952833492e-05,
      "loss": 1.0474,
      "step": 2219
    },
    {
      "epoch": 1.1922083151572096,
      "grad_norm": 0.625,
      "learning_rate": 4.33508953890058e-05,
      "loss": 0.9169,
      "step": 2220
    },
    {
      "epoch": 1.1927452098922855,
      "grad_norm": 0.7109375,
      "learning_rate": 4.334515915560713e-05,
      "loss": 0.96,
      "step": 2221
    },
    {
      "epoch": 1.1932821046273614,
      "grad_norm": 0.65234375,
      "learning_rate": 4.333942082879348e-05,
      "loss": 0.8072,
      "step": 2222
    },
    {
      "epoch": 1.1938189993624375,
      "grad_norm": 0.6640625,
      "learning_rate": 4.3333680409219655e-05,
      "loss": 0.8069,
      "step": 2223
    },
    {
      "epoch": 1.1943558940975134,
      "grad_norm": 0.6171875,
      "learning_rate": 4.332793789754072e-05,
      "loss": 0.8196,
      "step": 2224
    },
    {
      "epoch": 1.1948927888325895,
      "grad_norm": 0.6796875,
      "learning_rate": 4.3322193294411974e-05,
      "loss": 0.8526,
      "step": 2225
    },
    {
      "epoch": 1.1954296835676654,
      "grad_norm": 0.71875,
      "learning_rate": 4.331644660048893e-05,
      "loss": 0.8704,
      "step": 2226
    },
    {
      "epoch": 1.1959665783027416,
      "grad_norm": 0.60546875,
      "learning_rate": 4.3310697816427374e-05,
      "loss": 0.9687,
      "step": 2227
    },
    {
      "epoch": 1.1965034730378175,
      "grad_norm": 0.62890625,
      "learning_rate": 4.330494694288331e-05,
      "loss": 0.863,
      "step": 2228
    },
    {
      "epoch": 1.1970403677728936,
      "grad_norm": 0.66015625,
      "learning_rate": 4.3299193980512995e-05,
      "loss": 0.8124,
      "step": 2229
    },
    {
      "epoch": 1.1975772625079695,
      "grad_norm": 0.65625,
      "learning_rate": 4.3293438929972894e-05,
      "loss": 0.8963,
      "step": 2230
    },
    {
      "epoch": 1.1981141572430456,
      "grad_norm": 0.640625,
      "learning_rate": 4.328768179191975e-05,
      "loss": 1.0648,
      "step": 2231
    },
    {
      "epoch": 1.1986510519781215,
      "grad_norm": 0.59765625,
      "learning_rate": 4.328192256701052e-05,
      "loss": 0.7347,
      "step": 2232
    },
    {
      "epoch": 1.1991879467131976,
      "grad_norm": 0.66015625,
      "learning_rate": 4.3276161255902407e-05,
      "loss": 0.8145,
      "step": 2233
    },
    {
      "epoch": 1.1997248414482735,
      "grad_norm": 0.8125,
      "learning_rate": 4.327039785925283e-05,
      "loss": 0.811,
      "step": 2234
    },
    {
      "epoch": 1.2002617361833496,
      "grad_norm": 0.7109375,
      "learning_rate": 4.3264632377719496e-05,
      "loss": 1.0834,
      "step": 2235
    },
    {
      "epoch": 1.2007986309184255,
      "grad_norm": 0.66015625,
      "learning_rate": 4.3258864811960296e-05,
      "loss": 0.9029,
      "step": 2236
    },
    {
      "epoch": 1.2013355256535017,
      "grad_norm": 0.6796875,
      "learning_rate": 4.3253095162633396e-05,
      "loss": 1.0308,
      "step": 2237
    },
    {
      "epoch": 1.2018724203885776,
      "grad_norm": 0.74609375,
      "learning_rate": 4.324732343039718e-05,
      "loss": 0.9287,
      "step": 2238
    },
    {
      "epoch": 1.2024093151236537,
      "grad_norm": 0.73046875,
      "learning_rate": 4.324154961591028e-05,
      "loss": 0.9113,
      "step": 2239
    },
    {
      "epoch": 1.2029462098587296,
      "grad_norm": 0.64453125,
      "learning_rate": 4.323577371983155e-05,
      "loss": 0.7997,
      "step": 2240
    },
    {
      "epoch": 1.2034831045938055,
      "grad_norm": 0.77734375,
      "learning_rate": 4.322999574282011e-05,
      "loss": 0.9625,
      "step": 2241
    },
    {
      "epoch": 1.2040199993288816,
      "grad_norm": 0.8125,
      "learning_rate": 4.3224215685535294e-05,
      "loss": 0.8329,
      "step": 2242
    },
    {
      "epoch": 1.2045568940639575,
      "grad_norm": 0.65625,
      "learning_rate": 4.321843354863667e-05,
      "loss": 0.841,
      "step": 2243
    },
    {
      "epoch": 1.2050937887990336,
      "grad_norm": 0.71875,
      "learning_rate": 4.321264933278406e-05,
      "loss": 0.8994,
      "step": 2244
    },
    {
      "epoch": 1.2056306835341095,
      "grad_norm": 0.72265625,
      "learning_rate": 4.3206863038637516e-05,
      "loss": 1.1515,
      "step": 2245
    },
    {
      "epoch": 1.2061675782691856,
      "grad_norm": 0.70703125,
      "learning_rate": 4.3201074666857334e-05,
      "loss": 0.8285,
      "step": 2246
    },
    {
      "epoch": 1.2067044730042615,
      "grad_norm": 0.609375,
      "learning_rate": 4.3195284218104024e-05,
      "loss": 0.6893,
      "step": 2247
    },
    {
      "epoch": 1.2072413677393377,
      "grad_norm": 0.8046875,
      "learning_rate": 4.318949169303837e-05,
      "loss": 0.8844,
      "step": 2248
    },
    {
      "epoch": 1.2077782624744136,
      "grad_norm": 0.83984375,
      "learning_rate": 4.318369709232136e-05,
      "loss": 1.1114,
      "step": 2249
    },
    {
      "epoch": 1.2083151572094897,
      "grad_norm": 0.86328125,
      "learning_rate": 4.317790041661424e-05,
      "loss": 0.9558,
      "step": 2250
    },
    {
      "epoch": 1.2083151572094897,
      "eval_loss": 0.9185804724693298,
      "eval_runtime": 45.9295,
      "eval_samples_per_second": 20.945,
      "eval_steps_per_second": 20.945,
      "step": 2250
    },
    {
      "epoch": 1.2088520519445656,
      "grad_norm": 0.74609375,
      "learning_rate": 4.317210166657847e-05,
      "loss": 0.7758,
      "step": 2251
    },
    {
      "epoch": 1.2093889466796417,
      "grad_norm": 0.8515625,
      "learning_rate": 4.316630084287577e-05,
      "loss": 0.8318,
      "step": 2252
    },
    {
      "epoch": 1.2099258414147176,
      "grad_norm": 0.7109375,
      "learning_rate": 4.316049794616809e-05,
      "loss": 0.866,
      "step": 2253
    },
    {
      "epoch": 1.2104627361497937,
      "grad_norm": 0.80859375,
      "learning_rate": 4.3154692977117604e-05,
      "loss": 0.9432,
      "step": 2254
    },
    {
      "epoch": 1.2109996308848696,
      "grad_norm": 0.8359375,
      "learning_rate": 4.314888593638674e-05,
      "loss": 0.9956,
      "step": 2255
    },
    {
      "epoch": 1.2115365256199455,
      "grad_norm": 0.7421875,
      "learning_rate": 4.314307682463816e-05,
      "loss": 1.0258,
      "step": 2256
    },
    {
      "epoch": 1.2120734203550216,
      "grad_norm": 0.60546875,
      "learning_rate": 4.313726564253474e-05,
      "loss": 0.7928,
      "step": 2257
    },
    {
      "epoch": 1.2126103150900978,
      "grad_norm": 0.62890625,
      "learning_rate": 4.3131452390739626e-05,
      "loss": 0.772,
      "step": 2258
    },
    {
      "epoch": 1.2131472098251737,
      "grad_norm": 0.8125,
      "learning_rate": 4.312563706991618e-05,
      "loss": 0.8915,
      "step": 2259
    },
    {
      "epoch": 1.2136841045602496,
      "grad_norm": 0.71484375,
      "learning_rate": 4.3119819680728e-05,
      "loss": 0.8551,
      "step": 2260
    },
    {
      "epoch": 1.2142209992953257,
      "grad_norm": 0.87890625,
      "learning_rate": 4.3114000223838927e-05,
      "loss": 1.0303,
      "step": 2261
    },
    {
      "epoch": 1.2147578940304016,
      "grad_norm": 0.7734375,
      "learning_rate": 4.310817869991303e-05,
      "loss": 1.0987,
      "step": 2262
    },
    {
      "epoch": 1.2152947887654777,
      "grad_norm": 0.78515625,
      "learning_rate": 4.310235510961462e-05,
      "loss": 0.7569,
      "step": 2263
    },
    {
      "epoch": 1.2158316835005536,
      "grad_norm": 0.76171875,
      "learning_rate": 4.309652945360825e-05,
      "loss": 0.8945,
      "step": 2264
    },
    {
      "epoch": 1.2163685782356297,
      "grad_norm": 0.6875,
      "learning_rate": 4.3090701732558694e-05,
      "loss": 0.8193,
      "step": 2265
    },
    {
      "epoch": 1.2169054729707056,
      "grad_norm": 0.79296875,
      "learning_rate": 4.308487194713097e-05,
      "loss": 0.9874,
      "step": 2266
    },
    {
      "epoch": 1.2174423677057817,
      "grad_norm": 0.7734375,
      "learning_rate": 4.307904009799033e-05,
      "loss": 0.9096,
      "step": 2267
    },
    {
      "epoch": 1.2179792624408576,
      "grad_norm": 0.796875,
      "learning_rate": 4.307320618580226e-05,
      "loss": 1.0407,
      "step": 2268
    },
    {
      "epoch": 1.2185161571759338,
      "grad_norm": 0.76953125,
      "learning_rate": 4.306737021123249e-05,
      "loss": 0.9619,
      "step": 2269
    },
    {
      "epoch": 1.2190530519110097,
      "grad_norm": 0.66015625,
      "learning_rate": 4.306153217494697e-05,
      "loss": 0.8103,
      "step": 2270
    },
    {
      "epoch": 1.2195899466460858,
      "grad_norm": 0.78515625,
      "learning_rate": 4.30556920776119e-05,
      "loss": 0.8183,
      "step": 2271
    },
    {
      "epoch": 1.2201268413811617,
      "grad_norm": 0.98828125,
      "learning_rate": 4.304984991989371e-05,
      "loss": 1.0441,
      "step": 2272
    },
    {
      "epoch": 1.2206637361162378,
      "grad_norm": 1.0,
      "learning_rate": 4.304400570245906e-05,
      "loss": 0.7898,
      "step": 2273
    },
    {
      "epoch": 1.2212006308513137,
      "grad_norm": 0.6875,
      "learning_rate": 4.3038159425974846e-05,
      "loss": 0.9046,
      "step": 2274
    },
    {
      "epoch": 1.2217375255863896,
      "grad_norm": 0.8203125,
      "learning_rate": 4.3032311091108214e-05,
      "loss": 0.8699,
      "step": 2275
    },
    {
      "epoch": 1.2222744203214657,
      "grad_norm": 0.796875,
      "learning_rate": 4.302646069852652e-05,
      "loss": 0.7927,
      "step": 2276
    },
    {
      "epoch": 1.2228113150565418,
      "grad_norm": 0.7734375,
      "learning_rate": 4.3020608248897373e-05,
      "loss": 0.879,
      "step": 2277
    },
    {
      "epoch": 1.2233482097916177,
      "grad_norm": 0.74609375,
      "learning_rate": 4.301475374288861e-05,
      "loss": 0.9867,
      "step": 2278
    },
    {
      "epoch": 1.2238851045266936,
      "grad_norm": 0.67578125,
      "learning_rate": 4.300889718116831e-05,
      "loss": 0.8386,
      "step": 2279
    },
    {
      "epoch": 1.2244219992617698,
      "grad_norm": 0.78125,
      "learning_rate": 4.300303856440478e-05,
      "loss": 0.9709,
      "step": 2280
    },
    {
      "epoch": 1.2249588939968457,
      "grad_norm": 0.58984375,
      "learning_rate": 4.299717789326656e-05,
      "loss": 0.7167,
      "step": 2281
    },
    {
      "epoch": 1.2254957887319218,
      "grad_norm": 0.734375,
      "learning_rate": 4.299131516842242e-05,
      "loss": 0.9082,
      "step": 2282
    },
    {
      "epoch": 1.2260326834669977,
      "grad_norm": 0.63671875,
      "learning_rate": 4.298545039054138e-05,
      "loss": 0.7985,
      "step": 2283
    },
    {
      "epoch": 1.2265695782020738,
      "grad_norm": 0.6171875,
      "learning_rate": 4.297958356029269e-05,
      "loss": 0.6688,
      "step": 2284
    },
    {
      "epoch": 1.2271064729371497,
      "grad_norm": 0.67578125,
      "learning_rate": 4.2973714678345824e-05,
      "loss": 0.7596,
      "step": 2285
    },
    {
      "epoch": 1.2276433676722258,
      "grad_norm": 0.6171875,
      "learning_rate": 4.2967843745370484e-05,
      "loss": 0.8478,
      "step": 2286
    },
    {
      "epoch": 1.2281802624073017,
      "grad_norm": 0.734375,
      "learning_rate": 4.296197076203663e-05,
      "loss": 0.8947,
      "step": 2287
    },
    {
      "epoch": 1.2287171571423778,
      "grad_norm": 0.73046875,
      "learning_rate": 4.2956095729014444e-05,
      "loss": 1.0706,
      "step": 2288
    },
    {
      "epoch": 1.2292540518774537,
      "grad_norm": 0.69921875,
      "learning_rate": 4.2950218646974346e-05,
      "loss": 0.7206,
      "step": 2289
    },
    {
      "epoch": 1.2297909466125299,
      "grad_norm": 0.62890625,
      "learning_rate": 4.294433951658697e-05,
      "loss": 0.6709,
      "step": 2290
    },
    {
      "epoch": 1.2303278413476058,
      "grad_norm": 0.69140625,
      "learning_rate": 4.293845833852321e-05,
      "loss": 1.0079,
      "step": 2291
    },
    {
      "epoch": 1.2308647360826819,
      "grad_norm": 0.703125,
      "learning_rate": 4.293257511345419e-05,
      "loss": 0.8778,
      "step": 2292
    },
    {
      "epoch": 1.2314016308177578,
      "grad_norm": 0.6796875,
      "learning_rate": 4.292668984205124e-05,
      "loss": 1.0551,
      "step": 2293
    },
    {
      "epoch": 1.2319385255528337,
      "grad_norm": 0.76953125,
      "learning_rate": 4.292080252498596e-05,
      "loss": 0.9318,
      "step": 2294
    },
    {
      "epoch": 1.2324754202879098,
      "grad_norm": 0.80859375,
      "learning_rate": 4.291491316293017e-05,
      "loss": 0.8448,
      "step": 2295
    },
    {
      "epoch": 1.233012315022986,
      "grad_norm": 0.703125,
      "learning_rate": 4.2909021756555906e-05,
      "loss": 0.9453,
      "step": 2296
    },
    {
      "epoch": 1.2335492097580618,
      "grad_norm": 0.8515625,
      "learning_rate": 4.2903128306535466e-05,
      "loss": 0.9568,
      "step": 2297
    },
    {
      "epoch": 1.2340861044931377,
      "grad_norm": 0.69921875,
      "learning_rate": 4.289723281354135e-05,
      "loss": 0.9516,
      "step": 2298
    },
    {
      "epoch": 1.2346229992282138,
      "grad_norm": 0.69921875,
      "learning_rate": 4.2891335278246335e-05,
      "loss": 0.8181,
      "step": 2299
    },
    {
      "epoch": 1.2351598939632897,
      "grad_norm": 0.69140625,
      "learning_rate": 4.288543570132338e-05,
      "loss": 1.0422,
      "step": 2300
    },
    {
      "epoch": 1.2351598939632897,
      "eval_loss": 0.919032633304596,
      "eval_runtime": 46.5908,
      "eval_samples_per_second": 20.648,
      "eval_steps_per_second": 20.648,
      "step": 2300
    },
    {
      "epoch": 1.2356967886983659,
      "grad_norm": 0.6484375,
      "learning_rate": 4.2879534083445716e-05,
      "loss": 0.8635,
      "step": 2301
    },
    {
      "epoch": 1.2362336834334418,
      "grad_norm": 0.87890625,
      "learning_rate": 4.2873630425286784e-05,
      "loss": 0.872,
      "step": 2302
    },
    {
      "epoch": 1.2367705781685179,
      "grad_norm": 0.71875,
      "learning_rate": 4.286772472752027e-05,
      "loss": 0.8259,
      "step": 2303
    },
    {
      "epoch": 1.2373074729035938,
      "grad_norm": 0.7109375,
      "learning_rate": 4.2861816990820084e-05,
      "loss": 0.9112,
      "step": 2304
    },
    {
      "epoch": 1.23784436763867,
      "grad_norm": 0.76953125,
      "learning_rate": 4.285590721586039e-05,
      "loss": 0.8183,
      "step": 2305
    },
    {
      "epoch": 1.2383812623737458,
      "grad_norm": 0.8125,
      "learning_rate": 4.284999540331554e-05,
      "loss": 1.0442,
      "step": 2306
    },
    {
      "epoch": 1.238918157108822,
      "grad_norm": 0.734375,
      "learning_rate": 4.2844081553860174e-05,
      "loss": 1.0191,
      "step": 2307
    },
    {
      "epoch": 1.2394550518438978,
      "grad_norm": 0.78515625,
      "learning_rate": 4.2838165668169117e-05,
      "loss": 0.9365,
      "step": 2308
    },
    {
      "epoch": 1.239991946578974,
      "grad_norm": 0.60546875,
      "learning_rate": 4.283224774691746e-05,
      "loss": 0.7694,
      "step": 2309
    },
    {
      "epoch": 1.2405288413140498,
      "grad_norm": 0.68359375,
      "learning_rate": 4.282632779078051e-05,
      "loss": 0.8036,
      "step": 2310
    },
    {
      "epoch": 1.241065736049126,
      "grad_norm": 0.796875,
      "learning_rate": 4.2820405800433804e-05,
      "loss": 1.0061,
      "step": 2311
    },
    {
      "epoch": 1.2416026307842019,
      "grad_norm": 0.7578125,
      "learning_rate": 4.281448177655312e-05,
      "loss": 0.8638,
      "step": 2312
    },
    {
      "epoch": 1.2421395255192778,
      "grad_norm": 0.73046875,
      "learning_rate": 4.2808555719814466e-05,
      "loss": 0.7865,
      "step": 2313
    },
    {
      "epoch": 1.2426764202543539,
      "grad_norm": 0.6640625,
      "learning_rate": 4.280262763089407e-05,
      "loss": 0.9393,
      "step": 2314
    },
    {
      "epoch": 1.24321331498943,
      "grad_norm": 0.71484375,
      "learning_rate": 4.279669751046841e-05,
      "loss": 0.9858,
      "step": 2315
    },
    {
      "epoch": 1.243750209724506,
      "grad_norm": 0.74609375,
      "learning_rate": 4.27907653592142e-05,
      "loss": 0.993,
      "step": 2316
    },
    {
      "epoch": 1.2442871044595818,
      "grad_norm": 0.5703125,
      "learning_rate": 4.278483117780835e-05,
      "loss": 0.7039,
      "step": 2317
    },
    {
      "epoch": 1.244823999194658,
      "grad_norm": 0.66015625,
      "learning_rate": 4.277889496692804e-05,
      "loss": 0.8444,
      "step": 2318
    },
    {
      "epoch": 1.2453608939297338,
      "grad_norm": 0.58203125,
      "learning_rate": 4.2772956727250655e-05,
      "loss": 0.7538,
      "step": 2319
    },
    {
      "epoch": 1.24589778866481,
      "grad_norm": 0.7578125,
      "learning_rate": 4.2767016459453835e-05,
      "loss": 0.9166,
      "step": 2320
    },
    {
      "epoch": 1.2464346833998858,
      "grad_norm": 0.59765625,
      "learning_rate": 4.276107416421544e-05,
      "loss": 0.7658,
      "step": 2321
    },
    {
      "epoch": 1.246971578134962,
      "grad_norm": 0.58203125,
      "learning_rate": 4.2755129842213546e-05,
      "loss": 0.8692,
      "step": 2322
    },
    {
      "epoch": 1.2475084728700379,
      "grad_norm": 0.7421875,
      "learning_rate": 4.2749183494126496e-05,
      "loss": 0.8463,
      "step": 2323
    },
    {
      "epoch": 1.248045367605114,
      "grad_norm": 0.78515625,
      "learning_rate": 4.274323512063284e-05,
      "loss": 0.9421,
      "step": 2324
    },
    {
      "epoch": 1.2485822623401899,
      "grad_norm": 0.59765625,
      "learning_rate": 4.273728472241134e-05,
      "loss": 0.7189,
      "step": 2325
    },
    {
      "epoch": 1.249119157075266,
      "grad_norm": 0.69140625,
      "learning_rate": 4.2731332300141046e-05,
      "loss": 0.7346,
      "step": 2326
    },
    {
      "epoch": 1.249656051810342,
      "grad_norm": 0.625,
      "learning_rate": 4.272537785450118e-05,
      "loss": 0.9027,
      "step": 2327
    },
    {
      "epoch": 1.250192946545418,
      "grad_norm": 0.7578125,
      "learning_rate": 4.271942138617122e-05,
      "loss": 1.0147,
      "step": 2328
    },
    {
      "epoch": 1.250729841280494,
      "grad_norm": 0.76171875,
      "learning_rate": 4.2713462895830894e-05,
      "loss": 0.8928,
      "step": 2329
    },
    {
      "epoch": 1.25126673601557,
      "grad_norm": 0.8359375,
      "learning_rate": 4.270750238416013e-05,
      "loss": 0.8658,
      "step": 2330
    },
    {
      "epoch": 1.251803630750646,
      "grad_norm": 0.8984375,
      "learning_rate": 4.270153985183908e-05,
      "loss": 0.7505,
      "step": 2331
    },
    {
      "epoch": 1.2523405254857218,
      "grad_norm": 0.671875,
      "learning_rate": 4.269557529954818e-05,
      "loss": 0.8029,
      "step": 2332
    },
    {
      "epoch": 1.252877420220798,
      "grad_norm": 0.890625,
      "learning_rate": 4.268960872796802e-05,
      "loss": 0.9767,
      "step": 2333
    },
    {
      "epoch": 1.253414314955874,
      "grad_norm": 0.921875,
      "learning_rate": 4.26836401377795e-05,
      "loss": 0.9238,
      "step": 2334
    },
    {
      "epoch": 1.25395120969095,
      "grad_norm": 0.66796875,
      "learning_rate": 4.267766952966369e-05,
      "loss": 0.9478,
      "step": 2335
    },
    {
      "epoch": 1.2544881044260259,
      "grad_norm": 0.8828125,
      "learning_rate": 4.267169690430192e-05,
      "loss": 0.939,
      "step": 2336
    },
    {
      "epoch": 1.255024999161102,
      "grad_norm": 0.78515625,
      "learning_rate": 4.2665722262375734e-05,
      "loss": 0.8077,
      "step": 2337
    },
    {
      "epoch": 1.2555618938961781,
      "grad_norm": 0.73046875,
      "learning_rate": 4.2659745604566924e-05,
      "loss": 0.9556,
      "step": 2338
    },
    {
      "epoch": 1.256098788631254,
      "grad_norm": 0.62109375,
      "learning_rate": 4.26537669315575e-05,
      "loss": 0.7619,
      "step": 2339
    },
    {
      "epoch": 1.25663568336633,
      "grad_norm": 0.796875,
      "learning_rate": 4.2647786244029694e-05,
      "loss": 0.9437,
      "step": 2340
    },
    {
      "epoch": 1.257172578101406,
      "grad_norm": 0.71875,
      "learning_rate": 4.264180354266599e-05,
      "loss": 0.9256,
      "step": 2341
    },
    {
      "epoch": 1.257709472836482,
      "grad_norm": 1.0625,
      "learning_rate": 4.263581882814909e-05,
      "loss": 0.9118,
      "step": 2342
    },
    {
      "epoch": 1.258246367571558,
      "grad_norm": 0.8203125,
      "learning_rate": 4.262983210116193e-05,
      "loss": 1.0294,
      "step": 2343
    },
    {
      "epoch": 1.258783262306634,
      "grad_norm": 0.72265625,
      "learning_rate": 4.2623843362387645e-05,
      "loss": 1.1032,
      "step": 2344
    },
    {
      "epoch": 1.25932015704171,
      "grad_norm": 0.64453125,
      "learning_rate": 4.2617852612509656e-05,
      "loss": 0.8864,
      "step": 2345
    },
    {
      "epoch": 1.259857051776786,
      "grad_norm": 0.67578125,
      "learning_rate": 4.261185985221156e-05,
      "loss": 0.8555,
      "step": 2346
    },
    {
      "epoch": 1.260393946511862,
      "grad_norm": 0.6875,
      "learning_rate": 4.2605865082177235e-05,
      "loss": 0.7918,
      "step": 2347
    },
    {
      "epoch": 1.260930841246938,
      "grad_norm": 0.7109375,
      "learning_rate": 4.259986830309074e-05,
      "loss": 0.7688,
      "step": 2348
    },
    {
      "epoch": 1.2614677359820141,
      "grad_norm": 0.78125,
      "learning_rate": 4.2593869515636376e-05,
      "loss": 0.8333,
      "step": 2349
    },
    {
      "epoch": 1.26200463071709,
      "grad_norm": 0.80078125,
      "learning_rate": 4.2587868720498705e-05,
      "loss": 0.7353,
      "step": 2350
    },
    {
      "epoch": 1.26200463071709,
      "eval_loss": 0.9166867136955261,
      "eval_runtime": 45.6547,
      "eval_samples_per_second": 21.071,
      "eval_steps_per_second": 21.071,
      "step": 2350
    },
    {
      "epoch": 1.262541525452166,
      "grad_norm": 0.640625,
      "learning_rate": 4.258186591836248e-05,
      "loss": 0.773,
      "step": 2351
    },
    {
      "epoch": 1.263078420187242,
      "grad_norm": 0.64453125,
      "learning_rate": 4.257586110991268e-05,
      "loss": 0.7677,
      "step": 2352
    },
    {
      "epoch": 1.2636153149223182,
      "grad_norm": 0.65625,
      "learning_rate": 4.2569854295834566e-05,
      "loss": 0.9747,
      "step": 2353
    },
    {
      "epoch": 1.264152209657394,
      "grad_norm": 0.83203125,
      "learning_rate": 4.2563845476813566e-05,
      "loss": 0.8526,
      "step": 2354
    },
    {
      "epoch": 1.26468910439247,
      "grad_norm": 0.65625,
      "learning_rate": 4.2557834653535366e-05,
      "loss": 0.7428,
      "step": 2355
    },
    {
      "epoch": 1.265225999127546,
      "grad_norm": 0.68359375,
      "learning_rate": 4.2551821826685887e-05,
      "loss": 0.8741,
      "step": 2356
    },
    {
      "epoch": 1.2657628938626222,
      "grad_norm": 0.70703125,
      "learning_rate": 4.254580699695125e-05,
      "loss": 0.7939,
      "step": 2357
    },
    {
      "epoch": 1.266299788597698,
      "grad_norm": 0.75,
      "learning_rate": 4.253979016501784e-05,
      "loss": 0.9722,
      "step": 2358
    },
    {
      "epoch": 1.266836683332774,
      "grad_norm": 0.71484375,
      "learning_rate": 4.2533771331572263e-05,
      "loss": 1.0446,
      "step": 2359
    },
    {
      "epoch": 1.2673735780678501,
      "grad_norm": 0.6953125,
      "learning_rate": 4.2527750497301323e-05,
      "loss": 1.0864,
      "step": 2360
    },
    {
      "epoch": 1.267910472802926,
      "grad_norm": 0.69921875,
      "learning_rate": 4.2521727662892074e-05,
      "loss": 0.8485,
      "step": 2361
    },
    {
      "epoch": 1.2684473675380021,
      "grad_norm": 0.8046875,
      "learning_rate": 4.25157028290318e-05,
      "loss": 1.0127,
      "step": 2362
    },
    {
      "epoch": 1.268984262273078,
      "grad_norm": 0.8203125,
      "learning_rate": 4.250967599640803e-05,
      "loss": 0.8196,
      "step": 2363
    },
    {
      "epoch": 1.2695211570081542,
      "grad_norm": 0.7265625,
      "learning_rate": 4.250364716570847e-05,
      "loss": 0.8567,
      "step": 2364
    },
    {
      "epoch": 1.27005805174323,
      "grad_norm": 0.7265625,
      "learning_rate": 4.249761633762111e-05,
      "loss": 0.8167,
      "step": 2365
    },
    {
      "epoch": 1.2705949464783062,
      "grad_norm": 0.8125,
      "learning_rate": 4.249158351283414e-05,
      "loss": 0.9388,
      "step": 2366
    },
    {
      "epoch": 1.271131841213382,
      "grad_norm": 0.66015625,
      "learning_rate": 4.248554869203597e-05,
      "loss": 1.0163,
      "step": 2367
    },
    {
      "epoch": 1.2716687359484582,
      "grad_norm": 0.671875,
      "learning_rate": 4.247951187591527e-05,
      "loss": 0.8415,
      "step": 2368
    },
    {
      "epoch": 1.272205630683534,
      "grad_norm": 0.875,
      "learning_rate": 4.247347306516089e-05,
      "loss": 0.8779,
      "step": 2369
    },
    {
      "epoch": 1.27274252541861,
      "grad_norm": 0.7734375,
      "learning_rate": 4.246743226046196e-05,
      "loss": 0.8726,
      "step": 2370
    },
    {
      "epoch": 1.2732794201536861,
      "grad_norm": 0.72265625,
      "learning_rate": 4.24613894625078e-05,
      "loss": 0.8935,
      "step": 2371
    },
    {
      "epoch": 1.2738163148887622,
      "grad_norm": 0.6953125,
      "learning_rate": 4.245534467198797e-05,
      "loss": 0.8479,
      "step": 2372
    },
    {
      "epoch": 1.2743532096238381,
      "grad_norm": 1.0546875,
      "learning_rate": 4.244929788959225e-05,
      "loss": 0.9463,
      "step": 2373
    },
    {
      "epoch": 1.274890104358914,
      "grad_norm": 0.625,
      "learning_rate": 4.244324911601068e-05,
      "loss": 0.8583,
      "step": 2374
    },
    {
      "epoch": 1.2754269990939902,
      "grad_norm": 0.70703125,
      "learning_rate": 4.2437198351933484e-05,
      "loss": 0.9005,
      "step": 2375
    },
    {
      "epoch": 1.2759638938290663,
      "grad_norm": 0.69140625,
      "learning_rate": 4.243114559805112e-05,
      "loss": 0.7934,
      "step": 2376
    },
    {
      "epoch": 1.2765007885641422,
      "grad_norm": 0.65234375,
      "learning_rate": 4.24250908550543e-05,
      "loss": 0.7214,
      "step": 2377
    },
    {
      "epoch": 1.277037683299218,
      "grad_norm": 0.65234375,
      "learning_rate": 4.241903412363395e-05,
      "loss": 0.7903,
      "step": 2378
    },
    {
      "epoch": 1.2775745780342942,
      "grad_norm": 0.85546875,
      "learning_rate": 4.24129754044812e-05,
      "loss": 1.0774,
      "step": 2379
    },
    {
      "epoch": 1.27811147276937,
      "grad_norm": 0.640625,
      "learning_rate": 4.240691469828745e-05,
      "loss": 0.8575,
      "step": 2380
    },
    {
      "epoch": 1.2786483675044462,
      "grad_norm": 0.69140625,
      "learning_rate": 4.2400852005744284e-05,
      "loss": 1.0707,
      "step": 2381
    },
    {
      "epoch": 1.2791852622395221,
      "grad_norm": 0.6015625,
      "learning_rate": 4.239478732754354e-05,
      "loss": 0.7319,
      "step": 2382
    },
    {
      "epoch": 1.2797221569745982,
      "grad_norm": 0.90625,
      "learning_rate": 4.2388720664377276e-05,
      "loss": 1.0774,
      "step": 2383
    },
    {
      "epoch": 1.2802590517096741,
      "grad_norm": 0.703125,
      "learning_rate": 4.238265201693778e-05,
      "loss": 0.9633,
      "step": 2384
    },
    {
      "epoch": 1.2807959464447503,
      "grad_norm": 0.9375,
      "learning_rate": 4.2376581385917547e-05,
      "loss": 1.0594,
      "step": 2385
    },
    {
      "epoch": 1.2813328411798262,
      "grad_norm": 0.6640625,
      "learning_rate": 4.2370508772009334e-05,
      "loss": 0.9085,
      "step": 2386
    },
    {
      "epoch": 1.2818697359149023,
      "grad_norm": 0.61328125,
      "learning_rate": 4.236443417590607e-05,
      "loss": 0.7452,
      "step": 2387
    },
    {
      "epoch": 1.2824066306499782,
      "grad_norm": 0.6640625,
      "learning_rate": 4.235835759830098e-05,
      "loss": 0.7606,
      "step": 2388
    },
    {
      "epoch": 1.282943525385054,
      "grad_norm": 0.72265625,
      "learning_rate": 4.235227903988746e-05,
      "loss": 0.9656,
      "step": 2389
    },
    {
      "epoch": 1.2834804201201302,
      "grad_norm": 0.58203125,
      "learning_rate": 4.2346198501359146e-05,
      "loss": 0.8026,
      "step": 2390
    },
    {
      "epoch": 1.2840173148552063,
      "grad_norm": 0.75390625,
      "learning_rate": 4.234011598340992e-05,
      "loss": 0.9217,
      "step": 2391
    },
    {
      "epoch": 1.2845542095902822,
      "grad_norm": 0.76171875,
      "learning_rate": 4.233403148673386e-05,
      "loss": 0.8261,
      "step": 2392
    },
    {
      "epoch": 1.2850911043253581,
      "grad_norm": 0.69140625,
      "learning_rate": 4.2327945012025294e-05,
      "loss": 0.897,
      "step": 2393
    },
    {
      "epoch": 1.2856279990604342,
      "grad_norm": 0.828125,
      "learning_rate": 4.232185655997876e-05,
      "loss": 0.9855,
      "step": 2394
    },
    {
      "epoch": 1.2861648937955104,
      "grad_norm": 0.72265625,
      "learning_rate": 4.231576613128902e-05,
      "loss": 0.8463,
      "step": 2395
    },
    {
      "epoch": 1.2867017885305863,
      "grad_norm": 0.80078125,
      "learning_rate": 4.230967372665109e-05,
      "loss": 0.9179,
      "step": 2396
    },
    {
      "epoch": 1.2872386832656622,
      "grad_norm": 0.77734375,
      "learning_rate": 4.230357934676017e-05,
      "loss": 0.8223,
      "step": 2397
    },
    {
      "epoch": 1.2877755780007383,
      "grad_norm": 0.640625,
      "learning_rate": 4.229748299231172e-05,
      "loss": 0.6675,
      "step": 2398
    },
    {
      "epoch": 1.2883124727358142,
      "grad_norm": 0.7421875,
      "learning_rate": 4.22913846640014e-05,
      "loss": 0.8385,
      "step": 2399
    },
    {
      "epoch": 1.2888493674708903,
      "grad_norm": 0.69921875,
      "learning_rate": 4.228528436252512e-05,
      "loss": 0.924,
      "step": 2400
    },
    {
      "epoch": 1.2888493674708903,
      "eval_loss": 0.9168675541877747,
      "eval_runtime": 46.1564,
      "eval_samples_per_second": 20.842,
      "eval_steps_per_second": 20.842,
      "step": 2400
    },
    {
      "epoch": 1.2893862622059662,
      "grad_norm": 0.87109375,
      "learning_rate": 4.2279182088578996e-05,
      "loss": 0.922,
      "step": 2401
    },
    {
      "epoch": 1.2899231569410423,
      "grad_norm": 0.65234375,
      "learning_rate": 4.227307784285936e-05,
      "loss": 1.067,
      "step": 2402
    },
    {
      "epoch": 1.2904600516761182,
      "grad_norm": 0.73828125,
      "learning_rate": 4.22669716260628e-05,
      "loss": 0.8357,
      "step": 2403
    },
    {
      "epoch": 1.2909969464111943,
      "grad_norm": 0.68359375,
      "learning_rate": 4.226086343888611e-05,
      "loss": 0.7451,
      "step": 2404
    },
    {
      "epoch": 1.2915338411462702,
      "grad_norm": 0.69921875,
      "learning_rate": 4.2254753282026314e-05,
      "loss": 0.8641,
      "step": 2405
    },
    {
      "epoch": 1.2920707358813464,
      "grad_norm": 0.640625,
      "learning_rate": 4.224864115618066e-05,
      "loss": 0.8504,
      "step": 2406
    },
    {
      "epoch": 1.2926076306164223,
      "grad_norm": 0.69140625,
      "learning_rate": 4.22425270620466e-05,
      "loss": 0.8427,
      "step": 2407
    },
    {
      "epoch": 1.2931445253514982,
      "grad_norm": 0.71484375,
      "learning_rate": 4.223641100032185e-05,
      "loss": 0.93,
      "step": 2408
    },
    {
      "epoch": 1.2936814200865743,
      "grad_norm": 0.72265625,
      "learning_rate": 4.2230292971704324e-05,
      "loss": 0.9213,
      "step": 2409
    },
    {
      "epoch": 1.2942183148216504,
      "grad_norm": 1.1484375,
      "learning_rate": 4.222417297689217e-05,
      "loss": 0.915,
      "step": 2410
    },
    {
      "epoch": 1.2947552095567263,
      "grad_norm": 0.78515625,
      "learning_rate": 4.221805101658375e-05,
      "loss": 1.1016,
      "step": 2411
    },
    {
      "epoch": 1.2952921042918022,
      "grad_norm": 0.765625,
      "learning_rate": 4.2211927091477656e-05,
      "loss": 0.8728,
      "step": 2412
    },
    {
      "epoch": 1.2958289990268783,
      "grad_norm": 0.671875,
      "learning_rate": 4.220580120227271e-05,
      "loss": 1.0869,
      "step": 2413
    },
    {
      "epoch": 1.2963658937619544,
      "grad_norm": 0.9140625,
      "learning_rate": 4.2199673349667955e-05,
      "loss": 0.9189,
      "step": 2414
    },
    {
      "epoch": 1.2969027884970303,
      "grad_norm": 0.83984375,
      "learning_rate": 4.2193543534362654e-05,
      "loss": 1.0031,
      "step": 2415
    },
    {
      "epoch": 1.2974396832321062,
      "grad_norm": 0.7109375,
      "learning_rate": 4.2187411757056295e-05,
      "loss": 0.9205,
      "step": 2416
    },
    {
      "epoch": 1.2979765779671824,
      "grad_norm": 0.69921875,
      "learning_rate": 4.21812780184486e-05,
      "loss": 0.8935,
      "step": 2417
    },
    {
      "epoch": 1.2985134727022583,
      "grad_norm": 0.84375,
      "learning_rate": 4.217514231923949e-05,
      "loss": 1.1609,
      "step": 2418
    },
    {
      "epoch": 1.2990503674373344,
      "grad_norm": 0.74609375,
      "learning_rate": 4.2169004660129144e-05,
      "loss": 0.9533,
      "step": 2419
    },
    {
      "epoch": 1.2995872621724103,
      "grad_norm": 0.734375,
      "learning_rate": 4.2162865041817935e-05,
      "loss": 1.014,
      "step": 2420
    },
    {
      "epoch": 1.3001241569074864,
      "grad_norm": 0.77734375,
      "learning_rate": 4.215672346500648e-05,
      "loss": 0.9725,
      "step": 2421
    },
    {
      "epoch": 1.3006610516425623,
      "grad_norm": 0.69140625,
      "learning_rate": 4.21505799303956e-05,
      "loss": 0.8765,
      "step": 2422
    },
    {
      "epoch": 1.3011979463776384,
      "grad_norm": 0.6171875,
      "learning_rate": 4.2144434438686355e-05,
      "loss": 0.884,
      "step": 2423
    },
    {
      "epoch": 1.3017348411127143,
      "grad_norm": 0.74609375,
      "learning_rate": 4.2138286990580025e-05,
      "loss": 0.8088,
      "step": 2424
    },
    {
      "epoch": 1.3022717358477904,
      "grad_norm": 0.765625,
      "learning_rate": 4.21321375867781e-05,
      "loss": 0.9909,
      "step": 2425
    },
    {
      "epoch": 1.3028086305828663,
      "grad_norm": 0.68359375,
      "learning_rate": 4.212598622798233e-05,
      "loss": 0.9059,
      "step": 2426
    },
    {
      "epoch": 1.3033455253179422,
      "grad_norm": 0.6328125,
      "learning_rate": 4.211983291489465e-05,
      "loss": 0.7257,
      "step": 2427
    },
    {
      "epoch": 1.3038824200530184,
      "grad_norm": 0.6484375,
      "learning_rate": 4.211367764821722e-05,
      "loss": 0.7937,
      "step": 2428
    },
    {
      "epoch": 1.3044193147880945,
      "grad_norm": 0.640625,
      "learning_rate": 4.2107520428652445e-05,
      "loss": 0.7595,
      "step": 2429
    },
    {
      "epoch": 1.3049562095231704,
      "grad_norm": 0.66796875,
      "learning_rate": 4.210136125690294e-05,
      "loss": 0.8931,
      "step": 2430
    },
    {
      "epoch": 1.3054931042582463,
      "grad_norm": 0.86328125,
      "learning_rate": 4.209520013367155e-05,
      "loss": 0.8578,
      "step": 2431
    },
    {
      "epoch": 1.3060299989933224,
      "grad_norm": 0.6953125,
      "learning_rate": 4.2089037059661326e-05,
      "loss": 0.8389,
      "step": 2432
    },
    {
      "epoch": 1.3065668937283983,
      "grad_norm": 0.88671875,
      "learning_rate": 4.208287203557556e-05,
      "loss": 0.838,
      "step": 2433
    },
    {
      "epoch": 1.3071037884634744,
      "grad_norm": 0.8203125,
      "learning_rate": 4.207670506211776e-05,
      "loss": 1.0361,
      "step": 2434
    },
    {
      "epoch": 1.3076406831985503,
      "grad_norm": 0.68359375,
      "learning_rate": 4.2070536139991644e-05,
      "loss": 0.8776,
      "step": 2435
    },
    {
      "epoch": 1.3081775779336264,
      "grad_norm": 0.7734375,
      "learning_rate": 4.206436526990119e-05,
      "loss": 1.0119,
      "step": 2436
    },
    {
      "epoch": 1.3087144726687023,
      "grad_norm": 0.64453125,
      "learning_rate": 4.2058192452550535e-05,
      "loss": 0.837,
      "step": 2437
    },
    {
      "epoch": 1.3092513674037785,
      "grad_norm": 0.69921875,
      "learning_rate": 4.205201768864412e-05,
      "loss": 1.0506,
      "step": 2438
    },
    {
      "epoch": 1.3097882621388544,
      "grad_norm": 0.859375,
      "learning_rate": 4.204584097888652e-05,
      "loss": 0.9861,
      "step": 2439
    },
    {
      "epoch": 1.3103251568739305,
      "grad_norm": 0.71484375,
      "learning_rate": 4.203966232398261e-05,
      "loss": 0.8611,
      "step": 2440
    },
    {
      "epoch": 1.3108620516090064,
      "grad_norm": 0.61328125,
      "learning_rate": 4.203348172463743e-05,
      "loss": 0.8617,
      "step": 2441
    },
    {
      "epoch": 1.3113989463440825,
      "grad_norm": 0.73828125,
      "learning_rate": 4.202729918155628e-05,
      "loss": 0.8722,
      "step": 2442
    },
    {
      "epoch": 1.3119358410791584,
      "grad_norm": 0.671875,
      "learning_rate": 4.202111469544465e-05,
      "loss": 0.949,
      "step": 2443
    },
    {
      "epoch": 1.3124727358142345,
      "grad_norm": 0.6640625,
      "learning_rate": 4.201492826700829e-05,
      "loss": 0.9245,
      "step": 2444
    },
    {
      "epoch": 1.3130096305493104,
      "grad_norm": 0.71484375,
      "learning_rate": 4.200873989695313e-05,
      "loss": 0.7698,
      "step": 2445
    },
    {
      "epoch": 1.3135465252843863,
      "grad_norm": 0.72265625,
      "learning_rate": 4.200254958598535e-05,
      "loss": 0.8766,
      "step": 2446
    },
    {
      "epoch": 1.3140834200194624,
      "grad_norm": 0.85546875,
      "learning_rate": 4.1996357334811354e-05,
      "loss": 0.9836,
      "step": 2447
    },
    {
      "epoch": 1.3146203147545386,
      "grad_norm": 0.81640625,
      "learning_rate": 4.1990163144137735e-05,
      "loss": 0.9956,
      "step": 2448
    },
    {
      "epoch": 1.3151572094896145,
      "grad_norm": 0.5703125,
      "learning_rate": 4.1983967014671335e-05,
      "loss": 0.6642,
      "step": 2449
    },
    {
      "epoch": 1.3156941042246904,
      "grad_norm": 0.6875,
      "learning_rate": 4.197776894711922e-05,
      "loss": 0.8895,
      "step": 2450
    },
    {
      "epoch": 1.3156941042246904,
      "eval_loss": 0.915617823600769,
      "eval_runtime": 46.2183,
      "eval_samples_per_second": 20.814,
      "eval_steps_per_second": 20.814,
      "step": 2450
    },
    {
      "epoch": 1.3162309989597665,
      "grad_norm": 0.8984375,
      "learning_rate": 4.1971568942188665e-05,
      "loss": 0.7344,
      "step": 2451
    },
    {
      "epoch": 1.3167678936948424,
      "grad_norm": 0.78125,
      "learning_rate": 4.196536700058716e-05,
      "loss": 0.9511,
      "step": 2452
    },
    {
      "epoch": 1.3173047884299185,
      "grad_norm": 0.62890625,
      "learning_rate": 4.195916312302243e-05,
      "loss": 0.7909,
      "step": 2453
    },
    {
      "epoch": 1.3178416831649944,
      "grad_norm": 0.765625,
      "learning_rate": 4.1952957310202424e-05,
      "loss": 1.0219,
      "step": 2454
    },
    {
      "epoch": 1.3183785779000705,
      "grad_norm": 0.66015625,
      "learning_rate": 4.194674956283529e-05,
      "loss": 0.747,
      "step": 2455
    },
    {
      "epoch": 1.3189154726351464,
      "grad_norm": 0.69140625,
      "learning_rate": 4.194053988162943e-05,
      "loss": 0.9002,
      "step": 2456
    },
    {
      "epoch": 1.3194523673702225,
      "grad_norm": 0.6328125,
      "learning_rate": 4.193432826729342e-05,
      "loss": 0.7652,
      "step": 2457
    },
    {
      "epoch": 1.3199892621052984,
      "grad_norm": 0.66015625,
      "learning_rate": 4.1928114720536106e-05,
      "loss": 0.9451,
      "step": 2458
    },
    {
      "epoch": 1.3205261568403746,
      "grad_norm": 0.73828125,
      "learning_rate": 4.192189924206652e-05,
      "loss": 0.8087,
      "step": 2459
    },
    {
      "epoch": 1.3210630515754505,
      "grad_norm": 0.6796875,
      "learning_rate": 4.191568183259394e-05,
      "loss": 0.8087,
      "step": 2460
    },
    {
      "epoch": 1.3215999463105266,
      "grad_norm": 0.671875,
      "learning_rate": 4.1909462492827837e-05,
      "loss": 0.7977,
      "step": 2461
    },
    {
      "epoch": 1.3221368410456025,
      "grad_norm": 0.7265625,
      "learning_rate": 4.190324122347792e-05,
      "loss": 0.8794,
      "step": 2462
    },
    {
      "epoch": 1.3226737357806786,
      "grad_norm": 0.828125,
      "learning_rate": 4.189701802525412e-05,
      "loss": 0.9191,
      "step": 2463
    },
    {
      "epoch": 1.3232106305157545,
      "grad_norm": 0.5703125,
      "learning_rate": 4.189079289886659e-05,
      "loss": 0.7624,
      "step": 2464
    },
    {
      "epoch": 1.3237475252508304,
      "grad_norm": 0.71875,
      "learning_rate": 4.188456584502567e-05,
      "loss": 0.8994,
      "step": 2465
    },
    {
      "epoch": 1.3242844199859065,
      "grad_norm": 0.73828125,
      "learning_rate": 4.187833686444197e-05,
      "loss": 1.0305,
      "step": 2466
    },
    {
      "epoch": 1.3248213147209826,
      "grad_norm": 0.6171875,
      "learning_rate": 4.187210595782628e-05,
      "loss": 0.8002,
      "step": 2467
    },
    {
      "epoch": 1.3253582094560585,
      "grad_norm": 0.65234375,
      "learning_rate": 4.186587312588964e-05,
      "loss": 0.9054,
      "step": 2468
    },
    {
      "epoch": 1.3258951041911344,
      "grad_norm": 0.62890625,
      "learning_rate": 4.185963836934329e-05,
      "loss": 0.9195,
      "step": 2469
    },
    {
      "epoch": 1.3264319989262106,
      "grad_norm": 0.71484375,
      "learning_rate": 4.185340168889868e-05,
      "loss": 0.9654,
      "step": 2470
    },
    {
      "epoch": 1.3269688936612865,
      "grad_norm": 0.65234375,
      "learning_rate": 4.1847163085267514e-05,
      "loss": 0.9004,
      "step": 2471
    },
    {
      "epoch": 1.3275057883963626,
      "grad_norm": 0.87109375,
      "learning_rate": 4.18409225591617e-05,
      "loss": 0.7906,
      "step": 2472
    },
    {
      "epoch": 1.3280426831314385,
      "grad_norm": 0.6640625,
      "learning_rate": 4.183468011129333e-05,
      "loss": 0.8954,
      "step": 2473
    },
    {
      "epoch": 1.3285795778665146,
      "grad_norm": 0.6875,
      "learning_rate": 4.182843574237477e-05,
      "loss": 0.9238,
      "step": 2474
    },
    {
      "epoch": 1.3291164726015905,
      "grad_norm": 0.70703125,
      "learning_rate": 4.182218945311858e-05,
      "loss": 0.8575,
      "step": 2475
    },
    {
      "epoch": 1.3296533673366666,
      "grad_norm": 0.6328125,
      "learning_rate": 4.181594124423754e-05,
      "loss": 0.8888,
      "step": 2476
    },
    {
      "epoch": 1.3301902620717425,
      "grad_norm": 0.828125,
      "learning_rate": 4.180969111644464e-05,
      "loss": 1.0395,
      "step": 2477
    },
    {
      "epoch": 1.3307271568068186,
      "grad_norm": 0.6640625,
      "learning_rate": 4.180343907045311e-05,
      "loss": 0.8024,
      "step": 2478
    },
    {
      "epoch": 1.3312640515418945,
      "grad_norm": 0.6953125,
      "learning_rate": 4.179718510697638e-05,
      "loss": 0.8117,
      "step": 2479
    },
    {
      "epoch": 1.3318009462769707,
      "grad_norm": 0.953125,
      "learning_rate": 4.1790929226728114e-05,
      "loss": 0.9042,
      "step": 2480
    },
    {
      "epoch": 1.3323378410120466,
      "grad_norm": 0.7578125,
      "learning_rate": 4.1784671430422185e-05,
      "loss": 0.975,
      "step": 2481
    },
    {
      "epoch": 1.3328747357471227,
      "grad_norm": 0.60546875,
      "learning_rate": 4.177841171877268e-05,
      "loss": 1.008,
      "step": 2482
    },
    {
      "epoch": 1.3334116304821986,
      "grad_norm": 0.6328125,
      "learning_rate": 4.177215009249392e-05,
      "loss": 0.9612,
      "step": 2483
    },
    {
      "epoch": 1.3339485252172745,
      "grad_norm": 0.875,
      "learning_rate": 4.1765886552300435e-05,
      "loss": 1.0852,
      "step": 2484
    },
    {
      "epoch": 1.3344854199523506,
      "grad_norm": 0.75,
      "learning_rate": 4.175962109890696e-05,
      "loss": 1.0471,
      "step": 2485
    },
    {
      "epoch": 1.3350223146874267,
      "grad_norm": 0.78125,
      "learning_rate": 4.175335373302849e-05,
      "loss": 0.964,
      "step": 2486
    },
    {
      "epoch": 1.3355592094225026,
      "grad_norm": 0.71484375,
      "learning_rate": 4.174708445538018e-05,
      "loss": 0.8976,
      "step": 2487
    },
    {
      "epoch": 1.3360961041575785,
      "grad_norm": 0.62890625,
      "learning_rate": 4.174081326667746e-05,
      "loss": 0.9157,
      "step": 2488
    },
    {
      "epoch": 1.3366329988926546,
      "grad_norm": 0.79296875,
      "learning_rate": 4.173454016763594e-05,
      "loss": 0.736,
      "step": 2489
    },
    {
      "epoch": 1.3371698936277305,
      "grad_norm": 2.0625,
      "learning_rate": 4.172826515897146e-05,
      "loss": 0.8345,
      "step": 2490
    },
    {
      "epoch": 1.3377067883628067,
      "grad_norm": 0.796875,
      "learning_rate": 4.172198824140008e-05,
      "loss": 0.8717,
      "step": 2491
    },
    {
      "epoch": 1.3382436830978826,
      "grad_norm": 0.83203125,
      "learning_rate": 4.1715709415638074e-05,
      "loss": 0.819,
      "step": 2492
    },
    {
      "epoch": 1.3387805778329587,
      "grad_norm": 0.69140625,
      "learning_rate": 4.170942868240193e-05,
      "loss": 1.0067,
      "step": 2493
    },
    {
      "epoch": 1.3393174725680346,
      "grad_norm": 0.6796875,
      "learning_rate": 4.170314604240837e-05,
      "loss": 0.8044,
      "step": 2494
    },
    {
      "epoch": 1.3398543673031107,
      "grad_norm": 0.6953125,
      "learning_rate": 4.1696861496374315e-05,
      "loss": 0.8712,
      "step": 2495
    },
    {
      "epoch": 1.3403912620381866,
      "grad_norm": 0.64453125,
      "learning_rate": 4.169057504501692e-05,
      "loss": 0.9447,
      "step": 2496
    },
    {
      "epoch": 1.3409281567732627,
      "grad_norm": 0.78515625,
      "learning_rate": 4.168428668905354e-05,
      "loss": 0.8681,
      "step": 2497
    },
    {
      "epoch": 1.3414650515083386,
      "grad_norm": 0.7734375,
      "learning_rate": 4.167799642920176e-05,
      "loss": 1.0327,
      "step": 2498
    },
    {
      "epoch": 1.3420019462434145,
      "grad_norm": 0.8203125,
      "learning_rate": 4.167170426617938e-05,
      "loss": 0.7492,
      "step": 2499
    },
    {
      "epoch": 1.3425388409784906,
      "grad_norm": 0.62109375,
      "learning_rate": 4.166541020070441e-05,
      "loss": 0.9581,
      "step": 2500
    },
    {
      "epoch": 1.3425388409784906,
      "eval_loss": 0.9149090051651001,
      "eval_runtime": 45.9495,
      "eval_samples_per_second": 20.936,
      "eval_steps_per_second": 20.936,
      "step": 2500
    },
    {
      "epoch": 1.3430757357135668,
      "grad_norm": 0.7578125,
      "learning_rate": 4.165911423349509e-05,
      "loss": 1.0437,
      "step": 2501
    },
    {
      "epoch": 1.3436126304486427,
      "grad_norm": 0.671875,
      "learning_rate": 4.1652816365269866e-05,
      "loss": 0.9115,
      "step": 2502
    },
    {
      "epoch": 1.3441495251837186,
      "grad_norm": 0.765625,
      "learning_rate": 4.1646516596747395e-05,
      "loss": 0.8741,
      "step": 2503
    },
    {
      "epoch": 1.3446864199187947,
      "grad_norm": 0.66796875,
      "learning_rate": 4.1640214928646584e-05,
      "loss": 0.857,
      "step": 2504
    },
    {
      "epoch": 1.3452233146538708,
      "grad_norm": 0.67578125,
      "learning_rate": 4.163391136168651e-05,
      "loss": 0.8804,
      "step": 2505
    },
    {
      "epoch": 1.3457602093889467,
      "grad_norm": 0.81640625,
      "learning_rate": 4.162760589658649e-05,
      "loss": 0.7245,
      "step": 2506
    },
    {
      "epoch": 1.3462971041240226,
      "grad_norm": 0.6171875,
      "learning_rate": 4.162129853406607e-05,
      "loss": 0.8196,
      "step": 2507
    },
    {
      "epoch": 1.3468339988590987,
      "grad_norm": 0.7421875,
      "learning_rate": 4.1614989274845004e-05,
      "loss": 1.0052,
      "step": 2508
    },
    {
      "epoch": 1.3473708935941746,
      "grad_norm": 1.1171875,
      "learning_rate": 4.160867811964324e-05,
      "loss": 0.9977,
      "step": 2509
    },
    {
      "epoch": 1.3479077883292507,
      "grad_norm": 0.6796875,
      "learning_rate": 4.160236506918098e-05,
      "loss": 0.9316,
      "step": 2510
    },
    {
      "epoch": 1.3484446830643266,
      "grad_norm": 0.6875,
      "learning_rate": 4.159605012417861e-05,
      "loss": 0.9938,
      "step": 2511
    },
    {
      "epoch": 1.3489815777994028,
      "grad_norm": 0.609375,
      "learning_rate": 4.158973328535675e-05,
      "loss": 0.8452,
      "step": 2512
    },
    {
      "epoch": 1.3495184725344787,
      "grad_norm": 0.671875,
      "learning_rate": 4.158341455343622e-05,
      "loss": 0.7538,
      "step": 2513
    },
    {
      "epoch": 1.3500553672695548,
      "grad_norm": 0.62109375,
      "learning_rate": 4.1577093929138087e-05,
      "loss": 0.8579,
      "step": 2514
    },
    {
      "epoch": 1.3505922620046307,
      "grad_norm": 0.6796875,
      "learning_rate": 4.1570771413183606e-05,
      "loss": 0.7398,
      "step": 2515
    },
    {
      "epoch": 1.3511291567397068,
      "grad_norm": 1.046875,
      "learning_rate": 4.156444700629425e-05,
      "loss": 1.0352,
      "step": 2516
    },
    {
      "epoch": 1.3516660514747827,
      "grad_norm": 0.70703125,
      "learning_rate": 4.155812070919172e-05,
      "loss": 0.9783,
      "step": 2517
    },
    {
      "epoch": 1.3522029462098586,
      "grad_norm": 0.7890625,
      "learning_rate": 4.155179252259793e-05,
      "loss": 0.8666,
      "step": 2518
    },
    {
      "epoch": 1.3527398409449347,
      "grad_norm": 0.76953125,
      "learning_rate": 4.1545462447234995e-05,
      "loss": 1.0115,
      "step": 2519
    },
    {
      "epoch": 1.3532767356800108,
      "grad_norm": 0.5859375,
      "learning_rate": 4.1539130483825266e-05,
      "loss": 0.8279,
      "step": 2520
    },
    {
      "epoch": 1.3538136304150867,
      "grad_norm": 0.69140625,
      "learning_rate": 4.1532796633091296e-05,
      "loss": 0.853,
      "step": 2521
    },
    {
      "epoch": 1.3543505251501626,
      "grad_norm": 0.6328125,
      "learning_rate": 4.152646089575587e-05,
      "loss": 0.7318,
      "step": 2522
    },
    {
      "epoch": 1.3548874198852388,
      "grad_norm": 0.77734375,
      "learning_rate": 4.152012327254196e-05,
      "loss": 0.8343,
      "step": 2523
    },
    {
      "epoch": 1.3554243146203149,
      "grad_norm": 0.67578125,
      "learning_rate": 4.1513783764172765e-05,
      "loss": 0.7412,
      "step": 2524
    },
    {
      "epoch": 1.3559612093553908,
      "grad_norm": 0.92578125,
      "learning_rate": 4.150744237137171e-05,
      "loss": 0.9574,
      "step": 2525
    },
    {
      "epoch": 1.3564981040904667,
      "grad_norm": 0.62890625,
      "learning_rate": 4.1501099094862444e-05,
      "loss": 0.8173,
      "step": 2526
    },
    {
      "epoch": 1.3570349988255428,
      "grad_norm": 0.63671875,
      "learning_rate": 4.14947539353688e-05,
      "loss": 0.8508,
      "step": 2527
    },
    {
      "epoch": 1.3575718935606187,
      "grad_norm": 2.046875,
      "learning_rate": 4.148840689361484e-05,
      "loss": 0.8877,
      "step": 2528
    },
    {
      "epoch": 1.3581087882956948,
      "grad_norm": 0.76171875,
      "learning_rate": 4.1482057970324837e-05,
      "loss": 0.935,
      "step": 2529
    },
    {
      "epoch": 1.3586456830307707,
      "grad_norm": 0.78125,
      "learning_rate": 4.1475707166223296e-05,
      "loss": 0.8328,
      "step": 2530
    },
    {
      "epoch": 1.3591825777658468,
      "grad_norm": 0.70703125,
      "learning_rate": 4.1469354482034924e-05,
      "loss": 0.8265,
      "step": 2531
    },
    {
      "epoch": 1.3597194725009227,
      "grad_norm": 0.68359375,
      "learning_rate": 4.146299991848464e-05,
      "loss": 0.8488,
      "step": 2532
    },
    {
      "epoch": 1.3602563672359989,
      "grad_norm": 0.609375,
      "learning_rate": 4.1456643476297566e-05,
      "loss": 0.6766,
      "step": 2533
    },
    {
      "epoch": 1.3607932619710748,
      "grad_norm": 0.62890625,
      "learning_rate": 4.145028515619907e-05,
      "loss": 0.9599,
      "step": 2534
    },
    {
      "epoch": 1.3613301567061509,
      "grad_norm": 0.66796875,
      "learning_rate": 4.144392495891473e-05,
      "loss": 0.7632,
      "step": 2535
    },
    {
      "epoch": 1.3618670514412268,
      "grad_norm": 0.7265625,
      "learning_rate": 4.143756288517028e-05,
      "loss": 0.8555,
      "step": 2536
    },
    {
      "epoch": 1.3624039461763027,
      "grad_norm": 0.6328125,
      "learning_rate": 4.143119893569175e-05,
      "loss": 0.8442,
      "step": 2537
    },
    {
      "epoch": 1.3629408409113788,
      "grad_norm": 0.9140625,
      "learning_rate": 4.142483311120534e-05,
      "loss": 1.082,
      "step": 2538
    },
    {
      "epoch": 1.363477735646455,
      "grad_norm": 0.73046875,
      "learning_rate": 4.1418465412437467e-05,
      "loss": 0.847,
      "step": 2539
    },
    {
      "epoch": 1.3640146303815308,
      "grad_norm": 0.9140625,
      "learning_rate": 4.1412095840114764e-05,
      "loss": 0.9993,
      "step": 2540
    },
    {
      "epoch": 1.3645515251166067,
      "grad_norm": 0.6796875,
      "learning_rate": 4.140572439496409e-05,
      "loss": 0.9044,
      "step": 2541
    },
    {
      "epoch": 1.3650884198516828,
      "grad_norm": 0.86328125,
      "learning_rate": 4.1399351077712496e-05,
      "loss": 0.9764,
      "step": 2542
    },
    {
      "epoch": 1.365625314586759,
      "grad_norm": 0.63671875,
      "learning_rate": 4.1392975889087266e-05,
      "loss": 0.7391,
      "step": 2543
    },
    {
      "epoch": 1.3661622093218349,
      "grad_norm": 0.70703125,
      "learning_rate": 4.138659882981588e-05,
      "loss": 0.7217,
      "step": 2544
    },
    {
      "epoch": 1.3666991040569108,
      "grad_norm": 0.64453125,
      "learning_rate": 4.138021990062606e-05,
      "loss": 0.909,
      "step": 2545
    },
    {
      "epoch": 1.3672359987919869,
      "grad_norm": 0.6796875,
      "learning_rate": 4.13738391022457e-05,
      "loss": 0.9635,
      "step": 2546
    },
    {
      "epoch": 1.3677728935270628,
      "grad_norm": 0.70703125,
      "learning_rate": 4.136745643540295e-05,
      "loss": 0.7904,
      "step": 2547
    },
    {
      "epoch": 1.368309788262139,
      "grad_norm": 1.0234375,
      "learning_rate": 4.136107190082613e-05,
      "loss": 0.8989,
      "step": 2548
    },
    {
      "epoch": 1.3688466829972148,
      "grad_norm": 0.71484375,
      "learning_rate": 4.135468549924382e-05,
      "loss": 0.744,
      "step": 2549
    },
    {
      "epoch": 1.369383577732291,
      "grad_norm": 0.6875,
      "learning_rate": 4.134829723138477e-05,
      "loss": 0.7616,
      "step": 2550
    },
    {
      "epoch": 1.369383577732291,
      "eval_loss": 0.9148359894752502,
      "eval_runtime": 46.512,
      "eval_samples_per_second": 20.683,
      "eval_steps_per_second": 20.683,
      "step": 2550
    },
    {
      "epoch": 1.3699204724673668,
      "grad_norm": 0.72265625,
      "learning_rate": 4.1341907097977985e-05,
      "loss": 0.7684,
      "step": 2551
    },
    {
      "epoch": 1.370457367202443,
      "grad_norm": 0.6015625,
      "learning_rate": 4.133551509975264e-05,
      "loss": 0.8445,
      "step": 2552
    },
    {
      "epoch": 1.3709942619375188,
      "grad_norm": 0.65625,
      "learning_rate": 4.132912123743815e-05,
      "loss": 1.0183,
      "step": 2553
    },
    {
      "epoch": 1.371531156672595,
      "grad_norm": 0.640625,
      "learning_rate": 4.132272551176413e-05,
      "loss": 0.7382,
      "step": 2554
    },
    {
      "epoch": 1.3720680514076709,
      "grad_norm": 0.76953125,
      "learning_rate": 4.131632792346042e-05,
      "loss": 0.8819,
      "step": 2555
    },
    {
      "epoch": 1.3726049461427468,
      "grad_norm": 0.875,
      "learning_rate": 4.130992847325707e-05,
      "loss": 0.826,
      "step": 2556
    },
    {
      "epoch": 1.3731418408778229,
      "grad_norm": 0.76171875,
      "learning_rate": 4.1303527161884336e-05,
      "loss": 0.8119,
      "step": 2557
    },
    {
      "epoch": 1.373678735612899,
      "grad_norm": 0.81640625,
      "learning_rate": 4.129712399007267e-05,
      "loss": 0.8714,
      "step": 2558
    },
    {
      "epoch": 1.374215630347975,
      "grad_norm": 0.8125,
      "learning_rate": 4.129071895855279e-05,
      "loss": 1.1081,
      "step": 2559
    },
    {
      "epoch": 1.3747525250830508,
      "grad_norm": 0.64453125,
      "learning_rate": 4.128431206805557e-05,
      "loss": 0.7996,
      "step": 2560
    },
    {
      "epoch": 1.375289419818127,
      "grad_norm": 0.75390625,
      "learning_rate": 4.12779033193121e-05,
      "loss": 1.1367,
      "step": 2561
    },
    {
      "epoch": 1.375826314553203,
      "grad_norm": 0.7890625,
      "learning_rate": 4.1271492713053725e-05,
      "loss": 0.8228,
      "step": 2562
    },
    {
      "epoch": 1.376363209288279,
      "grad_norm": 0.76171875,
      "learning_rate": 4.126508025001198e-05,
      "loss": 0.9288,
      "step": 2563
    },
    {
      "epoch": 1.3769001040233548,
      "grad_norm": 0.91015625,
      "learning_rate": 4.125866593091859e-05,
      "loss": 1.1671,
      "step": 2564
    },
    {
      "epoch": 1.377436998758431,
      "grad_norm": 0.8984375,
      "learning_rate": 4.1252249756505514e-05,
      "loss": 0.9771,
      "step": 2565
    },
    {
      "epoch": 1.3779738934935069,
      "grad_norm": 0.82421875,
      "learning_rate": 4.1245831727504925e-05,
      "loss": 0.9341,
      "step": 2566
    },
    {
      "epoch": 1.378510788228583,
      "grad_norm": 0.75,
      "learning_rate": 4.123941184464921e-05,
      "loss": 1.0186,
      "step": 2567
    },
    {
      "epoch": 1.3790476829636589,
      "grad_norm": 0.58984375,
      "learning_rate": 4.1232990108670935e-05,
      "loss": 0.7287,
      "step": 2568
    },
    {
      "epoch": 1.379584577698735,
      "grad_norm": 0.7734375,
      "learning_rate": 4.122656652030292e-05,
      "loss": 0.8416,
      "step": 2569
    },
    {
      "epoch": 1.380121472433811,
      "grad_norm": 0.71875,
      "learning_rate": 4.122014108027817e-05,
      "loss": 1.0048,
      "step": 2570
    },
    {
      "epoch": 1.380658367168887,
      "grad_norm": 0.703125,
      "learning_rate": 4.121371378932991e-05,
      "loss": 0.9034,
      "step": 2571
    },
    {
      "epoch": 1.381195261903963,
      "grad_norm": 0.61328125,
      "learning_rate": 4.120728464819158e-05,
      "loss": 0.7797,
      "step": 2572
    },
    {
      "epoch": 1.381732156639039,
      "grad_norm": 0.59375,
      "learning_rate": 4.1200853657596826e-05,
      "loss": 0.7304,
      "step": 2573
    },
    {
      "epoch": 1.382269051374115,
      "grad_norm": 0.59375,
      "learning_rate": 4.11944208182795e-05,
      "loss": 0.7157,
      "step": 2574
    },
    {
      "epoch": 1.3828059461091908,
      "grad_norm": 0.67578125,
      "learning_rate": 4.118798613097366e-05,
      "loss": 0.9234,
      "step": 2575
    },
    {
      "epoch": 1.383342840844267,
      "grad_norm": 0.8203125,
      "learning_rate": 4.118154959641362e-05,
      "loss": 0.9838,
      "step": 2576
    },
    {
      "epoch": 1.383879735579343,
      "grad_norm": 0.65234375,
      "learning_rate": 4.117511121533384e-05,
      "loss": 0.7853,
      "step": 2577
    },
    {
      "epoch": 1.384416630314419,
      "grad_norm": 0.62890625,
      "learning_rate": 4.116867098846903e-05,
      "loss": 0.8374,
      "step": 2578
    },
    {
      "epoch": 1.3849535250494949,
      "grad_norm": 1.046875,
      "learning_rate": 4.116222891655409e-05,
      "loss": 1.1056,
      "step": 2579
    },
    {
      "epoch": 1.385490419784571,
      "grad_norm": 0.65234375,
      "learning_rate": 4.1155785000324165e-05,
      "loss": 0.8676,
      "step": 2580
    },
    {
      "epoch": 1.3860273145196471,
      "grad_norm": 0.734375,
      "learning_rate": 4.1149339240514574e-05,
      "loss": 0.8654,
      "step": 2581
    },
    {
      "epoch": 1.386564209254723,
      "grad_norm": 0.84765625,
      "learning_rate": 4.1142891637860855e-05,
      "loss": 0.9874,
      "step": 2582
    },
    {
      "epoch": 1.387101103989799,
      "grad_norm": 0.58203125,
      "learning_rate": 4.113644219309877e-05,
      "loss": 0.8685,
      "step": 2583
    },
    {
      "epoch": 1.387637998724875,
      "grad_norm": 0.7578125,
      "learning_rate": 4.112999090696427e-05,
      "loss": 1.203,
      "step": 2584
    },
    {
      "epoch": 1.388174893459951,
      "grad_norm": 0.6953125,
      "learning_rate": 4.112353778019355e-05,
      "loss": 0.9733,
      "step": 2585
    },
    {
      "epoch": 1.388711788195027,
      "grad_norm": 0.62109375,
      "learning_rate": 4.111708281352298e-05,
      "loss": 0.8137,
      "step": 2586
    },
    {
      "epoch": 1.389248682930103,
      "grad_norm": 0.64453125,
      "learning_rate": 4.1110626007689145e-05,
      "loss": 0.9915,
      "step": 2587
    },
    {
      "epoch": 1.389785577665179,
      "grad_norm": 0.69140625,
      "learning_rate": 4.1104167363428866e-05,
      "loss": 0.8851,
      "step": 2588
    },
    {
      "epoch": 1.390322472400255,
      "grad_norm": 0.67578125,
      "learning_rate": 4.109770688147915e-05,
      "loss": 0.8586,
      "step": 2589
    },
    {
      "epoch": 1.390859367135331,
      "grad_norm": 0.71484375,
      "learning_rate": 4.109124456257721e-05,
      "loss": 0.6883,
      "step": 2590
    },
    {
      "epoch": 1.391396261870407,
      "grad_norm": 0.6640625,
      "learning_rate": 4.10847804074605e-05,
      "loss": 0.9585,
      "step": 2591
    },
    {
      "epoch": 1.3919331566054831,
      "grad_norm": 0.609375,
      "learning_rate": 4.1078314416866636e-05,
      "loss": 0.7392,
      "step": 2592
    },
    {
      "epoch": 1.392470051340559,
      "grad_norm": 0.640625,
      "learning_rate": 4.107184659153348e-05,
      "loss": 1.0139,
      "step": 2593
    },
    {
      "epoch": 1.393006946075635,
      "grad_norm": 0.6796875,
      "learning_rate": 4.1065376932199104e-05,
      "loss": 0.8706,
      "step": 2594
    },
    {
      "epoch": 1.393543840810711,
      "grad_norm": 0.875,
      "learning_rate": 4.1058905439601766e-05,
      "loss": 0.9932,
      "step": 2595
    },
    {
      "epoch": 1.3940807355457872,
      "grad_norm": 0.72265625,
      "learning_rate": 4.1052432114479946e-05,
      "loss": 0.8429,
      "step": 2596
    },
    {
      "epoch": 1.394617630280863,
      "grad_norm": 0.75,
      "learning_rate": 4.104595695757233e-05,
      "loss": 0.7455,
      "step": 2597
    },
    {
      "epoch": 1.395154525015939,
      "grad_norm": 0.68359375,
      "learning_rate": 4.103947996961783e-05,
      "loss": 0.6963,
      "step": 2598
    },
    {
      "epoch": 1.395691419751015,
      "grad_norm": 0.8984375,
      "learning_rate": 4.103300115135553e-05,
      "loss": 0.9878,
      "step": 2599
    },
    {
      "epoch": 1.3962283144860912,
      "grad_norm": 0.64453125,
      "learning_rate": 4.1026520503524765e-05,
      "loss": 0.77,
      "step": 2600
    },
    {
      "epoch": 1.3962283144860912,
      "eval_loss": 0.914279043674469,
      "eval_runtime": 46.8263,
      "eval_samples_per_second": 20.544,
      "eval_steps_per_second": 20.544,
      "step": 2600
    },
    {
      "epoch": 1.396765209221167,
      "grad_norm": 0.703125,
      "learning_rate": 4.102003802686505e-05,
      "loss": 0.8192,
      "step": 2601
    },
    {
      "epoch": 1.397302103956243,
      "grad_norm": 0.671875,
      "learning_rate": 4.1013553722116114e-05,
      "loss": 0.8407,
      "step": 2602
    },
    {
      "epoch": 1.3978389986913191,
      "grad_norm": 0.84375,
      "learning_rate": 4.100706759001791e-05,
      "loss": 0.9569,
      "step": 2603
    },
    {
      "epoch": 1.398375893426395,
      "grad_norm": 0.70703125,
      "learning_rate": 4.1000579631310574e-05,
      "loss": 1.0608,
      "step": 2604
    },
    {
      "epoch": 1.3989127881614711,
      "grad_norm": 0.75,
      "learning_rate": 4.099408984673447e-05,
      "loss": 0.8907,
      "step": 2605
    },
    {
      "epoch": 1.399449682896547,
      "grad_norm": 0.76953125,
      "learning_rate": 4.098759823703018e-05,
      "loss": 0.9854,
      "step": 2606
    },
    {
      "epoch": 1.3999865776316232,
      "grad_norm": 0.6171875,
      "learning_rate": 4.098110480293845e-05,
      "loss": 0.8458,
      "step": 2607
    },
    {
      "epoch": 1.400523472366699,
      "grad_norm": 0.6953125,
      "learning_rate": 4.097460954520028e-05,
      "loss": 0.7117,
      "step": 2608
    },
    {
      "epoch": 1.4010603671017752,
      "grad_norm": 0.72265625,
      "learning_rate": 4.0968112464556864e-05,
      "loss": 0.8811,
      "step": 2609
    },
    {
      "epoch": 1.401597261836851,
      "grad_norm": 0.6875,
      "learning_rate": 4.096161356174959e-05,
      "loss": 1.094,
      "step": 2610
    },
    {
      "epoch": 1.4021341565719272,
      "grad_norm": 0.73828125,
      "learning_rate": 4.095511283752007e-05,
      "loss": 0.8327,
      "step": 2611
    },
    {
      "epoch": 1.402671051307003,
      "grad_norm": 0.67578125,
      "learning_rate": 4.094861029261013e-05,
      "loss": 0.768,
      "step": 2612
    },
    {
      "epoch": 1.403207946042079,
      "grad_norm": 0.7265625,
      "learning_rate": 4.0942105927761776e-05,
      "loss": 0.8954,
      "step": 2613
    },
    {
      "epoch": 1.4037448407771551,
      "grad_norm": 0.7109375,
      "learning_rate": 4.093559974371725e-05,
      "loss": 0.8423,
      "step": 2614
    },
    {
      "epoch": 1.4042817355122312,
      "grad_norm": 0.6953125,
      "learning_rate": 4.092909174121898e-05,
      "loss": 0.8002,
      "step": 2615
    },
    {
      "epoch": 1.4048186302473071,
      "grad_norm": 0.83203125,
      "learning_rate": 4.0922581921009615e-05,
      "loss": 1.0378,
      "step": 2616
    },
    {
      "epoch": 1.405355524982383,
      "grad_norm": 0.65625,
      "learning_rate": 4.091607028383201e-05,
      "loss": 0.8302,
      "step": 2617
    },
    {
      "epoch": 1.4058924197174592,
      "grad_norm": 0.75,
      "learning_rate": 4.090955683042923e-05,
      "loss": 0.8857,
      "step": 2618
    },
    {
      "epoch": 1.4064293144525353,
      "grad_norm": 0.66015625,
      "learning_rate": 4.0903041561544534e-05,
      "loss": 0.8998,
      "step": 2619
    },
    {
      "epoch": 1.4069662091876112,
      "grad_norm": 0.734375,
      "learning_rate": 4.08965244779214e-05,
      "loss": 0.9169,
      "step": 2620
    },
    {
      "epoch": 1.407503103922687,
      "grad_norm": 0.6796875,
      "learning_rate": 4.0890005580303516e-05,
      "loss": 0.7286,
      "step": 2621
    },
    {
      "epoch": 1.4080399986577632,
      "grad_norm": 0.703125,
      "learning_rate": 4.088348486943476e-05,
      "loss": 0.8804,
      "step": 2622
    },
    {
      "epoch": 1.408576893392839,
      "grad_norm": 0.76953125,
      "learning_rate": 4.087696234605923e-05,
      "loss": 0.9798,
      "step": 2623
    },
    {
      "epoch": 1.4091137881279152,
      "grad_norm": 0.70703125,
      "learning_rate": 4.087043801092123e-05,
      "loss": 0.9896,
      "step": 2624
    },
    {
      "epoch": 1.4096506828629911,
      "grad_norm": 0.64453125,
      "learning_rate": 4.0863911864765274e-05,
      "loss": 0.9417,
      "step": 2625
    },
    {
      "epoch": 1.4101875775980672,
      "grad_norm": 0.69921875,
      "learning_rate": 4.0857383908336076e-05,
      "loss": 0.948,
      "step": 2626
    },
    {
      "epoch": 1.4107244723331431,
      "grad_norm": 0.66015625,
      "learning_rate": 4.085085414237856e-05,
      "loss": 0.9259,
      "step": 2627
    },
    {
      "epoch": 1.4112613670682193,
      "grad_norm": 0.55859375,
      "learning_rate": 4.084432256763784e-05,
      "loss": 0.7512,
      "step": 2628
    },
    {
      "epoch": 1.4117982618032952,
      "grad_norm": 0.7578125,
      "learning_rate": 4.083778918485926e-05,
      "loss": 0.8592,
      "step": 2629
    },
    {
      "epoch": 1.4123351565383713,
      "grad_norm": 0.84375,
      "learning_rate": 4.083125399478838e-05,
      "loss": 0.8346,
      "step": 2630
    },
    {
      "epoch": 1.4128720512734472,
      "grad_norm": 0.73828125,
      "learning_rate": 4.0824716998170926e-05,
      "loss": 0.8997,
      "step": 2631
    },
    {
      "epoch": 1.413408946008523,
      "grad_norm": 0.65234375,
      "learning_rate": 4.081817819575285e-05,
      "loss": 0.9205,
      "step": 2632
    },
    {
      "epoch": 1.4139458407435992,
      "grad_norm": 0.6875,
      "learning_rate": 4.0811637588280335e-05,
      "loss": 0.8972,
      "step": 2633
    },
    {
      "epoch": 1.4144827354786753,
      "grad_norm": 0.6875,
      "learning_rate": 4.0805095176499724e-05,
      "loss": 0.8703,
      "step": 2634
    },
    {
      "epoch": 1.4150196302137512,
      "grad_norm": 0.59765625,
      "learning_rate": 4.07985509611576e-05,
      "loss": 0.7751,
      "step": 2635
    },
    {
      "epoch": 1.415556524948827,
      "grad_norm": 0.640625,
      "learning_rate": 4.079200494300074e-05,
      "loss": 0.7056,
      "step": 2636
    },
    {
      "epoch": 1.4160934196839032,
      "grad_norm": 0.83984375,
      "learning_rate": 4.0785457122776125e-05,
      "loss": 0.8981,
      "step": 2637
    },
    {
      "epoch": 1.4166303144189794,
      "grad_norm": 0.66015625,
      "learning_rate": 4.077890750123095e-05,
      "loss": 0.8363,
      "step": 2638
    },
    {
      "epoch": 1.4171672091540553,
      "grad_norm": 0.83984375,
      "learning_rate": 4.07723560791126e-05,
      "loss": 0.9147,
      "step": 2639
    },
    {
      "epoch": 1.4177041038891312,
      "grad_norm": 0.7265625,
      "learning_rate": 4.0765802857168687e-05,
      "loss": 0.9733,
      "step": 2640
    },
    {
      "epoch": 1.4182409986242073,
      "grad_norm": 0.66796875,
      "learning_rate": 4.0759247836147005e-05,
      "loss": 0.9805,
      "step": 2641
    },
    {
      "epoch": 1.4187778933592832,
      "grad_norm": 0.7265625,
      "learning_rate": 4.075269101679557e-05,
      "loss": 0.7929,
      "step": 2642
    },
    {
      "epoch": 1.4193147880943593,
      "grad_norm": 0.80859375,
      "learning_rate": 4.0746132399862605e-05,
      "loss": 0.8564,
      "step": 2643
    },
    {
      "epoch": 1.4198516828294352,
      "grad_norm": 0.5546875,
      "learning_rate": 4.073957198609651e-05,
      "loss": 0.7691,
      "step": 2644
    },
    {
      "epoch": 1.4203885775645113,
      "grad_norm": 0.6796875,
      "learning_rate": 4.073300977624594e-05,
      "loss": 0.8256,
      "step": 2645
    },
    {
      "epoch": 1.4209254722995872,
      "grad_norm": 0.65234375,
      "learning_rate": 4.072644577105971e-05,
      "loss": 0.8678,
      "step": 2646
    },
    {
      "epoch": 1.4214623670346633,
      "grad_norm": 0.66015625,
      "learning_rate": 4.071987997128685e-05,
      "loss": 0.8509,
      "step": 2647
    },
    {
      "epoch": 1.4219992617697392,
      "grad_norm": 0.69140625,
      "learning_rate": 4.071331237767662e-05,
      "loss": 0.9182,
      "step": 2648
    },
    {
      "epoch": 1.4225361565048154,
      "grad_norm": 0.640625,
      "learning_rate": 4.070674299097845e-05,
      "loss": 0.7763,
      "step": 2649
    },
    {
      "epoch": 1.4230730512398913,
      "grad_norm": 0.71875,
      "learning_rate": 4.070017181194199e-05,
      "loss": 0.8474,
      "step": 2650
    },
    {
      "epoch": 1.4230730512398913,
      "eval_loss": 0.9133580923080444,
      "eval_runtime": 46.4653,
      "eval_samples_per_second": 20.704,
      "eval_steps_per_second": 20.704,
      "step": 2650
    },
    {
      "epoch": 1.4236099459749672,
      "grad_norm": 0.76953125,
      "learning_rate": 4.06935988413171e-05,
      "loss": 0.7508,
      "step": 2651
    },
    {
      "epoch": 1.4241468407100433,
      "grad_norm": 0.68359375,
      "learning_rate": 4.068702407985384e-05,
      "loss": 0.9745,
      "step": 2652
    },
    {
      "epoch": 1.4246837354451194,
      "grad_norm": 0.68359375,
      "learning_rate": 4.068044752830248e-05,
      "loss": 0.7427,
      "step": 2653
    },
    {
      "epoch": 1.4252206301801953,
      "grad_norm": 0.71875,
      "learning_rate": 4.067386918741347e-05,
      "loss": 0.9359,
      "step": 2654
    },
    {
      "epoch": 1.4257575249152712,
      "grad_norm": 0.671875,
      "learning_rate": 4.066728905793749e-05,
      "loss": 0.8267,
      "step": 2655
    },
    {
      "epoch": 1.4262944196503473,
      "grad_norm": 0.68359375,
      "learning_rate": 4.0660707140625434e-05,
      "loss": 0.8708,
      "step": 2656
    },
    {
      "epoch": 1.4268313143854234,
      "grad_norm": 0.6953125,
      "learning_rate": 4.065412343622836e-05,
      "loss": 0.9951,
      "step": 2657
    },
    {
      "epoch": 1.4273682091204993,
      "grad_norm": 0.6875,
      "learning_rate": 4.064753794549756e-05,
      "loss": 0.8867,
      "step": 2658
    },
    {
      "epoch": 1.4279051038555752,
      "grad_norm": 0.86328125,
      "learning_rate": 4.064095066918451e-05,
      "loss": 1.146,
      "step": 2659
    },
    {
      "epoch": 1.4284419985906514,
      "grad_norm": 0.84375,
      "learning_rate": 4.063436160804092e-05,
      "loss": 0.774,
      "step": 2660
    },
    {
      "epoch": 1.4289788933257273,
      "grad_norm": 0.79296875,
      "learning_rate": 4.0627770762818684e-05,
      "loss": 0.8943,
      "step": 2661
    },
    {
      "epoch": 1.4295157880608034,
      "grad_norm": 0.609375,
      "learning_rate": 4.0621178134269884e-05,
      "loss": 0.8359,
      "step": 2662
    },
    {
      "epoch": 1.4300526827958793,
      "grad_norm": 0.62890625,
      "learning_rate": 4.061458372314684e-05,
      "loss": 0.8759,
      "step": 2663
    },
    {
      "epoch": 1.4305895775309554,
      "grad_norm": 0.5859375,
      "learning_rate": 4.060798753020204e-05,
      "loss": 0.8779,
      "step": 2664
    },
    {
      "epoch": 1.4311264722660313,
      "grad_norm": 0.79296875,
      "learning_rate": 4.060138955618821e-05,
      "loss": 0.9663,
      "step": 2665
    },
    {
      "epoch": 1.4316633670011074,
      "grad_norm": 0.81640625,
      "learning_rate": 4.059478980185826e-05,
      "loss": 1.3966,
      "step": 2666
    },
    {
      "epoch": 1.4322002617361833,
      "grad_norm": 0.8046875,
      "learning_rate": 4.058818826796529e-05,
      "loss": 0.9054,
      "step": 2667
    },
    {
      "epoch": 1.4327371564712594,
      "grad_norm": 0.7265625,
      "learning_rate": 4.058158495526264e-05,
      "loss": 0.9305,
      "step": 2668
    },
    {
      "epoch": 1.4332740512063353,
      "grad_norm": 0.7578125,
      "learning_rate": 4.0574979864503814e-05,
      "loss": 0.9078,
      "step": 2669
    },
    {
      "epoch": 1.4338109459414112,
      "grad_norm": 0.72265625,
      "learning_rate": 4.056837299644255e-05,
      "loss": 0.8311,
      "step": 2670
    },
    {
      "epoch": 1.4343478406764874,
      "grad_norm": 0.6328125,
      "learning_rate": 4.056176435183277e-05,
      "loss": 0.8182,
      "step": 2671
    },
    {
      "epoch": 1.4348847354115635,
      "grad_norm": 0.6875,
      "learning_rate": 4.0555153931428605e-05,
      "loss": 0.7843,
      "step": 2672
    },
    {
      "epoch": 1.4354216301466394,
      "grad_norm": 0.87109375,
      "learning_rate": 4.0548541735984385e-05,
      "loss": 1.1244,
      "step": 2673
    },
    {
      "epoch": 1.4359585248817153,
      "grad_norm": 0.7578125,
      "learning_rate": 4.054192776625464e-05,
      "loss": 1.0792,
      "step": 2674
    },
    {
      "epoch": 1.4364954196167914,
      "grad_norm": 0.76953125,
      "learning_rate": 4.053531202299412e-05,
      "loss": 1.007,
      "step": 2675
    },
    {
      "epoch": 1.4370323143518675,
      "grad_norm": 0.90625,
      "learning_rate": 4.052869450695776e-05,
      "loss": 0.9834,
      "step": 2676
    },
    {
      "epoch": 1.4375692090869434,
      "grad_norm": 0.76171875,
      "learning_rate": 4.0522075218900704e-05,
      "loss": 0.9715,
      "step": 2677
    },
    {
      "epoch": 1.4381061038220193,
      "grad_norm": 0.67578125,
      "learning_rate": 4.051545415957829e-05,
      "loss": 0.7714,
      "step": 2678
    },
    {
      "epoch": 1.4386429985570954,
      "grad_norm": 0.640625,
      "learning_rate": 4.050883132974607e-05,
      "loss": 0.7648,
      "step": 2679
    },
    {
      "epoch": 1.4391798932921713,
      "grad_norm": 0.7265625,
      "learning_rate": 4.050220673015979e-05,
      "loss": 0.9879,
      "step": 2680
    },
    {
      "epoch": 1.4397167880272475,
      "grad_norm": 0.671875,
      "learning_rate": 4.0495580361575405e-05,
      "loss": 0.8947,
      "step": 2681
    },
    {
      "epoch": 1.4402536827623234,
      "grad_norm": 0.7109375,
      "learning_rate": 4.0488952224749066e-05,
      "loss": 0.8378,
      "step": 2682
    },
    {
      "epoch": 1.4407905774973995,
      "grad_norm": 0.796875,
      "learning_rate": 4.048232232043713e-05,
      "loss": 0.9889,
      "step": 2683
    },
    {
      "epoch": 1.4413274722324754,
      "grad_norm": 0.765625,
      "learning_rate": 4.0475690649396134e-05,
      "loss": 0.952,
      "step": 2684
    },
    {
      "epoch": 1.4418643669675515,
      "grad_norm": 0.828125,
      "learning_rate": 4.0469057212382864e-05,
      "loss": 1.0912,
      "step": 2685
    },
    {
      "epoch": 1.4424012617026274,
      "grad_norm": 0.703125,
      "learning_rate": 4.0462422010154266e-05,
      "loss": 0.73,
      "step": 2686
    },
    {
      "epoch": 1.4429381564377035,
      "grad_norm": 0.7890625,
      "learning_rate": 4.04557850434675e-05,
      "loss": 0.7924,
      "step": 2687
    },
    {
      "epoch": 1.4434750511727794,
      "grad_norm": 0.7265625,
      "learning_rate": 4.044914631307993e-05,
      "loss": 0.8306,
      "step": 2688
    },
    {
      "epoch": 1.4440119459078553,
      "grad_norm": 0.6953125,
      "learning_rate": 4.044250581974911e-05,
      "loss": 0.872,
      "step": 2689
    },
    {
      "epoch": 1.4445488406429314,
      "grad_norm": 0.67578125,
      "learning_rate": 4.0435863564232826e-05,
      "loss": 0.96,
      "step": 2690
    },
    {
      "epoch": 1.4450857353780076,
      "grad_norm": 0.73828125,
      "learning_rate": 4.042921954728902e-05,
      "loss": 0.9316,
      "step": 2691
    },
    {
      "epoch": 1.4456226301130835,
      "grad_norm": 0.640625,
      "learning_rate": 4.042257376967586e-05,
      "loss": 0.8192,
      "step": 2692
    },
    {
      "epoch": 1.4461595248481594,
      "grad_norm": 0.734375,
      "learning_rate": 4.041592623215174e-05,
      "loss": 0.9176,
      "step": 2693
    },
    {
      "epoch": 1.4466964195832355,
      "grad_norm": 0.70703125,
      "learning_rate": 4.04092769354752e-05,
      "loss": 0.829,
      "step": 2694
    },
    {
      "epoch": 1.4472333143183116,
      "grad_norm": 0.66015625,
      "learning_rate": 4.040262588040503e-05,
      "loss": 0.7773,
      "step": 2695
    },
    {
      "epoch": 1.4477702090533875,
      "grad_norm": 0.69921875,
      "learning_rate": 4.0395973067700176e-05,
      "loss": 0.9758,
      "step": 2696
    },
    {
      "epoch": 1.4483071037884634,
      "grad_norm": 0.703125,
      "learning_rate": 4.038931849811982e-05,
      "loss": 0.8894,
      "step": 2697
    },
    {
      "epoch": 1.4488439985235395,
      "grad_norm": 0.69921875,
      "learning_rate": 4.038266217242334e-05,
      "loss": 0.8639,
      "step": 2698
    },
    {
      "epoch": 1.4493808932586154,
      "grad_norm": 0.6640625,
      "learning_rate": 4.03760040913703e-05,
      "loss": 0.7328,
      "step": 2699
    },
    {
      "epoch": 1.4499177879936915,
      "grad_norm": 0.76171875,
      "learning_rate": 4.0369344255720474e-05,
      "loss": 0.8242,
      "step": 2700
    },
    {
      "epoch": 1.4499177879936915,
      "eval_loss": 0.9133481979370117,
      "eval_runtime": 46.0758,
      "eval_samples_per_second": 20.879,
      "eval_steps_per_second": 20.879,
      "step": 2700
    },
    {
      "epoch": 1.4504546827287674,
      "grad_norm": 0.71875,
      "learning_rate": 4.036268266623382e-05,
      "loss": 0.8669,
      "step": 2701
    },
    {
      "epoch": 1.4509915774638436,
      "grad_norm": 0.8671875,
      "learning_rate": 4.035601932367054e-05,
      "loss": 0.9937,
      "step": 2702
    },
    {
      "epoch": 1.4515284721989195,
      "grad_norm": 0.65234375,
      "learning_rate": 4.034935422879097e-05,
      "loss": 0.8431,
      "step": 2703
    },
    {
      "epoch": 1.4520653669339956,
      "grad_norm": 0.55859375,
      "learning_rate": 4.03426873823557e-05,
      "loss": 0.756,
      "step": 2704
    },
    {
      "epoch": 1.4526022616690715,
      "grad_norm": 0.6953125,
      "learning_rate": 4.033601878512551e-05,
      "loss": 0.6655,
      "step": 2705
    },
    {
      "epoch": 1.4531391564041476,
      "grad_norm": 0.6640625,
      "learning_rate": 4.0329348437861354e-05,
      "loss": 0.8231,
      "step": 2706
    },
    {
      "epoch": 1.4536760511392235,
      "grad_norm": 0.73046875,
      "learning_rate": 4.0322676341324415e-05,
      "loss": 0.9908,
      "step": 2707
    },
    {
      "epoch": 1.4542129458742994,
      "grad_norm": 0.64453125,
      "learning_rate": 4.031600249627606e-05,
      "loss": 0.9434,
      "step": 2708
    },
    {
      "epoch": 1.4547498406093755,
      "grad_norm": 0.62890625,
      "learning_rate": 4.030932690347785e-05,
      "loss": 0.781,
      "step": 2709
    },
    {
      "epoch": 1.4552867353444516,
      "grad_norm": 0.7109375,
      "learning_rate": 4.030264956369157e-05,
      "loss": 0.852,
      "step": 2710
    },
    {
      "epoch": 1.4558236300795275,
      "grad_norm": 0.671875,
      "learning_rate": 4.029597047767919e-05,
      "loss": 0.8335,
      "step": 2711
    },
    {
      "epoch": 1.4563605248146034,
      "grad_norm": 0.66796875,
      "learning_rate": 4.028928964620286e-05,
      "loss": 0.6993,
      "step": 2712
    },
    {
      "epoch": 1.4568974195496796,
      "grad_norm": 0.69921875,
      "learning_rate": 4.028260707002496e-05,
      "loss": 0.9187,
      "step": 2713
    },
    {
      "epoch": 1.4574343142847557,
      "grad_norm": 0.74609375,
      "learning_rate": 4.0275922749908055e-05,
      "loss": 0.9053,
      "step": 2714
    },
    {
      "epoch": 1.4579712090198316,
      "grad_norm": 0.67578125,
      "learning_rate": 4.026923668661491e-05,
      "loss": 0.9959,
      "step": 2715
    },
    {
      "epoch": 1.4585081037549075,
      "grad_norm": 0.53515625,
      "learning_rate": 4.026254888090849e-05,
      "loss": 0.6651,
      "step": 2716
    },
    {
      "epoch": 1.4590449984899836,
      "grad_norm": 0.6640625,
      "learning_rate": 4.025585933355196e-05,
      "loss": 0.9886,
      "step": 2717
    },
    {
      "epoch": 1.4595818932250595,
      "grad_norm": 0.734375,
      "learning_rate": 4.024916804530868e-05,
      "loss": 1.0021,
      "step": 2718
    },
    {
      "epoch": 1.4601187879601356,
      "grad_norm": 0.7109375,
      "learning_rate": 4.024247501694221e-05,
      "loss": 0.902,
      "step": 2719
    },
    {
      "epoch": 1.4606556826952115,
      "grad_norm": 0.734375,
      "learning_rate": 4.0235780249216314e-05,
      "loss": 1.1045,
      "step": 2720
    },
    {
      "epoch": 1.4611925774302876,
      "grad_norm": 0.7109375,
      "learning_rate": 4.022908374289494e-05,
      "loss": 0.9118,
      "step": 2721
    },
    {
      "epoch": 1.4617294721653635,
      "grad_norm": 0.7109375,
      "learning_rate": 4.022238549874226e-05,
      "loss": 0.7801,
      "step": 2722
    },
    {
      "epoch": 1.4622663669004397,
      "grad_norm": 0.7734375,
      "learning_rate": 4.02156855175226e-05,
      "loss": 0.8942,
      "step": 2723
    },
    {
      "epoch": 1.4628032616355156,
      "grad_norm": 0.73828125,
      "learning_rate": 4.020898380000055e-05,
      "loss": 0.8445,
      "step": 2724
    },
    {
      "epoch": 1.4633401563705917,
      "grad_norm": 0.609375,
      "learning_rate": 4.020228034694084e-05,
      "loss": 0.7591,
      "step": 2725
    },
    {
      "epoch": 1.4638770511056676,
      "grad_norm": 0.7109375,
      "learning_rate": 4.0195575159108414e-05,
      "loss": 0.9358,
      "step": 2726
    },
    {
      "epoch": 1.4644139458407435,
      "grad_norm": 0.734375,
      "learning_rate": 4.018886823726842e-05,
      "loss": 1.0574,
      "step": 2727
    },
    {
      "epoch": 1.4649508405758196,
      "grad_norm": 1.0078125,
      "learning_rate": 4.018215958218622e-05,
      "loss": 1.0147,
      "step": 2728
    },
    {
      "epoch": 1.4654877353108957,
      "grad_norm": 0.6171875,
      "learning_rate": 4.0175449194627343e-05,
      "loss": 0.7976,
      "step": 2729
    },
    {
      "epoch": 1.4660246300459716,
      "grad_norm": 0.69921875,
      "learning_rate": 4.0168737075357535e-05,
      "loss": 0.8205,
      "step": 2730
    },
    {
      "epoch": 1.4665615247810475,
      "grad_norm": 0.73828125,
      "learning_rate": 4.016202322514272e-05,
      "loss": 0.8443,
      "step": 2731
    },
    {
      "epoch": 1.4670984195161236,
      "grad_norm": 0.7578125,
      "learning_rate": 4.015530764474905e-05,
      "loss": 0.9848,
      "step": 2732
    },
    {
      "epoch": 1.4676353142511998,
      "grad_norm": 0.69140625,
      "learning_rate": 4.014859033494285e-05,
      "loss": 0.9751,
      "step": 2733
    },
    {
      "epoch": 1.4681722089862757,
      "grad_norm": 0.83984375,
      "learning_rate": 4.014187129649065e-05,
      "loss": 0.8844,
      "step": 2734
    },
    {
      "epoch": 1.4687091037213515,
      "grad_norm": 0.7421875,
      "learning_rate": 4.013515053015918e-05,
      "loss": 0.8448,
      "step": 2735
    },
    {
      "epoch": 1.4692459984564277,
      "grad_norm": 0.79296875,
      "learning_rate": 4.0128428036715365e-05,
      "loss": 1.0072,
      "step": 2736
    },
    {
      "epoch": 1.4697828931915036,
      "grad_norm": 0.58203125,
      "learning_rate": 4.012170381692632e-05,
      "loss": 0.7699,
      "step": 2737
    },
    {
      "epoch": 1.4703197879265797,
      "grad_norm": 0.71484375,
      "learning_rate": 4.011497787155938e-05,
      "loss": 0.9128,
      "step": 2738
    },
    {
      "epoch": 1.4708566826616556,
      "grad_norm": 0.77734375,
      "learning_rate": 4.0108250201382036e-05,
      "loss": 0.8606,
      "step": 2739
    },
    {
      "epoch": 1.4713935773967317,
      "grad_norm": 0.69140625,
      "learning_rate": 4.010152080716202e-05,
      "loss": 0.9566,
      "step": 2740
    },
    {
      "epoch": 1.4719304721318076,
      "grad_norm": 0.72265625,
      "learning_rate": 4.0094789689667234e-05,
      "loss": 0.8603,
      "step": 2741
    },
    {
      "epoch": 1.4724673668668837,
      "grad_norm": 0.6640625,
      "learning_rate": 4.008805684966578e-05,
      "loss": 0.6741,
      "step": 2742
    },
    {
      "epoch": 1.4730042616019596,
      "grad_norm": 0.63671875,
      "learning_rate": 4.008132228792597e-05,
      "loss": 0.8336,
      "step": 2743
    },
    {
      "epoch": 1.4735411563370358,
      "grad_norm": 0.72265625,
      "learning_rate": 4.00745860052163e-05,
      "loss": 1.0235,
      "step": 2744
    },
    {
      "epoch": 1.4740780510721117,
      "grad_norm": 0.65625,
      "learning_rate": 4.0067848002305455e-05,
      "loss": 0.9239,
      "step": 2745
    },
    {
      "epoch": 1.4746149458071875,
      "grad_norm": 0.74609375,
      "learning_rate": 4.0061108279962336e-05,
      "loss": 0.9004,
      "step": 2746
    },
    {
      "epoch": 1.4751518405422637,
      "grad_norm": 0.6484375,
      "learning_rate": 4.005436683895603e-05,
      "loss": 0.7801,
      "step": 2747
    },
    {
      "epoch": 1.4756887352773398,
      "grad_norm": 0.66796875,
      "learning_rate": 4.004762368005583e-05,
      "loss": 0.887,
      "step": 2748
    },
    {
      "epoch": 1.4762256300124157,
      "grad_norm": 0.6171875,
      "learning_rate": 4.004087880403119e-05,
      "loss": 0.6446,
      "step": 2749
    },
    {
      "epoch": 1.4767625247474916,
      "grad_norm": 0.65625,
      "learning_rate": 4.0034132211651806e-05,
      "loss": 0.8491,
      "step": 2750
    },
    {
      "epoch": 1.4767625247474916,
      "eval_loss": 0.9130702018737793,
      "eval_runtime": 45.7937,
      "eval_samples_per_second": 21.007,
      "eval_steps_per_second": 21.007,
      "step": 2750
    },
    {
      "epoch": 1.4772994194825677,
      "grad_norm": 0.7734375,
      "learning_rate": 4.002738390368754e-05,
      "loss": 0.9299,
      "step": 2751
    },
    {
      "epoch": 1.4778363142176438,
      "grad_norm": 0.640625,
      "learning_rate": 4.0020633880908476e-05,
      "loss": 0.8185,
      "step": 2752
    },
    {
      "epoch": 1.4783732089527197,
      "grad_norm": 0.6171875,
      "learning_rate": 4.0013882144084855e-05,
      "loss": 0.7305,
      "step": 2753
    },
    {
      "epoch": 1.4789101036877956,
      "grad_norm": 0.6640625,
      "learning_rate": 4.000712869398715e-05,
      "loss": 0.813,
      "step": 2754
    },
    {
      "epoch": 1.4794469984228718,
      "grad_norm": 0.765625,
      "learning_rate": 4.0000373531386015e-05,
      "loss": 0.8307,
      "step": 2755
    },
    {
      "epoch": 1.4799838931579476,
      "grad_norm": 0.671875,
      "learning_rate": 3.999361665705229e-05,
      "loss": 0.9109,
      "step": 2756
    },
    {
      "epoch": 1.4805207878930238,
      "grad_norm": 0.7109375,
      "learning_rate": 3.9986858071757026e-05,
      "loss": 0.9161,
      "step": 2757
    },
    {
      "epoch": 1.4810576826280997,
      "grad_norm": 0.8125,
      "learning_rate": 3.998009777627146e-05,
      "loss": 0.9108,
      "step": 2758
    },
    {
      "epoch": 1.4815945773631758,
      "grad_norm": 0.63671875,
      "learning_rate": 3.9973335771367035e-05,
      "loss": 0.905,
      "step": 2759
    },
    {
      "epoch": 1.4821314720982517,
      "grad_norm": 0.71875,
      "learning_rate": 3.9966572057815373e-05,
      "loss": 0.8476,
      "step": 2760
    },
    {
      "epoch": 1.4826683668333278,
      "grad_norm": 0.7578125,
      "learning_rate": 3.9959806636388314e-05,
      "loss": 0.9646,
      "step": 2761
    },
    {
      "epoch": 1.4832052615684037,
      "grad_norm": 0.6171875,
      "learning_rate": 3.995303950785786e-05,
      "loss": 0.8415,
      "step": 2762
    },
    {
      "epoch": 1.4837421563034798,
      "grad_norm": 0.609375,
      "learning_rate": 3.994627067299623e-05,
      "loss": 1.0265,
      "step": 2763
    },
    {
      "epoch": 1.4842790510385557,
      "grad_norm": 0.9140625,
      "learning_rate": 3.9939500132575834e-05,
      "loss": 1.058,
      "step": 2764
    },
    {
      "epoch": 1.4848159457736316,
      "grad_norm": 0.69140625,
      "learning_rate": 3.993272788736928e-05,
      "loss": 0.8433,
      "step": 2765
    },
    {
      "epoch": 1.4853528405087078,
      "grad_norm": 0.9140625,
      "learning_rate": 3.992595393814938e-05,
      "loss": 1.1677,
      "step": 2766
    },
    {
      "epoch": 1.4858897352437839,
      "grad_norm": 0.640625,
      "learning_rate": 3.9919178285689104e-05,
      "loss": 0.9373,
      "step": 2767
    },
    {
      "epoch": 1.4864266299788598,
      "grad_norm": 0.671875,
      "learning_rate": 3.9912400930761654e-05,
      "loss": 0.7869,
      "step": 2768
    },
    {
      "epoch": 1.4869635247139357,
      "grad_norm": 0.609375,
      "learning_rate": 3.99056218741404e-05,
      "loss": 0.7845,
      "step": 2769
    },
    {
      "epoch": 1.4875004194490118,
      "grad_norm": 0.77734375,
      "learning_rate": 3.989884111659893e-05,
      "loss": 0.8824,
      "step": 2770
    },
    {
      "epoch": 1.488037314184088,
      "grad_norm": 0.68359375,
      "learning_rate": 3.989205865891101e-05,
      "loss": 0.7785,
      "step": 2771
    },
    {
      "epoch": 1.4885742089191638,
      "grad_norm": 0.75390625,
      "learning_rate": 3.98852745018506e-05,
      "loss": 0.7917,
      "step": 2772
    },
    {
      "epoch": 1.4891111036542397,
      "grad_norm": 0.703125,
      "learning_rate": 3.987848864619186e-05,
      "loss": 1.0772,
      "step": 2773
    },
    {
      "epoch": 1.4896479983893158,
      "grad_norm": 0.64453125,
      "learning_rate": 3.9871701092709155e-05,
      "loss": 0.9172,
      "step": 2774
    },
    {
      "epoch": 1.4901848931243917,
      "grad_norm": 0.69140625,
      "learning_rate": 3.9864911842177024e-05,
      "loss": 0.9747,
      "step": 2775
    },
    {
      "epoch": 1.4907217878594679,
      "grad_norm": 0.6875,
      "learning_rate": 3.985812089537019e-05,
      "loss": 0.8586,
      "step": 2776
    },
    {
      "epoch": 1.4912586825945437,
      "grad_norm": 0.6640625,
      "learning_rate": 3.98513282530636e-05,
      "loss": 0.8061,
      "step": 2777
    },
    {
      "epoch": 1.4917955773296199,
      "grad_norm": 0.75390625,
      "learning_rate": 3.984453391603239e-05,
      "loss": 1.0231,
      "step": 2778
    },
    {
      "epoch": 1.4923324720646958,
      "grad_norm": 0.7265625,
      "learning_rate": 3.983773788505186e-05,
      "loss": 1.0016,
      "step": 2779
    },
    {
      "epoch": 1.492869366799772,
      "grad_norm": 0.72265625,
      "learning_rate": 3.9830940160897535e-05,
      "loss": 0.9346,
      "step": 2780
    },
    {
      "epoch": 1.4934062615348478,
      "grad_norm": 0.76171875,
      "learning_rate": 3.982414074434512e-05,
      "loss": 0.894,
      "step": 2781
    },
    {
      "epoch": 1.493943156269924,
      "grad_norm": 0.59375,
      "learning_rate": 3.9817339636170515e-05,
      "loss": 0.7455,
      "step": 2782
    },
    {
      "epoch": 1.4944800510049998,
      "grad_norm": 0.6796875,
      "learning_rate": 3.981053683714981e-05,
      "loss": 0.7339,
      "step": 2783
    },
    {
      "epoch": 1.4950169457400757,
      "grad_norm": 0.6796875,
      "learning_rate": 3.980373234805928e-05,
      "loss": 0.9382,
      "step": 2784
    },
    {
      "epoch": 1.4955538404751518,
      "grad_norm": 0.890625,
      "learning_rate": 3.979692616967543e-05,
      "loss": 0.7895,
      "step": 2785
    },
    {
      "epoch": 1.496090735210228,
      "grad_norm": 0.65625,
      "learning_rate": 3.9790118302774915e-05,
      "loss": 0.8784,
      "step": 2786
    },
    {
      "epoch": 1.4966276299453039,
      "grad_norm": 0.71875,
      "learning_rate": 3.97833087481346e-05,
      "loss": 0.9706,
      "step": 2787
    },
    {
      "epoch": 1.4971645246803797,
      "grad_norm": 0.75390625,
      "learning_rate": 3.9776497506531536e-05,
      "loss": 0.7745,
      "step": 2788
    },
    {
      "epoch": 1.4977014194154559,
      "grad_norm": 0.6875,
      "learning_rate": 3.976968457874298e-05,
      "loss": 0.8367,
      "step": 2789
    },
    {
      "epoch": 1.498238314150532,
      "grad_norm": 0.71875,
      "learning_rate": 3.976286996554637e-05,
      "loss": 0.9974,
      "step": 2790
    },
    {
      "epoch": 1.498775208885608,
      "grad_norm": 0.734375,
      "learning_rate": 3.975605366771934e-05,
      "loss": 0.974,
      "step": 2791
    },
    {
      "epoch": 1.4993121036206838,
      "grad_norm": 0.8515625,
      "learning_rate": 3.9749235686039724e-05,
      "loss": 0.957,
      "step": 2792
    },
    {
      "epoch": 1.49984899835576,
      "grad_norm": 0.73046875,
      "learning_rate": 3.974241602128552e-05,
      "loss": 0.8976,
      "step": 2793
    },
    {
      "epoch": 1.500385893090836,
      "grad_norm": 0.65234375,
      "learning_rate": 3.9735594674234956e-05,
      "loss": 0.8671,
      "step": 2794
    },
    {
      "epoch": 1.500922787825912,
      "grad_norm": 0.90234375,
      "learning_rate": 3.972877164566643e-05,
      "loss": 0.8571,
      "step": 2795
    },
    {
      "epoch": 1.5014596825609878,
      "grad_norm": 0.73828125,
      "learning_rate": 3.972194693635853e-05,
      "loss": 1.0287,
      "step": 2796
    },
    {
      "epoch": 1.501996577296064,
      "grad_norm": 0.94140625,
      "learning_rate": 3.9715120547090046e-05,
      "loss": 1.0705,
      "step": 2797
    },
    {
      "epoch": 1.5025334720311398,
      "grad_norm": 0.76953125,
      "learning_rate": 3.970829247863996e-05,
      "loss": 0.8678,
      "step": 2798
    },
    {
      "epoch": 1.5030703667662157,
      "grad_norm": 0.671875,
      "learning_rate": 3.970146273178743e-05,
      "loss": 0.9473,
      "step": 2799
    },
    {
      "epoch": 1.5036072615012919,
      "grad_norm": 0.69921875,
      "learning_rate": 3.969463130731183e-05,
      "loss": 0.8286,
      "step": 2800
    },
    {
      "epoch": 1.5036072615012919,
      "eval_loss": 0.9118887186050415,
      "eval_runtime": 46.142,
      "eval_samples_per_second": 20.849,
      "eval_steps_per_second": 20.849,
      "step": 2800
    },
    {
      "epoch": 1.504144156236368,
      "grad_norm": 0.7421875,
      "learning_rate": 3.9687798205992705e-05,
      "loss": 0.7714,
      "step": 2801
    },
    {
      "epoch": 1.504681050971444,
      "grad_norm": 0.8046875,
      "learning_rate": 3.96809634286098e-05,
      "loss": 0.9371,
      "step": 2802
    },
    {
      "epoch": 1.5052179457065198,
      "grad_norm": 0.640625,
      "learning_rate": 3.9674126975943045e-05,
      "loss": 1.0391,
      "step": 2803
    },
    {
      "epoch": 1.505754840441596,
      "grad_norm": 0.7890625,
      "learning_rate": 3.966728884877256e-05,
      "loss": 0.9517,
      "step": 2804
    },
    {
      "epoch": 1.506291735176672,
      "grad_norm": 0.734375,
      "learning_rate": 3.966044904787869e-05,
      "loss": 0.8353,
      "step": 2805
    },
    {
      "epoch": 1.506828629911748,
      "grad_norm": 0.703125,
      "learning_rate": 3.9653607574041914e-05,
      "loss": 0.8159,
      "step": 2806
    },
    {
      "epoch": 1.5073655246468238,
      "grad_norm": 0.7734375,
      "learning_rate": 3.964676442804294e-05,
      "loss": 0.9357,
      "step": 2807
    },
    {
      "epoch": 1.5079024193819,
      "grad_norm": 0.83203125,
      "learning_rate": 3.963991961066266e-05,
      "loss": 0.8781,
      "step": 2808
    },
    {
      "epoch": 1.508439314116976,
      "grad_norm": 0.671875,
      "learning_rate": 3.9633073122682156e-05,
      "loss": 0.7878,
      "step": 2809
    },
    {
      "epoch": 1.508976208852052,
      "grad_norm": 0.6796875,
      "learning_rate": 3.962622496488269e-05,
      "loss": 0.8794,
      "step": 2810
    },
    {
      "epoch": 1.5095131035871279,
      "grad_norm": 0.65625,
      "learning_rate": 3.961937513804573e-05,
      "loss": 0.9357,
      "step": 2811
    },
    {
      "epoch": 1.510049998322204,
      "grad_norm": 0.796875,
      "learning_rate": 3.961252364295294e-05,
      "loss": 0.9121,
      "step": 2812
    },
    {
      "epoch": 1.5105868930572801,
      "grad_norm": 0.67578125,
      "learning_rate": 3.9605670480386134e-05,
      "loss": 0.8773,
      "step": 2813
    },
    {
      "epoch": 1.511123787792356,
      "grad_norm": 0.9921875,
      "learning_rate": 3.9598815651127364e-05,
      "loss": 0.9364,
      "step": 2814
    },
    {
      "epoch": 1.511660682527432,
      "grad_norm": 0.671875,
      "learning_rate": 3.9591959155958855e-05,
      "loss": 0.8412,
      "step": 2815
    },
    {
      "epoch": 1.512197577262508,
      "grad_norm": 0.67578125,
      "learning_rate": 3.958510099566301e-05,
      "loss": 0.9101,
      "step": 2816
    },
    {
      "epoch": 1.512734471997584,
      "grad_norm": 0.63671875,
      "learning_rate": 3.9578241171022436e-05,
      "loss": 0.9992,
      "step": 2817
    },
    {
      "epoch": 1.5132713667326598,
      "grad_norm": 0.69140625,
      "learning_rate": 3.9571379682819925e-05,
      "loss": 0.8525,
      "step": 2818
    },
    {
      "epoch": 1.513808261467736,
      "grad_norm": 0.69921875,
      "learning_rate": 3.956451653183846e-05,
      "loss": 0.8469,
      "step": 2819
    },
    {
      "epoch": 1.514345156202812,
      "grad_norm": 0.66796875,
      "learning_rate": 3.9557651718861224e-05,
      "loss": 0.8208,
      "step": 2820
    },
    {
      "epoch": 1.514882050937888,
      "grad_norm": 0.671875,
      "learning_rate": 3.955078524467156e-05,
      "loss": 0.8166,
      "step": 2821
    },
    {
      "epoch": 1.5154189456729639,
      "grad_norm": 0.76171875,
      "learning_rate": 3.9543917110053025e-05,
      "loss": 0.9855,
      "step": 2822
    },
    {
      "epoch": 1.51595584040804,
      "grad_norm": 0.84765625,
      "learning_rate": 3.9537047315789366e-05,
      "loss": 0.8982,
      "step": 2823
    },
    {
      "epoch": 1.516492735143116,
      "grad_norm": 0.7578125,
      "learning_rate": 3.953017586266452e-05,
      "loss": 0.943,
      "step": 2824
    },
    {
      "epoch": 1.517029629878192,
      "grad_norm": 0.66015625,
      "learning_rate": 3.9523302751462586e-05,
      "loss": 0.8065,
      "step": 2825
    },
    {
      "epoch": 1.517566524613268,
      "grad_norm": 0.63671875,
      "learning_rate": 3.95164279829679e-05,
      "loss": 0.9145,
      "step": 2826
    },
    {
      "epoch": 1.518103419348344,
      "grad_norm": 0.625,
      "learning_rate": 3.950955155796493e-05,
      "loss": 0.7812,
      "step": 2827
    },
    {
      "epoch": 1.5186403140834202,
      "grad_norm": 0.6953125,
      "learning_rate": 3.950267347723838e-05,
      "loss": 0.9267,
      "step": 2828
    },
    {
      "epoch": 1.519177208818496,
      "grad_norm": 0.65234375,
      "learning_rate": 3.949579374157314e-05,
      "loss": 0.6184,
      "step": 2829
    },
    {
      "epoch": 1.519714103553572,
      "grad_norm": 0.75390625,
      "learning_rate": 3.948891235175425e-05,
      "loss": 0.9621,
      "step": 2830
    },
    {
      "epoch": 1.520250998288648,
      "grad_norm": 0.66015625,
      "learning_rate": 3.948202930856697e-05,
      "loss": 0.8378,
      "step": 2831
    },
    {
      "epoch": 1.5207878930237242,
      "grad_norm": 0.765625,
      "learning_rate": 3.947514461279676e-05,
      "loss": 0.8502,
      "step": 2832
    },
    {
      "epoch": 1.5213247877588,
      "grad_norm": 0.5390625,
      "learning_rate": 3.946825826522922e-05,
      "loss": 0.7776,
      "step": 2833
    },
    {
      "epoch": 1.521861682493876,
      "grad_norm": 0.72265625,
      "learning_rate": 3.946137026665019e-05,
      "loss": 0.9325,
      "step": 2834
    },
    {
      "epoch": 1.522398577228952,
      "grad_norm": 0.7421875,
      "learning_rate": 3.945448061784568e-05,
      "loss": 0.7118,
      "step": 2835
    },
    {
      "epoch": 1.522935471964028,
      "grad_norm": 0.64453125,
      "learning_rate": 3.944758931960188e-05,
      "loss": 0.943,
      "step": 2836
    },
    {
      "epoch": 1.523472366699104,
      "grad_norm": 0.69140625,
      "learning_rate": 3.944069637270517e-05,
      "loss": 0.9437,
      "step": 2837
    },
    {
      "epoch": 1.52400926143418,
      "grad_norm": 0.71875,
      "learning_rate": 3.943380177794213e-05,
      "loss": 0.9411,
      "step": 2838
    },
    {
      "epoch": 1.5245461561692562,
      "grad_norm": 0.7890625,
      "learning_rate": 3.942690553609951e-05,
      "loss": 0.9696,
      "step": 2839
    },
    {
      "epoch": 1.525083050904332,
      "grad_norm": 0.73828125,
      "learning_rate": 3.942000764796427e-05,
      "loss": 0.8598,
      "step": 2840
    },
    {
      "epoch": 1.525619945639408,
      "grad_norm": 0.62890625,
      "learning_rate": 3.9413108114323555e-05,
      "loss": 0.75,
      "step": 2841
    },
    {
      "epoch": 1.526156840374484,
      "grad_norm": 0.59765625,
      "learning_rate": 3.940620693596467e-05,
      "loss": 0.7114,
      "step": 2842
    },
    {
      "epoch": 1.5266937351095602,
      "grad_norm": 0.765625,
      "learning_rate": 3.939930411367513e-05,
      "loss": 0.9994,
      "step": 2843
    },
    {
      "epoch": 1.527230629844636,
      "grad_norm": 0.73828125,
      "learning_rate": 3.939239964824263e-05,
      "loss": 0.8982,
      "step": 2844
    },
    {
      "epoch": 1.527767524579712,
      "grad_norm": 0.76953125,
      "learning_rate": 3.938549354045508e-05,
      "loss": 0.8759,
      "step": 2845
    },
    {
      "epoch": 1.528304419314788,
      "grad_norm": 0.7265625,
      "learning_rate": 3.9378585791100545e-05,
      "loss": 1.2196,
      "step": 2846
    },
    {
      "epoch": 1.5288413140498642,
      "grad_norm": 0.7109375,
      "learning_rate": 3.937167640096726e-05,
      "loss": 0.8423,
      "step": 2847
    },
    {
      "epoch": 1.5293782087849401,
      "grad_norm": 0.69140625,
      "learning_rate": 3.936476537084371e-05,
      "loss": 0.821,
      "step": 2848
    },
    {
      "epoch": 1.529915103520016,
      "grad_norm": 0.7734375,
      "learning_rate": 3.935785270151851e-05,
      "loss": 0.7115,
      "step": 2849
    },
    {
      "epoch": 1.5304519982550921,
      "grad_norm": 0.6328125,
      "learning_rate": 3.935093839378049e-05,
      "loss": 0.7373,
      "step": 2850
    },
    {
      "epoch": 1.5304519982550921,
      "eval_loss": 0.9116076231002808,
      "eval_runtime": 46.0198,
      "eval_samples_per_second": 20.904,
      "eval_steps_per_second": 20.904,
      "step": 2850
    },
    {
      "epoch": 1.5309888929901683,
      "grad_norm": 0.77734375,
      "learning_rate": 3.934402244841866e-05,
      "loss": 1.0201,
      "step": 2851
    },
    {
      "epoch": 1.5315257877252442,
      "grad_norm": 0.7890625,
      "learning_rate": 3.933710486622222e-05,
      "loss": 1.0381,
      "step": 2852
    },
    {
      "epoch": 1.53206268246032,
      "grad_norm": 0.77734375,
      "learning_rate": 3.933018564798055e-05,
      "loss": 0.9738,
      "step": 2853
    },
    {
      "epoch": 1.5325995771953962,
      "grad_norm": 0.69921875,
      "learning_rate": 3.932326479448322e-05,
      "loss": 0.8155,
      "step": 2854
    },
    {
      "epoch": 1.533136471930472,
      "grad_norm": 0.75390625,
      "learning_rate": 3.9316342306519977e-05,
      "loss": 0.7815,
      "step": 2855
    },
    {
      "epoch": 1.533673366665548,
      "grad_norm": 0.79296875,
      "learning_rate": 3.9309418184880786e-05,
      "loss": 0.8796,
      "step": 2856
    },
    {
      "epoch": 1.534210261400624,
      "grad_norm": 0.609375,
      "learning_rate": 3.930249243035576e-05,
      "loss": 0.7324,
      "step": 2857
    },
    {
      "epoch": 1.5347471561357002,
      "grad_norm": 0.7265625,
      "learning_rate": 3.9295565043735226e-05,
      "loss": 0.8448,
      "step": 2858
    },
    {
      "epoch": 1.5352840508707761,
      "grad_norm": 0.765625,
      "learning_rate": 3.928863602580968e-05,
      "loss": 0.7994,
      "step": 2859
    },
    {
      "epoch": 1.535820945605852,
      "grad_norm": 0.95703125,
      "learning_rate": 3.928170537736981e-05,
      "loss": 0.9257,
      "step": 2860
    },
    {
      "epoch": 1.5363578403409281,
      "grad_norm": 0.69921875,
      "learning_rate": 3.9274773099206496e-05,
      "loss": 0.8972,
      "step": 2861
    },
    {
      "epoch": 1.5368947350760043,
      "grad_norm": 0.6171875,
      "learning_rate": 3.92678391921108e-05,
      "loss": 0.8927,
      "step": 2862
    },
    {
      "epoch": 1.5374316298110802,
      "grad_norm": 0.75390625,
      "learning_rate": 3.926090365687395e-05,
      "loss": 0.9338,
      "step": 2863
    },
    {
      "epoch": 1.537968524546156,
      "grad_norm": 0.76953125,
      "learning_rate": 3.9253966494287405e-05,
      "loss": 0.9745,
      "step": 2864
    },
    {
      "epoch": 1.5385054192812322,
      "grad_norm": 0.7265625,
      "learning_rate": 3.924702770514277e-05,
      "loss": 1.0386,
      "step": 2865
    },
    {
      "epoch": 1.5390423140163083,
      "grad_norm": 0.8515625,
      "learning_rate": 3.924008729023185e-05,
      "loss": 0.9243,
      "step": 2866
    },
    {
      "epoch": 1.5395792087513842,
      "grad_norm": 0.7578125,
      "learning_rate": 3.923314525034664e-05,
      "loss": 0.7536,
      "step": 2867
    },
    {
      "epoch": 1.54011610348646,
      "grad_norm": 0.6875,
      "learning_rate": 3.92262015862793e-05,
      "loss": 0.8662,
      "step": 2868
    },
    {
      "epoch": 1.5406529982215362,
      "grad_norm": 0.74609375,
      "learning_rate": 3.921925629882219e-05,
      "loss": 0.9451,
      "step": 2869
    },
    {
      "epoch": 1.5411898929566124,
      "grad_norm": 0.671875,
      "learning_rate": 3.921230938876788e-05,
      "loss": 0.9485,
      "step": 2870
    },
    {
      "epoch": 1.5417267876916882,
      "grad_norm": 0.7890625,
      "learning_rate": 3.920536085690908e-05,
      "loss": 1.1434,
      "step": 2871
    },
    {
      "epoch": 1.5422636824267641,
      "grad_norm": 0.65234375,
      "learning_rate": 3.919841070403871e-05,
      "loss": 1.1414,
      "step": 2872
    },
    {
      "epoch": 1.5428005771618403,
      "grad_norm": 0.734375,
      "learning_rate": 3.919145893094987e-05,
      "loss": 0.7403,
      "step": 2873
    },
    {
      "epoch": 1.5433374718969162,
      "grad_norm": 0.75390625,
      "learning_rate": 3.9184505538435844e-05,
      "loss": 0.9405,
      "step": 2874
    },
    {
      "epoch": 1.543874366631992,
      "grad_norm": 0.71875,
      "learning_rate": 3.917755052729011e-05,
      "loss": 0.7924,
      "step": 2875
    },
    {
      "epoch": 1.5444112613670682,
      "grad_norm": 0.671875,
      "learning_rate": 3.9170593898306306e-05,
      "loss": 0.7892,
      "step": 2876
    },
    {
      "epoch": 1.5449481561021443,
      "grad_norm": 0.68359375,
      "learning_rate": 3.916363565227829e-05,
      "loss": 0.7871,
      "step": 2877
    },
    {
      "epoch": 1.5454850508372202,
      "grad_norm": 0.8515625,
      "learning_rate": 3.915667579000007e-05,
      "loss": 1.0402,
      "step": 2878
    },
    {
      "epoch": 1.546021945572296,
      "grad_norm": 0.796875,
      "learning_rate": 3.914971431226587e-05,
      "loss": 1.0749,
      "step": 2879
    },
    {
      "epoch": 1.5465588403073722,
      "grad_norm": 0.69921875,
      "learning_rate": 3.9142751219870075e-05,
      "loss": 0.8598,
      "step": 2880
    },
    {
      "epoch": 1.5470957350424484,
      "grad_norm": 0.7109375,
      "learning_rate": 3.9135786513607264e-05,
      "loss": 0.9311,
      "step": 2881
    },
    {
      "epoch": 1.5476326297775242,
      "grad_norm": 0.75390625,
      "learning_rate": 3.912882019427219e-05,
      "loss": 0.9104,
      "step": 2882
    },
    {
      "epoch": 1.5481695245126001,
      "grad_norm": 0.546875,
      "learning_rate": 3.912185226265981e-05,
      "loss": 0.7468,
      "step": 2883
    },
    {
      "epoch": 1.5487064192476763,
      "grad_norm": 0.6015625,
      "learning_rate": 3.911488271956525e-05,
      "loss": 0.6917,
      "step": 2884
    },
    {
      "epoch": 1.5492433139827524,
      "grad_norm": 0.69140625,
      "learning_rate": 3.910791156578382e-05,
      "loss": 0.9262,
      "step": 2885
    },
    {
      "epoch": 1.5497802087178283,
      "grad_norm": 0.703125,
      "learning_rate": 3.910093880211101e-05,
      "loss": 0.9821,
      "step": 2886
    },
    {
      "epoch": 1.5503171034529042,
      "grad_norm": 0.6640625,
      "learning_rate": 3.909396442934252e-05,
      "loss": 0.8115,
      "step": 2887
    },
    {
      "epoch": 1.5508539981879803,
      "grad_norm": 0.6796875,
      "learning_rate": 3.90869884482742e-05,
      "loss": 1.065,
      "step": 2888
    },
    {
      "epoch": 1.5513908929230564,
      "grad_norm": 0.73828125,
      "learning_rate": 3.9080010859702086e-05,
      "loss": 0.8692,
      "step": 2889
    },
    {
      "epoch": 1.5519277876581323,
      "grad_norm": 0.80859375,
      "learning_rate": 3.9073031664422444e-05,
      "loss": 0.9388,
      "step": 2890
    },
    {
      "epoch": 1.5524646823932082,
      "grad_norm": 0.80078125,
      "learning_rate": 3.906605086323166e-05,
      "loss": 0.712,
      "step": 2891
    },
    {
      "epoch": 1.5530015771282843,
      "grad_norm": 0.73046875,
      "learning_rate": 3.905906845692634e-05,
      "loss": 0.9071,
      "step": 2892
    },
    {
      "epoch": 1.5535384718633602,
      "grad_norm": 0.73046875,
      "learning_rate": 3.905208444630327e-05,
      "loss": 0.8565,
      "step": 2893
    },
    {
      "epoch": 1.5540753665984361,
      "grad_norm": 0.5859375,
      "learning_rate": 3.9045098832159406e-05,
      "loss": 0.8977,
      "step": 2894
    },
    {
      "epoch": 1.5546122613335123,
      "grad_norm": 0.6484375,
      "learning_rate": 3.9038111615291894e-05,
      "loss": 0.9619,
      "step": 2895
    },
    {
      "epoch": 1.5551491560685884,
      "grad_norm": 0.9296875,
      "learning_rate": 3.9031122796498074e-05,
      "loss": 0.8718,
      "step": 2896
    },
    {
      "epoch": 1.5556860508036643,
      "grad_norm": 0.6171875,
      "learning_rate": 3.902413237657545e-05,
      "loss": 0.8214,
      "step": 2897
    },
    {
      "epoch": 1.5562229455387402,
      "grad_norm": 0.625,
      "learning_rate": 3.901714035632172e-05,
      "loss": 0.9797,
      "step": 2898
    },
    {
      "epoch": 1.5567598402738163,
      "grad_norm": 0.6640625,
      "learning_rate": 3.901014673653477e-05,
      "loss": 0.9068,
      "step": 2899
    },
    {
      "epoch": 1.5572967350088924,
      "grad_norm": 0.82421875,
      "learning_rate": 3.900315151801265e-05,
      "loss": 1.1709,
      "step": 2900
    },
    {
      "epoch": 1.5572967350088924,
      "eval_loss": 0.9109301567077637,
      "eval_runtime": 45.6924,
      "eval_samples_per_second": 21.054,
      "eval_steps_per_second": 21.054,
      "step": 2900
    },
    {
      "epoch": 1.5578336297439683,
      "grad_norm": 0.609375,
      "learning_rate": 3.899615470155361e-05,
      "loss": 0.8851,
      "step": 2901
    },
    {
      "epoch": 1.5583705244790442,
      "grad_norm": 0.6328125,
      "learning_rate": 3.898915628795606e-05,
      "loss": 0.8582,
      "step": 2902
    },
    {
      "epoch": 1.5589074192141203,
      "grad_norm": 0.65625,
      "learning_rate": 3.898215627801862e-05,
      "loss": 0.8282,
      "step": 2903
    },
    {
      "epoch": 1.5594443139491965,
      "grad_norm": 0.9140625,
      "learning_rate": 3.8975154672540094e-05,
      "loss": 0.8162,
      "step": 2904
    },
    {
      "epoch": 1.5599812086842724,
      "grad_norm": 0.74609375,
      "learning_rate": 3.8968151472319425e-05,
      "loss": 0.8415,
      "step": 2905
    },
    {
      "epoch": 1.5605181034193483,
      "grad_norm": 0.66015625,
      "learning_rate": 3.896114667815579e-05,
      "loss": 0.9145,
      "step": 2906
    },
    {
      "epoch": 1.5610549981544244,
      "grad_norm": 0.7265625,
      "learning_rate": 3.8954140290848505e-05,
      "loss": 0.9159,
      "step": 2907
    },
    {
      "epoch": 1.5615918928895005,
      "grad_norm": 0.7109375,
      "learning_rate": 3.8947132311197106e-05,
      "loss": 0.9453,
      "step": 2908
    },
    {
      "epoch": 1.5621287876245764,
      "grad_norm": 0.69921875,
      "learning_rate": 3.8940122740001285e-05,
      "loss": 0.9828,
      "step": 2909
    },
    {
      "epoch": 1.5626656823596523,
      "grad_norm": 0.7890625,
      "learning_rate": 3.893311157806091e-05,
      "loss": 0.8762,
      "step": 2910
    },
    {
      "epoch": 1.5632025770947284,
      "grad_norm": 0.67578125,
      "learning_rate": 3.892609882617606e-05,
      "loss": 1.0379,
      "step": 2911
    },
    {
      "epoch": 1.5637394718298043,
      "grad_norm": 0.6640625,
      "learning_rate": 3.8919084485146985e-05,
      "loss": 0.8265,
      "step": 2912
    },
    {
      "epoch": 1.5642763665648802,
      "grad_norm": 0.6328125,
      "learning_rate": 3.89120685557741e-05,
      "loss": 0.9004,
      "step": 2913
    },
    {
      "epoch": 1.5648132612999563,
      "grad_norm": 0.6875,
      "learning_rate": 3.8905051038857996e-05,
      "loss": 0.7486,
      "step": 2914
    },
    {
      "epoch": 1.5653501560350325,
      "grad_norm": 0.765625,
      "learning_rate": 3.889803193519948e-05,
      "loss": 0.9516,
      "step": 2915
    },
    {
      "epoch": 1.5658870507701084,
      "grad_norm": 0.703125,
      "learning_rate": 3.8891011245599515e-05,
      "loss": 0.9366,
      "step": 2916
    },
    {
      "epoch": 1.5664239455051843,
      "grad_norm": 0.78125,
      "learning_rate": 3.888398897085927e-05,
      "loss": 0.9448,
      "step": 2917
    },
    {
      "epoch": 1.5669608402402604,
      "grad_norm": 0.87890625,
      "learning_rate": 3.887696511178003e-05,
      "loss": 1.0717,
      "step": 2918
    },
    {
      "epoch": 1.5674977349753365,
      "grad_norm": 0.7265625,
      "learning_rate": 3.8869939669163345e-05,
      "loss": 0.9569,
      "step": 2919
    },
    {
      "epoch": 1.5680346297104124,
      "grad_norm": 0.671875,
      "learning_rate": 3.8862912643810895e-05,
      "loss": 0.7483,
      "step": 2920
    },
    {
      "epoch": 1.5685715244454883,
      "grad_norm": 0.74609375,
      "learning_rate": 3.885588403652455e-05,
      "loss": 1.0418,
      "step": 2921
    },
    {
      "epoch": 1.5691084191805644,
      "grad_norm": 0.6953125,
      "learning_rate": 3.884885384810637e-05,
      "loss": 0.7563,
      "step": 2922
    },
    {
      "epoch": 1.5696453139156405,
      "grad_norm": 0.9296875,
      "learning_rate": 3.884182207935858e-05,
      "loss": 1.0287,
      "step": 2923
    },
    {
      "epoch": 1.5701822086507164,
      "grad_norm": 0.75390625,
      "learning_rate": 3.883478873108361e-05,
      "loss": 0.752,
      "step": 2924
    },
    {
      "epoch": 1.5707191033857923,
      "grad_norm": 0.81640625,
      "learning_rate": 3.882775380408403e-05,
      "loss": 0.9046,
      "step": 2925
    },
    {
      "epoch": 1.5712559981208685,
      "grad_norm": 0.58203125,
      "learning_rate": 3.882071729916263e-05,
      "loss": 0.78,
      "step": 2926
    },
    {
      "epoch": 1.5717928928559446,
      "grad_norm": 0.8046875,
      "learning_rate": 3.8813679217122354e-05,
      "loss": 0.9863,
      "step": 2927
    },
    {
      "epoch": 1.5723297875910205,
      "grad_norm": 0.9140625,
      "learning_rate": 3.880663955876635e-05,
      "loss": 1.0235,
      "step": 2928
    },
    {
      "epoch": 1.5728666823260964,
      "grad_norm": 0.62890625,
      "learning_rate": 3.8799598324897933e-05,
      "loss": 0.8813,
      "step": 2929
    },
    {
      "epoch": 1.5734035770611725,
      "grad_norm": 0.73046875,
      "learning_rate": 3.8792555516320584e-05,
      "loss": 1.1182,
      "step": 2930
    },
    {
      "epoch": 1.5739404717962484,
      "grad_norm": 0.7265625,
      "learning_rate": 3.8785511133837975e-05,
      "loss": 0.9417,
      "step": 2931
    },
    {
      "epoch": 1.5744773665313243,
      "grad_norm": 0.71875,
      "learning_rate": 3.877846517825397e-05,
      "loss": 0.9112,
      "step": 2932
    },
    {
      "epoch": 1.5750142612664004,
      "grad_norm": 0.796875,
      "learning_rate": 3.87714176503726e-05,
      "loss": 0.8963,
      "step": 2933
    },
    {
      "epoch": 1.5755511560014765,
      "grad_norm": 0.70703125,
      "learning_rate": 3.876436855099807e-05,
      "loss": 0.7229,
      "step": 2934
    },
    {
      "epoch": 1.5760880507365524,
      "grad_norm": 0.62109375,
      "learning_rate": 3.8757317880934786e-05,
      "loss": 0.829,
      "step": 2935
    },
    {
      "epoch": 1.5766249454716283,
      "grad_norm": 0.71875,
      "learning_rate": 3.87502656409873e-05,
      "loss": 0.932,
      "step": 2936
    },
    {
      "epoch": 1.5771618402067045,
      "grad_norm": 0.7578125,
      "learning_rate": 3.874321183196038e-05,
      "loss": 0.9727,
      "step": 2937
    },
    {
      "epoch": 1.5776987349417806,
      "grad_norm": 0.93359375,
      "learning_rate": 3.8736156454658944e-05,
      "loss": 0.9234,
      "step": 2938
    },
    {
      "epoch": 1.5782356296768565,
      "grad_norm": 0.6328125,
      "learning_rate": 3.8729099509888104e-05,
      "loss": 0.9148,
      "step": 2939
    },
    {
      "epoch": 1.5787725244119324,
      "grad_norm": 0.67578125,
      "learning_rate": 3.872204099845313e-05,
      "loss": 0.8338,
      "step": 2940
    },
    {
      "epoch": 1.5793094191470085,
      "grad_norm": 0.68359375,
      "learning_rate": 3.871498092115952e-05,
      "loss": 0.8973,
      "step": 2941
    },
    {
      "epoch": 1.5798463138820846,
      "grad_norm": 0.63671875,
      "learning_rate": 3.8707919278812897e-05,
      "loss": 0.8941,
      "step": 2942
    },
    {
      "epoch": 1.5803832086171605,
      "grad_norm": 0.7109375,
      "learning_rate": 3.870085607221908e-05,
      "loss": 1.052,
      "step": 2943
    },
    {
      "epoch": 1.5809201033522364,
      "grad_norm": 0.671875,
      "learning_rate": 3.869379130218409e-05,
      "loss": 0.9589,
      "step": 2944
    },
    {
      "epoch": 1.5814569980873125,
      "grad_norm": 0.73046875,
      "learning_rate": 3.868672496951409e-05,
      "loss": 0.81,
      "step": 2945
    },
    {
      "epoch": 1.5819938928223887,
      "grad_norm": 0.59375,
      "learning_rate": 3.867965707501544e-05,
      "loss": 0.8769,
      "step": 2946
    },
    {
      "epoch": 1.5825307875574646,
      "grad_norm": 0.65234375,
      "learning_rate": 3.8672587619494674e-05,
      "loss": 0.8472,
      "step": 2947
    },
    {
      "epoch": 1.5830676822925405,
      "grad_norm": 0.734375,
      "learning_rate": 3.866551660375851e-05,
      "loss": 0.9401,
      "step": 2948
    },
    {
      "epoch": 1.5836045770276166,
      "grad_norm": 1.390625,
      "learning_rate": 3.8658444028613846e-05,
      "loss": 0.861,
      "step": 2949
    },
    {
      "epoch": 1.5841414717626925,
      "grad_norm": 0.67578125,
      "learning_rate": 3.865136989486776e-05,
      "loss": 0.918,
      "step": 2950
    },
    {
      "epoch": 1.5841414717626925,
      "eval_loss": 0.9100096821784973,
      "eval_runtime": 46.743,
      "eval_samples_per_second": 20.581,
      "eval_steps_per_second": 20.581,
      "step": 2950
    },
    {
      "epoch": 1.5846783664977684,
      "grad_norm": 0.64453125,
      "learning_rate": 3.864429420332748e-05,
      "loss": 0.8551,
      "step": 2951
    },
    {
      "epoch": 1.5852152612328445,
      "grad_norm": 0.77734375,
      "learning_rate": 3.8637216954800435e-05,
      "loss": 1.0583,
      "step": 2952
    },
    {
      "epoch": 1.5857521559679206,
      "grad_norm": 0.671875,
      "learning_rate": 3.8630138150094244e-05,
      "loss": 0.7534,
      "step": 2953
    },
    {
      "epoch": 1.5862890507029965,
      "grad_norm": 0.82421875,
      "learning_rate": 3.862305779001667e-05,
      "loss": 0.8902,
      "step": 2954
    },
    {
      "epoch": 1.5868259454380724,
      "grad_norm": 0.78515625,
      "learning_rate": 3.861597587537568e-05,
      "loss": 1.058,
      "step": 2955
    },
    {
      "epoch": 1.5873628401731485,
      "grad_norm": 0.74609375,
      "learning_rate": 3.8608892406979416e-05,
      "loss": 0.8657,
      "step": 2956
    },
    {
      "epoch": 1.5878997349082247,
      "grad_norm": 0.7265625,
      "learning_rate": 3.860180738563618e-05,
      "loss": 0.814,
      "step": 2957
    },
    {
      "epoch": 1.5884366296433006,
      "grad_norm": 0.796875,
      "learning_rate": 3.859472081215447e-05,
      "loss": 0.7906,
      "step": 2958
    },
    {
      "epoch": 1.5889735243783765,
      "grad_norm": 0.6953125,
      "learning_rate": 3.8587632687342954e-05,
      "loss": 0.8657,
      "step": 2959
    },
    {
      "epoch": 1.5895104191134526,
      "grad_norm": 0.78125,
      "learning_rate": 3.858054301201047e-05,
      "loss": 0.7898,
      "step": 2960
    },
    {
      "epoch": 1.5900473138485287,
      "grad_norm": 0.6484375,
      "learning_rate": 3.857345178696605e-05,
      "loss": 0.9045,
      "step": 2961
    },
    {
      "epoch": 1.5905842085836046,
      "grad_norm": 0.9296875,
      "learning_rate": 3.856635901301888e-05,
      "loss": 1.0102,
      "step": 2962
    },
    {
      "epoch": 1.5911211033186805,
      "grad_norm": 0.6875,
      "learning_rate": 3.855926469097836e-05,
      "loss": 1.0804,
      "step": 2963
    },
    {
      "epoch": 1.5916579980537566,
      "grad_norm": 0.73046875,
      "learning_rate": 3.855216882165401e-05,
      "loss": 0.9125,
      "step": 2964
    },
    {
      "epoch": 1.5921948927888327,
      "grad_norm": 0.66015625,
      "learning_rate": 3.854507140585557e-05,
      "loss": 0.7551,
      "step": 2965
    },
    {
      "epoch": 1.5927317875239086,
      "grad_norm": 0.6875,
      "learning_rate": 3.8537972444392955e-05,
      "loss": 0.7987,
      "step": 2966
    },
    {
      "epoch": 1.5932686822589845,
      "grad_norm": 0.765625,
      "learning_rate": 3.853087193807624e-05,
      "loss": 0.8164,
      "step": 2967
    },
    {
      "epoch": 1.5938055769940607,
      "grad_norm": 0.71484375,
      "learning_rate": 3.852376988771568e-05,
      "loss": 0.9392,
      "step": 2968
    },
    {
      "epoch": 1.5943424717291366,
      "grad_norm": 0.65625,
      "learning_rate": 3.851666629412172e-05,
      "loss": 0.7026,
      "step": 2969
    },
    {
      "epoch": 1.5948793664642125,
      "grad_norm": 0.703125,
      "learning_rate": 3.850956115810495e-05,
      "loss": 0.8727,
      "step": 2970
    },
    {
      "epoch": 1.5954162611992886,
      "grad_norm": 0.77734375,
      "learning_rate": 3.850245448047617e-05,
      "loss": 1.0059,
      "step": 2971
    },
    {
      "epoch": 1.5959531559343647,
      "grad_norm": 0.73828125,
      "learning_rate": 3.849534626204634e-05,
      "loss": 0.9154,
      "step": 2972
    },
    {
      "epoch": 1.5964900506694406,
      "grad_norm": 0.90625,
      "learning_rate": 3.8488236503626594e-05,
      "loss": 0.8564,
      "step": 2973
    },
    {
      "epoch": 1.5970269454045165,
      "grad_norm": 0.8125,
      "learning_rate": 3.848112520602825e-05,
      "loss": 0.8836,
      "step": 2974
    },
    {
      "epoch": 1.5975638401395926,
      "grad_norm": 0.69140625,
      "learning_rate": 3.84740123700628e-05,
      "loss": 0.6721,
      "step": 2975
    },
    {
      "epoch": 1.5981007348746687,
      "grad_norm": 0.703125,
      "learning_rate": 3.8466897996541894e-05,
      "loss": 0.8799,
      "step": 2976
    },
    {
      "epoch": 1.5986376296097446,
      "grad_norm": 10.875,
      "learning_rate": 3.845978208627739e-05,
      "loss": 1.0333,
      "step": 2977
    },
    {
      "epoch": 1.5991745243448205,
      "grad_norm": 0.67578125,
      "learning_rate": 3.845266464008129e-05,
      "loss": 0.9886,
      "step": 2978
    },
    {
      "epoch": 1.5997114190798967,
      "grad_norm": 0.69921875,
      "learning_rate": 3.844554565876579e-05,
      "loss": 0.7612,
      "step": 2979
    },
    {
      "epoch": 1.6002483138149728,
      "grad_norm": 0.83984375,
      "learning_rate": 3.8438425143143265e-05,
      "loss": 1.0969,
      "step": 2980
    },
    {
      "epoch": 1.6007852085500487,
      "grad_norm": 0.74609375,
      "learning_rate": 3.843130309402624e-05,
      "loss": 0.806,
      "step": 2981
    },
    {
      "epoch": 1.6013221032851246,
      "grad_norm": 0.671875,
      "learning_rate": 3.842417951222744e-05,
      "loss": 0.7616,
      "step": 2982
    },
    {
      "epoch": 1.6018589980202007,
      "grad_norm": 0.67578125,
      "learning_rate": 3.8417054398559755e-05,
      "loss": 0.8587,
      "step": 2983
    },
    {
      "epoch": 1.6023958927552768,
      "grad_norm": 0.734375,
      "learning_rate": 3.840992775383626e-05,
      "loss": 0.9865,
      "step": 2984
    },
    {
      "epoch": 1.6029327874903525,
      "grad_norm": 0.65625,
      "learning_rate": 3.840279957887017e-05,
      "loss": 0.8252,
      "step": 2985
    },
    {
      "epoch": 1.6034696822254286,
      "grad_norm": 0.6875,
      "learning_rate": 3.8395669874474915e-05,
      "loss": 0.8268,
      "step": 2986
    },
    {
      "epoch": 1.6040065769605047,
      "grad_norm": 0.62109375,
      "learning_rate": 3.8388538641464086e-05,
      "loss": 0.8161,
      "step": 2987
    },
    {
      "epoch": 1.6045434716955806,
      "grad_norm": 0.671875,
      "learning_rate": 3.838140588065145e-05,
      "loss": 0.9033,
      "step": 2988
    },
    {
      "epoch": 1.6050803664306565,
      "grad_norm": 0.6484375,
      "learning_rate": 3.8374271592850954e-05,
      "loss": 0.7718,
      "step": 2989
    },
    {
      "epoch": 1.6056172611657327,
      "grad_norm": 0.63671875,
      "learning_rate": 3.836713577887668e-05,
      "loss": 0.6896,
      "step": 2990
    },
    {
      "epoch": 1.6061541559008088,
      "grad_norm": 0.7734375,
      "learning_rate": 3.835999843954294e-05,
      "loss": 0.9686,
      "step": 2991
    },
    {
      "epoch": 1.6066910506358847,
      "grad_norm": 0.58984375,
      "learning_rate": 3.835285957566418e-05,
      "loss": 0.675,
      "step": 2992
    },
    {
      "epoch": 1.6072279453709606,
      "grad_norm": 0.85546875,
      "learning_rate": 3.834571918805505e-05,
      "loss": 1.0189,
      "step": 2993
    },
    {
      "epoch": 1.6077648401060367,
      "grad_norm": 0.734375,
      "learning_rate": 3.833857727753035e-05,
      "loss": 0.7767,
      "step": 2994
    },
    {
      "epoch": 1.6083017348411128,
      "grad_norm": 0.66796875,
      "learning_rate": 3.833143384490506e-05,
      "loss": 0.872,
      "step": 2995
    },
    {
      "epoch": 1.6088386295761887,
      "grad_norm": 0.74609375,
      "learning_rate": 3.832428889099434e-05,
      "loss": 0.7987,
      "step": 2996
    },
    {
      "epoch": 1.6093755243112646,
      "grad_norm": 0.5859375,
      "learning_rate": 3.8317142416613515e-05,
      "loss": 0.7762,
      "step": 2997
    },
    {
      "epoch": 1.6099124190463407,
      "grad_norm": 0.84375,
      "learning_rate": 3.83099944225781e-05,
      "loss": 0.8574,
      "step": 2998
    },
    {
      "epoch": 1.6104493137814169,
      "grad_norm": 0.69140625,
      "learning_rate": 3.830284490970376e-05,
      "loss": 0.8994,
      "step": 2999
    },
    {
      "epoch": 1.6109862085164928,
      "grad_norm": 0.73046875,
      "learning_rate": 3.8295693878806347e-05,
      "loss": 0.8682,
      "step": 3000
    },
    {
      "epoch": 1.6109862085164928,
      "eval_loss": 0.9103614091873169,
      "eval_runtime": 47.1803,
      "eval_samples_per_second": 20.39,
      "eval_steps_per_second": 20.39,
      "step": 3000
    },
    {
      "epoch": 1.6115231032515687,
      "grad_norm": 0.734375,
      "learning_rate": 3.828854133070189e-05,
      "loss": 1.0039,
      "step": 3001
    },
    {
      "epoch": 1.6120599979866448,
      "grad_norm": 0.77734375,
      "learning_rate": 3.828138726620658e-05,
      "loss": 0.8209,
      "step": 3002
    },
    {
      "epoch": 1.612596892721721,
      "grad_norm": 0.7421875,
      "learning_rate": 3.827423168613679e-05,
      "loss": 1.0899,
      "step": 3003
    },
    {
      "epoch": 1.6131337874567966,
      "grad_norm": 0.875,
      "learning_rate": 3.826707459130906e-05,
      "loss": 1.0434,
      "step": 3004
    },
    {
      "epoch": 1.6136706821918727,
      "grad_norm": 0.68359375,
      "learning_rate": 3.82599159825401e-05,
      "loss": 1.0561,
      "step": 3005
    },
    {
      "epoch": 1.6142075769269488,
      "grad_norm": 0.70703125,
      "learning_rate": 3.8252755860646804e-05,
      "loss": 0.8762,
      "step": 3006
    },
    {
      "epoch": 1.6147444716620247,
      "grad_norm": 0.578125,
      "learning_rate": 3.824559422644624e-05,
      "loss": 0.671,
      "step": 3007
    },
    {
      "epoch": 1.6152813663971006,
      "grad_norm": 0.69140625,
      "learning_rate": 3.823843108075562e-05,
      "loss": 0.8227,
      "step": 3008
    },
    {
      "epoch": 1.6158182611321767,
      "grad_norm": 0.62109375,
      "learning_rate": 3.8231266424392375e-05,
      "loss": 0.7786,
      "step": 3009
    },
    {
      "epoch": 1.6163551558672529,
      "grad_norm": 0.734375,
      "learning_rate": 3.822410025817406e-05,
      "loss": 0.8902,
      "step": 3010
    },
    {
      "epoch": 1.6168920506023288,
      "grad_norm": 0.6015625,
      "learning_rate": 3.821693258291844e-05,
      "loss": 0.7087,
      "step": 3011
    },
    {
      "epoch": 1.6174289453374047,
      "grad_norm": 0.8359375,
      "learning_rate": 3.820976339944344e-05,
      "loss": 1.0657,
      "step": 3012
    },
    {
      "epoch": 1.6179658400724808,
      "grad_norm": 0.89453125,
      "learning_rate": 3.820259270856714e-05,
      "loss": 0.85,
      "step": 3013
    },
    {
      "epoch": 1.618502734807557,
      "grad_norm": 0.78515625,
      "learning_rate": 3.819542051110782e-05,
      "loss": 1.1769,
      "step": 3014
    },
    {
      "epoch": 1.6190396295426328,
      "grad_norm": 0.68359375,
      "learning_rate": 3.818824680788391e-05,
      "loss": 0.928,
      "step": 3015
    },
    {
      "epoch": 1.6195765242777087,
      "grad_norm": 0.609375,
      "learning_rate": 3.818107159971402e-05,
      "loss": 0.8648,
      "step": 3016
    },
    {
      "epoch": 1.6201134190127848,
      "grad_norm": 0.640625,
      "learning_rate": 3.8173894887416945e-05,
      "loss": 0.9321,
      "step": 3017
    },
    {
      "epoch": 1.620650313747861,
      "grad_norm": 0.71484375,
      "learning_rate": 3.816671667181162e-05,
      "loss": 0.976,
      "step": 3018
    },
    {
      "epoch": 1.6211872084829368,
      "grad_norm": 0.93359375,
      "learning_rate": 3.8159536953717185e-05,
      "loss": 0.9066,
      "step": 3019
    },
    {
      "epoch": 1.6217241032180127,
      "grad_norm": 0.7578125,
      "learning_rate": 3.8152355733952925e-05,
      "loss": 0.8672,
      "step": 3020
    },
    {
      "epoch": 1.6222609979530889,
      "grad_norm": 0.75390625,
      "learning_rate": 3.814517301333832e-05,
      "loss": 0.8575,
      "step": 3021
    },
    {
      "epoch": 1.622797892688165,
      "grad_norm": 0.5625,
      "learning_rate": 3.8137988792693006e-05,
      "loss": 0.7132,
      "step": 3022
    },
    {
      "epoch": 1.6233347874232407,
      "grad_norm": 0.70703125,
      "learning_rate": 3.8130803072836796e-05,
      "loss": 0.8218,
      "step": 3023
    },
    {
      "epoch": 1.6238716821583168,
      "grad_norm": 0.81640625,
      "learning_rate": 3.8123615854589657e-05,
      "loss": 0.8358,
      "step": 3024
    },
    {
      "epoch": 1.624408576893393,
      "grad_norm": 0.765625,
      "learning_rate": 3.8116427138771746e-05,
      "loss": 0.8898,
      "step": 3025
    },
    {
      "epoch": 1.6249454716284688,
      "grad_norm": 0.609375,
      "learning_rate": 3.810923692620341e-05,
      "loss": 0.8229,
      "step": 3026
    },
    {
      "epoch": 1.6254823663635447,
      "grad_norm": 0.8125,
      "learning_rate": 3.810204521770512e-05,
      "loss": 0.9023,
      "step": 3027
    },
    {
      "epoch": 1.6260192610986208,
      "grad_norm": 0.63671875,
      "learning_rate": 3.8094852014097544e-05,
      "loss": 0.7394,
      "step": 3028
    },
    {
      "epoch": 1.626556155833697,
      "grad_norm": 0.76171875,
      "learning_rate": 3.808765731620152e-05,
      "loss": 0.8538,
      "step": 3029
    },
    {
      "epoch": 1.6270930505687728,
      "grad_norm": 0.76953125,
      "learning_rate": 3.808046112483805e-05,
      "loss": 1.0332,
      "step": 3030
    },
    {
      "epoch": 1.6276299453038487,
      "grad_norm": 0.6875,
      "learning_rate": 3.807326344082832e-05,
      "loss": 0.7159,
      "step": 3031
    },
    {
      "epoch": 1.6281668400389249,
      "grad_norm": 0.6796875,
      "learning_rate": 3.8066064264993674e-05,
      "loss": 0.8199,
      "step": 3032
    },
    {
      "epoch": 1.628703734774001,
      "grad_norm": 0.8515625,
      "learning_rate": 3.805886359815563e-05,
      "loss": 0.8384,
      "step": 3033
    },
    {
      "epoch": 1.6292406295090769,
      "grad_norm": 0.6484375,
      "learning_rate": 3.8051661441135864e-05,
      "loss": 0.8304,
      "step": 3034
    },
    {
      "epoch": 1.6297775242441528,
      "grad_norm": 0.74609375,
      "learning_rate": 3.804445779475624e-05,
      "loss": 0.8786,
      "step": 3035
    },
    {
      "epoch": 1.630314418979229,
      "grad_norm": 0.73828125,
      "learning_rate": 3.803725265983879e-05,
      "loss": 0.8432,
      "step": 3036
    },
    {
      "epoch": 1.630851313714305,
      "grad_norm": 0.75390625,
      "learning_rate": 3.803004603720571e-05,
      "loss": 0.896,
      "step": 3037
    },
    {
      "epoch": 1.631388208449381,
      "grad_norm": 0.68359375,
      "learning_rate": 3.8022837927679373e-05,
      "loss": 0.7521,
      "step": 3038
    },
    {
      "epoch": 1.6319251031844568,
      "grad_norm": 0.6171875,
      "learning_rate": 3.8015628332082295e-05,
      "loss": 0.8765,
      "step": 3039
    },
    {
      "epoch": 1.632461997919533,
      "grad_norm": 1.078125,
      "learning_rate": 3.80084172512372e-05,
      "loss": 1.0539,
      "step": 3040
    },
    {
      "epoch": 1.632998892654609,
      "grad_norm": 0.65625,
      "learning_rate": 3.800120468596696e-05,
      "loss": 0.889,
      "step": 3041
    },
    {
      "epoch": 1.6335357873896847,
      "grad_norm": 0.640625,
      "learning_rate": 3.799399063709461e-05,
      "loss": 0.7775,
      "step": 3042
    },
    {
      "epoch": 1.6340726821247609,
      "grad_norm": 0.62890625,
      "learning_rate": 3.798677510544338e-05,
      "loss": 0.7852,
      "step": 3043
    },
    {
      "epoch": 1.634609576859837,
      "grad_norm": 0.69921875,
      "learning_rate": 3.797955809183665e-05,
      "loss": 1.0743,
      "step": 3044
    },
    {
      "epoch": 1.6351464715949129,
      "grad_norm": 0.78515625,
      "learning_rate": 3.797233959709796e-05,
      "loss": 0.8473,
      "step": 3045
    },
    {
      "epoch": 1.6356833663299888,
      "grad_norm": 1.3515625,
      "learning_rate": 3.796511962205104e-05,
      "loss": 0.8339,
      "step": 3046
    },
    {
      "epoch": 1.636220261065065,
      "grad_norm": 0.671875,
      "learning_rate": 3.795789816751979e-05,
      "loss": 0.7525,
      "step": 3047
    },
    {
      "epoch": 1.636757155800141,
      "grad_norm": 0.62890625,
      "learning_rate": 3.795067523432826e-05,
      "loss": 0.7052,
      "step": 3048
    },
    {
      "epoch": 1.637294050535217,
      "grad_norm": 0.68359375,
      "learning_rate": 3.794345082330067e-05,
      "loss": 0.8761,
      "step": 3049
    },
    {
      "epoch": 1.6378309452702928,
      "grad_norm": 0.62890625,
      "learning_rate": 3.7936224935261434e-05,
      "loss": 0.7289,
      "step": 3050
    },
    {
      "epoch": 1.6378309452702928,
      "eval_loss": 0.9097536206245422,
      "eval_runtime": 46.8458,
      "eval_samples_per_second": 20.535,
      "eval_steps_per_second": 20.535,
      "step": 3050
    },
    {
      "epoch": 1.638367840005369,
      "grad_norm": 0.71875,
      "learning_rate": 3.792899757103511e-05,
      "loss": 1.0245,
      "step": 3051
    },
    {
      "epoch": 1.638904734740445,
      "grad_norm": 0.7109375,
      "learning_rate": 3.792176873144642e-05,
      "loss": 0.8299,
      "step": 3052
    },
    {
      "epoch": 1.639441629475521,
      "grad_norm": 0.80078125,
      "learning_rate": 3.791453841732028e-05,
      "loss": 0.9915,
      "step": 3053
    },
    {
      "epoch": 1.6399785242105969,
      "grad_norm": 0.7578125,
      "learning_rate": 3.790730662948176e-05,
      "loss": 0.9616,
      "step": 3054
    },
    {
      "epoch": 1.640515418945673,
      "grad_norm": 0.67578125,
      "learning_rate": 3.79000733687561e-05,
      "loss": 0.8969,
      "step": 3055
    },
    {
      "epoch": 1.641052313680749,
      "grad_norm": 0.69140625,
      "learning_rate": 3.7892838635968696e-05,
      "loss": 0.8711,
      "step": 3056
    },
    {
      "epoch": 1.641589208415825,
      "grad_norm": 0.72265625,
      "learning_rate": 3.788560243194513e-05,
      "loss": 0.9227,
      "step": 3057
    },
    {
      "epoch": 1.642126103150901,
      "grad_norm": 0.59375,
      "learning_rate": 3.787836475751114e-05,
      "loss": 0.8242,
      "step": 3058
    },
    {
      "epoch": 1.642662997885977,
      "grad_norm": 0.66015625,
      "learning_rate": 3.7871125613492654e-05,
      "loss": 0.8317,
      "step": 3059
    },
    {
      "epoch": 1.6431998926210531,
      "grad_norm": 0.71484375,
      "learning_rate": 3.786388500071572e-05,
      "loss": 0.955,
      "step": 3060
    },
    {
      "epoch": 1.6437367873561288,
      "grad_norm": 0.7109375,
      "learning_rate": 3.785664292000661e-05,
      "loss": 0.8222,
      "step": 3061
    },
    {
      "epoch": 1.644273682091205,
      "grad_norm": 0.7265625,
      "learning_rate": 3.784939937219172e-05,
      "loss": 0.9555,
      "step": 3062
    },
    {
      "epoch": 1.644810576826281,
      "grad_norm": 0.640625,
      "learning_rate": 3.7842154358097635e-05,
      "loss": 0.6415,
      "step": 3063
    },
    {
      "epoch": 1.645347471561357,
      "grad_norm": 0.5703125,
      "learning_rate": 3.7834907878551116e-05,
      "loss": 0.7773,
      "step": 3064
    },
    {
      "epoch": 1.6458843662964329,
      "grad_norm": 0.78125,
      "learning_rate": 3.782765993437906e-05,
      "loss": 0.668,
      "step": 3065
    },
    {
      "epoch": 1.646421261031509,
      "grad_norm": 0.6953125,
      "learning_rate": 3.7820410526408556e-05,
      "loss": 1.02,
      "step": 3066
    },
    {
      "epoch": 1.646958155766585,
      "grad_norm": 0.58203125,
      "learning_rate": 3.781315965546686e-05,
      "loss": 0.6888,
      "step": 3067
    },
    {
      "epoch": 1.647495050501661,
      "grad_norm": 0.76171875,
      "learning_rate": 3.780590732238137e-05,
      "loss": 0.8736,
      "step": 3068
    },
    {
      "epoch": 1.648031945236737,
      "grad_norm": 0.796875,
      "learning_rate": 3.779865352797969e-05,
      "loss": 1.0496,
      "step": 3069
    },
    {
      "epoch": 1.648568839971813,
      "grad_norm": 0.58203125,
      "learning_rate": 3.779139827308956e-05,
      "loss": 0.8068,
      "step": 3070
    },
    {
      "epoch": 1.6491057347068891,
      "grad_norm": 0.83203125,
      "learning_rate": 3.778414155853889e-05,
      "loss": 1.0234,
      "step": 3071
    },
    {
      "epoch": 1.649642629441965,
      "grad_norm": 0.6015625,
      "learning_rate": 3.777688338515578e-05,
      "loss": 0.7916,
      "step": 3072
    },
    {
      "epoch": 1.650179524177041,
      "grad_norm": 0.65234375,
      "learning_rate": 3.7769623753768455e-05,
      "loss": 1.0565,
      "step": 3073
    },
    {
      "epoch": 1.650716418912117,
      "grad_norm": 0.8046875,
      "learning_rate": 3.7762362665205364e-05,
      "loss": 1.0026,
      "step": 3074
    },
    {
      "epoch": 1.6512533136471932,
      "grad_norm": 0.7265625,
      "learning_rate": 3.7755100120295064e-05,
      "loss": 0.9098,
      "step": 3075
    },
    {
      "epoch": 1.651790208382269,
      "grad_norm": 0.73828125,
      "learning_rate": 3.7747836119866305e-05,
      "loss": 0.8866,
      "step": 3076
    },
    {
      "epoch": 1.652327103117345,
      "grad_norm": 0.71875,
      "learning_rate": 3.774057066474801e-05,
      "loss": 0.7804,
      "step": 3077
    },
    {
      "epoch": 1.652863997852421,
      "grad_norm": 0.65234375,
      "learning_rate": 3.7733303755769255e-05,
      "loss": 1.0191,
      "step": 3078
    },
    {
      "epoch": 1.6534008925874972,
      "grad_norm": 0.55859375,
      "learning_rate": 3.7726035393759285e-05,
      "loss": 0.6412,
      "step": 3079
    },
    {
      "epoch": 1.653937787322573,
      "grad_norm": 0.6328125,
      "learning_rate": 3.771876557954752e-05,
      "loss": 0.8218,
      "step": 3080
    },
    {
      "epoch": 1.654474682057649,
      "grad_norm": 0.734375,
      "learning_rate": 3.771149431396353e-05,
      "loss": 0.9572,
      "step": 3081
    },
    {
      "epoch": 1.6550115767927251,
      "grad_norm": 0.6640625,
      "learning_rate": 3.7704221597837055e-05,
      "loss": 0.894,
      "step": 3082
    },
    {
      "epoch": 1.655548471527801,
      "grad_norm": 0.625,
      "learning_rate": 3.7696947431998015e-05,
      "loss": 0.8156,
      "step": 3083
    },
    {
      "epoch": 1.656085366262877,
      "grad_norm": 0.74609375,
      "learning_rate": 3.7689671817276477e-05,
      "loss": 0.9495,
      "step": 3084
    },
    {
      "epoch": 1.656622260997953,
      "grad_norm": 0.71484375,
      "learning_rate": 3.768239475450269e-05,
      "loss": 0.7342,
      "step": 3085
    },
    {
      "epoch": 1.6571591557330292,
      "grad_norm": 0.77734375,
      "learning_rate": 3.767511624450705e-05,
      "loss": 0.7942,
      "step": 3086
    },
    {
      "epoch": 1.657696050468105,
      "grad_norm": 0.703125,
      "learning_rate": 3.766783628812012e-05,
      "loss": 0.7667,
      "step": 3087
    },
    {
      "epoch": 1.658232945203181,
      "grad_norm": 0.93359375,
      "learning_rate": 3.766055488617265e-05,
      "loss": 1.1026,
      "step": 3088
    },
    {
      "epoch": 1.658769839938257,
      "grad_norm": 0.84765625,
      "learning_rate": 3.765327203949554e-05,
      "loss": 0.9175,
      "step": 3089
    },
    {
      "epoch": 1.6593067346733332,
      "grad_norm": 0.69140625,
      "learning_rate": 3.7645987748919836e-05,
      "loss": 0.8211,
      "step": 3090
    },
    {
      "epoch": 1.6598436294084091,
      "grad_norm": 0.6953125,
      "learning_rate": 3.763870201527679e-05,
      "loss": 0.8248,
      "step": 3091
    },
    {
      "epoch": 1.660380524143485,
      "grad_norm": 0.69140625,
      "learning_rate": 3.763141483939779e-05,
      "loss": 0.976,
      "step": 3092
    },
    {
      "epoch": 1.6609174188785611,
      "grad_norm": 0.6171875,
      "learning_rate": 3.762412622211439e-05,
      "loss": 0.869,
      "step": 3093
    },
    {
      "epoch": 1.6614543136136373,
      "grad_norm": 0.63671875,
      "learning_rate": 3.761683616425832e-05,
      "loss": 0.7032,
      "step": 3094
    },
    {
      "epoch": 1.6619912083487132,
      "grad_norm": 0.6875,
      "learning_rate": 3.7609544666661455e-05,
      "loss": 0.8989,
      "step": 3095
    },
    {
      "epoch": 1.662528103083789,
      "grad_norm": 0.55859375,
      "learning_rate": 3.7602251730155855e-05,
      "loss": 0.6087,
      "step": 3096
    },
    {
      "epoch": 1.6630649978188652,
      "grad_norm": 0.6328125,
      "learning_rate": 3.7594957355573756e-05,
      "loss": 0.8414,
      "step": 3097
    },
    {
      "epoch": 1.6636018925539413,
      "grad_norm": 0.703125,
      "learning_rate": 3.758766154374751e-05,
      "loss": 1.0477,
      "step": 3098
    },
    {
      "epoch": 1.664138787289017,
      "grad_norm": 0.6953125,
      "learning_rate": 3.7580364295509675e-05,
      "loss": 0.8542,
      "step": 3099
    },
    {
      "epoch": 1.664675682024093,
      "grad_norm": 0.7109375,
      "learning_rate": 3.7573065611692946e-05,
      "loss": 0.9615,
      "step": 3100
    },
    {
      "epoch": 1.664675682024093,
      "eval_loss": 0.9097515940666199,
      "eval_runtime": 48.3801,
      "eval_samples_per_second": 19.884,
      "eval_steps_per_second": 19.884,
      "step": 3100
    },
    {
      "epoch": 1.6652125767591692,
      "grad_norm": 0.69140625,
      "learning_rate": 3.756576549313022e-05,
      "loss": 0.8375,
      "step": 3101
    },
    {
      "epoch": 1.6657494714942451,
      "grad_norm": 0.671875,
      "learning_rate": 3.755846394065451e-05,
      "loss": 0.6936,
      "step": 3102
    },
    {
      "epoch": 1.666286366229321,
      "grad_norm": 0.7109375,
      "learning_rate": 3.755116095509903e-05,
      "loss": 0.919,
      "step": 3103
    },
    {
      "epoch": 1.6668232609643971,
      "grad_norm": 0.7109375,
      "learning_rate": 3.7543856537297126e-05,
      "loss": 0.9726,
      "step": 3104
    },
    {
      "epoch": 1.6673601556994733,
      "grad_norm": 0.640625,
      "learning_rate": 3.753655068808234e-05,
      "loss": 0.6599,
      "step": 3105
    },
    {
      "epoch": 1.6678970504345492,
      "grad_norm": 0.71484375,
      "learning_rate": 3.752924340828837e-05,
      "loss": 0.9401,
      "step": 3106
    },
    {
      "epoch": 1.668433945169625,
      "grad_norm": 0.796875,
      "learning_rate": 3.7521934698749046e-05,
      "loss": 0.8985,
      "step": 3107
    },
    {
      "epoch": 1.6689708399047012,
      "grad_norm": 0.859375,
      "learning_rate": 3.75146245602984e-05,
      "loss": 0.9108,
      "step": 3108
    },
    {
      "epoch": 1.6695077346397773,
      "grad_norm": 0.703125,
      "learning_rate": 3.7507312993770596e-05,
      "loss": 0.7507,
      "step": 3109
    },
    {
      "epoch": 1.6700446293748532,
      "grad_norm": 0.69140625,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 0.8606,
      "step": 3110
    },
    {
      "epoch": 1.670581524109929,
      "grad_norm": 0.6640625,
      "learning_rate": 3.749268557982111e-05,
      "loss": 0.6185,
      "step": 3111
    },
    {
      "epoch": 1.6711184188450052,
      "grad_norm": 0.57421875,
      "learning_rate": 3.748536973406857e-05,
      "loss": 0.6909,
      "step": 3112
    },
    {
      "epoch": 1.6716553135800813,
      "grad_norm": 0.62890625,
      "learning_rate": 3.747805246357724e-05,
      "loss": 0.85,
      "step": 3113
    },
    {
      "epoch": 1.6721922083151572,
      "grad_norm": 0.71875,
      "learning_rate": 3.74707337691821e-05,
      "loss": 0.991,
      "step": 3114
    },
    {
      "epoch": 1.6727291030502331,
      "grad_norm": 0.6640625,
      "learning_rate": 3.7463413651718303e-05,
      "loss": 0.8143,
      "step": 3115
    },
    {
      "epoch": 1.6732659977853093,
      "grad_norm": 0.79296875,
      "learning_rate": 3.745609211202118e-05,
      "loss": 0.919,
      "step": 3116
    },
    {
      "epoch": 1.6738028925203854,
      "grad_norm": 0.65234375,
      "learning_rate": 3.74487691509262e-05,
      "loss": 0.8302,
      "step": 3117
    },
    {
      "epoch": 1.674339787255461,
      "grad_norm": 0.7109375,
      "learning_rate": 3.7441444769269005e-05,
      "loss": 0.9914,
      "step": 3118
    },
    {
      "epoch": 1.6748766819905372,
      "grad_norm": 0.71875,
      "learning_rate": 3.743411896788541e-05,
      "loss": 0.9043,
      "step": 3119
    },
    {
      "epoch": 1.6754135767256133,
      "grad_norm": 0.73828125,
      "learning_rate": 3.742679174761137e-05,
      "loss": 0.9318,
      "step": 3120
    },
    {
      "epoch": 1.6759504714606892,
      "grad_norm": 0.75,
      "learning_rate": 3.741946310928303e-05,
      "loss": 0.7836,
      "step": 3121
    },
    {
      "epoch": 1.676487366195765,
      "grad_norm": 0.796875,
      "learning_rate": 3.741213305373666e-05,
      "loss": 0.9275,
      "step": 3122
    },
    {
      "epoch": 1.6770242609308412,
      "grad_norm": 0.75390625,
      "learning_rate": 3.740480158180872e-05,
      "loss": 0.8911,
      "step": 3123
    },
    {
      "epoch": 1.6775611556659173,
      "grad_norm": 0.6328125,
      "learning_rate": 3.739746869433583e-05,
      "loss": 0.807,
      "step": 3124
    },
    {
      "epoch": 1.6780980504009932,
      "grad_norm": 0.63671875,
      "learning_rate": 3.739013439215476e-05,
      "loss": 0.7685,
      "step": 3125
    },
    {
      "epoch": 1.6786349451360691,
      "grad_norm": 0.671875,
      "learning_rate": 3.738279867610244e-05,
      "loss": 1.0855,
      "step": 3126
    },
    {
      "epoch": 1.6791718398711453,
      "grad_norm": 0.76953125,
      "learning_rate": 3.7375461547015986e-05,
      "loss": 0.813,
      "step": 3127
    },
    {
      "epoch": 1.6797087346062214,
      "grad_norm": 0.80078125,
      "learning_rate": 3.736812300573264e-05,
      "loss": 1.0443,
      "step": 3128
    },
    {
      "epoch": 1.6802456293412973,
      "grad_norm": 0.69140625,
      "learning_rate": 3.736078305308983e-05,
      "loss": 0.7276,
      "step": 3129
    },
    {
      "epoch": 1.6807825240763732,
      "grad_norm": 0.7890625,
      "learning_rate": 3.735344168992515e-05,
      "loss": 0.8503,
      "step": 3130
    },
    {
      "epoch": 1.6813194188114493,
      "grad_norm": 0.82421875,
      "learning_rate": 3.734609891707632e-05,
      "loss": 0.7957,
      "step": 3131
    },
    {
      "epoch": 1.6818563135465254,
      "grad_norm": 0.7421875,
      "learning_rate": 3.733875473538125e-05,
      "loss": 0.8538,
      "step": 3132
    },
    {
      "epoch": 1.6823932082816013,
      "grad_norm": 0.640625,
      "learning_rate": 3.733140914567801e-05,
      "loss": 0.7315,
      "step": 3133
    },
    {
      "epoch": 1.6829301030166772,
      "grad_norm": 0.6953125,
      "learning_rate": 3.732406214880481e-05,
      "loss": 0.7941,
      "step": 3134
    },
    {
      "epoch": 1.6834669977517533,
      "grad_norm": 0.609375,
      "learning_rate": 3.731671374560007e-05,
      "loss": 0.7481,
      "step": 3135
    },
    {
      "epoch": 1.6840038924868292,
      "grad_norm": 0.69140625,
      "learning_rate": 3.7309363936902294e-05,
      "loss": 0.879,
      "step": 3136
    },
    {
      "epoch": 1.6845407872219051,
      "grad_norm": 0.8359375,
      "learning_rate": 3.730201272355022e-05,
      "loss": 1.0314,
      "step": 3137
    },
    {
      "epoch": 1.6850776819569813,
      "grad_norm": 0.7890625,
      "learning_rate": 3.729466010638269e-05,
      "loss": 0.9467,
      "step": 3138
    },
    {
      "epoch": 1.6856145766920574,
      "grad_norm": 1.0859375,
      "learning_rate": 3.728730608623875e-05,
      "loss": 0.8344,
      "step": 3139
    },
    {
      "epoch": 1.6861514714271333,
      "grad_norm": 0.625,
      "learning_rate": 3.727995066395758e-05,
      "loss": 0.8438,
      "step": 3140
    },
    {
      "epoch": 1.6866883661622092,
      "grad_norm": 0.71484375,
      "learning_rate": 3.727259384037852e-05,
      "loss": 0.8444,
      "step": 3141
    },
    {
      "epoch": 1.6872252608972853,
      "grad_norm": 0.7578125,
      "learning_rate": 3.726523561634109e-05,
      "loss": 0.9086,
      "step": 3142
    },
    {
      "epoch": 1.6877621556323614,
      "grad_norm": 0.6875,
      "learning_rate": 3.725787599268495e-05,
      "loss": 0.8877,
      "step": 3143
    },
    {
      "epoch": 1.6882990503674373,
      "grad_norm": 0.70703125,
      "learning_rate": 3.7250514970249926e-05,
      "loss": 1.1066,
      "step": 3144
    },
    {
      "epoch": 1.6888359451025132,
      "grad_norm": 0.62109375,
      "learning_rate": 3.724315254987599e-05,
      "loss": 0.9929,
      "step": 3145
    },
    {
      "epoch": 1.6893728398375893,
      "grad_norm": 0.63671875,
      "learning_rate": 3.7235788732403315e-05,
      "loss": 0.7913,
      "step": 3146
    },
    {
      "epoch": 1.6899097345726655,
      "grad_norm": 0.63671875,
      "learning_rate": 3.722842351867219e-05,
      "loss": 0.8937,
      "step": 3147
    },
    {
      "epoch": 1.6904466293077414,
      "grad_norm": 0.65234375,
      "learning_rate": 3.7221056909523075e-05,
      "loss": 0.8757,
      "step": 3148
    },
    {
      "epoch": 1.6909835240428173,
      "grad_norm": 0.76953125,
      "learning_rate": 3.721368890579661e-05,
      "loss": 0.8272,
      "step": 3149
    },
    {
      "epoch": 1.6915204187778934,
      "grad_norm": 0.6875,
      "learning_rate": 3.720631950833355e-05,
      "loss": 0.9054,
      "step": 3150
    },
    {
      "epoch": 1.6915204187778934,
      "eval_loss": 0.9101051688194275,
      "eval_runtime": 46.1851,
      "eval_samples_per_second": 20.829,
      "eval_steps_per_second": 20.829,
      "step": 3150
    },
    {
      "epoch": 1.6920573135129695,
      "grad_norm": 0.6484375,
      "learning_rate": 3.7198948717974866e-05,
      "loss": 0.8198,
      "step": 3151
    },
    {
      "epoch": 1.6925942082480454,
      "grad_norm": 0.703125,
      "learning_rate": 3.719157653556166e-05,
      "loss": 0.7609,
      "step": 3152
    },
    {
      "epoch": 1.6931311029831213,
      "grad_norm": 0.6328125,
      "learning_rate": 3.7184202961935156e-05,
      "loss": 0.8904,
      "step": 3153
    },
    {
      "epoch": 1.6936679977181974,
      "grad_norm": 0.77734375,
      "learning_rate": 3.71768279979368e-05,
      "loss": 1.0937,
      "step": 3154
    },
    {
      "epoch": 1.6942048924532733,
      "grad_norm": 0.6796875,
      "learning_rate": 3.716945164440816e-05,
      "loss": 0.8533,
      "step": 3155
    },
    {
      "epoch": 1.6947417871883492,
      "grad_norm": 0.65234375,
      "learning_rate": 3.7162073902190974e-05,
      "loss": 0.7931,
      "step": 3156
    },
    {
      "epoch": 1.6952786819234253,
      "grad_norm": 0.6796875,
      "learning_rate": 3.715469477212714e-05,
      "loss": 0.8886,
      "step": 3157
    },
    {
      "epoch": 1.6958155766585015,
      "grad_norm": 0.734375,
      "learning_rate": 3.7147314255058704e-05,
      "loss": 1.003,
      "step": 3158
    },
    {
      "epoch": 1.6963524713935774,
      "grad_norm": 0.7734375,
      "learning_rate": 3.7139932351827875e-05,
      "loss": 0.6984,
      "step": 3159
    },
    {
      "epoch": 1.6968893661286533,
      "grad_norm": 0.79296875,
      "learning_rate": 3.713254906327703e-05,
      "loss": 0.8247,
      "step": 3160
    },
    {
      "epoch": 1.6974262608637294,
      "grad_norm": 0.6875,
      "learning_rate": 3.712516439024868e-05,
      "loss": 0.9855,
      "step": 3161
    },
    {
      "epoch": 1.6979631555988055,
      "grad_norm": 0.8203125,
      "learning_rate": 3.7117778333585534e-05,
      "loss": 1.003,
      "step": 3162
    },
    {
      "epoch": 1.6985000503338814,
      "grad_norm": 0.796875,
      "learning_rate": 3.7110390894130414e-05,
      "loss": 0.9409,
      "step": 3163
    },
    {
      "epoch": 1.6990369450689573,
      "grad_norm": 0.703125,
      "learning_rate": 3.710300207272633e-05,
      "loss": 0.8235,
      "step": 3164
    },
    {
      "epoch": 1.6995738398040334,
      "grad_norm": 0.70703125,
      "learning_rate": 3.709561187021644e-05,
      "loss": 1.2149,
      "step": 3165
    },
    {
      "epoch": 1.7001107345391095,
      "grad_norm": 0.875,
      "learning_rate": 3.708822028744405e-05,
      "loss": 1.0149,
      "step": 3166
    },
    {
      "epoch": 1.7006476292741854,
      "grad_norm": 0.75,
      "learning_rate": 3.7080827325252635e-05,
      "loss": 0.8986,
      "step": 3167
    },
    {
      "epoch": 1.7011845240092613,
      "grad_norm": 0.6875,
      "learning_rate": 3.7073432984485846e-05,
      "loss": 0.9211,
      "step": 3168
    },
    {
      "epoch": 1.7017214187443375,
      "grad_norm": 0.57421875,
      "learning_rate": 3.706603726598745e-05,
      "loss": 0.7196,
      "step": 3169
    },
    {
      "epoch": 1.7022583134794136,
      "grad_norm": 0.83203125,
      "learning_rate": 3.7058640170601394e-05,
      "loss": 0.8942,
      "step": 3170
    },
    {
      "epoch": 1.7027952082144895,
      "grad_norm": 0.73828125,
      "learning_rate": 3.705124169917179e-05,
      "loss": 0.8141,
      "step": 3171
    },
    {
      "epoch": 1.7033321029495654,
      "grad_norm": 0.6796875,
      "learning_rate": 3.704384185254288e-05,
      "loss": 0.9116,
      "step": 3172
    },
    {
      "epoch": 1.7038689976846415,
      "grad_norm": 0.76953125,
      "learning_rate": 3.703644063155911e-05,
      "loss": 0.9862,
      "step": 3173
    },
    {
      "epoch": 1.7044058924197174,
      "grad_norm": 0.59765625,
      "learning_rate": 3.702903803706502e-05,
      "loss": 1.047,
      "step": 3174
    },
    {
      "epoch": 1.7049427871547933,
      "grad_norm": 0.75390625,
      "learning_rate": 3.7021634069905356e-05,
      "loss": 0.8986,
      "step": 3175
    },
    {
      "epoch": 1.7054796818898694,
      "grad_norm": 0.671875,
      "learning_rate": 3.7014228730925003e-05,
      "loss": 0.8705,
      "step": 3176
    },
    {
      "epoch": 1.7060165766249455,
      "grad_norm": 0.7109375,
      "learning_rate": 3.700682202096901e-05,
      "loss": 0.9543,
      "step": 3177
    },
    {
      "epoch": 1.7065534713600214,
      "grad_norm": 0.625,
      "learning_rate": 3.699941394088257e-05,
      "loss": 0.9731,
      "step": 3178
    },
    {
      "epoch": 1.7070903660950973,
      "grad_norm": 0.82421875,
      "learning_rate": 3.699200449151103e-05,
      "loss": 1.0328,
      "step": 3179
    },
    {
      "epoch": 1.7076272608301735,
      "grad_norm": 0.86328125,
      "learning_rate": 3.6984593673699926e-05,
      "loss": 1.0875,
      "step": 3180
    },
    {
      "epoch": 1.7081641555652496,
      "grad_norm": 0.7734375,
      "learning_rate": 3.69771814882949e-05,
      "loss": 1.0672,
      "step": 3181
    },
    {
      "epoch": 1.7087010503003255,
      "grad_norm": 0.5859375,
      "learning_rate": 3.69697679361418e-05,
      "loss": 0.9401,
      "step": 3182
    },
    {
      "epoch": 1.7092379450354014,
      "grad_norm": 0.8984375,
      "learning_rate": 3.696235301808657e-05,
      "loss": 0.9693,
      "step": 3183
    },
    {
      "epoch": 1.7097748397704775,
      "grad_norm": 0.66796875,
      "learning_rate": 3.695493673497539e-05,
      "loss": 0.7992,
      "step": 3184
    },
    {
      "epoch": 1.7103117345055536,
      "grad_norm": 0.7109375,
      "learning_rate": 3.6947519087654524e-05,
      "loss": 0.9487,
      "step": 3185
    },
    {
      "epoch": 1.7108486292406295,
      "grad_norm": 0.65234375,
      "learning_rate": 3.694010007697043e-05,
      "loss": 0.7009,
      "step": 3186
    },
    {
      "epoch": 1.7113855239757054,
      "grad_norm": 0.71875,
      "learning_rate": 3.693267970376971e-05,
      "loss": 0.8556,
      "step": 3187
    },
    {
      "epoch": 1.7119224187107815,
      "grad_norm": 0.6015625,
      "learning_rate": 3.6925257968899114e-05,
      "loss": 0.7355,
      "step": 3188
    },
    {
      "epoch": 1.7124593134458577,
      "grad_norm": 0.75390625,
      "learning_rate": 3.691783487320556e-05,
      "loss": 0.838,
      "step": 3189
    },
    {
      "epoch": 1.7129962081809336,
      "grad_norm": 0.76171875,
      "learning_rate": 3.691041041753613e-05,
      "loss": 1.0657,
      "step": 3190
    },
    {
      "epoch": 1.7135331029160095,
      "grad_norm": 0.73828125,
      "learning_rate": 3.6902984602738025e-05,
      "loss": 0.8307,
      "step": 3191
    },
    {
      "epoch": 1.7140699976510856,
      "grad_norm": 0.6484375,
      "learning_rate": 3.689555742965865e-05,
      "loss": 0.8304,
      "step": 3192
    },
    {
      "epoch": 1.7146068923861615,
      "grad_norm": 0.71484375,
      "learning_rate": 3.6888128899145524e-05,
      "loss": 0.8784,
      "step": 3193
    },
    {
      "epoch": 1.7151437871212374,
      "grad_norm": 0.640625,
      "learning_rate": 3.688069901204634e-05,
      "loss": 0.738,
      "step": 3194
    },
    {
      "epoch": 1.7156806818563135,
      "grad_norm": 0.6953125,
      "learning_rate": 3.6873267769208944e-05,
      "loss": 0.8378,
      "step": 3195
    },
    {
      "epoch": 1.7162175765913896,
      "grad_norm": 0.73046875,
      "learning_rate": 3.686583517148133e-05,
      "loss": 0.8341,
      "step": 3196
    },
    {
      "epoch": 1.7167544713264655,
      "grad_norm": 0.71484375,
      "learning_rate": 3.685840121971165e-05,
      "loss": 0.8852,
      "step": 3197
    },
    {
      "epoch": 1.7172913660615414,
      "grad_norm": 0.76953125,
      "learning_rate": 3.6850965914748225e-05,
      "loss": 0.8795,
      "step": 3198
    },
    {
      "epoch": 1.7178282607966175,
      "grad_norm": 0.6015625,
      "learning_rate": 3.68435292574395e-05,
      "loss": 0.8056,
      "step": 3199
    },
    {
      "epoch": 1.7183651555316937,
      "grad_norm": 0.71875,
      "learning_rate": 3.68360912486341e-05,
      "loss": 0.9033,
      "step": 3200
    },
    {
      "epoch": 1.7183651555316937,
      "eval_loss": 0.9093900918960571,
      "eval_runtime": 46.5819,
      "eval_samples_per_second": 20.652,
      "eval_steps_per_second": 20.652,
      "step": 3200
    },
    {
      "epoch": 1.7189020502667696,
      "grad_norm": 0.6640625,
      "learning_rate": 3.6828651889180807e-05,
      "loss": 0.9704,
      "step": 3201
    },
    {
      "epoch": 1.7194389450018455,
      "grad_norm": 1.1171875,
      "learning_rate": 3.682121117992853e-05,
      "loss": 1.1052,
      "step": 3202
    },
    {
      "epoch": 1.7199758397369216,
      "grad_norm": 0.73046875,
      "learning_rate": 3.681376912172636e-05,
      "loss": 0.8855,
      "step": 3203
    },
    {
      "epoch": 1.7205127344719977,
      "grad_norm": 0.66015625,
      "learning_rate": 3.680632571542351e-05,
      "loss": 0.9028,
      "step": 3204
    },
    {
      "epoch": 1.7210496292070736,
      "grad_norm": 0.86328125,
      "learning_rate": 3.679888096186939e-05,
      "loss": 0.9277,
      "step": 3205
    },
    {
      "epoch": 1.7215865239421495,
      "grad_norm": 0.66796875,
      "learning_rate": 3.6791434861913535e-05,
      "loss": 0.8151,
      "step": 3206
    },
    {
      "epoch": 1.7221234186772256,
      "grad_norm": 0.6328125,
      "learning_rate": 3.678398741640563e-05,
      "loss": 0.923,
      "step": 3207
    },
    {
      "epoch": 1.7226603134123017,
      "grad_norm": 0.80859375,
      "learning_rate": 3.677653862619552e-05,
      "loss": 1.0484,
      "step": 3208
    },
    {
      "epoch": 1.7231972081473776,
      "grad_norm": 0.55859375,
      "learning_rate": 3.676908849213323e-05,
      "loss": 0.6934,
      "step": 3209
    },
    {
      "epoch": 1.7237341028824535,
      "grad_norm": 1.078125,
      "learning_rate": 3.67616370150689e-05,
      "loss": 0.7035,
      "step": 3210
    },
    {
      "epoch": 1.7242709976175297,
      "grad_norm": 0.71484375,
      "learning_rate": 3.6754184195852826e-05,
      "loss": 0.9222,
      "step": 3211
    },
    {
      "epoch": 1.7248078923526056,
      "grad_norm": 0.65234375,
      "learning_rate": 3.6746730035335484e-05,
      "loss": 0.8372,
      "step": 3212
    },
    {
      "epoch": 1.7253447870876815,
      "grad_norm": 0.89453125,
      "learning_rate": 3.6739274534367484e-05,
      "loss": 0.8406,
      "step": 3213
    },
    {
      "epoch": 1.7258816818227576,
      "grad_norm": 0.671875,
      "learning_rate": 3.67318176937996e-05,
      "loss": 0.8601,
      "step": 3214
    },
    {
      "epoch": 1.7264185765578337,
      "grad_norm": 0.70703125,
      "learning_rate": 3.672435951448274e-05,
      "loss": 0.9916,
      "step": 3215
    },
    {
      "epoch": 1.7269554712929096,
      "grad_norm": 0.6484375,
      "learning_rate": 3.671689999726798e-05,
      "loss": 0.7553,
      "step": 3216
    },
    {
      "epoch": 1.7274923660279855,
      "grad_norm": 0.73046875,
      "learning_rate": 3.6709439143006556e-05,
      "loss": 0.9272,
      "step": 3217
    },
    {
      "epoch": 1.7280292607630616,
      "grad_norm": 0.8046875,
      "learning_rate": 3.670197695254984e-05,
      "loss": 0.9901,
      "step": 3218
    },
    {
      "epoch": 1.7285661554981377,
      "grad_norm": 0.71484375,
      "learning_rate": 3.669451342674935e-05,
      "loss": 0.8426,
      "step": 3219
    },
    {
      "epoch": 1.7291030502332136,
      "grad_norm": 0.7734375,
      "learning_rate": 3.668704856645678e-05,
      "loss": 1.0482,
      "step": 3220
    },
    {
      "epoch": 1.7296399449682895,
      "grad_norm": 0.70703125,
      "learning_rate": 3.667958237252397e-05,
      "loss": 0.787,
      "step": 3221
    },
    {
      "epoch": 1.7301768397033657,
      "grad_norm": 0.79296875,
      "learning_rate": 3.6672114845802895e-05,
      "loss": 0.9153,
      "step": 3222
    },
    {
      "epoch": 1.7307137344384418,
      "grad_norm": 0.64453125,
      "learning_rate": 3.666464598714571e-05,
      "loss": 0.7852,
      "step": 3223
    },
    {
      "epoch": 1.7312506291735177,
      "grad_norm": 0.76171875,
      "learning_rate": 3.665717579740469e-05,
      "loss": 0.8749,
      "step": 3224
    },
    {
      "epoch": 1.7317875239085936,
      "grad_norm": 0.63671875,
      "learning_rate": 3.6649704277432286e-05,
      "loss": 0.805,
      "step": 3225
    },
    {
      "epoch": 1.7323244186436697,
      "grad_norm": 0.6484375,
      "learning_rate": 3.66422314280811e-05,
      "loss": 0.8683,
      "step": 3226
    },
    {
      "epoch": 1.7328613133787458,
      "grad_norm": 0.69140625,
      "learning_rate": 3.663475725020386e-05,
      "loss": 0.9145,
      "step": 3227
    },
    {
      "epoch": 1.7333982081138217,
      "grad_norm": 0.6171875,
      "learning_rate": 3.662728174465348e-05,
      "loss": 0.851,
      "step": 3228
    },
    {
      "epoch": 1.7339351028488976,
      "grad_norm": 0.66796875,
      "learning_rate": 3.661980491228301e-05,
      "loss": 0.8579,
      "step": 3229
    },
    {
      "epoch": 1.7344719975839737,
      "grad_norm": 0.61328125,
      "learning_rate": 3.661232675394563e-05,
      "loss": 0.8754,
      "step": 3230
    },
    {
      "epoch": 1.7350088923190496,
      "grad_norm": 0.6484375,
      "learning_rate": 3.660484727049472e-05,
      "loss": 0.8597,
      "step": 3231
    },
    {
      "epoch": 1.7355457870541255,
      "grad_norm": 0.80859375,
      "learning_rate": 3.659736646278378e-05,
      "loss": 0.8489,
      "step": 3232
    },
    {
      "epoch": 1.7360826817892017,
      "grad_norm": 0.82421875,
      "learning_rate": 3.658988433166645e-05,
      "loss": 0.888,
      "step": 3233
    },
    {
      "epoch": 1.7366195765242778,
      "grad_norm": 0.66015625,
      "learning_rate": 3.6582400877996546e-05,
      "loss": 0.8524,
      "step": 3234
    },
    {
      "epoch": 1.7371564712593537,
      "grad_norm": 0.88671875,
      "learning_rate": 3.657491610262802e-05,
      "loss": 0.9846,
      "step": 3235
    },
    {
      "epoch": 1.7376933659944296,
      "grad_norm": 0.6640625,
      "learning_rate": 3.656743000641499e-05,
      "loss": 0.9523,
      "step": 3236
    },
    {
      "epoch": 1.7382302607295057,
      "grad_norm": 0.69140625,
      "learning_rate": 3.65599425902117e-05,
      "loss": 0.916,
      "step": 3237
    },
    {
      "epoch": 1.7387671554645818,
      "grad_norm": 0.63671875,
      "learning_rate": 3.655245385487256e-05,
      "loss": 0.8783,
      "step": 3238
    },
    {
      "epoch": 1.7393040501996577,
      "grad_norm": 0.87109375,
      "learning_rate": 3.654496380125214e-05,
      "loss": 0.8304,
      "step": 3239
    },
    {
      "epoch": 1.7398409449347336,
      "grad_norm": 0.625,
      "learning_rate": 3.653747243020515e-05,
      "loss": 0.8268,
      "step": 3240
    },
    {
      "epoch": 1.7403778396698097,
      "grad_norm": 0.8046875,
      "learning_rate": 3.652997974258644e-05,
      "loss": 1.0443,
      "step": 3241
    },
    {
      "epoch": 1.7409147344048859,
      "grad_norm": 0.81640625,
      "learning_rate": 3.6522485739251034e-05,
      "loss": 0.89,
      "step": 3242
    },
    {
      "epoch": 1.7414516291399618,
      "grad_norm": 0.73046875,
      "learning_rate": 3.651499042105407e-05,
      "loss": 0.6678,
      "step": 3243
    },
    {
      "epoch": 1.7419885238750377,
      "grad_norm": 0.703125,
      "learning_rate": 3.650749378885089e-05,
      "loss": 0.8595,
      "step": 3244
    },
    {
      "epoch": 1.7425254186101138,
      "grad_norm": 0.66015625,
      "learning_rate": 3.649999584349693e-05,
      "loss": 0.8412,
      "step": 3245
    },
    {
      "epoch": 1.74306231334519,
      "grad_norm": 0.7734375,
      "learning_rate": 3.649249658584781e-05,
      "loss": 0.805,
      "step": 3246
    },
    {
      "epoch": 1.7435992080802658,
      "grad_norm": 0.73828125,
      "learning_rate": 3.648499601675928e-05,
      "loss": 0.9899,
      "step": 3247
    },
    {
      "epoch": 1.7441361028153417,
      "grad_norm": 0.62890625,
      "learning_rate": 3.6477494137087276e-05,
      "loss": 0.8808,
      "step": 3248
    },
    {
      "epoch": 1.7446729975504178,
      "grad_norm": 0.85546875,
      "learning_rate": 3.6469990947687825e-05,
      "loss": 1.1073,
      "step": 3249
    },
    {
      "epoch": 1.7452098922854937,
      "grad_norm": 0.61328125,
      "learning_rate": 3.646248644941716e-05,
      "loss": 0.8673,
      "step": 3250
    },
    {
      "epoch": 1.7452098922854937,
      "eval_loss": 0.9094784259796143,
      "eval_runtime": 47.2403,
      "eval_samples_per_second": 20.364,
      "eval_steps_per_second": 20.364,
      "step": 3250
    },
    {
      "epoch": 1.7457467870205696,
      "grad_norm": 0.63671875,
      "learning_rate": 3.6454980643131627e-05,
      "loss": 1.0337,
      "step": 3251
    },
    {
      "epoch": 1.7462836817556457,
      "grad_norm": 0.625,
      "learning_rate": 3.6447473529687736e-05,
      "loss": 0.9485,
      "step": 3252
    },
    {
      "epoch": 1.7468205764907219,
      "grad_norm": 0.73828125,
      "learning_rate": 3.643996510994215e-05,
      "loss": 0.8813,
      "step": 3253
    },
    {
      "epoch": 1.7473574712257978,
      "grad_norm": 0.6640625,
      "learning_rate": 3.643245538475166e-05,
      "loss": 0.9719,
      "step": 3254
    },
    {
      "epoch": 1.7478943659608737,
      "grad_norm": 0.70703125,
      "learning_rate": 3.642494435497324e-05,
      "loss": 0.9222,
      "step": 3255
    },
    {
      "epoch": 1.7484312606959498,
      "grad_norm": 0.7109375,
      "learning_rate": 3.641743202146398e-05,
      "loss": 0.8616,
      "step": 3256
    },
    {
      "epoch": 1.748968155431026,
      "grad_norm": 0.7421875,
      "learning_rate": 3.6409918385081135e-05,
      "loss": 0.8157,
      "step": 3257
    },
    {
      "epoch": 1.7495050501661018,
      "grad_norm": 0.93359375,
      "learning_rate": 3.640240344668211e-05,
      "loss": 0.9544,
      "step": 3258
    },
    {
      "epoch": 1.7500419449011777,
      "grad_norm": 0.91796875,
      "learning_rate": 3.639488720712445e-05,
      "loss": 0.9077,
      "step": 3259
    },
    {
      "epoch": 1.7505788396362538,
      "grad_norm": 0.65234375,
      "learning_rate": 3.638736966726585e-05,
      "loss": 0.7943,
      "step": 3260
    },
    {
      "epoch": 1.75111573437133,
      "grad_norm": 0.67578125,
      "learning_rate": 3.637985082796417e-05,
      "loss": 1.0695,
      "step": 3261
    },
    {
      "epoch": 1.7516526291064058,
      "grad_norm": 0.65625,
      "learning_rate": 3.6372330690077396e-05,
      "loss": 0.8446,
      "step": 3262
    },
    {
      "epoch": 1.7521895238414817,
      "grad_norm": 0.66796875,
      "learning_rate": 3.6364809254463665e-05,
      "loss": 0.877,
      "step": 3263
    },
    {
      "epoch": 1.7527264185765579,
      "grad_norm": 0.6015625,
      "learning_rate": 3.635728652198128e-05,
      "loss": 0.8914,
      "step": 3264
    },
    {
      "epoch": 1.753263313311634,
      "grad_norm": 0.75390625,
      "learning_rate": 3.634976249348867e-05,
      "loss": 0.9122,
      "step": 3265
    },
    {
      "epoch": 1.7538002080467099,
      "grad_norm": 0.7734375,
      "learning_rate": 3.634223716984443e-05,
      "loss": 0.9247,
      "step": 3266
    },
    {
      "epoch": 1.7543371027817858,
      "grad_norm": 0.671875,
      "learning_rate": 3.633471055190729e-05,
      "loss": 0.694,
      "step": 3267
    },
    {
      "epoch": 1.754873997516862,
      "grad_norm": 0.80859375,
      "learning_rate": 3.632718264053613e-05,
      "loss": 0.9333,
      "step": 3268
    },
    {
      "epoch": 1.7554108922519378,
      "grad_norm": 0.71875,
      "learning_rate": 3.6319653436589996e-05,
      "loss": 0.8551,
      "step": 3269
    },
    {
      "epoch": 1.7559477869870137,
      "grad_norm": 0.63671875,
      "learning_rate": 3.631212294092805e-05,
      "loss": 0.8005,
      "step": 3270
    },
    {
      "epoch": 1.7564846817220898,
      "grad_norm": 0.90234375,
      "learning_rate": 3.630459115440961e-05,
      "loss": 1.0247,
      "step": 3271
    },
    {
      "epoch": 1.757021576457166,
      "grad_norm": 0.765625,
      "learning_rate": 3.629705807789416e-05,
      "loss": 0.8605,
      "step": 3272
    },
    {
      "epoch": 1.7575584711922418,
      "grad_norm": 0.69921875,
      "learning_rate": 3.6289523712241316e-05,
      "loss": 1.124,
      "step": 3273
    },
    {
      "epoch": 1.7580953659273177,
      "grad_norm": 0.5703125,
      "learning_rate": 3.6281988058310856e-05,
      "loss": 0.7418,
      "step": 3274
    },
    {
      "epoch": 1.7586322606623939,
      "grad_norm": 0.67578125,
      "learning_rate": 3.627445111696268e-05,
      "loss": 0.6977,
      "step": 3275
    },
    {
      "epoch": 1.75916915539747,
      "grad_norm": 0.71484375,
      "learning_rate": 3.626691288905684e-05,
      "loss": 0.7461,
      "step": 3276
    },
    {
      "epoch": 1.7597060501325459,
      "grad_norm": 0.80859375,
      "learning_rate": 3.625937337545357e-05,
      "loss": 0.8126,
      "step": 3277
    },
    {
      "epoch": 1.7602429448676218,
      "grad_norm": 0.66796875,
      "learning_rate": 3.6251832577013214e-05,
      "loss": 0.9651,
      "step": 3278
    },
    {
      "epoch": 1.760779839602698,
      "grad_norm": 0.78125,
      "learning_rate": 3.6244290494596254e-05,
      "loss": 0.8284,
      "step": 3279
    },
    {
      "epoch": 1.761316734337774,
      "grad_norm": 0.80078125,
      "learning_rate": 3.623674712906335e-05,
      "loss": 0.9531,
      "step": 3280
    },
    {
      "epoch": 1.76185362907285,
      "grad_norm": 0.671875,
      "learning_rate": 3.6229202481275306e-05,
      "loss": 0.7829,
      "step": 3281
    },
    {
      "epoch": 1.7623905238079258,
      "grad_norm": 0.67578125,
      "learning_rate": 3.622165655209305e-05,
      "loss": 0.764,
      "step": 3282
    },
    {
      "epoch": 1.762927418543002,
      "grad_norm": 0.75390625,
      "learning_rate": 3.621410934237767e-05,
      "loss": 1.1222,
      "step": 3283
    },
    {
      "epoch": 1.763464313278078,
      "grad_norm": 0.734375,
      "learning_rate": 3.62065608529904e-05,
      "loss": 1.0118,
      "step": 3284
    },
    {
      "epoch": 1.764001208013154,
      "grad_norm": 0.7421875,
      "learning_rate": 3.619901108479261e-05,
      "loss": 1.0181,
      "step": 3285
    },
    {
      "epoch": 1.7645381027482299,
      "grad_norm": 0.640625,
      "learning_rate": 3.619146003864585e-05,
      "loss": 0.8426,
      "step": 3286
    },
    {
      "epoch": 1.765074997483306,
      "grad_norm": 0.64453125,
      "learning_rate": 3.618390771541175e-05,
      "loss": 0.7738,
      "step": 3287
    },
    {
      "epoch": 1.7656118922183819,
      "grad_norm": 0.6953125,
      "learning_rate": 3.617635411595215e-05,
      "loss": 0.931,
      "step": 3288
    },
    {
      "epoch": 1.7661487869534578,
      "grad_norm": 0.73046875,
      "learning_rate": 3.616879924112902e-05,
      "loss": 0.7866,
      "step": 3289
    },
    {
      "epoch": 1.766685681688534,
      "grad_norm": 0.6796875,
      "learning_rate": 3.6161243091804446e-05,
      "loss": 0.9237,
      "step": 3290
    },
    {
      "epoch": 1.76722257642361,
      "grad_norm": 0.69921875,
      "learning_rate": 3.61536856688407e-05,
      "loss": 0.8205,
      "step": 3291
    },
    {
      "epoch": 1.767759471158686,
      "grad_norm": 0.75390625,
      "learning_rate": 3.614612697310016e-05,
      "loss": 1.009,
      "step": 3292
    },
    {
      "epoch": 1.7682963658937618,
      "grad_norm": 0.77734375,
      "learning_rate": 3.6138567005445375e-05,
      "loss": 1.1373,
      "step": 3293
    },
    {
      "epoch": 1.768833260628838,
      "grad_norm": 0.7578125,
      "learning_rate": 3.6131005766739046e-05,
      "loss": 1.1374,
      "step": 3294
    },
    {
      "epoch": 1.769370155363914,
      "grad_norm": 0.76171875,
      "learning_rate": 3.612344325784399e-05,
      "loss": 0.9462,
      "step": 3295
    },
    {
      "epoch": 1.76990705009899,
      "grad_norm": 0.7265625,
      "learning_rate": 3.611587947962319e-05,
      "loss": 1.124,
      "step": 3296
    },
    {
      "epoch": 1.7704439448340659,
      "grad_norm": 0.65625,
      "learning_rate": 3.610831443293977e-05,
      "loss": 0.8779,
      "step": 3297
    },
    {
      "epoch": 1.770980839569142,
      "grad_norm": 1.046875,
      "learning_rate": 3.6100748118657e-05,
      "loss": 0.8598,
      "step": 3298
    },
    {
      "epoch": 1.771517734304218,
      "grad_norm": 0.86328125,
      "learning_rate": 3.6093180537638294e-05,
      "loss": 0.8016,
      "step": 3299
    },
    {
      "epoch": 1.772054629039294,
      "grad_norm": 1.125,
      "learning_rate": 3.60856116907472e-05,
      "loss": 1.0133,
      "step": 3300
    },
    {
      "epoch": 1.772054629039294,
      "eval_loss": 0.9077935218811035,
      "eval_runtime": 49.1719,
      "eval_samples_per_second": 19.564,
      "eval_steps_per_second": 19.564,
      "step": 3300
    },
    {
      "epoch": 1.77259152377437,
      "grad_norm": 0.5625,
      "learning_rate": 3.607804157884742e-05,
      "loss": 0.771,
      "step": 3301
    },
    {
      "epoch": 1.773128418509446,
      "grad_norm": 0.671875,
      "learning_rate": 3.607047020280281e-05,
      "loss": 0.9158,
      "step": 3302
    },
    {
      "epoch": 1.7736653132445221,
      "grad_norm": 0.6328125,
      "learning_rate": 3.606289756347736e-05,
      "loss": 0.8109,
      "step": 3303
    },
    {
      "epoch": 1.774202207979598,
      "grad_norm": 0.70703125,
      "learning_rate": 3.605532366173518e-05,
      "loss": 0.933,
      "step": 3304
    },
    {
      "epoch": 1.774739102714674,
      "grad_norm": 0.9453125,
      "learning_rate": 3.6047748498440585e-05,
      "loss": 0.9528,
      "step": 3305
    },
    {
      "epoch": 1.77527599744975,
      "grad_norm": 0.85546875,
      "learning_rate": 3.604017207445797e-05,
      "loss": 0.807,
      "step": 3306
    },
    {
      "epoch": 1.775812892184826,
      "grad_norm": 0.65625,
      "learning_rate": 3.603259439065191e-05,
      "loss": 0.6761,
      "step": 3307
    },
    {
      "epoch": 1.7763497869199019,
      "grad_norm": 0.8984375,
      "learning_rate": 3.602501544788711e-05,
      "loss": 0.8852,
      "step": 3308
    },
    {
      "epoch": 1.776886681654978,
      "grad_norm": 0.6796875,
      "learning_rate": 3.6017435247028425e-05,
      "loss": 0.9164,
      "step": 3309
    },
    {
      "epoch": 1.777423576390054,
      "grad_norm": 0.80859375,
      "learning_rate": 3.600985378894086e-05,
      "loss": 0.8396,
      "step": 3310
    },
    {
      "epoch": 1.77796047112513,
      "grad_norm": 0.765625,
      "learning_rate": 3.6002271074489554e-05,
      "loss": 0.9967,
      "step": 3311
    },
    {
      "epoch": 1.778497365860206,
      "grad_norm": 0.6796875,
      "learning_rate": 3.599468710453977e-05,
      "loss": 0.9166,
      "step": 3312
    },
    {
      "epoch": 1.779034260595282,
      "grad_norm": 0.6484375,
      "learning_rate": 3.5987101879956965e-05,
      "loss": 0.7454,
      "step": 3313
    },
    {
      "epoch": 1.7795711553303581,
      "grad_norm": 0.5859375,
      "learning_rate": 3.597951540160669e-05,
      "loss": 0.792,
      "step": 3314
    },
    {
      "epoch": 1.780108050065434,
      "grad_norm": 0.734375,
      "learning_rate": 3.597192767035466e-05,
      "loss": 0.9001,
      "step": 3315
    },
    {
      "epoch": 1.78064494480051,
      "grad_norm": 0.84375,
      "learning_rate": 3.596433868706674e-05,
      "loss": 0.9441,
      "step": 3316
    },
    {
      "epoch": 1.781181839535586,
      "grad_norm": 0.7421875,
      "learning_rate": 3.595674845260891e-05,
      "loss": 0.8817,
      "step": 3317
    },
    {
      "epoch": 1.7817187342706622,
      "grad_norm": 0.8671875,
      "learning_rate": 3.594915696784734e-05,
      "loss": 1.0732,
      "step": 3318
    },
    {
      "epoch": 1.782255629005738,
      "grad_norm": 0.83203125,
      "learning_rate": 3.59415642336483e-05,
      "loss": 1.038,
      "step": 3319
    },
    {
      "epoch": 1.782792523740814,
      "grad_norm": 0.65625,
      "learning_rate": 3.593397025087821e-05,
      "loss": 0.8258,
      "step": 3320
    },
    {
      "epoch": 1.78332941847589,
      "grad_norm": 0.7109375,
      "learning_rate": 3.5926375020403655e-05,
      "loss": 0.7235,
      "step": 3321
    },
    {
      "epoch": 1.7838663132109662,
      "grad_norm": 0.7578125,
      "learning_rate": 3.5918778543091334e-05,
      "loss": 0.8121,
      "step": 3322
    },
    {
      "epoch": 1.7844032079460421,
      "grad_norm": 0.73046875,
      "learning_rate": 3.5911180819808103e-05,
      "loss": 1.0629,
      "step": 3323
    },
    {
      "epoch": 1.784940102681118,
      "grad_norm": 0.734375,
      "learning_rate": 3.590358185142098e-05,
      "loss": 0.8761,
      "step": 3324
    },
    {
      "epoch": 1.7854769974161941,
      "grad_norm": 0.7109375,
      "learning_rate": 3.589598163879706e-05,
      "loss": 0.8972,
      "step": 3325
    },
    {
      "epoch": 1.78601389215127,
      "grad_norm": 0.73046875,
      "learning_rate": 3.588838018280366e-05,
      "loss": 0.8558,
      "step": 3326
    },
    {
      "epoch": 1.786550786886346,
      "grad_norm": 0.6328125,
      "learning_rate": 3.588077748430819e-05,
      "loss": 0.6925,
      "step": 3327
    },
    {
      "epoch": 1.787087681621422,
      "grad_norm": 0.75,
      "learning_rate": 3.587317354417822e-05,
      "loss": 0.9284,
      "step": 3328
    },
    {
      "epoch": 1.7876245763564982,
      "grad_norm": 0.7265625,
      "learning_rate": 3.5865568363281443e-05,
      "loss": 1.1003,
      "step": 3329
    },
    {
      "epoch": 1.788161471091574,
      "grad_norm": 0.84765625,
      "learning_rate": 3.585796194248572e-05,
      "loss": 0.8994,
      "step": 3330
    },
    {
      "epoch": 1.78869836582665,
      "grad_norm": 0.76171875,
      "learning_rate": 3.585035428265903e-05,
      "loss": 1.1384,
      "step": 3331
    },
    {
      "epoch": 1.789235260561726,
      "grad_norm": 0.65625,
      "learning_rate": 3.584274538466951e-05,
      "loss": 0.8725,
      "step": 3332
    },
    {
      "epoch": 1.7897721552968022,
      "grad_norm": 0.640625,
      "learning_rate": 3.5835135249385433e-05,
      "loss": 0.9772,
      "step": 3333
    },
    {
      "epoch": 1.7903090500318781,
      "grad_norm": 0.625,
      "learning_rate": 3.5827523877675196e-05,
      "loss": 0.9266,
      "step": 3334
    },
    {
      "epoch": 1.790845944766954,
      "grad_norm": 0.69921875,
      "learning_rate": 3.5819911270407377e-05,
      "loss": 0.9851,
      "step": 3335
    },
    {
      "epoch": 1.7913828395020301,
      "grad_norm": 0.64453125,
      "learning_rate": 3.581229742845065e-05,
      "loss": 0.837,
      "step": 3336
    },
    {
      "epoch": 1.7919197342371063,
      "grad_norm": 0.7578125,
      "learning_rate": 3.580468235267387e-05,
      "loss": 0.8346,
      "step": 3337
    },
    {
      "epoch": 1.7924566289721822,
      "grad_norm": 0.68359375,
      "learning_rate": 3.5797066043946e-05,
      "loss": 0.9218,
      "step": 3338
    },
    {
      "epoch": 1.792993523707258,
      "grad_norm": 0.7265625,
      "learning_rate": 3.578944850313615e-05,
      "loss": 1.0078,
      "step": 3339
    },
    {
      "epoch": 1.7935304184423342,
      "grad_norm": 0.71484375,
      "learning_rate": 3.5781829731113605e-05,
      "loss": 0.7248,
      "step": 3340
    },
    {
      "epoch": 1.7940673131774103,
      "grad_norm": 0.7578125,
      "learning_rate": 3.577420972874774e-05,
      "loss": 0.9012,
      "step": 3341
    },
    {
      "epoch": 1.7946042079124862,
      "grad_norm": 0.640625,
      "learning_rate": 3.5766588496908095e-05,
      "loss": 0.5642,
      "step": 3342
    },
    {
      "epoch": 1.795141102647562,
      "grad_norm": 0.83984375,
      "learning_rate": 3.5758966036464356e-05,
      "loss": 0.9413,
      "step": 3343
    },
    {
      "epoch": 1.7956779973826382,
      "grad_norm": 0.72265625,
      "learning_rate": 3.5751342348286354e-05,
      "loss": 0.8961,
      "step": 3344
    },
    {
      "epoch": 1.7962148921177141,
      "grad_norm": 0.66015625,
      "learning_rate": 3.5743717433244025e-05,
      "loss": 0.8161,
      "step": 3345
    },
    {
      "epoch": 1.79675178685279,
      "grad_norm": 0.65234375,
      "learning_rate": 3.573609129220748e-05,
      "loss": 0.8581,
      "step": 3346
    },
    {
      "epoch": 1.7972886815878661,
      "grad_norm": 0.6875,
      "learning_rate": 3.572846392604695e-05,
      "loss": 0.9598,
      "step": 3347
    },
    {
      "epoch": 1.7978255763229423,
      "grad_norm": 0.75390625,
      "learning_rate": 3.572083533563283e-05,
      "loss": 0.7201,
      "step": 3348
    },
    {
      "epoch": 1.7983624710580182,
      "grad_norm": 0.80078125,
      "learning_rate": 3.571320552183563e-05,
      "loss": 0.92,
      "step": 3349
    },
    {
      "epoch": 1.798899365793094,
      "grad_norm": 0.625,
      "learning_rate": 3.570557448552601e-05,
      "loss": 0.8208,
      "step": 3350
    },
    {
      "epoch": 1.798899365793094,
      "eval_loss": 0.9074547290802002,
      "eval_runtime": 46.6661,
      "eval_samples_per_second": 20.615,
      "eval_steps_per_second": 20.615,
      "step": 3350
    },
    {
      "epoch": 1.7994362605281702,
      "grad_norm": 0.609375,
      "learning_rate": 3.569794222757475e-05,
      "loss": 0.8413,
      "step": 3351
    },
    {
      "epoch": 1.7999731552632463,
      "grad_norm": 0.7265625,
      "learning_rate": 3.569030874885282e-05,
      "loss": 0.7694,
      "step": 3352
    },
    {
      "epoch": 1.8005100499983222,
      "grad_norm": 0.9296875,
      "learning_rate": 3.568267405023128e-05,
      "loss": 0.8728,
      "step": 3353
    },
    {
      "epoch": 1.801046944733398,
      "grad_norm": 0.7734375,
      "learning_rate": 3.567503813258134e-05,
      "loss": 1.0077,
      "step": 3354
    },
    {
      "epoch": 1.8015838394684742,
      "grad_norm": 0.82421875,
      "learning_rate": 3.5667400996774356e-05,
      "loss": 0.911,
      "step": 3355
    },
    {
      "epoch": 1.8021207342035503,
      "grad_norm": 0.6015625,
      "learning_rate": 3.5659762643681834e-05,
      "loss": 0.7527,
      "step": 3356
    },
    {
      "epoch": 1.8026576289386262,
      "grad_norm": 0.80078125,
      "learning_rate": 3.56521230741754e-05,
      "loss": 0.9561,
      "step": 3357
    },
    {
      "epoch": 1.8031945236737021,
      "grad_norm": 0.84765625,
      "learning_rate": 3.564448228912682e-05,
      "loss": 0.9424,
      "step": 3358
    },
    {
      "epoch": 1.8037314184087783,
      "grad_norm": 0.76171875,
      "learning_rate": 3.563684028940801e-05,
      "loss": 0.8091,
      "step": 3359
    },
    {
      "epoch": 1.8042683131438544,
      "grad_norm": 0.6953125,
      "learning_rate": 3.562919707589102e-05,
      "loss": 0.8003,
      "step": 3360
    },
    {
      "epoch": 1.8048052078789303,
      "grad_norm": 0.80078125,
      "learning_rate": 3.562155264944803e-05,
      "loss": 0.9384,
      "step": 3361
    },
    {
      "epoch": 1.8053421026140062,
      "grad_norm": 0.69921875,
      "learning_rate": 3.5613907010951384e-05,
      "loss": 0.9769,
      "step": 3362
    },
    {
      "epoch": 1.8058789973490823,
      "grad_norm": 0.61328125,
      "learning_rate": 3.560626016127353e-05,
      "loss": 1.0228,
      "step": 3363
    },
    {
      "epoch": 1.8064158920841582,
      "grad_norm": 0.71484375,
      "learning_rate": 3.559861210128706e-05,
      "loss": 0.8613,
      "step": 3364
    },
    {
      "epoch": 1.806952786819234,
      "grad_norm": 0.6171875,
      "learning_rate": 3.559096283186475e-05,
      "loss": 0.8571,
      "step": 3365
    },
    {
      "epoch": 1.8074896815543102,
      "grad_norm": 0.75390625,
      "learning_rate": 3.5583312353879444e-05,
      "loss": 1.0133,
      "step": 3366
    },
    {
      "epoch": 1.8080265762893863,
      "grad_norm": 0.703125,
      "learning_rate": 3.5575660668204176e-05,
      "loss": 0.9155,
      "step": 3367
    },
    {
      "epoch": 1.8085634710244622,
      "grad_norm": 0.7421875,
      "learning_rate": 3.55680077757121e-05,
      "loss": 0.9624,
      "step": 3368
    },
    {
      "epoch": 1.8091003657595381,
      "grad_norm": 0.78125,
      "learning_rate": 3.556035367727649e-05,
      "loss": 1.0062,
      "step": 3369
    },
    {
      "epoch": 1.8096372604946143,
      "grad_norm": 0.69140625,
      "learning_rate": 3.55526983737708e-05,
      "loss": 0.8793,
      "step": 3370
    },
    {
      "epoch": 1.8101741552296904,
      "grad_norm": 0.65625,
      "learning_rate": 3.554504186606859e-05,
      "loss": 0.9257,
      "step": 3371
    },
    {
      "epoch": 1.8107110499647663,
      "grad_norm": 0.5859375,
      "learning_rate": 3.5537384155043546e-05,
      "loss": 0.6916,
      "step": 3372
    },
    {
      "epoch": 1.8112479446998422,
      "grad_norm": 0.70703125,
      "learning_rate": 3.5529725241569536e-05,
      "loss": 0.885,
      "step": 3373
    },
    {
      "epoch": 1.8117848394349183,
      "grad_norm": 0.63671875,
      "learning_rate": 3.552206512652053e-05,
      "loss": 0.8285,
      "step": 3374
    },
    {
      "epoch": 1.8123217341699944,
      "grad_norm": 0.64453125,
      "learning_rate": 3.5514403810770635e-05,
      "loss": 0.7443,
      "step": 3375
    },
    {
      "epoch": 1.8128586289050703,
      "grad_norm": 0.6953125,
      "learning_rate": 3.5506741295194115e-05,
      "loss": 0.8971,
      "step": 3376
    },
    {
      "epoch": 1.8133955236401462,
      "grad_norm": 0.59375,
      "learning_rate": 3.5499077580665355e-05,
      "loss": 0.7787,
      "step": 3377
    },
    {
      "epoch": 1.8139324183752223,
      "grad_norm": 0.95703125,
      "learning_rate": 3.549141266805888e-05,
      "loss": 0.6698,
      "step": 3378
    },
    {
      "epoch": 1.8144693131102985,
      "grad_norm": 0.92578125,
      "learning_rate": 3.548374655824936e-05,
      "loss": 0.8459,
      "step": 3379
    },
    {
      "epoch": 1.8150062078453744,
      "grad_norm": 0.76953125,
      "learning_rate": 3.547607925211159e-05,
      "loss": 0.9377,
      "step": 3380
    },
    {
      "epoch": 1.8155431025804503,
      "grad_norm": 0.91796875,
      "learning_rate": 3.546841075052051e-05,
      "loss": 0.9742,
      "step": 3381
    },
    {
      "epoch": 1.8160799973155264,
      "grad_norm": 0.8359375,
      "learning_rate": 3.546074105435119e-05,
      "loss": 1.0094,
      "step": 3382
    },
    {
      "epoch": 1.8166168920506023,
      "grad_norm": 0.85546875,
      "learning_rate": 3.545307016447884e-05,
      "loss": 0.9406,
      "step": 3383
    },
    {
      "epoch": 1.8171537867856782,
      "grad_norm": 0.73046875,
      "learning_rate": 3.5445398081778814e-05,
      "loss": 0.9025,
      "step": 3384
    },
    {
      "epoch": 1.8176906815207543,
      "grad_norm": 0.875,
      "learning_rate": 3.543772480712658e-05,
      "loss": 0.9076,
      "step": 3385
    },
    {
      "epoch": 1.8182275762558304,
      "grad_norm": 0.60546875,
      "learning_rate": 3.543005034139776e-05,
      "loss": 0.8067,
      "step": 3386
    },
    {
      "epoch": 1.8187644709909063,
      "grad_norm": 0.65234375,
      "learning_rate": 3.542237468546812e-05,
      "loss": 0.6781,
      "step": 3387
    },
    {
      "epoch": 1.8193013657259822,
      "grad_norm": 0.88671875,
      "learning_rate": 3.541469784021353e-05,
      "loss": 0.9106,
      "step": 3388
    },
    {
      "epoch": 1.8198382604610583,
      "grad_norm": 0.609375,
      "learning_rate": 3.540701980651003e-05,
      "loss": 0.6246,
      "step": 3389
    },
    {
      "epoch": 1.8203751551961345,
      "grad_norm": 0.671875,
      "learning_rate": 3.5399340585233774e-05,
      "loss": 0.9278,
      "step": 3390
    },
    {
      "epoch": 1.8209120499312104,
      "grad_norm": 0.703125,
      "learning_rate": 3.5391660177261064e-05,
      "loss": 0.9695,
      "step": 3391
    },
    {
      "epoch": 1.8214489446662863,
      "grad_norm": 0.67578125,
      "learning_rate": 3.5383978583468324e-05,
      "loss": 0.7959,
      "step": 3392
    },
    {
      "epoch": 1.8219858394013624,
      "grad_norm": 0.82421875,
      "learning_rate": 3.537629580473213e-05,
      "loss": 1.0399,
      "step": 3393
    },
    {
      "epoch": 1.8225227341364385,
      "grad_norm": 0.7734375,
      "learning_rate": 3.536861184192917e-05,
      "loss": 0.9706,
      "step": 3394
    },
    {
      "epoch": 1.8230596288715144,
      "grad_norm": 0.65625,
      "learning_rate": 3.53609266959363e-05,
      "loss": 0.7903,
      "step": 3395
    },
    {
      "epoch": 1.8235965236065903,
      "grad_norm": 0.74609375,
      "learning_rate": 3.535324036763048e-05,
      "loss": 0.9438,
      "step": 3396
    },
    {
      "epoch": 1.8241334183416664,
      "grad_norm": 0.6796875,
      "learning_rate": 3.534555285788882e-05,
      "loss": 0.9121,
      "step": 3397
    },
    {
      "epoch": 1.8246703130767425,
      "grad_norm": 0.6484375,
      "learning_rate": 3.5337864167588566e-05,
      "loss": 0.9203,
      "step": 3398
    },
    {
      "epoch": 1.8252072078118184,
      "grad_norm": 0.71875,
      "learning_rate": 3.533017429760709e-05,
      "loss": 0.8828,
      "step": 3399
    },
    {
      "epoch": 1.8257441025468943,
      "grad_norm": 0.69921875,
      "learning_rate": 3.5322483248821904e-05,
      "loss": 0.8854,
      "step": 3400
    },
    {
      "epoch": 1.8257441025468943,
      "eval_loss": 0.9071779847145081,
      "eval_runtime": 46.4737,
      "eval_samples_per_second": 20.7,
      "eval_steps_per_second": 20.7,
      "step": 3400
    },
    {
      "epoch": 1.8262809972819705,
      "grad_norm": 0.65625,
      "learning_rate": 3.531479102211066e-05,
      "loss": 0.9293,
      "step": 3401
    },
    {
      "epoch": 1.8268178920170464,
      "grad_norm": 0.78515625,
      "learning_rate": 3.5307097618351126e-05,
      "loss": 1.0142,
      "step": 3402
    },
    {
      "epoch": 1.8273547867521223,
      "grad_norm": 0.66015625,
      "learning_rate": 3.529940303842123e-05,
      "loss": 0.77,
      "step": 3403
    },
    {
      "epoch": 1.8278916814871984,
      "grad_norm": 0.6796875,
      "learning_rate": 3.529170728319903e-05,
      "loss": 0.8665,
      "step": 3404
    },
    {
      "epoch": 1.8284285762222745,
      "grad_norm": 0.8203125,
      "learning_rate": 3.528401035356267e-05,
      "loss": 0.9725,
      "step": 3405
    },
    {
      "epoch": 1.8289654709573504,
      "grad_norm": 0.67578125,
      "learning_rate": 3.527631225039051e-05,
      "loss": 0.7628,
      "step": 3406
    },
    {
      "epoch": 1.8295023656924263,
      "grad_norm": 0.6796875,
      "learning_rate": 3.526861297456097e-05,
      "loss": 0.954,
      "step": 3407
    },
    {
      "epoch": 1.8300392604275024,
      "grad_norm": 0.7421875,
      "learning_rate": 3.526091252695266e-05,
      "loss": 0.8412,
      "step": 3408
    },
    {
      "epoch": 1.8305761551625785,
      "grad_norm": 0.59765625,
      "learning_rate": 3.5253210908444276e-05,
      "loss": 0.7154,
      "step": 3409
    },
    {
      "epoch": 1.8311130498976544,
      "grad_norm": 0.73046875,
      "learning_rate": 3.5245508119914687e-05,
      "loss": 0.759,
      "step": 3410
    },
    {
      "epoch": 1.8316499446327303,
      "grad_norm": 0.6953125,
      "learning_rate": 3.5237804162242874e-05,
      "loss": 0.8561,
      "step": 3411
    },
    {
      "epoch": 1.8321868393678065,
      "grad_norm": 0.6328125,
      "learning_rate": 3.523009903630796e-05,
      "loss": 0.8855,
      "step": 3412
    },
    {
      "epoch": 1.8327237341028826,
      "grad_norm": 0.69140625,
      "learning_rate": 3.522239274298918e-05,
      "loss": 0.7588,
      "step": 3413
    },
    {
      "epoch": 1.8332606288379585,
      "grad_norm": 0.65625,
      "learning_rate": 3.521468528316594e-05,
      "loss": 0.7918,
      "step": 3414
    },
    {
      "epoch": 1.8337975235730344,
      "grad_norm": 0.8984375,
      "learning_rate": 3.520697665771774e-05,
      "loss": 0.9868,
      "step": 3415
    },
    {
      "epoch": 1.8343344183081105,
      "grad_norm": 0.73046875,
      "learning_rate": 3.5199266867524254e-05,
      "loss": 0.9829,
      "step": 3416
    },
    {
      "epoch": 1.8348713130431866,
      "grad_norm": 0.8359375,
      "learning_rate": 3.5191555913465254e-05,
      "loss": 0.9624,
      "step": 3417
    },
    {
      "epoch": 1.8354082077782625,
      "grad_norm": 0.671875,
      "learning_rate": 3.518384379642066e-05,
      "loss": 0.8279,
      "step": 3418
    },
    {
      "epoch": 1.8359451025133384,
      "grad_norm": 0.74609375,
      "learning_rate": 3.517613051727051e-05,
      "loss": 0.8205,
      "step": 3419
    },
    {
      "epoch": 1.8364819972484145,
      "grad_norm": 0.69921875,
      "learning_rate": 3.516841607689501e-05,
      "loss": 0.8706,
      "step": 3420
    },
    {
      "epoch": 1.8370188919834904,
      "grad_norm": 0.671875,
      "learning_rate": 3.5160700476174456e-05,
      "loss": 0.8847,
      "step": 3421
    },
    {
      "epoch": 1.8375557867185663,
      "grad_norm": 0.703125,
      "learning_rate": 3.515298371598931e-05,
      "loss": 0.9828,
      "step": 3422
    },
    {
      "epoch": 1.8380926814536425,
      "grad_norm": 0.82421875,
      "learning_rate": 3.5145265797220134e-05,
      "loss": 0.975,
      "step": 3423
    },
    {
      "epoch": 1.8386295761887186,
      "grad_norm": 0.58984375,
      "learning_rate": 3.513754672074765e-05,
      "loss": 0.8465,
      "step": 3424
    },
    {
      "epoch": 1.8391664709237945,
      "grad_norm": 0.66796875,
      "learning_rate": 3.5129826487452724e-05,
      "loss": 0.8347,
      "step": 3425
    },
    {
      "epoch": 1.8397033656588704,
      "grad_norm": 0.79296875,
      "learning_rate": 3.51221050982163e-05,
      "loss": 0.8954,
      "step": 3426
    },
    {
      "epoch": 1.8402402603939465,
      "grad_norm": 0.66015625,
      "learning_rate": 3.511438255391948e-05,
      "loss": 0.772,
      "step": 3427
    },
    {
      "epoch": 1.8407771551290226,
      "grad_norm": 0.79296875,
      "learning_rate": 3.510665885544354e-05,
      "loss": 1.1601,
      "step": 3428
    },
    {
      "epoch": 1.8413140498640985,
      "grad_norm": 0.68359375,
      "learning_rate": 3.509893400366984e-05,
      "loss": 0.9285,
      "step": 3429
    },
    {
      "epoch": 1.8418509445991744,
      "grad_norm": 0.74609375,
      "learning_rate": 3.509120799947987e-05,
      "loss": 0.7532,
      "step": 3430
    },
    {
      "epoch": 1.8423878393342505,
      "grad_norm": 0.68359375,
      "learning_rate": 3.508348084375527e-05,
      "loss": 0.8838,
      "step": 3431
    },
    {
      "epoch": 1.8429247340693267,
      "grad_norm": 0.92578125,
      "learning_rate": 3.5075752537377816e-05,
      "loss": 0.7139,
      "step": 3432
    },
    {
      "epoch": 1.8434616288044026,
      "grad_norm": 0.88671875,
      "learning_rate": 3.5068023081229396e-05,
      "loss": 0.9805,
      "step": 3433
    },
    {
      "epoch": 1.8439985235394785,
      "grad_norm": 1.171875,
      "learning_rate": 3.5060292476192035e-05,
      "loss": 0.8156,
      "step": 3434
    },
    {
      "epoch": 1.8445354182745546,
      "grad_norm": 0.6484375,
      "learning_rate": 3.5052560723147895e-05,
      "loss": 0.8591,
      "step": 3435
    },
    {
      "epoch": 1.8450723130096307,
      "grad_norm": 0.69140625,
      "learning_rate": 3.5044827822979286e-05,
      "loss": 0.8497,
      "step": 3436
    },
    {
      "epoch": 1.8456092077447066,
      "grad_norm": 0.59375,
      "learning_rate": 3.503709377656861e-05,
      "loss": 0.8224,
      "step": 3437
    },
    {
      "epoch": 1.8461461024797825,
      "grad_norm": 0.7890625,
      "learning_rate": 3.5029358584798424e-05,
      "loss": 1.0091,
      "step": 3438
    },
    {
      "epoch": 1.8466829972148586,
      "grad_norm": 0.68359375,
      "learning_rate": 3.5021622248551414e-05,
      "loss": 0.8467,
      "step": 3439
    },
    {
      "epoch": 1.8472198919499345,
      "grad_norm": 0.8984375,
      "learning_rate": 3.501388476871039e-05,
      "loss": 0.8345,
      "step": 3440
    },
    {
      "epoch": 1.8477567866850104,
      "grad_norm": 0.83984375,
      "learning_rate": 3.500614614615829e-05,
      "loss": 0.7775,
      "step": 3441
    },
    {
      "epoch": 1.8482936814200865,
      "grad_norm": 0.79296875,
      "learning_rate": 3.4998406381778203e-05,
      "loss": 0.7873,
      "step": 3442
    },
    {
      "epoch": 1.8488305761551627,
      "grad_norm": 0.7265625,
      "learning_rate": 3.499066547645333e-05,
      "loss": 1.1564,
      "step": 3443
    },
    {
      "epoch": 1.8493674708902386,
      "grad_norm": 0.69140625,
      "learning_rate": 3.4982923431067e-05,
      "loss": 0.7893,
      "step": 3444
    },
    {
      "epoch": 1.8499043656253145,
      "grad_norm": 0.71484375,
      "learning_rate": 3.497518024650269e-05,
      "loss": 0.8709,
      "step": 3445
    },
    {
      "epoch": 1.8504412603603906,
      "grad_norm": 0.76953125,
      "learning_rate": 3.496743592364399e-05,
      "loss": 0.9887,
      "step": 3446
    },
    {
      "epoch": 1.8509781550954667,
      "grad_norm": 0.7265625,
      "learning_rate": 3.4959690463374615e-05,
      "loss": 1.0011,
      "step": 3447
    },
    {
      "epoch": 1.8515150498305426,
      "grad_norm": 0.640625,
      "learning_rate": 3.495194386657843e-05,
      "loss": 0.9166,
      "step": 3448
    },
    {
      "epoch": 1.8520519445656185,
      "grad_norm": 0.64453125,
      "learning_rate": 3.494419613413942e-05,
      "loss": 0.7669,
      "step": 3449
    },
    {
      "epoch": 1.8525888393006946,
      "grad_norm": 1.0625,
      "learning_rate": 3.493644726694171e-05,
      "loss": 0.81,
      "step": 3450
    },
    {
      "epoch": 1.8525888393006946,
      "eval_loss": 0.9073987603187561,
      "eval_runtime": 46.54,
      "eval_samples_per_second": 20.67,
      "eval_steps_per_second": 20.67,
      "step": 3450
    },
    {
      "epoch": 1.8531257340357707,
      "grad_norm": 0.71484375,
      "learning_rate": 3.4928697265869515e-05,
      "loss": 0.8877,
      "step": 3451
    },
    {
      "epoch": 1.8536626287708466,
      "grad_norm": 0.6953125,
      "learning_rate": 3.492094613180723e-05,
      "loss": 0.8905,
      "step": 3452
    },
    {
      "epoch": 1.8541995235059225,
      "grad_norm": 0.80078125,
      "learning_rate": 3.491319386563936e-05,
      "loss": 0.9149,
      "step": 3453
    },
    {
      "epoch": 1.8547364182409987,
      "grad_norm": 0.62109375,
      "learning_rate": 3.490544046825052e-05,
      "loss": 0.8345,
      "step": 3454
    },
    {
      "epoch": 1.8552733129760748,
      "grad_norm": 0.6171875,
      "learning_rate": 3.4897685940525485e-05,
      "loss": 0.7826,
      "step": 3455
    },
    {
      "epoch": 1.8558102077111507,
      "grad_norm": 0.7734375,
      "learning_rate": 3.4889930283349145e-05,
      "loss": 0.9327,
      "step": 3456
    },
    {
      "epoch": 1.8563471024462266,
      "grad_norm": 0.703125,
      "learning_rate": 3.48821734976065e-05,
      "loss": 0.617,
      "step": 3457
    },
    {
      "epoch": 1.8568839971813027,
      "grad_norm": 0.62890625,
      "learning_rate": 3.487441558418273e-05,
      "loss": 0.7964,
      "step": 3458
    },
    {
      "epoch": 1.8574208919163786,
      "grad_norm": 0.66796875,
      "learning_rate": 3.4866656543963084e-05,
      "loss": 1.0588,
      "step": 3459
    },
    {
      "epoch": 1.8579577866514545,
      "grad_norm": 0.80859375,
      "learning_rate": 3.4858896377832966e-05,
      "loss": 0.9347,
      "step": 3460
    },
    {
      "epoch": 1.8584946813865306,
      "grad_norm": 0.76171875,
      "learning_rate": 3.485113508667793e-05,
      "loss": 0.9807,
      "step": 3461
    },
    {
      "epoch": 1.8590315761216067,
      "grad_norm": 0.66796875,
      "learning_rate": 3.484337267138362e-05,
      "loss": 1.0128,
      "step": 3462
    },
    {
      "epoch": 1.8595684708566826,
      "grad_norm": 0.7578125,
      "learning_rate": 3.4835609132835834e-05,
      "loss": 1.0562,
      "step": 3463
    },
    {
      "epoch": 1.8601053655917585,
      "grad_norm": 0.76953125,
      "learning_rate": 3.482784447192049e-05,
      "loss": 0.7675,
      "step": 3464
    },
    {
      "epoch": 1.8606422603268347,
      "grad_norm": 0.87109375,
      "learning_rate": 3.482007868952362e-05,
      "loss": 1.0954,
      "step": 3465
    },
    {
      "epoch": 1.8611791550619108,
      "grad_norm": 0.66015625,
      "learning_rate": 3.4812311786531426e-05,
      "loss": 0.8647,
      "step": 3466
    },
    {
      "epoch": 1.8617160497969867,
      "grad_norm": 0.7734375,
      "learning_rate": 3.4804543763830195e-05,
      "loss": 0.8386,
      "step": 3467
    },
    {
      "epoch": 1.8622529445320626,
      "grad_norm": 0.7734375,
      "learning_rate": 3.4796774622306345e-05,
      "loss": 0.8632,
      "step": 3468
    },
    {
      "epoch": 1.8627898392671387,
      "grad_norm": 0.8125,
      "learning_rate": 3.478900436284645e-05,
      "loss": 1.1912,
      "step": 3469
    },
    {
      "epoch": 1.8633267340022148,
      "grad_norm": 0.7109375,
      "learning_rate": 3.4781232986337184e-05,
      "loss": 0.7997,
      "step": 3470
    },
    {
      "epoch": 1.8638636287372907,
      "grad_norm": 0.8984375,
      "learning_rate": 3.4773460493665376e-05,
      "loss": 0.8008,
      "step": 3471
    },
    {
      "epoch": 1.8644005234723666,
      "grad_norm": 0.77734375,
      "learning_rate": 3.4765686885717945e-05,
      "loss": 0.9192,
      "step": 3472
    },
    {
      "epoch": 1.8649374182074427,
      "grad_norm": 0.78515625,
      "learning_rate": 3.475791216338197e-05,
      "loss": 0.8563,
      "step": 3473
    },
    {
      "epoch": 1.8654743129425189,
      "grad_norm": 0.69921875,
      "learning_rate": 3.475013632754465e-05,
      "loss": 1.0418,
      "step": 3474
    },
    {
      "epoch": 1.8660112076775948,
      "grad_norm": 0.765625,
      "learning_rate": 3.474235937909329e-05,
      "loss": 0.967,
      "step": 3475
    },
    {
      "epoch": 1.8665481024126707,
      "grad_norm": 0.72265625,
      "learning_rate": 3.473458131891535e-05,
      "loss": 0.7923,
      "step": 3476
    },
    {
      "epoch": 1.8670849971477468,
      "grad_norm": 0.7109375,
      "learning_rate": 3.472680214789841e-05,
      "loss": 0.9474,
      "step": 3477
    },
    {
      "epoch": 1.8676218918828227,
      "grad_norm": 0.6875,
      "learning_rate": 3.471902186693016e-05,
      "loss": 0.8369,
      "step": 3478
    },
    {
      "epoch": 1.8681587866178986,
      "grad_norm": 0.61328125,
      "learning_rate": 3.471124047689843e-05,
      "loss": 0.7111,
      "step": 3479
    },
    {
      "epoch": 1.8686956813529747,
      "grad_norm": 0.67578125,
      "learning_rate": 3.470345797869118e-05,
      "loss": 0.8305,
      "step": 3480
    },
    {
      "epoch": 1.8692325760880508,
      "grad_norm": 0.73046875,
      "learning_rate": 3.46956743731965e-05,
      "loss": 0.9186,
      "step": 3481
    },
    {
      "epoch": 1.8697694708231267,
      "grad_norm": 0.953125,
      "learning_rate": 3.4687889661302576e-05,
      "loss": 0.8437,
      "step": 3482
    },
    {
      "epoch": 1.8703063655582026,
      "grad_norm": 0.67578125,
      "learning_rate": 3.468010384389777e-05,
      "loss": 0.6737,
      "step": 3483
    },
    {
      "epoch": 1.8708432602932787,
      "grad_norm": 0.77734375,
      "learning_rate": 3.4672316921870504e-05,
      "loss": 0.8664,
      "step": 3484
    },
    {
      "epoch": 1.8713801550283549,
      "grad_norm": 0.76953125,
      "learning_rate": 3.46645288961094e-05,
      "loss": 0.8586,
      "step": 3485
    },
    {
      "epoch": 1.8719170497634308,
      "grad_norm": 0.859375,
      "learning_rate": 3.4656739767503163e-05,
      "loss": 1.059,
      "step": 3486
    },
    {
      "epoch": 1.8724539444985067,
      "grad_norm": 0.6875,
      "learning_rate": 3.464894953694061e-05,
      "loss": 0.7063,
      "step": 3487
    },
    {
      "epoch": 1.8729908392335828,
      "grad_norm": 0.640625,
      "learning_rate": 3.464115820531074e-05,
      "loss": 0.7961,
      "step": 3488
    },
    {
      "epoch": 1.873527733968659,
      "grad_norm": 0.71484375,
      "learning_rate": 3.463336577350261e-05,
      "loss": 0.8118,
      "step": 3489
    },
    {
      "epoch": 1.8740646287037348,
      "grad_norm": 0.83984375,
      "learning_rate": 3.462557224240545e-05,
      "loss": 0.8889,
      "step": 3490
    },
    {
      "epoch": 1.8746015234388107,
      "grad_norm": 0.6484375,
      "learning_rate": 3.46177776129086e-05,
      "loss": 0.8418,
      "step": 3491
    },
    {
      "epoch": 1.8751384181738868,
      "grad_norm": 0.7578125,
      "learning_rate": 3.4609981885901524e-05,
      "loss": 1.0085,
      "step": 3492
    },
    {
      "epoch": 1.875675312908963,
      "grad_norm": 0.81640625,
      "learning_rate": 3.4602185062273826e-05,
      "loss": 1.0661,
      "step": 3493
    },
    {
      "epoch": 1.8762122076440388,
      "grad_norm": 0.828125,
      "learning_rate": 3.459438714291519e-05,
      "loss": 1.0243,
      "step": 3494
    },
    {
      "epoch": 1.8767491023791147,
      "grad_norm": 0.65625,
      "learning_rate": 3.458658812871549e-05,
      "loss": 0.9302,
      "step": 3495
    },
    {
      "epoch": 1.8772859971141909,
      "grad_norm": 0.80859375,
      "learning_rate": 3.457878802056469e-05,
      "loss": 0.8507,
      "step": 3496
    },
    {
      "epoch": 1.8778228918492668,
      "grad_norm": 0.6953125,
      "learning_rate": 3.457098681935286e-05,
      "loss": 0.7617,
      "step": 3497
    },
    {
      "epoch": 1.8783597865843427,
      "grad_norm": 0.6484375,
      "learning_rate": 3.456318452597022e-05,
      "loss": 0.8152,
      "step": 3498
    },
    {
      "epoch": 1.8788966813194188,
      "grad_norm": 0.6796875,
      "learning_rate": 3.455538114130713e-05,
      "loss": 0.8631,
      "step": 3499
    },
    {
      "epoch": 1.879433576054495,
      "grad_norm": 0.703125,
      "learning_rate": 3.454757666625404e-05,
      "loss": 0.9013,
      "step": 3500
    },
    {
      "epoch": 1.879433576054495,
      "eval_loss": 0.9068979024887085,
      "eval_runtime": 47.2179,
      "eval_samples_per_second": 20.374,
      "eval_steps_per_second": 20.374,
      "step": 3500
    },
    {
      "epoch": 1.8799704707895708,
      "grad_norm": 0.6484375,
      "learning_rate": 3.4539771101701537e-05,
      "loss": 0.8096,
      "step": 3501
    },
    {
      "epoch": 1.8805073655246467,
      "grad_norm": 0.76171875,
      "learning_rate": 3.453196444854034e-05,
      "loss": 1.0139,
      "step": 3502
    },
    {
      "epoch": 1.8810442602597228,
      "grad_norm": 0.65234375,
      "learning_rate": 3.4524156707661296e-05,
      "loss": 0.7859,
      "step": 3503
    },
    {
      "epoch": 1.881581154994799,
      "grad_norm": 0.62890625,
      "learning_rate": 3.451634787995535e-05,
      "loss": 0.7599,
      "step": 3504
    },
    {
      "epoch": 1.8821180497298748,
      "grad_norm": 0.7734375,
      "learning_rate": 3.45085379663136e-05,
      "loss": 0.9438,
      "step": 3505
    },
    {
      "epoch": 1.8826549444649507,
      "grad_norm": 0.6796875,
      "learning_rate": 3.450072696762725e-05,
      "loss": 0.9305,
      "step": 3506
    },
    {
      "epoch": 1.8831918392000269,
      "grad_norm": 0.75390625,
      "learning_rate": 3.449291488478763e-05,
      "loss": 1.0345,
      "step": 3507
    },
    {
      "epoch": 1.883728733935103,
      "grad_norm": 0.76953125,
      "learning_rate": 3.448510171868622e-05,
      "loss": 0.9365,
      "step": 3508
    },
    {
      "epoch": 1.8842656286701789,
      "grad_norm": 0.671875,
      "learning_rate": 3.447728747021457e-05,
      "loss": 0.9009,
      "step": 3509
    },
    {
      "epoch": 1.8848025234052548,
      "grad_norm": 0.6640625,
      "learning_rate": 3.44694721402644e-05,
      "loss": 0.9301,
      "step": 3510
    },
    {
      "epoch": 1.885339418140331,
      "grad_norm": 0.7109375,
      "learning_rate": 3.446165572972755e-05,
      "loss": 0.9063,
      "step": 3511
    },
    {
      "epoch": 1.885876312875407,
      "grad_norm": 0.6875,
      "learning_rate": 3.4453838239495946e-05,
      "loss": 0.7689,
      "step": 3512
    },
    {
      "epoch": 1.886413207610483,
      "grad_norm": 0.64453125,
      "learning_rate": 3.444601967046168e-05,
      "loss": 0.7711,
      "step": 3513
    },
    {
      "epoch": 1.8869501023455588,
      "grad_norm": 0.78125,
      "learning_rate": 3.443820002351694e-05,
      "loss": 0.85,
      "step": 3514
    },
    {
      "epoch": 1.887486997080635,
      "grad_norm": 0.9609375,
      "learning_rate": 3.443037929955406e-05,
      "loss": 0.9214,
      "step": 3515
    },
    {
      "epoch": 1.8880238918157108,
      "grad_norm": 0.69140625,
      "learning_rate": 3.442255749946548e-05,
      "loss": 0.7868,
      "step": 3516
    },
    {
      "epoch": 1.8885607865507867,
      "grad_norm": 0.734375,
      "learning_rate": 3.441473462414375e-05,
      "loss": 0.8949,
      "step": 3517
    },
    {
      "epoch": 1.8890976812858629,
      "grad_norm": 0.66015625,
      "learning_rate": 3.440691067448159e-05,
      "loss": 0.7483,
      "step": 3518
    },
    {
      "epoch": 1.889634576020939,
      "grad_norm": 0.671875,
      "learning_rate": 3.439908565137178e-05,
      "loss": 0.9311,
      "step": 3519
    },
    {
      "epoch": 1.8901714707560149,
      "grad_norm": 0.96484375,
      "learning_rate": 3.439125955570725e-05,
      "loss": 1.1153,
      "step": 3520
    },
    {
      "epoch": 1.8907083654910908,
      "grad_norm": 0.67578125,
      "learning_rate": 3.438343238838111e-05,
      "loss": 0.8369,
      "step": 3521
    },
    {
      "epoch": 1.891245260226167,
      "grad_norm": 0.71484375,
      "learning_rate": 3.437560415028648e-05,
      "loss": 1.0822,
      "step": 3522
    },
    {
      "epoch": 1.891782154961243,
      "grad_norm": 0.7890625,
      "learning_rate": 3.4367774842316683e-05,
      "loss": 0.8995,
      "step": 3523
    },
    {
      "epoch": 1.892319049696319,
      "grad_norm": 0.59375,
      "learning_rate": 3.4359944465365156e-05,
      "loss": 0.8483,
      "step": 3524
    },
    {
      "epoch": 1.8928559444313948,
      "grad_norm": 0.8515625,
      "learning_rate": 3.4352113020325424e-05,
      "loss": 0.9264,
      "step": 3525
    },
    {
      "epoch": 1.893392839166471,
      "grad_norm": 0.58984375,
      "learning_rate": 3.434428050809116e-05,
      "loss": 0.7832,
      "step": 3526
    },
    {
      "epoch": 1.893929733901547,
      "grad_norm": 0.77734375,
      "learning_rate": 3.433644692955616e-05,
      "loss": 0.9841,
      "step": 3527
    },
    {
      "epoch": 1.894466628636623,
      "grad_norm": 0.65234375,
      "learning_rate": 3.432861228561432e-05,
      "loss": 0.9326,
      "step": 3528
    },
    {
      "epoch": 1.8950035233716989,
      "grad_norm": 0.73828125,
      "learning_rate": 3.432077657715969e-05,
      "loss": 1.0263,
      "step": 3529
    },
    {
      "epoch": 1.895540418106775,
      "grad_norm": 1.0078125,
      "learning_rate": 3.4312939805086413e-05,
      "loss": 0.8403,
      "step": 3530
    },
    {
      "epoch": 1.896077312841851,
      "grad_norm": 0.69140625,
      "learning_rate": 3.4305101970288764e-05,
      "loss": 0.9852,
      "step": 3531
    },
    {
      "epoch": 1.896614207576927,
      "grad_norm": 0.66796875,
      "learning_rate": 3.4297263073661136e-05,
      "loss": 1.024,
      "step": 3532
    },
    {
      "epoch": 1.897151102312003,
      "grad_norm": 0.80078125,
      "learning_rate": 3.4289423116098056e-05,
      "loss": 0.9328,
      "step": 3533
    },
    {
      "epoch": 1.897687997047079,
      "grad_norm": 0.65234375,
      "learning_rate": 3.428158209849416e-05,
      "loss": 0.9998,
      "step": 3534
    },
    {
      "epoch": 1.898224891782155,
      "grad_norm": 0.70703125,
      "learning_rate": 3.42737400217442e-05,
      "loss": 0.9602,
      "step": 3535
    },
    {
      "epoch": 1.8987617865172308,
      "grad_norm": 0.76171875,
      "learning_rate": 3.4265896886743054e-05,
      "loss": 1.1247,
      "step": 3536
    },
    {
      "epoch": 1.899298681252307,
      "grad_norm": 0.6640625,
      "learning_rate": 3.425805269438574e-05,
      "loss": 0.852,
      "step": 3537
    },
    {
      "epoch": 1.899835575987383,
      "grad_norm": 0.671875,
      "learning_rate": 3.4250207445567376e-05,
      "loss": 0.8087,
      "step": 3538
    },
    {
      "epoch": 1.900372470722459,
      "grad_norm": 0.69921875,
      "learning_rate": 3.424236114118319e-05,
      "loss": 0.787,
      "step": 3539
    },
    {
      "epoch": 1.9009093654575349,
      "grad_norm": 0.66796875,
      "learning_rate": 3.423451378212855e-05,
      "loss": 0.8916,
      "step": 3540
    },
    {
      "epoch": 1.901446260192611,
      "grad_norm": 0.66796875,
      "learning_rate": 3.4226665369298945e-05,
      "loss": 0.9666,
      "step": 3541
    },
    {
      "epoch": 1.901983154927687,
      "grad_norm": 0.625,
      "learning_rate": 3.4218815903589974e-05,
      "loss": 0.8278,
      "step": 3542
    },
    {
      "epoch": 1.902520049662763,
      "grad_norm": 0.69921875,
      "learning_rate": 3.421096538589736e-05,
      "loss": 0.7801,
      "step": 3543
    },
    {
      "epoch": 1.903056944397839,
      "grad_norm": 0.6875,
      "learning_rate": 3.4203113817116957e-05,
      "loss": 0.8558,
      "step": 3544
    },
    {
      "epoch": 1.903593839132915,
      "grad_norm": 0.73046875,
      "learning_rate": 3.4195261198144704e-05,
      "loss": 0.7254,
      "step": 3545
    },
    {
      "epoch": 1.9041307338679911,
      "grad_norm": 0.8671875,
      "learning_rate": 3.418740752987672e-05,
      "loss": 0.9899,
      "step": 3546
    },
    {
      "epoch": 1.904667628603067,
      "grad_norm": 0.72265625,
      "learning_rate": 3.417955281320917e-05,
      "loss": 1.0206,
      "step": 3547
    },
    {
      "epoch": 1.905204523338143,
      "grad_norm": 0.65625,
      "learning_rate": 3.41716970490384e-05,
      "loss": 0.92,
      "step": 3548
    },
    {
      "epoch": 1.905741418073219,
      "grad_norm": 0.6796875,
      "learning_rate": 3.416384023826085e-05,
      "loss": 0.7991,
      "step": 3549
    },
    {
      "epoch": 1.9062783128082952,
      "grad_norm": 0.85546875,
      "learning_rate": 3.415598238177307e-05,
      "loss": 0.8539,
      "step": 3550
    },
    {
      "epoch": 1.9062783128082952,
      "eval_loss": 0.9063857793807983,
      "eval_runtime": 46.3907,
      "eval_samples_per_second": 20.737,
      "eval_steps_per_second": 20.737,
      "step": 3550
    },
    {
      "epoch": 1.906815207543371,
      "grad_norm": 0.66015625,
      "learning_rate": 3.4148123480471764e-05,
      "loss": 0.9112,
      "step": 3551
    },
    {
      "epoch": 1.907352102278447,
      "grad_norm": 0.88671875,
      "learning_rate": 3.41402635352537e-05,
      "loss": 0.8875,
      "step": 3552
    },
    {
      "epoch": 1.907888997013523,
      "grad_norm": 0.66796875,
      "learning_rate": 3.413240254701582e-05,
      "loss": 0.8938,
      "step": 3553
    },
    {
      "epoch": 1.908425891748599,
      "grad_norm": 0.859375,
      "learning_rate": 3.412454051665516e-05,
      "loss": 0.9102,
      "step": 3554
    },
    {
      "epoch": 1.908962786483675,
      "grad_norm": 0.65234375,
      "learning_rate": 3.411667744506887e-05,
      "loss": 0.8858,
      "step": 3555
    },
    {
      "epoch": 1.909499681218751,
      "grad_norm": 0.6875,
      "learning_rate": 3.4108813333154236e-05,
      "loss": 0.8897,
      "step": 3556
    },
    {
      "epoch": 1.9100365759538271,
      "grad_norm": 0.6953125,
      "learning_rate": 3.4100948181808645e-05,
      "loss": 0.7843,
      "step": 3557
    },
    {
      "epoch": 1.910573470688903,
      "grad_norm": 0.734375,
      "learning_rate": 3.409308199192961e-05,
      "loss": 0.8233,
      "step": 3558
    },
    {
      "epoch": 1.911110365423979,
      "grad_norm": 0.7734375,
      "learning_rate": 3.408521476441477e-05,
      "loss": 0.9662,
      "step": 3559
    },
    {
      "epoch": 1.911647260159055,
      "grad_norm": 0.88671875,
      "learning_rate": 3.407734650016187e-05,
      "loss": 0.909,
      "step": 3560
    },
    {
      "epoch": 1.9121841548941312,
      "grad_norm": 0.94140625,
      "learning_rate": 3.4069477200068776e-05,
      "loss": 0.9564,
      "step": 3561
    },
    {
      "epoch": 1.912721049629207,
      "grad_norm": 0.578125,
      "learning_rate": 3.406160686503348e-05,
      "loss": 0.7775,
      "step": 3562
    },
    {
      "epoch": 1.913257944364283,
      "grad_norm": 0.66796875,
      "learning_rate": 3.405373549595408e-05,
      "loss": 0.9723,
      "step": 3563
    },
    {
      "epoch": 1.913794839099359,
      "grad_norm": 0.67578125,
      "learning_rate": 3.40458630937288e-05,
      "loss": 0.916,
      "step": 3564
    },
    {
      "epoch": 1.9143317338344352,
      "grad_norm": 0.71484375,
      "learning_rate": 3.4037989659256e-05,
      "loss": 0.752,
      "step": 3565
    },
    {
      "epoch": 1.9148686285695111,
      "grad_norm": 0.62109375,
      "learning_rate": 3.403011519343411e-05,
      "loss": 0.7295,
      "step": 3566
    },
    {
      "epoch": 1.915405523304587,
      "grad_norm": 0.72265625,
      "learning_rate": 3.4022239697161724e-05,
      "loss": 0.8212,
      "step": 3567
    },
    {
      "epoch": 1.9159424180396631,
      "grad_norm": 0.68359375,
      "learning_rate": 3.401436317133753e-05,
      "loss": 0.7814,
      "step": 3568
    },
    {
      "epoch": 1.9164793127747393,
      "grad_norm": 0.71484375,
      "learning_rate": 3.400648561686034e-05,
      "loss": 0.932,
      "step": 3569
    },
    {
      "epoch": 1.9170162075098152,
      "grad_norm": 0.71875,
      "learning_rate": 3.399860703462908e-05,
      "loss": 1.0507,
      "step": 3570
    },
    {
      "epoch": 1.917553102244891,
      "grad_norm": 0.72265625,
      "learning_rate": 3.3990727425542814e-05,
      "loss": 0.8958,
      "step": 3571
    },
    {
      "epoch": 1.9180899969799672,
      "grad_norm": 0.66796875,
      "learning_rate": 3.398284679050067e-05,
      "loss": 0.7969,
      "step": 3572
    },
    {
      "epoch": 1.918626891715043,
      "grad_norm": 0.77734375,
      "learning_rate": 3.397496513040196e-05,
      "loss": 0.9821,
      "step": 3573
    },
    {
      "epoch": 1.919163786450119,
      "grad_norm": 0.8046875,
      "learning_rate": 3.396708244614607e-05,
      "loss": 1.1287,
      "step": 3574
    },
    {
      "epoch": 1.919700681185195,
      "grad_norm": 0.8203125,
      "learning_rate": 3.39591987386325e-05,
      "loss": 0.8573,
      "step": 3575
    },
    {
      "epoch": 1.9202375759202712,
      "grad_norm": 0.69140625,
      "learning_rate": 3.395131400876092e-05,
      "loss": 0.7293,
      "step": 3576
    },
    {
      "epoch": 1.9207744706553471,
      "grad_norm": 0.75390625,
      "learning_rate": 3.3943428257431034e-05,
      "loss": 0.8483,
      "step": 3577
    },
    {
      "epoch": 1.921311365390423,
      "grad_norm": 0.65625,
      "learning_rate": 3.393554148554273e-05,
      "loss": 0.8788,
      "step": 3578
    },
    {
      "epoch": 1.9218482601254991,
      "grad_norm": 0.75,
      "learning_rate": 3.392765369399599e-05,
      "loss": 0.9468,
      "step": 3579
    },
    {
      "epoch": 1.9223851548605753,
      "grad_norm": 0.78515625,
      "learning_rate": 3.391976488369089e-05,
      "loss": 0.7401,
      "step": 3580
    },
    {
      "epoch": 1.9229220495956512,
      "grad_norm": 0.75390625,
      "learning_rate": 3.3911875055527666e-05,
      "loss": 0.8656,
      "step": 3581
    },
    {
      "epoch": 1.923458944330727,
      "grad_norm": 0.6953125,
      "learning_rate": 3.390398421040664e-05,
      "loss": 0.9307,
      "step": 3582
    },
    {
      "epoch": 1.9239958390658032,
      "grad_norm": 0.65625,
      "learning_rate": 3.389609234922824e-05,
      "loss": 0.9472,
      "step": 3583
    },
    {
      "epoch": 1.9245327338008793,
      "grad_norm": 0.609375,
      "learning_rate": 3.388819947289307e-05,
      "loss": 0.7706,
      "step": 3584
    },
    {
      "epoch": 1.9250696285359552,
      "grad_norm": 0.578125,
      "learning_rate": 3.3880305582301766e-05,
      "loss": 0.7794,
      "step": 3585
    },
    {
      "epoch": 1.925606523271031,
      "grad_norm": 0.8125,
      "learning_rate": 3.387241067835513e-05,
      "loss": 1.0654,
      "step": 3586
    },
    {
      "epoch": 1.9261434180061072,
      "grad_norm": 0.671875,
      "learning_rate": 3.386451476195408e-05,
      "loss": 0.8464,
      "step": 3587
    },
    {
      "epoch": 1.9266803127411833,
      "grad_norm": 0.6796875,
      "learning_rate": 3.3856617833999635e-05,
      "loss": 0.7479,
      "step": 3588
    },
    {
      "epoch": 1.927217207476259,
      "grad_norm": 0.76171875,
      "learning_rate": 3.384871989539293e-05,
      "loss": 0.7932,
      "step": 3589
    },
    {
      "epoch": 1.9277541022113351,
      "grad_norm": 0.65625,
      "learning_rate": 3.3840820947035234e-05,
      "loss": 0.7046,
      "step": 3590
    },
    {
      "epoch": 1.9282909969464113,
      "grad_norm": 0.75,
      "learning_rate": 3.38329209898279e-05,
      "loss": 0.9476,
      "step": 3591
    },
    {
      "epoch": 1.9288278916814872,
      "grad_norm": 0.6875,
      "learning_rate": 3.3825020024672434e-05,
      "loss": 0.8789,
      "step": 3592
    },
    {
      "epoch": 1.929364786416563,
      "grad_norm": 0.78125,
      "learning_rate": 3.381711805247041e-05,
      "loss": 0.9042,
      "step": 3593
    },
    {
      "epoch": 1.9299016811516392,
      "grad_norm": 0.6015625,
      "learning_rate": 3.380921507412356e-05,
      "loss": 0.7703,
      "step": 3594
    },
    {
      "epoch": 1.9304385758867153,
      "grad_norm": 0.68359375,
      "learning_rate": 3.380131109053371e-05,
      "loss": 0.9292,
      "step": 3595
    },
    {
      "epoch": 1.9309754706217912,
      "grad_norm": 0.640625,
      "learning_rate": 3.379340610260281e-05,
      "loss": 0.9689,
      "step": 3596
    },
    {
      "epoch": 1.931512365356867,
      "grad_norm": 0.625,
      "learning_rate": 3.378550011123291e-05,
      "loss": 0.7642,
      "step": 3597
    },
    {
      "epoch": 1.9320492600919432,
      "grad_norm": 0.73828125,
      "learning_rate": 3.377759311732619e-05,
      "loss": 0.9143,
      "step": 3598
    },
    {
      "epoch": 1.9325861548270193,
      "grad_norm": 0.71484375,
      "learning_rate": 3.3769685121784935e-05,
      "loss": 0.8401,
      "step": 3599
    },
    {
      "epoch": 1.9331230495620952,
      "grad_norm": 0.703125,
      "learning_rate": 3.376177612551155e-05,
      "loss": 0.7346,
      "step": 3600
    },
    {
      "epoch": 1.9331230495620952,
      "eval_loss": 0.9065521955490112,
      "eval_runtime": 46.0053,
      "eval_samples_per_second": 20.911,
      "eval_steps_per_second": 20.911,
      "step": 3600
    },
    {
      "epoch": 1.9336599442971711,
      "grad_norm": 0.6953125,
      "learning_rate": 3.375386612940855e-05,
      "loss": 1.0007,
      "step": 3601
    },
    {
      "epoch": 1.9341968390322473,
      "grad_norm": 0.66796875,
      "learning_rate": 3.374595513437856e-05,
      "loss": 0.8313,
      "step": 3602
    },
    {
      "epoch": 1.9347337337673234,
      "grad_norm": 0.640625,
      "learning_rate": 3.3738043141324336e-05,
      "loss": 0.799,
      "step": 3603
    },
    {
      "epoch": 1.9352706285023993,
      "grad_norm": 0.6640625,
      "learning_rate": 3.373013015114873e-05,
      "loss": 0.8217,
      "step": 3604
    },
    {
      "epoch": 1.9358075232374752,
      "grad_norm": 0.75390625,
      "learning_rate": 3.372221616475472e-05,
      "loss": 0.8775,
      "step": 3605
    },
    {
      "epoch": 1.9363444179725513,
      "grad_norm": 0.69921875,
      "learning_rate": 3.3714301183045385e-05,
      "loss": 0.8575,
      "step": 3606
    },
    {
      "epoch": 1.9368813127076274,
      "grad_norm": 0.7265625,
      "learning_rate": 3.370638520692393e-05,
      "loss": 1.1144,
      "step": 3607
    },
    {
      "epoch": 1.937418207442703,
      "grad_norm": 0.61328125,
      "learning_rate": 3.369846823729366e-05,
      "loss": 0.7982,
      "step": 3608
    },
    {
      "epoch": 1.9379551021777792,
      "grad_norm": 0.76953125,
      "learning_rate": 3.369055027505801e-05,
      "loss": 0.8817,
      "step": 3609
    },
    {
      "epoch": 1.9384919969128553,
      "grad_norm": 0.73046875,
      "learning_rate": 3.3682631321120504e-05,
      "loss": 0.749,
      "step": 3610
    },
    {
      "epoch": 1.9390288916479312,
      "grad_norm": 0.62890625,
      "learning_rate": 3.3674711376384824e-05,
      "loss": 0.7168,
      "step": 3611
    },
    {
      "epoch": 1.9395657863830071,
      "grad_norm": 0.6484375,
      "learning_rate": 3.366679044175471e-05,
      "loss": 0.73,
      "step": 3612
    },
    {
      "epoch": 1.9401026811180833,
      "grad_norm": 0.60546875,
      "learning_rate": 3.365886851813406e-05,
      "loss": 0.819,
      "step": 3613
    },
    {
      "epoch": 1.9406395758531594,
      "grad_norm": 0.83984375,
      "learning_rate": 3.365094560642685e-05,
      "loss": 0.8556,
      "step": 3614
    },
    {
      "epoch": 1.9411764705882353,
      "grad_norm": 0.875,
      "learning_rate": 3.36430217075372e-05,
      "loss": 0.8645,
      "step": 3615
    },
    {
      "epoch": 1.9417133653233112,
      "grad_norm": 0.71484375,
      "learning_rate": 3.363509682236931e-05,
      "loss": 0.8661,
      "step": 3616
    },
    {
      "epoch": 1.9422502600583873,
      "grad_norm": 0.62890625,
      "learning_rate": 3.362717095182751e-05,
      "loss": 0.8436,
      "step": 3617
    },
    {
      "epoch": 1.9427871547934634,
      "grad_norm": 0.70703125,
      "learning_rate": 3.361924409681626e-05,
      "loss": 0.9804,
      "step": 3618
    },
    {
      "epoch": 1.9433240495285393,
      "grad_norm": 0.6171875,
      "learning_rate": 3.36113162582401e-05,
      "loss": 0.8653,
      "step": 3619
    },
    {
      "epoch": 1.9438609442636152,
      "grad_norm": 0.80859375,
      "learning_rate": 3.360338743700371e-05,
      "loss": 0.9412,
      "step": 3620
    },
    {
      "epoch": 1.9443978389986913,
      "grad_norm": 0.69921875,
      "learning_rate": 3.359545763401185e-05,
      "loss": 0.6573,
      "step": 3621
    },
    {
      "epoch": 1.9449347337337675,
      "grad_norm": 0.66015625,
      "learning_rate": 3.358752685016943e-05,
      "loss": 0.9258,
      "step": 3622
    },
    {
      "epoch": 1.9454716284688434,
      "grad_norm": 0.6953125,
      "learning_rate": 3.357959508638144e-05,
      "loss": 0.8442,
      "step": 3623
    },
    {
      "epoch": 1.9460085232039193,
      "grad_norm": 0.6015625,
      "learning_rate": 3.3571662343553e-05,
      "loss": 0.7089,
      "step": 3624
    },
    {
      "epoch": 1.9465454179389954,
      "grad_norm": 0.66015625,
      "learning_rate": 3.356372862258934e-05,
      "loss": 0.8417,
      "step": 3625
    },
    {
      "epoch": 1.9470823126740715,
      "grad_norm": 0.65625,
      "learning_rate": 3.355579392439578e-05,
      "loss": 0.894,
      "step": 3626
    },
    {
      "epoch": 1.9476192074091472,
      "grad_norm": 0.71875,
      "learning_rate": 3.354785824987779e-05,
      "loss": 0.9974,
      "step": 3627
    },
    {
      "epoch": 1.9481561021442233,
      "grad_norm": 0.71875,
      "learning_rate": 3.3539921599940925e-05,
      "loss": 0.7023,
      "step": 3628
    },
    {
      "epoch": 1.9486929968792994,
      "grad_norm": 0.58984375,
      "learning_rate": 3.353198397549086e-05,
      "loss": 0.7872,
      "step": 3629
    },
    {
      "epoch": 1.9492298916143753,
      "grad_norm": 0.78515625,
      "learning_rate": 3.352404537743337e-05,
      "loss": 0.9705,
      "step": 3630
    },
    {
      "epoch": 1.9497667863494512,
      "grad_norm": 0.75,
      "learning_rate": 3.351610580667436e-05,
      "loss": 1.1664,
      "step": 3631
    },
    {
      "epoch": 1.9503036810845273,
      "grad_norm": 0.78125,
      "learning_rate": 3.350816526411982e-05,
      "loss": 0.9449,
      "step": 3632
    },
    {
      "epoch": 1.9508405758196035,
      "grad_norm": 0.703125,
      "learning_rate": 3.350022375067588e-05,
      "loss": 0.924,
      "step": 3633
    },
    {
      "epoch": 1.9513774705546794,
      "grad_norm": 0.63671875,
      "learning_rate": 3.3492281267248774e-05,
      "loss": 0.9257,
      "step": 3634
    },
    {
      "epoch": 1.9519143652897553,
      "grad_norm": 0.6328125,
      "learning_rate": 3.348433781474481e-05,
      "loss": 0.8889,
      "step": 3635
    },
    {
      "epoch": 1.9524512600248314,
      "grad_norm": 0.81640625,
      "learning_rate": 3.347639339407047e-05,
      "loss": 0.8841,
      "step": 3636
    },
    {
      "epoch": 1.9529881547599075,
      "grad_norm": 0.6875,
      "learning_rate": 3.346844800613229e-05,
      "loss": 0.9369,
      "step": 3637
    },
    {
      "epoch": 1.9535250494949834,
      "grad_norm": 0.7265625,
      "learning_rate": 3.346050165183695e-05,
      "loss": 0.9178,
      "step": 3638
    },
    {
      "epoch": 1.9540619442300593,
      "grad_norm": 0.6640625,
      "learning_rate": 3.3452554332091234e-05,
      "loss": 0.8964,
      "step": 3639
    },
    {
      "epoch": 1.9545988389651354,
      "grad_norm": 0.69140625,
      "learning_rate": 3.344460604780202e-05,
      "loss": 0.8634,
      "step": 3640
    },
    {
      "epoch": 1.9551357337002115,
      "grad_norm": 0.71875,
      "learning_rate": 3.343665679987631e-05,
      "loss": 0.7428,
      "step": 3641
    },
    {
      "epoch": 1.9556726284352874,
      "grad_norm": 0.68359375,
      "learning_rate": 3.342870658922123e-05,
      "loss": 0.9646,
      "step": 3642
    },
    {
      "epoch": 1.9562095231703633,
      "grad_norm": 0.765625,
      "learning_rate": 3.342075541674398e-05,
      "loss": 0.996,
      "step": 3643
    },
    {
      "epoch": 1.9567464179054395,
      "grad_norm": 0.65234375,
      "learning_rate": 3.3412803283351895e-05,
      "loss": 0.8717,
      "step": 3644
    },
    {
      "epoch": 1.9572833126405156,
      "grad_norm": 0.6953125,
      "learning_rate": 3.3404850189952416e-05,
      "loss": 0.8542,
      "step": 3645
    },
    {
      "epoch": 1.9578202073755913,
      "grad_norm": 0.765625,
      "learning_rate": 3.339689613745309e-05,
      "loss": 0.8331,
      "step": 3646
    },
    {
      "epoch": 1.9583571021106674,
      "grad_norm": 0.76171875,
      "learning_rate": 3.3388941126761584e-05,
      "loss": 0.9255,
      "step": 3647
    },
    {
      "epoch": 1.9588939968457435,
      "grad_norm": 0.75390625,
      "learning_rate": 3.3380985158785654e-05,
      "loss": 0.9237,
      "step": 3648
    },
    {
      "epoch": 1.9594308915808194,
      "grad_norm": 0.78515625,
      "learning_rate": 3.337302823443318e-05,
      "loss": 1.1332,
      "step": 3649
    },
    {
      "epoch": 1.9599677863158953,
      "grad_norm": 0.90625,
      "learning_rate": 3.336507035461215e-05,
      "loss": 0.9698,
      "step": 3650
    },
    {
      "epoch": 1.9599677863158953,
      "eval_loss": 0.9060580730438232,
      "eval_runtime": 46.6383,
      "eval_samples_per_second": 20.627,
      "eval_steps_per_second": 20.627,
      "step": 3650
    },
    {
      "epoch": 1.9605046810509714,
      "grad_norm": 0.7734375,
      "learning_rate": 3.335711152023066e-05,
      "loss": 0.9829,
      "step": 3651
    },
    {
      "epoch": 1.9610415757860475,
      "grad_norm": 0.703125,
      "learning_rate": 3.3349151732196904e-05,
      "loss": 0.9752,
      "step": 3652
    },
    {
      "epoch": 1.9615784705211234,
      "grad_norm": 0.83203125,
      "learning_rate": 3.3341190991419204e-05,
      "loss": 1.0297,
      "step": 3653
    },
    {
      "epoch": 1.9621153652561993,
      "grad_norm": 0.66015625,
      "learning_rate": 3.333322929880598e-05,
      "loss": 0.9164,
      "step": 3654
    },
    {
      "epoch": 1.9626522599912755,
      "grad_norm": 0.73046875,
      "learning_rate": 3.332526665526577e-05,
      "loss": 0.8236,
      "step": 3655
    },
    {
      "epoch": 1.9631891547263516,
      "grad_norm": 0.66796875,
      "learning_rate": 3.331730306170719e-05,
      "loss": 0.8501,
      "step": 3656
    },
    {
      "epoch": 1.9637260494614275,
      "grad_norm": 0.70703125,
      "learning_rate": 3.3309338519039015e-05,
      "loss": 0.9423,
      "step": 3657
    },
    {
      "epoch": 1.9642629441965034,
      "grad_norm": 1.28125,
      "learning_rate": 3.330137302817007e-05,
      "loss": 1.0223,
      "step": 3658
    },
    {
      "epoch": 1.9647998389315795,
      "grad_norm": 0.63671875,
      "learning_rate": 3.329340659000934e-05,
      "loss": 0.8156,
      "step": 3659
    },
    {
      "epoch": 1.9653367336666556,
      "grad_norm": 0.75,
      "learning_rate": 3.32854392054659e-05,
      "loss": 0.8239,
      "step": 3660
    },
    {
      "epoch": 1.9658736284017315,
      "grad_norm": 0.59375,
      "learning_rate": 3.327747087544891e-05,
      "loss": 0.6689,
      "step": 3661
    },
    {
      "epoch": 1.9664105231368074,
      "grad_norm": 0.890625,
      "learning_rate": 3.326950160086766e-05,
      "loss": 1.016,
      "step": 3662
    },
    {
      "epoch": 1.9669474178718835,
      "grad_norm": 0.640625,
      "learning_rate": 3.3261531382631574e-05,
      "loss": 0.7536,
      "step": 3663
    },
    {
      "epoch": 1.9674843126069597,
      "grad_norm": 0.91015625,
      "learning_rate": 3.325356022165012e-05,
      "loss": 1.2018,
      "step": 3664
    },
    {
      "epoch": 1.9680212073420353,
      "grad_norm": 0.66015625,
      "learning_rate": 3.324558811883292e-05,
      "loss": 0.9699,
      "step": 3665
    },
    {
      "epoch": 1.9685581020771115,
      "grad_norm": 0.80859375,
      "learning_rate": 3.3237615075089704e-05,
      "loss": 0.997,
      "step": 3666
    },
    {
      "epoch": 1.9690949968121876,
      "grad_norm": 0.69921875,
      "learning_rate": 3.3229641091330274e-05,
      "loss": 0.8508,
      "step": 3667
    },
    {
      "epoch": 1.9696318915472635,
      "grad_norm": 0.8203125,
      "learning_rate": 3.322166616846458e-05,
      "loss": 0.9971,
      "step": 3668
    },
    {
      "epoch": 1.9701687862823394,
      "grad_norm": 0.82421875,
      "learning_rate": 3.3213690307402666e-05,
      "loss": 0.976,
      "step": 3669
    },
    {
      "epoch": 1.9707056810174155,
      "grad_norm": 0.6484375,
      "learning_rate": 3.320571350905466e-05,
      "loss": 0.9101,
      "step": 3670
    },
    {
      "epoch": 1.9712425757524916,
      "grad_norm": 0.74609375,
      "learning_rate": 3.319773577433083e-05,
      "loss": 0.9938,
      "step": 3671
    },
    {
      "epoch": 1.9717794704875675,
      "grad_norm": 0.703125,
      "learning_rate": 3.3189757104141536e-05,
      "loss": 0.949,
      "step": 3672
    },
    {
      "epoch": 1.9723163652226434,
      "grad_norm": 0.73828125,
      "learning_rate": 3.318177749939723e-05,
      "loss": 1.0143,
      "step": 3673
    },
    {
      "epoch": 1.9728532599577195,
      "grad_norm": 0.74609375,
      "learning_rate": 3.31737969610085e-05,
      "loss": 0.8021,
      "step": 3674
    },
    {
      "epoch": 1.9733901546927957,
      "grad_norm": 0.6171875,
      "learning_rate": 3.316581548988603e-05,
      "loss": 0.8257,
      "step": 3675
    },
    {
      "epoch": 1.9739270494278716,
      "grad_norm": 0.640625,
      "learning_rate": 3.31578330869406e-05,
      "loss": 0.7043,
      "step": 3676
    },
    {
      "epoch": 1.9744639441629475,
      "grad_norm": 0.859375,
      "learning_rate": 3.314984975308311e-05,
      "loss": 0.8382,
      "step": 3677
    },
    {
      "epoch": 1.9750008388980236,
      "grad_norm": 0.65234375,
      "learning_rate": 3.3141865489224546e-05,
      "loss": 0.7103,
      "step": 3678
    },
    {
      "epoch": 1.9755377336330997,
      "grad_norm": 0.85546875,
      "learning_rate": 3.313388029627601e-05,
      "loss": 0.799,
      "step": 3679
    },
    {
      "epoch": 1.9760746283681756,
      "grad_norm": 0.62109375,
      "learning_rate": 3.312589417514874e-05,
      "loss": 0.8044,
      "step": 3680
    },
    {
      "epoch": 1.9766115231032515,
      "grad_norm": 0.7578125,
      "learning_rate": 3.311790712675403e-05,
      "loss": 0.8228,
      "step": 3681
    },
    {
      "epoch": 1.9771484178383276,
      "grad_norm": 0.828125,
      "learning_rate": 3.310991915200332e-05,
      "loss": 1.1587,
      "step": 3682
    },
    {
      "epoch": 1.9776853125734037,
      "grad_norm": 0.66796875,
      "learning_rate": 3.3101930251808134e-05,
      "loss": 0.7751,
      "step": 3683
    },
    {
      "epoch": 1.9782222073084794,
      "grad_norm": 0.71484375,
      "learning_rate": 3.309394042708009e-05,
      "loss": 0.6996,
      "step": 3684
    },
    {
      "epoch": 1.9787591020435555,
      "grad_norm": 0.90234375,
      "learning_rate": 3.308594967873095e-05,
      "loss": 0.8803,
      "step": 3685
    },
    {
      "epoch": 1.9792959967786317,
      "grad_norm": 0.609375,
      "learning_rate": 3.307795800767256e-05,
      "loss": 0.9486,
      "step": 3686
    },
    {
      "epoch": 1.9798328915137076,
      "grad_norm": 0.68359375,
      "learning_rate": 3.3069965414816853e-05,
      "loss": 0.74,
      "step": 3687
    },
    {
      "epoch": 1.9803697862487835,
      "grad_norm": 0.90234375,
      "learning_rate": 3.306197190107589e-05,
      "loss": 0.8933,
      "step": 3688
    },
    {
      "epoch": 1.9809066809838596,
      "grad_norm": 0.8515625,
      "learning_rate": 3.305397746736184e-05,
      "loss": 1.0735,
      "step": 3689
    },
    {
      "epoch": 1.9814435757189357,
      "grad_norm": 0.6015625,
      "learning_rate": 3.304598211458697e-05,
      "loss": 0.8555,
      "step": 3690
    },
    {
      "epoch": 1.9819804704540116,
      "grad_norm": 0.85546875,
      "learning_rate": 3.3037985843663643e-05,
      "loss": 1.0123,
      "step": 3691
    },
    {
      "epoch": 1.9825173651890875,
      "grad_norm": 0.69921875,
      "learning_rate": 3.302998865550434e-05,
      "loss": 0.7485,
      "step": 3692
    },
    {
      "epoch": 1.9830542599241636,
      "grad_norm": 0.7109375,
      "learning_rate": 3.302199055102163e-05,
      "loss": 0.8797,
      "step": 3693
    },
    {
      "epoch": 1.9835911546592397,
      "grad_norm": 0.73828125,
      "learning_rate": 3.301399153112822e-05,
      "loss": 1.0695,
      "step": 3694
    },
    {
      "epoch": 1.9841280493943156,
      "grad_norm": 0.59765625,
      "learning_rate": 3.3005991596736884e-05,
      "loss": 0.7878,
      "step": 3695
    },
    {
      "epoch": 1.9846649441293915,
      "grad_norm": 0.578125,
      "learning_rate": 3.299799074876052e-05,
      "loss": 0.7779,
      "step": 3696
    },
    {
      "epoch": 1.9852018388644677,
      "grad_norm": 0.6875,
      "learning_rate": 3.298998898811213e-05,
      "loss": 0.7775,
      "step": 3697
    },
    {
      "epoch": 1.9857387335995438,
      "grad_norm": 0.83203125,
      "learning_rate": 3.29819863157048e-05,
      "loss": 0.9006,
      "step": 3698
    },
    {
      "epoch": 1.9862756283346197,
      "grad_norm": 0.75390625,
      "learning_rate": 3.2973982732451755e-05,
      "loss": 0.7849,
      "step": 3699
    },
    {
      "epoch": 1.9868125230696956,
      "grad_norm": 0.6640625,
      "learning_rate": 3.296597823926629e-05,
      "loss": 0.7256,
      "step": 3700
    },
    {
      "epoch": 1.9868125230696956,
      "eval_loss": 0.9062338471412659,
      "eval_runtime": 46.4266,
      "eval_samples_per_second": 20.721,
      "eval_steps_per_second": 20.721,
      "step": 3700
    },
    {
      "epoch": 1.9873494178047717,
      "grad_norm": 0.72265625,
      "learning_rate": 3.295797283706183e-05,
      "loss": 0.8203,
      "step": 3701
    },
    {
      "epoch": 1.9878863125398478,
      "grad_norm": 0.61328125,
      "learning_rate": 3.29499665267519e-05,
      "loss": 0.8417,
      "step": 3702
    },
    {
      "epoch": 1.9884232072749235,
      "grad_norm": 0.65625,
      "learning_rate": 3.2941959309250094e-05,
      "loss": 0.8639,
      "step": 3703
    },
    {
      "epoch": 1.9889601020099996,
      "grad_norm": 0.640625,
      "learning_rate": 3.293395118547017e-05,
      "loss": 0.9724,
      "step": 3704
    },
    {
      "epoch": 1.9894969967450757,
      "grad_norm": 0.89453125,
      "learning_rate": 3.292594215632593e-05,
      "loss": 1.0296,
      "step": 3705
    },
    {
      "epoch": 1.9900338914801516,
      "grad_norm": 0.765625,
      "learning_rate": 3.2917932222731325e-05,
      "loss": 0.8509,
      "step": 3706
    },
    {
      "epoch": 1.9905707862152275,
      "grad_norm": 0.65234375,
      "learning_rate": 3.290992138560038e-05,
      "loss": 0.8924,
      "step": 3707
    },
    {
      "epoch": 1.9911076809503037,
      "grad_norm": 0.84375,
      "learning_rate": 3.290190964584723e-05,
      "loss": 0.9232,
      "step": 3708
    },
    {
      "epoch": 1.9916445756853798,
      "grad_norm": 0.8203125,
      "learning_rate": 3.289389700438611e-05,
      "loss": 1.0032,
      "step": 3709
    },
    {
      "epoch": 1.9921814704204557,
      "grad_norm": 0.8125,
      "learning_rate": 3.2885883462131394e-05,
      "loss": 0.7262,
      "step": 3710
    },
    {
      "epoch": 1.9927183651555316,
      "grad_norm": 0.609375,
      "learning_rate": 3.28778690199975e-05,
      "loss": 0.9565,
      "step": 3711
    },
    {
      "epoch": 1.9932552598906077,
      "grad_norm": 0.66015625,
      "learning_rate": 3.2869853678898974e-05,
      "loss": 0.8002,
      "step": 3712
    },
    {
      "epoch": 1.9937921546256838,
      "grad_norm": 0.75390625,
      "learning_rate": 3.286183743975049e-05,
      "loss": 0.9041,
      "step": 3713
    },
    {
      "epoch": 1.9943290493607597,
      "grad_norm": 0.71875,
      "learning_rate": 3.2853820303466794e-05,
      "loss": 0.9013,
      "step": 3714
    },
    {
      "epoch": 1.9948659440958356,
      "grad_norm": 0.671875,
      "learning_rate": 3.284580227096273e-05,
      "loss": 0.8556,
      "step": 3715
    },
    {
      "epoch": 1.9954028388309117,
      "grad_norm": 0.76953125,
      "learning_rate": 3.283778334315328e-05,
      "loss": 0.8141,
      "step": 3716
    },
    {
      "epoch": 1.9959397335659879,
      "grad_norm": 0.65625,
      "learning_rate": 3.2829763520953486e-05,
      "loss": 0.7905,
      "step": 3717
    },
    {
      "epoch": 1.9964766283010638,
      "grad_norm": 0.6796875,
      "learning_rate": 3.2821742805278525e-05,
      "loss": 0.9433,
      "step": 3718
    },
    {
      "epoch": 1.9970135230361397,
      "grad_norm": 0.984375,
      "learning_rate": 3.281372119704366e-05,
      "loss": 0.9418,
      "step": 3719
    },
    {
      "epoch": 1.9975504177712158,
      "grad_norm": 0.71484375,
      "learning_rate": 3.2805698697164234e-05,
      "loss": 0.8866,
      "step": 3720
    },
    {
      "epoch": 1.998087312506292,
      "grad_norm": 0.6484375,
      "learning_rate": 3.279767530655575e-05,
      "loss": 0.7644,
      "step": 3721
    },
    {
      "epoch": 1.9986242072413676,
      "grad_norm": 0.65234375,
      "learning_rate": 3.278965102613376e-05,
      "loss": 0.7099,
      "step": 3722
    },
    {
      "epoch": 1.9991611019764437,
      "grad_norm": 0.625,
      "learning_rate": 3.278162585681394e-05,
      "loss": 0.8385,
      "step": 3723
    },
    {
      "epoch": 1.9996979967115198,
      "grad_norm": 0.82421875,
      "learning_rate": 3.2773599799512066e-05,
      "loss": 0.7835,
      "step": 3724
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.46875,
      "learning_rate": 3.2765572855144006e-05,
      "loss": 0.5844,
      "step": 3725
    },
    {
      "epoch": 2.000536894735076,
      "grad_norm": 0.61328125,
      "learning_rate": 3.2757545024625744e-05,
      "loss": 0.896,
      "step": 3726
    },
    {
      "epoch": 2.001073789470152,
      "grad_norm": 0.7578125,
      "learning_rate": 3.274951630887335e-05,
      "loss": 0.8918,
      "step": 3727
    },
    {
      "epoch": 2.001610684205228,
      "grad_norm": 0.75390625,
      "learning_rate": 3.2741486708803005e-05,
      "loss": 0.7393,
      "step": 3728
    },
    {
      "epoch": 2.002147578940304,
      "grad_norm": 0.703125,
      "learning_rate": 3.273345622533099e-05,
      "loss": 0.937,
      "step": 3729
    },
    {
      "epoch": 2.00268447367538,
      "grad_norm": 0.65625,
      "learning_rate": 3.272542485937369e-05,
      "loss": 0.7374,
      "step": 3730
    },
    {
      "epoch": 2.003221368410456,
      "grad_norm": 0.7578125,
      "learning_rate": 3.271739261184757e-05,
      "loss": 0.8095,
      "step": 3731
    },
    {
      "epoch": 2.003758263145532,
      "grad_norm": 0.671875,
      "learning_rate": 3.2709359483669227e-05,
      "loss": 0.7696,
      "step": 3732
    },
    {
      "epoch": 2.004295157880608,
      "grad_norm": 0.7421875,
      "learning_rate": 3.2701325475755326e-05,
      "loss": 0.8414,
      "step": 3733
    },
    {
      "epoch": 2.004832052615684,
      "grad_norm": 0.640625,
      "learning_rate": 3.269329058902266e-05,
      "loss": 0.7175,
      "step": 3734
    },
    {
      "epoch": 2.00536894735076,
      "grad_norm": 0.66015625,
      "learning_rate": 3.268525482438813e-05,
      "loss": 0.9309,
      "step": 3735
    },
    {
      "epoch": 2.005905842085836,
      "grad_norm": 0.58984375,
      "learning_rate": 3.267721818276868e-05,
      "loss": 0.8925,
      "step": 3736
    },
    {
      "epoch": 2.006442736820912,
      "grad_norm": 0.68359375,
      "learning_rate": 3.2669180665081414e-05,
      "loss": 0.9805,
      "step": 3737
    },
    {
      "epoch": 2.0069796315559882,
      "grad_norm": 0.69921875,
      "learning_rate": 3.266114227224351e-05,
      "loss": 0.8372,
      "step": 3738
    },
    {
      "epoch": 2.007516526291064,
      "grad_norm": 0.9140625,
      "learning_rate": 3.2653103005172257e-05,
      "loss": 0.7892,
      "step": 3739
    },
    {
      "epoch": 2.00805342102614,
      "grad_norm": 0.58203125,
      "learning_rate": 3.264506286478503e-05,
      "loss": 0.6366,
      "step": 3740
    },
    {
      "epoch": 2.008590315761216,
      "grad_norm": 0.703125,
      "learning_rate": 3.2637021851999315e-05,
      "loss": 1.0804,
      "step": 3741
    },
    {
      "epoch": 2.0091272104962923,
      "grad_norm": 0.72265625,
      "learning_rate": 3.262897996773269e-05,
      "loss": 0.9327,
      "step": 3742
    },
    {
      "epoch": 2.009664105231368,
      "grad_norm": 0.7578125,
      "learning_rate": 3.2620937212902835e-05,
      "loss": 0.9353,
      "step": 3743
    },
    {
      "epoch": 2.010200999966444,
      "grad_norm": 0.85546875,
      "learning_rate": 3.2612893588427536e-05,
      "loss": 1.0168,
      "step": 3744
    },
    {
      "epoch": 2.01073789470152,
      "grad_norm": 0.7421875,
      "learning_rate": 3.260484909522467e-05,
      "loss": 0.895,
      "step": 3745
    },
    {
      "epoch": 2.011274789436596,
      "grad_norm": 0.6171875,
      "learning_rate": 3.259680373421221e-05,
      "loss": 0.8307,
      "step": 3746
    },
    {
      "epoch": 2.011811684171672,
      "grad_norm": 0.62890625,
      "learning_rate": 3.2588757506308236e-05,
      "loss": 0.7808,
      "step": 3747
    },
    {
      "epoch": 2.012348578906748,
      "grad_norm": 0.94140625,
      "learning_rate": 3.258071041243094e-05,
      "loss": 1.1381,
      "step": 3748
    },
    {
      "epoch": 2.0128854736418242,
      "grad_norm": 0.6640625,
      "learning_rate": 3.257266245349858e-05,
      "loss": 0.7828,
      "step": 3749
    },
    {
      "epoch": 2.0134223683769,
      "grad_norm": 0.8125,
      "learning_rate": 3.256461363042952e-05,
      "loss": 0.8813,
      "step": 3750
    },
    {
      "epoch": 2.0134223683769,
      "eval_loss": 0.905744194984436,
      "eval_runtime": 46.622,
      "eval_samples_per_second": 20.634,
      "eval_steps_per_second": 20.634,
      "step": 3750
    },
    {
      "epoch": 2.013959263111976,
      "grad_norm": 0.6171875,
      "learning_rate": 3.255656394414226e-05,
      "loss": 0.7973,
      "step": 3751
    },
    {
      "epoch": 2.014496157847052,
      "grad_norm": 0.765625,
      "learning_rate": 3.254851339555535e-05,
      "loss": 0.8319,
      "step": 3752
    },
    {
      "epoch": 2.0150330525821283,
      "grad_norm": 0.5390625,
      "learning_rate": 3.254046198558748e-05,
      "loss": 0.5987,
      "step": 3753
    },
    {
      "epoch": 2.015569947317204,
      "grad_norm": 0.56640625,
      "learning_rate": 3.253240971515739e-05,
      "loss": 0.745,
      "step": 3754
    },
    {
      "epoch": 2.01610684205228,
      "grad_norm": 0.75390625,
      "learning_rate": 3.252435658518397e-05,
      "loss": 0.775,
      "step": 3755
    },
    {
      "epoch": 2.016643736787356,
      "grad_norm": 0.70703125,
      "learning_rate": 3.251630259658618e-05,
      "loss": 0.8454,
      "step": 3756
    },
    {
      "epoch": 2.0171806315224323,
      "grad_norm": 0.734375,
      "learning_rate": 3.2508247750283065e-05,
      "loss": 0.7807,
      "step": 3757
    },
    {
      "epoch": 2.017717526257508,
      "grad_norm": 0.62109375,
      "learning_rate": 3.25001920471938e-05,
      "loss": 0.6699,
      "step": 3758
    },
    {
      "epoch": 2.018254420992584,
      "grad_norm": 0.73828125,
      "learning_rate": 3.249213548823765e-05,
      "loss": 0.7912,
      "step": 3759
    },
    {
      "epoch": 2.0187913157276602,
      "grad_norm": 0.7265625,
      "learning_rate": 3.2484078074333954e-05,
      "loss": 0.8884,
      "step": 3760
    },
    {
      "epoch": 2.0193282104627364,
      "grad_norm": 0.75,
      "learning_rate": 3.247601980640217e-05,
      "loss": 0.8316,
      "step": 3761
    },
    {
      "epoch": 2.019865105197812,
      "grad_norm": 0.64453125,
      "learning_rate": 3.2467960685361854e-05,
      "loss": 0.8054,
      "step": 3762
    },
    {
      "epoch": 2.020401999932888,
      "grad_norm": 0.6953125,
      "learning_rate": 3.245990071213265e-05,
      "loss": 0.7922,
      "step": 3763
    },
    {
      "epoch": 2.0209388946679643,
      "grad_norm": 0.69140625,
      "learning_rate": 3.245183988763429e-05,
      "loss": 0.7621,
      "step": 3764
    },
    {
      "epoch": 2.02147578940304,
      "grad_norm": 0.70703125,
      "learning_rate": 3.2443778212786646e-05,
      "loss": 0.8941,
      "step": 3765
    },
    {
      "epoch": 2.022012684138116,
      "grad_norm": 0.64453125,
      "learning_rate": 3.2435715688509626e-05,
      "loss": 0.8907,
      "step": 3766
    },
    {
      "epoch": 2.022549578873192,
      "grad_norm": 0.69921875,
      "learning_rate": 3.242765231572329e-05,
      "loss": 0.8796,
      "step": 3767
    },
    {
      "epoch": 2.0230864736082683,
      "grad_norm": 0.68359375,
      "learning_rate": 3.241958809534776e-05,
      "loss": 0.7591,
      "step": 3768
    },
    {
      "epoch": 2.023623368343344,
      "grad_norm": 0.828125,
      "learning_rate": 3.241152302830326e-05,
      "loss": 0.8742,
      "step": 3769
    },
    {
      "epoch": 2.02416026307842,
      "grad_norm": 0.82421875,
      "learning_rate": 3.240345711551014e-05,
      "loss": 1.0462,
      "step": 3770
    },
    {
      "epoch": 2.0246971578134962,
      "grad_norm": 0.6640625,
      "learning_rate": 3.2395390357888785e-05,
      "loss": 0.9092,
      "step": 3771
    },
    {
      "epoch": 2.0252340525485724,
      "grad_norm": 0.5546875,
      "learning_rate": 3.2387322756359745e-05,
      "loss": 0.614,
      "step": 3772
    },
    {
      "epoch": 2.025770947283648,
      "grad_norm": 0.6328125,
      "learning_rate": 3.2379254311843634e-05,
      "loss": 0.7836,
      "step": 3773
    },
    {
      "epoch": 2.026307842018724,
      "grad_norm": 0.6171875,
      "learning_rate": 3.237118502526114e-05,
      "loss": 0.8289,
      "step": 3774
    },
    {
      "epoch": 2.0268447367538003,
      "grad_norm": 0.73828125,
      "learning_rate": 3.236311489753309e-05,
      "loss": 0.8063,
      "step": 3775
    },
    {
      "epoch": 2.0273816314888764,
      "grad_norm": 0.6484375,
      "learning_rate": 3.235504392958039e-05,
      "loss": 0.7739,
      "step": 3776
    },
    {
      "epoch": 2.027918526223952,
      "grad_norm": 0.7109375,
      "learning_rate": 3.234697212232403e-05,
      "loss": 0.7872,
      "step": 3777
    },
    {
      "epoch": 2.028455420959028,
      "grad_norm": 0.734375,
      "learning_rate": 3.233889947668511e-05,
      "loss": 0.8459,
      "step": 3778
    },
    {
      "epoch": 2.0289923156941043,
      "grad_norm": 0.6875,
      "learning_rate": 3.2330825993584827e-05,
      "loss": 0.7905,
      "step": 3779
    },
    {
      "epoch": 2.0295292104291804,
      "grad_norm": 0.59375,
      "learning_rate": 3.232275167394444e-05,
      "loss": 0.952,
      "step": 3780
    },
    {
      "epoch": 2.030066105164256,
      "grad_norm": 0.62890625,
      "learning_rate": 3.2314676518685374e-05,
      "loss": 0.9142,
      "step": 3781
    },
    {
      "epoch": 2.0306029998993322,
      "grad_norm": 0.6015625,
      "learning_rate": 3.230660052872907e-05,
      "loss": 0.8182,
      "step": 3782
    },
    {
      "epoch": 2.0311398946344084,
      "grad_norm": 0.7890625,
      "learning_rate": 3.229852370499712e-05,
      "loss": 0.8295,
      "step": 3783
    },
    {
      "epoch": 2.031676789369484,
      "grad_norm": 0.62890625,
      "learning_rate": 3.229044604841119e-05,
      "loss": 0.8785,
      "step": 3784
    },
    {
      "epoch": 2.03221368410456,
      "grad_norm": 0.57421875,
      "learning_rate": 3.228236755989303e-05,
      "loss": 0.7369,
      "step": 3785
    },
    {
      "epoch": 2.0327505788396363,
      "grad_norm": 0.65625,
      "learning_rate": 3.227428824036452e-05,
      "loss": 0.7634,
      "step": 3786
    },
    {
      "epoch": 2.0332874735747124,
      "grad_norm": 0.6015625,
      "learning_rate": 3.2266208090747594e-05,
      "loss": 0.776,
      "step": 3787
    },
    {
      "epoch": 2.033824368309788,
      "grad_norm": 0.703125,
      "learning_rate": 3.22581271119643e-05,
      "loss": 0.8237,
      "step": 3788
    },
    {
      "epoch": 2.034361263044864,
      "grad_norm": 0.703125,
      "learning_rate": 3.2250045304936795e-05,
      "loss": 0.7268,
      "step": 3789
    },
    {
      "epoch": 2.0348981577799403,
      "grad_norm": 0.6171875,
      "learning_rate": 3.2241962670587314e-05,
      "loss": 0.7474,
      "step": 3790
    },
    {
      "epoch": 2.0354350525150164,
      "grad_norm": 0.5859375,
      "learning_rate": 3.223387920983816e-05,
      "loss": 0.7071,
      "step": 3791
    },
    {
      "epoch": 2.035971947250092,
      "grad_norm": 0.609375,
      "learning_rate": 3.222579492361179e-05,
      "loss": 0.7037,
      "step": 3792
    },
    {
      "epoch": 2.0365088419851682,
      "grad_norm": 0.66796875,
      "learning_rate": 3.221770981283071e-05,
      "loss": 0.9173,
      "step": 3793
    },
    {
      "epoch": 2.0370457367202444,
      "grad_norm": 0.6015625,
      "learning_rate": 3.220962387841754e-05,
      "loss": 0.8577,
      "step": 3794
    },
    {
      "epoch": 2.0375826314553205,
      "grad_norm": 0.67578125,
      "learning_rate": 3.220153712129498e-05,
      "loss": 1.0043,
      "step": 3795
    },
    {
      "epoch": 2.038119526190396,
      "grad_norm": 0.6640625,
      "learning_rate": 3.2193449542385835e-05,
      "loss": 0.926,
      "step": 3796
    },
    {
      "epoch": 2.0386564209254723,
      "grad_norm": 0.6484375,
      "learning_rate": 3.218536114261299e-05,
      "loss": 0.8775,
      "step": 3797
    },
    {
      "epoch": 2.0391933156605484,
      "grad_norm": 0.625,
      "learning_rate": 3.217727192289946e-05,
      "loss": 0.7468,
      "step": 3798
    },
    {
      "epoch": 2.0397302103956245,
      "grad_norm": 0.71875,
      "learning_rate": 3.21691818841683e-05,
      "loss": 0.8912,
      "step": 3799
    },
    {
      "epoch": 2.0402671051307,
      "grad_norm": 0.6875,
      "learning_rate": 3.216109102734269e-05,
      "loss": 1.1117,
      "step": 3800
    },
    {
      "epoch": 2.0402671051307,
      "eval_loss": 0.906833827495575,
      "eval_runtime": 46.2156,
      "eval_samples_per_second": 20.815,
      "eval_steps_per_second": 20.815,
      "step": 3800
    },
    {
      "epoch": 2.0408039998657763,
      "grad_norm": 0.68359375,
      "learning_rate": 3.215299935334592e-05,
      "loss": 0.7149,
      "step": 3801
    },
    {
      "epoch": 2.0413408946008524,
      "grad_norm": 0.66015625,
      "learning_rate": 3.214490686310133e-05,
      "loss": 0.7853,
      "step": 3802
    },
    {
      "epoch": 2.041877789335928,
      "grad_norm": 0.7578125,
      "learning_rate": 3.213681355753239e-05,
      "loss": 0.8345,
      "step": 3803
    },
    {
      "epoch": 2.0424146840710042,
      "grad_norm": 0.69140625,
      "learning_rate": 3.2128719437562646e-05,
      "loss": 0.944,
      "step": 3804
    },
    {
      "epoch": 2.0429515788060804,
      "grad_norm": 0.6953125,
      "learning_rate": 3.212062450411572e-05,
      "loss": 0.8661,
      "step": 3805
    },
    {
      "epoch": 2.0434884735411565,
      "grad_norm": 0.69921875,
      "learning_rate": 3.211252875811538e-05,
      "loss": 1.0284,
      "step": 3806
    },
    {
      "epoch": 2.044025368276232,
      "grad_norm": 0.62109375,
      "learning_rate": 3.2104432200485426e-05,
      "loss": 0.775,
      "step": 3807
    },
    {
      "epoch": 2.0445622630113083,
      "grad_norm": 0.671875,
      "learning_rate": 3.2096334832149785e-05,
      "loss": 0.8019,
      "step": 3808
    },
    {
      "epoch": 2.0450991577463844,
      "grad_norm": 0.6640625,
      "learning_rate": 3.208823665403248e-05,
      "loss": 0.742,
      "step": 3809
    },
    {
      "epoch": 2.0456360524814605,
      "grad_norm": 0.703125,
      "learning_rate": 3.2080137667057595e-05,
      "loss": 0.7986,
      "step": 3810
    },
    {
      "epoch": 2.046172947216536,
      "grad_norm": 0.66015625,
      "learning_rate": 3.207203787214936e-05,
      "loss": 0.7775,
      "step": 3811
    },
    {
      "epoch": 2.0467098419516123,
      "grad_norm": 0.703125,
      "learning_rate": 3.206393727023203e-05,
      "loss": 0.8301,
      "step": 3812
    },
    {
      "epoch": 2.0472467366866884,
      "grad_norm": 0.63671875,
      "learning_rate": 3.205583586223e-05,
      "loss": 0.8504,
      "step": 3813
    },
    {
      "epoch": 2.0477836314217646,
      "grad_norm": 0.6953125,
      "learning_rate": 3.2047733649067746e-05,
      "loss": 0.7674,
      "step": 3814
    },
    {
      "epoch": 2.0483205261568402,
      "grad_norm": 0.671875,
      "learning_rate": 3.203963063166983e-05,
      "loss": 0.8156,
      "step": 3815
    },
    {
      "epoch": 2.0488574208919164,
      "grad_norm": 0.7890625,
      "learning_rate": 3.203152681096091e-05,
      "loss": 1.0049,
      "step": 3816
    },
    {
      "epoch": 2.0493943156269925,
      "grad_norm": 0.640625,
      "learning_rate": 3.202342218786574e-05,
      "loss": 0.7347,
      "step": 3817
    },
    {
      "epoch": 2.0499312103620686,
      "grad_norm": 0.703125,
      "learning_rate": 3.201531676330914e-05,
      "loss": 0.9347,
      "step": 3818
    },
    {
      "epoch": 2.0504681050971443,
      "grad_norm": 0.703125,
      "learning_rate": 3.2007210538216066e-05,
      "loss": 0.8745,
      "step": 3819
    },
    {
      "epoch": 2.0510049998322204,
      "grad_norm": 0.6484375,
      "learning_rate": 3.199910351351153e-05,
      "loss": 0.8997,
      "step": 3820
    },
    {
      "epoch": 2.0515418945672965,
      "grad_norm": 0.703125,
      "learning_rate": 3.1990995690120635e-05,
      "loss": 0.6886,
      "step": 3821
    },
    {
      "epoch": 2.052078789302372,
      "grad_norm": 0.57421875,
      "learning_rate": 3.19828870689686e-05,
      "loss": 0.6798,
      "step": 3822
    },
    {
      "epoch": 2.0526156840374483,
      "grad_norm": 0.71484375,
      "learning_rate": 3.1974777650980735e-05,
      "loss": 0.8744,
      "step": 3823
    },
    {
      "epoch": 2.0531525787725244,
      "grad_norm": 0.62109375,
      "learning_rate": 3.19666674370824e-05,
      "loss": 0.7508,
      "step": 3824
    },
    {
      "epoch": 2.0536894735076006,
      "grad_norm": 0.6875,
      "learning_rate": 3.195855642819909e-05,
      "loss": 0.9087,
      "step": 3825
    },
    {
      "epoch": 2.0542263682426762,
      "grad_norm": 0.72265625,
      "learning_rate": 3.195044462525636e-05,
      "loss": 1.0194,
      "step": 3826
    },
    {
      "epoch": 2.0547632629777524,
      "grad_norm": 0.6953125,
      "learning_rate": 3.194233202917989e-05,
      "loss": 0.9374,
      "step": 3827
    },
    {
      "epoch": 2.0553001577128285,
      "grad_norm": 0.8515625,
      "learning_rate": 3.193421864089541e-05,
      "loss": 0.9167,
      "step": 3828
    },
    {
      "epoch": 2.0558370524479046,
      "grad_norm": 0.6484375,
      "learning_rate": 3.1926104461328766e-05,
      "loss": 0.9575,
      "step": 3829
    },
    {
      "epoch": 2.0563739471829803,
      "grad_norm": 0.5859375,
      "learning_rate": 3.19179894914059e-05,
      "loss": 0.8427,
      "step": 3830
    },
    {
      "epoch": 2.0569108419180564,
      "grad_norm": 0.703125,
      "learning_rate": 3.1909873732052816e-05,
      "loss": 0.9534,
      "step": 3831
    },
    {
      "epoch": 2.0574477366531325,
      "grad_norm": 0.76953125,
      "learning_rate": 3.190175718419564e-05,
      "loss": 0.9799,
      "step": 3832
    },
    {
      "epoch": 2.0579846313882086,
      "grad_norm": 0.765625,
      "learning_rate": 3.189363984876056e-05,
      "loss": 0.7339,
      "step": 3833
    },
    {
      "epoch": 2.0585215261232843,
      "grad_norm": 0.625,
      "learning_rate": 3.188552172667387e-05,
      "loss": 0.9575,
      "step": 3834
    },
    {
      "epoch": 2.0590584208583604,
      "grad_norm": 0.69921875,
      "learning_rate": 3.187740281886195e-05,
      "loss": 0.783,
      "step": 3835
    },
    {
      "epoch": 2.0595953155934366,
      "grad_norm": 0.6953125,
      "learning_rate": 3.186928312625129e-05,
      "loss": 0.8004,
      "step": 3836
    },
    {
      "epoch": 2.0601322103285127,
      "grad_norm": 0.77734375,
      "learning_rate": 3.1861162649768415e-05,
      "loss": 1.0324,
      "step": 3837
    },
    {
      "epoch": 2.0606691050635884,
      "grad_norm": 0.78125,
      "learning_rate": 3.185304139034e-05,
      "loss": 0.968,
      "step": 3838
    },
    {
      "epoch": 2.0612059997986645,
      "grad_norm": 0.75,
      "learning_rate": 3.184491934889278e-05,
      "loss": 0.8289,
      "step": 3839
    },
    {
      "epoch": 2.0617428945337406,
      "grad_norm": 0.69140625,
      "learning_rate": 3.183679652635357e-05,
      "loss": 0.8071,
      "step": 3840
    },
    {
      "epoch": 2.0622797892688163,
      "grad_norm": 0.83203125,
      "learning_rate": 3.182867292364929e-05,
      "loss": 0.8421,
      "step": 3841
    },
    {
      "epoch": 2.0628166840038924,
      "grad_norm": 0.66796875,
      "learning_rate": 3.182054854170697e-05,
      "loss": 0.8974,
      "step": 3842
    },
    {
      "epoch": 2.0633535787389685,
      "grad_norm": 1.1484375,
      "learning_rate": 3.181242338145367e-05,
      "loss": 0.9089,
      "step": 3843
    },
    {
      "epoch": 2.0638904734740446,
      "grad_norm": 0.6640625,
      "learning_rate": 3.18042974438166e-05,
      "loss": 0.8187,
      "step": 3844
    },
    {
      "epoch": 2.0644273682091203,
      "grad_norm": 0.71875,
      "learning_rate": 3.179617072972302e-05,
      "loss": 0.721,
      "step": 3845
    },
    {
      "epoch": 2.0649642629441964,
      "grad_norm": 0.734375,
      "learning_rate": 3.178804324010028e-05,
      "loss": 0.8762,
      "step": 3846
    },
    {
      "epoch": 2.0655011576792726,
      "grad_norm": 0.58984375,
      "learning_rate": 3.1779914975875856e-05,
      "loss": 0.795,
      "step": 3847
    },
    {
      "epoch": 2.0660380524143487,
      "grad_norm": 0.71484375,
      "learning_rate": 3.1771785937977274e-05,
      "loss": 0.8996,
      "step": 3848
    },
    {
      "epoch": 2.0665749471494244,
      "grad_norm": 0.7734375,
      "learning_rate": 3.176365612733216e-05,
      "loss": 0.7479,
      "step": 3849
    },
    {
      "epoch": 2.0671118418845005,
      "grad_norm": 0.578125,
      "learning_rate": 3.175552554486822e-05,
      "loss": 0.766,
      "step": 3850
    },
    {
      "epoch": 2.0671118418845005,
      "eval_loss": 0.9062168598175049,
      "eval_runtime": 46.226,
      "eval_samples_per_second": 20.811,
      "eval_steps_per_second": 20.811,
      "step": 3850
    },
    {
      "epoch": 2.0676487366195766,
      "grad_norm": 0.6640625,
      "learning_rate": 3.174739419151326e-05,
      "loss": 0.9449,
      "step": 3851
    },
    {
      "epoch": 2.0681856313546527,
      "grad_norm": 0.63671875,
      "learning_rate": 3.173926206819518e-05,
      "loss": 0.8738,
      "step": 3852
    },
    {
      "epoch": 2.0687225260897284,
      "grad_norm": 0.7421875,
      "learning_rate": 3.173112917584196e-05,
      "loss": 0.937,
      "step": 3853
    },
    {
      "epoch": 2.0692594208248045,
      "grad_norm": 0.80078125,
      "learning_rate": 3.172299551538164e-05,
      "loss": 0.9554,
      "step": 3854
    },
    {
      "epoch": 2.0697963155598806,
      "grad_norm": 0.56640625,
      "learning_rate": 3.171486108774241e-05,
      "loss": 0.7058,
      "step": 3855
    },
    {
      "epoch": 2.0703332102949563,
      "grad_norm": 0.765625,
      "learning_rate": 3.1706725893852484e-05,
      "loss": 0.9259,
      "step": 3856
    },
    {
      "epoch": 2.0708701050300324,
      "grad_norm": 0.67578125,
      "learning_rate": 3.16985899346402e-05,
      "loss": 0.8209,
      "step": 3857
    },
    {
      "epoch": 2.0714069997651086,
      "grad_norm": 0.67578125,
      "learning_rate": 3.169045321103398e-05,
      "loss": 1.0139,
      "step": 3858
    },
    {
      "epoch": 2.0719438945001847,
      "grad_norm": 0.640625,
      "learning_rate": 3.168231572396231e-05,
      "loss": 0.824,
      "step": 3859
    },
    {
      "epoch": 2.0724807892352604,
      "grad_norm": 0.73046875,
      "learning_rate": 3.167417747435379e-05,
      "loss": 0.9313,
      "step": 3860
    },
    {
      "epoch": 2.0730176839703365,
      "grad_norm": 0.66796875,
      "learning_rate": 3.166603846313711e-05,
      "loss": 0.9802,
      "step": 3861
    },
    {
      "epoch": 2.0735545787054126,
      "grad_norm": 0.83984375,
      "learning_rate": 3.165789869124102e-05,
      "loss": 0.7642,
      "step": 3862
    },
    {
      "epoch": 2.0740914734404887,
      "grad_norm": 0.703125,
      "learning_rate": 3.1649758159594365e-05,
      "loss": 0.8664,
      "step": 3863
    },
    {
      "epoch": 2.0746283681755644,
      "grad_norm": 0.72265625,
      "learning_rate": 3.1641616869126103e-05,
      "loss": 0.952,
      "step": 3864
    },
    {
      "epoch": 2.0751652629106405,
      "grad_norm": 0.65625,
      "learning_rate": 3.163347482076524e-05,
      "loss": 1.0794,
      "step": 3865
    },
    {
      "epoch": 2.0757021576457166,
      "grad_norm": 0.765625,
      "learning_rate": 3.16253320154409e-05,
      "loss": 0.8358,
      "step": 3866
    },
    {
      "epoch": 2.0762390523807928,
      "grad_norm": 0.68359375,
      "learning_rate": 3.161718845408227e-05,
      "loss": 0.9434,
      "step": 3867
    },
    {
      "epoch": 2.0767759471158684,
      "grad_norm": 0.64453125,
      "learning_rate": 3.160904413761863e-05,
      "loss": 0.6919,
      "step": 3868
    },
    {
      "epoch": 2.0773128418509446,
      "grad_norm": 0.66796875,
      "learning_rate": 3.1600899066979376e-05,
      "loss": 0.8196,
      "step": 3869
    },
    {
      "epoch": 2.0778497365860207,
      "grad_norm": 0.671875,
      "learning_rate": 3.159275324309393e-05,
      "loss": 0.8148,
      "step": 3870
    },
    {
      "epoch": 2.078386631321097,
      "grad_norm": 0.72265625,
      "learning_rate": 3.158460666689185e-05,
      "loss": 1.1962,
      "step": 3871
    },
    {
      "epoch": 2.0789235260561725,
      "grad_norm": 0.671875,
      "learning_rate": 3.1576459339302763e-05,
      "loss": 0.8825,
      "step": 3872
    },
    {
      "epoch": 2.0794604207912486,
      "grad_norm": 0.7265625,
      "learning_rate": 3.156831126125638e-05,
      "loss": 0.9035,
      "step": 3873
    },
    {
      "epoch": 2.0799973155263247,
      "grad_norm": 0.69140625,
      "learning_rate": 3.156016243368251e-05,
      "loss": 0.9358,
      "step": 3874
    },
    {
      "epoch": 2.0805342102614004,
      "grad_norm": 0.78515625,
      "learning_rate": 3.1552012857511014e-05,
      "loss": 0.8911,
      "step": 3875
    },
    {
      "epoch": 2.0810711049964765,
      "grad_norm": 0.81640625,
      "learning_rate": 3.1543862533671886e-05,
      "loss": 1.2351,
      "step": 3876
    },
    {
      "epoch": 2.0816079997315526,
      "grad_norm": 0.671875,
      "learning_rate": 3.1535711463095166e-05,
      "loss": 0.8813,
      "step": 3877
    },
    {
      "epoch": 2.0821448944666288,
      "grad_norm": 0.65234375,
      "learning_rate": 3.152755964671101e-05,
      "loss": 0.8181,
      "step": 3878
    },
    {
      "epoch": 2.0826817892017044,
      "grad_norm": 0.6015625,
      "learning_rate": 3.1519407085449627e-05,
      "loss": 0.819,
      "step": 3879
    },
    {
      "epoch": 2.0832186839367806,
      "grad_norm": 0.73046875,
      "learning_rate": 3.151125378024133e-05,
      "loss": 0.8729,
      "step": 3880
    },
    {
      "epoch": 2.0837555786718567,
      "grad_norm": 0.734375,
      "learning_rate": 3.150309973201653e-05,
      "loss": 0.872,
      "step": 3881
    },
    {
      "epoch": 2.084292473406933,
      "grad_norm": 0.546875,
      "learning_rate": 3.149494494170569e-05,
      "loss": 0.6089,
      "step": 3882
    },
    {
      "epoch": 2.0848293681420085,
      "grad_norm": 0.67578125,
      "learning_rate": 3.1486789410239376e-05,
      "loss": 0.8081,
      "step": 3883
    },
    {
      "epoch": 2.0853662628770846,
      "grad_norm": 0.76171875,
      "learning_rate": 3.1478633138548235e-05,
      "loss": 1.0491,
      "step": 3884
    },
    {
      "epoch": 2.0859031576121607,
      "grad_norm": 0.7421875,
      "learning_rate": 3.147047612756302e-05,
      "loss": 0.9228,
      "step": 3885
    },
    {
      "epoch": 2.086440052347237,
      "grad_norm": 0.76953125,
      "learning_rate": 3.146231837821454e-05,
      "loss": 0.8107,
      "step": 3886
    },
    {
      "epoch": 2.0869769470823125,
      "grad_norm": 0.671875,
      "learning_rate": 3.1454159891433685e-05,
      "loss": 0.9009,
      "step": 3887
    },
    {
      "epoch": 2.0875138418173886,
      "grad_norm": 0.6875,
      "learning_rate": 3.144600066815145e-05,
      "loss": 0.9105,
      "step": 3888
    },
    {
      "epoch": 2.0880507365524648,
      "grad_norm": 0.69140625,
      "learning_rate": 3.1437840709298914e-05,
      "loss": 0.9284,
      "step": 3889
    },
    {
      "epoch": 2.088587631287541,
      "grad_norm": 0.6328125,
      "learning_rate": 3.142968001580722e-05,
      "loss": 0.7902,
      "step": 3890
    },
    {
      "epoch": 2.0891245260226166,
      "grad_norm": 0.734375,
      "learning_rate": 3.142151858860761e-05,
      "loss": 0.9536,
      "step": 3891
    },
    {
      "epoch": 2.0896614207576927,
      "grad_norm": 0.63671875,
      "learning_rate": 3.1413356428631414e-05,
      "loss": 1.0451,
      "step": 3892
    },
    {
      "epoch": 2.090198315492769,
      "grad_norm": 0.73046875,
      "learning_rate": 3.140519353681003e-05,
      "loss": 0.8377,
      "step": 3893
    },
    {
      "epoch": 2.0907352102278445,
      "grad_norm": 0.765625,
      "learning_rate": 3.139702991407494e-05,
      "loss": 0.7791,
      "step": 3894
    },
    {
      "epoch": 2.0912721049629206,
      "grad_norm": 0.6328125,
      "learning_rate": 3.138886556135773e-05,
      "loss": 0.7932,
      "step": 3895
    },
    {
      "epoch": 2.0918089996979967,
      "grad_norm": 0.71484375,
      "learning_rate": 3.138070047959005e-05,
      "loss": 0.775,
      "step": 3896
    },
    {
      "epoch": 2.092345894433073,
      "grad_norm": 0.66796875,
      "learning_rate": 3.137253466970364e-05,
      "loss": 0.7762,
      "step": 3897
    },
    {
      "epoch": 2.0928827891681485,
      "grad_norm": 0.69921875,
      "learning_rate": 3.1364368132630316e-05,
      "loss": 0.932,
      "step": 3898
    },
    {
      "epoch": 2.0934196839032246,
      "grad_norm": 0.73046875,
      "learning_rate": 3.1356200869302e-05,
      "loss": 0.8704,
      "step": 3899
    },
    {
      "epoch": 2.0939565786383008,
      "grad_norm": 0.7265625,
      "learning_rate": 3.1348032880650675e-05,
      "loss": 0.8469,
      "step": 3900
    },
    {
      "epoch": 2.0939565786383008,
      "eval_loss": 0.9066072702407837,
      "eval_runtime": 46.1752,
      "eval_samples_per_second": 20.834,
      "eval_steps_per_second": 20.834,
      "step": 3900
    },
    {
      "epoch": 2.094493473373377,
      "grad_norm": 0.75,
      "learning_rate": 3.133986416760839e-05,
      "loss": 0.9143,
      "step": 3901
    },
    {
      "epoch": 2.0950303681084526,
      "grad_norm": 0.84765625,
      "learning_rate": 3.133169473110733e-05,
      "loss": 0.7819,
      "step": 3902
    },
    {
      "epoch": 2.0955672628435287,
      "grad_norm": 0.61328125,
      "learning_rate": 3.132352457207971e-05,
      "loss": 0.9131,
      "step": 3903
    },
    {
      "epoch": 2.096104157578605,
      "grad_norm": 0.71875,
      "learning_rate": 3.1315353691457864e-05,
      "loss": 0.8014,
      "step": 3904
    },
    {
      "epoch": 2.096641052313681,
      "grad_norm": 0.60546875,
      "learning_rate": 3.130718209017417e-05,
      "loss": 0.7388,
      "step": 3905
    },
    {
      "epoch": 2.0971779470487566,
      "grad_norm": 0.6640625,
      "learning_rate": 3.1299009769161133e-05,
      "loss": 0.6695,
      "step": 3906
    },
    {
      "epoch": 2.0977148417838327,
      "grad_norm": 0.80078125,
      "learning_rate": 3.129083672935131e-05,
      "loss": 0.863,
      "step": 3907
    },
    {
      "epoch": 2.098251736518909,
      "grad_norm": 0.8515625,
      "learning_rate": 3.1282662971677356e-05,
      "loss": 1.1562,
      "step": 3908
    },
    {
      "epoch": 2.098788631253985,
      "grad_norm": 0.79296875,
      "learning_rate": 3.127448849707198e-05,
      "loss": 0.9662,
      "step": 3909
    },
    {
      "epoch": 2.0993255259890606,
      "grad_norm": 0.69140625,
      "learning_rate": 3.126631330646802e-05,
      "loss": 0.8017,
      "step": 3910
    },
    {
      "epoch": 2.0998624207241368,
      "grad_norm": 0.82421875,
      "learning_rate": 3.1258137400798345e-05,
      "loss": 0.7675,
      "step": 3911
    },
    {
      "epoch": 2.100399315459213,
      "grad_norm": 0.59375,
      "learning_rate": 3.1249960780995946e-05,
      "loss": 0.7322,
      "step": 3912
    },
    {
      "epoch": 2.1009362101942886,
      "grad_norm": 0.67578125,
      "learning_rate": 3.1241783447993865e-05,
      "loss": 0.8878,
      "step": 3913
    },
    {
      "epoch": 2.1014731049293647,
      "grad_norm": 0.56640625,
      "learning_rate": 3.123360540272525e-05,
      "loss": 0.7652,
      "step": 3914
    },
    {
      "epoch": 2.102009999664441,
      "grad_norm": 0.734375,
      "learning_rate": 3.122542664612332e-05,
      "loss": 0.7856,
      "step": 3915
    },
    {
      "epoch": 2.102546894399517,
      "grad_norm": 0.78125,
      "learning_rate": 3.121724717912138e-05,
      "loss": 0.8477,
      "step": 3916
    },
    {
      "epoch": 2.1030837891345926,
      "grad_norm": 0.71875,
      "learning_rate": 3.120906700265278e-05,
      "loss": 0.7319,
      "step": 3917
    },
    {
      "epoch": 2.1036206838696687,
      "grad_norm": 0.59375,
      "learning_rate": 3.1200886117651016e-05,
      "loss": 0.7902,
      "step": 3918
    },
    {
      "epoch": 2.104157578604745,
      "grad_norm": 0.6875,
      "learning_rate": 3.119270452504962e-05,
      "loss": 0.8681,
      "step": 3919
    },
    {
      "epoch": 2.104694473339821,
      "grad_norm": 0.91015625,
      "learning_rate": 3.118452222578221e-05,
      "loss": 1.0723,
      "step": 3920
    },
    {
      "epoch": 2.1052313680748966,
      "grad_norm": 0.7265625,
      "learning_rate": 3.1176339220782494e-05,
      "loss": 0.898,
      "step": 3921
    },
    {
      "epoch": 2.1057682628099728,
      "grad_norm": 0.6171875,
      "learning_rate": 3.1168155510984255e-05,
      "loss": 0.8005,
      "step": 3922
    },
    {
      "epoch": 2.106305157545049,
      "grad_norm": 0.7734375,
      "learning_rate": 3.1159971097321355e-05,
      "loss": 0.9795,
      "step": 3923
    },
    {
      "epoch": 2.106842052280125,
      "grad_norm": 0.73828125,
      "learning_rate": 3.1151785980727755e-05,
      "loss": 0.9722,
      "step": 3924
    },
    {
      "epoch": 2.1073789470152007,
      "grad_norm": 0.66015625,
      "learning_rate": 3.114360016213745e-05,
      "loss": 0.7907,
      "step": 3925
    },
    {
      "epoch": 2.107915841750277,
      "grad_norm": 0.75390625,
      "learning_rate": 3.1135413642484576e-05,
      "loss": 0.9692,
      "step": 3926
    },
    {
      "epoch": 2.108452736485353,
      "grad_norm": 0.59375,
      "learning_rate": 3.112722642270331e-05,
      "loss": 0.9058,
      "step": 3927
    },
    {
      "epoch": 2.108989631220429,
      "grad_norm": 0.60546875,
      "learning_rate": 3.111903850372791e-05,
      "loss": 0.7444,
      "step": 3928
    },
    {
      "epoch": 2.1095265259555047,
      "grad_norm": 0.7265625,
      "learning_rate": 3.1110849886492724e-05,
      "loss": 0.8687,
      "step": 3929
    },
    {
      "epoch": 2.110063420690581,
      "grad_norm": 0.84375,
      "learning_rate": 3.1102660571932184e-05,
      "loss": 0.7473,
      "step": 3930
    },
    {
      "epoch": 2.110600315425657,
      "grad_norm": 0.57421875,
      "learning_rate": 3.109447056098078e-05,
      "loss": 0.656,
      "step": 3931
    },
    {
      "epoch": 2.1111372101607326,
      "grad_norm": 0.61328125,
      "learning_rate": 3.1086279854573116e-05,
      "loss": 0.979,
      "step": 3932
    },
    {
      "epoch": 2.1116741048958088,
      "grad_norm": 0.76171875,
      "learning_rate": 3.1078088453643827e-05,
      "loss": 0.8921,
      "step": 3933
    },
    {
      "epoch": 2.112210999630885,
      "grad_norm": 0.671875,
      "learning_rate": 3.1069896359127686e-05,
      "loss": 0.7848,
      "step": 3934
    },
    {
      "epoch": 2.112747894365961,
      "grad_norm": 0.69140625,
      "learning_rate": 3.10617035719595e-05,
      "loss": 0.6612,
      "step": 3935
    },
    {
      "epoch": 2.1132847891010367,
      "grad_norm": 0.64453125,
      "learning_rate": 3.1053510093074166e-05,
      "loss": 0.8053,
      "step": 3936
    },
    {
      "epoch": 2.113821683836113,
      "grad_norm": 0.73828125,
      "learning_rate": 3.1045315923406676e-05,
      "loss": 0.8986,
      "step": 3937
    },
    {
      "epoch": 2.114358578571189,
      "grad_norm": 0.703125,
      "learning_rate": 3.103712106389208e-05,
      "loss": 0.9231,
      "step": 3938
    },
    {
      "epoch": 2.114895473306265,
      "grad_norm": 0.66796875,
      "learning_rate": 3.1028925515465504e-05,
      "loss": 0.6654,
      "step": 3939
    },
    {
      "epoch": 2.1154323680413407,
      "grad_norm": 0.63671875,
      "learning_rate": 3.102072927906219e-05,
      "loss": 0.7507,
      "step": 3940
    },
    {
      "epoch": 2.115969262776417,
      "grad_norm": 0.6796875,
      "learning_rate": 3.101253235561742e-05,
      "loss": 0.7998,
      "step": 3941
    },
    {
      "epoch": 2.116506157511493,
      "grad_norm": 0.63671875,
      "learning_rate": 3.100433474606656e-05,
      "loss": 1.0736,
      "step": 3942
    },
    {
      "epoch": 2.117043052246569,
      "grad_norm": 0.65234375,
      "learning_rate": 3.099613645134507e-05,
      "loss": 0.749,
      "step": 3943
    },
    {
      "epoch": 2.1175799469816448,
      "grad_norm": 0.55859375,
      "learning_rate": 3.0987937472388475e-05,
      "loss": 0.7379,
      "step": 3944
    },
    {
      "epoch": 2.118116841716721,
      "grad_norm": 0.62109375,
      "learning_rate": 3.097973781013238e-05,
      "loss": 0.697,
      "step": 3945
    },
    {
      "epoch": 2.118653736451797,
      "grad_norm": 0.84765625,
      "learning_rate": 3.097153746551248e-05,
      "loss": 1.005,
      "step": 3946
    },
    {
      "epoch": 2.119190631186873,
      "grad_norm": 0.671875,
      "learning_rate": 3.0963336439464526e-05,
      "loss": 0.8847,
      "step": 3947
    },
    {
      "epoch": 2.119727525921949,
      "grad_norm": 0.6953125,
      "learning_rate": 3.095513473292436e-05,
      "loss": 0.8669,
      "step": 3948
    },
    {
      "epoch": 2.120264420657025,
      "grad_norm": 0.625,
      "learning_rate": 3.094693234682792e-05,
      "loss": 0.7202,
      "step": 3949
    },
    {
      "epoch": 2.120801315392101,
      "grad_norm": 0.71875,
      "learning_rate": 3.0938729282111176e-05,
      "loss": 0.9628,
      "step": 3950
    },
    {
      "epoch": 2.120801315392101,
      "eval_loss": 0.906258225440979,
      "eval_runtime": 46.8303,
      "eval_samples_per_second": 20.542,
      "eval_steps_per_second": 20.542,
      "step": 3950
    },
    {
      "epoch": 2.1213382101271767,
      "grad_norm": 0.625,
      "learning_rate": 3.093052553971022e-05,
      "loss": 0.7971,
      "step": 3951
    },
    {
      "epoch": 2.121875104862253,
      "grad_norm": 0.65234375,
      "learning_rate": 3.0922321120561186e-05,
      "loss": 0.8604,
      "step": 3952
    },
    {
      "epoch": 2.122411999597329,
      "grad_norm": 0.7890625,
      "learning_rate": 3.091411602560032e-05,
      "loss": 0.8655,
      "step": 3953
    },
    {
      "epoch": 2.122948894332405,
      "grad_norm": 0.625,
      "learning_rate": 3.090591025576392e-05,
      "loss": 0.7767,
      "step": 3954
    },
    {
      "epoch": 2.1234857890674808,
      "grad_norm": 0.71484375,
      "learning_rate": 3.089770381198836e-05,
      "loss": 0.8675,
      "step": 3955
    },
    {
      "epoch": 2.124022683802557,
      "grad_norm": 0.62109375,
      "learning_rate": 3.088949669521011e-05,
      "loss": 0.833,
      "step": 3956
    },
    {
      "epoch": 2.124559578537633,
      "grad_norm": 0.73046875,
      "learning_rate": 3.088128890636571e-05,
      "loss": 0.8068,
      "step": 3957
    },
    {
      "epoch": 2.125096473272709,
      "grad_norm": 0.66796875,
      "learning_rate": 3.087308044639176e-05,
      "loss": 0.924,
      "step": 3958
    },
    {
      "epoch": 2.125633368007785,
      "grad_norm": 0.75390625,
      "learning_rate": 3.0864871316224955e-05,
      "loss": 0.7982,
      "step": 3959
    },
    {
      "epoch": 2.126170262742861,
      "grad_norm": 0.6953125,
      "learning_rate": 3.0856661516802054e-05,
      "loss": 1.0301,
      "step": 3960
    },
    {
      "epoch": 2.126707157477937,
      "grad_norm": 0.69921875,
      "learning_rate": 3.084845104905991e-05,
      "loss": 0.8342,
      "step": 3961
    },
    {
      "epoch": 2.127244052213013,
      "grad_norm": 0.76171875,
      "learning_rate": 3.084023991393544e-05,
      "loss": 1.0664,
      "step": 3962
    },
    {
      "epoch": 2.127780946948089,
      "grad_norm": 0.61328125,
      "learning_rate": 3.083202811236564e-05,
      "loss": 0.8307,
      "step": 3963
    },
    {
      "epoch": 2.128317841683165,
      "grad_norm": 0.703125,
      "learning_rate": 3.082381564528756e-05,
      "loss": 1.0423,
      "step": 3964
    },
    {
      "epoch": 2.128854736418241,
      "grad_norm": 0.78515625,
      "learning_rate": 3.081560251363837e-05,
      "loss": 1.0167,
      "step": 3965
    },
    {
      "epoch": 2.129391631153317,
      "grad_norm": 0.65625,
      "learning_rate": 3.0807388718355286e-05,
      "loss": 0.8246,
      "step": 3966
    },
    {
      "epoch": 2.129928525888393,
      "grad_norm": 0.6875,
      "learning_rate": 3.07991742603756e-05,
      "loss": 0.7927,
      "step": 3967
    },
    {
      "epoch": 2.130465420623469,
      "grad_norm": 0.68359375,
      "learning_rate": 3.07909591406367e-05,
      "loss": 0.6947,
      "step": 3968
    },
    {
      "epoch": 2.131002315358545,
      "grad_norm": 0.61328125,
      "learning_rate": 3.078274336007601e-05,
      "loss": 0.7996,
      "step": 3969
    },
    {
      "epoch": 2.131539210093621,
      "grad_norm": 0.6328125,
      "learning_rate": 3.077452691963108e-05,
      "loss": 0.8553,
      "step": 3970
    },
    {
      "epoch": 2.132076104828697,
      "grad_norm": 0.65234375,
      "learning_rate": 3.076630982023951e-05,
      "loss": 0.8175,
      "step": 3971
    },
    {
      "epoch": 2.132612999563773,
      "grad_norm": 0.609375,
      "learning_rate": 3.075809206283895e-05,
      "loss": 0.9002,
      "step": 3972
    },
    {
      "epoch": 2.133149894298849,
      "grad_norm": 0.5546875,
      "learning_rate": 3.0749873648367164e-05,
      "loss": 0.6067,
      "step": 3973
    },
    {
      "epoch": 2.133686789033925,
      "grad_norm": 0.640625,
      "learning_rate": 3.074165457776199e-05,
      "loss": 0.7013,
      "step": 3974
    },
    {
      "epoch": 2.134223683769001,
      "grad_norm": 0.6484375,
      "learning_rate": 3.073343485196131e-05,
      "loss": 0.8546,
      "step": 3975
    },
    {
      "epoch": 2.134760578504077,
      "grad_norm": 0.72265625,
      "learning_rate": 3.072521447190311e-05,
      "loss": 0.9153,
      "step": 3976
    },
    {
      "epoch": 2.135297473239153,
      "grad_norm": 0.6796875,
      "learning_rate": 3.071699343852542e-05,
      "loss": 0.8468,
      "step": 3977
    },
    {
      "epoch": 2.135834367974229,
      "grad_norm": 0.65625,
      "learning_rate": 3.0708771752766394e-05,
      "loss": 0.8384,
      "step": 3978
    },
    {
      "epoch": 2.136371262709305,
      "grad_norm": 0.796875,
      "learning_rate": 3.070054941556422e-05,
      "loss": 0.8359,
      "step": 3979
    },
    {
      "epoch": 2.136908157444381,
      "grad_norm": 0.6328125,
      "learning_rate": 3.069232642785715e-05,
      "loss": 0.6869,
      "step": 3980
    },
    {
      "epoch": 2.1374450521794572,
      "grad_norm": 0.6875,
      "learning_rate": 3.068410279058355e-05,
      "loss": 1.1633,
      "step": 3981
    },
    {
      "epoch": 2.137981946914533,
      "grad_norm": 0.66015625,
      "learning_rate": 3.067587850468184e-05,
      "loss": 0.9488,
      "step": 3982
    },
    {
      "epoch": 2.138518841649609,
      "grad_norm": 0.6484375,
      "learning_rate": 3.066765357109052e-05,
      "loss": 0.8224,
      "step": 3983
    },
    {
      "epoch": 2.139055736384685,
      "grad_norm": 0.78125,
      "learning_rate": 3.0659427990748154e-05,
      "loss": 1.083,
      "step": 3984
    },
    {
      "epoch": 2.1395926311197613,
      "grad_norm": 0.76171875,
      "learning_rate": 3.065120176459338e-05,
      "loss": 0.8218,
      "step": 3985
    },
    {
      "epoch": 2.140129525854837,
      "grad_norm": 0.703125,
      "learning_rate": 3.0642974893564906e-05,
      "loss": 0.9032,
      "step": 3986
    },
    {
      "epoch": 2.140666420589913,
      "grad_norm": 0.671875,
      "learning_rate": 3.063474737860155e-05,
      "loss": 0.864,
      "step": 3987
    },
    {
      "epoch": 2.141203315324989,
      "grad_norm": 0.69140625,
      "learning_rate": 3.062651922064215e-05,
      "loss": 0.7869,
      "step": 3988
    },
    {
      "epoch": 2.141740210060065,
      "grad_norm": 0.6796875,
      "learning_rate": 3.0618290420625656e-05,
      "loss": 0.8007,
      "step": 3989
    },
    {
      "epoch": 2.142277104795141,
      "grad_norm": 0.60546875,
      "learning_rate": 3.0610060979491076e-05,
      "loss": 0.8238,
      "step": 3990
    },
    {
      "epoch": 2.142813999530217,
      "grad_norm": 0.640625,
      "learning_rate": 3.060183089817749e-05,
      "loss": 1.0372,
      "step": 3991
    },
    {
      "epoch": 2.1433508942652932,
      "grad_norm": 0.75,
      "learning_rate": 3.059360017762407e-05,
      "loss": 0.8935,
      "step": 3992
    },
    {
      "epoch": 2.143887789000369,
      "grad_norm": 0.73828125,
      "learning_rate": 3.058536881877002e-05,
      "loss": 1.0801,
      "step": 3993
    },
    {
      "epoch": 2.144424683735445,
      "grad_norm": 0.58984375,
      "learning_rate": 3.0577136822554645e-05,
      "loss": 0.8193,
      "step": 3994
    },
    {
      "epoch": 2.144961578470521,
      "grad_norm": 0.70703125,
      "learning_rate": 3.0568904189917346e-05,
      "loss": 0.9261,
      "step": 3995
    },
    {
      "epoch": 2.1454984732055973,
      "grad_norm": 0.7578125,
      "learning_rate": 3.056067092179755e-05,
      "loss": 1.0825,
      "step": 3996
    },
    {
      "epoch": 2.146035367940673,
      "grad_norm": 0.77734375,
      "learning_rate": 3.055243701913478e-05,
      "loss": 1.0628,
      "step": 3997
    },
    {
      "epoch": 2.146572262675749,
      "grad_norm": 0.6640625,
      "learning_rate": 3.054420248286863e-05,
      "loss": 0.6996,
      "step": 3998
    },
    {
      "epoch": 2.147109157410825,
      "grad_norm": 0.796875,
      "learning_rate": 3.0535967313938766e-05,
      "loss": 1.0626,
      "step": 3999
    },
    {
      "epoch": 2.1476460521459013,
      "grad_norm": 0.765625,
      "learning_rate": 3.052773151328493e-05,
      "loss": 0.9167,
      "step": 4000
    },
    {
      "epoch": 2.1476460521459013,
      "eval_loss": 0.906206488609314,
      "eval_runtime": 47.0017,
      "eval_samples_per_second": 20.467,
      "eval_steps_per_second": 20.467,
      "step": 4000
    },
    {
      "epoch": 2.148182946880977,
      "grad_norm": 0.59375,
      "learning_rate": 3.051949508184692e-05,
      "loss": 0.7831,
      "step": 4001
    },
    {
      "epoch": 2.148719841616053,
      "grad_norm": 0.70703125,
      "learning_rate": 3.051125802056462e-05,
      "loss": 0.8307,
      "step": 4002
    },
    {
      "epoch": 2.1492567363511292,
      "grad_norm": 0.7421875,
      "learning_rate": 3.0503020330377985e-05,
      "loss": 0.9071,
      "step": 4003
    },
    {
      "epoch": 2.1497936310862054,
      "grad_norm": 0.7421875,
      "learning_rate": 3.0494782012227048e-05,
      "loss": 0.8799,
      "step": 4004
    },
    {
      "epoch": 2.150330525821281,
      "grad_norm": 0.58203125,
      "learning_rate": 3.048654306705189e-05,
      "loss": 0.72,
      "step": 4005
    },
    {
      "epoch": 2.150867420556357,
      "grad_norm": 0.6953125,
      "learning_rate": 3.047830349579269e-05,
      "loss": 0.9297,
      "step": 4006
    },
    {
      "epoch": 2.1514043152914333,
      "grad_norm": 0.8984375,
      "learning_rate": 3.0470063299389684e-05,
      "loss": 0.7495,
      "step": 4007
    },
    {
      "epoch": 2.151941210026509,
      "grad_norm": 0.87890625,
      "learning_rate": 3.0461822478783185e-05,
      "loss": 0.8434,
      "step": 4008
    },
    {
      "epoch": 2.152478104761585,
      "grad_norm": 0.69921875,
      "learning_rate": 3.045358103491357e-05,
      "loss": 0.9524,
      "step": 4009
    },
    {
      "epoch": 2.153014999496661,
      "grad_norm": 0.66796875,
      "learning_rate": 3.0445338968721287e-05,
      "loss": 0.8771,
      "step": 4010
    },
    {
      "epoch": 2.1535518942317373,
      "grad_norm": 0.76171875,
      "learning_rate": 3.0437096281146877e-05,
      "loss": 1.0115,
      "step": 4011
    },
    {
      "epoch": 2.154088788966813,
      "grad_norm": 0.8203125,
      "learning_rate": 3.0428852973130933e-05,
      "loss": 1.0401,
      "step": 4012
    },
    {
      "epoch": 2.154625683701889,
      "grad_norm": 0.73828125,
      "learning_rate": 3.0420609045614103e-05,
      "loss": 0.9135,
      "step": 4013
    },
    {
      "epoch": 2.1551625784369652,
      "grad_norm": 0.75,
      "learning_rate": 3.0412364499537137e-05,
      "loss": 0.9513,
      "step": 4014
    },
    {
      "epoch": 2.1556994731720414,
      "grad_norm": 0.6796875,
      "learning_rate": 3.0404119335840843e-05,
      "loss": 0.9136,
      "step": 4015
    },
    {
      "epoch": 2.156236367907117,
      "grad_norm": 0.64453125,
      "learning_rate": 3.0395873555466088e-05,
      "loss": 0.6858,
      "step": 4016
    },
    {
      "epoch": 2.156773262642193,
      "grad_norm": 0.578125,
      "learning_rate": 3.038762715935383e-05,
      "loss": 0.8168,
      "step": 4017
    },
    {
      "epoch": 2.1573101573772693,
      "grad_norm": 0.73046875,
      "learning_rate": 3.037938014844508e-05,
      "loss": 0.9354,
      "step": 4018
    },
    {
      "epoch": 2.1578470521123454,
      "grad_norm": 0.6328125,
      "learning_rate": 3.037113252368094e-05,
      "loss": 0.818,
      "step": 4019
    },
    {
      "epoch": 2.158383946847421,
      "grad_norm": 0.64453125,
      "learning_rate": 3.0362884286002558e-05,
      "loss": 0.841,
      "step": 4020
    },
    {
      "epoch": 2.158920841582497,
      "grad_norm": 0.78125,
      "learning_rate": 3.035463543635116e-05,
      "loss": 1.0041,
      "step": 4021
    },
    {
      "epoch": 2.1594577363175733,
      "grad_norm": 0.578125,
      "learning_rate": 3.0346385975668047e-05,
      "loss": 0.7238,
      "step": 4022
    },
    {
      "epoch": 2.1599946310526494,
      "grad_norm": 0.6171875,
      "learning_rate": 3.0338135904894594e-05,
      "loss": 0.8383,
      "step": 4023
    },
    {
      "epoch": 2.160531525787725,
      "grad_norm": 0.703125,
      "learning_rate": 3.032988522497222e-05,
      "loss": 1.1517,
      "step": 4024
    },
    {
      "epoch": 2.1610684205228012,
      "grad_norm": 0.59765625,
      "learning_rate": 3.032163393684246e-05,
      "loss": 0.7259,
      "step": 4025
    },
    {
      "epoch": 2.1616053152578774,
      "grad_norm": 0.66796875,
      "learning_rate": 3.0313382041446858e-05,
      "loss": 0.7314,
      "step": 4026
    },
    {
      "epoch": 2.162142209992953,
      "grad_norm": 0.796875,
      "learning_rate": 3.0305129539727085e-05,
      "loss": 0.903,
      "step": 4027
    },
    {
      "epoch": 2.162679104728029,
      "grad_norm": 0.74609375,
      "learning_rate": 3.0296876432624848e-05,
      "loss": 0.8733,
      "step": 4028
    },
    {
      "epoch": 2.1632159994631053,
      "grad_norm": 0.671875,
      "learning_rate": 3.0288622721081933e-05,
      "loss": 0.7457,
      "step": 4029
    },
    {
      "epoch": 2.1637528941981814,
      "grad_norm": 0.82421875,
      "learning_rate": 3.028036840604019e-05,
      "loss": 0.9793,
      "step": 4030
    },
    {
      "epoch": 2.164289788933257,
      "grad_norm": 0.73828125,
      "learning_rate": 3.027211348844154e-05,
      "loss": 0.913,
      "step": 4031
    },
    {
      "epoch": 2.164826683668333,
      "grad_norm": 0.6015625,
      "learning_rate": 3.0263857969227966e-05,
      "loss": 0.7289,
      "step": 4032
    },
    {
      "epoch": 2.1653635784034093,
      "grad_norm": 0.71875,
      "learning_rate": 3.0255601849341554e-05,
      "loss": 0.8946,
      "step": 4033
    },
    {
      "epoch": 2.1659004731384854,
      "grad_norm": 0.7265625,
      "learning_rate": 3.02473451297244e-05,
      "loss": 1.0224,
      "step": 4034
    },
    {
      "epoch": 2.166437367873561,
      "grad_norm": 0.6171875,
      "learning_rate": 3.0239087811318715e-05,
      "loss": 0.7468,
      "step": 4035
    },
    {
      "epoch": 2.1669742626086372,
      "grad_norm": 0.66796875,
      "learning_rate": 3.023082989506677e-05,
      "loss": 0.7348,
      "step": 4036
    },
    {
      "epoch": 2.1675111573437134,
      "grad_norm": 0.6640625,
      "learning_rate": 3.0222571381910884e-05,
      "loss": 0.8514,
      "step": 4037
    },
    {
      "epoch": 2.1680480520787895,
      "grad_norm": 0.65625,
      "learning_rate": 3.0214312272793472e-05,
      "loss": 0.8658,
      "step": 4038
    },
    {
      "epoch": 2.168584946813865,
      "grad_norm": 0.765625,
      "learning_rate": 3.0206052568656996e-05,
      "loss": 1.0283,
      "step": 4039
    },
    {
      "epoch": 2.1691218415489413,
      "grad_norm": 0.6015625,
      "learning_rate": 3.0197792270443982e-05,
      "loss": 0.7336,
      "step": 4040
    },
    {
      "epoch": 2.1696587362840174,
      "grad_norm": 0.61328125,
      "learning_rate": 3.0189531379097057e-05,
      "loss": 0.7364,
      "step": 4041
    },
    {
      "epoch": 2.1701956310190935,
      "grad_norm": 0.65234375,
      "learning_rate": 3.018126989555888e-05,
      "loss": 0.774,
      "step": 4042
    },
    {
      "epoch": 2.170732525754169,
      "grad_norm": 0.6953125,
      "learning_rate": 3.017300782077218e-05,
      "loss": 0.8639,
      "step": 4043
    },
    {
      "epoch": 2.1712694204892453,
      "grad_norm": 0.640625,
      "learning_rate": 3.016474515567979e-05,
      "loss": 0.8877,
      "step": 4044
    },
    {
      "epoch": 2.1718063152243214,
      "grad_norm": 0.75,
      "learning_rate": 3.015648190122457e-05,
      "loss": 0.7921,
      "step": 4045
    },
    {
      "epoch": 2.172343209959397,
      "grad_norm": 0.7265625,
      "learning_rate": 3.014821805834946e-05,
      "loss": 0.8884,
      "step": 4046
    },
    {
      "epoch": 2.1728801046944732,
      "grad_norm": 0.64453125,
      "learning_rate": 3.0139953627997475e-05,
      "loss": 0.856,
      "step": 4047
    },
    {
      "epoch": 2.1734169994295494,
      "grad_norm": 0.75,
      "learning_rate": 3.0131688611111685e-05,
      "loss": 0.892,
      "step": 4048
    },
    {
      "epoch": 2.1739538941646255,
      "grad_norm": 0.70703125,
      "learning_rate": 3.0123423008635236e-05,
      "loss": 0.9259,
      "step": 4049
    },
    {
      "epoch": 2.174490788899701,
      "grad_norm": 0.65625,
      "learning_rate": 3.011515682151135e-05,
      "loss": 0.8287,
      "step": 4050
    },
    {
      "epoch": 2.174490788899701,
      "eval_loss": 0.9057601690292358,
      "eval_runtime": 46.7747,
      "eval_samples_per_second": 20.567,
      "eval_steps_per_second": 20.567,
      "step": 4050
    },
    {
      "epoch": 2.1750276836347773,
      "grad_norm": 0.70703125,
      "learning_rate": 3.0106890050683278e-05,
      "loss": 1.0291,
      "step": 4051
    },
    {
      "epoch": 2.1755645783698534,
      "grad_norm": 0.7109375,
      "learning_rate": 3.0098622697094387e-05,
      "loss": 0.8809,
      "step": 4052
    },
    {
      "epoch": 2.1761014731049295,
      "grad_norm": 0.62109375,
      "learning_rate": 3.009035476168808e-05,
      "loss": 0.7938,
      "step": 4053
    },
    {
      "epoch": 2.176638367840005,
      "grad_norm": 0.6875,
      "learning_rate": 3.008208624540782e-05,
      "loss": 0.8107,
      "step": 4054
    },
    {
      "epoch": 2.1771752625750813,
      "grad_norm": 0.64453125,
      "learning_rate": 3.007381714919717e-05,
      "loss": 0.5469,
      "step": 4055
    },
    {
      "epoch": 2.1777121573101574,
      "grad_norm": 0.75390625,
      "learning_rate": 3.0065547473999723e-05,
      "loss": 1.1463,
      "step": 4056
    },
    {
      "epoch": 2.1782490520452336,
      "grad_norm": 0.703125,
      "learning_rate": 3.0057277220759156e-05,
      "loss": 1.0264,
      "step": 4057
    },
    {
      "epoch": 2.1787859467803092,
      "grad_norm": 0.73828125,
      "learning_rate": 3.004900639041922e-05,
      "loss": 0.775,
      "step": 4058
    },
    {
      "epoch": 2.1793228415153854,
      "grad_norm": 1.25,
      "learning_rate": 3.004073498392371e-05,
      "loss": 0.9768,
      "step": 4059
    },
    {
      "epoch": 2.1798597362504615,
      "grad_norm": 0.796875,
      "learning_rate": 3.0032463002216505e-05,
      "loss": 0.918,
      "step": 4060
    },
    {
      "epoch": 2.1803966309855376,
      "grad_norm": 0.67578125,
      "learning_rate": 3.002419044624154e-05,
      "loss": 0.8603,
      "step": 4061
    },
    {
      "epoch": 2.1809335257206133,
      "grad_norm": 0.79296875,
      "learning_rate": 3.001591731694282e-05,
      "loss": 0.9137,
      "step": 4062
    },
    {
      "epoch": 2.1814704204556894,
      "grad_norm": 0.60546875,
      "learning_rate": 3.000764361526441e-05,
      "loss": 0.8619,
      "step": 4063
    },
    {
      "epoch": 2.1820073151907655,
      "grad_norm": 0.8125,
      "learning_rate": 2.9999369342150446e-05,
      "loss": 0.7769,
      "step": 4064
    },
    {
      "epoch": 2.182544209925841,
      "grad_norm": 0.84375,
      "learning_rate": 2.9991094498545126e-05,
      "loss": 0.8459,
      "step": 4065
    },
    {
      "epoch": 2.1830811046609173,
      "grad_norm": 0.84765625,
      "learning_rate": 2.998281908539272e-05,
      "loss": 0.9062,
      "step": 4066
    },
    {
      "epoch": 2.1836179993959934,
      "grad_norm": 0.8828125,
      "learning_rate": 2.9974543103637558e-05,
      "loss": 0.9375,
      "step": 4067
    },
    {
      "epoch": 2.1841548941310696,
      "grad_norm": 0.70703125,
      "learning_rate": 2.9966266554224027e-05,
      "loss": 0.7708,
      "step": 4068
    },
    {
      "epoch": 2.1846917888661452,
      "grad_norm": 0.703125,
      "learning_rate": 2.995798943809659e-05,
      "loss": 0.9111,
      "step": 4069
    },
    {
      "epoch": 2.1852286836012214,
      "grad_norm": 0.609375,
      "learning_rate": 2.9949711756199772e-05,
      "loss": 0.6897,
      "step": 4070
    },
    {
      "epoch": 2.1857655783362975,
      "grad_norm": 0.64453125,
      "learning_rate": 2.9941433509478156e-05,
      "loss": 0.8613,
      "step": 4071
    },
    {
      "epoch": 2.1863024730713736,
      "grad_norm": 0.7578125,
      "learning_rate": 2.9933154698876403e-05,
      "loss": 1.0753,
      "step": 4072
    },
    {
      "epoch": 2.1868393678064493,
      "grad_norm": 0.85546875,
      "learning_rate": 2.992487532533922e-05,
      "loss": 0.9237,
      "step": 4073
    },
    {
      "epoch": 2.1873762625415254,
      "grad_norm": 0.69140625,
      "learning_rate": 2.9916595389811397e-05,
      "loss": 0.8755,
      "step": 4074
    },
    {
      "epoch": 2.1879131572766015,
      "grad_norm": 0.73046875,
      "learning_rate": 2.9908314893237776e-05,
      "loss": 0.8316,
      "step": 4075
    },
    {
      "epoch": 2.1884500520116776,
      "grad_norm": 0.68359375,
      "learning_rate": 2.9900033836563263e-05,
      "loss": 0.9919,
      "step": 4076
    },
    {
      "epoch": 2.1889869467467533,
      "grad_norm": 0.74609375,
      "learning_rate": 2.9891752220732838e-05,
      "loss": 0.9578,
      "step": 4077
    },
    {
      "epoch": 2.1895238414818294,
      "grad_norm": 0.62109375,
      "learning_rate": 2.9883470046691543e-05,
      "loss": 0.8043,
      "step": 4078
    },
    {
      "epoch": 2.1900607362169056,
      "grad_norm": 0.63671875,
      "learning_rate": 2.987518731538446e-05,
      "loss": 0.7975,
      "step": 4079
    },
    {
      "epoch": 2.1905976309519817,
      "grad_norm": 0.70703125,
      "learning_rate": 2.986690402775677e-05,
      "loss": 0.9073,
      "step": 4080
    },
    {
      "epoch": 2.1911345256870574,
      "grad_norm": 0.77734375,
      "learning_rate": 2.9858620184753695e-05,
      "loss": 0.8354,
      "step": 4081
    },
    {
      "epoch": 2.1916714204221335,
      "grad_norm": 0.5859375,
      "learning_rate": 2.9850335787320526e-05,
      "loss": 0.7135,
      "step": 4082
    },
    {
      "epoch": 2.1922083151572096,
      "grad_norm": 0.6640625,
      "learning_rate": 2.984205083640263e-05,
      "loss": 0.6937,
      "step": 4083
    },
    {
      "epoch": 2.1927452098922853,
      "grad_norm": 0.61328125,
      "learning_rate": 2.9833765332945402e-05,
      "loss": 0.7958,
      "step": 4084
    },
    {
      "epoch": 2.1932821046273614,
      "grad_norm": 0.90625,
      "learning_rate": 2.9825479277894343e-05,
      "loss": 1.0131,
      "step": 4085
    },
    {
      "epoch": 2.1938189993624375,
      "grad_norm": 0.7265625,
      "learning_rate": 2.9817192672194984e-05,
      "loss": 0.8615,
      "step": 4086
    },
    {
      "epoch": 2.1943558940975136,
      "grad_norm": 0.69140625,
      "learning_rate": 2.980890551679293e-05,
      "loss": 0.8079,
      "step": 4087
    },
    {
      "epoch": 2.1948927888325893,
      "grad_norm": 0.66796875,
      "learning_rate": 2.980061781263387e-05,
      "loss": 0.7669,
      "step": 4088
    },
    {
      "epoch": 2.1954296835676654,
      "grad_norm": 0.7109375,
      "learning_rate": 2.9792329560663507e-05,
      "loss": 0.721,
      "step": 4089
    },
    {
      "epoch": 2.1959665783027416,
      "grad_norm": 0.66015625,
      "learning_rate": 2.9784040761827658e-05,
      "loss": 0.8621,
      "step": 4090
    },
    {
      "epoch": 2.1965034730378177,
      "grad_norm": 0.7421875,
      "learning_rate": 2.977575141707218e-05,
      "loss": 1.0204,
      "step": 4091
    },
    {
      "epoch": 2.1970403677728934,
      "grad_norm": 0.63671875,
      "learning_rate": 2.9767461527342983e-05,
      "loss": 0.8186,
      "step": 4092
    },
    {
      "epoch": 2.1975772625079695,
      "grad_norm": 0.74609375,
      "learning_rate": 2.9759171093586047e-05,
      "loss": 1.0321,
      "step": 4093
    },
    {
      "epoch": 2.1981141572430456,
      "grad_norm": 0.6640625,
      "learning_rate": 2.975088011674742e-05,
      "loss": 0.8158,
      "step": 4094
    },
    {
      "epoch": 2.1986510519781217,
      "grad_norm": 0.87109375,
      "learning_rate": 2.9742588597773204e-05,
      "loss": 0.9527,
      "step": 4095
    },
    {
      "epoch": 2.1991879467131974,
      "grad_norm": 0.6953125,
      "learning_rate": 2.9734296537609575e-05,
      "loss": 0.773,
      "step": 4096
    },
    {
      "epoch": 2.1997248414482735,
      "grad_norm": 0.6484375,
      "learning_rate": 2.9726003937202757e-05,
      "loss": 0.7991,
      "step": 4097
    },
    {
      "epoch": 2.2002617361833496,
      "grad_norm": 0.70703125,
      "learning_rate": 2.9717710797499025e-05,
      "loss": 0.7457,
      "step": 4098
    },
    {
      "epoch": 2.2007986309184258,
      "grad_norm": 0.6796875,
      "learning_rate": 2.970941711944476e-05,
      "loss": 0.827,
      "step": 4099
    },
    {
      "epoch": 2.2013355256535014,
      "grad_norm": 0.8046875,
      "learning_rate": 2.970112290398636e-05,
      "loss": 0.866,
      "step": 4100
    },
    {
      "epoch": 2.2013355256535014,
      "eval_loss": 0.9053437113761902,
      "eval_runtime": 46.5428,
      "eval_samples_per_second": 20.669,
      "eval_steps_per_second": 20.669,
      "step": 4100
    },
    {
      "epoch": 2.2018724203885776,
      "grad_norm": 0.65625,
      "learning_rate": 2.96928281520703e-05,
      "loss": 0.9705,
      "step": 4101
    },
    {
      "epoch": 2.2024093151236537,
      "grad_norm": 0.6484375,
      "learning_rate": 2.9684532864643122e-05,
      "loss": 0.7725,
      "step": 4102
    },
    {
      "epoch": 2.2029462098587294,
      "grad_norm": 0.83984375,
      "learning_rate": 2.9676237042651412e-05,
      "loss": 1.1466,
      "step": 4103
    },
    {
      "epoch": 2.2034831045938055,
      "grad_norm": 0.70703125,
      "learning_rate": 2.966794068704184e-05,
      "loss": 0.9232,
      "step": 4104
    },
    {
      "epoch": 2.2040199993288816,
      "grad_norm": 0.83203125,
      "learning_rate": 2.9659643798761123e-05,
      "loss": 0.8079,
      "step": 4105
    },
    {
      "epoch": 2.2045568940639577,
      "grad_norm": 0.63671875,
      "learning_rate": 2.9651346378756028e-05,
      "loss": 0.7778,
      "step": 4106
    },
    {
      "epoch": 2.2050937887990334,
      "grad_norm": 0.796875,
      "learning_rate": 2.9643048427973417e-05,
      "loss": 0.8648,
      "step": 4107
    },
    {
      "epoch": 2.2056306835341095,
      "grad_norm": 0.7734375,
      "learning_rate": 2.9634749947360178e-05,
      "loss": 0.993,
      "step": 4108
    },
    {
      "epoch": 2.2061675782691856,
      "grad_norm": 0.71484375,
      "learning_rate": 2.962645093786328e-05,
      "loss": 0.8057,
      "step": 4109
    },
    {
      "epoch": 2.2067044730042618,
      "grad_norm": 0.66796875,
      "learning_rate": 2.961815140042974e-05,
      "loss": 0.7218,
      "step": 4110
    },
    {
      "epoch": 2.2072413677393374,
      "grad_norm": 1.0703125,
      "learning_rate": 2.9609851336006632e-05,
      "loss": 0.9607,
      "step": 4111
    },
    {
      "epoch": 2.2077782624744136,
      "grad_norm": 0.79296875,
      "learning_rate": 2.9601550745541112e-05,
      "loss": 0.9372,
      "step": 4112
    },
    {
      "epoch": 2.2083151572094897,
      "grad_norm": 0.6796875,
      "learning_rate": 2.9593249629980386e-05,
      "loss": 0.919,
      "step": 4113
    },
    {
      "epoch": 2.208852051944566,
      "grad_norm": 0.8203125,
      "learning_rate": 2.95849479902717e-05,
      "loss": 0.9681,
      "step": 4114
    },
    {
      "epoch": 2.2093889466796415,
      "grad_norm": 0.62890625,
      "learning_rate": 2.9576645827362386e-05,
      "loss": 0.8681,
      "step": 4115
    },
    {
      "epoch": 2.2099258414147176,
      "grad_norm": 0.81640625,
      "learning_rate": 2.9568343142199824e-05,
      "loss": 0.8769,
      "step": 4116
    },
    {
      "epoch": 2.2104627361497937,
      "grad_norm": 0.76953125,
      "learning_rate": 2.956003993573146e-05,
      "loss": 0.9149,
      "step": 4117
    },
    {
      "epoch": 2.21099963088487,
      "grad_norm": 0.69140625,
      "learning_rate": 2.955173620890479e-05,
      "loss": 0.8632,
      "step": 4118
    },
    {
      "epoch": 2.2115365256199455,
      "grad_norm": 0.703125,
      "learning_rate": 2.9543431962667372e-05,
      "loss": 0.8461,
      "step": 4119
    },
    {
      "epoch": 2.2120734203550216,
      "grad_norm": 0.65234375,
      "learning_rate": 2.9535127197966828e-05,
      "loss": 0.8977,
      "step": 4120
    },
    {
      "epoch": 2.2126103150900978,
      "grad_norm": 0.72265625,
      "learning_rate": 2.9526821915750847e-05,
      "loss": 0.9606,
      "step": 4121
    },
    {
      "epoch": 2.2131472098251734,
      "grad_norm": 0.65234375,
      "learning_rate": 2.9518516116967145e-05,
      "loss": 0.8072,
      "step": 4122
    },
    {
      "epoch": 2.2136841045602496,
      "grad_norm": 0.6328125,
      "learning_rate": 2.9510209802563537e-05,
      "loss": 0.843,
      "step": 4123
    },
    {
      "epoch": 2.2142209992953257,
      "grad_norm": 0.84375,
      "learning_rate": 2.9501902973487877e-05,
      "loss": 1.114,
      "step": 4124
    },
    {
      "epoch": 2.214757894030402,
      "grad_norm": 0.671875,
      "learning_rate": 2.9493595630688076e-05,
      "loss": 0.8705,
      "step": 4125
    },
    {
      "epoch": 2.2152947887654775,
      "grad_norm": 1.3671875,
      "learning_rate": 2.9485287775112103e-05,
      "loss": 0.9069,
      "step": 4126
    },
    {
      "epoch": 2.2158316835005536,
      "grad_norm": 0.73046875,
      "learning_rate": 2.9476979407707995e-05,
      "loss": 0.8303,
      "step": 4127
    },
    {
      "epoch": 2.2163685782356297,
      "grad_norm": 0.65625,
      "learning_rate": 2.9468670529423835e-05,
      "loss": 0.9907,
      "step": 4128
    },
    {
      "epoch": 2.216905472970706,
      "grad_norm": 0.7890625,
      "learning_rate": 2.9460361141207782e-05,
      "loss": 1.0083,
      "step": 4129
    },
    {
      "epoch": 2.2174423677057815,
      "grad_norm": 0.70703125,
      "learning_rate": 2.945205124400804e-05,
      "loss": 0.7639,
      "step": 4130
    },
    {
      "epoch": 2.2179792624408576,
      "grad_norm": 0.578125,
      "learning_rate": 2.9443740838772865e-05,
      "loss": 0.7762,
      "step": 4131
    },
    {
      "epoch": 2.2185161571759338,
      "grad_norm": 0.65625,
      "learning_rate": 2.9435429926450586e-05,
      "loss": 0.8291,
      "step": 4132
    },
    {
      "epoch": 2.21905305191101,
      "grad_norm": 0.65234375,
      "learning_rate": 2.9427118507989586e-05,
      "loss": 0.7603,
      "step": 4133
    },
    {
      "epoch": 2.2195899466460856,
      "grad_norm": 0.77734375,
      "learning_rate": 2.94188065843383e-05,
      "loss": 1.0706,
      "step": 4134
    },
    {
      "epoch": 2.2201268413811617,
      "grad_norm": 0.6875,
      "learning_rate": 2.9410494156445216e-05,
      "loss": 0.9553,
      "step": 4135
    },
    {
      "epoch": 2.220663736116238,
      "grad_norm": 0.82421875,
      "learning_rate": 2.9402181225258894e-05,
      "loss": 0.7941,
      "step": 4136
    },
    {
      "epoch": 2.221200630851314,
      "grad_norm": 0.671875,
      "learning_rate": 2.9393867791727947e-05,
      "loss": 0.895,
      "step": 4137
    },
    {
      "epoch": 2.2217375255863896,
      "grad_norm": 0.56640625,
      "learning_rate": 2.938555385680105e-05,
      "loss": 0.6571,
      "step": 4138
    },
    {
      "epoch": 2.2222744203214657,
      "grad_norm": 0.7890625,
      "learning_rate": 2.9377239421426905e-05,
      "loss": 0.9236,
      "step": 4139
    },
    {
      "epoch": 2.222811315056542,
      "grad_norm": 0.78125,
      "learning_rate": 2.9368924486554317e-05,
      "loss": 0.9655,
      "step": 4140
    },
    {
      "epoch": 2.2233482097916175,
      "grad_norm": 0.75,
      "learning_rate": 2.9360609053132116e-05,
      "loss": 0.901,
      "step": 4141
    },
    {
      "epoch": 2.2238851045266936,
      "grad_norm": 0.6328125,
      "learning_rate": 2.9352293122109202e-05,
      "loss": 0.7299,
      "step": 4142
    },
    {
      "epoch": 2.2244219992617698,
      "grad_norm": 0.55859375,
      "learning_rate": 2.934397669443452e-05,
      "loss": 0.6925,
      "step": 4143
    },
    {
      "epoch": 2.224958893996846,
      "grad_norm": 0.70703125,
      "learning_rate": 2.933565977105709e-05,
      "loss": 0.8147,
      "step": 4144
    },
    {
      "epoch": 2.2254957887319216,
      "grad_norm": 0.921875,
      "learning_rate": 2.932734235292597e-05,
      "loss": 1.0086,
      "step": 4145
    },
    {
      "epoch": 2.2260326834669977,
      "grad_norm": 0.69921875,
      "learning_rate": 2.9319024440990296e-05,
      "loss": 1.1011,
      "step": 4146
    },
    {
      "epoch": 2.226569578202074,
      "grad_norm": 0.7109375,
      "learning_rate": 2.931070603619922e-05,
      "loss": 0.9351,
      "step": 4147
    },
    {
      "epoch": 2.22710647293715,
      "grad_norm": 0.65234375,
      "learning_rate": 2.9302387139502003e-05,
      "loss": 0.8345,
      "step": 4148
    },
    {
      "epoch": 2.2276433676722256,
      "grad_norm": 0.71484375,
      "learning_rate": 2.9294067751847933e-05,
      "loss": 0.9992,
      "step": 4149
    },
    {
      "epoch": 2.2281802624073017,
      "grad_norm": 0.640625,
      "learning_rate": 2.9285747874186342e-05,
      "loss": 0.9124,
      "step": 4150
    },
    {
      "epoch": 2.2281802624073017,
      "eval_loss": 0.9054576754570007,
      "eval_runtime": 46.6126,
      "eval_samples_per_second": 20.638,
      "eval_steps_per_second": 20.638,
      "step": 4150
    },
    {
      "epoch": 2.228717157142378,
      "grad_norm": 0.8125,
      "learning_rate": 2.9277427507466654e-05,
      "loss": 0.8543,
      "step": 4151
    },
    {
      "epoch": 2.229254051877454,
      "grad_norm": 0.54296875,
      "learning_rate": 2.926910665263831e-05,
      "loss": 0.6306,
      "step": 4152
    },
    {
      "epoch": 2.2297909466125296,
      "grad_norm": 0.62890625,
      "learning_rate": 2.926078531065084e-05,
      "loss": 0.7358,
      "step": 4153
    },
    {
      "epoch": 2.2303278413476058,
      "grad_norm": 0.69140625,
      "learning_rate": 2.9252463482453802e-05,
      "loss": 0.8626,
      "step": 4154
    },
    {
      "epoch": 2.230864736082682,
      "grad_norm": 0.64453125,
      "learning_rate": 2.924414116899683e-05,
      "loss": 0.8718,
      "step": 4155
    },
    {
      "epoch": 2.231401630817758,
      "grad_norm": 0.73046875,
      "learning_rate": 2.9235818371229606e-05,
      "loss": 0.7792,
      "step": 4156
    },
    {
      "epoch": 2.2319385255528337,
      "grad_norm": 0.65234375,
      "learning_rate": 2.9227495090101863e-05,
      "loss": 0.7288,
      "step": 4157
    },
    {
      "epoch": 2.23247542028791,
      "grad_norm": 0.6484375,
      "learning_rate": 2.921917132656339e-05,
      "loss": 0.9498,
      "step": 4158
    },
    {
      "epoch": 2.233012315022986,
      "grad_norm": 0.7109375,
      "learning_rate": 2.9210847081564042e-05,
      "loss": 0.9954,
      "step": 4159
    },
    {
      "epoch": 2.2335492097580616,
      "grad_norm": 0.6875,
      "learning_rate": 2.920252235605371e-05,
      "loss": 0.8263,
      "step": 4160
    },
    {
      "epoch": 2.2340861044931377,
      "grad_norm": 0.56640625,
      "learning_rate": 2.9194197150982357e-05,
      "loss": 0.7274,
      "step": 4161
    },
    {
      "epoch": 2.234622999228214,
      "grad_norm": 0.62890625,
      "learning_rate": 2.918587146729999e-05,
      "loss": 0.7906,
      "step": 4162
    },
    {
      "epoch": 2.23515989396329,
      "grad_norm": 0.73046875,
      "learning_rate": 2.917754530595669e-05,
      "loss": 0.926,
      "step": 4163
    },
    {
      "epoch": 2.2356967886983656,
      "grad_norm": 0.6171875,
      "learning_rate": 2.916921866790256e-05,
      "loss": 0.7595,
      "step": 4164
    },
    {
      "epoch": 2.2362336834334418,
      "grad_norm": 0.859375,
      "learning_rate": 2.916089155408778e-05,
      "loss": 0.9828,
      "step": 4165
    },
    {
      "epoch": 2.236770578168518,
      "grad_norm": 0.60546875,
      "learning_rate": 2.915256396546258e-05,
      "loss": 0.7342,
      "step": 4166
    },
    {
      "epoch": 2.237307472903594,
      "grad_norm": 0.83203125,
      "learning_rate": 2.914423590297725e-05,
      "loss": 0.8831,
      "step": 4167
    },
    {
      "epoch": 2.2378443676386697,
      "grad_norm": 0.74609375,
      "learning_rate": 2.9135907367582117e-05,
      "loss": 0.9026,
      "step": 4168
    },
    {
      "epoch": 2.238381262373746,
      "grad_norm": 0.63671875,
      "learning_rate": 2.912757836022757e-05,
      "loss": 0.8294,
      "step": 4169
    },
    {
      "epoch": 2.238918157108822,
      "grad_norm": 0.66015625,
      "learning_rate": 2.911924888186407e-05,
      "loss": 0.868,
      "step": 4170
    },
    {
      "epoch": 2.239455051843898,
      "grad_norm": 0.72265625,
      "learning_rate": 2.9110918933442104e-05,
      "loss": 0.8918,
      "step": 4171
    },
    {
      "epoch": 2.2399919465789737,
      "grad_norm": 0.68359375,
      "learning_rate": 2.9102588515912225e-05,
      "loss": 0.8856,
      "step": 4172
    },
    {
      "epoch": 2.24052884131405,
      "grad_norm": 0.73828125,
      "learning_rate": 2.909425763022504e-05,
      "loss": 1.046,
      "step": 4173
    },
    {
      "epoch": 2.241065736049126,
      "grad_norm": 0.75390625,
      "learning_rate": 2.9085926277331214e-05,
      "loss": 0.9102,
      "step": 4174
    },
    {
      "epoch": 2.241602630784202,
      "grad_norm": 0.59375,
      "learning_rate": 2.907759445818145e-05,
      "loss": 0.7687,
      "step": 4175
    },
    {
      "epoch": 2.2421395255192778,
      "grad_norm": 0.69140625,
      "learning_rate": 2.906926217372653e-05,
      "loss": 0.9139,
      "step": 4176
    },
    {
      "epoch": 2.242676420254354,
      "grad_norm": 0.65625,
      "learning_rate": 2.9060929424917254e-05,
      "loss": 0.9126,
      "step": 4177
    },
    {
      "epoch": 2.24321331498943,
      "grad_norm": 0.7421875,
      "learning_rate": 2.9052596212704507e-05,
      "loss": 1.0514,
      "step": 4178
    },
    {
      "epoch": 2.2437502097245057,
      "grad_norm": 0.62890625,
      "learning_rate": 2.9044262538039206e-05,
      "loss": 0.8539,
      "step": 4179
    },
    {
      "epoch": 2.244287104459582,
      "grad_norm": 0.73046875,
      "learning_rate": 2.9035928401872336e-05,
      "loss": 1.0017,
      "step": 4180
    },
    {
      "epoch": 2.244823999194658,
      "grad_norm": 0.5859375,
      "learning_rate": 2.9027593805154927e-05,
      "loss": 0.7458,
      "step": 4181
    },
    {
      "epoch": 2.245360893929734,
      "grad_norm": 0.91796875,
      "learning_rate": 2.9019258748838058e-05,
      "loss": 0.7076,
      "step": 4182
    },
    {
      "epoch": 2.2458977886648097,
      "grad_norm": 0.796875,
      "learning_rate": 2.901092323387286e-05,
      "loss": 0.9812,
      "step": 4183
    },
    {
      "epoch": 2.246434683399886,
      "grad_norm": 0.6328125,
      "learning_rate": 2.9002587261210534e-05,
      "loss": 0.8256,
      "step": 4184
    },
    {
      "epoch": 2.246971578134962,
      "grad_norm": 0.6796875,
      "learning_rate": 2.8994250831802307e-05,
      "loss": 0.7752,
      "step": 4185
    },
    {
      "epoch": 2.247508472870038,
      "grad_norm": 0.60546875,
      "learning_rate": 2.898591394659948e-05,
      "loss": 0.5932,
      "step": 4186
    },
    {
      "epoch": 2.2480453676051138,
      "grad_norm": 0.6015625,
      "learning_rate": 2.8977576606553398e-05,
      "loss": 0.6987,
      "step": 4187
    },
    {
      "epoch": 2.24858226234019,
      "grad_norm": 0.98046875,
      "learning_rate": 2.8969238812615452e-05,
      "loss": 1.0829,
      "step": 4188
    },
    {
      "epoch": 2.249119157075266,
      "grad_norm": 0.6171875,
      "learning_rate": 2.8960900565737086e-05,
      "loss": 0.8561,
      "step": 4189
    },
    {
      "epoch": 2.249656051810342,
      "grad_norm": 0.703125,
      "learning_rate": 2.895256186686981e-05,
      "loss": 0.7907,
      "step": 4190
    },
    {
      "epoch": 2.250192946545418,
      "grad_norm": 0.78125,
      "learning_rate": 2.8944222716965163e-05,
      "loss": 1.0313,
      "step": 4191
    },
    {
      "epoch": 2.250729841280494,
      "grad_norm": 0.79296875,
      "learning_rate": 2.893588311697476e-05,
      "loss": 0.9696,
      "step": 4192
    },
    {
      "epoch": 2.25126673601557,
      "grad_norm": 0.6484375,
      "learning_rate": 2.8927543067850248e-05,
      "loss": 0.6992,
      "step": 4193
    },
    {
      "epoch": 2.251803630750646,
      "grad_norm": 0.69921875,
      "learning_rate": 2.8919202570543336e-05,
      "loss": 0.7765,
      "step": 4194
    },
    {
      "epoch": 2.252340525485722,
      "grad_norm": 0.68359375,
      "learning_rate": 2.8910861626005776e-05,
      "loss": 0.6895,
      "step": 4195
    },
    {
      "epoch": 2.252877420220798,
      "grad_norm": 0.609375,
      "learning_rate": 2.890252023518938e-05,
      "loss": 0.7656,
      "step": 4196
    },
    {
      "epoch": 2.253414314955874,
      "grad_norm": 0.671875,
      "learning_rate": 2.8894178399046002e-05,
      "loss": 0.6924,
      "step": 4197
    },
    {
      "epoch": 2.2539512096909498,
      "grad_norm": 0.5390625,
      "learning_rate": 2.888583611852755e-05,
      "loss": 0.7085,
      "step": 4198
    },
    {
      "epoch": 2.254488104426026,
      "grad_norm": 0.6171875,
      "learning_rate": 2.8877493394585982e-05,
      "loss": 0.817,
      "step": 4199
    },
    {
      "epoch": 2.255024999161102,
      "grad_norm": 0.6796875,
      "learning_rate": 2.886915022817332e-05,
      "loss": 0.722,
      "step": 4200
    },
    {
      "epoch": 2.255024999161102,
      "eval_loss": 0.9057018160820007,
      "eval_runtime": 46.7193,
      "eval_samples_per_second": 20.591,
      "eval_steps_per_second": 20.591,
      "step": 4200
    },
    {
      "epoch": 2.255561893896178,
      "grad_norm": 0.57421875,
      "learning_rate": 2.8860806620241627e-05,
      "loss": 0.9053,
      "step": 4201
    },
    {
      "epoch": 2.256098788631254,
      "grad_norm": 0.7265625,
      "learning_rate": 2.885246257174299e-05,
      "loss": 0.9915,
      "step": 4202
    },
    {
      "epoch": 2.25663568336633,
      "grad_norm": 0.703125,
      "learning_rate": 2.8844118083629596e-05,
      "loss": 0.9003,
      "step": 4203
    },
    {
      "epoch": 2.257172578101406,
      "grad_norm": 0.78515625,
      "learning_rate": 2.883577315685364e-05,
      "loss": 0.9268,
      "step": 4204
    },
    {
      "epoch": 2.257709472836482,
      "grad_norm": 0.54296875,
      "learning_rate": 2.8827427792367394e-05,
      "loss": 0.7192,
      "step": 4205
    },
    {
      "epoch": 2.258246367571558,
      "grad_norm": 0.6640625,
      "learning_rate": 2.881908199112317e-05,
      "loss": 0.8311,
      "step": 4206
    },
    {
      "epoch": 2.258783262306634,
      "grad_norm": 0.578125,
      "learning_rate": 2.8810735754073315e-05,
      "loss": 0.5093,
      "step": 4207
    },
    {
      "epoch": 2.25932015704171,
      "grad_norm": 0.6875,
      "learning_rate": 2.880238908217026e-05,
      "loss": 0.851,
      "step": 4208
    },
    {
      "epoch": 2.259857051776786,
      "grad_norm": 0.921875,
      "learning_rate": 2.8794041976366463e-05,
      "loss": 0.7708,
      "step": 4209
    },
    {
      "epoch": 2.260393946511862,
      "grad_norm": 0.66796875,
      "learning_rate": 2.878569443761442e-05,
      "loss": 0.8941,
      "step": 4210
    },
    {
      "epoch": 2.260930841246938,
      "grad_norm": 0.6796875,
      "learning_rate": 2.87773464668667e-05,
      "loss": 0.8557,
      "step": 4211
    },
    {
      "epoch": 2.261467735982014,
      "grad_norm": 0.6328125,
      "learning_rate": 2.8768998065075915e-05,
      "loss": 0.7612,
      "step": 4212
    },
    {
      "epoch": 2.2620046307170902,
      "grad_norm": 0.66015625,
      "learning_rate": 2.8760649233194714e-05,
      "loss": 0.8401,
      "step": 4213
    },
    {
      "epoch": 2.262541525452166,
      "grad_norm": 0.6796875,
      "learning_rate": 2.8752299972175824e-05,
      "loss": 0.8999,
      "step": 4214
    },
    {
      "epoch": 2.263078420187242,
      "grad_norm": 0.66796875,
      "learning_rate": 2.8743950282971977e-05,
      "loss": 0.8894,
      "step": 4215
    },
    {
      "epoch": 2.263615314922318,
      "grad_norm": 0.66015625,
      "learning_rate": 2.873560016653599e-05,
      "loss": 0.8505,
      "step": 4216
    },
    {
      "epoch": 2.264152209657394,
      "grad_norm": 0.6796875,
      "learning_rate": 2.8727249623820717e-05,
      "loss": 0.893,
      "step": 4217
    },
    {
      "epoch": 2.26468910439247,
      "grad_norm": 0.671875,
      "learning_rate": 2.8718898655779066e-05,
      "loss": 0.7213,
      "step": 4218
    },
    {
      "epoch": 2.265225999127546,
      "grad_norm": 0.77734375,
      "learning_rate": 2.871054726336398e-05,
      "loss": 1.1274,
      "step": 4219
    },
    {
      "epoch": 2.265762893862622,
      "grad_norm": 0.7421875,
      "learning_rate": 2.8702195447528456e-05,
      "loss": 1.0309,
      "step": 4220
    },
    {
      "epoch": 2.266299788597698,
      "grad_norm": 0.6484375,
      "learning_rate": 2.8693843209225545e-05,
      "loss": 0.9051,
      "step": 4221
    },
    {
      "epoch": 2.266836683332774,
      "grad_norm": 0.703125,
      "learning_rate": 2.8685490549408363e-05,
      "loss": 0.7285,
      "step": 4222
    },
    {
      "epoch": 2.26737357806785,
      "grad_norm": 0.59765625,
      "learning_rate": 2.867713746903002e-05,
      "loss": 0.6962,
      "step": 4223
    },
    {
      "epoch": 2.2679104728029262,
      "grad_norm": 0.70703125,
      "learning_rate": 2.8668783969043723e-05,
      "loss": 0.9066,
      "step": 4224
    },
    {
      "epoch": 2.268447367538002,
      "grad_norm": 0.84375,
      "learning_rate": 2.866043005040272e-05,
      "loss": 0.7681,
      "step": 4225
    },
    {
      "epoch": 2.268984262273078,
      "grad_norm": 0.84765625,
      "learning_rate": 2.8652075714060295e-05,
      "loss": 0.8514,
      "step": 4226
    },
    {
      "epoch": 2.269521157008154,
      "grad_norm": 0.76953125,
      "learning_rate": 2.8643720960969778e-05,
      "loss": 1.0333,
      "step": 4227
    },
    {
      "epoch": 2.2700580517432303,
      "grad_norm": 0.69140625,
      "learning_rate": 2.8635365792084556e-05,
      "loss": 0.7421,
      "step": 4228
    },
    {
      "epoch": 2.270594946478306,
      "grad_norm": 0.73828125,
      "learning_rate": 2.8627010208358053e-05,
      "loss": 0.845,
      "step": 4229
    },
    {
      "epoch": 2.271131841213382,
      "grad_norm": 0.6796875,
      "learning_rate": 2.861865421074376e-05,
      "loss": 0.8226,
      "step": 4230
    },
    {
      "epoch": 2.271668735948458,
      "grad_norm": 0.734375,
      "learning_rate": 2.8610297800195194e-05,
      "loss": 0.8869,
      "step": 4231
    },
    {
      "epoch": 2.2722056306835343,
      "grad_norm": 0.62109375,
      "learning_rate": 2.8601940977665927e-05,
      "loss": 0.7535,
      "step": 4232
    },
    {
      "epoch": 2.27274252541861,
      "grad_norm": 0.69921875,
      "learning_rate": 2.8593583744109574e-05,
      "loss": 0.8447,
      "step": 4233
    },
    {
      "epoch": 2.273279420153686,
      "grad_norm": 0.7265625,
      "learning_rate": 2.858522610047982e-05,
      "loss": 1.0643,
      "step": 4234
    },
    {
      "epoch": 2.2738163148887622,
      "grad_norm": 0.66796875,
      "learning_rate": 2.8576868047730354e-05,
      "loss": 0.8225,
      "step": 4235
    },
    {
      "epoch": 2.274353209623838,
      "grad_norm": 0.56640625,
      "learning_rate": 2.856850958681495e-05,
      "loss": 0.6639,
      "step": 4236
    },
    {
      "epoch": 2.274890104358914,
      "grad_norm": 0.8359375,
      "learning_rate": 2.8560150718687413e-05,
      "loss": 0.9331,
      "step": 4237
    },
    {
      "epoch": 2.27542699909399,
      "grad_norm": 0.640625,
      "learning_rate": 2.8551791444301583e-05,
      "loss": 0.7256,
      "step": 4238
    },
    {
      "epoch": 2.2759638938290663,
      "grad_norm": 0.69921875,
      "learning_rate": 2.8543431764611383e-05,
      "loss": 0.7501,
      "step": 4239
    },
    {
      "epoch": 2.276500788564142,
      "grad_norm": 0.734375,
      "learning_rate": 2.8535071680570734e-05,
      "loss": 0.9966,
      "step": 4240
    },
    {
      "epoch": 2.277037683299218,
      "grad_norm": 0.82421875,
      "learning_rate": 2.852671119313365e-05,
      "loss": 0.8804,
      "step": 4241
    },
    {
      "epoch": 2.277574578034294,
      "grad_norm": 0.65234375,
      "learning_rate": 2.8518350303254147e-05,
      "loss": 0.7867,
      "step": 4242
    },
    {
      "epoch": 2.2781114727693703,
      "grad_norm": 0.69921875,
      "learning_rate": 2.8509989011886325e-05,
      "loss": 0.8485,
      "step": 4243
    },
    {
      "epoch": 2.278648367504446,
      "grad_norm": 0.95703125,
      "learning_rate": 2.8501627319984296e-05,
      "loss": 0.9357,
      "step": 4244
    },
    {
      "epoch": 2.279185262239522,
      "grad_norm": 0.71484375,
      "learning_rate": 2.8493265228502253e-05,
      "loss": 0.8212,
      "step": 4245
    },
    {
      "epoch": 2.2797221569745982,
      "grad_norm": 0.72265625,
      "learning_rate": 2.8484902738394403e-05,
      "loss": 0.8409,
      "step": 4246
    },
    {
      "epoch": 2.2802590517096744,
      "grad_norm": 0.66015625,
      "learning_rate": 2.847653985061503e-05,
      "loss": 0.729,
      "step": 4247
    },
    {
      "epoch": 2.28079594644475,
      "grad_norm": 0.64453125,
      "learning_rate": 2.846817656611842e-05,
      "loss": 0.7594,
      "step": 4248
    },
    {
      "epoch": 2.281332841179826,
      "grad_norm": 0.6875,
      "learning_rate": 2.845981288585895e-05,
      "loss": 0.7443,
      "step": 4249
    },
    {
      "epoch": 2.2818697359149023,
      "grad_norm": 0.80859375,
      "learning_rate": 2.8451448810791014e-05,
      "loss": 0.956,
      "step": 4250
    },
    {
      "epoch": 2.2818697359149023,
      "eval_loss": 0.9055709838867188,
      "eval_runtime": 46.8289,
      "eval_samples_per_second": 20.543,
      "eval_steps_per_second": 20.543,
      "step": 4250
    },
    {
      "epoch": 2.2824066306499784,
      "grad_norm": 0.703125,
      "learning_rate": 2.844308434186906e-05,
      "loss": 0.8582,
      "step": 4251
    },
    {
      "epoch": 2.282943525385054,
      "grad_norm": 0.76953125,
      "learning_rate": 2.8434719480047577e-05,
      "loss": 0.8337,
      "step": 4252
    },
    {
      "epoch": 2.28348042012013,
      "grad_norm": 0.703125,
      "learning_rate": 2.8426354226281105e-05,
      "loss": 0.9313,
      "step": 4253
    },
    {
      "epoch": 2.2840173148552063,
      "grad_norm": 0.8984375,
      "learning_rate": 2.8417988581524217e-05,
      "loss": 0.9046,
      "step": 4254
    },
    {
      "epoch": 2.284554209590282,
      "grad_norm": 0.6953125,
      "learning_rate": 2.8409622546731558e-05,
      "loss": 0.873,
      "step": 4255
    },
    {
      "epoch": 2.285091104325358,
      "grad_norm": 0.6875,
      "learning_rate": 2.8401256122857784e-05,
      "loss": 0.8176,
      "step": 4256
    },
    {
      "epoch": 2.2856279990604342,
      "grad_norm": 0.8125,
      "learning_rate": 2.8392889310857612e-05,
      "loss": 0.8911,
      "step": 4257
    },
    {
      "epoch": 2.2861648937955104,
      "grad_norm": 0.69140625,
      "learning_rate": 2.8384522111685803e-05,
      "loss": 1.085,
      "step": 4258
    },
    {
      "epoch": 2.286701788530586,
      "grad_norm": 0.671875,
      "learning_rate": 2.837615452629716e-05,
      "loss": 0.6702,
      "step": 4259
    },
    {
      "epoch": 2.287238683265662,
      "grad_norm": 0.59765625,
      "learning_rate": 2.836778655564653e-05,
      "loss": 0.6827,
      "step": 4260
    },
    {
      "epoch": 2.2877755780007383,
      "grad_norm": 0.8046875,
      "learning_rate": 2.8359418200688805e-05,
      "loss": 0.9794,
      "step": 4261
    },
    {
      "epoch": 2.2883124727358144,
      "grad_norm": 0.7109375,
      "learning_rate": 2.835104946237891e-05,
      "loss": 0.9336,
      "step": 4262
    },
    {
      "epoch": 2.28884936747089,
      "grad_norm": 0.703125,
      "learning_rate": 2.8342680341671845e-05,
      "loss": 0.9572,
      "step": 4263
    },
    {
      "epoch": 2.289386262205966,
      "grad_norm": 0.8125,
      "learning_rate": 2.833431083952262e-05,
      "loss": 0.7498,
      "step": 4264
    },
    {
      "epoch": 2.2899231569410423,
      "grad_norm": 0.64453125,
      "learning_rate": 2.83259409568863e-05,
      "loss": 0.9209,
      "step": 4265
    },
    {
      "epoch": 2.2904600516761184,
      "grad_norm": 0.7421875,
      "learning_rate": 2.8317570694717998e-05,
      "loss": 0.7977,
      "step": 4266
    },
    {
      "epoch": 2.290996946411194,
      "grad_norm": 0.8046875,
      "learning_rate": 2.8309200053972868e-05,
      "loss": 1.0071,
      "step": 4267
    },
    {
      "epoch": 2.2915338411462702,
      "grad_norm": 0.61328125,
      "learning_rate": 2.8300829035606102e-05,
      "loss": 0.7656,
      "step": 4268
    },
    {
      "epoch": 2.2920707358813464,
      "grad_norm": 0.6171875,
      "learning_rate": 2.829245764057294e-05,
      "loss": 0.6606,
      "step": 4269
    },
    {
      "epoch": 2.2926076306164225,
      "grad_norm": 0.703125,
      "learning_rate": 2.8284085869828665e-05,
      "loss": 0.7588,
      "step": 4270
    },
    {
      "epoch": 2.293144525351498,
      "grad_norm": 0.63671875,
      "learning_rate": 2.82757137243286e-05,
      "loss": 0.842,
      "step": 4271
    },
    {
      "epoch": 2.2936814200865743,
      "grad_norm": 0.671875,
      "learning_rate": 2.826734120502813e-05,
      "loss": 0.7822,
      "step": 4272
    },
    {
      "epoch": 2.2942183148216504,
      "grad_norm": 0.61328125,
      "learning_rate": 2.8258968312882638e-05,
      "loss": 0.6891,
      "step": 4273
    },
    {
      "epoch": 2.294755209556726,
      "grad_norm": 0.73828125,
      "learning_rate": 2.8250595048847594e-05,
      "loss": 0.9425,
      "step": 4274
    },
    {
      "epoch": 2.295292104291802,
      "grad_norm": 0.6953125,
      "learning_rate": 2.824222141387849e-05,
      "loss": 0.9684,
      "step": 4275
    },
    {
      "epoch": 2.2958289990268783,
      "grad_norm": 0.6875,
      "learning_rate": 2.8233847408930857e-05,
      "loss": 0.861,
      "step": 4276
    },
    {
      "epoch": 2.2963658937619544,
      "grad_norm": 0.671875,
      "learning_rate": 2.822547303496029e-05,
      "loss": 0.8275,
      "step": 4277
    },
    {
      "epoch": 2.29690278849703,
      "grad_norm": 0.76953125,
      "learning_rate": 2.8217098292922393e-05,
      "loss": 0.7691,
      "step": 4278
    },
    {
      "epoch": 2.2974396832321062,
      "grad_norm": 0.7734375,
      "learning_rate": 2.8208723183772844e-05,
      "loss": 1.0145,
      "step": 4279
    },
    {
      "epoch": 2.2979765779671824,
      "grad_norm": 0.6640625,
      "learning_rate": 2.820034770846735e-05,
      "loss": 0.7483,
      "step": 4280
    },
    {
      "epoch": 2.2985134727022585,
      "grad_norm": 0.640625,
      "learning_rate": 2.8191971867961652e-05,
      "loss": 0.6169,
      "step": 4281
    },
    {
      "epoch": 2.299050367437334,
      "grad_norm": 0.7421875,
      "learning_rate": 2.8183595663211536e-05,
      "loss": 0.708,
      "step": 4282
    },
    {
      "epoch": 2.2995872621724103,
      "grad_norm": 0.8125,
      "learning_rate": 2.8175219095172838e-05,
      "loss": 1.0034,
      "step": 4283
    },
    {
      "epoch": 2.3001241569074864,
      "grad_norm": 0.71875,
      "learning_rate": 2.816684216480143e-05,
      "loss": 0.8352,
      "step": 4284
    },
    {
      "epoch": 2.3006610516425625,
      "grad_norm": 0.77734375,
      "learning_rate": 2.8158464873053237e-05,
      "loss": 1.0078,
      "step": 4285
    },
    {
      "epoch": 2.301197946377638,
      "grad_norm": 0.75390625,
      "learning_rate": 2.815008722088419e-05,
      "loss": 0.8073,
      "step": 4286
    },
    {
      "epoch": 2.3017348411127143,
      "grad_norm": 0.68359375,
      "learning_rate": 2.81417092092503e-05,
      "loss": 0.8694,
      "step": 4287
    },
    {
      "epoch": 2.3022717358477904,
      "grad_norm": 0.703125,
      "learning_rate": 2.8133330839107608e-05,
      "loss": 0.9218,
      "step": 4288
    },
    {
      "epoch": 2.3028086305828666,
      "grad_norm": 0.78515625,
      "learning_rate": 2.8124952111412183e-05,
      "loss": 0.9545,
      "step": 4289
    },
    {
      "epoch": 2.3033455253179422,
      "grad_norm": 0.6328125,
      "learning_rate": 2.8116573027120152e-05,
      "loss": 0.8647,
      "step": 4290
    },
    {
      "epoch": 2.3038824200530184,
      "grad_norm": 0.61328125,
      "learning_rate": 2.810819358718767e-05,
      "loss": 0.695,
      "step": 4291
    },
    {
      "epoch": 2.3044193147880945,
      "grad_norm": 0.9140625,
      "learning_rate": 2.8099813792570928e-05,
      "loss": 0.8527,
      "step": 4292
    },
    {
      "epoch": 2.30495620952317,
      "grad_norm": 0.7265625,
      "learning_rate": 2.809143364422619e-05,
      "loss": 0.7821,
      "step": 4293
    },
    {
      "epoch": 2.3054931042582463,
      "grad_norm": 0.6875,
      "learning_rate": 2.8083053143109717e-05,
      "loss": 0.9497,
      "step": 4294
    },
    {
      "epoch": 2.3060299989933224,
      "grad_norm": 0.83984375,
      "learning_rate": 2.8074672290177832e-05,
      "loss": 1.2043,
      "step": 4295
    },
    {
      "epoch": 2.3065668937283985,
      "grad_norm": 0.68359375,
      "learning_rate": 2.806629108638691e-05,
      "loss": 0.8137,
      "step": 4296
    },
    {
      "epoch": 2.307103788463474,
      "grad_norm": 0.6953125,
      "learning_rate": 2.8057909532693338e-05,
      "loss": 0.9738,
      "step": 4297
    },
    {
      "epoch": 2.3076406831985503,
      "grad_norm": 0.63671875,
      "learning_rate": 2.8049527630053562e-05,
      "loss": 0.6856,
      "step": 4298
    },
    {
      "epoch": 2.3081775779336264,
      "grad_norm": 0.66015625,
      "learning_rate": 2.8041145379424073e-05,
      "loss": 0.8664,
      "step": 4299
    },
    {
      "epoch": 2.3087144726687026,
      "grad_norm": 0.6015625,
      "learning_rate": 2.8032762781761374e-05,
      "loss": 0.6837,
      "step": 4300
    },
    {
      "epoch": 2.3087144726687026,
      "eval_loss": 0.9049941897392273,
      "eval_runtime": 46.0915,
      "eval_samples_per_second": 20.872,
      "eval_steps_per_second": 20.872,
      "step": 4300
    },
    {
      "epoch": 2.3092513674037782,
      "grad_norm": 0.70703125,
      "learning_rate": 2.8024379838022036e-05,
      "loss": 0.7317,
      "step": 4301
    },
    {
      "epoch": 2.3097882621388544,
      "grad_norm": 0.85546875,
      "learning_rate": 2.8015996549162664e-05,
      "loss": 0.8046,
      "step": 4302
    },
    {
      "epoch": 2.3103251568739305,
      "grad_norm": 0.578125,
      "learning_rate": 2.800761291613988e-05,
      "loss": 0.7216,
      "step": 4303
    },
    {
      "epoch": 2.3108620516090066,
      "grad_norm": 0.75,
      "learning_rate": 2.7999228939910382e-05,
      "loss": 0.9202,
      "step": 4304
    },
    {
      "epoch": 2.3113989463440823,
      "grad_norm": 0.8359375,
      "learning_rate": 2.799084462143088e-05,
      "loss": 0.8792,
      "step": 4305
    },
    {
      "epoch": 2.3119358410791584,
      "grad_norm": 0.69921875,
      "learning_rate": 2.798245996165813e-05,
      "loss": 0.8558,
      "step": 4306
    },
    {
      "epoch": 2.3124727358142345,
      "grad_norm": 0.7109375,
      "learning_rate": 2.7974074961548928e-05,
      "loss": 0.7988,
      "step": 4307
    },
    {
      "epoch": 2.3130096305493106,
      "grad_norm": 0.75,
      "learning_rate": 2.796568962206011e-05,
      "loss": 0.9657,
      "step": 4308
    },
    {
      "epoch": 2.3135465252843863,
      "grad_norm": 0.59375,
      "learning_rate": 2.7957303944148543e-05,
      "loss": 0.8314,
      "step": 4309
    },
    {
      "epoch": 2.3140834200194624,
      "grad_norm": 0.74609375,
      "learning_rate": 2.7948917928771158e-05,
      "loss": 1.2084,
      "step": 4310
    },
    {
      "epoch": 2.3146203147545386,
      "grad_norm": 0.7265625,
      "learning_rate": 2.7940531576884876e-05,
      "loss": 0.9992,
      "step": 4311
    },
    {
      "epoch": 2.3151572094896142,
      "grad_norm": 0.6640625,
      "learning_rate": 2.793214488944671e-05,
      "loss": 0.8511,
      "step": 4312
    },
    {
      "epoch": 2.3156941042246904,
      "grad_norm": 0.7890625,
      "learning_rate": 2.792375786741368e-05,
      "loss": 0.9781,
      "step": 4313
    },
    {
      "epoch": 2.3162309989597665,
      "grad_norm": 0.7109375,
      "learning_rate": 2.7915370511742855e-05,
      "loss": 0.889,
      "step": 4314
    },
    {
      "epoch": 2.3167678936948426,
      "grad_norm": 0.7265625,
      "learning_rate": 2.7906982823391332e-05,
      "loss": 0.6994,
      "step": 4315
    },
    {
      "epoch": 2.3173047884299183,
      "grad_norm": 0.7890625,
      "learning_rate": 2.7898594803316253e-05,
      "loss": 0.7376,
      "step": 4316
    },
    {
      "epoch": 2.3178416831649944,
      "grad_norm": 0.78125,
      "learning_rate": 2.7890206452474794e-05,
      "loss": 0.9348,
      "step": 4317
    },
    {
      "epoch": 2.3183785779000705,
      "grad_norm": 0.68359375,
      "learning_rate": 2.7881817771824188e-05,
      "loss": 0.7226,
      "step": 4318
    },
    {
      "epoch": 2.3189154726351466,
      "grad_norm": 0.69140625,
      "learning_rate": 2.787342876232167e-05,
      "loss": 0.7594,
      "step": 4319
    },
    {
      "epoch": 2.3194523673702223,
      "grad_norm": 0.66796875,
      "learning_rate": 2.7865039424924542e-05,
      "loss": 0.8509,
      "step": 4320
    },
    {
      "epoch": 2.3199892621052984,
      "grad_norm": 0.8359375,
      "learning_rate": 2.7856649760590136e-05,
      "loss": 0.8446,
      "step": 4321
    },
    {
      "epoch": 2.3205261568403746,
      "grad_norm": 0.9765625,
      "learning_rate": 2.7848259770275814e-05,
      "loss": 0.9816,
      "step": 4322
    },
    {
      "epoch": 2.3210630515754507,
      "grad_norm": 0.74609375,
      "learning_rate": 2.7839869454938983e-05,
      "loss": 0.9402,
      "step": 4323
    },
    {
      "epoch": 2.3215999463105264,
      "grad_norm": 0.8046875,
      "learning_rate": 2.7831478815537087e-05,
      "loss": 0.9757,
      "step": 4324
    },
    {
      "epoch": 2.3221368410456025,
      "grad_norm": 1.0703125,
      "learning_rate": 2.782308785302759e-05,
      "loss": 0.966,
      "step": 4325
    },
    {
      "epoch": 2.3226737357806786,
      "grad_norm": 0.6484375,
      "learning_rate": 2.7814696568368027e-05,
      "loss": 0.8309,
      "step": 4326
    },
    {
      "epoch": 2.3232106305157547,
      "grad_norm": 0.7265625,
      "learning_rate": 2.7806304962515938e-05,
      "loss": 0.8635,
      "step": 4327
    },
    {
      "epoch": 2.3237475252508304,
      "grad_norm": 0.68359375,
      "learning_rate": 2.7797913036428914e-05,
      "loss": 0.7369,
      "step": 4328
    },
    {
      "epoch": 2.3242844199859065,
      "grad_norm": 0.734375,
      "learning_rate": 2.7789520791064578e-05,
      "loss": 0.8928,
      "step": 4329
    },
    {
      "epoch": 2.3248213147209826,
      "grad_norm": 0.72265625,
      "learning_rate": 2.778112822738059e-05,
      "loss": 0.7585,
      "step": 4330
    },
    {
      "epoch": 2.3253582094560583,
      "grad_norm": 0.640625,
      "learning_rate": 2.777273534633466e-05,
      "loss": 0.702,
      "step": 4331
    },
    {
      "epoch": 2.3258951041911344,
      "grad_norm": 0.70703125,
      "learning_rate": 2.776434214888451e-05,
      "loss": 0.9025,
      "step": 4332
    },
    {
      "epoch": 2.3264319989262106,
      "grad_norm": 0.75,
      "learning_rate": 2.7755948635987906e-05,
      "loss": 0.9347,
      "step": 4333
    },
    {
      "epoch": 2.3269688936612867,
      "grad_norm": 0.66796875,
      "learning_rate": 2.7747554808602665e-05,
      "loss": 0.8172,
      "step": 4334
    },
    {
      "epoch": 2.3275057883963624,
      "grad_norm": 0.75390625,
      "learning_rate": 2.7739160667686637e-05,
      "loss": 0.907,
      "step": 4335
    },
    {
      "epoch": 2.3280426831314385,
      "grad_norm": 0.7734375,
      "learning_rate": 2.7730766214197674e-05,
      "loss": 0.8014,
      "step": 4336
    },
    {
      "epoch": 2.3285795778665146,
      "grad_norm": 0.62890625,
      "learning_rate": 2.772237144909371e-05,
      "loss": 0.8979,
      "step": 4337
    },
    {
      "epoch": 2.3291164726015907,
      "grad_norm": 0.71875,
      "learning_rate": 2.7713976373332686e-05,
      "loss": 0.8025,
      "step": 4338
    },
    {
      "epoch": 2.3296533673366664,
      "grad_norm": 0.71875,
      "learning_rate": 2.7705580987872588e-05,
      "loss": 0.8223,
      "step": 4339
    },
    {
      "epoch": 2.3301902620717425,
      "grad_norm": 0.66015625,
      "learning_rate": 2.7697185293671434e-05,
      "loss": 0.7898,
      "step": 4340
    },
    {
      "epoch": 2.3307271568068186,
      "grad_norm": 0.70703125,
      "learning_rate": 2.768878929168728e-05,
      "loss": 0.806,
      "step": 4341
    },
    {
      "epoch": 2.3312640515418948,
      "grad_norm": 0.62890625,
      "learning_rate": 2.7680392982878223e-05,
      "loss": 0.9565,
      "step": 4342
    },
    {
      "epoch": 2.3318009462769704,
      "grad_norm": 0.61328125,
      "learning_rate": 2.7671996368202374e-05,
      "loss": 0.9082,
      "step": 4343
    },
    {
      "epoch": 2.3323378410120466,
      "grad_norm": 0.68359375,
      "learning_rate": 2.766359944861791e-05,
      "loss": 0.8444,
      "step": 4344
    },
    {
      "epoch": 2.3328747357471227,
      "grad_norm": 0.640625,
      "learning_rate": 2.7655202225083015e-05,
      "loss": 0.991,
      "step": 4345
    },
    {
      "epoch": 2.333411630482199,
      "grad_norm": 0.6796875,
      "learning_rate": 2.7646804698555917e-05,
      "loss": 0.8506,
      "step": 4346
    },
    {
      "epoch": 2.3339485252172745,
      "grad_norm": 0.66015625,
      "learning_rate": 2.763840686999488e-05,
      "loss": 0.8369,
      "step": 4347
    },
    {
      "epoch": 2.3344854199523506,
      "grad_norm": 0.73046875,
      "learning_rate": 2.763000874035822e-05,
      "loss": 0.9519,
      "step": 4348
    },
    {
      "epoch": 2.3350223146874267,
      "grad_norm": 0.66015625,
      "learning_rate": 2.762161031060425e-05,
      "loss": 0.7414,
      "step": 4349
    },
    {
      "epoch": 2.3355592094225024,
      "grad_norm": 0.84375,
      "learning_rate": 2.761321158169134e-05,
      "loss": 1.0191,
      "step": 4350
    },
    {
      "epoch": 2.3355592094225024,
      "eval_loss": 0.9044721126556396,
      "eval_runtime": 46.2168,
      "eval_samples_per_second": 20.815,
      "eval_steps_per_second": 20.815,
      "step": 4350
    },
    {
      "epoch": 2.3360961041575785,
      "grad_norm": 0.6953125,
      "learning_rate": 2.76048125545779e-05,
      "loss": 1.0184,
      "step": 4351
    },
    {
      "epoch": 2.3366329988926546,
      "grad_norm": 0.703125,
      "learning_rate": 2.759641323022235e-05,
      "loss": 0.6846,
      "step": 4352
    },
    {
      "epoch": 2.3371698936277308,
      "grad_norm": 0.640625,
      "learning_rate": 2.758801360958318e-05,
      "loss": 0.8757,
      "step": 4353
    },
    {
      "epoch": 2.3377067883628064,
      "grad_norm": 0.61328125,
      "learning_rate": 2.7579613693618884e-05,
      "loss": 0.8561,
      "step": 4354
    },
    {
      "epoch": 2.3382436830978826,
      "grad_norm": 0.69921875,
      "learning_rate": 2.7571213483287984e-05,
      "loss": 0.8019,
      "step": 4355
    },
    {
      "epoch": 2.3387805778329587,
      "grad_norm": 0.6796875,
      "learning_rate": 2.756281297954908e-05,
      "loss": 0.7533,
      "step": 4356
    },
    {
      "epoch": 2.339317472568035,
      "grad_norm": 0.62109375,
      "learning_rate": 2.7554412183360757e-05,
      "loss": 0.8404,
      "step": 4357
    },
    {
      "epoch": 2.3398543673031105,
      "grad_norm": 0.77734375,
      "learning_rate": 2.7546011095681638e-05,
      "loss": 0.8433,
      "step": 4358
    },
    {
      "epoch": 2.3403912620381866,
      "grad_norm": 0.59375,
      "learning_rate": 2.7537609717470425e-05,
      "loss": 0.9055,
      "step": 4359
    },
    {
      "epoch": 2.3409281567732627,
      "grad_norm": 0.71875,
      "learning_rate": 2.7529208049685807e-05,
      "loss": 0.8937,
      "step": 4360
    },
    {
      "epoch": 2.341465051508339,
      "grad_norm": 0.86328125,
      "learning_rate": 2.7520806093286516e-05,
      "loss": 0.798,
      "step": 4361
    },
    {
      "epoch": 2.3420019462434145,
      "grad_norm": 0.7265625,
      "learning_rate": 2.7512403849231332e-05,
      "loss": 1.106,
      "step": 4362
    },
    {
      "epoch": 2.3425388409784906,
      "grad_norm": 0.65234375,
      "learning_rate": 2.750400131847905e-05,
      "loss": 0.7766,
      "step": 4363
    },
    {
      "epoch": 2.3430757357135668,
      "grad_norm": 0.6640625,
      "learning_rate": 2.7495598501988502e-05,
      "loss": 0.7044,
      "step": 4364
    },
    {
      "epoch": 2.343612630448643,
      "grad_norm": 0.640625,
      "learning_rate": 2.7487195400718568e-05,
      "loss": 0.8082,
      "step": 4365
    },
    {
      "epoch": 2.3441495251837186,
      "grad_norm": 0.7734375,
      "learning_rate": 2.747879201562813e-05,
      "loss": 0.927,
      "step": 4366
    },
    {
      "epoch": 2.3446864199187947,
      "grad_norm": 0.91015625,
      "learning_rate": 2.747038834767614e-05,
      "loss": 0.9076,
      "step": 4367
    },
    {
      "epoch": 2.345223314653871,
      "grad_norm": 0.625,
      "learning_rate": 2.7461984397821555e-05,
      "loss": 0.7095,
      "step": 4368
    },
    {
      "epoch": 2.3457602093889465,
      "grad_norm": 0.62109375,
      "learning_rate": 2.7453580167023367e-05,
      "loss": 0.8225,
      "step": 4369
    },
    {
      "epoch": 2.3462971041240226,
      "grad_norm": 0.72265625,
      "learning_rate": 2.7445175656240617e-05,
      "loss": 0.8181,
      "step": 4370
    },
    {
      "epoch": 2.3468339988590987,
      "grad_norm": 0.68359375,
      "learning_rate": 2.7436770866432356e-05,
      "loss": 0.8887,
      "step": 4371
    },
    {
      "epoch": 2.347370893594175,
      "grad_norm": 0.65234375,
      "learning_rate": 2.7428365798557676e-05,
      "loss": 0.7915,
      "step": 4372
    },
    {
      "epoch": 2.3479077883292505,
      "grad_norm": 0.88671875,
      "learning_rate": 2.741996045357571e-05,
      "loss": 0.7064,
      "step": 4373
    },
    {
      "epoch": 2.3484446830643266,
      "grad_norm": 0.7109375,
      "learning_rate": 2.7411554832445608e-05,
      "loss": 0.8799,
      "step": 4374
    },
    {
      "epoch": 2.3489815777994028,
      "grad_norm": 0.56640625,
      "learning_rate": 2.740314893612656e-05,
      "loss": 0.762,
      "step": 4375
    },
    {
      "epoch": 2.349518472534479,
      "grad_norm": 0.640625,
      "learning_rate": 2.7394742765577785e-05,
      "loss": 0.8482,
      "step": 4376
    },
    {
      "epoch": 2.3500553672695546,
      "grad_norm": 0.69140625,
      "learning_rate": 2.7386336321758534e-05,
      "loss": 0.8223,
      "step": 4377
    },
    {
      "epoch": 2.3505922620046307,
      "grad_norm": 0.765625,
      "learning_rate": 2.737792960562809e-05,
      "loss": 0.8307,
      "step": 4378
    },
    {
      "epoch": 2.351129156739707,
      "grad_norm": 0.640625,
      "learning_rate": 2.736952261814576e-05,
      "loss": 0.7123,
      "step": 4379
    },
    {
      "epoch": 2.351666051474783,
      "grad_norm": 0.953125,
      "learning_rate": 2.736111536027089e-05,
      "loss": 0.9244,
      "step": 4380
    },
    {
      "epoch": 2.3522029462098586,
      "grad_norm": 0.68359375,
      "learning_rate": 2.7352707832962865e-05,
      "loss": 0.8727,
      "step": 4381
    },
    {
      "epoch": 2.3527398409449347,
      "grad_norm": 0.60546875,
      "learning_rate": 2.734430003718107e-05,
      "loss": 0.8808,
      "step": 4382
    },
    {
      "epoch": 2.353276735680011,
      "grad_norm": 0.640625,
      "learning_rate": 2.733589197388496e-05,
      "loss": 0.6586,
      "step": 4383
    },
    {
      "epoch": 2.353813630415087,
      "grad_norm": 0.73046875,
      "learning_rate": 2.7327483644033992e-05,
      "loss": 0.9529,
      "step": 4384
    },
    {
      "epoch": 2.3543505251501626,
      "grad_norm": 0.65625,
      "learning_rate": 2.7319075048587666e-05,
      "loss": 0.649,
      "step": 4385
    },
    {
      "epoch": 2.3548874198852388,
      "grad_norm": 0.6875,
      "learning_rate": 2.731066618850551e-05,
      "loss": 0.7666,
      "step": 4386
    },
    {
      "epoch": 2.355424314620315,
      "grad_norm": 0.77734375,
      "learning_rate": 2.7302257064747077e-05,
      "loss": 0.9609,
      "step": 4387
    },
    {
      "epoch": 2.3559612093553906,
      "grad_norm": 0.5546875,
      "learning_rate": 2.7293847678271954e-05,
      "loss": 0.6664,
      "step": 4388
    },
    {
      "epoch": 2.3564981040904667,
      "grad_norm": 0.6875,
      "learning_rate": 2.7285438030039768e-05,
      "loss": 0.8773,
      "step": 4389
    },
    {
      "epoch": 2.357034998825543,
      "grad_norm": 0.7109375,
      "learning_rate": 2.7277028121010162e-05,
      "loss": 0.8702,
      "step": 4390
    },
    {
      "epoch": 2.357571893560619,
      "grad_norm": 0.89453125,
      "learning_rate": 2.7268617952142806e-05,
      "loss": 0.9229,
      "step": 4391
    },
    {
      "epoch": 2.3581087882956946,
      "grad_norm": 0.69921875,
      "learning_rate": 2.7260207524397413e-05,
      "loss": 0.9931,
      "step": 4392
    },
    {
      "epoch": 2.3586456830307707,
      "grad_norm": 0.7109375,
      "learning_rate": 2.7251796838733716e-05,
      "loss": 0.7213,
      "step": 4393
    },
    {
      "epoch": 2.359182577765847,
      "grad_norm": 0.62109375,
      "learning_rate": 2.724338589611149e-05,
      "loss": 0.7953,
      "step": 4394
    },
    {
      "epoch": 2.359719472500923,
      "grad_norm": 0.62890625,
      "learning_rate": 2.723497469749052e-05,
      "loss": 0.6998,
      "step": 4395
    },
    {
      "epoch": 2.3602563672359986,
      "grad_norm": 0.69140625,
      "learning_rate": 2.722656324383062e-05,
      "loss": 0.812,
      "step": 4396
    },
    {
      "epoch": 2.3607932619710748,
      "grad_norm": 0.7109375,
      "learning_rate": 2.7218151536091674e-05,
      "loss": 1.0321,
      "step": 4397
    },
    {
      "epoch": 2.361330156706151,
      "grad_norm": 0.578125,
      "learning_rate": 2.7209739575233544e-05,
      "loss": 0.693,
      "step": 4398
    },
    {
      "epoch": 2.361867051441227,
      "grad_norm": 0.5859375,
      "learning_rate": 2.720132736221614e-05,
      "loss": 0.7624,
      "step": 4399
    },
    {
      "epoch": 2.3624039461763027,
      "grad_norm": 0.62890625,
      "learning_rate": 2.7192914897999404e-05,
      "loss": 0.9707,
      "step": 4400
    },
    {
      "epoch": 2.3624039461763027,
      "eval_loss": 0.9049584865570068,
      "eval_runtime": 45.9822,
      "eval_samples_per_second": 20.921,
      "eval_steps_per_second": 20.921,
      "step": 4400
    },
    {
      "epoch": 2.362940840911379,
      "grad_norm": 0.74609375,
      "learning_rate": 2.7184502183543314e-05,
      "loss": 0.886,
      "step": 4401
    },
    {
      "epoch": 2.363477735646455,
      "grad_norm": 0.984375,
      "learning_rate": 2.717608921980786e-05,
      "loss": 0.857,
      "step": 4402
    },
    {
      "epoch": 2.364014630381531,
      "grad_norm": 0.6875,
      "learning_rate": 2.7167676007753067e-05,
      "loss": 0.6939,
      "step": 4403
    },
    {
      "epoch": 2.3645515251166067,
      "grad_norm": 0.65234375,
      "learning_rate": 2.715926254833898e-05,
      "loss": 0.6873,
      "step": 4404
    },
    {
      "epoch": 2.365088419851683,
      "grad_norm": 0.76171875,
      "learning_rate": 2.7150848842525704e-05,
      "loss": 0.8498,
      "step": 4405
    },
    {
      "epoch": 2.365625314586759,
      "grad_norm": 0.7578125,
      "learning_rate": 2.714243489127333e-05,
      "loss": 0.9702,
      "step": 4406
    },
    {
      "epoch": 2.3661622093218346,
      "grad_norm": 0.73046875,
      "learning_rate": 2.7134020695542012e-05,
      "loss": 0.91,
      "step": 4407
    },
    {
      "epoch": 2.3666991040569108,
      "grad_norm": 0.66796875,
      "learning_rate": 2.7125606256291902e-05,
      "loss": 0.6683,
      "step": 4408
    },
    {
      "epoch": 2.367235998791987,
      "grad_norm": 0.66796875,
      "learning_rate": 2.7117191574483197e-05,
      "loss": 0.9067,
      "step": 4409
    },
    {
      "epoch": 2.367772893527063,
      "grad_norm": 0.69140625,
      "learning_rate": 2.7108776651076118e-05,
      "loss": 1.0223,
      "step": 4410
    },
    {
      "epoch": 2.3683097882621387,
      "grad_norm": 0.703125,
      "learning_rate": 2.7100361487030933e-05,
      "loss": 0.8826,
      "step": 4411
    },
    {
      "epoch": 2.368846682997215,
      "grad_norm": 0.8046875,
      "learning_rate": 2.7091946083307896e-05,
      "loss": 1.0356,
      "step": 4412
    },
    {
      "epoch": 2.369383577732291,
      "grad_norm": 0.5546875,
      "learning_rate": 2.708353044086731e-05,
      "loss": 0.6455,
      "step": 4413
    },
    {
      "epoch": 2.369920472467367,
      "grad_norm": 0.75,
      "learning_rate": 2.7075114560669518e-05,
      "loss": 0.9449,
      "step": 4414
    },
    {
      "epoch": 2.3704573672024427,
      "grad_norm": 0.65234375,
      "learning_rate": 2.7066698443674883e-05,
      "loss": 0.7601,
      "step": 4415
    },
    {
      "epoch": 2.370994261937519,
      "grad_norm": 0.6875,
      "learning_rate": 2.7058282090843778e-05,
      "loss": 0.8004,
      "step": 4416
    },
    {
      "epoch": 2.371531156672595,
      "grad_norm": 0.69921875,
      "learning_rate": 2.7049865503136616e-05,
      "loss": 0.7394,
      "step": 4417
    },
    {
      "epoch": 2.372068051407671,
      "grad_norm": 0.63671875,
      "learning_rate": 2.7041448681513842e-05,
      "loss": 0.6308,
      "step": 4418
    },
    {
      "epoch": 2.3726049461427468,
      "grad_norm": 0.7265625,
      "learning_rate": 2.7033031626935934e-05,
      "loss": 0.8589,
      "step": 4419
    },
    {
      "epoch": 2.373141840877823,
      "grad_norm": 0.671875,
      "learning_rate": 2.7024614340363365e-05,
      "loss": 0.9038,
      "step": 4420
    },
    {
      "epoch": 2.373678735612899,
      "grad_norm": 0.64453125,
      "learning_rate": 2.7016196822756644e-05,
      "loss": 0.8403,
      "step": 4421
    },
    {
      "epoch": 2.374215630347975,
      "grad_norm": 0.95703125,
      "learning_rate": 2.700777907507635e-05,
      "loss": 0.8906,
      "step": 4422
    },
    {
      "epoch": 2.374752525083051,
      "grad_norm": 0.66796875,
      "learning_rate": 2.6999361098283033e-05,
      "loss": 0.942,
      "step": 4423
    },
    {
      "epoch": 2.375289419818127,
      "grad_norm": 0.87109375,
      "learning_rate": 2.699094289333729e-05,
      "loss": 0.9625,
      "step": 4424
    },
    {
      "epoch": 2.375826314553203,
      "grad_norm": 0.6171875,
      "learning_rate": 2.6982524461199753e-05,
      "loss": 0.7712,
      "step": 4425
    },
    {
      "epoch": 2.3763632092882787,
      "grad_norm": 0.72265625,
      "learning_rate": 2.697410580283107e-05,
      "loss": 0.9637,
      "step": 4426
    },
    {
      "epoch": 2.376900104023355,
      "grad_norm": 0.67578125,
      "learning_rate": 2.696568691919191e-05,
      "loss": 0.7743,
      "step": 4427
    },
    {
      "epoch": 2.377436998758431,
      "grad_norm": 0.63671875,
      "learning_rate": 2.6957267811242986e-05,
      "loss": 0.7868,
      "step": 4428
    },
    {
      "epoch": 2.377973893493507,
      "grad_norm": 0.609375,
      "learning_rate": 2.694884847994501e-05,
      "loss": 1.0199,
      "step": 4429
    },
    {
      "epoch": 2.3785107882285828,
      "grad_norm": 0.74609375,
      "learning_rate": 2.6940428926258748e-05,
      "loss": 1.0205,
      "step": 4430
    },
    {
      "epoch": 2.379047682963659,
      "grad_norm": 0.734375,
      "learning_rate": 2.6932009151144973e-05,
      "loss": 0.9248,
      "step": 4431
    },
    {
      "epoch": 2.379584577698735,
      "grad_norm": 0.7109375,
      "learning_rate": 2.692358915556449e-05,
      "loss": 0.9396,
      "step": 4432
    },
    {
      "epoch": 2.380121472433811,
      "grad_norm": 0.78515625,
      "learning_rate": 2.6915168940478115e-05,
      "loss": 0.9513,
      "step": 4433
    },
    {
      "epoch": 2.380658367168887,
      "grad_norm": 0.59375,
      "learning_rate": 2.6906748506846713e-05,
      "loss": 0.8836,
      "step": 4434
    },
    {
      "epoch": 2.381195261903963,
      "grad_norm": 0.7734375,
      "learning_rate": 2.6898327855631155e-05,
      "loss": 0.9377,
      "step": 4435
    },
    {
      "epoch": 2.381732156639039,
      "grad_norm": 0.61328125,
      "learning_rate": 2.6889906987792362e-05,
      "loss": 0.7775,
      "step": 4436
    },
    {
      "epoch": 2.382269051374115,
      "grad_norm": 0.6640625,
      "learning_rate": 2.688148590429124e-05,
      "loss": 0.85,
      "step": 4437
    },
    {
      "epoch": 2.382805946109191,
      "grad_norm": 0.69921875,
      "learning_rate": 2.6873064606088744e-05,
      "loss": 0.779,
      "step": 4438
    },
    {
      "epoch": 2.383342840844267,
      "grad_norm": 0.63671875,
      "learning_rate": 2.6864643094145857e-05,
      "loss": 0.7394,
      "step": 4439
    },
    {
      "epoch": 2.383879735579343,
      "grad_norm": 0.7421875,
      "learning_rate": 2.685622136942359e-05,
      "loss": 0.9771,
      "step": 4440
    },
    {
      "epoch": 2.384416630314419,
      "grad_norm": 0.765625,
      "learning_rate": 2.684779943288295e-05,
      "loss": 0.8868,
      "step": 4441
    },
    {
      "epoch": 2.384953525049495,
      "grad_norm": 0.69921875,
      "learning_rate": 2.6839377285484997e-05,
      "loss": 0.7228,
      "step": 4442
    },
    {
      "epoch": 2.385490419784571,
      "grad_norm": 0.72265625,
      "learning_rate": 2.6830954928190794e-05,
      "loss": 0.6917,
      "step": 4443
    },
    {
      "epoch": 2.386027314519647,
      "grad_norm": 0.68359375,
      "learning_rate": 2.6822532361961455e-05,
      "loss": 0.7938,
      "step": 4444
    },
    {
      "epoch": 2.386564209254723,
      "grad_norm": 0.8828125,
      "learning_rate": 2.6814109587758086e-05,
      "loss": 0.7497,
      "step": 4445
    },
    {
      "epoch": 2.387101103989799,
      "grad_norm": 0.62890625,
      "learning_rate": 2.6805686606541847e-05,
      "loss": 0.9161,
      "step": 4446
    },
    {
      "epoch": 2.387637998724875,
      "grad_norm": 0.609375,
      "learning_rate": 2.6797263419273893e-05,
      "loss": 0.7484,
      "step": 4447
    },
    {
      "epoch": 2.388174893459951,
      "grad_norm": 0.83984375,
      "learning_rate": 2.6788840026915423e-05,
      "loss": 0.8598,
      "step": 4448
    },
    {
      "epoch": 2.388711788195027,
      "grad_norm": 0.66796875,
      "learning_rate": 2.678041643042766e-05,
      "loss": 0.8954,
      "step": 4449
    },
    {
      "epoch": 2.389248682930103,
      "grad_norm": 0.72265625,
      "learning_rate": 2.6771992630771824e-05,
      "loss": 0.9852,
      "step": 4450
    },
    {
      "epoch": 2.389248682930103,
      "eval_loss": 0.9053536057472229,
      "eval_runtime": 46.1188,
      "eval_samples_per_second": 20.859,
      "eval_steps_per_second": 20.859,
      "step": 4450
    },
    {
      "epoch": 2.389785577665179,
      "grad_norm": 0.6640625,
      "learning_rate": 2.676356862890919e-05,
      "loss": 0.7657,
      "step": 4451
    },
    {
      "epoch": 2.390322472400255,
      "grad_norm": 0.69140625,
      "learning_rate": 2.6755144425801048e-05,
      "loss": 0.8893,
      "step": 4452
    },
    {
      "epoch": 2.390859367135331,
      "grad_norm": 0.703125,
      "learning_rate": 2.67467200224087e-05,
      "loss": 0.9203,
      "step": 4453
    },
    {
      "epoch": 2.391396261870407,
      "grad_norm": 0.734375,
      "learning_rate": 2.6738295419693473e-05,
      "loss": 0.9228,
      "step": 4454
    },
    {
      "epoch": 2.391933156605483,
      "grad_norm": 0.71484375,
      "learning_rate": 2.6729870618616724e-05,
      "loss": 0.7608,
      "step": 4455
    },
    {
      "epoch": 2.3924700513405592,
      "grad_norm": 0.9453125,
      "learning_rate": 2.6721445620139835e-05,
      "loss": 1.0952,
      "step": 4456
    },
    {
      "epoch": 2.393006946075635,
      "grad_norm": 0.76171875,
      "learning_rate": 2.6713020425224195e-05,
      "loss": 0.9739,
      "step": 4457
    },
    {
      "epoch": 2.393543840810711,
      "grad_norm": 0.640625,
      "learning_rate": 2.6704595034831236e-05,
      "loss": 0.9615,
      "step": 4458
    },
    {
      "epoch": 2.394080735545787,
      "grad_norm": 0.6015625,
      "learning_rate": 2.6696169449922386e-05,
      "loss": 0.8106,
      "step": 4459
    },
    {
      "epoch": 2.3946176302808633,
      "grad_norm": 0.7421875,
      "learning_rate": 2.668774367145913e-05,
      "loss": 0.9132,
      "step": 4460
    },
    {
      "epoch": 2.395154525015939,
      "grad_norm": 0.7109375,
      "learning_rate": 2.6679317700402956e-05,
      "loss": 0.8524,
      "step": 4461
    },
    {
      "epoch": 2.395691419751015,
      "grad_norm": 0.82421875,
      "learning_rate": 2.6670891537715354e-05,
      "loss": 0.9832,
      "step": 4462
    },
    {
      "epoch": 2.396228314486091,
      "grad_norm": 0.7421875,
      "learning_rate": 2.666246518435787e-05,
      "loss": 0.8812,
      "step": 4463
    },
    {
      "epoch": 2.396765209221167,
      "grad_norm": 0.6953125,
      "learning_rate": 2.6654038641292056e-05,
      "loss": 0.7877,
      "step": 4464
    },
    {
      "epoch": 2.397302103956243,
      "grad_norm": 0.73828125,
      "learning_rate": 2.6645611909479484e-05,
      "loss": 0.844,
      "step": 4465
    },
    {
      "epoch": 2.397838998691319,
      "grad_norm": 0.7421875,
      "learning_rate": 2.663718498988176e-05,
      "loss": 0.9999,
      "step": 4466
    },
    {
      "epoch": 2.3983758934263952,
      "grad_norm": 0.8203125,
      "learning_rate": 2.662875788346048e-05,
      "loss": 1.0071,
      "step": 4467
    },
    {
      "epoch": 2.398912788161471,
      "grad_norm": 0.6328125,
      "learning_rate": 2.6620330591177313e-05,
      "loss": 0.7417,
      "step": 4468
    },
    {
      "epoch": 2.399449682896547,
      "grad_norm": 0.66796875,
      "learning_rate": 2.6611903113993897e-05,
      "loss": 1.0774,
      "step": 4469
    },
    {
      "epoch": 2.399986577631623,
      "grad_norm": 0.72265625,
      "learning_rate": 2.660347545287193e-05,
      "loss": 1.0166,
      "step": 4470
    },
    {
      "epoch": 2.4005234723666993,
      "grad_norm": 0.83203125,
      "learning_rate": 2.6595047608773103e-05,
      "loss": 0.848,
      "step": 4471
    },
    {
      "epoch": 2.401060367101775,
      "grad_norm": 0.5625,
      "learning_rate": 2.658661958265915e-05,
      "loss": 0.781,
      "step": 4472
    },
    {
      "epoch": 2.401597261836851,
      "grad_norm": 0.73828125,
      "learning_rate": 2.65781913754918e-05,
      "loss": 0.9022,
      "step": 4473
    },
    {
      "epoch": 2.402134156571927,
      "grad_norm": 0.57421875,
      "learning_rate": 2.656976298823284e-05,
      "loss": 0.7464,
      "step": 4474
    },
    {
      "epoch": 2.4026710513070033,
      "grad_norm": 0.64453125,
      "learning_rate": 2.6561334421844037e-05,
      "loss": 0.8152,
      "step": 4475
    },
    {
      "epoch": 2.403207946042079,
      "grad_norm": 0.6875,
      "learning_rate": 2.6552905677287197e-05,
      "loss": 0.83,
      "step": 4476
    },
    {
      "epoch": 2.403744840777155,
      "grad_norm": 0.8125,
      "learning_rate": 2.654447675552416e-05,
      "loss": 0.8393,
      "step": 4477
    },
    {
      "epoch": 2.4042817355122312,
      "grad_norm": 0.7265625,
      "learning_rate": 2.6536047657516772e-05,
      "loss": 0.8914,
      "step": 4478
    },
    {
      "epoch": 2.4048186302473074,
      "grad_norm": 0.69921875,
      "learning_rate": 2.6527618384226887e-05,
      "loss": 0.7263,
      "step": 4479
    },
    {
      "epoch": 2.405355524982383,
      "grad_norm": 0.765625,
      "learning_rate": 2.65191889366164e-05,
      "loss": 0.9114,
      "step": 4480
    },
    {
      "epoch": 2.405892419717459,
      "grad_norm": 0.71484375,
      "learning_rate": 2.651075931564721e-05,
      "loss": 0.8877,
      "step": 4481
    },
    {
      "epoch": 2.4064293144525353,
      "grad_norm": 0.80859375,
      "learning_rate": 2.6502329522281262e-05,
      "loss": 0.7874,
      "step": 4482
    },
    {
      "epoch": 2.406966209187611,
      "grad_norm": 0.7421875,
      "learning_rate": 2.6493899557480485e-05,
      "loss": 0.9375,
      "step": 4483
    },
    {
      "epoch": 2.407503103922687,
      "grad_norm": 0.5546875,
      "learning_rate": 2.6485469422206843e-05,
      "loss": 0.7289,
      "step": 4484
    },
    {
      "epoch": 2.408039998657763,
      "grad_norm": 0.609375,
      "learning_rate": 2.6477039117422335e-05,
      "loss": 0.752,
      "step": 4485
    },
    {
      "epoch": 2.4085768933928393,
      "grad_norm": 0.67578125,
      "learning_rate": 2.6468608644088953e-05,
      "loss": 0.975,
      "step": 4486
    },
    {
      "epoch": 2.409113788127915,
      "grad_norm": 0.6640625,
      "learning_rate": 2.6460178003168734e-05,
      "loss": 1.0165,
      "step": 4487
    },
    {
      "epoch": 2.409650682862991,
      "grad_norm": 0.7578125,
      "learning_rate": 2.645174719562371e-05,
      "loss": 0.9423,
      "step": 4488
    },
    {
      "epoch": 2.4101875775980672,
      "grad_norm": 0.6484375,
      "learning_rate": 2.6443316222415943e-05,
      "loss": 1.0277,
      "step": 4489
    },
    {
      "epoch": 2.4107244723331434,
      "grad_norm": 0.671875,
      "learning_rate": 2.6434885084507517e-05,
      "loss": 0.8967,
      "step": 4490
    },
    {
      "epoch": 2.411261367068219,
      "grad_norm": 0.69921875,
      "learning_rate": 2.6426453782860545e-05,
      "loss": 0.8638,
      "step": 4491
    },
    {
      "epoch": 2.411798261803295,
      "grad_norm": 0.81640625,
      "learning_rate": 2.6418022318437114e-05,
      "loss": 0.9491,
      "step": 4492
    },
    {
      "epoch": 2.4123351565383713,
      "grad_norm": 0.72265625,
      "learning_rate": 2.640959069219938e-05,
      "loss": 0.9281,
      "step": 4493
    },
    {
      "epoch": 2.4128720512734474,
      "grad_norm": 0.62890625,
      "learning_rate": 2.6401158905109506e-05,
      "loss": 0.9745,
      "step": 4494
    },
    {
      "epoch": 2.413408946008523,
      "grad_norm": 0.7734375,
      "learning_rate": 2.6392726958129653e-05,
      "loss": 0.9925,
      "step": 4495
    },
    {
      "epoch": 2.413945840743599,
      "grad_norm": 0.76171875,
      "learning_rate": 2.6384294852222018e-05,
      "loss": 0.8337,
      "step": 4496
    },
    {
      "epoch": 2.4144827354786753,
      "grad_norm": 0.57421875,
      "learning_rate": 2.637586258834881e-05,
      "loss": 0.6912,
      "step": 4497
    },
    {
      "epoch": 2.4150196302137514,
      "grad_norm": 0.78125,
      "learning_rate": 2.636743016747225e-05,
      "loss": 0.9982,
      "step": 4498
    },
    {
      "epoch": 2.415556524948827,
      "grad_norm": 0.80859375,
      "learning_rate": 2.63589975905546e-05,
      "loss": 0.8983,
      "step": 4499
    },
    {
      "epoch": 2.4160934196839032,
      "grad_norm": 0.70703125,
      "learning_rate": 2.6350564858558107e-05,
      "loss": 0.8172,
      "step": 4500
    },
    {
      "epoch": 2.4160934196839032,
      "eval_loss": 0.9049860835075378,
      "eval_runtime": 45.937,
      "eval_samples_per_second": 20.942,
      "eval_steps_per_second": 20.942,
      "step": 4500
    },
    {
      "epoch": 2.4166303144189794,
      "grad_norm": 0.7109375,
      "learning_rate": 2.6342131972445067e-05,
      "loss": 0.8297,
      "step": 4501
    },
    {
      "epoch": 2.417167209154055,
      "grad_norm": 0.921875,
      "learning_rate": 2.633369893317777e-05,
      "loss": 0.9613,
      "step": 4502
    },
    {
      "epoch": 2.417704103889131,
      "grad_norm": 0.65625,
      "learning_rate": 2.6325265741718537e-05,
      "loss": 0.7996,
      "step": 4503
    },
    {
      "epoch": 2.4182409986242073,
      "grad_norm": 0.6796875,
      "learning_rate": 2.6316832399029696e-05,
      "loss": 0.6878,
      "step": 4504
    },
    {
      "epoch": 2.4187778933592834,
      "grad_norm": 0.71484375,
      "learning_rate": 2.63083989060736e-05,
      "loss": 0.9636,
      "step": 4505
    },
    {
      "epoch": 2.419314788094359,
      "grad_norm": 0.84375,
      "learning_rate": 2.6299965263812616e-05,
      "loss": 0.9298,
      "step": 4506
    },
    {
      "epoch": 2.419851682829435,
      "grad_norm": 0.85546875,
      "learning_rate": 2.6291531473209148e-05,
      "loss": 0.7767,
      "step": 4507
    },
    {
      "epoch": 2.4203885775645113,
      "grad_norm": 0.71484375,
      "learning_rate": 2.6283097535225565e-05,
      "loss": 0.699,
      "step": 4508
    },
    {
      "epoch": 2.4209254722995874,
      "grad_norm": 0.703125,
      "learning_rate": 2.6274663450824318e-05,
      "loss": 0.7578,
      "step": 4509
    },
    {
      "epoch": 2.421462367034663,
      "grad_norm": 0.6484375,
      "learning_rate": 2.6266229220967818e-05,
      "loss": 0.8686,
      "step": 4510
    },
    {
      "epoch": 2.4219992617697392,
      "grad_norm": 0.61328125,
      "learning_rate": 2.625779484661854e-05,
      "loss": 0.6708,
      "step": 4511
    },
    {
      "epoch": 2.4225361565048154,
      "grad_norm": 0.69921875,
      "learning_rate": 2.6249360328738932e-05,
      "loss": 0.7383,
      "step": 4512
    },
    {
      "epoch": 2.423073051239891,
      "grad_norm": 0.59765625,
      "learning_rate": 2.6240925668291496e-05,
      "loss": 0.7092,
      "step": 4513
    },
    {
      "epoch": 2.423609945974967,
      "grad_norm": 0.62890625,
      "learning_rate": 2.6232490866238714e-05,
      "loss": 0.9468,
      "step": 4514
    },
    {
      "epoch": 2.4241468407100433,
      "grad_norm": 0.75390625,
      "learning_rate": 2.622405592354312e-05,
      "loss": 0.8644,
      "step": 4515
    },
    {
      "epoch": 2.4246837354451194,
      "grad_norm": 0.67578125,
      "learning_rate": 2.621562084116726e-05,
      "loss": 0.9886,
      "step": 4516
    },
    {
      "epoch": 2.4252206301801955,
      "grad_norm": 0.7109375,
      "learning_rate": 2.6207185620073648e-05,
      "loss": 0.75,
      "step": 4517
    },
    {
      "epoch": 2.425757524915271,
      "grad_norm": 0.76953125,
      "learning_rate": 2.6198750261224873e-05,
      "loss": 0.7946,
      "step": 4518
    },
    {
      "epoch": 2.4262944196503473,
      "grad_norm": 0.74609375,
      "learning_rate": 2.6190314765583517e-05,
      "loss": 0.8414,
      "step": 4519
    },
    {
      "epoch": 2.4268313143854234,
      "grad_norm": 0.58984375,
      "learning_rate": 2.6181879134112173e-05,
      "loss": 0.7495,
      "step": 4520
    },
    {
      "epoch": 2.427368209120499,
      "grad_norm": 0.703125,
      "learning_rate": 2.6173443367773452e-05,
      "loss": 0.9884,
      "step": 4521
    },
    {
      "epoch": 2.4279051038555752,
      "grad_norm": 0.703125,
      "learning_rate": 2.6165007467529973e-05,
      "loss": 0.9648,
      "step": 4522
    },
    {
      "epoch": 2.4284419985906514,
      "grad_norm": 0.68359375,
      "learning_rate": 2.6156571434344394e-05,
      "loss": 0.7987,
      "step": 4523
    },
    {
      "epoch": 2.4289788933257275,
      "grad_norm": 0.69921875,
      "learning_rate": 2.614813526917938e-05,
      "loss": 0.8769,
      "step": 4524
    },
    {
      "epoch": 2.429515788060803,
      "grad_norm": 0.71875,
      "learning_rate": 2.6139698972997573e-05,
      "loss": 0.8598,
      "step": 4525
    },
    {
      "epoch": 2.4300526827958793,
      "grad_norm": 0.62890625,
      "learning_rate": 2.6131262546761693e-05,
      "loss": 0.9152,
      "step": 4526
    },
    {
      "epoch": 2.4305895775309554,
      "grad_norm": 0.8125,
      "learning_rate": 2.6122825991434425e-05,
      "loss": 0.8841,
      "step": 4527
    },
    {
      "epoch": 2.4311264722660315,
      "grad_norm": 0.74609375,
      "learning_rate": 2.611438930797849e-05,
      "loss": 0.8423,
      "step": 4528
    },
    {
      "epoch": 2.431663367001107,
      "grad_norm": 0.7421875,
      "learning_rate": 2.6105952497356628e-05,
      "loss": 0.926,
      "step": 4529
    },
    {
      "epoch": 2.4322002617361833,
      "grad_norm": 0.81640625,
      "learning_rate": 2.609751556053157e-05,
      "loss": 0.866,
      "step": 4530
    },
    {
      "epoch": 2.4327371564712594,
      "grad_norm": 0.671875,
      "learning_rate": 2.6089078498466096e-05,
      "loss": 0.8663,
      "step": 4531
    },
    {
      "epoch": 2.433274051206335,
      "grad_norm": 0.73828125,
      "learning_rate": 2.608064131212298e-05,
      "loss": 0.7887,
      "step": 4532
    },
    {
      "epoch": 2.4338109459414112,
      "grad_norm": 0.75,
      "learning_rate": 2.607220400246499e-05,
      "loss": 0.8505,
      "step": 4533
    },
    {
      "epoch": 2.4343478406764874,
      "grad_norm": 0.76171875,
      "learning_rate": 2.606376657045495e-05,
      "loss": 0.8075,
      "step": 4534
    },
    {
      "epoch": 2.4348847354115635,
      "grad_norm": 0.61328125,
      "learning_rate": 2.6055329017055673e-05,
      "loss": 0.8791,
      "step": 4535
    },
    {
      "epoch": 2.4354216301466396,
      "grad_norm": 0.76171875,
      "learning_rate": 2.604689134322999e-05,
      "loss": 1.1443,
      "step": 4536
    },
    {
      "epoch": 2.4359585248817153,
      "grad_norm": 0.89453125,
      "learning_rate": 2.6038453549940755e-05,
      "loss": 0.9011,
      "step": 4537
    },
    {
      "epoch": 2.4364954196167914,
      "grad_norm": 0.625,
      "learning_rate": 2.603001563815081e-05,
      "loss": 0.797,
      "step": 4538
    },
    {
      "epoch": 2.4370323143518675,
      "grad_norm": 0.75390625,
      "learning_rate": 2.602157760882304e-05,
      "loss": 0.9502,
      "step": 4539
    },
    {
      "epoch": 2.437569209086943,
      "grad_norm": 0.65625,
      "learning_rate": 2.6013139462920328e-05,
      "loss": 0.85,
      "step": 4540
    },
    {
      "epoch": 2.4381061038220193,
      "grad_norm": 0.69921875,
      "learning_rate": 2.6004701201405568e-05,
      "loss": 0.7806,
      "step": 4541
    },
    {
      "epoch": 2.4386429985570954,
      "grad_norm": 0.6875,
      "learning_rate": 2.5996262825241686e-05,
      "loss": 0.7601,
      "step": 4542
    },
    {
      "epoch": 2.4391798932921716,
      "grad_norm": 0.8046875,
      "learning_rate": 2.598782433539159e-05,
      "loss": 0.7733,
      "step": 4543
    },
    {
      "epoch": 2.4397167880272472,
      "grad_norm": 0.75,
      "learning_rate": 2.5979385732818228e-05,
      "loss": 0.8172,
      "step": 4544
    },
    {
      "epoch": 2.4402536827623234,
      "grad_norm": 0.73828125,
      "learning_rate": 2.5970947018484564e-05,
      "loss": 0.824,
      "step": 4545
    },
    {
      "epoch": 2.4407905774973995,
      "grad_norm": 0.69140625,
      "learning_rate": 2.5962508193353542e-05,
      "loss": 0.8564,
      "step": 4546
    },
    {
      "epoch": 2.4413274722324756,
      "grad_norm": 1.0,
      "learning_rate": 2.5954069258388137e-05,
      "loss": 1.0499,
      "step": 4547
    },
    {
      "epoch": 2.4418643669675513,
      "grad_norm": 0.71875,
      "learning_rate": 2.594563021455135e-05,
      "loss": 0.8332,
      "step": 4548
    },
    {
      "epoch": 2.4424012617026274,
      "grad_norm": 0.72265625,
      "learning_rate": 2.5937191062806187e-05,
      "loss": 0.7358,
      "step": 4549
    },
    {
      "epoch": 2.4429381564377035,
      "grad_norm": 0.73046875,
      "learning_rate": 2.5928751804115652e-05,
      "loss": 0.979,
      "step": 4550
    },
    {
      "epoch": 2.4429381564377035,
      "eval_loss": 0.9049565196037292,
      "eval_runtime": 47.1158,
      "eval_samples_per_second": 20.418,
      "eval_steps_per_second": 20.418,
      "step": 4550
    },
    {
      "epoch": 2.443475051172779,
      "grad_norm": 0.76953125,
      "learning_rate": 2.592031243944278e-05,
      "loss": 1.0015,
      "step": 4551
    },
    {
      "epoch": 2.4440119459078553,
      "grad_norm": 0.69140625,
      "learning_rate": 2.5911872969750595e-05,
      "loss": 0.7109,
      "step": 4552
    },
    {
      "epoch": 2.4445488406429314,
      "grad_norm": 0.68359375,
      "learning_rate": 2.5903433396002164e-05,
      "loss": 0.8791,
      "step": 4553
    },
    {
      "epoch": 2.4450857353780076,
      "grad_norm": 0.765625,
      "learning_rate": 2.5894993719160536e-05,
      "loss": 0.9258,
      "step": 4554
    },
    {
      "epoch": 2.4456226301130837,
      "grad_norm": 0.71484375,
      "learning_rate": 2.588655394018879e-05,
      "loss": 0.8183,
      "step": 4555
    },
    {
      "epoch": 2.4461595248481594,
      "grad_norm": 0.62890625,
      "learning_rate": 2.587811406005001e-05,
      "loss": 0.8501,
      "step": 4556
    },
    {
      "epoch": 2.4466964195832355,
      "grad_norm": 1.03125,
      "learning_rate": 2.58696740797073e-05,
      "loss": 0.8437,
      "step": 4557
    },
    {
      "epoch": 2.4472333143183116,
      "grad_norm": 0.73046875,
      "learning_rate": 2.5861234000123767e-05,
      "loss": 0.8139,
      "step": 4558
    },
    {
      "epoch": 2.4477702090533873,
      "grad_norm": 0.6875,
      "learning_rate": 2.5852793822262528e-05,
      "loss": 0.9045,
      "step": 4559
    },
    {
      "epoch": 2.4483071037884634,
      "grad_norm": 0.66015625,
      "learning_rate": 2.584435354708671e-05,
      "loss": 0.7598,
      "step": 4560
    },
    {
      "epoch": 2.4488439985235395,
      "grad_norm": 0.70703125,
      "learning_rate": 2.583591317555945e-05,
      "loss": 0.9663,
      "step": 4561
    },
    {
      "epoch": 2.4493808932586156,
      "grad_norm": 0.70703125,
      "learning_rate": 2.5827472708643928e-05,
      "loss": 0.9587,
      "step": 4562
    },
    {
      "epoch": 2.4499177879936913,
      "grad_norm": 0.67578125,
      "learning_rate": 2.581903214730328e-05,
      "loss": 0.8771,
      "step": 4563
    },
    {
      "epoch": 2.4504546827287674,
      "grad_norm": 0.84765625,
      "learning_rate": 2.5810591492500695e-05,
      "loss": 0.9174,
      "step": 4564
    },
    {
      "epoch": 2.4509915774638436,
      "grad_norm": 0.73046875,
      "learning_rate": 2.5802150745199355e-05,
      "loss": 0.788,
      "step": 4565
    },
    {
      "epoch": 2.4515284721989197,
      "grad_norm": 0.58984375,
      "learning_rate": 2.5793709906362463e-05,
      "loss": 0.6663,
      "step": 4566
    },
    {
      "epoch": 2.4520653669339953,
      "grad_norm": 0.6875,
      "learning_rate": 2.578526897695321e-05,
      "loss": 0.9211,
      "step": 4567
    },
    {
      "epoch": 2.4526022616690715,
      "grad_norm": 0.73828125,
      "learning_rate": 2.577682795793483e-05,
      "loss": 1.0319,
      "step": 4568
    },
    {
      "epoch": 2.4531391564041476,
      "grad_norm": 0.7109375,
      "learning_rate": 2.576838685027053e-05,
      "loss": 0.8483,
      "step": 4569
    },
    {
      "epoch": 2.4536760511392233,
      "grad_norm": 0.66015625,
      "learning_rate": 2.5759945654923575e-05,
      "loss": 0.7972,
      "step": 4570
    },
    {
      "epoch": 2.4542129458742994,
      "grad_norm": 0.71875,
      "learning_rate": 2.575150437285718e-05,
      "loss": 0.916,
      "step": 4571
    },
    {
      "epoch": 2.4547498406093755,
      "grad_norm": 0.75390625,
      "learning_rate": 2.5743063005034628e-05,
      "loss": 0.7371,
      "step": 4572
    },
    {
      "epoch": 2.4552867353444516,
      "grad_norm": 0.73828125,
      "learning_rate": 2.573462155241918e-05,
      "loss": 0.7676,
      "step": 4573
    },
    {
      "epoch": 2.4558236300795278,
      "grad_norm": 0.79296875,
      "learning_rate": 2.572618001597411e-05,
      "loss": 0.851,
      "step": 4574
    },
    {
      "epoch": 2.4563605248146034,
      "grad_norm": 0.66796875,
      "learning_rate": 2.57177383966627e-05,
      "loss": 0.7246,
      "step": 4575
    },
    {
      "epoch": 2.4568974195496796,
      "grad_norm": 0.69140625,
      "learning_rate": 2.570929669544825e-05,
      "loss": 0.9511,
      "step": 4576
    },
    {
      "epoch": 2.4574343142847557,
      "grad_norm": 0.85546875,
      "learning_rate": 2.570085491329406e-05,
      "loss": 0.8849,
      "step": 4577
    },
    {
      "epoch": 2.4579712090198313,
      "grad_norm": 0.6953125,
      "learning_rate": 2.5692413051163457e-05,
      "loss": 0.8535,
      "step": 4578
    },
    {
      "epoch": 2.4585081037549075,
      "grad_norm": 0.82421875,
      "learning_rate": 2.5683971110019755e-05,
      "loss": 0.807,
      "step": 4579
    },
    {
      "epoch": 2.4590449984899836,
      "grad_norm": 0.6796875,
      "learning_rate": 2.5675529090826277e-05,
      "loss": 0.9063,
      "step": 4580
    },
    {
      "epoch": 2.4595818932250597,
      "grad_norm": 0.67578125,
      "learning_rate": 2.5667086994546387e-05,
      "loss": 0.7945,
      "step": 4581
    },
    {
      "epoch": 2.4601187879601354,
      "grad_norm": 0.65234375,
      "learning_rate": 2.565864482214342e-05,
      "loss": 0.8773,
      "step": 4582
    },
    {
      "epoch": 2.4606556826952115,
      "grad_norm": 0.66015625,
      "learning_rate": 2.565020257458074e-05,
      "loss": 0.6864,
      "step": 4583
    },
    {
      "epoch": 2.4611925774302876,
      "grad_norm": 0.71484375,
      "learning_rate": 2.5641760252821706e-05,
      "loss": 0.8717,
      "step": 4584
    },
    {
      "epoch": 2.4617294721653638,
      "grad_norm": 0.74609375,
      "learning_rate": 2.5633317857829697e-05,
      "loss": 0.7126,
      "step": 4585
    },
    {
      "epoch": 2.4622663669004394,
      "grad_norm": 0.61328125,
      "learning_rate": 2.562487539056811e-05,
      "loss": 0.8311,
      "step": 4586
    },
    {
      "epoch": 2.4628032616355156,
      "grad_norm": 0.6953125,
      "learning_rate": 2.5616432852000326e-05,
      "loss": 0.8652,
      "step": 4587
    },
    {
      "epoch": 2.4633401563705917,
      "grad_norm": 0.62890625,
      "learning_rate": 2.560799024308974e-05,
      "loss": 0.6975,
      "step": 4588
    },
    {
      "epoch": 2.4638770511056673,
      "grad_norm": 0.6953125,
      "learning_rate": 2.5599547564799777e-05,
      "loss": 0.885,
      "step": 4589
    },
    {
      "epoch": 2.4644139458407435,
      "grad_norm": 0.6171875,
      "learning_rate": 2.5591104818093837e-05,
      "loss": 0.9691,
      "step": 4590
    },
    {
      "epoch": 2.4649508405758196,
      "grad_norm": 0.578125,
      "learning_rate": 2.5582662003935353e-05,
      "loss": 0.5909,
      "step": 4591
    },
    {
      "epoch": 2.4654877353108957,
      "grad_norm": 0.66796875,
      "learning_rate": 2.5574219123287758e-05,
      "loss": 0.6404,
      "step": 4592
    },
    {
      "epoch": 2.466024630045972,
      "grad_norm": 0.73828125,
      "learning_rate": 2.556577617711448e-05,
      "loss": 0.9185,
      "step": 4593
    },
    {
      "epoch": 2.4665615247810475,
      "grad_norm": 0.87109375,
      "learning_rate": 2.5557333166378982e-05,
      "loss": 0.9454,
      "step": 4594
    },
    {
      "epoch": 2.4670984195161236,
      "grad_norm": 0.62890625,
      "learning_rate": 2.554889009204472e-05,
      "loss": 0.909,
      "step": 4595
    },
    {
      "epoch": 2.4676353142511998,
      "grad_norm": 0.70703125,
      "learning_rate": 2.5540446955075137e-05,
      "loss": 0.9791,
      "step": 4596
    },
    {
      "epoch": 2.4681722089862754,
      "grad_norm": 0.66796875,
      "learning_rate": 2.5532003756433714e-05,
      "loss": 0.9859,
      "step": 4597
    },
    {
      "epoch": 2.4687091037213515,
      "grad_norm": 0.6640625,
      "learning_rate": 2.5523560497083926e-05,
      "loss": 0.9977,
      "step": 4598
    },
    {
      "epoch": 2.4692459984564277,
      "grad_norm": 0.82421875,
      "learning_rate": 2.551511717798925e-05,
      "loss": 0.8019,
      "step": 4599
    },
    {
      "epoch": 2.469782893191504,
      "grad_norm": 0.7890625,
      "learning_rate": 2.5506673800113194e-05,
      "loss": 0.9173,
      "step": 4600
    },
    {
      "epoch": 2.469782893191504,
      "eval_loss": 0.9041844606399536,
      "eval_runtime": 46.0744,
      "eval_samples_per_second": 20.879,
      "eval_steps_per_second": 20.879,
      "step": 4600
    },
    {
      "epoch": 2.4703197879265795,
      "grad_norm": 0.6328125,
      "learning_rate": 2.5498230364419234e-05,
      "loss": 0.8684,
      "step": 4601
    },
    {
      "epoch": 2.4708566826616556,
      "grad_norm": 0.69140625,
      "learning_rate": 2.548978687187087e-05,
      "loss": 0.7525,
      "step": 4602
    },
    {
      "epoch": 2.4713935773967317,
      "grad_norm": 0.68359375,
      "learning_rate": 2.5481343323431634e-05,
      "loss": 0.8157,
      "step": 4603
    },
    {
      "epoch": 2.471930472131808,
      "grad_norm": 0.67578125,
      "learning_rate": 2.5472899720065025e-05,
      "loss": 0.6885,
      "step": 4604
    },
    {
      "epoch": 2.4724673668668835,
      "grad_norm": 0.7109375,
      "learning_rate": 2.5464456062734566e-05,
      "loss": 0.9621,
      "step": 4605
    },
    {
      "epoch": 2.4730042616019596,
      "grad_norm": 0.69921875,
      "learning_rate": 2.545601235240379e-05,
      "loss": 0.847,
      "step": 4606
    },
    {
      "epoch": 2.4735411563370358,
      "grad_norm": 0.63671875,
      "learning_rate": 2.5447568590036232e-05,
      "loss": 0.8625,
      "step": 4607
    },
    {
      "epoch": 2.4740780510721114,
      "grad_norm": 0.703125,
      "learning_rate": 2.543912477659543e-05,
      "loss": 0.9182,
      "step": 4608
    },
    {
      "epoch": 2.4746149458071875,
      "grad_norm": 0.64453125,
      "learning_rate": 2.543068091304493e-05,
      "loss": 0.849,
      "step": 4609
    },
    {
      "epoch": 2.4751518405422637,
      "grad_norm": 0.625,
      "learning_rate": 2.5422237000348276e-05,
      "loss": 0.6942,
      "step": 4610
    },
    {
      "epoch": 2.47568873527734,
      "grad_norm": 0.83203125,
      "learning_rate": 2.5413793039469043e-05,
      "loss": 1.0324,
      "step": 4611
    },
    {
      "epoch": 2.476225630012416,
      "grad_norm": 0.6328125,
      "learning_rate": 2.540534903137078e-05,
      "loss": 0.7678,
      "step": 4612
    },
    {
      "epoch": 2.4767625247474916,
      "grad_norm": 0.6171875,
      "learning_rate": 2.539690497701706e-05,
      "loss": 0.6931,
      "step": 4613
    },
    {
      "epoch": 2.4772994194825677,
      "grad_norm": 0.69921875,
      "learning_rate": 2.5388460877371455e-05,
      "loss": 0.7073,
      "step": 4614
    },
    {
      "epoch": 2.477836314217644,
      "grad_norm": 0.73828125,
      "learning_rate": 2.538001673339754e-05,
      "loss": 0.921,
      "step": 4615
    },
    {
      "epoch": 2.4783732089527195,
      "grad_norm": 0.72265625,
      "learning_rate": 2.5371572546058915e-05,
      "loss": 0.8605,
      "step": 4616
    },
    {
      "epoch": 2.4789101036877956,
      "grad_norm": 0.59375,
      "learning_rate": 2.5363128316319153e-05,
      "loss": 0.7955,
      "step": 4617
    },
    {
      "epoch": 2.4794469984228718,
      "grad_norm": 0.72265625,
      "learning_rate": 2.535468404514184e-05,
      "loss": 0.7811,
      "step": 4618
    },
    {
      "epoch": 2.479983893157948,
      "grad_norm": 0.60546875,
      "learning_rate": 2.5346239733490594e-05,
      "loss": 0.7526,
      "step": 4619
    },
    {
      "epoch": 2.4805207878930235,
      "grad_norm": 0.67578125,
      "learning_rate": 2.533779538232901e-05,
      "loss": 0.7551,
      "step": 4620
    },
    {
      "epoch": 2.4810576826280997,
      "grad_norm": 0.65234375,
      "learning_rate": 2.5329350992620697e-05,
      "loss": 0.7493,
      "step": 4621
    },
    {
      "epoch": 2.481594577363176,
      "grad_norm": 0.6796875,
      "learning_rate": 2.532090656532926e-05,
      "loss": 0.957,
      "step": 4622
    },
    {
      "epoch": 2.482131472098252,
      "grad_norm": 0.64453125,
      "learning_rate": 2.5312462101418326e-05,
      "loss": 0.8087,
      "step": 4623
    },
    {
      "epoch": 2.4826683668333276,
      "grad_norm": 1.15625,
      "learning_rate": 2.53040176018515e-05,
      "loss": 0.6707,
      "step": 4624
    },
    {
      "epoch": 2.4832052615684037,
      "grad_norm": 0.7578125,
      "learning_rate": 2.5295573067592428e-05,
      "loss": 0.9298,
      "step": 4625
    },
    {
      "epoch": 2.48374215630348,
      "grad_norm": 0.73046875,
      "learning_rate": 2.5287128499604712e-05,
      "loss": 0.9217,
      "step": 4626
    },
    {
      "epoch": 2.4842790510385555,
      "grad_norm": 0.58203125,
      "learning_rate": 2.5278683898852006e-05,
      "loss": 0.7417,
      "step": 4627
    },
    {
      "epoch": 2.4848159457736316,
      "grad_norm": 0.828125,
      "learning_rate": 2.527023926629794e-05,
      "loss": 0.8912,
      "step": 4628
    },
    {
      "epoch": 2.4853528405087078,
      "grad_norm": 0.62109375,
      "learning_rate": 2.5261794602906145e-05,
      "loss": 0.8226,
      "step": 4629
    },
    {
      "epoch": 2.485889735243784,
      "grad_norm": 0.57421875,
      "learning_rate": 2.5253349909640278e-05,
      "loss": 0.6685,
      "step": 4630
    },
    {
      "epoch": 2.48642662997886,
      "grad_norm": 0.92578125,
      "learning_rate": 2.5244905187463973e-05,
      "loss": 0.8736,
      "step": 4631
    },
    {
      "epoch": 2.4869635247139357,
      "grad_norm": 0.78125,
      "learning_rate": 2.523646043734088e-05,
      "loss": 0.83,
      "step": 4632
    },
    {
      "epoch": 2.487500419449012,
      "grad_norm": 0.796875,
      "learning_rate": 2.522801566023467e-05,
      "loss": 0.9054,
      "step": 4633
    },
    {
      "epoch": 2.488037314184088,
      "grad_norm": 0.67578125,
      "learning_rate": 2.521957085710897e-05,
      "loss": 0.9039,
      "step": 4634
    },
    {
      "epoch": 2.4885742089191636,
      "grad_norm": 0.62109375,
      "learning_rate": 2.5211126028927464e-05,
      "loss": 0.7891,
      "step": 4635
    },
    {
      "epoch": 2.4891111036542397,
      "grad_norm": 0.72265625,
      "learning_rate": 2.5202681176653803e-05,
      "loss": 0.9205,
      "step": 4636
    },
    {
      "epoch": 2.489647998389316,
      "grad_norm": 0.6640625,
      "learning_rate": 2.5194236301251657e-05,
      "loss": 0.6967,
      "step": 4637
    },
    {
      "epoch": 2.490184893124392,
      "grad_norm": 0.58984375,
      "learning_rate": 2.5185791403684682e-05,
      "loss": 0.7991,
      "step": 4638
    },
    {
      "epoch": 2.4907217878594676,
      "grad_norm": 0.6015625,
      "learning_rate": 2.517734648491656e-05,
      "loss": 0.7712,
      "step": 4639
    },
    {
      "epoch": 2.4912586825945437,
      "grad_norm": 0.640625,
      "learning_rate": 2.516890154591095e-05,
      "loss": 0.7971,
      "step": 4640
    },
    {
      "epoch": 2.49179557732962,
      "grad_norm": 0.71875,
      "learning_rate": 2.5160456587631544e-05,
      "loss": 0.9565,
      "step": 4641
    },
    {
      "epoch": 2.492332472064696,
      "grad_norm": 0.66796875,
      "learning_rate": 2.515201161104201e-05,
      "loss": 0.9156,
      "step": 4642
    },
    {
      "epoch": 2.4928693667997717,
      "grad_norm": 0.73046875,
      "learning_rate": 2.514356661710601e-05,
      "loss": 0.8908,
      "step": 4643
    },
    {
      "epoch": 2.493406261534848,
      "grad_norm": 0.6484375,
      "learning_rate": 2.513512160678726e-05,
      "loss": 0.7587,
      "step": 4644
    },
    {
      "epoch": 2.493943156269924,
      "grad_norm": 0.7109375,
      "learning_rate": 2.512667658104941e-05,
      "loss": 0.7787,
      "step": 4645
    },
    {
      "epoch": 2.4944800510049996,
      "grad_norm": 0.69140625,
      "learning_rate": 2.511823154085617e-05,
      "loss": 0.8485,
      "step": 4646
    },
    {
      "epoch": 2.4950169457400757,
      "grad_norm": 0.6796875,
      "learning_rate": 2.510978648717121e-05,
      "loss": 0.7542,
      "step": 4647
    },
    {
      "epoch": 2.495553840475152,
      "grad_norm": 0.62109375,
      "learning_rate": 2.5101341420958212e-05,
      "loss": 0.8525,
      "step": 4648
    },
    {
      "epoch": 2.496090735210228,
      "grad_norm": 0.7421875,
      "learning_rate": 2.509289634318088e-05,
      "loss": 0.8682,
      "step": 4649
    },
    {
      "epoch": 2.496627629945304,
      "grad_norm": 0.7890625,
      "learning_rate": 2.508445125480291e-05,
      "loss": 0.8936,
      "step": 4650
    },
    {
      "epoch": 2.496627629945304,
      "eval_loss": 0.9043004512786865,
      "eval_runtime": 46.2228,
      "eval_samples_per_second": 20.812,
      "eval_steps_per_second": 20.812,
      "step": 4650
    },
    {
      "epoch": 2.4971645246803797,
      "grad_norm": 0.78515625,
      "learning_rate": 2.507600615678797e-05,
      "loss": 0.7684,
      "step": 4651
    },
    {
      "epoch": 2.497701419415456,
      "grad_norm": 0.83203125,
      "learning_rate": 2.506756105009977e-05,
      "loss": 0.9203,
      "step": 4652
    },
    {
      "epoch": 2.498238314150532,
      "grad_norm": 0.7109375,
      "learning_rate": 2.5059115935701992e-05,
      "loss": 1.0308,
      "step": 4653
    },
    {
      "epoch": 2.4987752088856077,
      "grad_norm": 0.59375,
      "learning_rate": 2.5050670814558346e-05,
      "loss": 0.7377,
      "step": 4654
    },
    {
      "epoch": 2.499312103620684,
      "grad_norm": 0.6171875,
      "learning_rate": 2.5042225687632514e-05,
      "loss": 0.8715,
      "step": 4655
    },
    {
      "epoch": 2.49984899835576,
      "grad_norm": 0.71875,
      "learning_rate": 2.5033780555888193e-05,
      "loss": 0.9159,
      "step": 4656
    },
    {
      "epoch": 2.500385893090836,
      "grad_norm": 0.65625,
      "learning_rate": 2.5025335420289086e-05,
      "loss": 0.8335,
      "step": 4657
    },
    {
      "epoch": 2.500922787825912,
      "grad_norm": 0.73046875,
      "learning_rate": 2.5016890281798893e-05,
      "loss": 1.0118,
      "step": 4658
    },
    {
      "epoch": 2.501459682560988,
      "grad_norm": 0.65625,
      "learning_rate": 2.5008445141381294e-05,
      "loss": 0.7462,
      "step": 4659
    },
    {
      "epoch": 2.501996577296064,
      "grad_norm": 0.64453125,
      "learning_rate": 2.5e-05,
      "loss": 0.8913,
      "step": 4660
    },
    {
      "epoch": 2.50253347203114,
      "grad_norm": 0.75,
      "learning_rate": 2.4991554858618712e-05,
      "loss": 0.8428,
      "step": 4661
    },
    {
      "epoch": 2.5030703667662157,
      "grad_norm": 0.68359375,
      "learning_rate": 2.4983109718201112e-05,
      "loss": 1.022,
      "step": 4662
    },
    {
      "epoch": 2.503607261501292,
      "grad_norm": 0.68359375,
      "learning_rate": 2.4974664579710913e-05,
      "loss": 0.8398,
      "step": 4663
    },
    {
      "epoch": 2.504144156236368,
      "grad_norm": 0.6875,
      "learning_rate": 2.496621944411181e-05,
      "loss": 0.879,
      "step": 4664
    },
    {
      "epoch": 2.5046810509714437,
      "grad_norm": 0.65625,
      "learning_rate": 2.4957774312367495e-05,
      "loss": 0.8861,
      "step": 4665
    },
    {
      "epoch": 2.50521794570652,
      "grad_norm": 0.7265625,
      "learning_rate": 2.4949329185441666e-05,
      "loss": 0.8208,
      "step": 4666
    },
    {
      "epoch": 2.505754840441596,
      "grad_norm": 0.65234375,
      "learning_rate": 2.494088406429801e-05,
      "loss": 0.8499,
      "step": 4667
    },
    {
      "epoch": 2.506291735176672,
      "grad_norm": 0.72265625,
      "learning_rate": 2.493243894990024e-05,
      "loss": 0.8782,
      "step": 4668
    },
    {
      "epoch": 2.506828629911748,
      "grad_norm": 0.69921875,
      "learning_rate": 2.4923993843212037e-05,
      "loss": 0.768,
      "step": 4669
    },
    {
      "epoch": 2.507365524646824,
      "grad_norm": 1.0703125,
      "learning_rate": 2.4915548745197097e-05,
      "loss": 0.8833,
      "step": 4670
    },
    {
      "epoch": 2.5079024193819,
      "grad_norm": 0.921875,
      "learning_rate": 2.4907103656819118e-05,
      "loss": 0.972,
      "step": 4671
    },
    {
      "epoch": 2.508439314116976,
      "grad_norm": 0.6171875,
      "learning_rate": 2.4898658579041787e-05,
      "loss": 0.8168,
      "step": 4672
    },
    {
      "epoch": 2.5089762088520517,
      "grad_norm": 0.76953125,
      "learning_rate": 2.4890213512828803e-05,
      "loss": 1.0264,
      "step": 4673
    },
    {
      "epoch": 2.509513103587128,
      "grad_norm": 0.83203125,
      "learning_rate": 2.4881768459143844e-05,
      "loss": 0.8988,
      "step": 4674
    },
    {
      "epoch": 2.510049998322204,
      "grad_norm": 0.63671875,
      "learning_rate": 2.487332341895059e-05,
      "loss": 0.6743,
      "step": 4675
    },
    {
      "epoch": 2.51058689305728,
      "grad_norm": 0.68359375,
      "learning_rate": 2.4864878393212748e-05,
      "loss": 0.7572,
      "step": 4676
    },
    {
      "epoch": 2.5111237877923562,
      "grad_norm": 0.70703125,
      "learning_rate": 2.485643338289399e-05,
      "loss": 0.8108,
      "step": 4677
    },
    {
      "epoch": 2.511660682527432,
      "grad_norm": 0.7109375,
      "learning_rate": 2.4847988388957997e-05,
      "loss": 0.7347,
      "step": 4678
    },
    {
      "epoch": 2.512197577262508,
      "grad_norm": 0.6484375,
      "learning_rate": 2.483954341236846e-05,
      "loss": 0.9389,
      "step": 4679
    },
    {
      "epoch": 2.512734471997584,
      "grad_norm": 0.6796875,
      "learning_rate": 2.483109845408905e-05,
      "loss": 1.0592,
      "step": 4680
    },
    {
      "epoch": 2.51327136673266,
      "grad_norm": 0.6640625,
      "learning_rate": 2.4822653515083453e-05,
      "loss": 0.7301,
      "step": 4681
    },
    {
      "epoch": 2.513808261467736,
      "grad_norm": 0.76953125,
      "learning_rate": 2.481420859631533e-05,
      "loss": 1.0802,
      "step": 4682
    },
    {
      "epoch": 2.514345156202812,
      "grad_norm": 0.93359375,
      "learning_rate": 2.4805763698748352e-05,
      "loss": 0.8456,
      "step": 4683
    },
    {
      "epoch": 2.5148820509378877,
      "grad_norm": 0.7421875,
      "learning_rate": 2.4797318823346203e-05,
      "loss": 0.8313,
      "step": 4684
    },
    {
      "epoch": 2.515418945672964,
      "grad_norm": 0.76171875,
      "learning_rate": 2.478887397107254e-05,
      "loss": 0.8066,
      "step": 4685
    },
    {
      "epoch": 2.51595584040804,
      "grad_norm": 0.62109375,
      "learning_rate": 2.4780429142891033e-05,
      "loss": 0.7529,
      "step": 4686
    },
    {
      "epoch": 2.516492735143116,
      "grad_norm": 0.66796875,
      "learning_rate": 2.4771984339765338e-05,
      "loss": 0.7963,
      "step": 4687
    },
    {
      "epoch": 2.5170296298781922,
      "grad_norm": 0.73046875,
      "learning_rate": 2.476353956265912e-05,
      "loss": 0.9988,
      "step": 4688
    },
    {
      "epoch": 2.517566524613268,
      "grad_norm": 0.5859375,
      "learning_rate": 2.4755094812536033e-05,
      "loss": 0.8237,
      "step": 4689
    },
    {
      "epoch": 2.518103419348344,
      "grad_norm": 0.6796875,
      "learning_rate": 2.474665009035973e-05,
      "loss": 0.864,
      "step": 4690
    },
    {
      "epoch": 2.51864031408342,
      "grad_norm": 0.7421875,
      "learning_rate": 2.4738205397093864e-05,
      "loss": 0.7942,
      "step": 4691
    },
    {
      "epoch": 2.519177208818496,
      "grad_norm": 0.73828125,
      "learning_rate": 2.472976073370207e-05,
      "loss": 0.8739,
      "step": 4692
    },
    {
      "epoch": 2.519714103553572,
      "grad_norm": 0.6484375,
      "learning_rate": 2.4721316101148e-05,
      "loss": 0.7663,
      "step": 4693
    },
    {
      "epoch": 2.520250998288648,
      "grad_norm": 0.6328125,
      "learning_rate": 2.471287150039529e-05,
      "loss": 0.9158,
      "step": 4694
    },
    {
      "epoch": 2.520787893023724,
      "grad_norm": 0.69140625,
      "learning_rate": 2.4704426932407578e-05,
      "loss": 0.7818,
      "step": 4695
    },
    {
      "epoch": 2.5213247877588003,
      "grad_norm": 0.78125,
      "learning_rate": 2.4695982398148497e-05,
      "loss": 0.7843,
      "step": 4696
    },
    {
      "epoch": 2.521861682493876,
      "grad_norm": 0.58203125,
      "learning_rate": 2.468753789858168e-05,
      "loss": 0.7717,
      "step": 4697
    },
    {
      "epoch": 2.522398577228952,
      "grad_norm": 0.66796875,
      "learning_rate": 2.467909343467075e-05,
      "loss": 0.8969,
      "step": 4698
    },
    {
      "epoch": 2.5229354719640282,
      "grad_norm": 0.66015625,
      "learning_rate": 2.4670649007379315e-05,
      "loss": 0.8058,
      "step": 4699
    },
    {
      "epoch": 2.523472366699104,
      "grad_norm": 0.796875,
      "learning_rate": 2.4662204617670994e-05,
      "loss": 0.6992,
      "step": 4700
    },
    {
      "epoch": 2.523472366699104,
      "eval_loss": 0.9045475125312805,
      "eval_runtime": 45.9539,
      "eval_samples_per_second": 20.934,
      "eval_steps_per_second": 20.934,
      "step": 4700
    },
    {
      "epoch": 2.52400926143418,
      "grad_norm": 0.8046875,
      "learning_rate": 2.465376026650941e-05,
      "loss": 0.9277,
      "step": 4701
    },
    {
      "epoch": 2.524546156169256,
      "grad_norm": 0.74609375,
      "learning_rate": 2.4645315954858165e-05,
      "loss": 0.9898,
      "step": 4702
    },
    {
      "epoch": 2.525083050904332,
      "grad_norm": 0.59765625,
      "learning_rate": 2.4636871683680856e-05,
      "loss": 0.7192,
      "step": 4703
    },
    {
      "epoch": 2.525619945639408,
      "grad_norm": 0.71484375,
      "learning_rate": 2.462842745394109e-05,
      "loss": 0.7924,
      "step": 4704
    },
    {
      "epoch": 2.526156840374484,
      "grad_norm": 0.765625,
      "learning_rate": 2.461998326660246e-05,
      "loss": 0.9676,
      "step": 4705
    },
    {
      "epoch": 2.52669373510956,
      "grad_norm": 0.7265625,
      "learning_rate": 2.4611539122628544e-05,
      "loss": 1.0498,
      "step": 4706
    },
    {
      "epoch": 2.5272306298446363,
      "grad_norm": 1.3359375,
      "learning_rate": 2.460309502298295e-05,
      "loss": 0.8755,
      "step": 4707
    },
    {
      "epoch": 2.527767524579712,
      "grad_norm": 0.6015625,
      "learning_rate": 2.4594650968629226e-05,
      "loss": 0.7964,
      "step": 4708
    },
    {
      "epoch": 2.528304419314788,
      "grad_norm": 0.69921875,
      "learning_rate": 2.4586206960530963e-05,
      "loss": 0.9157,
      "step": 4709
    },
    {
      "epoch": 2.5288413140498642,
      "grad_norm": 0.77734375,
      "learning_rate": 2.4577762999651726e-05,
      "loss": 0.9714,
      "step": 4710
    },
    {
      "epoch": 2.52937820878494,
      "grad_norm": 0.76171875,
      "learning_rate": 2.4569319086955075e-05,
      "loss": 0.7733,
      "step": 4711
    },
    {
      "epoch": 2.529915103520016,
      "grad_norm": 0.63671875,
      "learning_rate": 2.4560875223404573e-05,
      "loss": 0.7672,
      "step": 4712
    },
    {
      "epoch": 2.530451998255092,
      "grad_norm": 0.61328125,
      "learning_rate": 2.455243140996377e-05,
      "loss": 0.7962,
      "step": 4713
    },
    {
      "epoch": 2.5309888929901683,
      "grad_norm": 0.765625,
      "learning_rate": 2.4543987647596206e-05,
      "loss": 0.7733,
      "step": 4714
    },
    {
      "epoch": 2.5315257877252444,
      "grad_norm": 0.65625,
      "learning_rate": 2.4535543937265443e-05,
      "loss": 0.7989,
      "step": 4715
    },
    {
      "epoch": 2.53206268246032,
      "grad_norm": 0.640625,
      "learning_rate": 2.4527100279934985e-05,
      "loss": 0.8543,
      "step": 4716
    },
    {
      "epoch": 2.532599577195396,
      "grad_norm": 1.2109375,
      "learning_rate": 2.4518656676568376e-05,
      "loss": 0.9989,
      "step": 4717
    },
    {
      "epoch": 2.5331364719304723,
      "grad_norm": 0.6171875,
      "learning_rate": 2.4510213128129132e-05,
      "loss": 0.7076,
      "step": 4718
    },
    {
      "epoch": 2.533673366665548,
      "grad_norm": 0.69140625,
      "learning_rate": 2.4501769635580772e-05,
      "loss": 0.8046,
      "step": 4719
    },
    {
      "epoch": 2.534210261400624,
      "grad_norm": 0.72265625,
      "learning_rate": 2.4493326199886812e-05,
      "loss": 0.8405,
      "step": 4720
    },
    {
      "epoch": 2.5347471561357002,
      "grad_norm": 0.62109375,
      "learning_rate": 2.4484882822010748e-05,
      "loss": 0.8945,
      "step": 4721
    },
    {
      "epoch": 2.535284050870776,
      "grad_norm": 0.7109375,
      "learning_rate": 2.447643950291608e-05,
      "loss": 0.9577,
      "step": 4722
    },
    {
      "epoch": 2.535820945605852,
      "grad_norm": 0.62109375,
      "learning_rate": 2.4467996243566295e-05,
      "loss": 0.8425,
      "step": 4723
    },
    {
      "epoch": 2.536357840340928,
      "grad_norm": 0.7890625,
      "learning_rate": 2.445955304492487e-05,
      "loss": 1.0657,
      "step": 4724
    },
    {
      "epoch": 2.5368947350760043,
      "grad_norm": 0.7734375,
      "learning_rate": 2.445110990795529e-05,
      "loss": 0.9352,
      "step": 4725
    },
    {
      "epoch": 2.5374316298110804,
      "grad_norm": 0.6953125,
      "learning_rate": 2.444266683362102e-05,
      "loss": 0.836,
      "step": 4726
    },
    {
      "epoch": 2.537968524546156,
      "grad_norm": 0.6171875,
      "learning_rate": 2.4434223822885526e-05,
      "loss": 0.7781,
      "step": 4727
    },
    {
      "epoch": 2.538505419281232,
      "grad_norm": 0.6171875,
      "learning_rate": 2.4425780876712248e-05,
      "loss": 0.7379,
      "step": 4728
    },
    {
      "epoch": 2.5390423140163083,
      "grad_norm": 0.74609375,
      "learning_rate": 2.4417337996064653e-05,
      "loss": 0.9583,
      "step": 4729
    },
    {
      "epoch": 2.539579208751384,
      "grad_norm": 0.6875,
      "learning_rate": 2.440889518190617e-05,
      "loss": 0.8795,
      "step": 4730
    },
    {
      "epoch": 2.54011610348646,
      "grad_norm": 0.76953125,
      "learning_rate": 2.4400452435200226e-05,
      "loss": 0.9928,
      "step": 4731
    },
    {
      "epoch": 2.5406529982215362,
      "grad_norm": 0.7734375,
      "learning_rate": 2.439200975691027e-05,
      "loss": 0.9659,
      "step": 4732
    },
    {
      "epoch": 2.5411898929566124,
      "grad_norm": 0.59765625,
      "learning_rate": 2.4383567147999683e-05,
      "loss": 0.9078,
      "step": 4733
    },
    {
      "epoch": 2.5417267876916885,
      "grad_norm": 0.62890625,
      "learning_rate": 2.43751246094319e-05,
      "loss": 0.666,
      "step": 4734
    },
    {
      "epoch": 2.542263682426764,
      "grad_norm": 0.671875,
      "learning_rate": 2.436668214217031e-05,
      "loss": 0.9913,
      "step": 4735
    },
    {
      "epoch": 2.5428005771618403,
      "grad_norm": 0.66015625,
      "learning_rate": 2.43582397471783e-05,
      "loss": 0.8997,
      "step": 4736
    },
    {
      "epoch": 2.5433374718969164,
      "grad_norm": 0.75390625,
      "learning_rate": 2.434979742541927e-05,
      "loss": 0.9264,
      "step": 4737
    },
    {
      "epoch": 2.543874366631992,
      "grad_norm": 0.6796875,
      "learning_rate": 2.4341355177856587e-05,
      "loss": 0.9566,
      "step": 4738
    },
    {
      "epoch": 2.544411261367068,
      "grad_norm": 0.78125,
      "learning_rate": 2.4332913005453612e-05,
      "loss": 1.0158,
      "step": 4739
    },
    {
      "epoch": 2.5449481561021443,
      "grad_norm": 0.67578125,
      "learning_rate": 2.432447090917373e-05,
      "loss": 0.8305,
      "step": 4740
    },
    {
      "epoch": 2.54548505083722,
      "grad_norm": 0.66796875,
      "learning_rate": 2.4316028889980254e-05,
      "loss": 1.0086,
      "step": 4741
    },
    {
      "epoch": 2.546021945572296,
      "grad_norm": 0.71484375,
      "learning_rate": 2.4307586948836552e-05,
      "loss": 0.8123,
      "step": 4742
    },
    {
      "epoch": 2.5465588403073722,
      "grad_norm": 0.8203125,
      "learning_rate": 2.4299145086705947e-05,
      "loss": 0.9146,
      "step": 4743
    },
    {
      "epoch": 2.5470957350424484,
      "grad_norm": 0.7421875,
      "learning_rate": 2.4290703304551756e-05,
      "loss": 0.8819,
      "step": 4744
    },
    {
      "epoch": 2.5476326297775245,
      "grad_norm": 0.77734375,
      "learning_rate": 2.4282261603337306e-05,
      "loss": 0.9958,
      "step": 4745
    },
    {
      "epoch": 2.5481695245126,
      "grad_norm": 0.68359375,
      "learning_rate": 2.4273819984025896e-05,
      "loss": 0.8316,
      "step": 4746
    },
    {
      "epoch": 2.5487064192476763,
      "grad_norm": 0.73828125,
      "learning_rate": 2.4265378447580818e-05,
      "loss": 0.9154,
      "step": 4747
    },
    {
      "epoch": 2.5492433139827524,
      "grad_norm": 0.80078125,
      "learning_rate": 2.4256936994965368e-05,
      "loss": 0.9037,
      "step": 4748
    },
    {
      "epoch": 2.549780208717828,
      "grad_norm": 0.79296875,
      "learning_rate": 2.424849562714282e-05,
      "loss": 0.8984,
      "step": 4749
    },
    {
      "epoch": 2.550317103452904,
      "grad_norm": 0.85546875,
      "learning_rate": 2.4240054345076438e-05,
      "loss": 0.79,
      "step": 4750
    },
    {
      "epoch": 2.550317103452904,
      "eval_loss": 0.9044550657272339,
      "eval_runtime": 46.4744,
      "eval_samples_per_second": 20.7,
      "eval_steps_per_second": 20.7,
      "step": 4750
    },
    {
      "epoch": 2.5508539981879803,
      "grad_norm": 0.64453125,
      "learning_rate": 2.4231613149729478e-05,
      "loss": 0.7006,
      "step": 4751
    },
    {
      "epoch": 2.5513908929230564,
      "grad_norm": 0.78125,
      "learning_rate": 2.422317204206518e-05,
      "loss": 0.8896,
      "step": 4752
    },
    {
      "epoch": 2.5519277876581326,
      "grad_norm": 0.79296875,
      "learning_rate": 2.4214731023046793e-05,
      "loss": 0.8646,
      "step": 4753
    },
    {
      "epoch": 2.5524646823932082,
      "grad_norm": 0.64453125,
      "learning_rate": 2.4206290093637546e-05,
      "loss": 0.8121,
      "step": 4754
    },
    {
      "epoch": 2.5530015771282843,
      "grad_norm": 0.78515625,
      "learning_rate": 2.419784925480064e-05,
      "loss": 0.7799,
      "step": 4755
    },
    {
      "epoch": 2.5535384718633605,
      "grad_norm": 0.6640625,
      "learning_rate": 2.4189408507499304e-05,
      "loss": 1.0234,
      "step": 4756
    },
    {
      "epoch": 2.554075366598436,
      "grad_norm": 0.6171875,
      "learning_rate": 2.4180967852696725e-05,
      "loss": 0.7871,
      "step": 4757
    },
    {
      "epoch": 2.5546122613335123,
      "grad_norm": 0.76953125,
      "learning_rate": 2.417252729135608e-05,
      "loss": 0.768,
      "step": 4758
    },
    {
      "epoch": 2.5551491560685884,
      "grad_norm": 0.64453125,
      "learning_rate": 2.4164086824440552e-05,
      "loss": 0.8422,
      "step": 4759
    },
    {
      "epoch": 2.555686050803664,
      "grad_norm": 0.66796875,
      "learning_rate": 2.4155646452913296e-05,
      "loss": 0.9319,
      "step": 4760
    },
    {
      "epoch": 2.55622294553874,
      "grad_norm": 0.74609375,
      "learning_rate": 2.414720617773748e-05,
      "loss": 0.8486,
      "step": 4761
    },
    {
      "epoch": 2.5567598402738163,
      "grad_norm": 0.53125,
      "learning_rate": 2.413876599987624e-05,
      "loss": 0.598,
      "step": 4762
    },
    {
      "epoch": 2.5572967350088924,
      "grad_norm": 0.62109375,
      "learning_rate": 2.41303259202927e-05,
      "loss": 0.67,
      "step": 4763
    },
    {
      "epoch": 2.5578336297439686,
      "grad_norm": 0.6875,
      "learning_rate": 2.4121885939949987e-05,
      "loss": 0.8766,
      "step": 4764
    },
    {
      "epoch": 2.5583705244790442,
      "grad_norm": 0.671875,
      "learning_rate": 2.4113446059811217e-05,
      "loss": 0.7272,
      "step": 4765
    },
    {
      "epoch": 2.5589074192141203,
      "grad_norm": 0.609375,
      "learning_rate": 2.4105006280839473e-05,
      "loss": 0.7251,
      "step": 4766
    },
    {
      "epoch": 2.5594443139491965,
      "grad_norm": 0.78125,
      "learning_rate": 2.409656660399785e-05,
      "loss": 0.7694,
      "step": 4767
    },
    {
      "epoch": 2.559981208684272,
      "grad_norm": 0.73828125,
      "learning_rate": 2.4088127030249415e-05,
      "loss": 0.8735,
      "step": 4768
    },
    {
      "epoch": 2.5605181034193483,
      "grad_norm": 0.63671875,
      "learning_rate": 2.4079687560557226e-05,
      "loss": 0.774,
      "step": 4769
    },
    {
      "epoch": 2.5610549981544244,
      "grad_norm": 0.7421875,
      "learning_rate": 2.407124819588435e-05,
      "loss": 1.0585,
      "step": 4770
    },
    {
      "epoch": 2.5615918928895005,
      "grad_norm": 0.6171875,
      "learning_rate": 2.406280893719382e-05,
      "loss": 0.7963,
      "step": 4771
    },
    {
      "epoch": 2.5621287876245766,
      "grad_norm": 0.65625,
      "learning_rate": 2.4054369785448648e-05,
      "loss": 0.6606,
      "step": 4772
    },
    {
      "epoch": 2.5626656823596523,
      "grad_norm": 0.66796875,
      "learning_rate": 2.4045930741611862e-05,
      "loss": 0.8388,
      "step": 4773
    },
    {
      "epoch": 2.5632025770947284,
      "grad_norm": 0.53125,
      "learning_rate": 2.4037491806646474e-05,
      "loss": 0.7488,
      "step": 4774
    },
    {
      "epoch": 2.5637394718298046,
      "grad_norm": 0.69921875,
      "learning_rate": 2.402905298151545e-05,
      "loss": 0.7551,
      "step": 4775
    },
    {
      "epoch": 2.5642763665648802,
      "grad_norm": 0.64453125,
      "learning_rate": 2.4020614267181778e-05,
      "loss": 0.7906,
      "step": 4776
    },
    {
      "epoch": 2.5648132612999563,
      "grad_norm": 0.5625,
      "learning_rate": 2.4012175664608412e-05,
      "loss": 0.816,
      "step": 4777
    },
    {
      "epoch": 2.5653501560350325,
      "grad_norm": 0.796875,
      "learning_rate": 2.4003737174758323e-05,
      "loss": 0.9098,
      "step": 4778
    },
    {
      "epoch": 2.565887050770108,
      "grad_norm": 0.63671875,
      "learning_rate": 2.3995298798594438e-05,
      "loss": 0.9021,
      "step": 4779
    },
    {
      "epoch": 2.5664239455051843,
      "grad_norm": 0.984375,
      "learning_rate": 2.3986860537079675e-05,
      "loss": 0.8583,
      "step": 4780
    },
    {
      "epoch": 2.5669608402402604,
      "grad_norm": 0.74609375,
      "learning_rate": 2.3978422391176963e-05,
      "loss": 0.8389,
      "step": 4781
    },
    {
      "epoch": 2.5674977349753365,
      "grad_norm": 0.6875,
      "learning_rate": 2.3969984361849198e-05,
      "loss": 0.7442,
      "step": 4782
    },
    {
      "epoch": 2.5680346297104126,
      "grad_norm": 0.984375,
      "learning_rate": 2.3961546450059257e-05,
      "loss": 0.8782,
      "step": 4783
    },
    {
      "epoch": 2.5685715244454883,
      "grad_norm": 0.72265625,
      "learning_rate": 2.3953108656770016e-05,
      "loss": 0.7903,
      "step": 4784
    },
    {
      "epoch": 2.5691084191805644,
      "grad_norm": 0.80859375,
      "learning_rate": 2.394467098294433e-05,
      "loss": 0.9673,
      "step": 4785
    },
    {
      "epoch": 2.5696453139156405,
      "grad_norm": 0.859375,
      "learning_rate": 2.3936233429545054e-05,
      "loss": 0.9672,
      "step": 4786
    },
    {
      "epoch": 2.5701822086507162,
      "grad_norm": 0.80859375,
      "learning_rate": 2.392779599753502e-05,
      "loss": 0.9942,
      "step": 4787
    },
    {
      "epoch": 2.5707191033857923,
      "grad_norm": 0.79296875,
      "learning_rate": 2.391935868787703e-05,
      "loss": 0.8621,
      "step": 4788
    },
    {
      "epoch": 2.5712559981208685,
      "grad_norm": 0.58203125,
      "learning_rate": 2.3910921501533906e-05,
      "loss": 0.8573,
      "step": 4789
    },
    {
      "epoch": 2.5717928928559446,
      "grad_norm": 0.65234375,
      "learning_rate": 2.390248443946843e-05,
      "loss": 0.8559,
      "step": 4790
    },
    {
      "epoch": 2.5723297875910207,
      "grad_norm": 0.65625,
      "learning_rate": 2.3894047502643385e-05,
      "loss": 0.7038,
      "step": 4791
    },
    {
      "epoch": 2.5728666823260964,
      "grad_norm": 0.65234375,
      "learning_rate": 2.388561069202152e-05,
      "loss": 0.757,
      "step": 4792
    },
    {
      "epoch": 2.5734035770611725,
      "grad_norm": 0.6015625,
      "learning_rate": 2.387717400856558e-05,
      "loss": 0.6814,
      "step": 4793
    },
    {
      "epoch": 2.5739404717962486,
      "grad_norm": 0.66796875,
      "learning_rate": 2.3868737453238316e-05,
      "loss": 0.7257,
      "step": 4794
    },
    {
      "epoch": 2.5744773665313243,
      "grad_norm": 0.69140625,
      "learning_rate": 2.386030102700243e-05,
      "loss": 0.8394,
      "step": 4795
    },
    {
      "epoch": 2.5750142612664004,
      "grad_norm": 0.69921875,
      "learning_rate": 2.385186473082063e-05,
      "loss": 0.7631,
      "step": 4796
    },
    {
      "epoch": 2.5755511560014765,
      "grad_norm": 0.7109375,
      "learning_rate": 2.38434285656556e-05,
      "loss": 0.984,
      "step": 4797
    },
    {
      "epoch": 2.5760880507365522,
      "grad_norm": 0.66796875,
      "learning_rate": 2.3834992532470023e-05,
      "loss": 0.737,
      "step": 4798
    },
    {
      "epoch": 2.5766249454716283,
      "grad_norm": 0.66015625,
      "learning_rate": 2.382655663222656e-05,
      "loss": 0.7902,
      "step": 4799
    },
    {
      "epoch": 2.5771618402067045,
      "grad_norm": 0.6953125,
      "learning_rate": 2.3818120865887836e-05,
      "loss": 0.7661,
      "step": 4800
    },
    {
      "epoch": 2.5771618402067045,
      "eval_loss": 0.9042898416519165,
      "eval_runtime": 45.9124,
      "eval_samples_per_second": 20.953,
      "eval_steps_per_second": 20.953,
      "step": 4800
    },
    {
      "epoch": 2.5776987349417806,
      "grad_norm": 0.85546875,
      "learning_rate": 2.3809685234416486e-05,
      "loss": 0.8461,
      "step": 4801
    },
    {
      "epoch": 2.5782356296768567,
      "grad_norm": 0.71484375,
      "learning_rate": 2.380124973877513e-05,
      "loss": 0.881,
      "step": 4802
    },
    {
      "epoch": 2.5787725244119324,
      "grad_norm": 0.59765625,
      "learning_rate": 2.3792814379926358e-05,
      "loss": 0.7544,
      "step": 4803
    },
    {
      "epoch": 2.5793094191470085,
      "grad_norm": 0.7890625,
      "learning_rate": 2.3784379158832747e-05,
      "loss": 0.8336,
      "step": 4804
    },
    {
      "epoch": 2.5798463138820846,
      "grad_norm": 0.703125,
      "learning_rate": 2.3775944076456874e-05,
      "loss": 0.8285,
      "step": 4805
    },
    {
      "epoch": 2.5803832086171603,
      "grad_norm": 0.67578125,
      "learning_rate": 2.376750913376129e-05,
      "loss": 0.7354,
      "step": 4806
    },
    {
      "epoch": 2.5809201033522364,
      "grad_norm": 0.74609375,
      "learning_rate": 2.3759074331708517e-05,
      "loss": 0.8723,
      "step": 4807
    },
    {
      "epoch": 2.5814569980873125,
      "grad_norm": 0.66015625,
      "learning_rate": 2.375063967126108e-05,
      "loss": 0.7881,
      "step": 4808
    },
    {
      "epoch": 2.5819938928223887,
      "grad_norm": 0.79296875,
      "learning_rate": 2.374220515338147e-05,
      "loss": 0.8679,
      "step": 4809
    },
    {
      "epoch": 2.582530787557465,
      "grad_norm": 0.6796875,
      "learning_rate": 2.3733770779032184e-05,
      "loss": 0.8832,
      "step": 4810
    },
    {
      "epoch": 2.5830676822925405,
      "grad_norm": 0.63671875,
      "learning_rate": 2.372533654917569e-05,
      "loss": 1.0323,
      "step": 4811
    },
    {
      "epoch": 2.5836045770276166,
      "grad_norm": 0.6171875,
      "learning_rate": 2.371690246477444e-05,
      "loss": 0.8778,
      "step": 4812
    },
    {
      "epoch": 2.5841414717626927,
      "grad_norm": 0.62890625,
      "learning_rate": 2.3708468526790858e-05,
      "loss": 0.8339,
      "step": 4813
    },
    {
      "epoch": 2.5846783664977684,
      "grad_norm": 0.59375,
      "learning_rate": 2.3700034736187383e-05,
      "loss": 0.9372,
      "step": 4814
    },
    {
      "epoch": 2.5852152612328445,
      "grad_norm": 0.63671875,
      "learning_rate": 2.3691601093926404e-05,
      "loss": 0.704,
      "step": 4815
    },
    {
      "epoch": 2.5857521559679206,
      "grad_norm": 0.72265625,
      "learning_rate": 2.368316760097032e-05,
      "loss": 0.9305,
      "step": 4816
    },
    {
      "epoch": 2.5862890507029963,
      "grad_norm": 0.67578125,
      "learning_rate": 2.367473425828148e-05,
      "loss": 0.9001,
      "step": 4817
    },
    {
      "epoch": 2.5868259454380724,
      "grad_norm": 0.72265625,
      "learning_rate": 2.3666301066822236e-05,
      "loss": 0.9246,
      "step": 4818
    },
    {
      "epoch": 2.5873628401731485,
      "grad_norm": 0.7109375,
      "learning_rate": 2.365786802755494e-05,
      "loss": 0.9648,
      "step": 4819
    },
    {
      "epoch": 2.5878997349082247,
      "grad_norm": 0.71875,
      "learning_rate": 2.36494351414419e-05,
      "loss": 0.9637,
      "step": 4820
    },
    {
      "epoch": 2.588436629643301,
      "grad_norm": 0.7265625,
      "learning_rate": 2.36410024094454e-05,
      "loss": 0.8621,
      "step": 4821
    },
    {
      "epoch": 2.5889735243783765,
      "grad_norm": 0.87109375,
      "learning_rate": 2.363256983252775e-05,
      "loss": 0.7851,
      "step": 4822
    },
    {
      "epoch": 2.5895104191134526,
      "grad_norm": 0.6328125,
      "learning_rate": 2.3624137411651196e-05,
      "loss": 0.7886,
      "step": 4823
    },
    {
      "epoch": 2.5900473138485287,
      "grad_norm": 0.6953125,
      "learning_rate": 2.361570514777799e-05,
      "loss": 0.804,
      "step": 4824
    },
    {
      "epoch": 2.5905842085836044,
      "grad_norm": 0.6796875,
      "learning_rate": 2.3607273041870356e-05,
      "loss": 0.8343,
      "step": 4825
    },
    {
      "epoch": 2.5911211033186805,
      "grad_norm": 0.68359375,
      "learning_rate": 2.3598841094890496e-05,
      "loss": 0.8373,
      "step": 4826
    },
    {
      "epoch": 2.5916579980537566,
      "grad_norm": 0.75390625,
      "learning_rate": 2.359040930780062e-05,
      "loss": 0.7907,
      "step": 4827
    },
    {
      "epoch": 2.5921948927888327,
      "grad_norm": 0.60546875,
      "learning_rate": 2.3581977681562895e-05,
      "loss": 0.898,
      "step": 4828
    },
    {
      "epoch": 2.592731787523909,
      "grad_norm": 0.671875,
      "learning_rate": 2.3573546217139464e-05,
      "loss": 0.7337,
      "step": 4829
    },
    {
      "epoch": 2.5932686822589845,
      "grad_norm": 0.7109375,
      "learning_rate": 2.3565114915492482e-05,
      "loss": 0.9152,
      "step": 4830
    },
    {
      "epoch": 2.5938055769940607,
      "grad_norm": 1.3515625,
      "learning_rate": 2.355668377758406e-05,
      "loss": 0.8682,
      "step": 4831
    },
    {
      "epoch": 2.594342471729137,
      "grad_norm": 0.60546875,
      "learning_rate": 2.3548252804376288e-05,
      "loss": 0.6563,
      "step": 4832
    },
    {
      "epoch": 2.5948793664642125,
      "grad_norm": 0.74609375,
      "learning_rate": 2.3539821996831278e-05,
      "loss": 0.8853,
      "step": 4833
    },
    {
      "epoch": 2.5954162611992886,
      "grad_norm": 0.796875,
      "learning_rate": 2.353139135591105e-05,
      "loss": 0.9459,
      "step": 4834
    },
    {
      "epoch": 2.5959531559343647,
      "grad_norm": 0.66796875,
      "learning_rate": 2.352296088257767e-05,
      "loss": 0.9339,
      "step": 4835
    },
    {
      "epoch": 2.5964900506694404,
      "grad_norm": 0.6640625,
      "learning_rate": 2.3514530577793163e-05,
      "loss": 0.8762,
      "step": 4836
    },
    {
      "epoch": 2.5970269454045165,
      "grad_norm": 0.828125,
      "learning_rate": 2.350610044251952e-05,
      "loss": 0.9001,
      "step": 4837
    },
    {
      "epoch": 2.5975638401395926,
      "grad_norm": 0.84765625,
      "learning_rate": 2.3497670477718744e-05,
      "loss": 0.8743,
      "step": 4838
    },
    {
      "epoch": 2.5981007348746687,
      "grad_norm": 0.6328125,
      "learning_rate": 2.348924068435279e-05,
      "loss": 0.8777,
      "step": 4839
    },
    {
      "epoch": 2.598637629609745,
      "grad_norm": 0.828125,
      "learning_rate": 2.34808110633836e-05,
      "loss": 0.7591,
      "step": 4840
    },
    {
      "epoch": 2.5991745243448205,
      "grad_norm": 0.6171875,
      "learning_rate": 2.3472381615773126e-05,
      "loss": 0.7358,
      "step": 4841
    },
    {
      "epoch": 2.5997114190798967,
      "grad_norm": 0.79296875,
      "learning_rate": 2.3463952342483237e-05,
      "loss": 0.9206,
      "step": 4842
    },
    {
      "epoch": 2.600248313814973,
      "grad_norm": 0.7578125,
      "learning_rate": 2.3455523244475846e-05,
      "loss": 0.9042,
      "step": 4843
    },
    {
      "epoch": 2.6007852085500485,
      "grad_norm": 0.73046875,
      "learning_rate": 2.3447094322712805e-05,
      "loss": 1.0895,
      "step": 4844
    },
    {
      "epoch": 2.6013221032851246,
      "grad_norm": 0.79296875,
      "learning_rate": 2.343866557815597e-05,
      "loss": 0.9379,
      "step": 4845
    },
    {
      "epoch": 2.6018589980202007,
      "grad_norm": 0.63671875,
      "learning_rate": 2.3430237011767167e-05,
      "loss": 0.8434,
      "step": 4846
    },
    {
      "epoch": 2.602395892755277,
      "grad_norm": 0.6640625,
      "learning_rate": 2.3421808624508202e-05,
      "loss": 0.7852,
      "step": 4847
    },
    {
      "epoch": 2.6029327874903525,
      "grad_norm": 0.6875,
      "learning_rate": 2.341338041734086e-05,
      "loss": 0.8722,
      "step": 4848
    },
    {
      "epoch": 2.6034696822254286,
      "grad_norm": 0.62890625,
      "learning_rate": 2.3404952391226906e-05,
      "loss": 0.973,
      "step": 4849
    },
    {
      "epoch": 2.6040065769605047,
      "grad_norm": 0.6328125,
      "learning_rate": 2.3396524547128072e-05,
      "loss": 0.9067,
      "step": 4850
    },
    {
      "epoch": 2.6040065769605047,
      "eval_loss": 0.9035809636116028,
      "eval_runtime": 46.2018,
      "eval_samples_per_second": 20.822,
      "eval_steps_per_second": 20.822,
      "step": 4850
    },
    {
      "epoch": 2.604543471695581,
      "grad_norm": 0.66015625,
      "learning_rate": 2.3388096886006105e-05,
      "loss": 0.833,
      "step": 4851
    },
    {
      "epoch": 2.6050803664306565,
      "grad_norm": 0.6640625,
      "learning_rate": 2.3379669408822696e-05,
      "loss": 0.8819,
      "step": 4852
    },
    {
      "epoch": 2.6056172611657327,
      "grad_norm": 0.68359375,
      "learning_rate": 2.3371242116539525e-05,
      "loss": 0.911,
      "step": 4853
    },
    {
      "epoch": 2.606154155900809,
      "grad_norm": 0.671875,
      "learning_rate": 2.336281501011825e-05,
      "loss": 1.0088,
      "step": 4854
    },
    {
      "epoch": 2.6066910506358845,
      "grad_norm": 0.625,
      "learning_rate": 2.3354388090520522e-05,
      "loss": 0.7627,
      "step": 4855
    },
    {
      "epoch": 2.6072279453709606,
      "grad_norm": 0.60546875,
      "learning_rate": 2.3345961358707953e-05,
      "loss": 0.741,
      "step": 4856
    },
    {
      "epoch": 2.6077648401060367,
      "grad_norm": 0.63671875,
      "learning_rate": 2.333753481564213e-05,
      "loss": 0.7867,
      "step": 4857
    },
    {
      "epoch": 2.608301734841113,
      "grad_norm": 0.6328125,
      "learning_rate": 2.332910846228466e-05,
      "loss": 0.8172,
      "step": 4858
    },
    {
      "epoch": 2.608838629576189,
      "grad_norm": 0.95703125,
      "learning_rate": 2.3320682299597054e-05,
      "loss": 1.0437,
      "step": 4859
    },
    {
      "epoch": 2.6093755243112646,
      "grad_norm": 0.9765625,
      "learning_rate": 2.331225632854087e-05,
      "loss": 1.0193,
      "step": 4860
    },
    {
      "epoch": 2.6099124190463407,
      "grad_norm": 0.70703125,
      "learning_rate": 2.3303830550077617e-05,
      "loss": 0.7396,
      "step": 4861
    },
    {
      "epoch": 2.610449313781417,
      "grad_norm": 0.62109375,
      "learning_rate": 2.329540496516877e-05,
      "loss": 0.8171,
      "step": 4862
    },
    {
      "epoch": 2.6109862085164925,
      "grad_norm": 0.71875,
      "learning_rate": 2.3286979574775807e-05,
      "loss": 0.7973,
      "step": 4863
    },
    {
      "epoch": 2.6115231032515687,
      "grad_norm": 0.62890625,
      "learning_rate": 2.327855437986017e-05,
      "loss": 0.8042,
      "step": 4864
    },
    {
      "epoch": 2.612059997986645,
      "grad_norm": 0.6171875,
      "learning_rate": 2.3270129381383275e-05,
      "loss": 0.7434,
      "step": 4865
    },
    {
      "epoch": 2.612596892721721,
      "grad_norm": 0.8515625,
      "learning_rate": 2.326170458030654e-05,
      "loss": 0.9505,
      "step": 4866
    },
    {
      "epoch": 2.6131337874567966,
      "grad_norm": 0.828125,
      "learning_rate": 2.3253279977591308e-05,
      "loss": 0.8996,
      "step": 4867
    },
    {
      "epoch": 2.6136706821918727,
      "grad_norm": 0.6171875,
      "learning_rate": 2.3244855574198958e-05,
      "loss": 0.679,
      "step": 4868
    },
    {
      "epoch": 2.614207576926949,
      "grad_norm": 0.62109375,
      "learning_rate": 2.3236431371090814e-05,
      "loss": 0.8053,
      "step": 4869
    },
    {
      "epoch": 2.614744471662025,
      "grad_norm": 0.74609375,
      "learning_rate": 2.3228007369228178e-05,
      "loss": 0.9276,
      "step": 4870
    },
    {
      "epoch": 2.6152813663971006,
      "grad_norm": 0.6484375,
      "learning_rate": 2.3219583569572347e-05,
      "loss": 0.7546,
      "step": 4871
    },
    {
      "epoch": 2.6158182611321767,
      "grad_norm": 0.6484375,
      "learning_rate": 2.321115997308458e-05,
      "loss": 0.8459,
      "step": 4872
    },
    {
      "epoch": 2.616355155867253,
      "grad_norm": 0.6953125,
      "learning_rate": 2.3202736580726106e-05,
      "loss": 0.8358,
      "step": 4873
    },
    {
      "epoch": 2.6168920506023285,
      "grad_norm": 0.6484375,
      "learning_rate": 2.3194313393458156e-05,
      "loss": 0.8025,
      "step": 4874
    },
    {
      "epoch": 2.6174289453374047,
      "grad_norm": 0.55078125,
      "learning_rate": 2.3185890412241917e-05,
      "loss": 0.7274,
      "step": 4875
    },
    {
      "epoch": 2.617965840072481,
      "grad_norm": 0.6875,
      "learning_rate": 2.3177467638038555e-05,
      "loss": 0.8235,
      "step": 4876
    },
    {
      "epoch": 2.618502734807557,
      "grad_norm": 0.65234375,
      "learning_rate": 2.3169045071809215e-05,
      "loss": 0.9587,
      "step": 4877
    },
    {
      "epoch": 2.619039629542633,
      "grad_norm": 0.7734375,
      "learning_rate": 2.3160622714515012e-05,
      "loss": 0.9466,
      "step": 4878
    },
    {
      "epoch": 2.6195765242777087,
      "grad_norm": 0.7734375,
      "learning_rate": 2.3152200567117057e-05,
      "loss": 0.8643,
      "step": 4879
    },
    {
      "epoch": 2.620113419012785,
      "grad_norm": 1.0625,
      "learning_rate": 2.314377863057642e-05,
      "loss": 0.8484,
      "step": 4880
    },
    {
      "epoch": 2.620650313747861,
      "grad_norm": 0.74609375,
      "learning_rate": 2.313535690585414e-05,
      "loss": 0.7703,
      "step": 4881
    },
    {
      "epoch": 2.6211872084829366,
      "grad_norm": 0.703125,
      "learning_rate": 2.3126935393911255e-05,
      "loss": 0.8477,
      "step": 4882
    },
    {
      "epoch": 2.6217241032180127,
      "grad_norm": 0.75390625,
      "learning_rate": 2.311851409570877e-05,
      "loss": 1.0015,
      "step": 4883
    },
    {
      "epoch": 2.622260997953089,
      "grad_norm": 0.55078125,
      "learning_rate": 2.3110093012207647e-05,
      "loss": 0.6186,
      "step": 4884
    },
    {
      "epoch": 2.622797892688165,
      "grad_norm": 0.671875,
      "learning_rate": 2.310167214436885e-05,
      "loss": 0.8147,
      "step": 4885
    },
    {
      "epoch": 2.6233347874232407,
      "grad_norm": 0.6953125,
      "learning_rate": 2.309325149315329e-05,
      "loss": 0.7688,
      "step": 4886
    },
    {
      "epoch": 2.623871682158317,
      "grad_norm": 0.68359375,
      "learning_rate": 2.308483105952189e-05,
      "loss": 0.8862,
      "step": 4887
    },
    {
      "epoch": 2.624408576893393,
      "grad_norm": 0.61328125,
      "learning_rate": 2.307641084443552e-05,
      "loss": 0.8191,
      "step": 4888
    },
    {
      "epoch": 2.624945471628469,
      "grad_norm": 0.67578125,
      "learning_rate": 2.3067990848855026e-05,
      "loss": 0.6825,
      "step": 4889
    },
    {
      "epoch": 2.6254823663635447,
      "grad_norm": 0.6796875,
      "learning_rate": 2.3059571073741248e-05,
      "loss": 0.7945,
      "step": 4890
    },
    {
      "epoch": 2.626019261098621,
      "grad_norm": 0.59765625,
      "learning_rate": 2.3051151520054994e-05,
      "loss": 0.6775,
      "step": 4891
    },
    {
      "epoch": 2.626556155833697,
      "grad_norm": 0.734375,
      "learning_rate": 2.3042732188757023e-05,
      "loss": 1.176,
      "step": 4892
    },
    {
      "epoch": 2.6270930505687726,
      "grad_norm": 0.640625,
      "learning_rate": 2.3034313080808095e-05,
      "loss": 0.8802,
      "step": 4893
    },
    {
      "epoch": 2.6276299453038487,
      "grad_norm": 0.66015625,
      "learning_rate": 2.302589419716894e-05,
      "loss": 0.7208,
      "step": 4894
    },
    {
      "epoch": 2.628166840038925,
      "grad_norm": 0.65234375,
      "learning_rate": 2.301747553880025e-05,
      "loss": 0.7477,
      "step": 4895
    },
    {
      "epoch": 2.628703734774001,
      "grad_norm": 0.8203125,
      "learning_rate": 2.3009057106662714e-05,
      "loss": 1.0084,
      "step": 4896
    },
    {
      "epoch": 2.629240629509077,
      "grad_norm": 0.72265625,
      "learning_rate": 2.3000638901716976e-05,
      "loss": 0.87,
      "step": 4897
    },
    {
      "epoch": 2.629777524244153,
      "grad_norm": 0.6640625,
      "learning_rate": 2.2992220924923652e-05,
      "loss": 0.7963,
      "step": 4898
    },
    {
      "epoch": 2.630314418979229,
      "grad_norm": 0.83203125,
      "learning_rate": 2.2983803177243352e-05,
      "loss": 0.9892,
      "step": 4899
    },
    {
      "epoch": 2.630851313714305,
      "grad_norm": 0.7578125,
      "learning_rate": 2.297538565963665e-05,
      "loss": 0.7251,
      "step": 4900
    },
    {
      "epoch": 2.630851313714305,
      "eval_loss": 0.9034591317176819,
      "eval_runtime": 46.1079,
      "eval_samples_per_second": 20.864,
      "eval_steps_per_second": 20.864,
      "step": 4900
    },
    {
      "epoch": 2.6313882084493807,
      "grad_norm": 0.75,
      "learning_rate": 2.296696837306408e-05,
      "loss": 0.9207,
      "step": 4901
    },
    {
      "epoch": 2.631925103184457,
      "grad_norm": 0.69140625,
      "learning_rate": 2.295855131848616e-05,
      "loss": 0.86,
      "step": 4902
    },
    {
      "epoch": 2.632461997919533,
      "grad_norm": 0.84765625,
      "learning_rate": 2.2950134496863386e-05,
      "loss": 1.002,
      "step": 4903
    },
    {
      "epoch": 2.632998892654609,
      "grad_norm": 0.8125,
      "learning_rate": 2.2941717909156228e-05,
      "loss": 0.8904,
      "step": 4904
    },
    {
      "epoch": 2.6335357873896847,
      "grad_norm": 0.66015625,
      "learning_rate": 2.2933301556325123e-05,
      "loss": 0.8564,
      "step": 4905
    },
    {
      "epoch": 2.634072682124761,
      "grad_norm": 0.73046875,
      "learning_rate": 2.2924885439330477e-05,
      "loss": 0.8905,
      "step": 4906
    },
    {
      "epoch": 2.634609576859837,
      "grad_norm": 0.68359375,
      "learning_rate": 2.291646955913269e-05,
      "loss": 0.7895,
      "step": 4907
    },
    {
      "epoch": 2.635146471594913,
      "grad_norm": 0.68359375,
      "learning_rate": 2.2908053916692117e-05,
      "loss": 0.8444,
      "step": 4908
    },
    {
      "epoch": 2.635683366329989,
      "grad_norm": 0.70703125,
      "learning_rate": 2.289963851296908e-05,
      "loss": 0.828,
      "step": 4909
    },
    {
      "epoch": 2.636220261065065,
      "grad_norm": 0.81640625,
      "learning_rate": 2.2891223348923884e-05,
      "loss": 1.0982,
      "step": 4910
    },
    {
      "epoch": 2.636757155800141,
      "grad_norm": 0.73046875,
      "learning_rate": 2.2882808425516805e-05,
      "loss": 0.7755,
      "step": 4911
    },
    {
      "epoch": 2.6372940505352167,
      "grad_norm": 1.609375,
      "learning_rate": 2.2874393743708104e-05,
      "loss": 0.9889,
      "step": 4912
    },
    {
      "epoch": 2.637830945270293,
      "grad_norm": 0.68359375,
      "learning_rate": 2.2865979304457997e-05,
      "loss": 0.7403,
      "step": 4913
    },
    {
      "epoch": 2.638367840005369,
      "grad_norm": 0.69921875,
      "learning_rate": 2.2857565108726667e-05,
      "loss": 0.8106,
      "step": 4914
    },
    {
      "epoch": 2.638904734740445,
      "grad_norm": 0.6015625,
      "learning_rate": 2.2849151157474298e-05,
      "loss": 0.7078,
      "step": 4915
    },
    {
      "epoch": 2.639441629475521,
      "grad_norm": 0.6015625,
      "learning_rate": 2.2840737451661018e-05,
      "loss": 0.7153,
      "step": 4916
    },
    {
      "epoch": 2.639978524210597,
      "grad_norm": 0.80859375,
      "learning_rate": 2.2832323992246945e-05,
      "loss": 0.8945,
      "step": 4917
    },
    {
      "epoch": 2.640515418945673,
      "grad_norm": 0.7109375,
      "learning_rate": 2.2823910780192152e-05,
      "loss": 0.8071,
      "step": 4918
    },
    {
      "epoch": 2.641052313680749,
      "grad_norm": 0.58203125,
      "learning_rate": 2.281549781645669e-05,
      "loss": 0.7665,
      "step": 4919
    },
    {
      "epoch": 2.641589208415825,
      "grad_norm": 0.73046875,
      "learning_rate": 2.2807085102000598e-05,
      "loss": 0.8724,
      "step": 4920
    },
    {
      "epoch": 2.642126103150901,
      "grad_norm": 0.890625,
      "learning_rate": 2.2798672637783865e-05,
      "loss": 0.9079,
      "step": 4921
    },
    {
      "epoch": 2.642662997885977,
      "grad_norm": 0.8203125,
      "learning_rate": 2.279026042476646e-05,
      "loss": 0.9932,
      "step": 4922
    },
    {
      "epoch": 2.643199892621053,
      "grad_norm": 0.7109375,
      "learning_rate": 2.2781848463908328e-05,
      "loss": 0.9764,
      "step": 4923
    },
    {
      "epoch": 2.643736787356129,
      "grad_norm": 0.74609375,
      "learning_rate": 2.2773436756169378e-05,
      "loss": 0.917,
      "step": 4924
    },
    {
      "epoch": 2.644273682091205,
      "grad_norm": 0.71875,
      "learning_rate": 2.2765025302509497e-05,
      "loss": 1.0003,
      "step": 4925
    },
    {
      "epoch": 2.644810576826281,
      "grad_norm": 0.640625,
      "learning_rate": 2.2756614103888523e-05,
      "loss": 0.8843,
      "step": 4926
    },
    {
      "epoch": 2.645347471561357,
      "grad_norm": 0.59765625,
      "learning_rate": 2.2748203161266286e-05,
      "loss": 0.7537,
      "step": 4927
    },
    {
      "epoch": 2.645884366296433,
      "grad_norm": 0.78125,
      "learning_rate": 2.2739792475602596e-05,
      "loss": 0.874,
      "step": 4928
    },
    {
      "epoch": 2.646421261031509,
      "grad_norm": 0.6171875,
      "learning_rate": 2.2731382047857204e-05,
      "loss": 0.8862,
      "step": 4929
    },
    {
      "epoch": 2.646958155766585,
      "grad_norm": 0.78515625,
      "learning_rate": 2.272297187898984e-05,
      "loss": 0.8466,
      "step": 4930
    },
    {
      "epoch": 2.647495050501661,
      "grad_norm": 0.9609375,
      "learning_rate": 2.271456196996023e-05,
      "loss": 0.8819,
      "step": 4931
    },
    {
      "epoch": 2.648031945236737,
      "grad_norm": 0.671875,
      "learning_rate": 2.2706152321728045e-05,
      "loss": 0.7642,
      "step": 4932
    },
    {
      "epoch": 2.648568839971813,
      "grad_norm": 0.65625,
      "learning_rate": 2.269774293525293e-05,
      "loss": 0.8886,
      "step": 4933
    },
    {
      "epoch": 2.649105734706889,
      "grad_norm": 0.703125,
      "learning_rate": 2.26893338114945e-05,
      "loss": 0.9149,
      "step": 4934
    },
    {
      "epoch": 2.6496426294419653,
      "grad_norm": 0.66015625,
      "learning_rate": 2.2680924951412337e-05,
      "loss": 0.7346,
      "step": 4935
    },
    {
      "epoch": 2.650179524177041,
      "grad_norm": 0.74609375,
      "learning_rate": 2.267251635596601e-05,
      "loss": 0.8143,
      "step": 4936
    },
    {
      "epoch": 2.650716418912117,
      "grad_norm": 0.7265625,
      "learning_rate": 2.2664108026115043e-05,
      "loss": 0.9506,
      "step": 4937
    },
    {
      "epoch": 2.651253313647193,
      "grad_norm": 0.6875,
      "learning_rate": 2.2655699962818934e-05,
      "loss": 0.8035,
      "step": 4938
    },
    {
      "epoch": 2.651790208382269,
      "grad_norm": 0.65625,
      "learning_rate": 2.2647292167037144e-05,
      "loss": 0.7886,
      "step": 4939
    },
    {
      "epoch": 2.652327103117345,
      "grad_norm": 0.6875,
      "learning_rate": 2.263888463972911e-05,
      "loss": 0.7561,
      "step": 4940
    },
    {
      "epoch": 2.652863997852421,
      "grad_norm": 0.6953125,
      "learning_rate": 2.2630477381854247e-05,
      "loss": 0.839,
      "step": 4941
    },
    {
      "epoch": 2.6534008925874972,
      "grad_norm": 0.6484375,
      "learning_rate": 2.2622070394371926e-05,
      "loss": 0.7929,
      "step": 4942
    },
    {
      "epoch": 2.653937787322573,
      "grad_norm": 0.609375,
      "learning_rate": 2.261366367824148e-05,
      "loss": 0.7112,
      "step": 4943
    },
    {
      "epoch": 2.654474682057649,
      "grad_norm": 0.81640625,
      "learning_rate": 2.2605257234422224e-05,
      "loss": 0.8821,
      "step": 4944
    },
    {
      "epoch": 2.655011576792725,
      "grad_norm": 0.66015625,
      "learning_rate": 2.2596851063873446e-05,
      "loss": 0.892,
      "step": 4945
    },
    {
      "epoch": 2.6555484715278013,
      "grad_norm": 0.6328125,
      "learning_rate": 2.25884451675544e-05,
      "loss": 0.8032,
      "step": 4946
    },
    {
      "epoch": 2.656085366262877,
      "grad_norm": 0.703125,
      "learning_rate": 2.2580039546424292e-05,
      "loss": 0.9186,
      "step": 4947
    },
    {
      "epoch": 2.656622260997953,
      "grad_norm": 0.71875,
      "learning_rate": 2.2571634201442326e-05,
      "loss": 0.8775,
      "step": 4948
    },
    {
      "epoch": 2.657159155733029,
      "grad_norm": 0.7421875,
      "learning_rate": 2.2563229133567647e-05,
      "loss": 0.807,
      "step": 4949
    },
    {
      "epoch": 2.657696050468105,
      "grad_norm": 0.66015625,
      "learning_rate": 2.2554824343759392e-05,
      "loss": 0.7873,
      "step": 4950
    },
    {
      "epoch": 2.657696050468105,
      "eval_loss": 0.9035736918449402,
      "eval_runtime": 46.1198,
      "eval_samples_per_second": 20.859,
      "eval_steps_per_second": 20.859,
      "step": 4950
    },
    {
      "epoch": 2.658232945203181,
      "grad_norm": 2.859375,
      "learning_rate": 2.254641983297664e-05,
      "loss": 0.9557,
      "step": 4951
    },
    {
      "epoch": 2.658769839938257,
      "grad_norm": 0.7421875,
      "learning_rate": 2.253801560217845e-05,
      "loss": 0.8769,
      "step": 4952
    },
    {
      "epoch": 2.6593067346733332,
      "grad_norm": 0.81640625,
      "learning_rate": 2.2529611652323866e-05,
      "loss": 1.0339,
      "step": 4953
    },
    {
      "epoch": 2.6598436294084093,
      "grad_norm": 0.6640625,
      "learning_rate": 2.2521207984371872e-05,
      "loss": 0.9272,
      "step": 4954
    },
    {
      "epoch": 2.660380524143485,
      "grad_norm": 0.73828125,
      "learning_rate": 2.2512804599281438e-05,
      "loss": 0.9604,
      "step": 4955
    },
    {
      "epoch": 2.660917418878561,
      "grad_norm": 0.69921875,
      "learning_rate": 2.25044014980115e-05,
      "loss": 0.8449,
      "step": 4956
    },
    {
      "epoch": 2.6614543136136373,
      "grad_norm": 0.65234375,
      "learning_rate": 2.249599868152096e-05,
      "loss": 0.8504,
      "step": 4957
    },
    {
      "epoch": 2.661991208348713,
      "grad_norm": 0.6640625,
      "learning_rate": 2.248759615076868e-05,
      "loss": 0.9573,
      "step": 4958
    },
    {
      "epoch": 2.662528103083789,
      "grad_norm": 0.66015625,
      "learning_rate": 2.247919390671349e-05,
      "loss": 0.7302,
      "step": 4959
    },
    {
      "epoch": 2.663064997818865,
      "grad_norm": 0.72265625,
      "learning_rate": 2.24707919503142e-05,
      "loss": 1.0578,
      "step": 4960
    },
    {
      "epoch": 2.6636018925539413,
      "grad_norm": 0.58984375,
      "learning_rate": 2.2462390282529577e-05,
      "loss": 0.7379,
      "step": 4961
    },
    {
      "epoch": 2.664138787289017,
      "grad_norm": 0.8671875,
      "learning_rate": 2.2453988904318364e-05,
      "loss": 0.9434,
      "step": 4962
    },
    {
      "epoch": 2.664675682024093,
      "grad_norm": 0.66796875,
      "learning_rate": 2.2445587816639252e-05,
      "loss": 0.8265,
      "step": 4963
    },
    {
      "epoch": 2.6652125767591692,
      "grad_norm": 0.7109375,
      "learning_rate": 2.2437187020450926e-05,
      "loss": 1.0917,
      "step": 4964
    },
    {
      "epoch": 2.6657494714942453,
      "grad_norm": 0.6640625,
      "learning_rate": 2.2428786516712015e-05,
      "loss": 0.8258,
      "step": 4965
    },
    {
      "epoch": 2.666286366229321,
      "grad_norm": 0.75390625,
      "learning_rate": 2.242038630638112e-05,
      "loss": 1.0332,
      "step": 4966
    },
    {
      "epoch": 2.666823260964397,
      "grad_norm": 0.61328125,
      "learning_rate": 2.241198639041683e-05,
      "loss": 0.7824,
      "step": 4967
    },
    {
      "epoch": 2.6673601556994733,
      "grad_norm": 0.64453125,
      "learning_rate": 2.2403586769777652e-05,
      "loss": 0.9146,
      "step": 4968
    },
    {
      "epoch": 2.667897050434549,
      "grad_norm": 0.59765625,
      "learning_rate": 2.239518744542211e-05,
      "loss": 0.7834,
      "step": 4969
    },
    {
      "epoch": 2.668433945169625,
      "grad_norm": 0.68359375,
      "learning_rate": 2.238678841830867e-05,
      "loss": 0.7471,
      "step": 4970
    },
    {
      "epoch": 2.668970839904701,
      "grad_norm": 0.6484375,
      "learning_rate": 2.2378389689395754e-05,
      "loss": 0.7633,
      "step": 4971
    },
    {
      "epoch": 2.6695077346397773,
      "grad_norm": 0.6015625,
      "learning_rate": 2.2369991259641785e-05,
      "loss": 0.8676,
      "step": 4972
    },
    {
      "epoch": 2.6700446293748534,
      "grad_norm": 0.73828125,
      "learning_rate": 2.2361593130005118e-05,
      "loss": 0.8312,
      "step": 4973
    },
    {
      "epoch": 2.670581524109929,
      "grad_norm": 0.69140625,
      "learning_rate": 2.235319530144409e-05,
      "loss": 0.7971,
      "step": 4974
    },
    {
      "epoch": 2.6711184188450052,
      "grad_norm": 0.65234375,
      "learning_rate": 2.2344797774916998e-05,
      "loss": 0.833,
      "step": 4975
    },
    {
      "epoch": 2.6716553135800813,
      "grad_norm": 0.73828125,
      "learning_rate": 2.2336400551382096e-05,
      "loss": 0.8202,
      "step": 4976
    },
    {
      "epoch": 2.672192208315157,
      "grad_norm": 0.6484375,
      "learning_rate": 2.232800363179763e-05,
      "loss": 0.743,
      "step": 4977
    },
    {
      "epoch": 2.672729103050233,
      "grad_norm": 0.6640625,
      "learning_rate": 2.2319607017121786e-05,
      "loss": 0.844,
      "step": 4978
    },
    {
      "epoch": 2.6732659977853093,
      "grad_norm": 0.703125,
      "learning_rate": 2.2311210708312726e-05,
      "loss": 0.8646,
      "step": 4979
    },
    {
      "epoch": 2.6738028925203854,
      "grad_norm": 0.79296875,
      "learning_rate": 2.230281470632857e-05,
      "loss": 0.962,
      "step": 4980
    },
    {
      "epoch": 2.674339787255461,
      "grad_norm": 0.6484375,
      "learning_rate": 2.2294419012127418e-05,
      "loss": 1.1699,
      "step": 4981
    },
    {
      "epoch": 2.674876681990537,
      "grad_norm": 0.6328125,
      "learning_rate": 2.228602362666732e-05,
      "loss": 0.849,
      "step": 4982
    },
    {
      "epoch": 2.6754135767256133,
      "grad_norm": 0.77734375,
      "learning_rate": 2.2277628550906293e-05,
      "loss": 0.9349,
      "step": 4983
    },
    {
      "epoch": 2.6759504714606894,
      "grad_norm": 0.59375,
      "learning_rate": 2.226923378580234e-05,
      "loss": 0.6634,
      "step": 4984
    },
    {
      "epoch": 2.676487366195765,
      "grad_norm": 0.73828125,
      "learning_rate": 2.2260839332313375e-05,
      "loss": 0.8338,
      "step": 4985
    },
    {
      "epoch": 2.6770242609308412,
      "grad_norm": 0.83203125,
      "learning_rate": 2.2252445191397337e-05,
      "loss": 0.839,
      "step": 4986
    },
    {
      "epoch": 2.6775611556659173,
      "grad_norm": 0.69140625,
      "learning_rate": 2.22440513640121e-05,
      "loss": 0.8034,
      "step": 4987
    },
    {
      "epoch": 2.678098050400993,
      "grad_norm": 0.72265625,
      "learning_rate": 2.2235657851115497e-05,
      "loss": 0.9674,
      "step": 4988
    },
    {
      "epoch": 2.678634945136069,
      "grad_norm": 0.66796875,
      "learning_rate": 2.2227264653665345e-05,
      "loss": 0.7859,
      "step": 4989
    },
    {
      "epoch": 2.6791718398711453,
      "grad_norm": 0.65625,
      "learning_rate": 2.221887177261941e-05,
      "loss": 0.7488,
      "step": 4990
    },
    {
      "epoch": 2.6797087346062214,
      "grad_norm": 0.65625,
      "learning_rate": 2.221047920893542e-05,
      "loss": 0.6674,
      "step": 4991
    },
    {
      "epoch": 2.6802456293412975,
      "grad_norm": 0.81640625,
      "learning_rate": 2.22020869635711e-05,
      "loss": 0.8728,
      "step": 4992
    },
    {
      "epoch": 2.680782524076373,
      "grad_norm": 0.68359375,
      "learning_rate": 2.2193695037484068e-05,
      "loss": 0.91,
      "step": 4993
    },
    {
      "epoch": 2.6813194188114493,
      "grad_norm": 0.79296875,
      "learning_rate": 2.2185303431631982e-05,
      "loss": 0.9383,
      "step": 4994
    },
    {
      "epoch": 2.6818563135465254,
      "grad_norm": 0.7421875,
      "learning_rate": 2.2176912146972414e-05,
      "loss": 0.8167,
      "step": 4995
    },
    {
      "epoch": 2.682393208281601,
      "grad_norm": 0.78515625,
      "learning_rate": 2.216852118446292e-05,
      "loss": 1.0551,
      "step": 4996
    },
    {
      "epoch": 2.6829301030166772,
      "grad_norm": 0.88671875,
      "learning_rate": 2.216013054506102e-05,
      "loss": 0.9115,
      "step": 4997
    },
    {
      "epoch": 2.6834669977517533,
      "grad_norm": 0.578125,
      "learning_rate": 2.215174022972419e-05,
      "loss": 0.7588,
      "step": 4998
    },
    {
      "epoch": 2.684003892486829,
      "grad_norm": 0.62109375,
      "learning_rate": 2.2143350239409867e-05,
      "loss": 0.7433,
      "step": 4999
    },
    {
      "epoch": 2.684540787221905,
      "grad_norm": 0.734375,
      "learning_rate": 2.213496057507547e-05,
      "loss": 0.8441,
      "step": 5000
    },
    {
      "epoch": 2.684540787221905,
      "eval_loss": 0.903358519077301,
      "eval_runtime": 45.9635,
      "eval_samples_per_second": 20.93,
      "eval_steps_per_second": 20.93,
      "step": 5000
    },
    {
      "epoch": 2.6850776819569813,
      "grad_norm": 0.7578125,
      "learning_rate": 2.212657123767834e-05,
      "loss": 0.9228,
      "step": 5001
    },
    {
      "epoch": 2.6856145766920574,
      "grad_norm": 0.72265625,
      "learning_rate": 2.2118182228175824e-05,
      "loss": 0.7715,
      "step": 5002
    },
    {
      "epoch": 2.6861514714271335,
      "grad_norm": 0.65625,
      "learning_rate": 2.2109793547525215e-05,
      "loss": 0.7831,
      "step": 5003
    },
    {
      "epoch": 2.686688366162209,
      "grad_norm": 0.66796875,
      "learning_rate": 2.2101405196683756e-05,
      "loss": 0.7604,
      "step": 5004
    },
    {
      "epoch": 2.6872252608972853,
      "grad_norm": 0.75390625,
      "learning_rate": 2.2093017176608677e-05,
      "loss": 0.948,
      "step": 5005
    },
    {
      "epoch": 2.6877621556323614,
      "grad_norm": 0.74609375,
      "learning_rate": 2.208462948825715e-05,
      "loss": 0.9593,
      "step": 5006
    },
    {
      "epoch": 2.688299050367437,
      "grad_norm": 0.69140625,
      "learning_rate": 2.2076242132586317e-05,
      "loss": 0.7838,
      "step": 5007
    },
    {
      "epoch": 2.6888359451025132,
      "grad_norm": 0.65625,
      "learning_rate": 2.2067855110553286e-05,
      "loss": 0.666,
      "step": 5008
    },
    {
      "epoch": 2.6893728398375893,
      "grad_norm": 0.76171875,
      "learning_rate": 2.205946842311513e-05,
      "loss": 0.9892,
      "step": 5009
    },
    {
      "epoch": 2.6899097345726655,
      "grad_norm": 0.79296875,
      "learning_rate": 2.2051082071228854e-05,
      "loss": 0.8428,
      "step": 5010
    },
    {
      "epoch": 2.6904466293077416,
      "grad_norm": 0.65234375,
      "learning_rate": 2.2042696055851463e-05,
      "loss": 0.9376,
      "step": 5011
    },
    {
      "epoch": 2.6909835240428173,
      "grad_norm": 0.703125,
      "learning_rate": 2.2034310377939895e-05,
      "loss": 0.7812,
      "step": 5012
    },
    {
      "epoch": 2.6915204187778934,
      "grad_norm": 0.640625,
      "learning_rate": 2.2025925038451075e-05,
      "loss": 0.8864,
      "step": 5013
    },
    {
      "epoch": 2.6920573135129695,
      "grad_norm": 0.65234375,
      "learning_rate": 2.2017540038341875e-05,
      "loss": 0.8902,
      "step": 5014
    },
    {
      "epoch": 2.692594208248045,
      "grad_norm": 0.69921875,
      "learning_rate": 2.2009155378569117e-05,
      "loss": 0.8702,
      "step": 5015
    },
    {
      "epoch": 2.6931311029831213,
      "grad_norm": 0.7265625,
      "learning_rate": 2.2000771060089617e-05,
      "loss": 0.9905,
      "step": 5016
    },
    {
      "epoch": 2.6936679977181974,
      "grad_norm": 0.7265625,
      "learning_rate": 2.199238708386012e-05,
      "loss": 1.1197,
      "step": 5017
    },
    {
      "epoch": 2.694204892453273,
      "grad_norm": 0.78515625,
      "learning_rate": 2.198400345083735e-05,
      "loss": 0.9472,
      "step": 5018
    },
    {
      "epoch": 2.6947417871883492,
      "grad_norm": 0.59375,
      "learning_rate": 2.1975620161977973e-05,
      "loss": 0.7182,
      "step": 5019
    },
    {
      "epoch": 2.6952786819234253,
      "grad_norm": 0.640625,
      "learning_rate": 2.1967237218238628e-05,
      "loss": 0.8,
      "step": 5020
    },
    {
      "epoch": 2.6958155766585015,
      "grad_norm": 0.6796875,
      "learning_rate": 2.1958854620575936e-05,
      "loss": 0.83,
      "step": 5021
    },
    {
      "epoch": 2.6963524713935776,
      "grad_norm": 0.6875,
      "learning_rate": 2.195047236994644e-05,
      "loss": 0.8183,
      "step": 5022
    },
    {
      "epoch": 2.6968893661286533,
      "grad_norm": 0.83984375,
      "learning_rate": 2.1942090467306668e-05,
      "loss": 0.8642,
      "step": 5023
    },
    {
      "epoch": 2.6974262608637294,
      "grad_norm": 0.6875,
      "learning_rate": 2.193370891361309e-05,
      "loss": 0.7442,
      "step": 5024
    },
    {
      "epoch": 2.6979631555988055,
      "grad_norm": 0.74609375,
      "learning_rate": 2.1925327709822167e-05,
      "loss": 0.9279,
      "step": 5025
    },
    {
      "epoch": 2.698500050333881,
      "grad_norm": 0.58984375,
      "learning_rate": 2.1916946856890293e-05,
      "loss": 0.7511,
      "step": 5026
    },
    {
      "epoch": 2.6990369450689573,
      "grad_norm": 0.73046875,
      "learning_rate": 2.1908566355773817e-05,
      "loss": 0.8585,
      "step": 5027
    },
    {
      "epoch": 2.6995738398040334,
      "grad_norm": 0.86328125,
      "learning_rate": 2.1900186207429078e-05,
      "loss": 0.9793,
      "step": 5028
    },
    {
      "epoch": 2.7001107345391095,
      "grad_norm": 0.58984375,
      "learning_rate": 2.1891806412812338e-05,
      "loss": 0.7375,
      "step": 5029
    },
    {
      "epoch": 2.7006476292741857,
      "grad_norm": 0.74609375,
      "learning_rate": 2.1883426972879853e-05,
      "loss": 0.9499,
      "step": 5030
    },
    {
      "epoch": 2.7011845240092613,
      "grad_norm": 0.625,
      "learning_rate": 2.1875047888587823e-05,
      "loss": 0.814,
      "step": 5031
    },
    {
      "epoch": 2.7017214187443375,
      "grad_norm": 0.78515625,
      "learning_rate": 2.186666916089239e-05,
      "loss": 0.7674,
      "step": 5032
    },
    {
      "epoch": 2.7022583134794136,
      "grad_norm": 0.69921875,
      "learning_rate": 2.1858290790749698e-05,
      "loss": 0.7517,
      "step": 5033
    },
    {
      "epoch": 2.7027952082144893,
      "grad_norm": 0.6875,
      "learning_rate": 2.1849912779115818e-05,
      "loss": 0.8155,
      "step": 5034
    },
    {
      "epoch": 2.7033321029495654,
      "grad_norm": 0.74609375,
      "learning_rate": 2.1841535126946776e-05,
      "loss": 0.9909,
      "step": 5035
    },
    {
      "epoch": 2.7038689976846415,
      "grad_norm": 0.73046875,
      "learning_rate": 2.1833157835198574e-05,
      "loss": 0.9182,
      "step": 5036
    },
    {
      "epoch": 2.704405892419717,
      "grad_norm": 0.6484375,
      "learning_rate": 2.1824780904827164e-05,
      "loss": 0.8444,
      "step": 5037
    },
    {
      "epoch": 2.7049427871547933,
      "grad_norm": 0.69140625,
      "learning_rate": 2.181640433678847e-05,
      "loss": 0.8912,
      "step": 5038
    },
    {
      "epoch": 2.7054796818898694,
      "grad_norm": 1.03125,
      "learning_rate": 2.1808028132038354e-05,
      "loss": 0.7917,
      "step": 5039
    },
    {
      "epoch": 2.7060165766249455,
      "grad_norm": 0.56640625,
      "learning_rate": 2.179965229153265e-05,
      "loss": 0.6682,
      "step": 5040
    },
    {
      "epoch": 2.7065534713600217,
      "grad_norm": 0.7421875,
      "learning_rate": 2.1791276816227152e-05,
      "loss": 0.7958,
      "step": 5041
    },
    {
      "epoch": 2.7070903660950973,
      "grad_norm": 0.61328125,
      "learning_rate": 2.178290170707761e-05,
      "loss": 0.9435,
      "step": 5042
    },
    {
      "epoch": 2.7076272608301735,
      "grad_norm": 0.6640625,
      "learning_rate": 2.177452696503972e-05,
      "loss": 0.796,
      "step": 5043
    },
    {
      "epoch": 2.7081641555652496,
      "grad_norm": 0.6796875,
      "learning_rate": 2.176615259106915e-05,
      "loss": 0.8683,
      "step": 5044
    },
    {
      "epoch": 2.7087010503003253,
      "grad_norm": 0.58984375,
      "learning_rate": 2.1757778586121517e-05,
      "loss": 0.7923,
      "step": 5045
    },
    {
      "epoch": 2.7092379450354014,
      "grad_norm": 0.73046875,
      "learning_rate": 2.1749404951152412e-05,
      "loss": 0.6945,
      "step": 5046
    },
    {
      "epoch": 2.7097748397704775,
      "grad_norm": 0.67578125,
      "learning_rate": 2.1741031687117368e-05,
      "loss": 0.8413,
      "step": 5047
    },
    {
      "epoch": 2.7103117345055536,
      "grad_norm": 0.671875,
      "learning_rate": 2.1732658794971874e-05,
      "loss": 1.0605,
      "step": 5048
    },
    {
      "epoch": 2.7108486292406297,
      "grad_norm": 0.62890625,
      "learning_rate": 2.1724286275671396e-05,
      "loss": 0.7467,
      "step": 5049
    },
    {
      "epoch": 2.7113855239757054,
      "grad_norm": 0.70703125,
      "learning_rate": 2.1715914130171337e-05,
      "loss": 0.9242,
      "step": 5050
    },
    {
      "epoch": 2.7113855239757054,
      "eval_loss": 0.9034383296966553,
      "eval_runtime": 46.065,
      "eval_samples_per_second": 20.884,
      "eval_steps_per_second": 20.884,
      "step": 5050
    },
    {
      "epoch": 2.7119224187107815,
      "grad_norm": 0.671875,
      "learning_rate": 2.170754235942707e-05,
      "loss": 0.6494,
      "step": 5051
    },
    {
      "epoch": 2.7124593134458577,
      "grad_norm": 0.66796875,
      "learning_rate": 2.169917096439391e-05,
      "loss": 0.8253,
      "step": 5052
    },
    {
      "epoch": 2.7129962081809333,
      "grad_norm": 0.6640625,
      "learning_rate": 2.169079994602714e-05,
      "loss": 0.8121,
      "step": 5053
    },
    {
      "epoch": 2.7135331029160095,
      "grad_norm": 0.77734375,
      "learning_rate": 2.1682429305282008e-05,
      "loss": 0.8765,
      "step": 5054
    },
    {
      "epoch": 2.7140699976510856,
      "grad_norm": 0.8125,
      "learning_rate": 2.1674059043113707e-05,
      "loss": 1.0917,
      "step": 5055
    },
    {
      "epoch": 2.7146068923861613,
      "grad_norm": 0.71484375,
      "learning_rate": 2.1665689160477383e-05,
      "loss": 1.0088,
      "step": 5056
    },
    {
      "epoch": 2.7151437871212374,
      "grad_norm": 0.765625,
      "learning_rate": 2.1657319658328158e-05,
      "loss": 0.8097,
      "step": 5057
    },
    {
      "epoch": 2.7156806818563135,
      "grad_norm": 0.99609375,
      "learning_rate": 2.1648950537621087e-05,
      "loss": 1.2574,
      "step": 5058
    },
    {
      "epoch": 2.7162175765913896,
      "grad_norm": 0.61328125,
      "learning_rate": 2.1640581799311208e-05,
      "loss": 0.8028,
      "step": 5059
    },
    {
      "epoch": 2.7167544713264657,
      "grad_norm": 0.72265625,
      "learning_rate": 2.1632213444353482e-05,
      "loss": 0.7571,
      "step": 5060
    },
    {
      "epoch": 2.7172913660615414,
      "grad_norm": 0.72265625,
      "learning_rate": 2.1623845473702848e-05,
      "loss": 0.743,
      "step": 5061
    },
    {
      "epoch": 2.7178282607966175,
      "grad_norm": 0.67578125,
      "learning_rate": 2.1615477888314203e-05,
      "loss": 0.9261,
      "step": 5062
    },
    {
      "epoch": 2.7183651555316937,
      "grad_norm": 0.78515625,
      "learning_rate": 2.1607110689142393e-05,
      "loss": 0.8333,
      "step": 5063
    },
    {
      "epoch": 2.7189020502667693,
      "grad_norm": 0.69140625,
      "learning_rate": 2.1598743877142222e-05,
      "loss": 0.8227,
      "step": 5064
    },
    {
      "epoch": 2.7194389450018455,
      "grad_norm": 0.6640625,
      "learning_rate": 2.159037745326844e-05,
      "loss": 0.925,
      "step": 5065
    },
    {
      "epoch": 2.7199758397369216,
      "grad_norm": 0.80078125,
      "learning_rate": 2.1582011418475782e-05,
      "loss": 0.973,
      "step": 5066
    },
    {
      "epoch": 2.7205127344719977,
      "grad_norm": 0.9140625,
      "learning_rate": 2.15736457737189e-05,
      "loss": 0.9659,
      "step": 5067
    },
    {
      "epoch": 2.721049629207074,
      "grad_norm": 0.609375,
      "learning_rate": 2.1565280519952435e-05,
      "loss": 0.7824,
      "step": 5068
    },
    {
      "epoch": 2.7215865239421495,
      "grad_norm": 0.66796875,
      "learning_rate": 2.1556915658130954e-05,
      "loss": 0.8146,
      "step": 5069
    },
    {
      "epoch": 2.7221234186772256,
      "grad_norm": 0.76171875,
      "learning_rate": 2.154855118920899e-05,
      "loss": 0.9196,
      "step": 5070
    },
    {
      "epoch": 2.7226603134123017,
      "grad_norm": 1.59375,
      "learning_rate": 2.154018711414106e-05,
      "loss": 1.2276,
      "step": 5071
    },
    {
      "epoch": 2.7231972081473774,
      "grad_norm": 0.7578125,
      "learning_rate": 2.1531823433881588e-05,
      "loss": 0.8539,
      "step": 5072
    },
    {
      "epoch": 2.7237341028824535,
      "grad_norm": 0.76171875,
      "learning_rate": 2.152346014938498e-05,
      "loss": 0.7373,
      "step": 5073
    },
    {
      "epoch": 2.7242709976175297,
      "grad_norm": 0.78125,
      "learning_rate": 2.1515097261605596e-05,
      "loss": 0.6589,
      "step": 5074
    },
    {
      "epoch": 2.7248078923526053,
      "grad_norm": 0.73046875,
      "learning_rate": 2.150673477149775e-05,
      "loss": 0.9421,
      "step": 5075
    },
    {
      "epoch": 2.7253447870876815,
      "grad_norm": 0.65625,
      "learning_rate": 2.149837268001571e-05,
      "loss": 0.8746,
      "step": 5076
    },
    {
      "epoch": 2.7258816818227576,
      "grad_norm": 0.64453125,
      "learning_rate": 2.149001098811369e-05,
      "loss": 0.6623,
      "step": 5077
    },
    {
      "epoch": 2.7264185765578337,
      "grad_norm": 0.6484375,
      "learning_rate": 2.148164969674586e-05,
      "loss": 0.8815,
      "step": 5078
    },
    {
      "epoch": 2.72695547129291,
      "grad_norm": 0.6640625,
      "learning_rate": 2.147328880686636e-05,
      "loss": 0.8938,
      "step": 5079
    },
    {
      "epoch": 2.7274923660279855,
      "grad_norm": 0.74609375,
      "learning_rate": 2.146492831942927e-05,
      "loss": 0.8871,
      "step": 5080
    },
    {
      "epoch": 2.7280292607630616,
      "grad_norm": 0.984375,
      "learning_rate": 2.145656823538862e-05,
      "loss": 0.9893,
      "step": 5081
    },
    {
      "epoch": 2.7285661554981377,
      "grad_norm": 0.65625,
      "learning_rate": 2.1448208555698416e-05,
      "loss": 0.7905,
      "step": 5082
    },
    {
      "epoch": 2.7291030502332134,
      "grad_norm": 0.77734375,
      "learning_rate": 2.1439849281312593e-05,
      "loss": 1.0244,
      "step": 5083
    },
    {
      "epoch": 2.7296399449682895,
      "grad_norm": 0.6953125,
      "learning_rate": 2.143149041318506e-05,
      "loss": 0.7836,
      "step": 5084
    },
    {
      "epoch": 2.7301768397033657,
      "grad_norm": 0.73046875,
      "learning_rate": 2.1423131952269655e-05,
      "loss": 0.8897,
      "step": 5085
    },
    {
      "epoch": 2.730713734438442,
      "grad_norm": 0.6875,
      "learning_rate": 2.1414773899520187e-05,
      "loss": 0.9864,
      "step": 5086
    },
    {
      "epoch": 2.731250629173518,
      "grad_norm": 0.60546875,
      "learning_rate": 2.1406416255890428e-05,
      "loss": 0.6947,
      "step": 5087
    },
    {
      "epoch": 2.7317875239085936,
      "grad_norm": 0.60546875,
      "learning_rate": 2.139805902233408e-05,
      "loss": 0.8749,
      "step": 5088
    },
    {
      "epoch": 2.7323244186436697,
      "grad_norm": 0.66796875,
      "learning_rate": 2.1389702199804808e-05,
      "loss": 0.9514,
      "step": 5089
    },
    {
      "epoch": 2.732861313378746,
      "grad_norm": 0.671875,
      "learning_rate": 2.1381345789256245e-05,
      "loss": 0.7669,
      "step": 5090
    },
    {
      "epoch": 2.7333982081138215,
      "grad_norm": 0.7421875,
      "learning_rate": 2.137298979164195e-05,
      "loss": 1.098,
      "step": 5091
    },
    {
      "epoch": 2.7339351028488976,
      "grad_norm": 0.7109375,
      "learning_rate": 2.1364634207915447e-05,
      "loss": 1.1326,
      "step": 5092
    },
    {
      "epoch": 2.7344719975839737,
      "grad_norm": 0.69140625,
      "learning_rate": 2.1356279039030235e-05,
      "loss": 0.8915,
      "step": 5093
    },
    {
      "epoch": 2.7350088923190494,
      "grad_norm": 0.66796875,
      "learning_rate": 2.1347924285939714e-05,
      "loss": 0.8269,
      "step": 5094
    },
    {
      "epoch": 2.7355457870541255,
      "grad_norm": 0.74609375,
      "learning_rate": 2.1339569949597284e-05,
      "loss": 0.8303,
      "step": 5095
    },
    {
      "epoch": 2.7360826817892017,
      "grad_norm": 0.84765625,
      "learning_rate": 2.133121603095628e-05,
      "loss": 0.9537,
      "step": 5096
    },
    {
      "epoch": 2.736619576524278,
      "grad_norm": 0.765625,
      "learning_rate": 2.1322862530969984e-05,
      "loss": 0.7892,
      "step": 5097
    },
    {
      "epoch": 2.737156471259354,
      "grad_norm": 0.78515625,
      "learning_rate": 2.1314509450591647e-05,
      "loss": 1.0223,
      "step": 5098
    },
    {
      "epoch": 2.7376933659944296,
      "grad_norm": 1.015625,
      "learning_rate": 2.130615679077445e-05,
      "loss": 0.9654,
      "step": 5099
    },
    {
      "epoch": 2.7382302607295057,
      "grad_norm": 0.9765625,
      "learning_rate": 2.1297804552471547e-05,
      "loss": 0.8931,
      "step": 5100
    },
    {
      "epoch": 2.7382302607295057,
      "eval_loss": 0.9029310345649719,
      "eval_runtime": 47.2098,
      "eval_samples_per_second": 20.377,
      "eval_steps_per_second": 20.377,
      "step": 5100
    },
    {
      "epoch": 2.738767155464582,
      "grad_norm": 0.70703125,
      "learning_rate": 2.1289452736636035e-05,
      "loss": 0.7365,
      "step": 5101
    },
    {
      "epoch": 2.7393040501996575,
      "grad_norm": 0.66015625,
      "learning_rate": 2.1281101344220937e-05,
      "loss": 0.8493,
      "step": 5102
    },
    {
      "epoch": 2.7398409449347336,
      "grad_norm": 0.6875,
      "learning_rate": 2.1272750376179286e-05,
      "loss": 0.8451,
      "step": 5103
    },
    {
      "epoch": 2.7403778396698097,
      "grad_norm": 0.6328125,
      "learning_rate": 2.1264399833464015e-05,
      "loss": 0.8994,
      "step": 5104
    },
    {
      "epoch": 2.740914734404886,
      "grad_norm": 0.73046875,
      "learning_rate": 2.1256049717028032e-05,
      "loss": 0.9397,
      "step": 5105
    },
    {
      "epoch": 2.741451629139962,
      "grad_norm": 0.69921875,
      "learning_rate": 2.1247700027824182e-05,
      "loss": 0.6846,
      "step": 5106
    },
    {
      "epoch": 2.7419885238750377,
      "grad_norm": 0.609375,
      "learning_rate": 2.1239350766805285e-05,
      "loss": 0.738,
      "step": 5107
    },
    {
      "epoch": 2.742525418610114,
      "grad_norm": 0.7890625,
      "learning_rate": 2.1231001934924088e-05,
      "loss": 1.0414,
      "step": 5108
    },
    {
      "epoch": 2.74306231334519,
      "grad_norm": 0.7734375,
      "learning_rate": 2.12226535331333e-05,
      "loss": 0.8378,
      "step": 5109
    },
    {
      "epoch": 2.7435992080802656,
      "grad_norm": 0.69921875,
      "learning_rate": 2.1214305562385592e-05,
      "loss": 0.8825,
      "step": 5110
    },
    {
      "epoch": 2.7441361028153417,
      "grad_norm": 0.71484375,
      "learning_rate": 2.1205958023633546e-05,
      "loss": 0.8217,
      "step": 5111
    },
    {
      "epoch": 2.744672997550418,
      "grad_norm": 0.74609375,
      "learning_rate": 2.1197610917829742e-05,
      "loss": 0.8131,
      "step": 5112
    },
    {
      "epoch": 2.7452098922854935,
      "grad_norm": 0.65625,
      "learning_rate": 2.1189264245926687e-05,
      "loss": 0.7979,
      "step": 5113
    },
    {
      "epoch": 2.7457467870205696,
      "grad_norm": 0.62109375,
      "learning_rate": 2.1180918008876836e-05,
      "loss": 0.7882,
      "step": 5114
    },
    {
      "epoch": 2.7462836817556457,
      "grad_norm": 0.56640625,
      "learning_rate": 2.1172572207632608e-05,
      "loss": 0.7587,
      "step": 5115
    },
    {
      "epoch": 2.746820576490722,
      "grad_norm": 0.71484375,
      "learning_rate": 2.1164226843146365e-05,
      "loss": 0.9854,
      "step": 5116
    },
    {
      "epoch": 2.747357471225798,
      "grad_norm": 0.72265625,
      "learning_rate": 2.1155881916370407e-05,
      "loss": 0.8934,
      "step": 5117
    },
    {
      "epoch": 2.7478943659608737,
      "grad_norm": 0.69921875,
      "learning_rate": 2.114753742825702e-05,
      "loss": 0.8471,
      "step": 5118
    },
    {
      "epoch": 2.74843126069595,
      "grad_norm": 0.5859375,
      "learning_rate": 2.1139193379758386e-05,
      "loss": 0.8186,
      "step": 5119
    },
    {
      "epoch": 2.748968155431026,
      "grad_norm": 0.71875,
      "learning_rate": 2.1130849771826683e-05,
      "loss": 0.8046,
      "step": 5120
    },
    {
      "epoch": 2.7495050501661016,
      "grad_norm": 0.9375,
      "learning_rate": 2.112250660541402e-05,
      "loss": 1.1443,
      "step": 5121
    },
    {
      "epoch": 2.7500419449011777,
      "grad_norm": 0.85546875,
      "learning_rate": 2.1114163881472457e-05,
      "loss": 0.903,
      "step": 5122
    },
    {
      "epoch": 2.750578839636254,
      "grad_norm": 0.69140625,
      "learning_rate": 2.1105821600954007e-05,
      "loss": 0.9971,
      "step": 5123
    },
    {
      "epoch": 2.75111573437133,
      "grad_norm": 0.703125,
      "learning_rate": 2.109747976481063e-05,
      "loss": 0.7847,
      "step": 5124
    },
    {
      "epoch": 2.751652629106406,
      "grad_norm": 0.72265625,
      "learning_rate": 2.1089138373994223e-05,
      "loss": 0.9203,
      "step": 5125
    },
    {
      "epoch": 2.7521895238414817,
      "grad_norm": 0.69921875,
      "learning_rate": 2.1080797429456677e-05,
      "loss": 0.8991,
      "step": 5126
    },
    {
      "epoch": 2.752726418576558,
      "grad_norm": 0.7109375,
      "learning_rate": 2.1072456932149758e-05,
      "loss": 0.9665,
      "step": 5127
    },
    {
      "epoch": 2.753263313311634,
      "grad_norm": 0.66796875,
      "learning_rate": 2.106411688302525e-05,
      "loss": 0.9092,
      "step": 5128
    },
    {
      "epoch": 2.7538002080467097,
      "grad_norm": 0.64453125,
      "learning_rate": 2.1055777283034843e-05,
      "loss": 0.8094,
      "step": 5129
    },
    {
      "epoch": 2.754337102781786,
      "grad_norm": 0.625,
      "learning_rate": 2.1047438133130196e-05,
      "loss": 0.8451,
      "step": 5130
    },
    {
      "epoch": 2.754873997516862,
      "grad_norm": 0.72265625,
      "learning_rate": 2.1039099434262917e-05,
      "loss": 0.8954,
      "step": 5131
    },
    {
      "epoch": 2.7554108922519376,
      "grad_norm": 0.6171875,
      "learning_rate": 2.1030761187384557e-05,
      "loss": 0.7139,
      "step": 5132
    },
    {
      "epoch": 2.7559477869870137,
      "grad_norm": 0.74609375,
      "learning_rate": 2.10224233934466e-05,
      "loss": 0.6447,
      "step": 5133
    },
    {
      "epoch": 2.75648468172209,
      "grad_norm": 0.6796875,
      "learning_rate": 2.1014086053400516e-05,
      "loss": 0.879,
      "step": 5134
    },
    {
      "epoch": 2.757021576457166,
      "grad_norm": 0.68359375,
      "learning_rate": 2.1005749168197696e-05,
      "loss": 0.9523,
      "step": 5135
    },
    {
      "epoch": 2.757558471192242,
      "grad_norm": 0.70703125,
      "learning_rate": 2.0997412738789475e-05,
      "loss": 0.7948,
      "step": 5136
    },
    {
      "epoch": 2.7580953659273177,
      "grad_norm": 0.703125,
      "learning_rate": 2.0989076766127148e-05,
      "loss": 0.8665,
      "step": 5137
    },
    {
      "epoch": 2.758632260662394,
      "grad_norm": 0.7109375,
      "learning_rate": 2.0980741251161948e-05,
      "loss": 0.7218,
      "step": 5138
    },
    {
      "epoch": 2.75916915539747,
      "grad_norm": 0.87890625,
      "learning_rate": 2.097240619484508e-05,
      "loss": 0.9672,
      "step": 5139
    },
    {
      "epoch": 2.7597060501325457,
      "grad_norm": 0.7734375,
      "learning_rate": 2.0964071598127666e-05,
      "loss": 0.968,
      "step": 5140
    },
    {
      "epoch": 2.760242944867622,
      "grad_norm": 0.68359375,
      "learning_rate": 2.0955737461960793e-05,
      "loss": 0.8452,
      "step": 5141
    },
    {
      "epoch": 2.760779839602698,
      "grad_norm": 0.69921875,
      "learning_rate": 2.0947403787295495e-05,
      "loss": 0.8077,
      "step": 5142
    },
    {
      "epoch": 2.761316734337774,
      "grad_norm": 0.640625,
      "learning_rate": 2.0939070575082752e-05,
      "loss": 0.882,
      "step": 5143
    },
    {
      "epoch": 2.76185362907285,
      "grad_norm": 0.66015625,
      "learning_rate": 2.093073782627348e-05,
      "loss": 0.9186,
      "step": 5144
    },
    {
      "epoch": 2.762390523807926,
      "grad_norm": 0.85546875,
      "learning_rate": 2.0922405541818555e-05,
      "loss": 0.8085,
      "step": 5145
    },
    {
      "epoch": 2.762927418543002,
      "grad_norm": 0.65234375,
      "learning_rate": 2.0914073722668788e-05,
      "loss": 0.9689,
      "step": 5146
    },
    {
      "epoch": 2.763464313278078,
      "grad_norm": 0.63671875,
      "learning_rate": 2.090574236977496e-05,
      "loss": 0.7498,
      "step": 5147
    },
    {
      "epoch": 2.7640012080131537,
      "grad_norm": 0.734375,
      "learning_rate": 2.089741148408778e-05,
      "loss": 0.7993,
      "step": 5148
    },
    {
      "epoch": 2.76453810274823,
      "grad_norm": 0.76953125,
      "learning_rate": 2.0889081066557902e-05,
      "loss": 0.7807,
      "step": 5149
    },
    {
      "epoch": 2.765074997483306,
      "grad_norm": 0.75390625,
      "learning_rate": 2.088075111813593e-05,
      "loss": 1.0847,
      "step": 5150
    },
    {
      "epoch": 2.765074997483306,
      "eval_loss": 0.9028027057647705,
      "eval_runtime": 45.7427,
      "eval_samples_per_second": 21.031,
      "eval_steps_per_second": 21.031,
      "step": 5150
    },
    {
      "epoch": 2.7656118922183817,
      "grad_norm": 0.6171875,
      "learning_rate": 2.0872421639772426e-05,
      "loss": 0.8422,
      "step": 5151
    },
    {
      "epoch": 2.766148786953458,
      "grad_norm": 0.63671875,
      "learning_rate": 2.0864092632417892e-05,
      "loss": 0.8761,
      "step": 5152
    },
    {
      "epoch": 2.766685681688534,
      "grad_norm": 0.65234375,
      "learning_rate": 2.0855764097022758e-05,
      "loss": 0.6654,
      "step": 5153
    },
    {
      "epoch": 2.76722257642361,
      "grad_norm": 0.5859375,
      "learning_rate": 2.0847436034537428e-05,
      "loss": 0.7115,
      "step": 5154
    },
    {
      "epoch": 2.767759471158686,
      "grad_norm": 0.65234375,
      "learning_rate": 2.083910844591222e-05,
      "loss": 0.8087,
      "step": 5155
    },
    {
      "epoch": 2.768296365893762,
      "grad_norm": 0.91015625,
      "learning_rate": 2.0830781332097446e-05,
      "loss": 0.867,
      "step": 5156
    },
    {
      "epoch": 2.768833260628838,
      "grad_norm": 0.671875,
      "learning_rate": 2.0822454694043317e-05,
      "loss": 0.8529,
      "step": 5157
    },
    {
      "epoch": 2.769370155363914,
      "grad_norm": 0.61328125,
      "learning_rate": 2.0814128532700004e-05,
      "loss": 1.0199,
      "step": 5158
    },
    {
      "epoch": 2.7699070500989897,
      "grad_norm": 0.6875,
      "learning_rate": 2.0805802849017646e-05,
      "loss": 0.9396,
      "step": 5159
    },
    {
      "epoch": 2.770443944834066,
      "grad_norm": 0.71484375,
      "learning_rate": 2.07974776439463e-05,
      "loss": 0.8423,
      "step": 5160
    },
    {
      "epoch": 2.770980839569142,
      "grad_norm": 0.72265625,
      "learning_rate": 2.078915291843597e-05,
      "loss": 0.9325,
      "step": 5161
    },
    {
      "epoch": 2.771517734304218,
      "grad_norm": 0.734375,
      "learning_rate": 2.078082867343662e-05,
      "loss": 0.904,
      "step": 5162
    },
    {
      "epoch": 2.7720546290392942,
      "grad_norm": 0.75,
      "learning_rate": 2.0772504909898143e-05,
      "loss": 0.854,
      "step": 5163
    },
    {
      "epoch": 2.77259152377437,
      "grad_norm": 0.6484375,
      "learning_rate": 2.07641816287704e-05,
      "loss": 0.7798,
      "step": 5164
    },
    {
      "epoch": 2.773128418509446,
      "grad_norm": 0.70703125,
      "learning_rate": 2.075585883100317e-05,
      "loss": 0.6959,
      "step": 5165
    },
    {
      "epoch": 2.773665313244522,
      "grad_norm": 0.76953125,
      "learning_rate": 2.0747536517546197e-05,
      "loss": 1.0618,
      "step": 5166
    },
    {
      "epoch": 2.774202207979598,
      "grad_norm": 0.61328125,
      "learning_rate": 2.0739214689349163e-05,
      "loss": 0.7171,
      "step": 5167
    },
    {
      "epoch": 2.774739102714674,
      "grad_norm": 0.75,
      "learning_rate": 2.0730893347361695e-05,
      "loss": 0.8049,
      "step": 5168
    },
    {
      "epoch": 2.77527599744975,
      "grad_norm": 0.671875,
      "learning_rate": 2.0722572492533356e-05,
      "loss": 0.9721,
      "step": 5169
    },
    {
      "epoch": 2.7758128921848257,
      "grad_norm": 0.62109375,
      "learning_rate": 2.0714252125813667e-05,
      "loss": 0.8227,
      "step": 5170
    },
    {
      "epoch": 2.776349786919902,
      "grad_norm": 0.6328125,
      "learning_rate": 2.0705932248152073e-05,
      "loss": 0.8242,
      "step": 5171
    },
    {
      "epoch": 2.776886681654978,
      "grad_norm": 0.68359375,
      "learning_rate": 2.0697612860498e-05,
      "loss": 0.9323,
      "step": 5172
    },
    {
      "epoch": 2.777423576390054,
      "grad_norm": 0.65234375,
      "learning_rate": 2.0689293963800785e-05,
      "loss": 0.8015,
      "step": 5173
    },
    {
      "epoch": 2.7779604711251302,
      "grad_norm": 0.62109375,
      "learning_rate": 2.0680975559009713e-05,
      "loss": 0.784,
      "step": 5174
    },
    {
      "epoch": 2.778497365860206,
      "grad_norm": 0.8984375,
      "learning_rate": 2.067265764707403e-05,
      "loss": 0.9371,
      "step": 5175
    },
    {
      "epoch": 2.779034260595282,
      "grad_norm": 0.65234375,
      "learning_rate": 2.0664340228942913e-05,
      "loss": 0.6046,
      "step": 5176
    },
    {
      "epoch": 2.779571155330358,
      "grad_norm": 0.69140625,
      "learning_rate": 2.0656023305565485e-05,
      "loss": 0.9095,
      "step": 5177
    },
    {
      "epoch": 2.780108050065434,
      "grad_norm": 0.82421875,
      "learning_rate": 2.064770687789081e-05,
      "loss": 0.7477,
      "step": 5178
    },
    {
      "epoch": 2.78064494480051,
      "grad_norm": 0.625,
      "learning_rate": 2.0639390946867886e-05,
      "loss": 0.8375,
      "step": 5179
    },
    {
      "epoch": 2.781181839535586,
      "grad_norm": 0.6328125,
      "learning_rate": 2.0631075513445685e-05,
      "loss": 0.862,
      "step": 5180
    },
    {
      "epoch": 2.781718734270662,
      "grad_norm": 0.7421875,
      "learning_rate": 2.0622760578573097e-05,
      "loss": 1.1194,
      "step": 5181
    },
    {
      "epoch": 2.7822556290057383,
      "grad_norm": 0.62890625,
      "learning_rate": 2.0614446143198956e-05,
      "loss": 0.749,
      "step": 5182
    },
    {
      "epoch": 2.782792523740814,
      "grad_norm": 0.5703125,
      "learning_rate": 2.0606132208272052e-05,
      "loss": 0.7441,
      "step": 5183
    },
    {
      "epoch": 2.78332941847589,
      "grad_norm": 0.984375,
      "learning_rate": 2.0597818774741108e-05,
      "loss": 0.9655,
      "step": 5184
    },
    {
      "epoch": 2.7838663132109662,
      "grad_norm": 0.8359375,
      "learning_rate": 2.0589505843554797e-05,
      "loss": 0.8296,
      "step": 5185
    },
    {
      "epoch": 2.784403207946042,
      "grad_norm": 0.7734375,
      "learning_rate": 2.0581193415661716e-05,
      "loss": 0.8359,
      "step": 5186
    },
    {
      "epoch": 2.784940102681118,
      "grad_norm": 0.6953125,
      "learning_rate": 2.057288149201042e-05,
      "loss": 0.841,
      "step": 5187
    },
    {
      "epoch": 2.785476997416194,
      "grad_norm": 0.7421875,
      "learning_rate": 2.0564570073549417e-05,
      "loss": 0.7948,
      "step": 5188
    },
    {
      "epoch": 2.78601389215127,
      "grad_norm": 0.58984375,
      "learning_rate": 2.055625916122714e-05,
      "loss": 0.6271,
      "step": 5189
    },
    {
      "epoch": 2.786550786886346,
      "grad_norm": 0.6484375,
      "learning_rate": 2.0547948755991967e-05,
      "loss": 0.7313,
      "step": 5190
    },
    {
      "epoch": 2.787087681621422,
      "grad_norm": 0.91015625,
      "learning_rate": 2.0539638858792214e-05,
      "loss": 0.9432,
      "step": 5191
    },
    {
      "epoch": 2.787624576356498,
      "grad_norm": 0.71875,
      "learning_rate": 2.0531329470576164e-05,
      "loss": 0.8472,
      "step": 5192
    },
    {
      "epoch": 2.7881614710915743,
      "grad_norm": 0.6640625,
      "learning_rate": 2.0523020592292015e-05,
      "loss": 0.764,
      "step": 5193
    },
    {
      "epoch": 2.78869836582665,
      "grad_norm": 0.71875,
      "learning_rate": 2.0514712224887906e-05,
      "loss": 0.8106,
      "step": 5194
    },
    {
      "epoch": 2.789235260561726,
      "grad_norm": 0.73046875,
      "learning_rate": 2.0506404369311937e-05,
      "loss": 0.9758,
      "step": 5195
    },
    {
      "epoch": 2.7897721552968022,
      "grad_norm": 0.67578125,
      "learning_rate": 2.049809702651213e-05,
      "loss": 0.9376,
      "step": 5196
    },
    {
      "epoch": 2.790309050031878,
      "grad_norm": 0.6796875,
      "learning_rate": 2.0489790197436466e-05,
      "loss": 0.7836,
      "step": 5197
    },
    {
      "epoch": 2.790845944766954,
      "grad_norm": 0.625,
      "learning_rate": 2.0481483883032858e-05,
      "loss": 0.7543,
      "step": 5198
    },
    {
      "epoch": 2.79138283950203,
      "grad_norm": 0.734375,
      "learning_rate": 2.047317808424916e-05,
      "loss": 0.9031,
      "step": 5199
    },
    {
      "epoch": 2.7919197342371063,
      "grad_norm": 0.75,
      "learning_rate": 2.0464872802033175e-05,
      "loss": 0.7797,
      "step": 5200
    },
    {
      "epoch": 2.7919197342371063,
      "eval_loss": 0.9027925729751587,
      "eval_runtime": 46.0281,
      "eval_samples_per_second": 20.9,
      "eval_steps_per_second": 20.9,
      "step": 5200
    },
    {
      "epoch": 2.7924566289721824,
      "grad_norm": 0.65234375,
      "learning_rate": 2.0456568037332634e-05,
      "loss": 0.8983,
      "step": 5201
    },
    {
      "epoch": 2.792993523707258,
      "grad_norm": 0.67578125,
      "learning_rate": 2.0448263791095224e-05,
      "loss": 1.1263,
      "step": 5202
    },
    {
      "epoch": 2.793530418442334,
      "grad_norm": 0.70703125,
      "learning_rate": 2.0439960064268555e-05,
      "loss": 0.7796,
      "step": 5203
    },
    {
      "epoch": 2.7940673131774103,
      "grad_norm": 1.4765625,
      "learning_rate": 2.043165685780018e-05,
      "loss": 0.8716,
      "step": 5204
    },
    {
      "epoch": 2.794604207912486,
      "grad_norm": 0.984375,
      "learning_rate": 2.042335417263762e-05,
      "loss": 0.8917,
      "step": 5205
    },
    {
      "epoch": 2.795141102647562,
      "grad_norm": 0.6171875,
      "learning_rate": 2.0415052009728308e-05,
      "loss": 0.7989,
      "step": 5206
    },
    {
      "epoch": 2.7956779973826382,
      "grad_norm": 0.6484375,
      "learning_rate": 2.040675037001962e-05,
      "loss": 0.7786,
      "step": 5207
    },
    {
      "epoch": 2.796214892117714,
      "grad_norm": 0.69921875,
      "learning_rate": 2.0398449254458887e-05,
      "loss": 1.0738,
      "step": 5208
    },
    {
      "epoch": 2.79675178685279,
      "grad_norm": 0.61328125,
      "learning_rate": 2.039014866399337e-05,
      "loss": 0.7783,
      "step": 5209
    },
    {
      "epoch": 2.797288681587866,
      "grad_norm": 0.66015625,
      "learning_rate": 2.0381848599570276e-05,
      "loss": 0.8454,
      "step": 5210
    },
    {
      "epoch": 2.7978255763229423,
      "grad_norm": 0.828125,
      "learning_rate": 2.0373549062136732e-05,
      "loss": 0.9527,
      "step": 5211
    },
    {
      "epoch": 2.7983624710580184,
      "grad_norm": 0.69921875,
      "learning_rate": 2.0365250052639824e-05,
      "loss": 0.8622,
      "step": 5212
    },
    {
      "epoch": 2.798899365793094,
      "grad_norm": 0.7578125,
      "learning_rate": 2.0356951572026585e-05,
      "loss": 0.7361,
      "step": 5213
    },
    {
      "epoch": 2.79943626052817,
      "grad_norm": 0.6640625,
      "learning_rate": 2.0348653621243974e-05,
      "loss": 0.6588,
      "step": 5214
    },
    {
      "epoch": 2.7999731552632463,
      "grad_norm": 0.67578125,
      "learning_rate": 2.0340356201238883e-05,
      "loss": 0.8215,
      "step": 5215
    },
    {
      "epoch": 2.800510049998322,
      "grad_norm": 0.83203125,
      "learning_rate": 2.033205931295816e-05,
      "loss": 0.7829,
      "step": 5216
    },
    {
      "epoch": 2.801046944733398,
      "grad_norm": 0.7109375,
      "learning_rate": 2.032376295734859e-05,
      "loss": 0.7765,
      "step": 5217
    },
    {
      "epoch": 2.8015838394684742,
      "grad_norm": 0.62890625,
      "learning_rate": 2.031546713535688e-05,
      "loss": 0.7968,
      "step": 5218
    },
    {
      "epoch": 2.8021207342035503,
      "grad_norm": 0.671875,
      "learning_rate": 2.030717184792971e-05,
      "loss": 0.8357,
      "step": 5219
    },
    {
      "epoch": 2.8026576289386265,
      "grad_norm": 0.63671875,
      "learning_rate": 2.0298877096013648e-05,
      "loss": 0.9425,
      "step": 5220
    },
    {
      "epoch": 2.803194523673702,
      "grad_norm": 0.77734375,
      "learning_rate": 2.0290582880555244e-05,
      "loss": 0.8235,
      "step": 5221
    },
    {
      "epoch": 2.8037314184087783,
      "grad_norm": 0.80859375,
      "learning_rate": 2.0282289202500977e-05,
      "loss": 1.013,
      "step": 5222
    },
    {
      "epoch": 2.8042683131438544,
      "grad_norm": 0.703125,
      "learning_rate": 2.0273996062797252e-05,
      "loss": 0.7613,
      "step": 5223
    },
    {
      "epoch": 2.80480520787893,
      "grad_norm": 0.63671875,
      "learning_rate": 2.0265703462390427e-05,
      "loss": 0.9224,
      "step": 5224
    },
    {
      "epoch": 2.805342102614006,
      "grad_norm": 0.7421875,
      "learning_rate": 2.02574114022268e-05,
      "loss": 0.8286,
      "step": 5225
    },
    {
      "epoch": 2.8058789973490823,
      "grad_norm": 0.73046875,
      "learning_rate": 2.0249119883252585e-05,
      "loss": 0.9074,
      "step": 5226
    },
    {
      "epoch": 2.806415892084158,
      "grad_norm": 0.6953125,
      "learning_rate": 2.0240828906413963e-05,
      "loss": 1.018,
      "step": 5227
    },
    {
      "epoch": 2.806952786819234,
      "grad_norm": 0.71484375,
      "learning_rate": 2.0232538472657026e-05,
      "loss": 0.7941,
      "step": 5228
    },
    {
      "epoch": 2.8074896815543102,
      "grad_norm": 0.6796875,
      "learning_rate": 2.0224248582927827e-05,
      "loss": 0.6987,
      "step": 5229
    },
    {
      "epoch": 2.8080265762893863,
      "grad_norm": 0.61328125,
      "learning_rate": 2.0215959238172345e-05,
      "loss": 0.8979,
      "step": 5230
    },
    {
      "epoch": 2.8085634710244625,
      "grad_norm": 0.75,
      "learning_rate": 2.0207670439336495e-05,
      "loss": 0.7204,
      "step": 5231
    },
    {
      "epoch": 2.809100365759538,
      "grad_norm": 0.66015625,
      "learning_rate": 2.019938218736614e-05,
      "loss": 0.9128,
      "step": 5232
    },
    {
      "epoch": 2.8096372604946143,
      "grad_norm": 0.828125,
      "learning_rate": 2.0191094483207073e-05,
      "loss": 0.8289,
      "step": 5233
    },
    {
      "epoch": 2.8101741552296904,
      "grad_norm": 0.68359375,
      "learning_rate": 2.0182807327805025e-05,
      "loss": 0.9231,
      "step": 5234
    },
    {
      "epoch": 2.810711049964766,
      "grad_norm": 0.73828125,
      "learning_rate": 2.0174520722105673e-05,
      "loss": 0.9004,
      "step": 5235
    },
    {
      "epoch": 2.811247944699842,
      "grad_norm": 0.80078125,
      "learning_rate": 2.016623466705461e-05,
      "loss": 0.8392,
      "step": 5236
    },
    {
      "epoch": 2.8117848394349183,
      "grad_norm": 0.7890625,
      "learning_rate": 2.015794916359738e-05,
      "loss": 0.9063,
      "step": 5237
    },
    {
      "epoch": 2.8123217341699944,
      "grad_norm": 0.77734375,
      "learning_rate": 2.0149664212679476e-05,
      "loss": 0.8567,
      "step": 5238
    },
    {
      "epoch": 2.8128586289050705,
      "grad_norm": 0.703125,
      "learning_rate": 2.014137981524631e-05,
      "loss": 0.7895,
      "step": 5239
    },
    {
      "epoch": 2.8133955236401462,
      "grad_norm": 0.83984375,
      "learning_rate": 2.0133095972243233e-05,
      "loss": 1.0232,
      "step": 5240
    },
    {
      "epoch": 2.8139324183752223,
      "grad_norm": 0.65234375,
      "learning_rate": 2.0124812684615542e-05,
      "loss": 0.8287,
      "step": 5241
    },
    {
      "epoch": 2.8144693131102985,
      "grad_norm": 0.6796875,
      "learning_rate": 2.0116529953308466e-05,
      "loss": 0.8465,
      "step": 5242
    },
    {
      "epoch": 2.815006207845374,
      "grad_norm": 0.6328125,
      "learning_rate": 2.0108247779267157e-05,
      "loss": 0.9057,
      "step": 5243
    },
    {
      "epoch": 2.8155431025804503,
      "grad_norm": 0.828125,
      "learning_rate": 2.0099966163436746e-05,
      "loss": 0.7271,
      "step": 5244
    },
    {
      "epoch": 2.8160799973155264,
      "grad_norm": 0.6640625,
      "learning_rate": 2.009168510676223e-05,
      "loss": 0.8052,
      "step": 5245
    },
    {
      "epoch": 2.816616892050602,
      "grad_norm": 0.671875,
      "learning_rate": 2.0083404610188612e-05,
      "loss": 0.7657,
      "step": 5246
    },
    {
      "epoch": 2.817153786785678,
      "grad_norm": 0.765625,
      "learning_rate": 2.007512467466079e-05,
      "loss": 0.8261,
      "step": 5247
    },
    {
      "epoch": 2.8176906815207543,
      "grad_norm": 0.67578125,
      "learning_rate": 2.0066845301123603e-05,
      "loss": 0.76,
      "step": 5248
    },
    {
      "epoch": 2.8182275762558304,
      "grad_norm": 0.6171875,
      "learning_rate": 2.0058566490521847e-05,
      "loss": 0.8264,
      "step": 5249
    },
    {
      "epoch": 2.8187644709909065,
      "grad_norm": 0.64453125,
      "learning_rate": 2.0050288243800233e-05,
      "loss": 0.7537,
      "step": 5250
    },
    {
      "epoch": 2.8187644709909065,
      "eval_loss": 0.9029639959335327,
      "eval_runtime": 46.1523,
      "eval_samples_per_second": 20.844,
      "eval_steps_per_second": 20.844,
      "step": 5250
    },
    {
      "epoch": 2.8193013657259822,
      "grad_norm": 0.82421875,
      "learning_rate": 2.0042010561903408e-05,
      "loss": 0.9437,
      "step": 5251
    },
    {
      "epoch": 2.8198382604610583,
      "grad_norm": 0.74609375,
      "learning_rate": 2.0033733445775982e-05,
      "loss": 0.8539,
      "step": 5252
    },
    {
      "epoch": 2.8203751551961345,
      "grad_norm": 0.7265625,
      "learning_rate": 2.0025456896362445e-05,
      "loss": 0.7846,
      "step": 5253
    },
    {
      "epoch": 2.82091204993121,
      "grad_norm": 0.59375,
      "learning_rate": 2.001718091460728e-05,
      "loss": 0.6701,
      "step": 5254
    },
    {
      "epoch": 2.8214489446662863,
      "grad_norm": 0.56640625,
      "learning_rate": 2.0008905501454877e-05,
      "loss": 0.6474,
      "step": 5255
    },
    {
      "epoch": 2.8219858394013624,
      "grad_norm": 1.140625,
      "learning_rate": 2.0000630657849556e-05,
      "loss": 0.765,
      "step": 5256
    },
    {
      "epoch": 2.8225227341364385,
      "grad_norm": 0.6484375,
      "learning_rate": 1.9992356384735593e-05,
      "loss": 0.6924,
      "step": 5257
    },
    {
      "epoch": 2.8230596288715146,
      "grad_norm": 0.625,
      "learning_rate": 1.9984082683057186e-05,
      "loss": 0.7839,
      "step": 5258
    },
    {
      "epoch": 2.8235965236065903,
      "grad_norm": 0.57421875,
      "learning_rate": 1.9975809553758463e-05,
      "loss": 0.8914,
      "step": 5259
    },
    {
      "epoch": 2.8241334183416664,
      "grad_norm": 0.77734375,
      "learning_rate": 1.9967536997783494e-05,
      "loss": 0.9146,
      "step": 5260
    },
    {
      "epoch": 2.8246703130767425,
      "grad_norm": 0.828125,
      "learning_rate": 1.9959265016076294e-05,
      "loss": 0.9914,
      "step": 5261
    },
    {
      "epoch": 2.825207207811818,
      "grad_norm": 1.015625,
      "learning_rate": 1.9950993609580786e-05,
      "loss": 0.9179,
      "step": 5262
    },
    {
      "epoch": 2.8257441025468943,
      "grad_norm": 0.640625,
      "learning_rate": 1.994272277924085e-05,
      "loss": 0.7413,
      "step": 5263
    },
    {
      "epoch": 2.8262809972819705,
      "grad_norm": 0.67578125,
      "learning_rate": 1.9934452526000283e-05,
      "loss": 0.8451,
      "step": 5264
    },
    {
      "epoch": 2.826817892017046,
      "grad_norm": 0.78125,
      "learning_rate": 1.9926182850802837e-05,
      "loss": 0.8916,
      "step": 5265
    },
    {
      "epoch": 2.8273547867521223,
      "grad_norm": 0.734375,
      "learning_rate": 1.9917913754592183e-05,
      "loss": 0.8789,
      "step": 5266
    },
    {
      "epoch": 2.8278916814871984,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9909645238311923e-05,
      "loss": 1.0534,
      "step": 5267
    },
    {
      "epoch": 2.8284285762222745,
      "grad_norm": 0.77734375,
      "learning_rate": 1.9901377302905612e-05,
      "loss": 0.9451,
      "step": 5268
    },
    {
      "epoch": 2.8289654709573506,
      "grad_norm": 0.640625,
      "learning_rate": 1.9893109949316725e-05,
      "loss": 0.7487,
      "step": 5269
    },
    {
      "epoch": 2.8295023656924263,
      "grad_norm": 0.79296875,
      "learning_rate": 1.988484317848866e-05,
      "loss": 0.9931,
      "step": 5270
    },
    {
      "epoch": 2.8300392604275024,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9876576991364766e-05,
      "loss": 1.1299,
      "step": 5271
    },
    {
      "epoch": 2.8305761551625785,
      "grad_norm": 0.76171875,
      "learning_rate": 1.9868311388888318e-05,
      "loss": 1.0833,
      "step": 5272
    },
    {
      "epoch": 2.831113049897654,
      "grad_norm": 0.71484375,
      "learning_rate": 1.986004637200253e-05,
      "loss": 0.7378,
      "step": 5273
    },
    {
      "epoch": 2.8316499446327303,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9851781941650542e-05,
      "loss": 0.9345,
      "step": 5274
    },
    {
      "epoch": 2.8321868393678065,
      "grad_norm": 0.62890625,
      "learning_rate": 1.984351809877544e-05,
      "loss": 0.8584,
      "step": 5275
    },
    {
      "epoch": 2.8327237341028826,
      "grad_norm": 0.76171875,
      "learning_rate": 1.983525484432021e-05,
      "loss": 0.8483,
      "step": 5276
    },
    {
      "epoch": 2.8332606288379587,
      "grad_norm": 0.6796875,
      "learning_rate": 1.982699217922782e-05,
      "loss": 0.8624,
      "step": 5277
    },
    {
      "epoch": 2.8337975235730344,
      "grad_norm": 0.765625,
      "learning_rate": 1.981873010444113e-05,
      "loss": 0.8226,
      "step": 5278
    },
    {
      "epoch": 2.8343344183081105,
      "grad_norm": 0.65234375,
      "learning_rate": 1.9810468620902952e-05,
      "loss": 0.7661,
      "step": 5279
    },
    {
      "epoch": 2.8348713130431866,
      "grad_norm": 0.65234375,
      "learning_rate": 1.980220772955602e-05,
      "loss": 0.8675,
      "step": 5280
    },
    {
      "epoch": 2.8354082077782623,
      "grad_norm": 0.81640625,
      "learning_rate": 1.9793947431343013e-05,
      "loss": 0.9995,
      "step": 5281
    },
    {
      "epoch": 2.8359451025133384,
      "grad_norm": 0.828125,
      "learning_rate": 1.978568772720653e-05,
      "loss": 0.9801,
      "step": 5282
    },
    {
      "epoch": 2.8364819972484145,
      "grad_norm": 0.7421875,
      "learning_rate": 1.9777428618089118e-05,
      "loss": 0.8755,
      "step": 5283
    },
    {
      "epoch": 2.83701889198349,
      "grad_norm": 0.5625,
      "learning_rate": 1.976917010493323e-05,
      "loss": 0.6681,
      "step": 5284
    },
    {
      "epoch": 2.8375557867185663,
      "grad_norm": 0.5703125,
      "learning_rate": 1.9760912188681284e-05,
      "loss": 0.7422,
      "step": 5285
    },
    {
      "epoch": 2.8380926814536425,
      "grad_norm": 0.6328125,
      "learning_rate": 1.9752654870275608e-05,
      "loss": 0.8308,
      "step": 5286
    },
    {
      "epoch": 2.8386295761887186,
      "grad_norm": 0.765625,
      "learning_rate": 1.9744398150658462e-05,
      "loss": 0.9757,
      "step": 5287
    },
    {
      "epoch": 2.8391664709237947,
      "grad_norm": 0.68359375,
      "learning_rate": 1.973614203077204e-05,
      "loss": 0.7586,
      "step": 5288
    },
    {
      "epoch": 2.8397033656588704,
      "grad_norm": 0.671875,
      "learning_rate": 1.972788651155847e-05,
      "loss": 0.6954,
      "step": 5289
    },
    {
      "epoch": 2.8402402603939465,
      "grad_norm": 0.640625,
      "learning_rate": 1.9719631593959816e-05,
      "loss": 0.869,
      "step": 5290
    },
    {
      "epoch": 2.8407771551290226,
      "grad_norm": 0.62109375,
      "learning_rate": 1.9711377278918073e-05,
      "loss": 0.8526,
      "step": 5291
    },
    {
      "epoch": 2.8413140498640983,
      "grad_norm": 0.73046875,
      "learning_rate": 1.970312356737515e-05,
      "loss": 0.7922,
      "step": 5292
    },
    {
      "epoch": 2.8418509445991744,
      "grad_norm": 0.6328125,
      "learning_rate": 1.9694870460272914e-05,
      "loss": 0.831,
      "step": 5293
    },
    {
      "epoch": 2.8423878393342505,
      "grad_norm": 0.75390625,
      "learning_rate": 1.9686617958553148e-05,
      "loss": 0.9719,
      "step": 5294
    },
    {
      "epoch": 2.8429247340693267,
      "grad_norm": 0.8125,
      "learning_rate": 1.9678366063157556e-05,
      "loss": 0.9645,
      "step": 5295
    },
    {
      "epoch": 2.843461628804403,
      "grad_norm": 0.7578125,
      "learning_rate": 1.9670114775027786e-05,
      "loss": 0.9471,
      "step": 5296
    },
    {
      "epoch": 2.8439985235394785,
      "grad_norm": 0.9921875,
      "learning_rate": 1.9661864095105415e-05,
      "loss": 0.8779,
      "step": 5297
    },
    {
      "epoch": 2.8445354182745546,
      "grad_norm": 0.64453125,
      "learning_rate": 1.9653614024331956e-05,
      "loss": 0.845,
      "step": 5298
    },
    {
      "epoch": 2.8450723130096307,
      "grad_norm": 0.625,
      "learning_rate": 1.9645364563648847e-05,
      "loss": 0.7228,
      "step": 5299
    },
    {
      "epoch": 2.8456092077447064,
      "grad_norm": 0.57421875,
      "learning_rate": 1.9637115713997444e-05,
      "loss": 0.7131,
      "step": 5300
    },
    {
      "epoch": 2.8456092077447064,
      "eval_loss": 0.9029813408851624,
      "eval_runtime": 45.6367,
      "eval_samples_per_second": 21.08,
      "eval_steps_per_second": 21.08,
      "step": 5300
    },
    {
      "epoch": 2.8461461024797825,
      "grad_norm": 0.78515625,
      "learning_rate": 1.962886747631906e-05,
      "loss": 0.8761,
      "step": 5301
    },
    {
      "epoch": 2.8466829972148586,
      "grad_norm": 0.75,
      "learning_rate": 1.9620619851554916e-05,
      "loss": 0.8243,
      "step": 5302
    },
    {
      "epoch": 2.8472198919499343,
      "grad_norm": 0.7265625,
      "learning_rate": 1.961237284064618e-05,
      "loss": 0.8589,
      "step": 5303
    },
    {
      "epoch": 2.8477567866850104,
      "grad_norm": 0.66015625,
      "learning_rate": 1.960412644453392e-05,
      "loss": 0.8168,
      "step": 5304
    },
    {
      "epoch": 2.8482936814200865,
      "grad_norm": 0.76953125,
      "learning_rate": 1.9595880664159166e-05,
      "loss": 0.9777,
      "step": 5305
    },
    {
      "epoch": 2.8488305761551627,
      "grad_norm": 0.5625,
      "learning_rate": 1.958763550046287e-05,
      "loss": 0.6433,
      "step": 5306
    },
    {
      "epoch": 2.849367470890239,
      "grad_norm": 0.57421875,
      "learning_rate": 1.9579390954385907e-05,
      "loss": 0.6728,
      "step": 5307
    },
    {
      "epoch": 2.8499043656253145,
      "grad_norm": 0.6328125,
      "learning_rate": 1.9571147026869076e-05,
      "loss": 1.0079,
      "step": 5308
    },
    {
      "epoch": 2.8504412603603906,
      "grad_norm": 0.66796875,
      "learning_rate": 1.9562903718853122e-05,
      "loss": 0.8398,
      "step": 5309
    },
    {
      "epoch": 2.8509781550954667,
      "grad_norm": 0.9375,
      "learning_rate": 1.9554661031278712e-05,
      "loss": 0.9488,
      "step": 5310
    },
    {
      "epoch": 2.8515150498305424,
      "grad_norm": 0.5546875,
      "learning_rate": 1.9546418965086442e-05,
      "loss": 0.7689,
      "step": 5311
    },
    {
      "epoch": 2.8520519445656185,
      "grad_norm": 0.69921875,
      "learning_rate": 1.9538177521216827e-05,
      "loss": 1.0181,
      "step": 5312
    },
    {
      "epoch": 2.8525888393006946,
      "grad_norm": 0.69921875,
      "learning_rate": 1.952993670061032e-05,
      "loss": 0.9337,
      "step": 5313
    },
    {
      "epoch": 2.8531257340357707,
      "grad_norm": 0.62109375,
      "learning_rate": 1.9521696504207315e-05,
      "loss": 0.7985,
      "step": 5314
    },
    {
      "epoch": 2.853662628770847,
      "grad_norm": 0.70703125,
      "learning_rate": 1.9513456932948114e-05,
      "loss": 0.7185,
      "step": 5315
    },
    {
      "epoch": 2.8541995235059225,
      "grad_norm": 0.640625,
      "learning_rate": 1.950521798777296e-05,
      "loss": 0.9152,
      "step": 5316
    },
    {
      "epoch": 2.8547364182409987,
      "grad_norm": 0.6875,
      "learning_rate": 1.949697966962201e-05,
      "loss": 0.8969,
      "step": 5317
    },
    {
      "epoch": 2.855273312976075,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9488741979435382e-05,
      "loss": 0.8396,
      "step": 5318
    },
    {
      "epoch": 2.8558102077111505,
      "grad_norm": 0.67578125,
      "learning_rate": 1.948050491815309e-05,
      "loss": 0.797,
      "step": 5319
    },
    {
      "epoch": 2.8563471024462266,
      "grad_norm": 0.6640625,
      "learning_rate": 1.947226848671508e-05,
      "loss": 0.858,
      "step": 5320
    },
    {
      "epoch": 2.8568839971813027,
      "grad_norm": 0.75390625,
      "learning_rate": 1.946403268606124e-05,
      "loss": 0.9204,
      "step": 5321
    },
    {
      "epoch": 2.8574208919163784,
      "grad_norm": 0.6875,
      "learning_rate": 1.945579751713137e-05,
      "loss": 0.8311,
      "step": 5322
    },
    {
      "epoch": 2.8579577866514545,
      "grad_norm": 0.61328125,
      "learning_rate": 1.9447562980865226e-05,
      "loss": 0.717,
      "step": 5323
    },
    {
      "epoch": 2.8584946813865306,
      "grad_norm": 0.7578125,
      "learning_rate": 1.9439329078202457e-05,
      "loss": 0.9078,
      "step": 5324
    },
    {
      "epoch": 2.8590315761216067,
      "grad_norm": 0.8046875,
      "learning_rate": 1.9431095810082657e-05,
      "loss": 0.9643,
      "step": 5325
    },
    {
      "epoch": 2.859568470856683,
      "grad_norm": 0.71875,
      "learning_rate": 1.942286317744535e-05,
      "loss": 0.9782,
      "step": 5326
    },
    {
      "epoch": 2.8601053655917585,
      "grad_norm": 0.70703125,
      "learning_rate": 1.941463118122999e-05,
      "loss": 0.984,
      "step": 5327
    },
    {
      "epoch": 2.8606422603268347,
      "grad_norm": 0.66015625,
      "learning_rate": 1.940639982237595e-05,
      "loss": 0.8385,
      "step": 5328
    },
    {
      "epoch": 2.861179155061911,
      "grad_norm": 0.65625,
      "learning_rate": 1.9398169101822517e-05,
      "loss": 0.9262,
      "step": 5329
    },
    {
      "epoch": 2.8617160497969865,
      "grad_norm": 0.6640625,
      "learning_rate": 1.938993902050893e-05,
      "loss": 0.9427,
      "step": 5330
    },
    {
      "epoch": 2.8622529445320626,
      "grad_norm": 0.671875,
      "learning_rate": 1.938170957937435e-05,
      "loss": 0.7477,
      "step": 5331
    },
    {
      "epoch": 2.8627898392671387,
      "grad_norm": 0.6953125,
      "learning_rate": 1.9373480779357856e-05,
      "loss": 0.8386,
      "step": 5332
    },
    {
      "epoch": 2.863326734002215,
      "grad_norm": 0.63671875,
      "learning_rate": 1.9365252621398455e-05,
      "loss": 0.8151,
      "step": 5333
    },
    {
      "epoch": 2.863863628737291,
      "grad_norm": 0.64453125,
      "learning_rate": 1.9357025106435093e-05,
      "loss": 0.8419,
      "step": 5334
    },
    {
      "epoch": 2.8644005234723666,
      "grad_norm": 0.65625,
      "learning_rate": 1.934879823540663e-05,
      "loss": 0.7782,
      "step": 5335
    },
    {
      "epoch": 2.8649374182074427,
      "grad_norm": 0.66015625,
      "learning_rate": 1.934057200925186e-05,
      "loss": 0.5859,
      "step": 5336
    },
    {
      "epoch": 2.865474312942519,
      "grad_norm": 0.7109375,
      "learning_rate": 1.9332346428909485e-05,
      "loss": 0.8881,
      "step": 5337
    },
    {
      "epoch": 2.8660112076775945,
      "grad_norm": 0.72265625,
      "learning_rate": 1.932412149531816e-05,
      "loss": 0.9106,
      "step": 5338
    },
    {
      "epoch": 2.8665481024126707,
      "grad_norm": 0.74609375,
      "learning_rate": 1.9315897209416453e-05,
      "loss": 0.9241,
      "step": 5339
    },
    {
      "epoch": 2.867084997147747,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9307673572142856e-05,
      "loss": 0.6796,
      "step": 5340
    },
    {
      "epoch": 2.8676218918828225,
      "grad_norm": 0.80078125,
      "learning_rate": 1.9299450584435787e-05,
      "loss": 1.0987,
      "step": 5341
    },
    {
      "epoch": 2.8681587866178986,
      "grad_norm": 0.76171875,
      "learning_rate": 1.9291228247233605e-05,
      "loss": 0.8761,
      "step": 5342
    },
    {
      "epoch": 2.8686956813529747,
      "grad_norm": 0.69140625,
      "learning_rate": 1.9283006561474574e-05,
      "loss": 0.8641,
      "step": 5343
    },
    {
      "epoch": 2.869232576088051,
      "grad_norm": 0.6640625,
      "learning_rate": 1.9274785528096893e-05,
      "loss": 0.8586,
      "step": 5344
    },
    {
      "epoch": 2.869769470823127,
      "grad_norm": 0.68359375,
      "learning_rate": 1.92665651480387e-05,
      "loss": 0.8617,
      "step": 5345
    },
    {
      "epoch": 2.8703063655582026,
      "grad_norm": 0.66015625,
      "learning_rate": 1.9258345422238018e-05,
      "loss": 0.7381,
      "step": 5346
    },
    {
      "epoch": 2.8708432602932787,
      "grad_norm": 0.6796875,
      "learning_rate": 1.9250126351632835e-05,
      "loss": 0.8398,
      "step": 5347
    },
    {
      "epoch": 2.871380155028355,
      "grad_norm": 0.7109375,
      "learning_rate": 1.9241907937161056e-05,
      "loss": 0.6899,
      "step": 5348
    },
    {
      "epoch": 2.8719170497634305,
      "grad_norm": 0.7421875,
      "learning_rate": 1.9233690179760498e-05,
      "loss": 0.7721,
      "step": 5349
    },
    {
      "epoch": 2.8724539444985067,
      "grad_norm": 0.7109375,
      "learning_rate": 1.9225473080368916e-05,
      "loss": 0.8321,
      "step": 5350
    },
    {
      "epoch": 2.8724539444985067,
      "eval_loss": 0.9029684662818909,
      "eval_runtime": 45.8297,
      "eval_samples_per_second": 20.991,
      "eval_steps_per_second": 20.991,
      "step": 5350
    },
    {
      "epoch": 2.872990839233583,
      "grad_norm": 0.6328125,
      "learning_rate": 1.9217256639923986e-05,
      "loss": 0.7561,
      "step": 5351
    },
    {
      "epoch": 2.873527733968659,
      "grad_norm": 0.7421875,
      "learning_rate": 1.9209040859363302e-05,
      "loss": 0.9548,
      "step": 5352
    },
    {
      "epoch": 2.874064628703735,
      "grad_norm": 0.69140625,
      "learning_rate": 1.9200825739624405e-05,
      "loss": 0.7987,
      "step": 5353
    },
    {
      "epoch": 2.8746015234388107,
      "grad_norm": 0.71875,
      "learning_rate": 1.919261128164472e-05,
      "loss": 0.8068,
      "step": 5354
    },
    {
      "epoch": 2.875138418173887,
      "grad_norm": 0.61328125,
      "learning_rate": 1.9184397486361637e-05,
      "loss": 0.7436,
      "step": 5355
    },
    {
      "epoch": 2.875675312908963,
      "grad_norm": 0.734375,
      "learning_rate": 1.9176184354712445e-05,
      "loss": 0.7345,
      "step": 5356
    },
    {
      "epoch": 2.8762122076440386,
      "grad_norm": 1.09375,
      "learning_rate": 1.916797188763437e-05,
      "loss": 0.7977,
      "step": 5357
    },
    {
      "epoch": 2.8767491023791147,
      "grad_norm": 0.7890625,
      "learning_rate": 1.915976008606456e-05,
      "loss": 0.9964,
      "step": 5358
    },
    {
      "epoch": 2.877285997114191,
      "grad_norm": 0.70703125,
      "learning_rate": 1.915154895094009e-05,
      "loss": 1.0197,
      "step": 5359
    },
    {
      "epoch": 2.8778228918492665,
      "grad_norm": 0.765625,
      "learning_rate": 1.914333848319795e-05,
      "loss": 0.8793,
      "step": 5360
    },
    {
      "epoch": 2.8783597865843427,
      "grad_norm": 0.78125,
      "learning_rate": 1.9135128683775057e-05,
      "loss": 0.9739,
      "step": 5361
    },
    {
      "epoch": 2.878896681319419,
      "grad_norm": 0.5859375,
      "learning_rate": 1.9126919553608253e-05,
      "loss": 0.8764,
      "step": 5362
    },
    {
      "epoch": 2.879433576054495,
      "grad_norm": 0.63671875,
      "learning_rate": 1.91187110936343e-05,
      "loss": 0.8114,
      "step": 5363
    },
    {
      "epoch": 2.879970470789571,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9110503304789892e-05,
      "loss": 0.9944,
      "step": 5364
    },
    {
      "epoch": 2.8805073655246467,
      "grad_norm": 0.6484375,
      "learning_rate": 1.9102296188011645e-05,
      "loss": 0.6921,
      "step": 5365
    },
    {
      "epoch": 2.881044260259723,
      "grad_norm": 0.765625,
      "learning_rate": 1.9094089744236084e-05,
      "loss": 0.8673,
      "step": 5366
    },
    {
      "epoch": 2.881581154994799,
      "grad_norm": 0.67578125,
      "learning_rate": 1.9085883974399682e-05,
      "loss": 0.9318,
      "step": 5367
    },
    {
      "epoch": 2.8821180497298746,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9077678879438814e-05,
      "loss": 0.9419,
      "step": 5368
    },
    {
      "epoch": 2.8826549444649507,
      "grad_norm": 0.65625,
      "learning_rate": 1.9069474460289782e-05,
      "loss": 0.7571,
      "step": 5369
    },
    {
      "epoch": 2.883191839200027,
      "grad_norm": 0.62890625,
      "learning_rate": 1.9061270717888834e-05,
      "loss": 0.6775,
      "step": 5370
    },
    {
      "epoch": 2.883728733935103,
      "grad_norm": 1.078125,
      "learning_rate": 1.9053067653172092e-05,
      "loss": 0.6983,
      "step": 5371
    },
    {
      "epoch": 2.884265628670179,
      "grad_norm": 0.63671875,
      "learning_rate": 1.9044865267075643e-05,
      "loss": 0.8665,
      "step": 5372
    },
    {
      "epoch": 2.884802523405255,
      "grad_norm": 0.74609375,
      "learning_rate": 1.9036663560535483e-05,
      "loss": 0.8226,
      "step": 5373
    },
    {
      "epoch": 2.885339418140331,
      "grad_norm": 0.6015625,
      "learning_rate": 1.9028462534487526e-05,
      "loss": 0.7879,
      "step": 5374
    },
    {
      "epoch": 2.885876312875407,
      "grad_norm": 0.64453125,
      "learning_rate": 1.902026218986762e-05,
      "loss": 0.8318,
      "step": 5375
    },
    {
      "epoch": 2.8864132076104827,
      "grad_norm": 0.77734375,
      "learning_rate": 1.901206252761153e-05,
      "loss": 1.0547,
      "step": 5376
    },
    {
      "epoch": 2.886950102345559,
      "grad_norm": 0.64453125,
      "learning_rate": 1.900386354865493e-05,
      "loss": 0.7827,
      "step": 5377
    },
    {
      "epoch": 2.887486997080635,
      "grad_norm": 0.91015625,
      "learning_rate": 1.899566525393345e-05,
      "loss": 0.9078,
      "step": 5378
    },
    {
      "epoch": 2.8880238918157106,
      "grad_norm": 0.6875,
      "learning_rate": 1.8987467644382585e-05,
      "loss": 0.8543,
      "step": 5379
    },
    {
      "epoch": 2.8885607865507867,
      "grad_norm": 0.734375,
      "learning_rate": 1.8979270720937813e-05,
      "loss": 1.018,
      "step": 5380
    },
    {
      "epoch": 2.889097681285863,
      "grad_norm": 0.72265625,
      "learning_rate": 1.8971074484534495e-05,
      "loss": 0.9453,
      "step": 5381
    },
    {
      "epoch": 2.889634576020939,
      "grad_norm": 0.71875,
      "learning_rate": 1.8962878936107926e-05,
      "loss": 0.7863,
      "step": 5382
    },
    {
      "epoch": 2.890171470756015,
      "grad_norm": 0.63671875,
      "learning_rate": 1.895468407659333e-05,
      "loss": 0.7151,
      "step": 5383
    },
    {
      "epoch": 2.890708365491091,
      "grad_norm": 0.66796875,
      "learning_rate": 1.8946489906925836e-05,
      "loss": 0.8228,
      "step": 5384
    },
    {
      "epoch": 2.891245260226167,
      "grad_norm": 0.7734375,
      "learning_rate": 1.89382964280405e-05,
      "loss": 0.6157,
      "step": 5385
    },
    {
      "epoch": 2.891782154961243,
      "grad_norm": 0.70703125,
      "learning_rate": 1.8930103640872313e-05,
      "loss": 0.8695,
      "step": 5386
    },
    {
      "epoch": 2.8923190496963187,
      "grad_norm": 0.62890625,
      "learning_rate": 1.8921911546356176e-05,
      "loss": 0.7433,
      "step": 5387
    },
    {
      "epoch": 2.892855944431395,
      "grad_norm": 0.6484375,
      "learning_rate": 1.8913720145426897e-05,
      "loss": 0.7066,
      "step": 5388
    },
    {
      "epoch": 2.893392839166471,
      "grad_norm": 0.72265625,
      "learning_rate": 1.8905529439019226e-05,
      "loss": 0.6999,
      "step": 5389
    },
    {
      "epoch": 2.893929733901547,
      "grad_norm": 0.75390625,
      "learning_rate": 1.8897339428067822e-05,
      "loss": 0.6704,
      "step": 5390
    },
    {
      "epoch": 2.894466628636623,
      "grad_norm": 0.69140625,
      "learning_rate": 1.888915011350728e-05,
      "loss": 0.8328,
      "step": 5391
    },
    {
      "epoch": 2.895003523371699,
      "grad_norm": 0.671875,
      "learning_rate": 1.8880961496272092e-05,
      "loss": 0.8337,
      "step": 5392
    },
    {
      "epoch": 2.895540418106775,
      "grad_norm": 0.625,
      "learning_rate": 1.8872773577296687e-05,
      "loss": 0.7285,
      "step": 5393
    },
    {
      "epoch": 2.896077312841851,
      "grad_norm": 1.1953125,
      "learning_rate": 1.886458635751542e-05,
      "loss": 0.7891,
      "step": 5394
    },
    {
      "epoch": 2.896614207576927,
      "grad_norm": 0.73828125,
      "learning_rate": 1.885639983786255e-05,
      "loss": 0.7902,
      "step": 5395
    },
    {
      "epoch": 2.897151102312003,
      "grad_norm": 0.62109375,
      "learning_rate": 1.8848214019272255e-05,
      "loss": 0.698,
      "step": 5396
    },
    {
      "epoch": 2.897687997047079,
      "grad_norm": 0.72265625,
      "learning_rate": 1.884002890267865e-05,
      "loss": 0.9232,
      "step": 5397
    },
    {
      "epoch": 2.8982248917821547,
      "grad_norm": 0.8125,
      "learning_rate": 1.8831844489015748e-05,
      "loss": 0.7353,
      "step": 5398
    },
    {
      "epoch": 2.898761786517231,
      "grad_norm": 0.6328125,
      "learning_rate": 1.8823660779217512e-05,
      "loss": 0.7497,
      "step": 5399
    },
    {
      "epoch": 2.899298681252307,
      "grad_norm": 0.76953125,
      "learning_rate": 1.8815477774217797e-05,
      "loss": 0.7554,
      "step": 5400
    },
    {
      "epoch": 2.899298681252307,
      "eval_loss": 0.9032303690910339,
      "eval_runtime": 45.9451,
      "eval_samples_per_second": 20.938,
      "eval_steps_per_second": 20.938,
      "step": 5400
    },
    {
      "epoch": 2.899835575987383,
      "grad_norm": 0.7109375,
      "learning_rate": 1.8807295474950386e-05,
      "loss": 0.8578,
      "step": 5401
    },
    {
      "epoch": 2.900372470722459,
      "grad_norm": 0.5859375,
      "learning_rate": 1.8799113882348983e-05,
      "loss": 0.6351,
      "step": 5402
    },
    {
      "epoch": 2.900909365457535,
      "grad_norm": 0.6796875,
      "learning_rate": 1.8790932997347225e-05,
      "loss": 0.8753,
      "step": 5403
    },
    {
      "epoch": 2.901446260192611,
      "grad_norm": 0.703125,
      "learning_rate": 1.8782752820878634e-05,
      "loss": 0.8204,
      "step": 5404
    },
    {
      "epoch": 2.901983154927687,
      "grad_norm": 0.63671875,
      "learning_rate": 1.8774573353876685e-05,
      "loss": 0.8633,
      "step": 5405
    },
    {
      "epoch": 2.902520049662763,
      "grad_norm": 0.6015625,
      "learning_rate": 1.8766394597274757e-05,
      "loss": 0.6699,
      "step": 5406
    },
    {
      "epoch": 2.903056944397839,
      "grad_norm": 0.65625,
      "learning_rate": 1.8758216552006137e-05,
      "loss": 0.9089,
      "step": 5407
    },
    {
      "epoch": 2.903593839132915,
      "grad_norm": 0.67578125,
      "learning_rate": 1.8750039219004063e-05,
      "loss": 0.7657,
      "step": 5408
    },
    {
      "epoch": 2.904130733867991,
      "grad_norm": 0.66015625,
      "learning_rate": 1.8741862599201658e-05,
      "loss": 0.755,
      "step": 5409
    },
    {
      "epoch": 2.9046676286030673,
      "grad_norm": 0.75,
      "learning_rate": 1.8733686693531985e-05,
      "loss": 1.0561,
      "step": 5410
    },
    {
      "epoch": 2.905204523338143,
      "grad_norm": 0.74609375,
      "learning_rate": 1.8725511502928017e-05,
      "loss": 0.8596,
      "step": 5411
    },
    {
      "epoch": 2.905741418073219,
      "grad_norm": 0.60546875,
      "learning_rate": 1.8717337028322656e-05,
      "loss": 0.7712,
      "step": 5412
    },
    {
      "epoch": 2.906278312808295,
      "grad_norm": 0.63671875,
      "learning_rate": 1.8709163270648696e-05,
      "loss": 0.7858,
      "step": 5413
    },
    {
      "epoch": 2.906815207543371,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8700990230838876e-05,
      "loss": 0.8348,
      "step": 5414
    },
    {
      "epoch": 2.907352102278447,
      "grad_norm": 0.68359375,
      "learning_rate": 1.8692817909825832e-05,
      "loss": 0.8343,
      "step": 5415
    },
    {
      "epoch": 2.907888997013523,
      "grad_norm": 0.671875,
      "learning_rate": 1.868464630854215e-05,
      "loss": 0.8145,
      "step": 5416
    },
    {
      "epoch": 2.908425891748599,
      "grad_norm": 0.65625,
      "learning_rate": 1.867647542792029e-05,
      "loss": 0.7907,
      "step": 5417
    },
    {
      "epoch": 2.908962786483675,
      "grad_norm": 0.6640625,
      "learning_rate": 1.866830526889267e-05,
      "loss": 0.8069,
      "step": 5418
    },
    {
      "epoch": 2.909499681218751,
      "grad_norm": 0.734375,
      "learning_rate": 1.8660135832391607e-05,
      "loss": 0.9163,
      "step": 5419
    },
    {
      "epoch": 2.910036575953827,
      "grad_norm": 0.64453125,
      "learning_rate": 1.8651967119349338e-05,
      "loss": 0.7605,
      "step": 5420
    },
    {
      "epoch": 2.9105734706889033,
      "grad_norm": 0.7890625,
      "learning_rate": 1.8643799130698004e-05,
      "loss": 0.8737,
      "step": 5421
    },
    {
      "epoch": 2.911110365423979,
      "grad_norm": 0.69140625,
      "learning_rate": 1.8635631867369686e-05,
      "loss": 0.6631,
      "step": 5422
    },
    {
      "epoch": 2.911647260159055,
      "grad_norm": 0.7734375,
      "learning_rate": 1.8627465330296366e-05,
      "loss": 0.7786,
      "step": 5423
    },
    {
      "epoch": 2.912184154894131,
      "grad_norm": 0.68359375,
      "learning_rate": 1.8619299520409955e-05,
      "loss": 0.6868,
      "step": 5424
    },
    {
      "epoch": 2.912721049629207,
      "grad_norm": 0.64453125,
      "learning_rate": 1.8611134438642278e-05,
      "loss": 1.0167,
      "step": 5425
    },
    {
      "epoch": 2.913257944364283,
      "grad_norm": 0.75,
      "learning_rate": 1.860297008592506e-05,
      "loss": 0.9778,
      "step": 5426
    },
    {
      "epoch": 2.913794839099359,
      "grad_norm": 0.7265625,
      "learning_rate": 1.859480646318998e-05,
      "loss": 0.833,
      "step": 5427
    },
    {
      "epoch": 2.9143317338344352,
      "grad_norm": 0.71484375,
      "learning_rate": 1.858664357136859e-05,
      "loss": 1.0274,
      "step": 5428
    },
    {
      "epoch": 2.9148686285695113,
      "grad_norm": 0.75,
      "learning_rate": 1.8578481411392396e-05,
      "loss": 0.9675,
      "step": 5429
    },
    {
      "epoch": 2.915405523304587,
      "grad_norm": 0.77734375,
      "learning_rate": 1.857031998419279e-05,
      "loss": 1.069,
      "step": 5430
    },
    {
      "epoch": 2.915942418039663,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8562159290701092e-05,
      "loss": 1.1774,
      "step": 5431
    },
    {
      "epoch": 2.9164793127747393,
      "grad_norm": 0.8984375,
      "learning_rate": 1.8553999331848555e-05,
      "loss": 0.9284,
      "step": 5432
    },
    {
      "epoch": 2.917016207509815,
      "grad_norm": 0.74609375,
      "learning_rate": 1.854584010856632e-05,
      "loss": 1.0889,
      "step": 5433
    },
    {
      "epoch": 2.917553102244891,
      "grad_norm": 0.67578125,
      "learning_rate": 1.8537681621785465e-05,
      "loss": 0.7558,
      "step": 5434
    },
    {
      "epoch": 2.918089996979967,
      "grad_norm": 0.79296875,
      "learning_rate": 1.852952387243698e-05,
      "loss": 0.8962,
      "step": 5435
    },
    {
      "epoch": 2.918626891715043,
      "grad_norm": 0.66015625,
      "learning_rate": 1.852136686145176e-05,
      "loss": 0.9257,
      "step": 5436
    },
    {
      "epoch": 2.919163786450119,
      "grad_norm": 0.70703125,
      "learning_rate": 1.8513210589760637e-05,
      "loss": 0.7801,
      "step": 5437
    },
    {
      "epoch": 2.919700681185195,
      "grad_norm": 0.75,
      "learning_rate": 1.8505055058294322e-05,
      "loss": 0.7405,
      "step": 5438
    },
    {
      "epoch": 2.9202375759202712,
      "grad_norm": 0.640625,
      "learning_rate": 1.849690026798348e-05,
      "loss": 0.863,
      "step": 5439
    },
    {
      "epoch": 2.9207744706553473,
      "grad_norm": 0.6328125,
      "learning_rate": 1.8488746219758674e-05,
      "loss": 0.7822,
      "step": 5440
    },
    {
      "epoch": 2.921311365390423,
      "grad_norm": 0.5859375,
      "learning_rate": 1.8480592914550383e-05,
      "loss": 0.6784,
      "step": 5441
    },
    {
      "epoch": 2.921848260125499,
      "grad_norm": 0.76953125,
      "learning_rate": 1.8472440353289e-05,
      "loss": 0.9939,
      "step": 5442
    },
    {
      "epoch": 2.9223851548605753,
      "grad_norm": 0.64453125,
      "learning_rate": 1.8464288536904833e-05,
      "loss": 0.8815,
      "step": 5443
    },
    {
      "epoch": 2.922922049595651,
      "grad_norm": 0.734375,
      "learning_rate": 1.8456137466328116e-05,
      "loss": 0.8956,
      "step": 5444
    },
    {
      "epoch": 2.923458944330727,
      "grad_norm": 0.77734375,
      "learning_rate": 1.844798714248899e-05,
      "loss": 0.8806,
      "step": 5445
    },
    {
      "epoch": 2.923995839065803,
      "grad_norm": 0.66796875,
      "learning_rate": 1.8439837566317505e-05,
      "loss": 0.8622,
      "step": 5446
    },
    {
      "epoch": 2.9245327338008793,
      "grad_norm": 0.67578125,
      "learning_rate": 1.843168873874363e-05,
      "loss": 1.0269,
      "step": 5447
    },
    {
      "epoch": 2.9250696285359554,
      "grad_norm": 0.65234375,
      "learning_rate": 1.8423540660697242e-05,
      "loss": 0.733,
      "step": 5448
    },
    {
      "epoch": 2.925606523271031,
      "grad_norm": 0.76171875,
      "learning_rate": 1.841539333310816e-05,
      "loss": 0.7116,
      "step": 5449
    },
    {
      "epoch": 2.926143418006107,
      "grad_norm": 0.6796875,
      "learning_rate": 1.840724675690608e-05,
      "loss": 0.8003,
      "step": 5450
    },
    {
      "epoch": 2.926143418006107,
      "eval_loss": 0.9032182693481445,
      "eval_runtime": 47.7071,
      "eval_samples_per_second": 20.165,
      "eval_steps_per_second": 20.165,
      "step": 5450
    },
    {
      "epoch": 2.9266803127411833,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8399100933020633e-05,
      "loss": 0.9192,
      "step": 5451
    },
    {
      "epoch": 2.927217207476259,
      "grad_norm": 0.77734375,
      "learning_rate": 1.839095586238137e-05,
      "loss": 1.0464,
      "step": 5452
    },
    {
      "epoch": 2.927754102211335,
      "grad_norm": 0.828125,
      "learning_rate": 1.838281154591774e-05,
      "loss": 0.8922,
      "step": 5453
    },
    {
      "epoch": 2.9282909969464113,
      "grad_norm": 0.69140625,
      "learning_rate": 1.8374667984559112e-05,
      "loss": 0.8073,
      "step": 5454
    },
    {
      "epoch": 2.928827891681487,
      "grad_norm": 0.703125,
      "learning_rate": 1.836652517923477e-05,
      "loss": 0.9222,
      "step": 5455
    },
    {
      "epoch": 2.929364786416563,
      "grad_norm": 0.5859375,
      "learning_rate": 1.8358383130873902e-05,
      "loss": 0.8636,
      "step": 5456
    },
    {
      "epoch": 2.929901681151639,
      "grad_norm": 0.7890625,
      "learning_rate": 1.8350241840405638e-05,
      "loss": 0.975,
      "step": 5457
    },
    {
      "epoch": 2.9304385758867153,
      "grad_norm": 0.74609375,
      "learning_rate": 1.834210130875899e-05,
      "loss": 0.8365,
      "step": 5458
    },
    {
      "epoch": 2.9309754706217914,
      "grad_norm": 0.640625,
      "learning_rate": 1.8333961536862893e-05,
      "loss": 0.9949,
      "step": 5459
    },
    {
      "epoch": 2.931512365356867,
      "grad_norm": 0.65234375,
      "learning_rate": 1.8325822525646208e-05,
      "loss": 0.6634,
      "step": 5460
    },
    {
      "epoch": 2.932049260091943,
      "grad_norm": 0.71484375,
      "learning_rate": 1.8317684276037696e-05,
      "loss": 1.008,
      "step": 5461
    },
    {
      "epoch": 2.9325861548270193,
      "grad_norm": 0.6953125,
      "learning_rate": 1.8309546788966037e-05,
      "loss": 0.8497,
      "step": 5462
    },
    {
      "epoch": 2.933123049562095,
      "grad_norm": 0.6953125,
      "learning_rate": 1.830141006535981e-05,
      "loss": 0.99,
      "step": 5463
    },
    {
      "epoch": 2.933659944297171,
      "grad_norm": 0.70703125,
      "learning_rate": 1.8293274106147525e-05,
      "loss": 0.8016,
      "step": 5464
    },
    {
      "epoch": 2.9341968390322473,
      "grad_norm": 0.64453125,
      "learning_rate": 1.8285138912257597e-05,
      "loss": 0.7113,
      "step": 5465
    },
    {
      "epoch": 2.9347337337673234,
      "grad_norm": 0.69921875,
      "learning_rate": 1.827700448461836e-05,
      "loss": 0.8979,
      "step": 5466
    },
    {
      "epoch": 2.9352706285023995,
      "grad_norm": 0.71484375,
      "learning_rate": 1.8268870824158045e-05,
      "loss": 0.8675,
      "step": 5467
    },
    {
      "epoch": 2.935807523237475,
      "grad_norm": 0.6953125,
      "learning_rate": 1.826073793180482e-05,
      "loss": 0.9291,
      "step": 5468
    },
    {
      "epoch": 2.9363444179725513,
      "grad_norm": 0.62109375,
      "learning_rate": 1.825260580848674e-05,
      "loss": 0.7463,
      "step": 5469
    },
    {
      "epoch": 2.9368813127076274,
      "grad_norm": 0.640625,
      "learning_rate": 1.8244474455131792e-05,
      "loss": 0.9951,
      "step": 5470
    },
    {
      "epoch": 2.937418207442703,
      "grad_norm": 0.6484375,
      "learning_rate": 1.8236343872667855e-05,
      "loss": 0.8259,
      "step": 5471
    },
    {
      "epoch": 2.937955102177779,
      "grad_norm": 0.875,
      "learning_rate": 1.822821406202273e-05,
      "loss": 1.0088,
      "step": 5472
    },
    {
      "epoch": 2.9384919969128553,
      "grad_norm": 0.68359375,
      "learning_rate": 1.8220085024124146e-05,
      "loss": 0.8069,
      "step": 5473
    },
    {
      "epoch": 2.939028891647931,
      "grad_norm": 0.8359375,
      "learning_rate": 1.821195675989972e-05,
      "loss": 0.8615,
      "step": 5474
    },
    {
      "epoch": 2.939565786383007,
      "grad_norm": 0.8125,
      "learning_rate": 1.8203829270276988e-05,
      "loss": 1.0052,
      "step": 5475
    },
    {
      "epoch": 2.9401026811180833,
      "grad_norm": 0.74609375,
      "learning_rate": 1.8195702556183407e-05,
      "loss": 1.0683,
      "step": 5476
    },
    {
      "epoch": 2.9406395758531594,
      "grad_norm": 0.70703125,
      "learning_rate": 1.818757661854633e-05,
      "loss": 0.9592,
      "step": 5477
    },
    {
      "epoch": 2.9411764705882355,
      "grad_norm": 0.640625,
      "learning_rate": 1.8179451458293034e-05,
      "loss": 0.8372,
      "step": 5478
    },
    {
      "epoch": 2.941713365323311,
      "grad_norm": 0.734375,
      "learning_rate": 1.8171327076350714e-05,
      "loss": 0.8697,
      "step": 5479
    },
    {
      "epoch": 2.9422502600583873,
      "grad_norm": 0.62890625,
      "learning_rate": 1.8163203473646435e-05,
      "loss": 0.7804,
      "step": 5480
    },
    {
      "epoch": 2.9427871547934634,
      "grad_norm": 0.5859375,
      "learning_rate": 1.815508065110723e-05,
      "loss": 0.7558,
      "step": 5481
    },
    {
      "epoch": 2.943324049528539,
      "grad_norm": 0.6953125,
      "learning_rate": 1.8146958609660007e-05,
      "loss": 0.9593,
      "step": 5482
    },
    {
      "epoch": 2.943860944263615,
      "grad_norm": 0.6640625,
      "learning_rate": 1.8138837350231584e-05,
      "loss": 0.8378,
      "step": 5483
    },
    {
      "epoch": 2.9443978389986913,
      "grad_norm": 0.7421875,
      "learning_rate": 1.8130716873748718e-05,
      "loss": 0.8555,
      "step": 5484
    },
    {
      "epoch": 2.9449347337337675,
      "grad_norm": 0.77734375,
      "learning_rate": 1.812259718113805e-05,
      "loss": 0.7301,
      "step": 5485
    },
    {
      "epoch": 2.9454716284688436,
      "grad_norm": 0.640625,
      "learning_rate": 1.8114478273326135e-05,
      "loss": 0.8048,
      "step": 5486
    },
    {
      "epoch": 2.9460085232039193,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8106360151239454e-05,
      "loss": 0.8567,
      "step": 5487
    },
    {
      "epoch": 2.9465454179389954,
      "grad_norm": 0.61328125,
      "learning_rate": 1.809824281580437e-05,
      "loss": 0.7145,
      "step": 5488
    },
    {
      "epoch": 2.9470823126740715,
      "grad_norm": 0.90234375,
      "learning_rate": 1.809012626794719e-05,
      "loss": 0.9999,
      "step": 5489
    },
    {
      "epoch": 2.947619207409147,
      "grad_norm": 0.73828125,
      "learning_rate": 1.808201050859411e-05,
      "loss": 0.9137,
      "step": 5490
    },
    {
      "epoch": 2.9481561021442233,
      "grad_norm": 0.6796875,
      "learning_rate": 1.8073895538671236e-05,
      "loss": 0.8003,
      "step": 5491
    },
    {
      "epoch": 2.9486929968792994,
      "grad_norm": 0.9375,
      "learning_rate": 1.8065781359104596e-05,
      "loss": 0.809,
      "step": 5492
    },
    {
      "epoch": 2.949229891614375,
      "grad_norm": 0.58203125,
      "learning_rate": 1.8057667970820115e-05,
      "loss": 0.7794,
      "step": 5493
    },
    {
      "epoch": 2.949766786349451,
      "grad_norm": 0.671875,
      "learning_rate": 1.804955537474364e-05,
      "loss": 0.886,
      "step": 5494
    },
    {
      "epoch": 2.9503036810845273,
      "grad_norm": 0.703125,
      "learning_rate": 1.804144357180091e-05,
      "loss": 0.9306,
      "step": 5495
    },
    {
      "epoch": 2.9508405758196035,
      "grad_norm": 0.78515625,
      "learning_rate": 1.803333256291761e-05,
      "loss": 0.8701,
      "step": 5496
    },
    {
      "epoch": 2.9513774705546796,
      "grad_norm": 0.72265625,
      "learning_rate": 1.802522234901927e-05,
      "loss": 0.6668,
      "step": 5497
    },
    {
      "epoch": 2.9519143652897553,
      "grad_norm": 0.63671875,
      "learning_rate": 1.80171129310314e-05,
      "loss": 0.6868,
      "step": 5498
    },
    {
      "epoch": 2.9524512600248314,
      "grad_norm": 0.9296875,
      "learning_rate": 1.800900430987937e-05,
      "loss": 1.0572,
      "step": 5499
    },
    {
      "epoch": 2.9529881547599075,
      "grad_norm": 0.80859375,
      "learning_rate": 1.800089648648848e-05,
      "loss": 0.862,
      "step": 5500
    },
    {
      "epoch": 2.9529881547599075,
      "eval_loss": 0.9033634662628174,
      "eval_runtime": 46.6055,
      "eval_samples_per_second": 20.641,
      "eval_steps_per_second": 20.641,
      "step": 5500
    },
    {
      "epoch": 2.953525049494983,
      "grad_norm": 0.7265625,
      "learning_rate": 1.799278946178394e-05,
      "loss": 0.8409,
      "step": 5501
    },
    {
      "epoch": 2.9540619442300593,
      "grad_norm": 0.7265625,
      "learning_rate": 1.7984683236690863e-05,
      "loss": 0.8949,
      "step": 5502
    },
    {
      "epoch": 2.9545988389651354,
      "grad_norm": 0.84765625,
      "learning_rate": 1.7976577812134265e-05,
      "loss": 0.9617,
      "step": 5503
    },
    {
      "epoch": 2.9551357337002115,
      "grad_norm": 0.671875,
      "learning_rate": 1.79684731890391e-05,
      "loss": 0.7503,
      "step": 5504
    },
    {
      "epoch": 2.9556726284352877,
      "grad_norm": 0.734375,
      "learning_rate": 1.7960369368330173e-05,
      "loss": 0.8546,
      "step": 5505
    },
    {
      "epoch": 2.9562095231703633,
      "grad_norm": 0.81640625,
      "learning_rate": 1.795226635093226e-05,
      "loss": 0.7963,
      "step": 5506
    },
    {
      "epoch": 2.9567464179054395,
      "grad_norm": 0.66796875,
      "learning_rate": 1.7944164137770002e-05,
      "loss": 0.845,
      "step": 5507
    },
    {
      "epoch": 2.9572833126405156,
      "grad_norm": 0.69140625,
      "learning_rate": 1.7936062729767973e-05,
      "loss": 0.9114,
      "step": 5508
    },
    {
      "epoch": 2.9578202073755913,
      "grad_norm": 0.8046875,
      "learning_rate": 1.7927962127850646e-05,
      "loss": 0.9163,
      "step": 5509
    },
    {
      "epoch": 2.9583571021106674,
      "grad_norm": 0.73046875,
      "learning_rate": 1.79198623329424e-05,
      "loss": 0.7591,
      "step": 5510
    },
    {
      "epoch": 2.9588939968457435,
      "grad_norm": 0.7265625,
      "learning_rate": 1.791176334596752e-05,
      "loss": 0.7553,
      "step": 5511
    },
    {
      "epoch": 2.959430891580819,
      "grad_norm": 0.6796875,
      "learning_rate": 1.7903665167850224e-05,
      "loss": 0.9116,
      "step": 5512
    },
    {
      "epoch": 2.9599677863158953,
      "grad_norm": 0.609375,
      "learning_rate": 1.7895567799514584e-05,
      "loss": 0.7526,
      "step": 5513
    },
    {
      "epoch": 2.9605046810509714,
      "grad_norm": 0.70703125,
      "learning_rate": 1.7887471241884632e-05,
      "loss": 0.8126,
      "step": 5514
    },
    {
      "epoch": 2.9610415757860475,
      "grad_norm": 0.66015625,
      "learning_rate": 1.7879375495884284e-05,
      "loss": 0.9015,
      "step": 5515
    },
    {
      "epoch": 2.9615784705211237,
      "grad_norm": 0.64453125,
      "learning_rate": 1.7871280562437366e-05,
      "loss": 0.7862,
      "step": 5516
    },
    {
      "epoch": 2.9621153652561993,
      "grad_norm": 0.7265625,
      "learning_rate": 1.7863186442467613e-05,
      "loss": 1.1442,
      "step": 5517
    },
    {
      "epoch": 2.9626522599912755,
      "grad_norm": 0.68359375,
      "learning_rate": 1.785509313689867e-05,
      "loss": 0.8277,
      "step": 5518
    },
    {
      "epoch": 2.9631891547263516,
      "grad_norm": 0.74609375,
      "learning_rate": 1.784700064665408e-05,
      "loss": 0.9182,
      "step": 5519
    },
    {
      "epoch": 2.9637260494614273,
      "grad_norm": 0.6328125,
      "learning_rate": 1.7838908972657303e-05,
      "loss": 0.8057,
      "step": 5520
    },
    {
      "epoch": 2.9642629441965034,
      "grad_norm": 0.65625,
      "learning_rate": 1.783081811583171e-05,
      "loss": 0.794,
      "step": 5521
    },
    {
      "epoch": 2.9647998389315795,
      "grad_norm": 0.64453125,
      "learning_rate": 1.7822728077100552e-05,
      "loss": 0.9761,
      "step": 5522
    },
    {
      "epoch": 2.9653367336666556,
      "grad_norm": 0.65234375,
      "learning_rate": 1.7814638857387015e-05,
      "loss": 0.8062,
      "step": 5523
    },
    {
      "epoch": 2.9658736284017317,
      "grad_norm": 0.75390625,
      "learning_rate": 1.7806550457614174e-05,
      "loss": 0.9379,
      "step": 5524
    },
    {
      "epoch": 2.9664105231368074,
      "grad_norm": 0.66015625,
      "learning_rate": 1.7798462878705025e-05,
      "loss": 0.8716,
      "step": 5525
    },
    {
      "epoch": 2.9669474178718835,
      "grad_norm": 0.6484375,
      "learning_rate": 1.7790376121582465e-05,
      "loss": 0.7737,
      "step": 5526
    },
    {
      "epoch": 2.9674843126069597,
      "grad_norm": 1.8828125,
      "learning_rate": 1.7782290187169294e-05,
      "loss": 0.9156,
      "step": 5527
    },
    {
      "epoch": 2.9680212073420353,
      "grad_norm": 0.640625,
      "learning_rate": 1.7774205076388206e-05,
      "loss": 0.9634,
      "step": 5528
    },
    {
      "epoch": 2.9685581020771115,
      "grad_norm": 0.6953125,
      "learning_rate": 1.7766120790161845e-05,
      "loss": 0.8445,
      "step": 5529
    },
    {
      "epoch": 2.9690949968121876,
      "grad_norm": 0.609375,
      "learning_rate": 1.77580373294127e-05,
      "loss": 0.726,
      "step": 5530
    },
    {
      "epoch": 2.9696318915472633,
      "grad_norm": 0.72265625,
      "learning_rate": 1.774995469506321e-05,
      "loss": 0.8796,
      "step": 5531
    },
    {
      "epoch": 2.9701687862823394,
      "grad_norm": 1.0390625,
      "learning_rate": 1.7741872888035702e-05,
      "loss": 1.0207,
      "step": 5532
    },
    {
      "epoch": 2.9707056810174155,
      "grad_norm": 0.7109375,
      "learning_rate": 1.773379190925241e-05,
      "loss": 0.8211,
      "step": 5533
    },
    {
      "epoch": 2.9712425757524916,
      "grad_norm": 0.68359375,
      "learning_rate": 1.7725711759635487e-05,
      "loss": 0.8441,
      "step": 5534
    },
    {
      "epoch": 2.9717794704875677,
      "grad_norm": 0.72265625,
      "learning_rate": 1.771763244010697e-05,
      "loss": 1.0759,
      "step": 5535
    },
    {
      "epoch": 2.9723163652226434,
      "grad_norm": 0.83203125,
      "learning_rate": 1.7709553951588813e-05,
      "loss": 0.7086,
      "step": 5536
    },
    {
      "epoch": 2.9728532599577195,
      "grad_norm": 0.70703125,
      "learning_rate": 1.770147629500288e-05,
      "loss": 0.9243,
      "step": 5537
    },
    {
      "epoch": 2.9733901546927957,
      "grad_norm": 0.6328125,
      "learning_rate": 1.7693399471270937e-05,
      "loss": 0.7882,
      "step": 5538
    },
    {
      "epoch": 2.9739270494278713,
      "grad_norm": 0.6484375,
      "learning_rate": 1.768532348131464e-05,
      "loss": 0.8293,
      "step": 5539
    },
    {
      "epoch": 2.9744639441629475,
      "grad_norm": 0.69921875,
      "learning_rate": 1.7677248326055564e-05,
      "loss": 0.7795,
      "step": 5540
    },
    {
      "epoch": 2.9750008388980236,
      "grad_norm": 0.7421875,
      "learning_rate": 1.7669174006415186e-05,
      "loss": 0.8374,
      "step": 5541
    },
    {
      "epoch": 2.9755377336330997,
      "grad_norm": 0.69921875,
      "learning_rate": 1.7661100523314894e-05,
      "loss": 0.8082,
      "step": 5542
    },
    {
      "epoch": 2.976074628368176,
      "grad_norm": 0.63671875,
      "learning_rate": 1.7653027877675976e-05,
      "loss": 0.692,
      "step": 5543
    },
    {
      "epoch": 2.9766115231032515,
      "grad_norm": 0.8125,
      "learning_rate": 1.764495607041961e-05,
      "loss": 1.015,
      "step": 5544
    },
    {
      "epoch": 2.9771484178383276,
      "grad_norm": 0.64453125,
      "learning_rate": 1.7636885102466907e-05,
      "loss": 0.7693,
      "step": 5545
    },
    {
      "epoch": 2.9776853125734037,
      "grad_norm": 0.55078125,
      "learning_rate": 1.7628814974738866e-05,
      "loss": 0.7557,
      "step": 5546
    },
    {
      "epoch": 2.9782222073084794,
      "grad_norm": 0.69140625,
      "learning_rate": 1.762074568815638e-05,
      "loss": 0.9253,
      "step": 5547
    },
    {
      "epoch": 2.9787591020435555,
      "grad_norm": 0.703125,
      "learning_rate": 1.761267724364026e-05,
      "loss": 0.8267,
      "step": 5548
    },
    {
      "epoch": 2.9792959967786317,
      "grad_norm": 0.6171875,
      "learning_rate": 1.7604609642111214e-05,
      "loss": 0.7756,
      "step": 5549
    },
    {
      "epoch": 2.9798328915137073,
      "grad_norm": 0.70703125,
      "learning_rate": 1.759654288448987e-05,
      "loss": 0.9439,
      "step": 5550
    },
    {
      "epoch": 2.9798328915137073,
      "eval_loss": 0.9031280875205994,
      "eval_runtime": 46.8345,
      "eval_samples_per_second": 20.54,
      "eval_steps_per_second": 20.54,
      "step": 5550
    },
    {
      "epoch": 2.9803697862487835,
      "grad_norm": 0.66796875,
      "learning_rate": 1.758847697169674e-05,
      "loss": 0.882,
      "step": 5551
    },
    {
      "epoch": 2.9809066809838596,
      "grad_norm": 0.73828125,
      "learning_rate": 1.758041190465224e-05,
      "loss": 0.9173,
      "step": 5552
    },
    {
      "epoch": 2.9814435757189357,
      "grad_norm": 0.7265625,
      "learning_rate": 1.7572347684276708e-05,
      "loss": 0.7461,
      "step": 5553
    },
    {
      "epoch": 2.981980470454012,
      "grad_norm": 0.75390625,
      "learning_rate": 1.7564284311490376e-05,
      "loss": 1.1007,
      "step": 5554
    },
    {
      "epoch": 2.9825173651890875,
      "grad_norm": 0.7421875,
      "learning_rate": 1.7556221787213363e-05,
      "loss": 0.7541,
      "step": 5555
    },
    {
      "epoch": 2.9830542599241636,
      "grad_norm": 0.671875,
      "learning_rate": 1.7548160112365713e-05,
      "loss": 0.9148,
      "step": 5556
    },
    {
      "epoch": 2.9835911546592397,
      "grad_norm": 0.609375,
      "learning_rate": 1.7540099287867358e-05,
      "loss": 0.8838,
      "step": 5557
    },
    {
      "epoch": 2.9841280493943154,
      "grad_norm": 0.68359375,
      "learning_rate": 1.7532039314638155e-05,
      "loss": 0.8607,
      "step": 5558
    },
    {
      "epoch": 2.9846649441293915,
      "grad_norm": 0.7890625,
      "learning_rate": 1.7523980193597836e-05,
      "loss": 1.0443,
      "step": 5559
    },
    {
      "epoch": 2.9852018388644677,
      "grad_norm": 0.81640625,
      "learning_rate": 1.7515921925666052e-05,
      "loss": 1.0869,
      "step": 5560
    },
    {
      "epoch": 2.985738733599544,
      "grad_norm": 0.63671875,
      "learning_rate": 1.7507864511762355e-05,
      "loss": 0.7088,
      "step": 5561
    },
    {
      "epoch": 2.98627562833462,
      "grad_norm": 0.66796875,
      "learning_rate": 1.74998079528062e-05,
      "loss": 0.874,
      "step": 5562
    },
    {
      "epoch": 2.9868125230696956,
      "grad_norm": 0.71484375,
      "learning_rate": 1.749175224971694e-05,
      "loss": 0.7606,
      "step": 5563
    },
    {
      "epoch": 2.9873494178047717,
      "grad_norm": 0.6640625,
      "learning_rate": 1.7483697403413836e-05,
      "loss": 0.6888,
      "step": 5564
    },
    {
      "epoch": 2.987886312539848,
      "grad_norm": 0.70703125,
      "learning_rate": 1.7475643414816033e-05,
      "loss": 0.9098,
      "step": 5565
    },
    {
      "epoch": 2.9884232072749235,
      "grad_norm": 0.9140625,
      "learning_rate": 1.7467590284842612e-05,
      "loss": 1.0294,
      "step": 5566
    },
    {
      "epoch": 2.9889601020099996,
      "grad_norm": 0.66015625,
      "learning_rate": 1.7459538014412528e-05,
      "loss": 0.7936,
      "step": 5567
    },
    {
      "epoch": 2.9894969967450757,
      "grad_norm": 0.6328125,
      "learning_rate": 1.745148660444465e-05,
      "loss": 0.8168,
      "step": 5568
    },
    {
      "epoch": 2.9900338914801514,
      "grad_norm": 0.72265625,
      "learning_rate": 1.7443436055857742e-05,
      "loss": 0.774,
      "step": 5569
    },
    {
      "epoch": 2.9905707862152275,
      "grad_norm": 0.67578125,
      "learning_rate": 1.7435386369570478e-05,
      "loss": 0.8802,
      "step": 5570
    },
    {
      "epoch": 2.9911076809503037,
      "grad_norm": 0.75390625,
      "learning_rate": 1.7427337546501435e-05,
      "loss": 0.8458,
      "step": 5571
    },
    {
      "epoch": 2.99164457568538,
      "grad_norm": 0.5859375,
      "learning_rate": 1.741928958756907e-05,
      "loss": 0.648,
      "step": 5572
    },
    {
      "epoch": 2.992181470420456,
      "grad_norm": 0.74609375,
      "learning_rate": 1.741124249369177e-05,
      "loss": 0.944,
      "step": 5573
    },
    {
      "epoch": 2.9927183651555316,
      "grad_norm": 0.62109375,
      "learning_rate": 1.7403196265787797e-05,
      "loss": 0.8706,
      "step": 5574
    },
    {
      "epoch": 2.9932552598906077,
      "grad_norm": 0.7109375,
      "learning_rate": 1.7395150904775338e-05,
      "loss": 0.8605,
      "step": 5575
    },
    {
      "epoch": 2.993792154625684,
      "grad_norm": 0.765625,
      "learning_rate": 1.7387106411572474e-05,
      "loss": 0.9711,
      "step": 5576
    },
    {
      "epoch": 2.9943290493607595,
      "grad_norm": 0.61328125,
      "learning_rate": 1.7379062787097167e-05,
      "loss": 0.8551,
      "step": 5577
    },
    {
      "epoch": 2.9948659440958356,
      "grad_norm": 0.64453125,
      "learning_rate": 1.7371020032267315e-05,
      "loss": 0.8882,
      "step": 5578
    },
    {
      "epoch": 2.9954028388309117,
      "grad_norm": 0.65625,
      "learning_rate": 1.736297814800069e-05,
      "loss": 0.9511,
      "step": 5579
    },
    {
      "epoch": 2.995939733565988,
      "grad_norm": 0.8046875,
      "learning_rate": 1.7354937135214983e-05,
      "loss": 1.235,
      "step": 5580
    },
    {
      "epoch": 2.996476628301064,
      "grad_norm": 0.9609375,
      "learning_rate": 1.7346896994827756e-05,
      "loss": 1.3069,
      "step": 5581
    },
    {
      "epoch": 2.9970135230361397,
      "grad_norm": 0.73828125,
      "learning_rate": 1.7338857727756495e-05,
      "loss": 0.8365,
      "step": 5582
    },
    {
      "epoch": 2.997550417771216,
      "grad_norm": 0.625,
      "learning_rate": 1.7330819334918595e-05,
      "loss": 0.692,
      "step": 5583
    },
    {
      "epoch": 2.998087312506292,
      "grad_norm": 0.66796875,
      "learning_rate": 1.732278181723133e-05,
      "loss": 0.7502,
      "step": 5584
    },
    {
      "epoch": 2.9986242072413676,
      "grad_norm": 0.6328125,
      "learning_rate": 1.731474517561188e-05,
      "loss": 0.7588,
      "step": 5585
    },
    {
      "epoch": 2.9991611019764437,
      "grad_norm": 0.67578125,
      "learning_rate": 1.7306709410977333e-05,
      "loss": 0.6996,
      "step": 5586
    },
    {
      "epoch": 2.99969799671152,
      "grad_norm": 0.78125,
      "learning_rate": 1.7298674524244673e-05,
      "loss": 0.8848,
      "step": 5587
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.68359375,
      "learning_rate": 1.7290640516330786e-05,
      "loss": 0.4614,
      "step": 5588
    },
    {
      "epoch": 3.000536894735076,
      "grad_norm": 0.6328125,
      "learning_rate": 1.728260738815244e-05,
      "loss": 0.7966,
      "step": 5589
    },
    {
      "epoch": 3.001073789470152,
      "grad_norm": 0.609375,
      "learning_rate": 1.7274575140626318e-05,
      "loss": 0.8168,
      "step": 5590
    },
    {
      "epoch": 3.001610684205228,
      "grad_norm": 0.6328125,
      "learning_rate": 1.726654377466901e-05,
      "loss": 0.9058,
      "step": 5591
    },
    {
      "epoch": 3.002147578940304,
      "grad_norm": 0.76953125,
      "learning_rate": 1.7258513291196998e-05,
      "loss": 0.8953,
      "step": 5592
    },
    {
      "epoch": 3.00268447367538,
      "grad_norm": 0.69921875,
      "learning_rate": 1.725048369112665e-05,
      "loss": 0.8061,
      "step": 5593
    },
    {
      "epoch": 3.003221368410456,
      "grad_norm": 0.66015625,
      "learning_rate": 1.724245497537426e-05,
      "loss": 0.9316,
      "step": 5594
    },
    {
      "epoch": 3.003758263145532,
      "grad_norm": 1.1484375,
      "learning_rate": 1.7234427144855996e-05,
      "loss": 0.8271,
      "step": 5595
    },
    {
      "epoch": 3.004295157880608,
      "grad_norm": 0.63671875,
      "learning_rate": 1.7226400200487943e-05,
      "loss": 0.6916,
      "step": 5596
    },
    {
      "epoch": 3.004832052615684,
      "grad_norm": 0.69140625,
      "learning_rate": 1.721837414318607e-05,
      "loss": 0.6548,
      "step": 5597
    },
    {
      "epoch": 3.00536894735076,
      "grad_norm": 0.62109375,
      "learning_rate": 1.7210348973866243e-05,
      "loss": 0.9037,
      "step": 5598
    },
    {
      "epoch": 3.005905842085836,
      "grad_norm": 0.58984375,
      "learning_rate": 1.7202324693444256e-05,
      "loss": 0.8066,
      "step": 5599
    },
    {
      "epoch": 3.006442736820912,
      "grad_norm": 0.71875,
      "learning_rate": 1.7194301302835768e-05,
      "loss": 0.7934,
      "step": 5600
    },
    {
      "epoch": 3.006442736820912,
      "eval_loss": 0.9029846787452698,
      "eval_runtime": 47.4911,
      "eval_samples_per_second": 20.256,
      "eval_steps_per_second": 20.256,
      "step": 5600
    },
    {
      "epoch": 3.0069796315559882,
      "grad_norm": 0.6875,
      "learning_rate": 1.7186278802956353e-05,
      "loss": 0.7325,
      "step": 5601
    },
    {
      "epoch": 3.007516526291064,
      "grad_norm": 0.53515625,
      "learning_rate": 1.7178257194721477e-05,
      "loss": 0.7257,
      "step": 5602
    },
    {
      "epoch": 3.00805342102614,
      "grad_norm": 0.6484375,
      "learning_rate": 1.7170236479046516e-05,
      "loss": 0.8145,
      "step": 5603
    },
    {
      "epoch": 3.008590315761216,
      "grad_norm": 0.671875,
      "learning_rate": 1.716221665684672e-05,
      "loss": 0.8214,
      "step": 5604
    },
    {
      "epoch": 3.0091272104962923,
      "grad_norm": 0.640625,
      "learning_rate": 1.7154197729037274e-05,
      "loss": 0.6907,
      "step": 5605
    },
    {
      "epoch": 3.009664105231368,
      "grad_norm": 0.80078125,
      "learning_rate": 1.7146179696533212e-05,
      "loss": 0.6981,
      "step": 5606
    },
    {
      "epoch": 3.010200999966444,
      "grad_norm": 0.61328125,
      "learning_rate": 1.7138162560249517e-05,
      "loss": 0.7649,
      "step": 5607
    },
    {
      "epoch": 3.01073789470152,
      "grad_norm": 0.74609375,
      "learning_rate": 1.7130146321101032e-05,
      "loss": 0.7687,
      "step": 5608
    },
    {
      "epoch": 3.011274789436596,
      "grad_norm": 0.6953125,
      "learning_rate": 1.712213098000251e-05,
      "loss": 0.9491,
      "step": 5609
    },
    {
      "epoch": 3.011811684171672,
      "grad_norm": 0.73046875,
      "learning_rate": 1.711411653786861e-05,
      "loss": 0.8883,
      "step": 5610
    },
    {
      "epoch": 3.012348578906748,
      "grad_norm": 0.7109375,
      "learning_rate": 1.7106102995613888e-05,
      "loss": 0.8226,
      "step": 5611
    },
    {
      "epoch": 3.0128854736418242,
      "grad_norm": 0.6953125,
      "learning_rate": 1.709809035415278e-05,
      "loss": 0.6995,
      "step": 5612
    },
    {
      "epoch": 3.0134223683769,
      "grad_norm": 0.6796875,
      "learning_rate": 1.7090078614399635e-05,
      "loss": 0.8278,
      "step": 5613
    },
    {
      "epoch": 3.013959263111976,
      "grad_norm": 0.62890625,
      "learning_rate": 1.7082067777268678e-05,
      "loss": 0.9032,
      "step": 5614
    },
    {
      "epoch": 3.014496157847052,
      "grad_norm": 0.5234375,
      "learning_rate": 1.707405784367407e-05,
      "loss": 0.7106,
      "step": 5615
    },
    {
      "epoch": 3.0150330525821283,
      "grad_norm": 0.61328125,
      "learning_rate": 1.7066048814529834e-05,
      "loss": 0.6835,
      "step": 5616
    },
    {
      "epoch": 3.015569947317204,
      "grad_norm": 0.59375,
      "learning_rate": 1.7058040690749905e-05,
      "loss": 0.6236,
      "step": 5617
    },
    {
      "epoch": 3.01610684205228,
      "grad_norm": 0.8984375,
      "learning_rate": 1.7050033473248107e-05,
      "loss": 1.0277,
      "step": 5618
    },
    {
      "epoch": 3.016643736787356,
      "grad_norm": 0.625,
      "learning_rate": 1.704202716293817e-05,
      "loss": 0.9026,
      "step": 5619
    },
    {
      "epoch": 3.0171806315224323,
      "grad_norm": 0.76171875,
      "learning_rate": 1.703402176073371e-05,
      "loss": 0.78,
      "step": 5620
    },
    {
      "epoch": 3.017717526257508,
      "grad_norm": 0.63671875,
      "learning_rate": 1.702601726754825e-05,
      "loss": 0.8755,
      "step": 5621
    },
    {
      "epoch": 3.018254420992584,
      "grad_norm": 0.6328125,
      "learning_rate": 1.7018013684295212e-05,
      "loss": 0.6694,
      "step": 5622
    },
    {
      "epoch": 3.0187913157276602,
      "grad_norm": 0.58984375,
      "learning_rate": 1.7010011011887882e-05,
      "loss": 0.7758,
      "step": 5623
    },
    {
      "epoch": 3.0193282104627364,
      "grad_norm": 0.65234375,
      "learning_rate": 1.7002009251239488e-05,
      "loss": 0.7435,
      "step": 5624
    },
    {
      "epoch": 3.019865105197812,
      "grad_norm": 0.6484375,
      "learning_rate": 1.6994008403263122e-05,
      "loss": 0.7837,
      "step": 5625
    },
    {
      "epoch": 3.020401999932888,
      "grad_norm": 0.67578125,
      "learning_rate": 1.6986008468871783e-05,
      "loss": 0.904,
      "step": 5626
    },
    {
      "epoch": 3.0209388946679643,
      "grad_norm": 0.76953125,
      "learning_rate": 1.697800944897837e-05,
      "loss": 0.7706,
      "step": 5627
    },
    {
      "epoch": 3.02147578940304,
      "grad_norm": 0.8125,
      "learning_rate": 1.6970011344495667e-05,
      "loss": 0.999,
      "step": 5628
    },
    {
      "epoch": 3.022012684138116,
      "grad_norm": 0.6015625,
      "learning_rate": 1.696201415633636e-05,
      "loss": 0.9001,
      "step": 5629
    },
    {
      "epoch": 3.022549578873192,
      "grad_norm": 0.7265625,
      "learning_rate": 1.6954017885413044e-05,
      "loss": 0.9004,
      "step": 5630
    },
    {
      "epoch": 3.0230864736082683,
      "grad_norm": 0.6875,
      "learning_rate": 1.6946022532638163e-05,
      "loss": 0.8519,
      "step": 5631
    },
    {
      "epoch": 3.023623368343344,
      "grad_norm": 0.61328125,
      "learning_rate": 1.6938028098924117e-05,
      "loss": 0.8961,
      "step": 5632
    },
    {
      "epoch": 3.02416026307842,
      "grad_norm": 0.62890625,
      "learning_rate": 1.693003458518316e-05,
      "loss": 0.8072,
      "step": 5633
    },
    {
      "epoch": 3.0246971578134962,
      "grad_norm": 0.67578125,
      "learning_rate": 1.6922041992327447e-05,
      "loss": 0.8388,
      "step": 5634
    },
    {
      "epoch": 3.0252340525485724,
      "grad_norm": 0.69921875,
      "learning_rate": 1.6914050321269047e-05,
      "loss": 0.7467,
      "step": 5635
    },
    {
      "epoch": 3.025770947283648,
      "grad_norm": 0.7109375,
      "learning_rate": 1.6906059572919908e-05,
      "loss": 0.6684,
      "step": 5636
    },
    {
      "epoch": 3.026307842018724,
      "grad_norm": 0.70703125,
      "learning_rate": 1.6898069748191868e-05,
      "loss": 0.9435,
      "step": 5637
    },
    {
      "epoch": 3.0268447367538003,
      "grad_norm": 0.6328125,
      "learning_rate": 1.6890080847996686e-05,
      "loss": 0.8938,
      "step": 5638
    },
    {
      "epoch": 3.0273816314888764,
      "grad_norm": 0.9375,
      "learning_rate": 1.6882092873245968e-05,
      "loss": 0.7075,
      "step": 5639
    },
    {
      "epoch": 3.027918526223952,
      "grad_norm": 0.64453125,
      "learning_rate": 1.6874105824851267e-05,
      "loss": 0.7713,
      "step": 5640
    },
    {
      "epoch": 3.028455420959028,
      "grad_norm": 0.6875,
      "learning_rate": 1.686611970372399e-05,
      "loss": 0.9216,
      "step": 5641
    },
    {
      "epoch": 3.0289923156941043,
      "grad_norm": 0.61328125,
      "learning_rate": 1.6858134510775463e-05,
      "loss": 0.7349,
      "step": 5642
    },
    {
      "epoch": 3.0295292104291804,
      "grad_norm": 0.640625,
      "learning_rate": 1.68501502469169e-05,
      "loss": 0.8568,
      "step": 5643
    },
    {
      "epoch": 3.030066105164256,
      "grad_norm": 0.59765625,
      "learning_rate": 1.6842166913059402e-05,
      "loss": 0.9279,
      "step": 5644
    },
    {
      "epoch": 3.0306029998993322,
      "grad_norm": 0.6796875,
      "learning_rate": 1.683418451011397e-05,
      "loss": 0.8226,
      "step": 5645
    },
    {
      "epoch": 3.0311398946344084,
      "grad_norm": 0.6953125,
      "learning_rate": 1.6826203038991498e-05,
      "loss": 0.9037,
      "step": 5646
    },
    {
      "epoch": 3.031676789369484,
      "grad_norm": 0.7109375,
      "learning_rate": 1.6818222500602775e-05,
      "loss": 0.6855,
      "step": 5647
    },
    {
      "epoch": 3.03221368410456,
      "grad_norm": 0.69140625,
      "learning_rate": 1.681024289585848e-05,
      "loss": 0.8555,
      "step": 5648
    },
    {
      "epoch": 3.0327505788396363,
      "grad_norm": 0.5703125,
      "learning_rate": 1.680226422566918e-05,
      "loss": 0.732,
      "step": 5649
    },
    {
      "epoch": 3.0332874735747124,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6794286490945342e-05,
      "loss": 0.7656,
      "step": 5650
    },
    {
      "epoch": 3.0332874735747124,
      "eval_loss": 0.9030352830886841,
      "eval_runtime": 47.4397,
      "eval_samples_per_second": 20.278,
      "eval_steps_per_second": 20.278,
      "step": 5650
    },
    {
      "epoch": 3.033824368309788,
      "grad_norm": 0.6640625,
      "learning_rate": 1.678630969259734e-05,
      "loss": 0.7524,
      "step": 5651
    },
    {
      "epoch": 3.034361263044864,
      "grad_norm": 0.73046875,
      "learning_rate": 1.677833383153542e-05,
      "loss": 0.7811,
      "step": 5652
    },
    {
      "epoch": 3.0348981577799403,
      "grad_norm": 0.65625,
      "learning_rate": 1.6770358908669732e-05,
      "loss": 0.8238,
      "step": 5653
    },
    {
      "epoch": 3.0354350525150164,
      "grad_norm": 0.90234375,
      "learning_rate": 1.6762384924910302e-05,
      "loss": 0.9373,
      "step": 5654
    },
    {
      "epoch": 3.035971947250092,
      "grad_norm": 0.77734375,
      "learning_rate": 1.6754411881167084e-05,
      "loss": 0.9124,
      "step": 5655
    },
    {
      "epoch": 3.0365088419851682,
      "grad_norm": 0.73828125,
      "learning_rate": 1.6746439778349888e-05,
      "loss": 0.8393,
      "step": 5656
    },
    {
      "epoch": 3.0370457367202444,
      "grad_norm": 0.83984375,
      "learning_rate": 1.6738468617368436e-05,
      "loss": 0.9503,
      "step": 5657
    },
    {
      "epoch": 3.0375826314553205,
      "grad_norm": 0.76171875,
      "learning_rate": 1.673049839913234e-05,
      "loss": 0.9281,
      "step": 5658
    },
    {
      "epoch": 3.038119526190396,
      "grad_norm": 0.56640625,
      "learning_rate": 1.6722529124551096e-05,
      "loss": 0.7913,
      "step": 5659
    },
    {
      "epoch": 3.0386564209254723,
      "grad_norm": 0.60546875,
      "learning_rate": 1.6714560794534108e-05,
      "loss": 0.7278,
      "step": 5660
    },
    {
      "epoch": 3.0391933156605484,
      "grad_norm": 0.71875,
      "learning_rate": 1.6706593409990663e-05,
      "loss": 0.9703,
      "step": 5661
    },
    {
      "epoch": 3.0397302103956245,
      "grad_norm": 0.72265625,
      "learning_rate": 1.6698626971829927e-05,
      "loss": 0.9035,
      "step": 5662
    },
    {
      "epoch": 3.0402671051307,
      "grad_norm": 0.734375,
      "learning_rate": 1.669066148096099e-05,
      "loss": 0.954,
      "step": 5663
    },
    {
      "epoch": 3.0408039998657763,
      "grad_norm": 0.66796875,
      "learning_rate": 1.6682696938292813e-05,
      "loss": 0.8916,
      "step": 5664
    },
    {
      "epoch": 3.0413408946008524,
      "grad_norm": 0.79296875,
      "learning_rate": 1.667473334473424e-05,
      "loss": 0.9074,
      "step": 5665
    },
    {
      "epoch": 3.041877789335928,
      "grad_norm": 0.66796875,
      "learning_rate": 1.6666770701194022e-05,
      "loss": 0.7338,
      "step": 5666
    },
    {
      "epoch": 3.0424146840710042,
      "grad_norm": 0.85546875,
      "learning_rate": 1.6658809008580795e-05,
      "loss": 0.9933,
      "step": 5667
    },
    {
      "epoch": 3.0429515788060804,
      "grad_norm": 0.69140625,
      "learning_rate": 1.6650848267803098e-05,
      "loss": 0.7882,
      "step": 5668
    },
    {
      "epoch": 3.0434884735411565,
      "grad_norm": 0.78125,
      "learning_rate": 1.6642888479769348e-05,
      "loss": 0.8334,
      "step": 5669
    },
    {
      "epoch": 3.044025368276232,
      "grad_norm": 0.7421875,
      "learning_rate": 1.6634929645387852e-05,
      "loss": 1.0332,
      "step": 5670
    },
    {
      "epoch": 3.0445622630113083,
      "grad_norm": 0.54296875,
      "learning_rate": 1.6626971765566822e-05,
      "loss": 0.7662,
      "step": 5671
    },
    {
      "epoch": 3.0450991577463844,
      "grad_norm": 0.71484375,
      "learning_rate": 1.6619014841214352e-05,
      "loss": 0.8795,
      "step": 5672
    },
    {
      "epoch": 3.0456360524814605,
      "grad_norm": 0.734375,
      "learning_rate": 1.6611058873238425e-05,
      "loss": 0.802,
      "step": 5673
    },
    {
      "epoch": 3.046172947216536,
      "grad_norm": 0.63671875,
      "learning_rate": 1.6603103862546913e-05,
      "loss": 0.7107,
      "step": 5674
    },
    {
      "epoch": 3.0467098419516123,
      "grad_norm": 0.6796875,
      "learning_rate": 1.659514981004759e-05,
      "loss": 0.7349,
      "step": 5675
    },
    {
      "epoch": 3.0472467366866884,
      "grad_norm": 0.6953125,
      "learning_rate": 1.6587196716648114e-05,
      "loss": 0.7133,
      "step": 5676
    },
    {
      "epoch": 3.0477836314217646,
      "grad_norm": 0.7109375,
      "learning_rate": 1.6579244583256028e-05,
      "loss": 0.7735,
      "step": 5677
    },
    {
      "epoch": 3.0483205261568402,
      "grad_norm": 0.828125,
      "learning_rate": 1.6571293410778778e-05,
      "loss": 0.8824,
      "step": 5678
    },
    {
      "epoch": 3.0488574208919164,
      "grad_norm": 0.63671875,
      "learning_rate": 1.656334320012369e-05,
      "loss": 0.9331,
      "step": 5679
    },
    {
      "epoch": 3.0493943156269925,
      "grad_norm": 0.625,
      "learning_rate": 1.6555393952197988e-05,
      "loss": 0.7424,
      "step": 5680
    },
    {
      "epoch": 3.0499312103620686,
      "grad_norm": 0.6015625,
      "learning_rate": 1.654744566790878e-05,
      "loss": 0.7131,
      "step": 5681
    },
    {
      "epoch": 3.0504681050971443,
      "grad_norm": 0.765625,
      "learning_rate": 1.653949834816306e-05,
      "loss": 0.8564,
      "step": 5682
    },
    {
      "epoch": 3.0510049998322204,
      "grad_norm": 0.66015625,
      "learning_rate": 1.6531551993867717e-05,
      "loss": 0.7673,
      "step": 5683
    },
    {
      "epoch": 3.0515418945672965,
      "grad_norm": 0.640625,
      "learning_rate": 1.6523606605929544e-05,
      "loss": 0.7774,
      "step": 5684
    },
    {
      "epoch": 3.052078789302372,
      "grad_norm": 0.71484375,
      "learning_rate": 1.6515662185255197e-05,
      "loss": 0.724,
      "step": 5685
    },
    {
      "epoch": 3.0526156840374483,
      "grad_norm": 0.75390625,
      "learning_rate": 1.650771873275124e-05,
      "loss": 0.8567,
      "step": 5686
    },
    {
      "epoch": 3.0531525787725244,
      "grad_norm": 0.71875,
      "learning_rate": 1.6499776249324123e-05,
      "loss": 0.892,
      "step": 5687
    },
    {
      "epoch": 3.0536894735076006,
      "grad_norm": 0.625,
      "learning_rate": 1.6491834735880185e-05,
      "loss": 0.8192,
      "step": 5688
    },
    {
      "epoch": 3.0542263682426762,
      "grad_norm": 0.63671875,
      "learning_rate": 1.6483894193325655e-05,
      "loss": 0.852,
      "step": 5689
    },
    {
      "epoch": 3.0547632629777524,
      "grad_norm": 0.6640625,
      "learning_rate": 1.6475954622566638e-05,
      "loss": 0.7537,
      "step": 5690
    },
    {
      "epoch": 3.0553001577128285,
      "grad_norm": 0.71484375,
      "learning_rate": 1.6468016024509145e-05,
      "loss": 0.9256,
      "step": 5691
    },
    {
      "epoch": 3.0558370524479046,
      "grad_norm": 0.72265625,
      "learning_rate": 1.6460078400059077e-05,
      "loss": 0.9278,
      "step": 5692
    },
    {
      "epoch": 3.0563739471829803,
      "grad_norm": 0.6015625,
      "learning_rate": 1.645214175012221e-05,
      "loss": 0.7843,
      "step": 5693
    },
    {
      "epoch": 3.0569108419180564,
      "grad_norm": 0.78125,
      "learning_rate": 1.6444206075604223e-05,
      "loss": 0.8648,
      "step": 5694
    },
    {
      "epoch": 3.0574477366531325,
      "grad_norm": 0.5546875,
      "learning_rate": 1.6436271377410666e-05,
      "loss": 0.6759,
      "step": 5695
    },
    {
      "epoch": 3.0579846313882086,
      "grad_norm": 0.63671875,
      "learning_rate": 1.6428337656447e-05,
      "loss": 0.6953,
      "step": 5696
    },
    {
      "epoch": 3.0585215261232843,
      "grad_norm": 0.640625,
      "learning_rate": 1.6420404913618563e-05,
      "loss": 0.8308,
      "step": 5697
    },
    {
      "epoch": 3.0590584208583604,
      "grad_norm": 0.64453125,
      "learning_rate": 1.6412473149830576e-05,
      "loss": 0.7212,
      "step": 5698
    },
    {
      "epoch": 3.0595953155934366,
      "grad_norm": 0.64453125,
      "learning_rate": 1.6404542365988154e-05,
      "loss": 0.6479,
      "step": 5699
    },
    {
      "epoch": 3.0601322103285127,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6396612562996296e-05,
      "loss": 1.0536,
      "step": 5700
    },
    {
      "epoch": 3.0601322103285127,
      "eval_loss": 0.9032897353172302,
      "eval_runtime": 46.7649,
      "eval_samples_per_second": 20.571,
      "eval_steps_per_second": 20.571,
      "step": 5700
    },
    {
      "epoch": 3.0606691050635884,
      "grad_norm": 0.53515625,
      "learning_rate": 1.63886837417599e-05,
      "loss": 0.9007,
      "step": 5701
    },
    {
      "epoch": 3.0612059997986645,
      "grad_norm": 0.65625,
      "learning_rate": 1.6380755903183745e-05,
      "loss": 0.7438,
      "step": 5702
    },
    {
      "epoch": 3.0617428945337406,
      "grad_norm": 0.70703125,
      "learning_rate": 1.6372829048172488e-05,
      "loss": 0.9285,
      "step": 5703
    },
    {
      "epoch": 3.0622797892688163,
      "grad_norm": 0.75,
      "learning_rate": 1.6364903177630698e-05,
      "loss": 0.8741,
      "step": 5704
    },
    {
      "epoch": 3.0628166840038924,
      "grad_norm": 0.67578125,
      "learning_rate": 1.6356978292462817e-05,
      "loss": 0.8718,
      "step": 5705
    },
    {
      "epoch": 3.0633535787389685,
      "grad_norm": 0.6640625,
      "learning_rate": 1.6349054393573157e-05,
      "loss": 0.8836,
      "step": 5706
    },
    {
      "epoch": 3.0638904734740446,
      "grad_norm": 0.76953125,
      "learning_rate": 1.634113148186595e-05,
      "loss": 0.7811,
      "step": 5707
    },
    {
      "epoch": 3.0644273682091203,
      "grad_norm": 0.79296875,
      "learning_rate": 1.633320955824529e-05,
      "loss": 0.883,
      "step": 5708
    },
    {
      "epoch": 3.0649642629441964,
      "grad_norm": 0.6796875,
      "learning_rate": 1.632528862361518e-05,
      "loss": 1.0312,
      "step": 5709
    },
    {
      "epoch": 3.0655011576792726,
      "grad_norm": 0.6640625,
      "learning_rate": 1.6317368678879495e-05,
      "loss": 0.7349,
      "step": 5710
    },
    {
      "epoch": 3.0660380524143487,
      "grad_norm": 0.62890625,
      "learning_rate": 1.6309449724942e-05,
      "loss": 0.82,
      "step": 5711
    },
    {
      "epoch": 3.0665749471494244,
      "grad_norm": 0.7734375,
      "learning_rate": 1.6301531762706346e-05,
      "loss": 0.8962,
      "step": 5712
    },
    {
      "epoch": 3.0671118418845005,
      "grad_norm": 0.69921875,
      "learning_rate": 1.629361479307608e-05,
      "loss": 0.8272,
      "step": 5713
    },
    {
      "epoch": 3.0676487366195766,
      "grad_norm": 0.56640625,
      "learning_rate": 1.6285698816954624e-05,
      "loss": 0.7819,
      "step": 5714
    },
    {
      "epoch": 3.0681856313546527,
      "grad_norm": 0.66015625,
      "learning_rate": 1.6277783835245293e-05,
      "loss": 0.7803,
      "step": 5715
    },
    {
      "epoch": 3.0687225260897284,
      "grad_norm": 0.7109375,
      "learning_rate": 1.6269869848851273e-05,
      "loss": 1.0371,
      "step": 5716
    },
    {
      "epoch": 3.0692594208248045,
      "grad_norm": 0.65234375,
      "learning_rate": 1.6261956858675667e-05,
      "loss": 0.7819,
      "step": 5717
    },
    {
      "epoch": 3.0697963155598806,
      "grad_norm": 0.69140625,
      "learning_rate": 1.6254044865621442e-05,
      "loss": 0.8434,
      "step": 5718
    },
    {
      "epoch": 3.0703332102949563,
      "grad_norm": 0.671875,
      "learning_rate": 1.6246133870591455e-05,
      "loss": 0.8229,
      "step": 5719
    },
    {
      "epoch": 3.0708701050300324,
      "grad_norm": 0.734375,
      "learning_rate": 1.6238223874488452e-05,
      "loss": 0.9756,
      "step": 5720
    },
    {
      "epoch": 3.0714069997651086,
      "grad_norm": 0.72265625,
      "learning_rate": 1.623031487821507e-05,
      "loss": 0.9734,
      "step": 5721
    },
    {
      "epoch": 3.0719438945001847,
      "grad_norm": 0.66015625,
      "learning_rate": 1.622240688267382e-05,
      "loss": 0.9777,
      "step": 5722
    },
    {
      "epoch": 3.0724807892352604,
      "grad_norm": 0.6171875,
      "learning_rate": 1.62144998887671e-05,
      "loss": 0.7713,
      "step": 5723
    },
    {
      "epoch": 3.0730176839703365,
      "grad_norm": 0.57421875,
      "learning_rate": 1.6206593897397193e-05,
      "loss": 0.6509,
      "step": 5724
    },
    {
      "epoch": 3.0735545787054126,
      "grad_norm": 0.6953125,
      "learning_rate": 1.619868890946629e-05,
      "loss": 0.926,
      "step": 5725
    },
    {
      "epoch": 3.0740914734404887,
      "grad_norm": 0.71484375,
      "learning_rate": 1.6190784925876445e-05,
      "loss": 0.8084,
      "step": 5726
    },
    {
      "epoch": 3.0746283681755644,
      "grad_norm": 0.94921875,
      "learning_rate": 1.6182881947529593e-05,
      "loss": 0.8113,
      "step": 5727
    },
    {
      "epoch": 3.0751652629106405,
      "grad_norm": 0.63671875,
      "learning_rate": 1.6174979975327576e-05,
      "loss": 0.849,
      "step": 5728
    },
    {
      "epoch": 3.0757021576457166,
      "grad_norm": 0.71875,
      "learning_rate": 1.6167079010172102e-05,
      "loss": 0.8111,
      "step": 5729
    },
    {
      "epoch": 3.0762390523807928,
      "grad_norm": 0.6640625,
      "learning_rate": 1.615917905296477e-05,
      "loss": 0.754,
      "step": 5730
    },
    {
      "epoch": 3.0767759471158684,
      "grad_norm": 0.64453125,
      "learning_rate": 1.615128010460708e-05,
      "loss": 0.8334,
      "step": 5731
    },
    {
      "epoch": 3.0773128418509446,
      "grad_norm": 0.54296875,
      "learning_rate": 1.6143382166000375e-05,
      "loss": 0.6678,
      "step": 5732
    },
    {
      "epoch": 3.0778497365860207,
      "grad_norm": 0.6328125,
      "learning_rate": 1.6135485238045932e-05,
      "loss": 0.7403,
      "step": 5733
    },
    {
      "epoch": 3.078386631321097,
      "grad_norm": 0.76171875,
      "learning_rate": 1.6127589321644883e-05,
      "loss": 0.806,
      "step": 5734
    },
    {
      "epoch": 3.0789235260561725,
      "grad_norm": 0.71875,
      "learning_rate": 1.6119694417698246e-05,
      "loss": 0.8301,
      "step": 5735
    },
    {
      "epoch": 3.0794604207912486,
      "grad_norm": 0.72265625,
      "learning_rate": 1.611180052710694e-05,
      "loss": 0.7907,
      "step": 5736
    },
    {
      "epoch": 3.0799973155263247,
      "grad_norm": 0.671875,
      "learning_rate": 1.6103907650771757e-05,
      "loss": 0.8658,
      "step": 5737
    },
    {
      "epoch": 3.0805342102614004,
      "grad_norm": 0.65625,
      "learning_rate": 1.609601578959337e-05,
      "loss": 0.8728,
      "step": 5738
    },
    {
      "epoch": 3.0810711049964765,
      "grad_norm": 0.6484375,
      "learning_rate": 1.6088124944472343e-05,
      "loss": 0.8796,
      "step": 5739
    },
    {
      "epoch": 3.0816079997315526,
      "grad_norm": 0.60546875,
      "learning_rate": 1.6080235116309114e-05,
      "loss": 0.7863,
      "step": 5740
    },
    {
      "epoch": 3.0821448944666288,
      "grad_norm": 0.71484375,
      "learning_rate": 1.6072346306004023e-05,
      "loss": 0.6969,
      "step": 5741
    },
    {
      "epoch": 3.0826817892017044,
      "grad_norm": 0.7109375,
      "learning_rate": 1.6064458514457275e-05,
      "loss": 0.8227,
      "step": 5742
    },
    {
      "epoch": 3.0832186839367806,
      "grad_norm": 0.7890625,
      "learning_rate": 1.6056571742568972e-05,
      "loss": 0.9655,
      "step": 5743
    },
    {
      "epoch": 3.0837555786718567,
      "grad_norm": 0.55078125,
      "learning_rate": 1.6048685991239087e-05,
      "loss": 0.8102,
      "step": 5744
    },
    {
      "epoch": 3.084292473406933,
      "grad_norm": 0.63671875,
      "learning_rate": 1.6040801261367493e-05,
      "loss": 0.8922,
      "step": 5745
    },
    {
      "epoch": 3.0848293681420085,
      "grad_norm": 0.671875,
      "learning_rate": 1.6032917553853936e-05,
      "loss": 0.8152,
      "step": 5746
    },
    {
      "epoch": 3.0853662628770846,
      "grad_norm": 0.59765625,
      "learning_rate": 1.602503486959805e-05,
      "loss": 0.7547,
      "step": 5747
    },
    {
      "epoch": 3.0859031576121607,
      "grad_norm": 0.6640625,
      "learning_rate": 1.601715320949934e-05,
      "loss": 0.8982,
      "step": 5748
    },
    {
      "epoch": 3.086440052347237,
      "grad_norm": 0.7109375,
      "learning_rate": 1.6009272574457202e-05,
      "loss": 0.8848,
      "step": 5749
    },
    {
      "epoch": 3.0869769470823125,
      "grad_norm": 0.73828125,
      "learning_rate": 1.6001392965370925e-05,
      "loss": 0.7046,
      "step": 5750
    },
    {
      "epoch": 3.0869769470823125,
      "eval_loss": 0.9032167196273804,
      "eval_runtime": 46.6926,
      "eval_samples_per_second": 20.603,
      "eval_steps_per_second": 20.603,
      "step": 5750
    },
    {
      "epoch": 3.0875138418173886,
      "grad_norm": 0.69921875,
      "learning_rate": 1.5993514383139667e-05,
      "loss": 0.8378,
      "step": 5751
    },
    {
      "epoch": 3.0880507365524648,
      "grad_norm": 0.89453125,
      "learning_rate": 1.5985636828662476e-05,
      "loss": 1.0233,
      "step": 5752
    },
    {
      "epoch": 3.088587631287541,
      "grad_norm": 0.65625,
      "learning_rate": 1.5977760302838282e-05,
      "loss": 1.1289,
      "step": 5753
    },
    {
      "epoch": 3.0891245260226166,
      "grad_norm": 0.62109375,
      "learning_rate": 1.5969884806565895e-05,
      "loss": 0.6874,
      "step": 5754
    },
    {
      "epoch": 3.0896614207576927,
      "grad_norm": 0.69921875,
      "learning_rate": 1.5962010340744005e-05,
      "loss": 0.786,
      "step": 5755
    },
    {
      "epoch": 3.090198315492769,
      "grad_norm": 0.66796875,
      "learning_rate": 1.5954136906271205e-05,
      "loss": 0.7122,
      "step": 5756
    },
    {
      "epoch": 3.0907352102278445,
      "grad_norm": 0.640625,
      "learning_rate": 1.5946264504045927e-05,
      "loss": 1.106,
      "step": 5757
    },
    {
      "epoch": 3.0912721049629206,
      "grad_norm": 0.62890625,
      "learning_rate": 1.5938393134966535e-05,
      "loss": 0.9231,
      "step": 5758
    },
    {
      "epoch": 3.0918089996979967,
      "grad_norm": 0.6796875,
      "learning_rate": 1.5930522799931237e-05,
      "loss": 0.8559,
      "step": 5759
    },
    {
      "epoch": 3.092345894433073,
      "grad_norm": 0.7578125,
      "learning_rate": 1.5922653499838137e-05,
      "loss": 0.9821,
      "step": 5760
    },
    {
      "epoch": 3.0928827891681485,
      "grad_norm": 0.60546875,
      "learning_rate": 1.5914785235585234e-05,
      "loss": 0.8043,
      "step": 5761
    },
    {
      "epoch": 3.0934196839032246,
      "grad_norm": 0.70703125,
      "learning_rate": 1.5906918008070392e-05,
      "loss": 0.8762,
      "step": 5762
    },
    {
      "epoch": 3.0939565786383008,
      "grad_norm": 0.69140625,
      "learning_rate": 1.589905181819135e-05,
      "loss": 0.7984,
      "step": 5763
    },
    {
      "epoch": 3.094493473373377,
      "grad_norm": 0.6953125,
      "learning_rate": 1.5891186666845773e-05,
      "loss": 0.9015,
      "step": 5764
    },
    {
      "epoch": 3.0950303681084526,
      "grad_norm": 0.71875,
      "learning_rate": 1.5883322554931133e-05,
      "loss": 0.8705,
      "step": 5765
    },
    {
      "epoch": 3.0955672628435287,
      "grad_norm": 0.62109375,
      "learning_rate": 1.5875459483344845e-05,
      "loss": 0.6684,
      "step": 5766
    },
    {
      "epoch": 3.096104157578605,
      "grad_norm": 0.61328125,
      "learning_rate": 1.5867597452984183e-05,
      "loss": 0.7554,
      "step": 5767
    },
    {
      "epoch": 3.096641052313681,
      "grad_norm": 0.859375,
      "learning_rate": 1.5859736464746306e-05,
      "loss": 0.8085,
      "step": 5768
    },
    {
      "epoch": 3.0971779470487566,
      "grad_norm": 0.63671875,
      "learning_rate": 1.585187651952825e-05,
      "loss": 0.778,
      "step": 5769
    },
    {
      "epoch": 3.0977148417838327,
      "grad_norm": 0.6484375,
      "learning_rate": 1.5844017618226935e-05,
      "loss": 0.8396,
      "step": 5770
    },
    {
      "epoch": 3.098251736518909,
      "grad_norm": 0.6953125,
      "learning_rate": 1.5836159761739156e-05,
      "loss": 0.8293,
      "step": 5771
    },
    {
      "epoch": 3.098788631253985,
      "grad_norm": 0.7265625,
      "learning_rate": 1.58283029509616e-05,
      "loss": 0.8152,
      "step": 5772
    },
    {
      "epoch": 3.0993255259890606,
      "grad_norm": 0.70703125,
      "learning_rate": 1.5820447186790837e-05,
      "loss": 0.7627,
      "step": 5773
    },
    {
      "epoch": 3.0998624207241368,
      "grad_norm": 0.65625,
      "learning_rate": 1.5812592470123296e-05,
      "loss": 0.7837,
      "step": 5774
    },
    {
      "epoch": 3.100399315459213,
      "grad_norm": 0.67578125,
      "learning_rate": 1.5804738801855302e-05,
      "loss": 0.7875,
      "step": 5775
    },
    {
      "epoch": 3.1009362101942886,
      "grad_norm": 0.65625,
      "learning_rate": 1.5796886182883053e-05,
      "loss": 0.858,
      "step": 5776
    },
    {
      "epoch": 3.1014731049293647,
      "grad_norm": 0.72265625,
      "learning_rate": 1.5789034614102645e-05,
      "loss": 0.8946,
      "step": 5777
    },
    {
      "epoch": 3.102009999664441,
      "grad_norm": 0.8984375,
      "learning_rate": 1.5781184096410032e-05,
      "loss": 1.16,
      "step": 5778
    },
    {
      "epoch": 3.102546894399517,
      "grad_norm": 0.6640625,
      "learning_rate": 1.577333463070106e-05,
      "loss": 0.7316,
      "step": 5779
    },
    {
      "epoch": 3.1030837891345926,
      "grad_norm": 0.6484375,
      "learning_rate": 1.576548621787145e-05,
      "loss": 0.8395,
      "step": 5780
    },
    {
      "epoch": 3.1036206838696687,
      "grad_norm": 0.7109375,
      "learning_rate": 1.575763885881682e-05,
      "loss": 1.0934,
      "step": 5781
    },
    {
      "epoch": 3.104157578604745,
      "grad_norm": 0.69140625,
      "learning_rate": 1.5749792554432637e-05,
      "loss": 0.7515,
      "step": 5782
    },
    {
      "epoch": 3.104694473339821,
      "grad_norm": 0.68359375,
      "learning_rate": 1.5741947305614262e-05,
      "loss": 0.8488,
      "step": 5783
    },
    {
      "epoch": 3.1052313680748966,
      "grad_norm": 0.60546875,
      "learning_rate": 1.573410311325695e-05,
      "loss": 0.915,
      "step": 5784
    },
    {
      "epoch": 3.1057682628099728,
      "grad_norm": 0.71484375,
      "learning_rate": 1.572625997825581e-05,
      "loss": 0.8477,
      "step": 5785
    },
    {
      "epoch": 3.106305157545049,
      "grad_norm": 0.56640625,
      "learning_rate": 1.571841790150585e-05,
      "loss": 0.807,
      "step": 5786
    },
    {
      "epoch": 3.106842052280125,
      "grad_norm": 0.546875,
      "learning_rate": 1.571057688390195e-05,
      "loss": 0.747,
      "step": 5787
    },
    {
      "epoch": 3.1073789470152007,
      "grad_norm": 0.78125,
      "learning_rate": 1.5702736926338863e-05,
      "loss": 1.0878,
      "step": 5788
    },
    {
      "epoch": 3.107915841750277,
      "grad_norm": 0.70703125,
      "learning_rate": 1.5694898029711248e-05,
      "loss": 0.7316,
      "step": 5789
    },
    {
      "epoch": 3.108452736485353,
      "grad_norm": 0.875,
      "learning_rate": 1.5687060194913596e-05,
      "loss": 0.8267,
      "step": 5790
    },
    {
      "epoch": 3.108989631220429,
      "grad_norm": 0.6875,
      "learning_rate": 1.5679223422840318e-05,
      "loss": 0.9023,
      "step": 5791
    },
    {
      "epoch": 3.1095265259555047,
      "grad_norm": 0.7734375,
      "learning_rate": 1.5671387714385682e-05,
      "loss": 1.0515,
      "step": 5792
    },
    {
      "epoch": 3.110063420690581,
      "grad_norm": 0.609375,
      "learning_rate": 1.5663553070443846e-05,
      "loss": 0.6579,
      "step": 5793
    },
    {
      "epoch": 3.110600315425657,
      "grad_norm": 0.7734375,
      "learning_rate": 1.5655719491908843e-05,
      "loss": 0.8048,
      "step": 5794
    },
    {
      "epoch": 3.1111372101607326,
      "grad_norm": 0.69921875,
      "learning_rate": 1.564788697967458e-05,
      "loss": 0.9001,
      "step": 5795
    },
    {
      "epoch": 3.1116741048958088,
      "grad_norm": 0.71875,
      "learning_rate": 1.5640055534634846e-05,
      "loss": 0.9241,
      "step": 5796
    },
    {
      "epoch": 3.112210999630885,
      "grad_norm": 0.64453125,
      "learning_rate": 1.5632225157683312e-05,
      "loss": 0.8496,
      "step": 5797
    },
    {
      "epoch": 3.112747894365961,
      "grad_norm": 0.67578125,
      "learning_rate": 1.5624395849713527e-05,
      "loss": 0.9764,
      "step": 5798
    },
    {
      "epoch": 3.1132847891010367,
      "grad_norm": 0.75,
      "learning_rate": 1.5616567611618905e-05,
      "loss": 0.8557,
      "step": 5799
    },
    {
      "epoch": 3.113821683836113,
      "grad_norm": 0.70703125,
      "learning_rate": 1.5608740444292746e-05,
      "loss": 0.7297,
      "step": 5800
    },
    {
      "epoch": 3.113821683836113,
      "eval_loss": 0.9028341770172119,
      "eval_runtime": 46.5022,
      "eval_samples_per_second": 20.687,
      "eval_steps_per_second": 20.687,
      "step": 5800
    },
    {
      "epoch": 3.114358578571189,
      "grad_norm": 1.0078125,
      "learning_rate": 1.560091434862823e-05,
      "loss": 0.9834,
      "step": 5801
    },
    {
      "epoch": 3.114895473306265,
      "grad_norm": 0.62890625,
      "learning_rate": 1.5593089325518422e-05,
      "loss": 0.9395,
      "step": 5802
    },
    {
      "epoch": 3.1154323680413407,
      "grad_norm": 0.62109375,
      "learning_rate": 1.558526537585625e-05,
      "loss": 0.8744,
      "step": 5803
    },
    {
      "epoch": 3.115969262776417,
      "grad_norm": 0.7109375,
      "learning_rate": 1.5577442500534522e-05,
      "loss": 0.9399,
      "step": 5804
    },
    {
      "epoch": 3.116506157511493,
      "grad_norm": 0.71484375,
      "learning_rate": 1.5569620700445936e-05,
      "loss": 0.7778,
      "step": 5805
    },
    {
      "epoch": 3.117043052246569,
      "grad_norm": 0.70703125,
      "learning_rate": 1.556179997648306e-05,
      "loss": 0.9556,
      "step": 5806
    },
    {
      "epoch": 3.1175799469816448,
      "grad_norm": 0.61328125,
      "learning_rate": 1.5553980329538326e-05,
      "loss": 0.8476,
      "step": 5807
    },
    {
      "epoch": 3.118116841716721,
      "grad_norm": 0.78125,
      "learning_rate": 1.5546161760504063e-05,
      "loss": 0.8915,
      "step": 5808
    },
    {
      "epoch": 3.118653736451797,
      "grad_norm": 0.60546875,
      "learning_rate": 1.553834427027246e-05,
      "loss": 0.7585,
      "step": 5809
    },
    {
      "epoch": 3.119190631186873,
      "grad_norm": 0.734375,
      "learning_rate": 1.55305278597356e-05,
      "loss": 1.0599,
      "step": 5810
    },
    {
      "epoch": 3.119727525921949,
      "grad_norm": 0.6171875,
      "learning_rate": 1.552271252978544e-05,
      "loss": 0.7973,
      "step": 5811
    },
    {
      "epoch": 3.120264420657025,
      "grad_norm": 0.56640625,
      "learning_rate": 1.5514898281313785e-05,
      "loss": 0.786,
      "step": 5812
    },
    {
      "epoch": 3.120801315392101,
      "grad_norm": 0.6953125,
      "learning_rate": 1.5507085115212366e-05,
      "loss": 0.8865,
      "step": 5813
    },
    {
      "epoch": 3.1213382101271767,
      "grad_norm": 0.65625,
      "learning_rate": 1.549927303237275e-05,
      "loss": 0.9879,
      "step": 5814
    },
    {
      "epoch": 3.121875104862253,
      "grad_norm": 0.703125,
      "learning_rate": 1.549146203368641e-05,
      "loss": 0.7729,
      "step": 5815
    },
    {
      "epoch": 3.122411999597329,
      "grad_norm": 0.79296875,
      "learning_rate": 1.5483652120044658e-05,
      "loss": 1.182,
      "step": 5816
    },
    {
      "epoch": 3.122948894332405,
      "grad_norm": 0.6953125,
      "learning_rate": 1.547584329233871e-05,
      "loss": 0.8433,
      "step": 5817
    },
    {
      "epoch": 3.1234857890674808,
      "grad_norm": 0.78125,
      "learning_rate": 1.5468035551459663e-05,
      "loss": 0.7269,
      "step": 5818
    },
    {
      "epoch": 3.124022683802557,
      "grad_norm": 0.765625,
      "learning_rate": 1.546022889829847e-05,
      "loss": 0.9403,
      "step": 5819
    },
    {
      "epoch": 3.124559578537633,
      "grad_norm": 0.64453125,
      "learning_rate": 1.5452423333745964e-05,
      "loss": 0.6856,
      "step": 5820
    },
    {
      "epoch": 3.125096473272709,
      "grad_norm": 0.72265625,
      "learning_rate": 1.544461885869287e-05,
      "loss": 0.9602,
      "step": 5821
    },
    {
      "epoch": 3.125633368007785,
      "grad_norm": 0.66015625,
      "learning_rate": 1.5436815474029777e-05,
      "loss": 0.9313,
      "step": 5822
    },
    {
      "epoch": 3.126170262742861,
      "grad_norm": 0.625,
      "learning_rate": 1.542901318064715e-05,
      "loss": 0.8657,
      "step": 5823
    },
    {
      "epoch": 3.126707157477937,
      "grad_norm": 0.77734375,
      "learning_rate": 1.5421211979435324e-05,
      "loss": 0.8971,
      "step": 5824
    },
    {
      "epoch": 3.127244052213013,
      "grad_norm": 0.82421875,
      "learning_rate": 1.5413411871284514e-05,
      "loss": 0.9795,
      "step": 5825
    },
    {
      "epoch": 3.127780946948089,
      "grad_norm": 0.73046875,
      "learning_rate": 1.540561285708481e-05,
      "loss": 0.8813,
      "step": 5826
    },
    {
      "epoch": 3.128317841683165,
      "grad_norm": 0.7578125,
      "learning_rate": 1.5397814937726186e-05,
      "loss": 0.8421,
      "step": 5827
    },
    {
      "epoch": 3.128854736418241,
      "grad_norm": 0.609375,
      "learning_rate": 1.5390018114098475e-05,
      "loss": 0.6864,
      "step": 5828
    },
    {
      "epoch": 3.129391631153317,
      "grad_norm": 0.7734375,
      "learning_rate": 1.53822223870914e-05,
      "loss": 0.8245,
      "step": 5829
    },
    {
      "epoch": 3.129928525888393,
      "grad_norm": 0.76171875,
      "learning_rate": 1.5374427757594552e-05,
      "loss": 0.9671,
      "step": 5830
    },
    {
      "epoch": 3.130465420623469,
      "grad_norm": 0.671875,
      "learning_rate": 1.5366634226497402e-05,
      "loss": 0.8747,
      "step": 5831
    },
    {
      "epoch": 3.131002315358545,
      "grad_norm": 0.61328125,
      "learning_rate": 1.5358841794689274e-05,
      "loss": 0.8506,
      "step": 5832
    },
    {
      "epoch": 3.131539210093621,
      "grad_norm": 0.671875,
      "learning_rate": 1.5351050463059396e-05,
      "loss": 0.8789,
      "step": 5833
    },
    {
      "epoch": 3.132076104828697,
      "grad_norm": 0.66015625,
      "learning_rate": 1.534326023249685e-05,
      "loss": 0.8538,
      "step": 5834
    },
    {
      "epoch": 3.132612999563773,
      "grad_norm": 0.6640625,
      "learning_rate": 1.5335471103890603e-05,
      "loss": 0.7922,
      "step": 5835
    },
    {
      "epoch": 3.133149894298849,
      "grad_norm": 0.65625,
      "learning_rate": 1.53276830781295e-05,
      "loss": 0.7255,
      "step": 5836
    },
    {
      "epoch": 3.133686789033925,
      "grad_norm": 0.83984375,
      "learning_rate": 1.5319896156102244e-05,
      "loss": 0.772,
      "step": 5837
    },
    {
      "epoch": 3.134223683769001,
      "grad_norm": 0.75390625,
      "learning_rate": 1.5312110338697426e-05,
      "loss": 0.928,
      "step": 5838
    },
    {
      "epoch": 3.134760578504077,
      "grad_norm": 0.6328125,
      "learning_rate": 1.5304325626803507e-05,
      "loss": 0.6858,
      "step": 5839
    },
    {
      "epoch": 3.135297473239153,
      "grad_norm": 0.6640625,
      "learning_rate": 1.5296542021308825e-05,
      "loss": 0.9071,
      "step": 5840
    },
    {
      "epoch": 3.135834367974229,
      "grad_norm": 0.65625,
      "learning_rate": 1.5288759523101578e-05,
      "loss": 0.9167,
      "step": 5841
    },
    {
      "epoch": 3.136371262709305,
      "grad_norm": 0.6875,
      "learning_rate": 1.5280978133069846e-05,
      "loss": 0.8813,
      "step": 5842
    },
    {
      "epoch": 3.136908157444381,
      "grad_norm": 0.64453125,
      "learning_rate": 1.52731978521016e-05,
      "loss": 0.7688,
      "step": 5843
    },
    {
      "epoch": 3.1374450521794572,
      "grad_norm": 0.66015625,
      "learning_rate": 1.5265418681084655e-05,
      "loss": 0.917,
      "step": 5844
    },
    {
      "epoch": 3.137981946914533,
      "grad_norm": 0.69140625,
      "learning_rate": 1.525764062090671e-05,
      "loss": 0.7868,
      "step": 5845
    },
    {
      "epoch": 3.138518841649609,
      "grad_norm": 0.703125,
      "learning_rate": 1.5249863672455355e-05,
      "loss": 0.7626,
      "step": 5846
    },
    {
      "epoch": 3.139055736384685,
      "grad_norm": 0.71875,
      "learning_rate": 1.524208783661803e-05,
      "loss": 0.8455,
      "step": 5847
    },
    {
      "epoch": 3.1395926311197613,
      "grad_norm": 0.57421875,
      "learning_rate": 1.5234313114282061e-05,
      "loss": 0.7365,
      "step": 5848
    },
    {
      "epoch": 3.140129525854837,
      "grad_norm": 0.7109375,
      "learning_rate": 1.5226539506334636e-05,
      "loss": 0.9351,
      "step": 5849
    },
    {
      "epoch": 3.140666420589913,
      "grad_norm": 0.71484375,
      "learning_rate": 1.5218767013662816e-05,
      "loss": 0.7948,
      "step": 5850
    },
    {
      "epoch": 3.140666420589913,
      "eval_loss": 0.9028377532958984,
      "eval_runtime": 46.8928,
      "eval_samples_per_second": 20.515,
      "eval_steps_per_second": 20.515,
      "step": 5850
    },
    {
      "epoch": 3.141203315324989,
      "grad_norm": 0.6015625,
      "learning_rate": 1.5210995637153558e-05,
      "loss": 0.7551,
      "step": 5851
    },
    {
      "epoch": 3.141740210060065,
      "grad_norm": 0.625,
      "learning_rate": 1.5203225377693662e-05,
      "loss": 0.8331,
      "step": 5852
    },
    {
      "epoch": 3.142277104795141,
      "grad_norm": 0.6953125,
      "learning_rate": 1.5195456236169811e-05,
      "loss": 0.6731,
      "step": 5853
    },
    {
      "epoch": 3.142813999530217,
      "grad_norm": 0.67578125,
      "learning_rate": 1.5187688213468577e-05,
      "loss": 0.8433,
      "step": 5854
    },
    {
      "epoch": 3.1433508942652932,
      "grad_norm": 0.80078125,
      "learning_rate": 1.5179921310476375e-05,
      "loss": 0.7952,
      "step": 5855
    },
    {
      "epoch": 3.143887789000369,
      "grad_norm": 0.71484375,
      "learning_rate": 1.5172155528079512e-05,
      "loss": 0.8765,
      "step": 5856
    },
    {
      "epoch": 3.144424683735445,
      "grad_norm": 0.69140625,
      "learning_rate": 1.5164390867164177e-05,
      "loss": 0.9628,
      "step": 5857
    },
    {
      "epoch": 3.144961578470521,
      "grad_norm": 0.9140625,
      "learning_rate": 1.5156627328616385e-05,
      "loss": 0.7196,
      "step": 5858
    },
    {
      "epoch": 3.1454984732055973,
      "grad_norm": 0.66015625,
      "learning_rate": 1.5148864913322081e-05,
      "loss": 0.8912,
      "step": 5859
    },
    {
      "epoch": 3.146035367940673,
      "grad_norm": 0.828125,
      "learning_rate": 1.5141103622167041e-05,
      "loss": 1.2283,
      "step": 5860
    },
    {
      "epoch": 3.146572262675749,
      "grad_norm": 0.64453125,
      "learning_rate": 1.5133343456036925e-05,
      "loss": 0.6317,
      "step": 5861
    },
    {
      "epoch": 3.147109157410825,
      "grad_norm": 0.7265625,
      "learning_rate": 1.5125584415817279e-05,
      "loss": 0.7639,
      "step": 5862
    },
    {
      "epoch": 3.1476460521459013,
      "grad_norm": 0.61328125,
      "learning_rate": 1.5117826502393499e-05,
      "loss": 0.8705,
      "step": 5863
    },
    {
      "epoch": 3.148182946880977,
      "grad_norm": 0.6796875,
      "learning_rate": 1.511006971665086e-05,
      "loss": 0.9229,
      "step": 5864
    },
    {
      "epoch": 3.148719841616053,
      "grad_norm": 0.71875,
      "learning_rate": 1.5102314059474521e-05,
      "loss": 0.881,
      "step": 5865
    },
    {
      "epoch": 3.1492567363511292,
      "grad_norm": 0.6875,
      "learning_rate": 1.509455953174948e-05,
      "loss": 0.8269,
      "step": 5866
    },
    {
      "epoch": 3.1497936310862054,
      "grad_norm": 0.640625,
      "learning_rate": 1.5086806134360648e-05,
      "loss": 0.7855,
      "step": 5867
    },
    {
      "epoch": 3.150330525821281,
      "grad_norm": 0.734375,
      "learning_rate": 1.5079053868192771e-05,
      "loss": 0.9143,
      "step": 5868
    },
    {
      "epoch": 3.150867420556357,
      "grad_norm": 0.65234375,
      "learning_rate": 1.5071302734130489e-05,
      "loss": 0.7661,
      "step": 5869
    },
    {
      "epoch": 3.1514043152914333,
      "grad_norm": 0.625,
      "learning_rate": 1.5063552733058294e-05,
      "loss": 0.774,
      "step": 5870
    },
    {
      "epoch": 3.151941210026509,
      "grad_norm": 0.62890625,
      "learning_rate": 1.5055803865860576e-05,
      "loss": 0.7571,
      "step": 5871
    },
    {
      "epoch": 3.152478104761585,
      "grad_norm": 0.65625,
      "learning_rate": 1.5048056133421567e-05,
      "loss": 0.746,
      "step": 5872
    },
    {
      "epoch": 3.153014999496661,
      "grad_norm": 0.6875,
      "learning_rate": 1.5040309536625391e-05,
      "loss": 0.7903,
      "step": 5873
    },
    {
      "epoch": 3.1535518942317373,
      "grad_norm": 0.5859375,
      "learning_rate": 1.5032564076356023e-05,
      "loss": 0.695,
      "step": 5874
    },
    {
      "epoch": 3.154088788966813,
      "grad_norm": 0.62109375,
      "learning_rate": 1.5024819753497316e-05,
      "loss": 0.885,
      "step": 5875
    },
    {
      "epoch": 3.154625683701889,
      "grad_norm": 0.7890625,
      "learning_rate": 1.5017076568933003e-05,
      "loss": 0.97,
      "step": 5876
    },
    {
      "epoch": 3.1551625784369652,
      "grad_norm": 0.984375,
      "learning_rate": 1.5009334523546675e-05,
      "loss": 0.948,
      "step": 5877
    },
    {
      "epoch": 3.1556994731720414,
      "grad_norm": 0.9296875,
      "learning_rate": 1.5001593618221799e-05,
      "loss": 0.9747,
      "step": 5878
    },
    {
      "epoch": 3.156236367907117,
      "grad_norm": 0.77734375,
      "learning_rate": 1.4993853853841713e-05,
      "loss": 0.742,
      "step": 5879
    },
    {
      "epoch": 3.156773262642193,
      "grad_norm": 0.66015625,
      "learning_rate": 1.4986115231289622e-05,
      "loss": 0.7533,
      "step": 5880
    },
    {
      "epoch": 3.1573101573772693,
      "grad_norm": 0.6484375,
      "learning_rate": 1.4978377751448592e-05,
      "loss": 0.8585,
      "step": 5881
    },
    {
      "epoch": 3.1578470521123454,
      "grad_norm": 0.953125,
      "learning_rate": 1.4970641415201589e-05,
      "loss": 0.8325,
      "step": 5882
    },
    {
      "epoch": 3.158383946847421,
      "grad_norm": 0.80859375,
      "learning_rate": 1.4962906223431398e-05,
      "loss": 1.0475,
      "step": 5883
    },
    {
      "epoch": 3.158920841582497,
      "grad_norm": 0.7421875,
      "learning_rate": 1.4955172177020722e-05,
      "loss": 0.8778,
      "step": 5884
    },
    {
      "epoch": 3.1594577363175733,
      "grad_norm": 0.6484375,
      "learning_rate": 1.4947439276852104e-05,
      "loss": 0.71,
      "step": 5885
    },
    {
      "epoch": 3.1599946310526494,
      "grad_norm": 0.7265625,
      "learning_rate": 1.493970752380797e-05,
      "loss": 0.8112,
      "step": 5886
    },
    {
      "epoch": 3.160531525787725,
      "grad_norm": 0.70703125,
      "learning_rate": 1.4931976918770613e-05,
      "loss": 0.7708,
      "step": 5887
    },
    {
      "epoch": 3.1610684205228012,
      "grad_norm": 0.74609375,
      "learning_rate": 1.4924247462622193e-05,
      "loss": 0.8073,
      "step": 5888
    },
    {
      "epoch": 3.1616053152578774,
      "grad_norm": 0.76953125,
      "learning_rate": 1.491651915624473e-05,
      "loss": 0.8567,
      "step": 5889
    },
    {
      "epoch": 3.162142209992953,
      "grad_norm": 0.71484375,
      "learning_rate": 1.4908792000520141e-05,
      "loss": 0.922,
      "step": 5890
    },
    {
      "epoch": 3.162679104728029,
      "grad_norm": 0.70703125,
      "learning_rate": 1.4901065996330166e-05,
      "loss": 0.8179,
      "step": 5891
    },
    {
      "epoch": 3.1632159994631053,
      "grad_norm": 0.7265625,
      "learning_rate": 1.4893341144556461e-05,
      "loss": 0.8959,
      "step": 5892
    },
    {
      "epoch": 3.1637528941981814,
      "grad_norm": 0.66796875,
      "learning_rate": 1.4885617446080518e-05,
      "loss": 0.7958,
      "step": 5893
    },
    {
      "epoch": 3.164289788933257,
      "grad_norm": 0.71875,
      "learning_rate": 1.4877894901783712e-05,
      "loss": 0.7645,
      "step": 5894
    },
    {
      "epoch": 3.164826683668333,
      "grad_norm": 0.81640625,
      "learning_rate": 1.4870173512547287e-05,
      "loss": 1.0871,
      "step": 5895
    },
    {
      "epoch": 3.1653635784034093,
      "grad_norm": 0.58203125,
      "learning_rate": 1.4862453279252348e-05,
      "loss": 0.8,
      "step": 5896
    },
    {
      "epoch": 3.1659004731384854,
      "grad_norm": 0.640625,
      "learning_rate": 1.4854734202779865e-05,
      "loss": 0.7329,
      "step": 5897
    },
    {
      "epoch": 3.166437367873561,
      "grad_norm": 0.6484375,
      "learning_rate": 1.4847016284010695e-05,
      "loss": 0.7705,
      "step": 5898
    },
    {
      "epoch": 3.1669742626086372,
      "grad_norm": 0.60546875,
      "learning_rate": 1.483929952382555e-05,
      "loss": 0.7396,
      "step": 5899
    },
    {
      "epoch": 3.1675111573437134,
      "grad_norm": 0.83984375,
      "learning_rate": 1.4831583923104999e-05,
      "loss": 0.7877,
      "step": 5900
    },
    {
      "epoch": 3.1675111573437134,
      "eval_loss": 0.9030345678329468,
      "eval_runtime": 46.1703,
      "eval_samples_per_second": 20.836,
      "eval_steps_per_second": 20.836,
      "step": 5900
    },
    {
      "epoch": 3.1680480520787895,
      "grad_norm": 0.82421875,
      "learning_rate": 1.4823869482729493e-05,
      "loss": 0.8308,
      "step": 5901
    },
    {
      "epoch": 3.168584946813865,
      "grad_norm": 0.5703125,
      "learning_rate": 1.4816156203579346e-05,
      "loss": 0.7827,
      "step": 5902
    },
    {
      "epoch": 3.1691218415489413,
      "grad_norm": 0.66796875,
      "learning_rate": 1.480844408653475e-05,
      "loss": 0.7977,
      "step": 5903
    },
    {
      "epoch": 3.1696587362840174,
      "grad_norm": 0.6015625,
      "learning_rate": 1.4800733132475747e-05,
      "loss": 0.8653,
      "step": 5904
    },
    {
      "epoch": 3.1701956310190935,
      "grad_norm": 0.80078125,
      "learning_rate": 1.4793023342282258e-05,
      "loss": 0.7437,
      "step": 5905
    },
    {
      "epoch": 3.170732525754169,
      "grad_norm": 0.83984375,
      "learning_rate": 1.4785314716834065e-05,
      "loss": 0.9841,
      "step": 5906
    },
    {
      "epoch": 3.1712694204892453,
      "grad_norm": 0.65234375,
      "learning_rate": 1.4777607257010826e-05,
      "loss": 0.8282,
      "step": 5907
    },
    {
      "epoch": 3.1718063152243214,
      "grad_norm": 0.79296875,
      "learning_rate": 1.4769900963692051e-05,
      "loss": 0.8326,
      "step": 5908
    },
    {
      "epoch": 3.172343209959397,
      "grad_norm": 0.69921875,
      "learning_rate": 1.476219583775713e-05,
      "loss": 0.9247,
      "step": 5909
    },
    {
      "epoch": 3.1728801046944732,
      "grad_norm": 0.72265625,
      "learning_rate": 1.475449188008532e-05,
      "loss": 1.0606,
      "step": 5910
    },
    {
      "epoch": 3.1734169994295494,
      "grad_norm": 0.74609375,
      "learning_rate": 1.4746789091555724e-05,
      "loss": 0.7653,
      "step": 5911
    },
    {
      "epoch": 3.1739538941646255,
      "grad_norm": 0.72265625,
      "learning_rate": 1.4739087473047347e-05,
      "loss": 0.8635,
      "step": 5912
    },
    {
      "epoch": 3.174490788899701,
      "grad_norm": 0.7734375,
      "learning_rate": 1.473138702543903e-05,
      "loss": 0.8409,
      "step": 5913
    },
    {
      "epoch": 3.1750276836347773,
      "grad_norm": 0.82421875,
      "learning_rate": 1.4723687749609493e-05,
      "loss": 0.8368,
      "step": 5914
    },
    {
      "epoch": 3.1755645783698534,
      "grad_norm": 0.9609375,
      "learning_rate": 1.4715989646437338e-05,
      "loss": 0.9605,
      "step": 5915
    },
    {
      "epoch": 3.1761014731049295,
      "grad_norm": 0.62109375,
      "learning_rate": 1.4708292716800987e-05,
      "loss": 0.7128,
      "step": 5916
    },
    {
      "epoch": 3.176638367840005,
      "grad_norm": 0.74609375,
      "learning_rate": 1.4700596961578771e-05,
      "loss": 0.8462,
      "step": 5917
    },
    {
      "epoch": 3.1771752625750813,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4692902381648877e-05,
      "loss": 0.8852,
      "step": 5918
    },
    {
      "epoch": 3.1777121573101574,
      "grad_norm": 0.8125,
      "learning_rate": 1.4685208977889344e-05,
      "loss": 0.7421,
      "step": 5919
    },
    {
      "epoch": 3.1782490520452336,
      "grad_norm": 0.66796875,
      "learning_rate": 1.46775167511781e-05,
      "loss": 0.7741,
      "step": 5920
    },
    {
      "epoch": 3.1787859467803092,
      "grad_norm": 0.61328125,
      "learning_rate": 1.4669825702392915e-05,
      "loss": 0.7951,
      "step": 5921
    },
    {
      "epoch": 3.1793228415153854,
      "grad_norm": 0.75,
      "learning_rate": 1.4662135832411436e-05,
      "loss": 0.9868,
      "step": 5922
    },
    {
      "epoch": 3.1798597362504615,
      "grad_norm": 0.703125,
      "learning_rate": 1.465444714211118e-05,
      "loss": 0.8983,
      "step": 5923
    },
    {
      "epoch": 3.1803966309855376,
      "grad_norm": 0.8359375,
      "learning_rate": 1.4646759632369528e-05,
      "loss": 0.9055,
      "step": 5924
    },
    {
      "epoch": 3.1809335257206133,
      "grad_norm": 0.61328125,
      "learning_rate": 1.4639073304063708e-05,
      "loss": 0.7099,
      "step": 5925
    },
    {
      "epoch": 3.1814704204556894,
      "grad_norm": 0.6328125,
      "learning_rate": 1.4631388158070836e-05,
      "loss": 0.8191,
      "step": 5926
    },
    {
      "epoch": 3.1820073151907655,
      "grad_norm": 0.87890625,
      "learning_rate": 1.4623704195267875e-05,
      "loss": 0.9476,
      "step": 5927
    },
    {
      "epoch": 3.182544209925841,
      "grad_norm": 0.703125,
      "learning_rate": 1.4616021416531678e-05,
      "loss": 0.8805,
      "step": 5928
    },
    {
      "epoch": 3.1830811046609173,
      "grad_norm": 0.6875,
      "learning_rate": 1.4608339822738942e-05,
      "loss": 0.8201,
      "step": 5929
    },
    {
      "epoch": 3.1836179993959934,
      "grad_norm": 0.71484375,
      "learning_rate": 1.4600659414766224e-05,
      "loss": 0.6948,
      "step": 5930
    },
    {
      "epoch": 3.1841548941310696,
      "grad_norm": 0.890625,
      "learning_rate": 1.4592980193489975e-05,
      "loss": 0.9722,
      "step": 5931
    },
    {
      "epoch": 3.1846917888661452,
      "grad_norm": 0.86328125,
      "learning_rate": 1.4585302159786478e-05,
      "loss": 0.9052,
      "step": 5932
    },
    {
      "epoch": 3.1852286836012214,
      "grad_norm": 0.71875,
      "learning_rate": 1.457762531453189e-05,
      "loss": 1.1633,
      "step": 5933
    },
    {
      "epoch": 3.1857655783362975,
      "grad_norm": 0.640625,
      "learning_rate": 1.4569949658602238e-05,
      "loss": 0.736,
      "step": 5934
    },
    {
      "epoch": 3.1863024730713736,
      "grad_norm": 0.5703125,
      "learning_rate": 1.4562275192873428e-05,
      "loss": 0.7624,
      "step": 5935
    },
    {
      "epoch": 3.1868393678064493,
      "grad_norm": 0.65234375,
      "learning_rate": 1.4554601918221192e-05,
      "loss": 0.7997,
      "step": 5936
    },
    {
      "epoch": 3.1873762625415254,
      "grad_norm": 0.77734375,
      "learning_rate": 1.4546929835521155e-05,
      "loss": 1.0569,
      "step": 5937
    },
    {
      "epoch": 3.1879131572766015,
      "grad_norm": 0.84375,
      "learning_rate": 1.4539258945648814e-05,
      "loss": 0.8691,
      "step": 5938
    },
    {
      "epoch": 3.1884500520116776,
      "grad_norm": 0.85546875,
      "learning_rate": 1.453158924947949e-05,
      "loss": 0.8963,
      "step": 5939
    },
    {
      "epoch": 3.1889869467467533,
      "grad_norm": 0.6484375,
      "learning_rate": 1.4523920747888406e-05,
      "loss": 0.7196,
      "step": 5940
    },
    {
      "epoch": 3.1895238414818294,
      "grad_norm": 0.609375,
      "learning_rate": 1.4516253441750655e-05,
      "loss": 0.7365,
      "step": 5941
    },
    {
      "epoch": 3.1900607362169056,
      "grad_norm": 0.78125,
      "learning_rate": 1.450858733194113e-05,
      "loss": 0.9657,
      "step": 5942
    },
    {
      "epoch": 3.1905976309519817,
      "grad_norm": 0.72265625,
      "learning_rate": 1.450092241933465e-05,
      "loss": 0.8072,
      "step": 5943
    },
    {
      "epoch": 3.1911345256870574,
      "grad_norm": 0.93359375,
      "learning_rate": 1.4493258704805896e-05,
      "loss": 0.8625,
      "step": 5944
    },
    {
      "epoch": 3.1916714204221335,
      "grad_norm": 0.73046875,
      "learning_rate": 1.4485596189229373e-05,
      "loss": 0.7525,
      "step": 5945
    },
    {
      "epoch": 3.1922083151572096,
      "grad_norm": 0.640625,
      "learning_rate": 1.4477934873479473e-05,
      "loss": 0.7938,
      "step": 5946
    },
    {
      "epoch": 3.1927452098922853,
      "grad_norm": 0.69140625,
      "learning_rate": 1.4470274758430468e-05,
      "loss": 0.8503,
      "step": 5947
    },
    {
      "epoch": 3.1932821046273614,
      "grad_norm": 0.6953125,
      "learning_rate": 1.4462615844956452e-05,
      "loss": 0.9595,
      "step": 5948
    },
    {
      "epoch": 3.1938189993624375,
      "grad_norm": 0.64453125,
      "learning_rate": 1.4454958133931423e-05,
      "loss": 0.7076,
      "step": 5949
    },
    {
      "epoch": 3.1943558940975136,
      "grad_norm": 0.7265625,
      "learning_rate": 1.4447301626229204e-05,
      "loss": 0.8918,
      "step": 5950
    },
    {
      "epoch": 3.1943558940975136,
      "eval_loss": 0.9028350114822388,
      "eval_runtime": 46.1726,
      "eval_samples_per_second": 20.835,
      "eval_steps_per_second": 20.835,
      "step": 5950
    },
    {
      "epoch": 3.1948927888325893,
      "grad_norm": 0.67578125,
      "learning_rate": 1.4439646322723516e-05,
      "loss": 0.895,
      "step": 5951
    },
    {
      "epoch": 3.1954296835676654,
      "grad_norm": 0.7578125,
      "learning_rate": 1.443199222428791e-05,
      "loss": 0.9363,
      "step": 5952
    },
    {
      "epoch": 3.1959665783027416,
      "grad_norm": 0.69140625,
      "learning_rate": 1.4424339331795824e-05,
      "loss": 0.9611,
      "step": 5953
    },
    {
      "epoch": 3.1965034730378177,
      "grad_norm": 0.671875,
      "learning_rate": 1.441668764612056e-05,
      "loss": 0.7588,
      "step": 5954
    },
    {
      "epoch": 3.1970403677728934,
      "grad_norm": 0.64453125,
      "learning_rate": 1.4409037168135254e-05,
      "loss": 0.8395,
      "step": 5955
    },
    {
      "epoch": 3.1975772625079695,
      "grad_norm": 0.60546875,
      "learning_rate": 1.4401387898712931e-05,
      "loss": 1.038,
      "step": 5956
    },
    {
      "epoch": 3.1981141572430456,
      "grad_norm": 0.7109375,
      "learning_rate": 1.4393739838726486e-05,
      "loss": 0.822,
      "step": 5957
    },
    {
      "epoch": 3.1986510519781217,
      "grad_norm": 0.67578125,
      "learning_rate": 1.4386092989048625e-05,
      "loss": 1.0552,
      "step": 5958
    },
    {
      "epoch": 3.1991879467131974,
      "grad_norm": 0.71484375,
      "learning_rate": 1.4378447350551968e-05,
      "loss": 0.789,
      "step": 5959
    },
    {
      "epoch": 3.1997248414482735,
      "grad_norm": 0.6328125,
      "learning_rate": 1.437080292410899e-05,
      "loss": 0.897,
      "step": 5960
    },
    {
      "epoch": 3.2002617361833496,
      "grad_norm": 0.6328125,
      "learning_rate": 1.4363159710591995e-05,
      "loss": 0.7468,
      "step": 5961
    },
    {
      "epoch": 3.2007986309184258,
      "grad_norm": 0.66796875,
      "learning_rate": 1.4355517710873184e-05,
      "loss": 0.8809,
      "step": 5962
    },
    {
      "epoch": 3.2013355256535014,
      "grad_norm": 0.8984375,
      "learning_rate": 1.434787692582461e-05,
      "loss": 0.8277,
      "step": 5963
    },
    {
      "epoch": 3.2018724203885776,
      "grad_norm": 0.75,
      "learning_rate": 1.434023735631817e-05,
      "loss": 0.8399,
      "step": 5964
    },
    {
      "epoch": 3.2024093151236537,
      "grad_norm": 0.84375,
      "learning_rate": 1.4332599003225638e-05,
      "loss": 1.0324,
      "step": 5965
    },
    {
      "epoch": 3.2029462098587294,
      "grad_norm": 0.734375,
      "learning_rate": 1.4324961867418668e-05,
      "loss": 0.7748,
      "step": 5966
    },
    {
      "epoch": 3.2034831045938055,
      "grad_norm": 0.7421875,
      "learning_rate": 1.4317325949768734e-05,
      "loss": 1.0084,
      "step": 5967
    },
    {
      "epoch": 3.2040199993288816,
      "grad_norm": 1.046875,
      "learning_rate": 1.4309691251147183e-05,
      "loss": 0.7897,
      "step": 5968
    },
    {
      "epoch": 3.2045568940639577,
      "grad_norm": 0.67578125,
      "learning_rate": 1.4302057772425246e-05,
      "loss": 0.817,
      "step": 5969
    },
    {
      "epoch": 3.2050937887990334,
      "grad_norm": 0.69140625,
      "learning_rate": 1.4294425514474002e-05,
      "loss": 0.9683,
      "step": 5970
    },
    {
      "epoch": 3.2056306835341095,
      "grad_norm": 0.63671875,
      "learning_rate": 1.4286794478164373e-05,
      "loss": 0.7078,
      "step": 5971
    },
    {
      "epoch": 3.2061675782691856,
      "grad_norm": 0.68359375,
      "learning_rate": 1.4279164664367179e-05,
      "loss": 0.8567,
      "step": 5972
    },
    {
      "epoch": 3.2067044730042618,
      "grad_norm": 0.67578125,
      "learning_rate": 1.4271536073953051e-05,
      "loss": 0.8417,
      "step": 5973
    },
    {
      "epoch": 3.2072413677393374,
      "grad_norm": 0.7734375,
      "learning_rate": 1.4263908707792534e-05,
      "loss": 0.9633,
      "step": 5974
    },
    {
      "epoch": 3.2077782624744136,
      "grad_norm": 0.59765625,
      "learning_rate": 1.4256282566755983e-05,
      "loss": 0.7639,
      "step": 5975
    },
    {
      "epoch": 3.2083151572094897,
      "grad_norm": 0.69140625,
      "learning_rate": 1.4248657651713659e-05,
      "loss": 0.8073,
      "step": 5976
    },
    {
      "epoch": 3.208852051944566,
      "grad_norm": 0.5859375,
      "learning_rate": 1.4241033963535644e-05,
      "loss": 0.567,
      "step": 5977
    },
    {
      "epoch": 3.2093889466796415,
      "grad_norm": 0.62109375,
      "learning_rate": 1.4233411503091904e-05,
      "loss": 0.7537,
      "step": 5978
    },
    {
      "epoch": 3.2099258414147176,
      "grad_norm": 0.8203125,
      "learning_rate": 1.422579027125227e-05,
      "loss": 0.951,
      "step": 5979
    },
    {
      "epoch": 3.2104627361497937,
      "grad_norm": 0.5390625,
      "learning_rate": 1.42181702688864e-05,
      "loss": 0.7827,
      "step": 5980
    },
    {
      "epoch": 3.21099963088487,
      "grad_norm": 0.63671875,
      "learning_rate": 1.4210551496863845e-05,
      "loss": 0.8011,
      "step": 5981
    },
    {
      "epoch": 3.2115365256199455,
      "grad_norm": 0.625,
      "learning_rate": 1.4202933956054015e-05,
      "loss": 0.8767,
      "step": 5982
    },
    {
      "epoch": 3.2120734203550216,
      "grad_norm": 0.62890625,
      "learning_rate": 1.419531764732614e-05,
      "loss": 0.8392,
      "step": 5983
    },
    {
      "epoch": 3.2126103150900978,
      "grad_norm": 0.82421875,
      "learning_rate": 1.4187702571549349e-05,
      "loss": 0.8689,
      "step": 5984
    },
    {
      "epoch": 3.2131472098251734,
      "grad_norm": 0.6796875,
      "learning_rate": 1.4180088729592633e-05,
      "loss": 0.8801,
      "step": 5985
    },
    {
      "epoch": 3.2136841045602496,
      "grad_norm": 0.76171875,
      "learning_rate": 1.4172476122324806e-05,
      "loss": 0.9403,
      "step": 5986
    },
    {
      "epoch": 3.2142209992953257,
      "grad_norm": 0.625,
      "learning_rate": 1.4164864750614576e-05,
      "loss": 0.7572,
      "step": 5987
    },
    {
      "epoch": 3.214757894030402,
      "grad_norm": 0.671875,
      "learning_rate": 1.4157254615330501e-05,
      "loss": 0.8931,
      "step": 5988
    },
    {
      "epoch": 3.2152947887654775,
      "grad_norm": 0.73828125,
      "learning_rate": 1.4149645717340975e-05,
      "loss": 0.9601,
      "step": 5989
    },
    {
      "epoch": 3.2158316835005536,
      "grad_norm": 0.796875,
      "learning_rate": 1.4142038057514284e-05,
      "loss": 1.0164,
      "step": 5990
    },
    {
      "epoch": 3.2163685782356297,
      "grad_norm": 0.8828125,
      "learning_rate": 1.4134431636718564e-05,
      "loss": 0.9013,
      "step": 5991
    },
    {
      "epoch": 3.216905472970706,
      "grad_norm": 0.67578125,
      "learning_rate": 1.4126826455821796e-05,
      "loss": 0.9508,
      "step": 5992
    },
    {
      "epoch": 3.2174423677057815,
      "grad_norm": 0.6875,
      "learning_rate": 1.4119222515691816e-05,
      "loss": 0.8455,
      "step": 5993
    },
    {
      "epoch": 3.2179792624408576,
      "grad_norm": 0.66015625,
      "learning_rate": 1.4111619817196341e-05,
      "loss": 0.7485,
      "step": 5994
    },
    {
      "epoch": 3.2185161571759338,
      "grad_norm": 0.68359375,
      "learning_rate": 1.4104018361202948e-05,
      "loss": 0.8297,
      "step": 5995
    },
    {
      "epoch": 3.21905305191101,
      "grad_norm": 0.7421875,
      "learning_rate": 1.4096418148579032e-05,
      "loss": 0.7641,
      "step": 5996
    },
    {
      "epoch": 3.2195899466460856,
      "grad_norm": 0.625,
      "learning_rate": 1.4088819180191889e-05,
      "loss": 0.7851,
      "step": 5997
    },
    {
      "epoch": 3.2201268413811617,
      "grad_norm": 0.7734375,
      "learning_rate": 1.4081221456908666e-05,
      "loss": 0.6875,
      "step": 5998
    },
    {
      "epoch": 3.220663736116238,
      "grad_norm": 0.74609375,
      "learning_rate": 1.4073624979596354e-05,
      "loss": 0.9539,
      "step": 5999
    },
    {
      "epoch": 3.221200630851314,
      "grad_norm": 0.7578125,
      "learning_rate": 1.406602974912179e-05,
      "loss": 0.8123,
      "step": 6000
    },
    {
      "epoch": 3.221200630851314,
      "eval_loss": 0.9029889106750488,
      "eval_runtime": 46.1284,
      "eval_samples_per_second": 20.855,
      "eval_steps_per_second": 20.855,
      "step": 6000
    },
    {
      "epoch": 3.2217375255863896,
      "grad_norm": 0.609375,
      "learning_rate": 1.405843576635171e-05,
      "loss": 0.7553,
      "step": 6001
    },
    {
      "epoch": 3.2222744203214657,
      "grad_norm": 0.6796875,
      "learning_rate": 1.4050843032152661e-05,
      "loss": 0.8135,
      "step": 6002
    },
    {
      "epoch": 3.222811315056542,
      "grad_norm": 0.62109375,
      "learning_rate": 1.4043251547391084e-05,
      "loss": 0.8064,
      "step": 6003
    },
    {
      "epoch": 3.2233482097916175,
      "grad_norm": 0.671875,
      "learning_rate": 1.4035661312933273e-05,
      "loss": 0.8289,
      "step": 6004
    },
    {
      "epoch": 3.2238851045266936,
      "grad_norm": 0.875,
      "learning_rate": 1.4028072329645345e-05,
      "loss": 0.8678,
      "step": 6005
    },
    {
      "epoch": 3.2244219992617698,
      "grad_norm": 0.67578125,
      "learning_rate": 1.4020484598393314e-05,
      "loss": 0.8762,
      "step": 6006
    },
    {
      "epoch": 3.224958893996846,
      "grad_norm": 0.73828125,
      "learning_rate": 1.4012898120043045e-05,
      "loss": 0.893,
      "step": 6007
    },
    {
      "epoch": 3.2254957887319216,
      "grad_norm": 0.69140625,
      "learning_rate": 1.4005312895460237e-05,
      "loss": 0.8491,
      "step": 6008
    },
    {
      "epoch": 3.2260326834669977,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3997728925510458e-05,
      "loss": 0.8506,
      "step": 6009
    },
    {
      "epoch": 3.226569578202074,
      "grad_norm": 0.61328125,
      "learning_rate": 1.399014621105914e-05,
      "loss": 0.9122,
      "step": 6010
    },
    {
      "epoch": 3.22710647293715,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3982564752971577e-05,
      "loss": 0.8655,
      "step": 6011
    },
    {
      "epoch": 3.2276433676722256,
      "grad_norm": 0.625,
      "learning_rate": 1.3974984552112894e-05,
      "loss": 0.6847,
      "step": 6012
    },
    {
      "epoch": 3.2281802624073017,
      "grad_norm": 0.84765625,
      "learning_rate": 1.3967405609348099e-05,
      "loss": 0.9188,
      "step": 6013
    },
    {
      "epoch": 3.228717157142378,
      "grad_norm": 0.66015625,
      "learning_rate": 1.3959827925542034e-05,
      "loss": 0.8532,
      "step": 6014
    },
    {
      "epoch": 3.229254051877454,
      "grad_norm": 0.63671875,
      "learning_rate": 1.3952251501559416e-05,
      "loss": 0.6954,
      "step": 6015
    },
    {
      "epoch": 3.2297909466125296,
      "grad_norm": 0.76171875,
      "learning_rate": 1.3944676338264816e-05,
      "loss": 0.83,
      "step": 6016
    },
    {
      "epoch": 3.2303278413476058,
      "grad_norm": 0.71484375,
      "learning_rate": 1.3937102436522654e-05,
      "loss": 0.8042,
      "step": 6017
    },
    {
      "epoch": 3.230864736082682,
      "grad_norm": 0.66015625,
      "learning_rate": 1.3929529797197193e-05,
      "loss": 0.8731,
      "step": 6018
    },
    {
      "epoch": 3.231401630817758,
      "grad_norm": 0.63671875,
      "learning_rate": 1.3921958421152576e-05,
      "loss": 0.8801,
      "step": 6019
    },
    {
      "epoch": 3.2319385255528337,
      "grad_norm": 1.2265625,
      "learning_rate": 1.3914388309252807e-05,
      "loss": 0.8929,
      "step": 6020
    },
    {
      "epoch": 3.23247542028791,
      "grad_norm": 0.62890625,
      "learning_rate": 1.3906819462361709e-05,
      "loss": 0.9557,
      "step": 6021
    },
    {
      "epoch": 3.233012315022986,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3899251881342994e-05,
      "loss": 1.0061,
      "step": 6022
    },
    {
      "epoch": 3.2335492097580616,
      "grad_norm": 0.6875,
      "learning_rate": 1.389168556706023e-05,
      "loss": 0.804,
      "step": 6023
    },
    {
      "epoch": 3.2340861044931377,
      "grad_norm": 0.67578125,
      "learning_rate": 1.388412052037682e-05,
      "loss": 0.7606,
      "step": 6024
    },
    {
      "epoch": 3.234622999228214,
      "grad_norm": 0.72265625,
      "learning_rate": 1.387655674215602e-05,
      "loss": 0.9004,
      "step": 6025
    },
    {
      "epoch": 3.23515989396329,
      "grad_norm": 0.703125,
      "learning_rate": 1.3868994233260969e-05,
      "loss": 0.7433,
      "step": 6026
    },
    {
      "epoch": 3.2356967886983656,
      "grad_norm": 0.7265625,
      "learning_rate": 1.386143299455463e-05,
      "loss": 0.7762,
      "step": 6027
    },
    {
      "epoch": 3.2362336834334418,
      "grad_norm": 0.69921875,
      "learning_rate": 1.3853873026899846e-05,
      "loss": 0.7771,
      "step": 6028
    },
    {
      "epoch": 3.236770578168518,
      "grad_norm": 0.62109375,
      "learning_rate": 1.3846314331159313e-05,
      "loss": 0.8065,
      "step": 6029
    },
    {
      "epoch": 3.237307472903594,
      "grad_norm": 0.65625,
      "learning_rate": 1.3838756908195555e-05,
      "loss": 0.6639,
      "step": 6030
    },
    {
      "epoch": 3.2378443676386697,
      "grad_norm": 0.79296875,
      "learning_rate": 1.383120075887098e-05,
      "loss": 0.991,
      "step": 6031
    },
    {
      "epoch": 3.238381262373746,
      "grad_norm": 0.6484375,
      "learning_rate": 1.3823645884047847e-05,
      "loss": 0.7454,
      "step": 6032
    },
    {
      "epoch": 3.238918157108822,
      "grad_norm": 0.7109375,
      "learning_rate": 1.3816092284588262e-05,
      "loss": 0.8378,
      "step": 6033
    },
    {
      "epoch": 3.239455051843898,
      "grad_norm": 0.66015625,
      "learning_rate": 1.3808539961354167e-05,
      "loss": 0.9516,
      "step": 6034
    },
    {
      "epoch": 3.2399919465789737,
      "grad_norm": 0.62109375,
      "learning_rate": 1.3800988915207386e-05,
      "loss": 0.7173,
      "step": 6035
    },
    {
      "epoch": 3.24052884131405,
      "grad_norm": 0.6875,
      "learning_rate": 1.379343914700961e-05,
      "loss": 0.91,
      "step": 6036
    },
    {
      "epoch": 3.241065736049126,
      "grad_norm": 0.6875,
      "learning_rate": 1.3785890657622335e-05,
      "loss": 0.8349,
      "step": 6037
    },
    {
      "epoch": 3.241602630784202,
      "grad_norm": 0.6171875,
      "learning_rate": 1.3778343447906949e-05,
      "loss": 0.7921,
      "step": 6038
    },
    {
      "epoch": 3.2421395255192778,
      "grad_norm": 0.75,
      "learning_rate": 1.3770797518724699e-05,
      "loss": 0.8189,
      "step": 6039
    },
    {
      "epoch": 3.242676420254354,
      "grad_norm": 0.703125,
      "learning_rate": 1.3763252870936649e-05,
      "loss": 0.9578,
      "step": 6040
    },
    {
      "epoch": 3.24321331498943,
      "grad_norm": 0.67578125,
      "learning_rate": 1.3755709505403756e-05,
      "loss": 0.9725,
      "step": 6041
    },
    {
      "epoch": 3.2437502097245057,
      "grad_norm": 0.72265625,
      "learning_rate": 1.3748167422986807e-05,
      "loss": 1.049,
      "step": 6042
    },
    {
      "epoch": 3.244287104459582,
      "grad_norm": 0.66015625,
      "learning_rate": 1.3740626624546438e-05,
      "loss": 0.9442,
      "step": 6043
    },
    {
      "epoch": 3.244823999194658,
      "grad_norm": 0.8828125,
      "learning_rate": 1.3733087110943161e-05,
      "loss": 0.736,
      "step": 6044
    },
    {
      "epoch": 3.245360893929734,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3725548883037334e-05,
      "loss": 0.7684,
      "step": 6045
    },
    {
      "epoch": 3.2458977886648097,
      "grad_norm": 0.5625,
      "learning_rate": 1.3718011941689152e-05,
      "loss": 0.7714,
      "step": 6046
    },
    {
      "epoch": 3.246434683399886,
      "grad_norm": 0.6484375,
      "learning_rate": 1.3710476287758683e-05,
      "loss": 0.7945,
      "step": 6047
    },
    {
      "epoch": 3.246971578134962,
      "grad_norm": 1.1171875,
      "learning_rate": 1.370294192210585e-05,
      "loss": 0.7305,
      "step": 6048
    },
    {
      "epoch": 3.247508472870038,
      "grad_norm": 0.64453125,
      "learning_rate": 1.3695408845590396e-05,
      "loss": 0.9207,
      "step": 6049
    },
    {
      "epoch": 3.2480453676051138,
      "grad_norm": 0.625,
      "learning_rate": 1.3687877059071966e-05,
      "loss": 0.7079,
      "step": 6050
    },
    {
      "epoch": 3.2480453676051138,
      "eval_loss": 0.9029369950294495,
      "eval_runtime": 46.8553,
      "eval_samples_per_second": 20.531,
      "eval_steps_per_second": 20.531,
      "step": 6050
    },
    {
      "epoch": 3.24858226234019,
      "grad_norm": 0.60546875,
      "learning_rate": 1.368034656341001e-05,
      "loss": 0.8576,
      "step": 6051
    },
    {
      "epoch": 3.249119157075266,
      "grad_norm": 0.6640625,
      "learning_rate": 1.3672817359463875e-05,
      "loss": 0.8994,
      "step": 6052
    },
    {
      "epoch": 3.249656051810342,
      "grad_norm": 0.58203125,
      "learning_rate": 1.366528944809271e-05,
      "loss": 0.825,
      "step": 6053
    },
    {
      "epoch": 3.250192946545418,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3657762830155568e-05,
      "loss": 0.9276,
      "step": 6054
    },
    {
      "epoch": 3.250729841280494,
      "grad_norm": 0.6171875,
      "learning_rate": 1.3650237506511331e-05,
      "loss": 0.8413,
      "step": 6055
    },
    {
      "epoch": 3.25126673601557,
      "grad_norm": 0.7265625,
      "learning_rate": 1.364271347801872e-05,
      "loss": 0.7741,
      "step": 6056
    },
    {
      "epoch": 3.251803630750646,
      "grad_norm": 0.6484375,
      "learning_rate": 1.3635190745536342e-05,
      "loss": 0.9985,
      "step": 6057
    },
    {
      "epoch": 3.252340525485722,
      "grad_norm": 0.74609375,
      "learning_rate": 1.362766930992262e-05,
      "loss": 1.05,
      "step": 6058
    },
    {
      "epoch": 3.252877420220798,
      "grad_norm": 0.8125,
      "learning_rate": 1.3620149172035837e-05,
      "loss": 0.7583,
      "step": 6059
    },
    {
      "epoch": 3.253414314955874,
      "grad_norm": 0.63671875,
      "learning_rate": 1.361263033273415e-05,
      "loss": 0.9433,
      "step": 6060
    },
    {
      "epoch": 3.2539512096909498,
      "grad_norm": 0.77734375,
      "learning_rate": 1.3605112792875562e-05,
      "loss": 0.9179,
      "step": 6061
    },
    {
      "epoch": 3.254488104426026,
      "grad_norm": 0.6484375,
      "learning_rate": 1.3597596553317899e-05,
      "loss": 0.7118,
      "step": 6062
    },
    {
      "epoch": 3.255024999161102,
      "grad_norm": 0.7421875,
      "learning_rate": 1.3590081614918864e-05,
      "loss": 0.8761,
      "step": 6063
    },
    {
      "epoch": 3.255561893896178,
      "grad_norm": 0.66796875,
      "learning_rate": 1.3582567978536026e-05,
      "loss": 0.8091,
      "step": 6064
    },
    {
      "epoch": 3.256098788631254,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3575055645026763e-05,
      "loss": 0.7951,
      "step": 6065
    },
    {
      "epoch": 3.25663568336633,
      "grad_norm": 0.65234375,
      "learning_rate": 1.3567544615248345e-05,
      "loss": 0.762,
      "step": 6066
    },
    {
      "epoch": 3.257172578101406,
      "grad_norm": 0.609375,
      "learning_rate": 1.3560034890057866e-05,
      "loss": 0.9841,
      "step": 6067
    },
    {
      "epoch": 3.257709472836482,
      "grad_norm": 0.6484375,
      "learning_rate": 1.3552526470312271e-05,
      "loss": 0.8665,
      "step": 6068
    },
    {
      "epoch": 3.258246367571558,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3545019356868378e-05,
      "loss": 0.793,
      "step": 6069
    },
    {
      "epoch": 3.258783262306634,
      "grad_norm": 0.6640625,
      "learning_rate": 1.3537513550582853e-05,
      "loss": 0.6832,
      "step": 6070
    },
    {
      "epoch": 3.25932015704171,
      "grad_norm": 0.62109375,
      "learning_rate": 1.353000905231218e-05,
      "loss": 0.8276,
      "step": 6071
    },
    {
      "epoch": 3.259857051776786,
      "grad_norm": 0.7421875,
      "learning_rate": 1.352250586291273e-05,
      "loss": 1.0,
      "step": 6072
    },
    {
      "epoch": 3.260393946511862,
      "grad_norm": 0.6640625,
      "learning_rate": 1.351500398324072e-05,
      "loss": 0.7929,
      "step": 6073
    },
    {
      "epoch": 3.260930841246938,
      "grad_norm": 0.7890625,
      "learning_rate": 1.3507503414152195e-05,
      "loss": 0.8872,
      "step": 6074
    },
    {
      "epoch": 3.261467735982014,
      "grad_norm": 0.703125,
      "learning_rate": 1.350000415650308e-05,
      "loss": 0.8404,
      "step": 6075
    },
    {
      "epoch": 3.2620046307170902,
      "grad_norm": 0.69921875,
      "learning_rate": 1.3492506211149115e-05,
      "loss": 0.6903,
      "step": 6076
    },
    {
      "epoch": 3.262541525452166,
      "grad_norm": 0.78515625,
      "learning_rate": 1.3485009578945934e-05,
      "loss": 1.026,
      "step": 6077
    },
    {
      "epoch": 3.263078420187242,
      "grad_norm": 0.77734375,
      "learning_rate": 1.3477514260748975e-05,
      "loss": 0.7941,
      "step": 6078
    },
    {
      "epoch": 3.263615314922318,
      "grad_norm": 0.78125,
      "learning_rate": 1.3470020257413558e-05,
      "loss": 1.0685,
      "step": 6079
    },
    {
      "epoch": 3.264152209657394,
      "grad_norm": 0.59765625,
      "learning_rate": 1.3462527569794853e-05,
      "loss": 0.891,
      "step": 6080
    },
    {
      "epoch": 3.26468910439247,
      "grad_norm": 0.66796875,
      "learning_rate": 1.3455036198747856e-05,
      "loss": 0.8868,
      "step": 6081
    },
    {
      "epoch": 3.265225999127546,
      "grad_norm": 0.62109375,
      "learning_rate": 1.3447546145127434e-05,
      "loss": 0.8021,
      "step": 6082
    },
    {
      "epoch": 3.265762893862622,
      "grad_norm": 0.8125,
      "learning_rate": 1.3440057409788315e-05,
      "loss": 1.0257,
      "step": 6083
    },
    {
      "epoch": 3.266299788597698,
      "grad_norm": 0.62890625,
      "learning_rate": 1.3432569993585023e-05,
      "loss": 0.6777,
      "step": 6084
    },
    {
      "epoch": 3.266836683332774,
      "grad_norm": 0.7265625,
      "learning_rate": 1.3425083897371981e-05,
      "loss": 0.7381,
      "step": 6085
    },
    {
      "epoch": 3.26737357806785,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3417599122003464e-05,
      "loss": 0.9388,
      "step": 6086
    },
    {
      "epoch": 3.2679104728029262,
      "grad_norm": 0.70703125,
      "learning_rate": 1.3410115668333556e-05,
      "loss": 0.767,
      "step": 6087
    },
    {
      "epoch": 3.268447367538002,
      "grad_norm": 0.6796875,
      "learning_rate": 1.3402633537216221e-05,
      "loss": 0.8137,
      "step": 6088
    },
    {
      "epoch": 3.268984262273078,
      "grad_norm": 0.71875,
      "learning_rate": 1.3395152729505281e-05,
      "loss": 0.8238,
      "step": 6089
    },
    {
      "epoch": 3.269521157008154,
      "grad_norm": 0.76953125,
      "learning_rate": 1.3387673246054367e-05,
      "loss": 0.8889,
      "step": 6090
    },
    {
      "epoch": 3.2700580517432303,
      "grad_norm": 0.69140625,
      "learning_rate": 1.3380195087716996e-05,
      "loss": 0.8642,
      "step": 6091
    },
    {
      "epoch": 3.270594946478306,
      "grad_norm": 0.7734375,
      "learning_rate": 1.3372718255346526e-05,
      "loss": 0.9319,
      "step": 6092
    },
    {
      "epoch": 3.271131841213382,
      "grad_norm": 0.66015625,
      "learning_rate": 1.336524274979615e-05,
      "loss": 0.9325,
      "step": 6093
    },
    {
      "epoch": 3.271668735948458,
      "grad_norm": 0.64453125,
      "learning_rate": 1.3357768571918911e-05,
      "loss": 0.6935,
      "step": 6094
    },
    {
      "epoch": 3.2722056306835343,
      "grad_norm": 0.94921875,
      "learning_rate": 1.3350295722567713e-05,
      "loss": 0.774,
      "step": 6095
    },
    {
      "epoch": 3.27274252541861,
      "grad_norm": 0.63671875,
      "learning_rate": 1.3342824202595317e-05,
      "loss": 0.6772,
      "step": 6096
    },
    {
      "epoch": 3.273279420153686,
      "grad_norm": 0.6875,
      "learning_rate": 1.3335354012854293e-05,
      "loss": 0.7624,
      "step": 6097
    },
    {
      "epoch": 3.2738163148887622,
      "grad_norm": 0.6640625,
      "learning_rate": 1.3327885154197105e-05,
      "loss": 0.9108,
      "step": 6098
    },
    {
      "epoch": 3.274353209623838,
      "grad_norm": 0.765625,
      "learning_rate": 1.3320417627476034e-05,
      "loss": 0.9702,
      "step": 6099
    },
    {
      "epoch": 3.274890104358914,
      "grad_norm": 0.6640625,
      "learning_rate": 1.3312951433543225e-05,
      "loss": 0.9428,
      "step": 6100
    },
    {
      "epoch": 3.274890104358914,
      "eval_loss": 0.9030158519744873,
      "eval_runtime": 46.3139,
      "eval_samples_per_second": 20.771,
      "eval_steps_per_second": 20.771,
      "step": 6100
    },
    {
      "epoch": 3.27542699909399,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3305486573250653e-05,
      "loss": 0.8324,
      "step": 6101
    },
    {
      "epoch": 3.2759638938290663,
      "grad_norm": 0.69921875,
      "learning_rate": 1.3298023047450176e-05,
      "loss": 0.861,
      "step": 6102
    },
    {
      "epoch": 3.276500788564142,
      "grad_norm": 0.72265625,
      "learning_rate": 1.3290560856993448e-05,
      "loss": 0.9047,
      "step": 6103
    },
    {
      "epoch": 3.277037683299218,
      "grad_norm": 0.78125,
      "learning_rate": 1.3283100002732015e-05,
      "loss": 0.7606,
      "step": 6104
    },
    {
      "epoch": 3.277574578034294,
      "grad_norm": 0.625,
      "learning_rate": 1.3275640485517266e-05,
      "loss": 0.8246,
      "step": 6105
    },
    {
      "epoch": 3.2781114727693703,
      "grad_norm": 0.69921875,
      "learning_rate": 1.3268182306200405e-05,
      "loss": 0.8,
      "step": 6106
    },
    {
      "epoch": 3.278648367504446,
      "grad_norm": 0.703125,
      "learning_rate": 1.3260725465632512e-05,
      "loss": 0.8437,
      "step": 6107
    },
    {
      "epoch": 3.279185262239522,
      "grad_norm": 0.7109375,
      "learning_rate": 1.325326996466453e-05,
      "loss": 0.6511,
      "step": 6108
    },
    {
      "epoch": 3.2797221569745982,
      "grad_norm": 0.59765625,
      "learning_rate": 1.3245815804147185e-05,
      "loss": 0.8475,
      "step": 6109
    },
    {
      "epoch": 3.2802590517096744,
      "grad_norm": 0.5859375,
      "learning_rate": 1.3238362984931113e-05,
      "loss": 0.686,
      "step": 6110
    },
    {
      "epoch": 3.28079594644475,
      "grad_norm": 0.625,
      "learning_rate": 1.3230911507866783e-05,
      "loss": 0.9385,
      "step": 6111
    },
    {
      "epoch": 3.281332841179826,
      "grad_norm": 0.6328125,
      "learning_rate": 1.3223461373804482e-05,
      "loss": 0.7253,
      "step": 6112
    },
    {
      "epoch": 3.2818697359149023,
      "grad_norm": 0.66796875,
      "learning_rate": 1.3216012583594379e-05,
      "loss": 0.6429,
      "step": 6113
    },
    {
      "epoch": 3.2824066306499784,
      "grad_norm": 0.671875,
      "learning_rate": 1.3208565138086479e-05,
      "loss": 0.7178,
      "step": 6114
    },
    {
      "epoch": 3.282943525385054,
      "grad_norm": 0.63671875,
      "learning_rate": 1.3201119038130616e-05,
      "loss": 0.8726,
      "step": 6115
    },
    {
      "epoch": 3.28348042012013,
      "grad_norm": 0.609375,
      "learning_rate": 1.319367428457649e-05,
      "loss": 0.6606,
      "step": 6116
    },
    {
      "epoch": 3.2840173148552063,
      "grad_norm": 0.6796875,
      "learning_rate": 1.3186230878273653e-05,
      "loss": 0.8512,
      "step": 6117
    },
    {
      "epoch": 3.284554209590282,
      "grad_norm": 0.76953125,
      "learning_rate": 1.317878882007148e-05,
      "loss": 0.9175,
      "step": 6118
    },
    {
      "epoch": 3.285091104325358,
      "grad_norm": 0.59375,
      "learning_rate": 1.3171348110819198e-05,
      "loss": 0.8343,
      "step": 6119
    },
    {
      "epoch": 3.2856279990604342,
      "grad_norm": 0.75,
      "learning_rate": 1.3163908751365895e-05,
      "loss": 0.9887,
      "step": 6120
    },
    {
      "epoch": 3.2861648937955104,
      "grad_norm": 0.75390625,
      "learning_rate": 1.3156470742560506e-05,
      "loss": 0.877,
      "step": 6121
    },
    {
      "epoch": 3.286701788530586,
      "grad_norm": 0.59765625,
      "learning_rate": 1.3149034085251783e-05,
      "loss": 0.9246,
      "step": 6122
    },
    {
      "epoch": 3.287238683265662,
      "grad_norm": 0.65234375,
      "learning_rate": 1.3141598780288348e-05,
      "loss": 0.8138,
      "step": 6123
    },
    {
      "epoch": 3.2877755780007383,
      "grad_norm": 0.69140625,
      "learning_rate": 1.3134164828518674e-05,
      "loss": 0.8826,
      "step": 6124
    },
    {
      "epoch": 3.2883124727358144,
      "grad_norm": 0.62890625,
      "learning_rate": 1.3126732230791067e-05,
      "loss": 0.78,
      "step": 6125
    },
    {
      "epoch": 3.28884936747089,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3119300987953664e-05,
      "loss": 0.9377,
      "step": 6126
    },
    {
      "epoch": 3.289386262205966,
      "grad_norm": 0.703125,
      "learning_rate": 1.3111871100854483e-05,
      "loss": 0.8684,
      "step": 6127
    },
    {
      "epoch": 3.2899231569410423,
      "grad_norm": 0.76171875,
      "learning_rate": 1.3104442570341352e-05,
      "loss": 0.8714,
      "step": 6128
    },
    {
      "epoch": 3.2904600516761184,
      "grad_norm": 0.6875,
      "learning_rate": 1.3097015397261969e-05,
      "loss": 0.7836,
      "step": 6129
    },
    {
      "epoch": 3.290996946411194,
      "grad_norm": 0.66796875,
      "learning_rate": 1.3089589582463879e-05,
      "loss": 0.8938,
      "step": 6130
    },
    {
      "epoch": 3.2915338411462702,
      "grad_norm": 0.69140625,
      "learning_rate": 1.3082165126794441e-05,
      "loss": 0.8421,
      "step": 6131
    },
    {
      "epoch": 3.2920707358813464,
      "grad_norm": 0.640625,
      "learning_rate": 1.307474203110089e-05,
      "loss": 0.7528,
      "step": 6132
    },
    {
      "epoch": 3.2926076306164225,
      "grad_norm": 0.6640625,
      "learning_rate": 1.30673202962303e-05,
      "loss": 0.7538,
      "step": 6133
    },
    {
      "epoch": 3.293144525351498,
      "grad_norm": 0.60546875,
      "learning_rate": 1.3059899923029584e-05,
      "loss": 0.6987,
      "step": 6134
    },
    {
      "epoch": 3.2936814200865743,
      "grad_norm": 0.57421875,
      "learning_rate": 1.3052480912345482e-05,
      "loss": 0.7734,
      "step": 6135
    },
    {
      "epoch": 3.2942183148216504,
      "grad_norm": 0.69921875,
      "learning_rate": 1.3045063265024612e-05,
      "loss": 1.0949,
      "step": 6136
    },
    {
      "epoch": 3.294755209556726,
      "grad_norm": 0.65625,
      "learning_rate": 1.3037646981913431e-05,
      "loss": 0.9803,
      "step": 6137
    },
    {
      "epoch": 3.295292104291802,
      "grad_norm": 0.7265625,
      "learning_rate": 1.303023206385821e-05,
      "loss": 0.9215,
      "step": 6138
    },
    {
      "epoch": 3.2958289990268783,
      "grad_norm": 0.7734375,
      "learning_rate": 1.3022818511705107e-05,
      "loss": 0.718,
      "step": 6139
    },
    {
      "epoch": 3.2963658937619544,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3015406326300079e-05,
      "loss": 0.8945,
      "step": 6140
    },
    {
      "epoch": 3.29690278849703,
      "grad_norm": 0.890625,
      "learning_rate": 1.3007995508488963e-05,
      "loss": 0.7963,
      "step": 6141
    },
    {
      "epoch": 3.2974396832321062,
      "grad_norm": 0.6640625,
      "learning_rate": 1.3000586059117437e-05,
      "loss": 0.8496,
      "step": 6142
    },
    {
      "epoch": 3.2979765779671824,
      "grad_norm": 0.609375,
      "learning_rate": 1.2993177979030996e-05,
      "loss": 1.0048,
      "step": 6143
    },
    {
      "epoch": 3.2985134727022585,
      "grad_norm": 0.69140625,
      "learning_rate": 1.2985771269074997e-05,
      "loss": 1.0115,
      "step": 6144
    },
    {
      "epoch": 3.299050367437334,
      "grad_norm": 0.8125,
      "learning_rate": 1.2978365930094643e-05,
      "loss": 0.7098,
      "step": 6145
    },
    {
      "epoch": 3.2995872621724103,
      "grad_norm": 0.59375,
      "learning_rate": 1.2970961962934988e-05,
      "loss": 0.625,
      "step": 6146
    },
    {
      "epoch": 3.3001241569074864,
      "grad_norm": 0.85546875,
      "learning_rate": 1.2963559368440897e-05,
      "loss": 0.9434,
      "step": 6147
    },
    {
      "epoch": 3.3006610516425625,
      "grad_norm": 0.6640625,
      "learning_rate": 1.2956158147457115e-05,
      "loss": 0.7545,
      "step": 6148
    },
    {
      "epoch": 3.301197946377638,
      "grad_norm": 0.5703125,
      "learning_rate": 1.2948758300828217e-05,
      "loss": 0.7369,
      "step": 6149
    },
    {
      "epoch": 3.3017348411127143,
      "grad_norm": 0.71875,
      "learning_rate": 1.2941359829398616e-05,
      "loss": 0.7774,
      "step": 6150
    },
    {
      "epoch": 3.3017348411127143,
      "eval_loss": 0.902991533279419,
      "eval_runtime": 46.0383,
      "eval_samples_per_second": 20.896,
      "eval_steps_per_second": 20.896,
      "step": 6150
    },
    {
      "epoch": 3.3022717358477904,
      "grad_norm": 0.7734375,
      "learning_rate": 1.2933962734012558e-05,
      "loss": 0.8595,
      "step": 6151
    },
    {
      "epoch": 3.3028086305828666,
      "grad_norm": 0.67578125,
      "learning_rate": 1.2926567015514168e-05,
      "loss": 0.7515,
      "step": 6152
    },
    {
      "epoch": 3.3033455253179422,
      "grad_norm": 0.80859375,
      "learning_rate": 1.2919172674747366e-05,
      "loss": 0.8776,
      "step": 6153
    },
    {
      "epoch": 3.3038824200530184,
      "grad_norm": 0.640625,
      "learning_rate": 1.2911779712555955e-05,
      "loss": 0.688,
      "step": 6154
    },
    {
      "epoch": 3.3044193147880945,
      "grad_norm": 0.76171875,
      "learning_rate": 1.2904388129783574e-05,
      "loss": 0.9123,
      "step": 6155
    },
    {
      "epoch": 3.30495620952317,
      "grad_norm": 0.8359375,
      "learning_rate": 1.2896997927273675e-05,
      "loss": 0.7135,
      "step": 6156
    },
    {
      "epoch": 3.3054931042582463,
      "grad_norm": 0.76953125,
      "learning_rate": 1.2889609105869585e-05,
      "loss": 0.7934,
      "step": 6157
    },
    {
      "epoch": 3.3060299989933224,
      "grad_norm": 0.6953125,
      "learning_rate": 1.288222166641447e-05,
      "loss": 0.8593,
      "step": 6158
    },
    {
      "epoch": 3.3065668937283985,
      "grad_norm": 0.83984375,
      "learning_rate": 1.2874835609751324e-05,
      "loss": 1.0064,
      "step": 6159
    },
    {
      "epoch": 3.307103788463474,
      "grad_norm": 0.69921875,
      "learning_rate": 1.286745093672298e-05,
      "loss": 0.9235,
      "step": 6160
    },
    {
      "epoch": 3.3076406831985503,
      "grad_norm": 0.84375,
      "learning_rate": 1.2860067648172124e-05,
      "loss": 0.8554,
      "step": 6161
    },
    {
      "epoch": 3.3081775779336264,
      "grad_norm": 0.765625,
      "learning_rate": 1.2852685744941307e-05,
      "loss": 0.7287,
      "step": 6162
    },
    {
      "epoch": 3.3087144726687026,
      "grad_norm": 0.65234375,
      "learning_rate": 1.2845305227872865e-05,
      "loss": 0.8195,
      "step": 6163
    },
    {
      "epoch": 3.3092513674037782,
      "grad_norm": 0.671875,
      "learning_rate": 1.2837926097809023e-05,
      "loss": 0.9491,
      "step": 6164
    },
    {
      "epoch": 3.3097882621388544,
      "grad_norm": 0.73828125,
      "learning_rate": 1.2830548355591843e-05,
      "loss": 0.8297,
      "step": 6165
    },
    {
      "epoch": 3.3103251568739305,
      "grad_norm": 0.78515625,
      "learning_rate": 1.2823172002063202e-05,
      "loss": 0.97,
      "step": 6166
    },
    {
      "epoch": 3.3108620516090066,
      "grad_norm": 0.8359375,
      "learning_rate": 1.2815797038064853e-05,
      "loss": 0.8428,
      "step": 6167
    },
    {
      "epoch": 3.3113989463440823,
      "grad_norm": 0.6328125,
      "learning_rate": 1.2808423464438357e-05,
      "loss": 0.796,
      "step": 6168
    },
    {
      "epoch": 3.3119358410791584,
      "grad_norm": 0.69921875,
      "learning_rate": 1.2801051282025133e-05,
      "loss": 0.7423,
      "step": 6169
    },
    {
      "epoch": 3.3124727358142345,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2793680491666443e-05,
      "loss": 1.0473,
      "step": 6170
    },
    {
      "epoch": 3.3130096305493106,
      "grad_norm": 0.78125,
      "learning_rate": 1.2786311094203401e-05,
      "loss": 0.813,
      "step": 6171
    },
    {
      "epoch": 3.3135465252843863,
      "grad_norm": 0.66015625,
      "learning_rate": 1.2778943090476925e-05,
      "loss": 0.8596,
      "step": 6172
    },
    {
      "epoch": 3.3140834200194624,
      "grad_norm": 0.66015625,
      "learning_rate": 1.277157648132781e-05,
      "loss": 0.7721,
      "step": 6173
    },
    {
      "epoch": 3.3146203147545386,
      "grad_norm": 0.6171875,
      "learning_rate": 1.2764211267596687e-05,
      "loss": 0.8599,
      "step": 6174
    },
    {
      "epoch": 3.3151572094896142,
      "grad_norm": 0.65234375,
      "learning_rate": 1.2756847450124004e-05,
      "loss": 0.6867,
      "step": 6175
    },
    {
      "epoch": 3.3156941042246904,
      "grad_norm": 0.61328125,
      "learning_rate": 1.2749485029750085e-05,
      "loss": 0.9036,
      "step": 6176
    },
    {
      "epoch": 3.3162309989597665,
      "grad_norm": 0.7578125,
      "learning_rate": 1.2742124007315055e-05,
      "loss": 1.0004,
      "step": 6177
    },
    {
      "epoch": 3.3167678936948426,
      "grad_norm": 0.68359375,
      "learning_rate": 1.2734764383658917e-05,
      "loss": 0.7567,
      "step": 6178
    },
    {
      "epoch": 3.3173047884299183,
      "grad_norm": 0.67578125,
      "learning_rate": 1.272740615962148e-05,
      "loss": 0.8134,
      "step": 6179
    },
    {
      "epoch": 3.3178416831649944,
      "grad_norm": 0.71484375,
      "learning_rate": 1.2720049336042422e-05,
      "loss": 0.8148,
      "step": 6180
    },
    {
      "epoch": 3.3183785779000705,
      "grad_norm": 0.703125,
      "learning_rate": 1.2712693913761253e-05,
      "loss": 0.9834,
      "step": 6181
    },
    {
      "epoch": 3.3189154726351466,
      "grad_norm": 0.63671875,
      "learning_rate": 1.2705339893617308e-05,
      "loss": 0.8446,
      "step": 6182
    },
    {
      "epoch": 3.3194523673702223,
      "grad_norm": 0.8046875,
      "learning_rate": 1.269798727644979e-05,
      "loss": 0.9565,
      "step": 6183
    },
    {
      "epoch": 3.3199892621052984,
      "grad_norm": 0.7109375,
      "learning_rate": 1.2690636063097716e-05,
      "loss": 0.9704,
      "step": 6184
    },
    {
      "epoch": 3.3205261568403746,
      "grad_norm": 0.72265625,
      "learning_rate": 1.2683286254399943e-05,
      "loss": 0.7425,
      "step": 6185
    },
    {
      "epoch": 3.3210630515754507,
      "grad_norm": 0.71484375,
      "learning_rate": 1.2675937851195185e-05,
      "loss": 1.0241,
      "step": 6186
    },
    {
      "epoch": 3.3215999463105264,
      "grad_norm": 0.77734375,
      "learning_rate": 1.2668590854322005e-05,
      "loss": 0.8633,
      "step": 6187
    },
    {
      "epoch": 3.3221368410456025,
      "grad_norm": 0.61328125,
      "learning_rate": 1.2661245264618759e-05,
      "loss": 0.7208,
      "step": 6188
    },
    {
      "epoch": 3.3226737357806786,
      "grad_norm": 0.5390625,
      "learning_rate": 1.2653901082923686e-05,
      "loss": 0.6294,
      "step": 6189
    },
    {
      "epoch": 3.3232106305157547,
      "grad_norm": 0.80859375,
      "learning_rate": 1.264655831007486e-05,
      "loss": 0.9833,
      "step": 6190
    },
    {
      "epoch": 3.3237475252508304,
      "grad_norm": 0.703125,
      "learning_rate": 1.2639216946910164e-05,
      "loss": 0.837,
      "step": 6191
    },
    {
      "epoch": 3.3242844199859065,
      "grad_norm": 0.73828125,
      "learning_rate": 1.2631876994267363e-05,
      "loss": 0.8409,
      "step": 6192
    },
    {
      "epoch": 3.3248213147209826,
      "grad_norm": 0.7109375,
      "learning_rate": 1.2624538452984025e-05,
      "loss": 0.7585,
      "step": 6193
    },
    {
      "epoch": 3.3253582094560583,
      "grad_norm": 0.703125,
      "learning_rate": 1.2617201323897565e-05,
      "loss": 0.8939,
      "step": 6194
    },
    {
      "epoch": 3.3258951041911344,
      "grad_norm": 0.57421875,
      "learning_rate": 1.2609865607845244e-05,
      "loss": 0.8083,
      "step": 6195
    },
    {
      "epoch": 3.3264319989262106,
      "grad_norm": 0.58203125,
      "learning_rate": 1.260253130566418e-05,
      "loss": 0.7417,
      "step": 6196
    },
    {
      "epoch": 3.3269688936612867,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2595198418191285e-05,
      "loss": 0.8626,
      "step": 6197
    },
    {
      "epoch": 3.3275057883963624,
      "grad_norm": 0.6484375,
      "learning_rate": 1.2587866946263344e-05,
      "loss": 0.7583,
      "step": 6198
    },
    {
      "epoch": 3.3280426831314385,
      "grad_norm": 0.66796875,
      "learning_rate": 1.2580536890716982e-05,
      "loss": 0.7866,
      "step": 6199
    },
    {
      "epoch": 3.3285795778665146,
      "grad_norm": 0.65234375,
      "learning_rate": 1.2573208252388629e-05,
      "loss": 0.8418,
      "step": 6200
    },
    {
      "epoch": 3.3285795778665146,
      "eval_loss": 0.9033068418502808,
      "eval_runtime": 46.3129,
      "eval_samples_per_second": 20.772,
      "eval_steps_per_second": 20.772,
      "step": 6200
    },
    {
      "epoch": 3.3291164726015907,
      "grad_norm": 0.640625,
      "learning_rate": 1.2565881032114596e-05,
      "loss": 0.8344,
      "step": 6201
    },
    {
      "epoch": 3.3296533673366664,
      "grad_norm": 0.59375,
      "learning_rate": 1.2558555230730994e-05,
      "loss": 0.7096,
      "step": 6202
    },
    {
      "epoch": 3.3301902620717425,
      "grad_norm": 0.75390625,
      "learning_rate": 1.2551230849073809e-05,
      "loss": 0.9951,
      "step": 6203
    },
    {
      "epoch": 3.3307271568068186,
      "grad_norm": 0.6328125,
      "learning_rate": 1.2543907887978823e-05,
      "loss": 0.7308,
      "step": 6204
    },
    {
      "epoch": 3.3312640515418948,
      "grad_norm": 0.65234375,
      "learning_rate": 1.2536586348281692e-05,
      "loss": 1.0247,
      "step": 6205
    },
    {
      "epoch": 3.3318009462769704,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2529266230817907e-05,
      "loss": 0.9658,
      "step": 6206
    },
    {
      "epoch": 3.3323378410120466,
      "grad_norm": 0.80078125,
      "learning_rate": 1.2521947536422762e-05,
      "loss": 0.8694,
      "step": 6207
    },
    {
      "epoch": 3.3328747357471227,
      "grad_norm": 0.75,
      "learning_rate": 1.2514630265931426e-05,
      "loss": 0.6699,
      "step": 6208
    },
    {
      "epoch": 3.333411630482199,
      "grad_norm": 0.7265625,
      "learning_rate": 1.250731442017891e-05,
      "loss": 0.7852,
      "step": 6209
    },
    {
      "epoch": 3.3339485252172745,
      "grad_norm": 0.62890625,
      "learning_rate": 1.2500000000000006e-05,
      "loss": 0.9484,
      "step": 6210
    },
    {
      "epoch": 3.3344854199523506,
      "grad_norm": 0.7265625,
      "learning_rate": 1.24926870062294e-05,
      "loss": 0.9421,
      "step": 6211
    },
    {
      "epoch": 3.3350223146874267,
      "grad_norm": 0.71484375,
      "learning_rate": 1.248537543970161e-05,
      "loss": 0.9167,
      "step": 6212
    },
    {
      "epoch": 3.3355592094225024,
      "grad_norm": 0.71875,
      "learning_rate": 1.247806530125096e-05,
      "loss": 1.0036,
      "step": 6213
    },
    {
      "epoch": 3.3360961041575785,
      "grad_norm": 0.6953125,
      "learning_rate": 1.2470756591711635e-05,
      "loss": 0.8089,
      "step": 6214
    },
    {
      "epoch": 3.3366329988926546,
      "grad_norm": 0.76171875,
      "learning_rate": 1.2463449311917661e-05,
      "loss": 1.0078,
      "step": 6215
    },
    {
      "epoch": 3.3371698936277308,
      "grad_norm": 0.68359375,
      "learning_rate": 1.2456143462702874e-05,
      "loss": 0.9643,
      "step": 6216
    },
    {
      "epoch": 3.3377067883628064,
      "grad_norm": 0.66796875,
      "learning_rate": 1.2448839044900985e-05,
      "loss": 0.8913,
      "step": 6217
    },
    {
      "epoch": 3.3382436830978826,
      "grad_norm": 0.6953125,
      "learning_rate": 1.2441536059345496e-05,
      "loss": 0.8453,
      "step": 6218
    },
    {
      "epoch": 3.3387805778329587,
      "grad_norm": 0.6328125,
      "learning_rate": 1.2434234506869794e-05,
      "loss": 0.7513,
      "step": 6219
    },
    {
      "epoch": 3.339317472568035,
      "grad_norm": 0.66015625,
      "learning_rate": 1.2426934388307056e-05,
      "loss": 0.8628,
      "step": 6220
    },
    {
      "epoch": 3.3398543673031105,
      "grad_norm": 0.70703125,
      "learning_rate": 1.241963570449033e-05,
      "loss": 0.7863,
      "step": 6221
    },
    {
      "epoch": 3.3403912620381866,
      "grad_norm": 0.66015625,
      "learning_rate": 1.2412338456252498e-05,
      "loss": 0.805,
      "step": 6222
    },
    {
      "epoch": 3.3409281567732627,
      "grad_norm": 0.69140625,
      "learning_rate": 1.2405042644426248e-05,
      "loss": 0.9499,
      "step": 6223
    },
    {
      "epoch": 3.341465051508339,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2397748269844144e-05,
      "loss": 0.8158,
      "step": 6224
    },
    {
      "epoch": 3.3420019462434145,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2390455333338547e-05,
      "loss": 1.0434,
      "step": 6225
    },
    {
      "epoch": 3.3425388409784906,
      "grad_norm": 0.7890625,
      "learning_rate": 1.2383163835741692e-05,
      "loss": 0.8421,
      "step": 6226
    },
    {
      "epoch": 3.3430757357135668,
      "grad_norm": 0.70703125,
      "learning_rate": 1.2375873777885617e-05,
      "loss": 0.8171,
      "step": 6227
    },
    {
      "epoch": 3.343612630448643,
      "grad_norm": 0.73828125,
      "learning_rate": 1.2368585160602222e-05,
      "loss": 0.9369,
      "step": 6228
    },
    {
      "epoch": 3.3441495251837186,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2361297984723213e-05,
      "loss": 0.8971,
      "step": 6229
    },
    {
      "epoch": 3.3446864199187947,
      "grad_norm": 0.78125,
      "learning_rate": 1.2354012251080163e-05,
      "loss": 1.0165,
      "step": 6230
    },
    {
      "epoch": 3.345223314653871,
      "grad_norm": 0.7109375,
      "learning_rate": 1.2346727960504473e-05,
      "loss": 0.8555,
      "step": 6231
    },
    {
      "epoch": 3.3457602093889465,
      "grad_norm": 0.62109375,
      "learning_rate": 1.2339445113827356e-05,
      "loss": 0.7551,
      "step": 6232
    },
    {
      "epoch": 3.3462971041240226,
      "grad_norm": 0.73828125,
      "learning_rate": 1.233216371187988e-05,
      "loss": 0.8144,
      "step": 6233
    },
    {
      "epoch": 3.3468339988590987,
      "grad_norm": 0.7109375,
      "learning_rate": 1.2324883755492971e-05,
      "loss": 0.7931,
      "step": 6234
    },
    {
      "epoch": 3.347370893594175,
      "grad_norm": 0.78515625,
      "learning_rate": 1.2317605245497323e-05,
      "loss": 0.7344,
      "step": 6235
    },
    {
      "epoch": 3.3479077883292505,
      "grad_norm": 0.69140625,
      "learning_rate": 1.2310328182723528e-05,
      "loss": 0.8355,
      "step": 6236
    },
    {
      "epoch": 3.3484446830643266,
      "grad_norm": 0.76953125,
      "learning_rate": 1.2303052568001996e-05,
      "loss": 0.7781,
      "step": 6237
    },
    {
      "epoch": 3.3489815777994028,
      "grad_norm": 0.66796875,
      "learning_rate": 1.2295778402162949e-05,
      "loss": 0.7674,
      "step": 6238
    },
    {
      "epoch": 3.349518472534479,
      "grad_norm": 0.734375,
      "learning_rate": 1.2288505686036475e-05,
      "loss": 0.7675,
      "step": 6239
    },
    {
      "epoch": 3.3500553672695546,
      "grad_norm": 0.6640625,
      "learning_rate": 1.228123442045249e-05,
      "loss": 0.8565,
      "step": 6240
    },
    {
      "epoch": 3.3505922620046307,
      "grad_norm": 0.8515625,
      "learning_rate": 1.2273964606240718e-05,
      "loss": 0.9623,
      "step": 6241
    },
    {
      "epoch": 3.351129156739707,
      "grad_norm": 0.83984375,
      "learning_rate": 1.2266696244230747e-05,
      "loss": 0.7674,
      "step": 6242
    },
    {
      "epoch": 3.351666051474783,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2259429335251996e-05,
      "loss": 0.8609,
      "step": 6243
    },
    {
      "epoch": 3.3522029462098586,
      "grad_norm": 0.6640625,
      "learning_rate": 1.2252163880133707e-05,
      "loss": 0.8189,
      "step": 6244
    },
    {
      "epoch": 3.3527398409449347,
      "grad_norm": 0.60546875,
      "learning_rate": 1.2244899879704947e-05,
      "loss": 0.8592,
      "step": 6245
    },
    {
      "epoch": 3.353276735680011,
      "grad_norm": 0.58984375,
      "learning_rate": 1.2237637334794639e-05,
      "loss": 0.9223,
      "step": 6246
    },
    {
      "epoch": 3.353813630415087,
      "grad_norm": 0.69140625,
      "learning_rate": 1.2230376246231546e-05,
      "loss": 0.9225,
      "step": 6247
    },
    {
      "epoch": 3.3543505251501626,
      "grad_norm": 0.6640625,
      "learning_rate": 1.2223116614844227e-05,
      "loss": 0.9264,
      "step": 6248
    },
    {
      "epoch": 3.3548874198852388,
      "grad_norm": 0.703125,
      "learning_rate": 1.2215858441461107e-05,
      "loss": 0.8881,
      "step": 6249
    },
    {
      "epoch": 3.355424314620315,
      "grad_norm": 0.75390625,
      "learning_rate": 1.2208601726910446e-05,
      "loss": 1.0364,
      "step": 6250
    },
    {
      "epoch": 3.355424314620315,
      "eval_loss": 0.9032419919967651,
      "eval_runtime": 46.3679,
      "eval_samples_per_second": 20.747,
      "eval_steps_per_second": 20.747,
      "step": 6250
    },
    {
      "epoch": 3.3559612093553906,
      "grad_norm": 0.87109375,
      "learning_rate": 1.220134647202032e-05,
      "loss": 0.9675,
      "step": 6251
    },
    {
      "epoch": 3.3564981040904667,
      "grad_norm": 0.68359375,
      "learning_rate": 1.2194092677618632e-05,
      "loss": 0.8381,
      "step": 6252
    },
    {
      "epoch": 3.357034998825543,
      "grad_norm": 0.640625,
      "learning_rate": 1.2186840344533154e-05,
      "loss": 0.8053,
      "step": 6253
    },
    {
      "epoch": 3.357571893560619,
      "grad_norm": 0.6796875,
      "learning_rate": 1.217958947359145e-05,
      "loss": 0.6974,
      "step": 6254
    },
    {
      "epoch": 3.3581087882956946,
      "grad_norm": 0.734375,
      "learning_rate": 1.2172340065620942e-05,
      "loss": 0.8768,
      "step": 6255
    },
    {
      "epoch": 3.3586456830307707,
      "grad_norm": 0.7578125,
      "learning_rate": 1.2165092121448893e-05,
      "loss": 0.8516,
      "step": 6256
    },
    {
      "epoch": 3.359182577765847,
      "grad_norm": 0.65625,
      "learning_rate": 1.2157845641902363e-05,
      "loss": 0.9253,
      "step": 6257
    },
    {
      "epoch": 3.359719472500923,
      "grad_norm": 0.66796875,
      "learning_rate": 1.2150600627808278e-05,
      "loss": 0.9366,
      "step": 6258
    },
    {
      "epoch": 3.3602563672359986,
      "grad_norm": 0.7265625,
      "learning_rate": 1.2143357079993395e-05,
      "loss": 0.9551,
      "step": 6259
    },
    {
      "epoch": 3.3607932619710748,
      "grad_norm": 0.72265625,
      "learning_rate": 1.2136114999284288e-05,
      "loss": 0.8466,
      "step": 6260
    },
    {
      "epoch": 3.361330156706151,
      "grad_norm": 0.71875,
      "learning_rate": 1.2128874386507355e-05,
      "loss": 1.0039,
      "step": 6261
    },
    {
      "epoch": 3.361867051441227,
      "grad_norm": 0.734375,
      "learning_rate": 1.2121635242488855e-05,
      "loss": 0.8287,
      "step": 6262
    },
    {
      "epoch": 3.3624039461763027,
      "grad_norm": 0.73046875,
      "learning_rate": 1.2114397568054874e-05,
      "loss": 0.8006,
      "step": 6263
    },
    {
      "epoch": 3.362940840911379,
      "grad_norm": 0.78515625,
      "learning_rate": 1.2107161364031305e-05,
      "loss": 0.8989,
      "step": 6264
    },
    {
      "epoch": 3.363477735646455,
      "grad_norm": 0.62109375,
      "learning_rate": 1.209992663124391e-05,
      "loss": 1.1105,
      "step": 6265
    },
    {
      "epoch": 3.364014630381531,
      "grad_norm": 0.6953125,
      "learning_rate": 1.209269337051824e-05,
      "loss": 0.9349,
      "step": 6266
    },
    {
      "epoch": 3.3645515251166067,
      "grad_norm": 0.53515625,
      "learning_rate": 1.2085461582679715e-05,
      "loss": 0.7596,
      "step": 6267
    },
    {
      "epoch": 3.365088419851683,
      "grad_norm": 0.890625,
      "learning_rate": 1.2078231268553585e-05,
      "loss": 0.9907,
      "step": 6268
    },
    {
      "epoch": 3.365625314586759,
      "grad_norm": 0.7109375,
      "learning_rate": 1.2071002428964905e-05,
      "loss": 0.9681,
      "step": 6269
    },
    {
      "epoch": 3.3661622093218346,
      "grad_norm": 0.67578125,
      "learning_rate": 1.2063775064738572e-05,
      "loss": 0.695,
      "step": 6270
    },
    {
      "epoch": 3.3666991040569108,
      "grad_norm": 0.6953125,
      "learning_rate": 1.2056549176699331e-05,
      "loss": 0.8351,
      "step": 6271
    },
    {
      "epoch": 3.367235998791987,
      "grad_norm": 0.66796875,
      "learning_rate": 1.2049324765671749e-05,
      "loss": 0.9598,
      "step": 6272
    },
    {
      "epoch": 3.367772893527063,
      "grad_norm": 0.7265625,
      "learning_rate": 1.2042101832480212e-05,
      "loss": 0.8427,
      "step": 6273
    },
    {
      "epoch": 3.3683097882621387,
      "grad_norm": 0.76953125,
      "learning_rate": 1.2034880377948953e-05,
      "loss": 0.7668,
      "step": 6274
    },
    {
      "epoch": 3.368846682997215,
      "grad_norm": 0.59375,
      "learning_rate": 1.2027660402902044e-05,
      "loss": 0.6131,
      "step": 6275
    },
    {
      "epoch": 3.369383577732291,
      "grad_norm": 0.625,
      "learning_rate": 1.2020441908163362e-05,
      "loss": 0.8759,
      "step": 6276
    },
    {
      "epoch": 3.369920472467367,
      "grad_norm": 0.69921875,
      "learning_rate": 1.2013224894556625e-05,
      "loss": 1.0069,
      "step": 6277
    },
    {
      "epoch": 3.3704573672024427,
      "grad_norm": 0.69921875,
      "learning_rate": 1.2006009362905396e-05,
      "loss": 0.7265,
      "step": 6278
    },
    {
      "epoch": 3.370994261937519,
      "grad_norm": 0.64453125,
      "learning_rate": 1.1998795314033048e-05,
      "loss": 0.7634,
      "step": 6279
    },
    {
      "epoch": 3.371531156672595,
      "grad_norm": 0.71875,
      "learning_rate": 1.19915827487628e-05,
      "loss": 0.768,
      "step": 6280
    },
    {
      "epoch": 3.372068051407671,
      "grad_norm": 0.65234375,
      "learning_rate": 1.198437166791771e-05,
      "loss": 0.7861,
      "step": 6281
    },
    {
      "epoch": 3.3726049461427468,
      "grad_norm": 0.640625,
      "learning_rate": 1.1977162072320632e-05,
      "loss": 0.6923,
      "step": 6282
    },
    {
      "epoch": 3.373141840877823,
      "grad_norm": 0.69140625,
      "learning_rate": 1.1969953962794284e-05,
      "loss": 0.9575,
      "step": 6283
    },
    {
      "epoch": 3.373678735612899,
      "grad_norm": 0.6640625,
      "learning_rate": 1.196274734016121e-05,
      "loss": 0.848,
      "step": 6284
    },
    {
      "epoch": 3.374215630347975,
      "grad_norm": 0.81640625,
      "learning_rate": 1.1955542205243767e-05,
      "loss": 0.8451,
      "step": 6285
    },
    {
      "epoch": 3.374752525083051,
      "grad_norm": 0.6953125,
      "learning_rate": 1.1948338558864144e-05,
      "loss": 0.7274,
      "step": 6286
    },
    {
      "epoch": 3.375289419818127,
      "grad_norm": 0.65234375,
      "learning_rate": 1.1941136401844377e-05,
      "loss": 0.6763,
      "step": 6287
    },
    {
      "epoch": 3.375826314553203,
      "grad_norm": 0.6171875,
      "learning_rate": 1.1933935735006333e-05,
      "loss": 0.8213,
      "step": 6288
    },
    {
      "epoch": 3.3763632092882787,
      "grad_norm": 0.71484375,
      "learning_rate": 1.192673655917168e-05,
      "loss": 0.989,
      "step": 6289
    },
    {
      "epoch": 3.376900104023355,
      "grad_norm": 0.6796875,
      "learning_rate": 1.1919538875161945e-05,
      "loss": 0.8226,
      "step": 6290
    },
    {
      "epoch": 3.377436998758431,
      "grad_norm": 0.78125,
      "learning_rate": 1.1912342683798486e-05,
      "loss": 0.859,
      "step": 6291
    },
    {
      "epoch": 3.377973893493507,
      "grad_norm": 0.6953125,
      "learning_rate": 1.190514798590246e-05,
      "loss": 0.8144,
      "step": 6292
    },
    {
      "epoch": 3.3785107882285828,
      "grad_norm": 0.57421875,
      "learning_rate": 1.1897954782294888e-05,
      "loss": 0.8307,
      "step": 6293
    },
    {
      "epoch": 3.379047682963659,
      "grad_norm": 0.859375,
      "learning_rate": 1.18907630737966e-05,
      "loss": 0.7958,
      "step": 6294
    },
    {
      "epoch": 3.379584577698735,
      "grad_norm": 0.60546875,
      "learning_rate": 1.1883572861228253e-05,
      "loss": 0.8073,
      "step": 6295
    },
    {
      "epoch": 3.380121472433811,
      "grad_norm": 0.6484375,
      "learning_rate": 1.1876384145410346e-05,
      "loss": 0.8633,
      "step": 6296
    },
    {
      "epoch": 3.380658367168887,
      "grad_norm": 0.64453125,
      "learning_rate": 1.1869196927163218e-05,
      "loss": 0.8272,
      "step": 6297
    },
    {
      "epoch": 3.381195261903963,
      "grad_norm": 0.7890625,
      "learning_rate": 1.1862011207306997e-05,
      "loss": 0.8016,
      "step": 6298
    },
    {
      "epoch": 3.381732156639039,
      "grad_norm": 0.77734375,
      "learning_rate": 1.1854826986661677e-05,
      "loss": 0.8162,
      "step": 6299
    },
    {
      "epoch": 3.382269051374115,
      "grad_norm": 0.86328125,
      "learning_rate": 1.1847644266047076e-05,
      "loss": 0.7611,
      "step": 6300
    },
    {
      "epoch": 3.382269051374115,
      "eval_loss": 0.9031023383140564,
      "eval_runtime": 46.2991,
      "eval_samples_per_second": 20.778,
      "eval_steps_per_second": 20.778,
      "step": 6300
    },
    {
      "epoch": 3.382805946109191,
      "grad_norm": 0.69140625,
      "learning_rate": 1.1840463046282827e-05,
      "loss": 0.8548,
      "step": 6301
    },
    {
      "epoch": 3.383342840844267,
      "grad_norm": 0.76953125,
      "learning_rate": 1.1833283328188385e-05,
      "loss": 0.8113,
      "step": 6302
    },
    {
      "epoch": 3.383879735579343,
      "grad_norm": 0.6875,
      "learning_rate": 1.1826105112583061e-05,
      "loss": 0.6172,
      "step": 6303
    },
    {
      "epoch": 3.384416630314419,
      "grad_norm": 0.7734375,
      "learning_rate": 1.1818928400285986e-05,
      "loss": 1.1469,
      "step": 6304
    },
    {
      "epoch": 3.384953525049495,
      "grad_norm": 0.68359375,
      "learning_rate": 1.1811753192116096e-05,
      "loss": 0.9609,
      "step": 6305
    },
    {
      "epoch": 3.385490419784571,
      "grad_norm": 0.703125,
      "learning_rate": 1.1804579488892182e-05,
      "loss": 0.8386,
      "step": 6306
    },
    {
      "epoch": 3.386027314519647,
      "grad_norm": 0.77734375,
      "learning_rate": 1.1797407291432864e-05,
      "loss": 1.017,
      "step": 6307
    },
    {
      "epoch": 3.386564209254723,
      "grad_norm": 0.6796875,
      "learning_rate": 1.1790236600556562e-05,
      "loss": 0.7762,
      "step": 6308
    },
    {
      "epoch": 3.387101103989799,
      "grad_norm": 0.765625,
      "learning_rate": 1.1783067417081561e-05,
      "loss": 0.865,
      "step": 6309
    },
    {
      "epoch": 3.387637998724875,
      "grad_norm": 0.69921875,
      "learning_rate": 1.1775899741825947e-05,
      "loss": 0.8347,
      "step": 6310
    },
    {
      "epoch": 3.388174893459951,
      "grad_norm": 0.69921875,
      "learning_rate": 1.1768733575607633e-05,
      "loss": 0.7458,
      "step": 6311
    },
    {
      "epoch": 3.388711788195027,
      "grad_norm": 0.62109375,
      "learning_rate": 1.1761568919244378e-05,
      "loss": 0.9071,
      "step": 6312
    },
    {
      "epoch": 3.389248682930103,
      "grad_norm": 0.72265625,
      "learning_rate": 1.175440577355377e-05,
      "loss": 1.0371,
      "step": 6313
    },
    {
      "epoch": 3.389785577665179,
      "grad_norm": 0.6796875,
      "learning_rate": 1.1747244139353197e-05,
      "loss": 0.8916,
      "step": 6314
    },
    {
      "epoch": 3.390322472400255,
      "grad_norm": 0.60546875,
      "learning_rate": 1.1740084017459898e-05,
      "loss": 0.8818,
      "step": 6315
    },
    {
      "epoch": 3.390859367135331,
      "grad_norm": 0.7421875,
      "learning_rate": 1.1732925408690948e-05,
      "loss": 0.7905,
      "step": 6316
    },
    {
      "epoch": 3.391396261870407,
      "grad_norm": 0.69140625,
      "learning_rate": 1.1725768313863211e-05,
      "loss": 0.875,
      "step": 6317
    },
    {
      "epoch": 3.391933156605483,
      "grad_norm": 0.734375,
      "learning_rate": 1.1718612733793424e-05,
      "loss": 0.7773,
      "step": 6318
    },
    {
      "epoch": 3.3924700513405592,
      "grad_norm": 0.7421875,
      "learning_rate": 1.171145866929812e-05,
      "loss": 1.0167,
      "step": 6319
    },
    {
      "epoch": 3.393006946075635,
      "grad_norm": 0.75,
      "learning_rate": 1.170430612119366e-05,
      "loss": 0.768,
      "step": 6320
    },
    {
      "epoch": 3.393543840810711,
      "grad_norm": 0.64453125,
      "learning_rate": 1.1697155090296244e-05,
      "loss": 0.7505,
      "step": 6321
    },
    {
      "epoch": 3.394080735545787,
      "grad_norm": 0.73046875,
      "learning_rate": 1.1690005577421908e-05,
      "loss": 0.9494,
      "step": 6322
    },
    {
      "epoch": 3.3946176302808633,
      "grad_norm": 0.609375,
      "learning_rate": 1.1682857583386485e-05,
      "loss": 0.9592,
      "step": 6323
    },
    {
      "epoch": 3.395154525015939,
      "grad_norm": 0.8359375,
      "learning_rate": 1.1675711109005664e-05,
      "loss": 0.8081,
      "step": 6324
    },
    {
      "epoch": 3.395691419751015,
      "grad_norm": 0.59375,
      "learning_rate": 1.1668566155094949e-05,
      "loss": 0.7074,
      "step": 6325
    },
    {
      "epoch": 3.396228314486091,
      "grad_norm": 0.64453125,
      "learning_rate": 1.1661422722469656e-05,
      "loss": 0.779,
      "step": 6326
    },
    {
      "epoch": 3.396765209221167,
      "grad_norm": 0.85546875,
      "learning_rate": 1.165428081194496e-05,
      "loss": 0.8498,
      "step": 6327
    },
    {
      "epoch": 3.397302103956243,
      "grad_norm": 0.78125,
      "learning_rate": 1.164714042433582e-05,
      "loss": 0.763,
      "step": 6328
    },
    {
      "epoch": 3.397838998691319,
      "grad_norm": 0.6640625,
      "learning_rate": 1.164000156045707e-05,
      "loss": 0.8777,
      "step": 6329
    },
    {
      "epoch": 3.3983758934263952,
      "grad_norm": 1.5078125,
      "learning_rate": 1.1632864221123324e-05,
      "loss": 1.0502,
      "step": 6330
    },
    {
      "epoch": 3.398912788161471,
      "grad_norm": 0.75,
      "learning_rate": 1.1625728407149054e-05,
      "loss": 0.8756,
      "step": 6331
    },
    {
      "epoch": 3.399449682896547,
      "grad_norm": 0.72265625,
      "learning_rate": 1.1618594119348548e-05,
      "loss": 0.9094,
      "step": 6332
    },
    {
      "epoch": 3.399986577631623,
      "grad_norm": 0.7109375,
      "learning_rate": 1.1611461358535908e-05,
      "loss": 0.957,
      "step": 6333
    },
    {
      "epoch": 3.4005234723666993,
      "grad_norm": 0.6640625,
      "learning_rate": 1.1604330125525079e-05,
      "loss": 0.8582,
      "step": 6334
    },
    {
      "epoch": 3.401060367101775,
      "grad_norm": 0.67578125,
      "learning_rate": 1.1597200421129844e-05,
      "loss": 0.9101,
      "step": 6335
    },
    {
      "epoch": 3.401597261836851,
      "grad_norm": 0.75,
      "learning_rate": 1.1590072246163756e-05,
      "loss": 0.809,
      "step": 6336
    },
    {
      "epoch": 3.402134156571927,
      "grad_norm": 0.65625,
      "learning_rate": 1.1582945601440246e-05,
      "loss": 0.7188,
      "step": 6337
    },
    {
      "epoch": 3.4026710513070033,
      "grad_norm": 0.65625,
      "learning_rate": 1.1575820487772566e-05,
      "loss": 0.768,
      "step": 6338
    },
    {
      "epoch": 3.403207946042079,
      "grad_norm": 0.96484375,
      "learning_rate": 1.1568696905973761e-05,
      "loss": 1.112,
      "step": 6339
    },
    {
      "epoch": 3.403744840777155,
      "grad_norm": 0.6796875,
      "learning_rate": 1.1561574856856736e-05,
      "loss": 0.8483,
      "step": 6340
    },
    {
      "epoch": 3.4042817355122312,
      "grad_norm": 0.7421875,
      "learning_rate": 1.1554454341234211e-05,
      "loss": 0.8113,
      "step": 6341
    },
    {
      "epoch": 3.4048186302473074,
      "grad_norm": 0.6875,
      "learning_rate": 1.154733535991871e-05,
      "loss": 0.7565,
      "step": 6342
    },
    {
      "epoch": 3.405355524982383,
      "grad_norm": 0.65234375,
      "learning_rate": 1.1540217913722621e-05,
      "loss": 0.7865,
      "step": 6343
    },
    {
      "epoch": 3.405892419717459,
      "grad_norm": 0.61328125,
      "learning_rate": 1.153310200345811e-05,
      "loss": 0.6929,
      "step": 6344
    },
    {
      "epoch": 3.4064293144525353,
      "grad_norm": 0.765625,
      "learning_rate": 1.1525987629937212e-05,
      "loss": 0.9575,
      "step": 6345
    },
    {
      "epoch": 3.406966209187611,
      "grad_norm": 0.671875,
      "learning_rate": 1.1518874793971754e-05,
      "loss": 0.9249,
      "step": 6346
    },
    {
      "epoch": 3.407503103922687,
      "grad_norm": 0.734375,
      "learning_rate": 1.1511763496373407e-05,
      "loss": 0.8732,
      "step": 6347
    },
    {
      "epoch": 3.408039998657763,
      "grad_norm": 0.70703125,
      "learning_rate": 1.1504653737953669e-05,
      "loss": 0.9182,
      "step": 6348
    },
    {
      "epoch": 3.4085768933928393,
      "grad_norm": 0.65625,
      "learning_rate": 1.1497545519523834e-05,
      "loss": 0.811,
      "step": 6349
    },
    {
      "epoch": 3.409113788127915,
      "grad_norm": 0.671875,
      "learning_rate": 1.149043884189506e-05,
      "loss": 0.8938,
      "step": 6350
    },
    {
      "epoch": 3.409113788127915,
      "eval_loss": 0.9030389189720154,
      "eval_runtime": 45.9493,
      "eval_samples_per_second": 20.936,
      "eval_steps_per_second": 20.936,
      "step": 6350
    },
    {
      "epoch": 3.409650682862991,
      "grad_norm": 0.6015625,
      "learning_rate": 1.1483333705878289e-05,
      "loss": 0.7719,
      "step": 6351
    },
    {
      "epoch": 3.4101875775980672,
      "grad_norm": 0.671875,
      "learning_rate": 1.1476230112284325e-05,
      "loss": 0.7383,
      "step": 6352
    },
    {
      "epoch": 3.4107244723331434,
      "grad_norm": 0.71484375,
      "learning_rate": 1.1469128061923761e-05,
      "loss": 0.8725,
      "step": 6353
    },
    {
      "epoch": 3.411261367068219,
      "grad_norm": 0.69921875,
      "learning_rate": 1.1462027555607052e-05,
      "loss": 1.072,
      "step": 6354
    },
    {
      "epoch": 3.411798261803295,
      "grad_norm": 0.69140625,
      "learning_rate": 1.1454928594144431e-05,
      "loss": 0.8541,
      "step": 6355
    },
    {
      "epoch": 3.4123351565383713,
      "grad_norm": 0.734375,
      "learning_rate": 1.1447831178345994e-05,
      "loss": 0.8959,
      "step": 6356
    },
    {
      "epoch": 3.4128720512734474,
      "grad_norm": 0.57421875,
      "learning_rate": 1.1440735309021652e-05,
      "loss": 0.7549,
      "step": 6357
    },
    {
      "epoch": 3.413408946008523,
      "grad_norm": 0.80859375,
      "learning_rate": 1.1433640986981114e-05,
      "loss": 1.0197,
      "step": 6358
    },
    {
      "epoch": 3.413945840743599,
      "grad_norm": 0.7109375,
      "learning_rate": 1.1426548213033947e-05,
      "loss": 0.7,
      "step": 6359
    },
    {
      "epoch": 3.4144827354786753,
      "grad_norm": 0.73046875,
      "learning_rate": 1.141945698798954e-05,
      "loss": 0.9265,
      "step": 6360
    },
    {
      "epoch": 3.4150196302137514,
      "grad_norm": 0.58203125,
      "learning_rate": 1.1412367312657058e-05,
      "loss": 0.7631,
      "step": 6361
    },
    {
      "epoch": 3.415556524948827,
      "grad_norm": 0.75,
      "learning_rate": 1.1405279187845536e-05,
      "loss": 0.8483,
      "step": 6362
    },
    {
      "epoch": 3.4160934196839032,
      "grad_norm": 0.93359375,
      "learning_rate": 1.1398192614363832e-05,
      "loss": 0.9273,
      "step": 6363
    },
    {
      "epoch": 3.4166303144189794,
      "grad_norm": 0.765625,
      "learning_rate": 1.1391107593020595e-05,
      "loss": 0.8118,
      "step": 6364
    },
    {
      "epoch": 3.417167209154055,
      "grad_norm": 0.58203125,
      "learning_rate": 1.1384024124624324e-05,
      "loss": 0.8016,
      "step": 6365
    },
    {
      "epoch": 3.417704103889131,
      "grad_norm": 0.69140625,
      "learning_rate": 1.1376942209983341e-05,
      "loss": 0.8617,
      "step": 6366
    },
    {
      "epoch": 3.4182409986242073,
      "grad_norm": 0.53515625,
      "learning_rate": 1.1369861849905764e-05,
      "loss": 0.6418,
      "step": 6367
    },
    {
      "epoch": 3.4187778933592834,
      "grad_norm": 0.6640625,
      "learning_rate": 1.1362783045199563e-05,
      "loss": 0.9412,
      "step": 6368
    },
    {
      "epoch": 3.419314788094359,
      "grad_norm": 0.58203125,
      "learning_rate": 1.1355705796672528e-05,
      "loss": 0.7868,
      "step": 6369
    },
    {
      "epoch": 3.419851682829435,
      "grad_norm": 0.8125,
      "learning_rate": 1.1348630105132253e-05,
      "loss": 0.8543,
      "step": 6370
    },
    {
      "epoch": 3.4203885775645113,
      "grad_norm": 0.73046875,
      "learning_rate": 1.1341555971386151e-05,
      "loss": 0.7752,
      "step": 6371
    },
    {
      "epoch": 3.4209254722995874,
      "grad_norm": 0.69140625,
      "learning_rate": 1.1334483396241486e-05,
      "loss": 0.8137,
      "step": 6372
    },
    {
      "epoch": 3.421462367034663,
      "grad_norm": 0.734375,
      "learning_rate": 1.1327412380505334e-05,
      "loss": 0.9258,
      "step": 6373
    },
    {
      "epoch": 3.4219992617697392,
      "grad_norm": 0.64453125,
      "learning_rate": 1.1320342924984567e-05,
      "loss": 0.9591,
      "step": 6374
    },
    {
      "epoch": 3.4225361565048154,
      "grad_norm": 0.72265625,
      "learning_rate": 1.1313275030485914e-05,
      "loss": 0.8314,
      "step": 6375
    },
    {
      "epoch": 3.423073051239891,
      "grad_norm": 0.71875,
      "learning_rate": 1.1306208697815917e-05,
      "loss": 0.9276,
      "step": 6376
    },
    {
      "epoch": 3.423609945974967,
      "grad_norm": 0.72265625,
      "learning_rate": 1.1299143927780926e-05,
      "loss": 0.8003,
      "step": 6377
    },
    {
      "epoch": 3.4241468407100433,
      "grad_norm": 0.89453125,
      "learning_rate": 1.129208072118711e-05,
      "loss": 0.8508,
      "step": 6378
    },
    {
      "epoch": 3.4246837354451194,
      "grad_norm": 0.70703125,
      "learning_rate": 1.128501907884049e-05,
      "loss": 0.8766,
      "step": 6379
    },
    {
      "epoch": 3.4252206301801955,
      "grad_norm": 0.66796875,
      "learning_rate": 1.1277959001546867e-05,
      "loss": 0.9165,
      "step": 6380
    },
    {
      "epoch": 3.425757524915271,
      "grad_norm": 0.69140625,
      "learning_rate": 1.1270900490111902e-05,
      "loss": 0.9921,
      "step": 6381
    },
    {
      "epoch": 3.4262944196503473,
      "grad_norm": 0.7578125,
      "learning_rate": 1.1263843545341063e-05,
      "loss": 1.0443,
      "step": 6382
    },
    {
      "epoch": 3.4268313143854234,
      "grad_norm": 0.69921875,
      "learning_rate": 1.1256788168039623e-05,
      "loss": 0.8628,
      "step": 6383
    },
    {
      "epoch": 3.427368209120499,
      "grad_norm": 0.765625,
      "learning_rate": 1.1249734359012692e-05,
      "loss": 0.8578,
      "step": 6384
    },
    {
      "epoch": 3.4279051038555752,
      "grad_norm": 0.71875,
      "learning_rate": 1.1242682119065218e-05,
      "loss": 1.0051,
      "step": 6385
    },
    {
      "epoch": 3.4284419985906514,
      "grad_norm": 0.71484375,
      "learning_rate": 1.1235631449001934e-05,
      "loss": 0.7941,
      "step": 6386
    },
    {
      "epoch": 3.4289788933257275,
      "grad_norm": 0.6015625,
      "learning_rate": 1.1228582349627403e-05,
      "loss": 0.932,
      "step": 6387
    },
    {
      "epoch": 3.429515788060803,
      "grad_norm": 0.625,
      "learning_rate": 1.122153482174603e-05,
      "loss": 0.7477,
      "step": 6388
    },
    {
      "epoch": 3.4300526827958793,
      "grad_norm": 0.6875,
      "learning_rate": 1.1214488866162032e-05,
      "loss": 0.9474,
      "step": 6389
    },
    {
      "epoch": 3.4305895775309554,
      "grad_norm": 0.640625,
      "learning_rate": 1.1207444483679422e-05,
      "loss": 0.8235,
      "step": 6390
    },
    {
      "epoch": 3.4311264722660315,
      "grad_norm": 0.65234375,
      "learning_rate": 1.1200401675102076e-05,
      "loss": 0.7388,
      "step": 6391
    },
    {
      "epoch": 3.431663367001107,
      "grad_norm": 0.7578125,
      "learning_rate": 1.119336044123365e-05,
      "loss": 0.8663,
      "step": 6392
    },
    {
      "epoch": 3.4322002617361833,
      "grad_norm": 0.69140625,
      "learning_rate": 1.118632078287764e-05,
      "loss": 0.7474,
      "step": 6393
    },
    {
      "epoch": 3.4327371564712594,
      "grad_norm": 0.73828125,
      "learning_rate": 1.1179282700837377e-05,
      "loss": 0.9538,
      "step": 6394
    },
    {
      "epoch": 3.433274051206335,
      "grad_norm": 0.79296875,
      "learning_rate": 1.1172246195915983e-05,
      "loss": 0.9978,
      "step": 6395
    },
    {
      "epoch": 3.4338109459414112,
      "grad_norm": 0.60546875,
      "learning_rate": 1.11652112689164e-05,
      "loss": 0.9181,
      "step": 6396
    },
    {
      "epoch": 3.4343478406764874,
      "grad_norm": 0.69921875,
      "learning_rate": 1.115817792064142e-05,
      "loss": 0.7215,
      "step": 6397
    },
    {
      "epoch": 3.4348847354115635,
      "grad_norm": 0.71875,
      "learning_rate": 1.1151146151893638e-05,
      "loss": 0.9043,
      "step": 6398
    },
    {
      "epoch": 3.4354216301466396,
      "grad_norm": 0.7109375,
      "learning_rate": 1.1144115963475452e-05,
      "loss": 0.7626,
      "step": 6399
    },
    {
      "epoch": 3.4359585248817153,
      "grad_norm": 0.68359375,
      "learning_rate": 1.1137087356189104e-05,
      "loss": 0.9085,
      "step": 6400
    },
    {
      "epoch": 3.4359585248817153,
      "eval_loss": 0.902956485748291,
      "eval_runtime": 45.9769,
      "eval_samples_per_second": 20.924,
      "eval_steps_per_second": 20.924,
      "step": 6400
    },
    {
      "epoch": 3.4364954196167914,
      "grad_norm": 0.62890625,
      "learning_rate": 1.1130060330836659e-05,
      "loss": 0.8724,
      "step": 6401
    },
    {
      "epoch": 3.4370323143518675,
      "grad_norm": 0.85546875,
      "learning_rate": 1.112303488821998e-05,
      "loss": 0.7619,
      "step": 6402
    },
    {
      "epoch": 3.437569209086943,
      "grad_norm": 0.7109375,
      "learning_rate": 1.1116011029140747e-05,
      "loss": 0.783,
      "step": 6403
    },
    {
      "epoch": 3.4381061038220193,
      "grad_norm": 0.78515625,
      "learning_rate": 1.1108988754400493e-05,
      "loss": 1.0795,
      "step": 6404
    },
    {
      "epoch": 3.4386429985570954,
      "grad_norm": 0.72265625,
      "learning_rate": 1.1101968064800525e-05,
      "loss": 0.9267,
      "step": 6405
    },
    {
      "epoch": 3.4391798932921716,
      "grad_norm": 1.4453125,
      "learning_rate": 1.1094948961142008e-05,
      "loss": 0.947,
      "step": 6406
    },
    {
      "epoch": 3.4397167880272472,
      "grad_norm": 0.7421875,
      "learning_rate": 1.1087931444225915e-05,
      "loss": 0.8666,
      "step": 6407
    },
    {
      "epoch": 3.4402536827623234,
      "grad_norm": 0.6953125,
      "learning_rate": 1.108091551485302e-05,
      "loss": 0.926,
      "step": 6408
    },
    {
      "epoch": 3.4407905774973995,
      "grad_norm": 0.9296875,
      "learning_rate": 1.1073901173823934e-05,
      "loss": 1.0292,
      "step": 6409
    },
    {
      "epoch": 3.4413274722324756,
      "grad_norm": 0.796875,
      "learning_rate": 1.1066888421939093e-05,
      "loss": 0.7754,
      "step": 6410
    },
    {
      "epoch": 3.4418643669675513,
      "grad_norm": 0.58203125,
      "learning_rate": 1.105987725999873e-05,
      "loss": 0.6972,
      "step": 6411
    },
    {
      "epoch": 3.4424012617026274,
      "grad_norm": 0.67578125,
      "learning_rate": 1.10528676888029e-05,
      "loss": 0.7376,
      "step": 6412
    },
    {
      "epoch": 3.4429381564377035,
      "grad_norm": 0.6953125,
      "learning_rate": 1.1045859709151496e-05,
      "loss": 0.9169,
      "step": 6413
    },
    {
      "epoch": 3.443475051172779,
      "grad_norm": 0.703125,
      "learning_rate": 1.1038853321844222e-05,
      "loss": 0.8072,
      "step": 6414
    },
    {
      "epoch": 3.4440119459078553,
      "grad_norm": 0.765625,
      "learning_rate": 1.1031848527680577e-05,
      "loss": 0.8713,
      "step": 6415
    },
    {
      "epoch": 3.4445488406429314,
      "grad_norm": 0.79296875,
      "learning_rate": 1.102484532745991e-05,
      "loss": 0.8295,
      "step": 6416
    },
    {
      "epoch": 3.4450857353780076,
      "grad_norm": 0.6484375,
      "learning_rate": 1.101784372198138e-05,
      "loss": 0.7872,
      "step": 6417
    },
    {
      "epoch": 3.4456226301130837,
      "grad_norm": 0.68359375,
      "learning_rate": 1.1010843712043941e-05,
      "loss": 0.8727,
      "step": 6418
    },
    {
      "epoch": 3.4461595248481594,
      "grad_norm": 0.84375,
      "learning_rate": 1.1003845298446402e-05,
      "loss": 0.753,
      "step": 6419
    },
    {
      "epoch": 3.4466964195832355,
      "grad_norm": 0.734375,
      "learning_rate": 1.0996848481987362e-05,
      "loss": 0.9506,
      "step": 6420
    },
    {
      "epoch": 3.4472333143183116,
      "grad_norm": 0.78515625,
      "learning_rate": 1.0989853263465235e-05,
      "loss": 0.9232,
      "step": 6421
    },
    {
      "epoch": 3.4477702090533873,
      "grad_norm": 1.0703125,
      "learning_rate": 1.0982859643678277e-05,
      "loss": 0.873,
      "step": 6422
    },
    {
      "epoch": 3.4483071037884634,
      "grad_norm": 0.640625,
      "learning_rate": 1.0975867623424554e-05,
      "loss": 0.8102,
      "step": 6423
    },
    {
      "epoch": 3.4488439985235395,
      "grad_norm": 0.70703125,
      "learning_rate": 1.0968877203501929e-05,
      "loss": 0.7943,
      "step": 6424
    },
    {
      "epoch": 3.4493808932586156,
      "grad_norm": 0.70703125,
      "learning_rate": 1.0961888384708105e-05,
      "loss": 0.8762,
      "step": 6425
    },
    {
      "epoch": 3.4499177879936913,
      "grad_norm": 0.59765625,
      "learning_rate": 1.09549011678406e-05,
      "loss": 0.5962,
      "step": 6426
    },
    {
      "epoch": 3.4504546827287674,
      "grad_norm": 0.65625,
      "learning_rate": 1.0947915553696742e-05,
      "loss": 0.6815,
      "step": 6427
    },
    {
      "epoch": 3.4509915774638436,
      "grad_norm": 0.86328125,
      "learning_rate": 1.0940931543073665e-05,
      "loss": 0.9124,
      "step": 6428
    },
    {
      "epoch": 3.4515284721989197,
      "grad_norm": 0.75390625,
      "learning_rate": 1.0933949136768343e-05,
      "loss": 0.9224,
      "step": 6429
    },
    {
      "epoch": 3.4520653669339953,
      "grad_norm": 0.703125,
      "learning_rate": 1.0926968335577564e-05,
      "loss": 0.9123,
      "step": 6430
    },
    {
      "epoch": 3.4526022616690715,
      "grad_norm": 0.73046875,
      "learning_rate": 1.0919989140297911e-05,
      "loss": 0.7516,
      "step": 6431
    },
    {
      "epoch": 3.4531391564041476,
      "grad_norm": 0.6328125,
      "learning_rate": 1.0913011551725805e-05,
      "loss": 0.7664,
      "step": 6432
    },
    {
      "epoch": 3.4536760511392233,
      "grad_norm": 0.92578125,
      "learning_rate": 1.0906035570657488e-05,
      "loss": 0.7161,
      "step": 6433
    },
    {
      "epoch": 3.4542129458742994,
      "grad_norm": 0.59375,
      "learning_rate": 1.0899061197888987e-05,
      "loss": 1.035,
      "step": 6434
    },
    {
      "epoch": 3.4547498406093755,
      "grad_norm": 0.66796875,
      "learning_rate": 1.0892088434216188e-05,
      "loss": 0.823,
      "step": 6435
    },
    {
      "epoch": 3.4552867353444516,
      "grad_norm": 0.76953125,
      "learning_rate": 1.0885117280434762e-05,
      "loss": 0.9475,
      "step": 6436
    },
    {
      "epoch": 3.4558236300795278,
      "grad_norm": 0.81640625,
      "learning_rate": 1.0878147737340194e-05,
      "loss": 0.9158,
      "step": 6437
    },
    {
      "epoch": 3.4563605248146034,
      "grad_norm": 0.64453125,
      "learning_rate": 1.087117980572781e-05,
      "loss": 0.9155,
      "step": 6438
    },
    {
      "epoch": 3.4568974195496796,
      "grad_norm": 0.765625,
      "learning_rate": 1.0864213486392743e-05,
      "loss": 0.8528,
      "step": 6439
    },
    {
      "epoch": 3.4574343142847557,
      "grad_norm": 0.77734375,
      "learning_rate": 1.0857248780129928e-05,
      "loss": 0.808,
      "step": 6440
    },
    {
      "epoch": 3.4579712090198313,
      "grad_norm": 0.6015625,
      "learning_rate": 1.0850285687734126e-05,
      "loss": 0.8141,
      "step": 6441
    },
    {
      "epoch": 3.4585081037549075,
      "grad_norm": 1.8125,
      "learning_rate": 1.0843324209999931e-05,
      "loss": 0.8891,
      "step": 6442
    },
    {
      "epoch": 3.4590449984899836,
      "grad_norm": 0.65234375,
      "learning_rate": 1.083636434772171e-05,
      "loss": 0.7788,
      "step": 6443
    },
    {
      "epoch": 3.4595818932250597,
      "grad_norm": 0.75,
      "learning_rate": 1.0829406101693698e-05,
      "loss": 0.8557,
      "step": 6444
    },
    {
      "epoch": 3.4601187879601354,
      "grad_norm": 0.63671875,
      "learning_rate": 1.0822449472709905e-05,
      "loss": 0.7947,
      "step": 6445
    },
    {
      "epoch": 3.4606556826952115,
      "grad_norm": 0.7421875,
      "learning_rate": 1.0815494461564162e-05,
      "loss": 0.8934,
      "step": 6446
    },
    {
      "epoch": 3.4611925774302876,
      "grad_norm": 0.6171875,
      "learning_rate": 1.0808541069050131e-05,
      "loss": 0.7493,
      "step": 6447
    },
    {
      "epoch": 3.4617294721653638,
      "grad_norm": 0.7109375,
      "learning_rate": 1.0801589295961298e-05,
      "loss": 0.9463,
      "step": 6448
    },
    {
      "epoch": 3.4622663669004394,
      "grad_norm": 0.69921875,
      "learning_rate": 1.0794639143090924e-05,
      "loss": 0.904,
      "step": 6449
    },
    {
      "epoch": 3.4628032616355156,
      "grad_norm": 0.63671875,
      "learning_rate": 1.0787690611232118e-05,
      "loss": 0.8015,
      "step": 6450
    },
    {
      "epoch": 3.4628032616355156,
      "eval_loss": 0.9030494093894958,
      "eval_runtime": 46.6563,
      "eval_samples_per_second": 20.619,
      "eval_steps_per_second": 20.619,
      "step": 6450
    },
    {
      "epoch": 3.4633401563705917,
      "grad_norm": 0.73046875,
      "learning_rate": 1.0780743701177808e-05,
      "loss": 0.683,
      "step": 6451
    },
    {
      "epoch": 3.4638770511056673,
      "grad_norm": 0.59765625,
      "learning_rate": 1.0773798413720703e-05,
      "loss": 0.6482,
      "step": 6452
    },
    {
      "epoch": 3.4644139458407435,
      "grad_norm": 0.7265625,
      "learning_rate": 1.0766854749653372e-05,
      "loss": 0.7797,
      "step": 6453
    },
    {
      "epoch": 3.4649508405758196,
      "grad_norm": 0.73046875,
      "learning_rate": 1.0759912709768153e-05,
      "loss": 0.903,
      "step": 6454
    },
    {
      "epoch": 3.4654877353108957,
      "grad_norm": 0.57421875,
      "learning_rate": 1.0752972294857236e-05,
      "loss": 0.752,
      "step": 6455
    },
    {
      "epoch": 3.466024630045972,
      "grad_norm": 0.7109375,
      "learning_rate": 1.0746033505712597e-05,
      "loss": 0.6841,
      "step": 6456
    },
    {
      "epoch": 3.4665615247810475,
      "grad_norm": 0.66796875,
      "learning_rate": 1.0739096343126046e-05,
      "loss": 1.0571,
      "step": 6457
    },
    {
      "epoch": 3.4670984195161236,
      "grad_norm": 0.625,
      "learning_rate": 1.0732160807889211e-05,
      "loss": 0.7725,
      "step": 6458
    },
    {
      "epoch": 3.4676353142511998,
      "grad_norm": 0.6484375,
      "learning_rate": 1.0725226900793506e-05,
      "loss": 0.7736,
      "step": 6459
    },
    {
      "epoch": 3.4681722089862754,
      "grad_norm": 0.67578125,
      "learning_rate": 1.0718294622630188e-05,
      "loss": 0.9082,
      "step": 6460
    },
    {
      "epoch": 3.4687091037213515,
      "grad_norm": 0.6484375,
      "learning_rate": 1.0711363974190334e-05,
      "loss": 0.9389,
      "step": 6461
    },
    {
      "epoch": 3.4692459984564277,
      "grad_norm": 0.73046875,
      "learning_rate": 1.0704434956264783e-05,
      "loss": 0.808,
      "step": 6462
    },
    {
      "epoch": 3.469782893191504,
      "grad_norm": 0.74609375,
      "learning_rate": 1.0697507569644244e-05,
      "loss": 0.927,
      "step": 6463
    },
    {
      "epoch": 3.4703197879265795,
      "grad_norm": 0.9140625,
      "learning_rate": 1.0690581815119228e-05,
      "loss": 0.8797,
      "step": 6464
    },
    {
      "epoch": 3.4708566826616556,
      "grad_norm": 0.65234375,
      "learning_rate": 1.068365769348003e-05,
      "loss": 1.0282,
      "step": 6465
    },
    {
      "epoch": 3.4713935773967317,
      "grad_norm": 0.6015625,
      "learning_rate": 1.0676735205516788e-05,
      "loss": 0.7311,
      "step": 6466
    },
    {
      "epoch": 3.471930472131808,
      "grad_norm": 0.74609375,
      "learning_rate": 1.0669814352019462e-05,
      "loss": 0.7908,
      "step": 6467
    },
    {
      "epoch": 3.4724673668668835,
      "grad_norm": 0.734375,
      "learning_rate": 1.0662895133777784e-05,
      "loss": 0.8146,
      "step": 6468
    },
    {
      "epoch": 3.4730042616019596,
      "grad_norm": 0.65625,
      "learning_rate": 1.0655977551581345e-05,
      "loss": 0.9094,
      "step": 6469
    },
    {
      "epoch": 3.4735411563370358,
      "grad_norm": 0.69140625,
      "learning_rate": 1.0649061606219513e-05,
      "loss": 0.8647,
      "step": 6470
    },
    {
      "epoch": 3.4740780510721114,
      "grad_norm": 0.6953125,
      "learning_rate": 1.0642147298481498e-05,
      "loss": 0.7078,
      "step": 6471
    },
    {
      "epoch": 3.4746149458071875,
      "grad_norm": 0.7265625,
      "learning_rate": 1.0635234629156296e-05,
      "loss": 0.8195,
      "step": 6472
    },
    {
      "epoch": 3.4751518405422637,
      "grad_norm": 0.6875,
      "learning_rate": 1.0628323599032738e-05,
      "loss": 0.6716,
      "step": 6473
    },
    {
      "epoch": 3.47568873527734,
      "grad_norm": 0.76171875,
      "learning_rate": 1.062141420889947e-05,
      "loss": 0.8325,
      "step": 6474
    },
    {
      "epoch": 3.476225630012416,
      "grad_norm": 0.73046875,
      "learning_rate": 1.0614506459544921e-05,
      "loss": 0.8454,
      "step": 6475
    },
    {
      "epoch": 3.4767625247474916,
      "grad_norm": 0.62109375,
      "learning_rate": 1.0607600351757369e-05,
      "loss": 0.7848,
      "step": 6476
    },
    {
      "epoch": 3.4772994194825677,
      "grad_norm": 0.6875,
      "learning_rate": 1.0600695886324877e-05,
      "loss": 0.869,
      "step": 6477
    },
    {
      "epoch": 3.477836314217644,
      "grad_norm": 0.6953125,
      "learning_rate": 1.0593793064035343e-05,
      "loss": 0.8325,
      "step": 6478
    },
    {
      "epoch": 3.4783732089527195,
      "grad_norm": 0.62890625,
      "learning_rate": 1.0586891885676451e-05,
      "loss": 0.8221,
      "step": 6479
    },
    {
      "epoch": 3.4789101036877956,
      "grad_norm": 0.68359375,
      "learning_rate": 1.0579992352035734e-05,
      "loss": 0.8656,
      "step": 6480
    },
    {
      "epoch": 3.4794469984228718,
      "grad_norm": 0.72265625,
      "learning_rate": 1.057309446390049e-05,
      "loss": 0.7419,
      "step": 6481
    },
    {
      "epoch": 3.479983893157948,
      "grad_norm": 0.5234375,
      "learning_rate": 1.0566198222057875e-05,
      "loss": 0.7012,
      "step": 6482
    },
    {
      "epoch": 3.4805207878930235,
      "grad_norm": 0.7265625,
      "learning_rate": 1.0559303627294839e-05,
      "loss": 0.861,
      "step": 6483
    },
    {
      "epoch": 3.4810576826280997,
      "grad_norm": 0.640625,
      "learning_rate": 1.0552410680398126e-05,
      "loss": 0.8149,
      "step": 6484
    },
    {
      "epoch": 3.481594577363176,
      "grad_norm": 0.69140625,
      "learning_rate": 1.054551938215432e-05,
      "loss": 0.787,
      "step": 6485
    },
    {
      "epoch": 3.482131472098252,
      "grad_norm": 0.71875,
      "learning_rate": 1.053862973334982e-05,
      "loss": 0.9327,
      "step": 6486
    },
    {
      "epoch": 3.4826683668333276,
      "grad_norm": 0.77734375,
      "learning_rate": 1.0531741734770788e-05,
      "loss": 0.7855,
      "step": 6487
    },
    {
      "epoch": 3.4832052615684037,
      "grad_norm": 0.6875,
      "learning_rate": 1.0524855387203253e-05,
      "loss": 0.6862,
      "step": 6488
    },
    {
      "epoch": 3.48374215630348,
      "grad_norm": 0.65625,
      "learning_rate": 1.0517970691433035e-05,
      "loss": 0.8203,
      "step": 6489
    },
    {
      "epoch": 3.4842790510385555,
      "grad_norm": 0.71484375,
      "learning_rate": 1.0511087648245757e-05,
      "loss": 0.9173,
      "step": 6490
    },
    {
      "epoch": 3.4848159457736316,
      "grad_norm": 0.6640625,
      "learning_rate": 1.0504206258426864e-05,
      "loss": 0.8391,
      "step": 6491
    },
    {
      "epoch": 3.4853528405087078,
      "grad_norm": 0.6953125,
      "learning_rate": 1.049732652276162e-05,
      "loss": 0.7,
      "step": 6492
    },
    {
      "epoch": 3.485889735243784,
      "grad_norm": 0.7109375,
      "learning_rate": 1.049044844203507e-05,
      "loss": 0.837,
      "step": 6493
    },
    {
      "epoch": 3.48642662997886,
      "grad_norm": 0.84375,
      "learning_rate": 1.0483572017032112e-05,
      "loss": 0.945,
      "step": 6494
    },
    {
      "epoch": 3.4869635247139357,
      "grad_norm": 0.640625,
      "learning_rate": 1.0476697248537415e-05,
      "loss": 0.716,
      "step": 6495
    },
    {
      "epoch": 3.487500419449012,
      "grad_norm": 0.69921875,
      "learning_rate": 1.0469824137335491e-05,
      "loss": 0.8421,
      "step": 6496
    },
    {
      "epoch": 3.488037314184088,
      "grad_norm": 0.76171875,
      "learning_rate": 1.0462952684210636e-05,
      "loss": 0.9197,
      "step": 6497
    },
    {
      "epoch": 3.4885742089191636,
      "grad_norm": 0.62890625,
      "learning_rate": 1.0456082889946974e-05,
      "loss": 0.7482,
      "step": 6498
    },
    {
      "epoch": 3.4891111036542397,
      "grad_norm": 0.6015625,
      "learning_rate": 1.044921475532845e-05,
      "loss": 0.8445,
      "step": 6499
    },
    {
      "epoch": 3.489647998389316,
      "grad_norm": 0.62109375,
      "learning_rate": 1.0442348281138783e-05,
      "loss": 0.7286,
      "step": 6500
    },
    {
      "epoch": 3.489647998389316,
      "eval_loss": 0.903020441532135,
      "eval_runtime": 46.0667,
      "eval_samples_per_second": 20.883,
      "eval_steps_per_second": 20.883,
      "step": 6500
    },
    {
      "epoch": 3.490184893124392,
      "grad_norm": 0.80078125,
      "learning_rate": 1.0435483468161536e-05,
      "loss": 0.9734,
      "step": 6501
    },
    {
      "epoch": 3.4907217878594676,
      "grad_norm": 0.78125,
      "learning_rate": 1.0428620317180076e-05,
      "loss": 0.8584,
      "step": 6502
    },
    {
      "epoch": 3.4912586825945437,
      "grad_norm": 0.734375,
      "learning_rate": 1.0421758828977573e-05,
      "loss": 0.9207,
      "step": 6503
    },
    {
      "epoch": 3.49179557732962,
      "grad_norm": 0.63671875,
      "learning_rate": 1.0414899004336995e-05,
      "loss": 0.757,
      "step": 6504
    },
    {
      "epoch": 3.492332472064696,
      "grad_norm": 0.83203125,
      "learning_rate": 1.0408040844041156e-05,
      "loss": 0.936,
      "step": 6505
    },
    {
      "epoch": 3.4928693667997717,
      "grad_norm": 0.76953125,
      "learning_rate": 1.040118434887264e-05,
      "loss": 0.857,
      "step": 6506
    },
    {
      "epoch": 3.493406261534848,
      "grad_norm": 0.6796875,
      "learning_rate": 1.0394329519613865e-05,
      "loss": 0.8578,
      "step": 6507
    },
    {
      "epoch": 3.493943156269924,
      "grad_norm": 0.69921875,
      "learning_rate": 1.0387476357047072e-05,
      "loss": 0.7709,
      "step": 6508
    },
    {
      "epoch": 3.4944800510049996,
      "grad_norm": 0.72265625,
      "learning_rate": 1.0380624861954267e-05,
      "loss": 0.8426,
      "step": 6509
    },
    {
      "epoch": 3.4950169457400757,
      "grad_norm": 0.76171875,
      "learning_rate": 1.0373775035117305e-05,
      "loss": 1.0863,
      "step": 6510
    },
    {
      "epoch": 3.495553840475152,
      "grad_norm": 0.78125,
      "learning_rate": 1.0366926877317848e-05,
      "loss": 1.0062,
      "step": 6511
    },
    {
      "epoch": 3.496090735210228,
      "grad_norm": 0.71484375,
      "learning_rate": 1.0360080389337345e-05,
      "loss": 1.0122,
      "step": 6512
    },
    {
      "epoch": 3.496627629945304,
      "grad_norm": 0.68359375,
      "learning_rate": 1.0353235571957061e-05,
      "loss": 0.7241,
      "step": 6513
    },
    {
      "epoch": 3.4971645246803797,
      "grad_norm": 0.62890625,
      "learning_rate": 1.0346392425958088e-05,
      "loss": 0.7633,
      "step": 6514
    },
    {
      "epoch": 3.497701419415456,
      "grad_norm": 0.7578125,
      "learning_rate": 1.0339550952121318e-05,
      "loss": 0.9641,
      "step": 6515
    },
    {
      "epoch": 3.498238314150532,
      "grad_norm": 0.6015625,
      "learning_rate": 1.0332711151227433e-05,
      "loss": 0.7376,
      "step": 6516
    },
    {
      "epoch": 3.4987752088856077,
      "grad_norm": 0.62109375,
      "learning_rate": 1.0325873024056955e-05,
      "loss": 0.8517,
      "step": 6517
    },
    {
      "epoch": 3.499312103620684,
      "grad_norm": 0.796875,
      "learning_rate": 1.0319036571390206e-05,
      "loss": 0.7533,
      "step": 6518
    },
    {
      "epoch": 3.49984899835576,
      "grad_norm": 1.296875,
      "learning_rate": 1.0312201794007294e-05,
      "loss": 0.7301,
      "step": 6519
    },
    {
      "epoch": 3.500385893090836,
      "grad_norm": 0.66796875,
      "learning_rate": 1.0305368692688174e-05,
      "loss": 0.8539,
      "step": 6520
    },
    {
      "epoch": 3.500922787825912,
      "grad_norm": 0.765625,
      "learning_rate": 1.0298537268212577e-05,
      "loss": 0.8464,
      "step": 6521
    },
    {
      "epoch": 3.501459682560988,
      "grad_norm": 0.6875,
      "learning_rate": 1.0291707521360047e-05,
      "loss": 0.9696,
      "step": 6522
    },
    {
      "epoch": 3.501996577296064,
      "grad_norm": 0.60546875,
      "learning_rate": 1.0284879452909956e-05,
      "loss": 0.9041,
      "step": 6523
    },
    {
      "epoch": 3.50253347203114,
      "grad_norm": 0.69140625,
      "learning_rate": 1.0278053063641479e-05,
      "loss": 0.8197,
      "step": 6524
    },
    {
      "epoch": 3.5030703667662157,
      "grad_norm": 0.58203125,
      "learning_rate": 1.0271228354333578e-05,
      "loss": 0.856,
      "step": 6525
    },
    {
      "epoch": 3.503607261501292,
      "grad_norm": 0.72265625,
      "learning_rate": 1.0264405325765044e-05,
      "loss": 0.7948,
      "step": 6526
    },
    {
      "epoch": 3.504144156236368,
      "grad_norm": 0.71875,
      "learning_rate": 1.0257583978714485e-05,
      "loss": 0.7216,
      "step": 6527
    },
    {
      "epoch": 3.5046810509714437,
      "grad_norm": 0.625,
      "learning_rate": 1.0250764313960292e-05,
      "loss": 0.7046,
      "step": 6528
    },
    {
      "epoch": 3.50521794570652,
      "grad_norm": 0.73046875,
      "learning_rate": 1.0243946332280665e-05,
      "loss": 0.8455,
      "step": 6529
    },
    {
      "epoch": 3.505754840441596,
      "grad_norm": 0.6953125,
      "learning_rate": 1.023713003445364e-05,
      "loss": 0.7219,
      "step": 6530
    },
    {
      "epoch": 3.506291735176672,
      "grad_norm": 0.6484375,
      "learning_rate": 1.0230315421257027e-05,
      "loss": 0.7459,
      "step": 6531
    },
    {
      "epoch": 3.506828629911748,
      "grad_norm": 0.61328125,
      "learning_rate": 1.0223502493468468e-05,
      "loss": 0.7541,
      "step": 6532
    },
    {
      "epoch": 3.507365524646824,
      "grad_norm": 0.74609375,
      "learning_rate": 1.021669125186541e-05,
      "loss": 0.7977,
      "step": 6533
    },
    {
      "epoch": 3.5079024193819,
      "grad_norm": 0.734375,
      "learning_rate": 1.0209881697225087e-05,
      "loss": 0.806,
      "step": 6534
    },
    {
      "epoch": 3.508439314116976,
      "grad_norm": 0.7265625,
      "learning_rate": 1.0203073830324567e-05,
      "loss": 0.9033,
      "step": 6535
    },
    {
      "epoch": 3.5089762088520517,
      "grad_norm": 0.734375,
      "learning_rate": 1.0196267651940716e-05,
      "loss": 0.9589,
      "step": 6536
    },
    {
      "epoch": 3.509513103587128,
      "grad_norm": 0.71875,
      "learning_rate": 1.0189463162850202e-05,
      "loss": 0.7795,
      "step": 6537
    },
    {
      "epoch": 3.510049998322204,
      "grad_norm": 0.71484375,
      "learning_rate": 1.0182660363829495e-05,
      "loss": 0.7505,
      "step": 6538
    },
    {
      "epoch": 3.51058689305728,
      "grad_norm": 0.5703125,
      "learning_rate": 1.0175859255654882e-05,
      "loss": 0.6832,
      "step": 6539
    },
    {
      "epoch": 3.5111237877923562,
      "grad_norm": 0.75,
      "learning_rate": 1.0169059839102471e-05,
      "loss": 1.062,
      "step": 6540
    },
    {
      "epoch": 3.511660682527432,
      "grad_norm": 0.69921875,
      "learning_rate": 1.0162262114948143e-05,
      "loss": 0.7038,
      "step": 6541
    },
    {
      "epoch": 3.512197577262508,
      "grad_norm": 0.64453125,
      "learning_rate": 1.0155466083967613e-05,
      "loss": 0.7424,
      "step": 6542
    },
    {
      "epoch": 3.512734471997584,
      "grad_norm": 0.78125,
      "learning_rate": 1.01486717469364e-05,
      "loss": 0.8705,
      "step": 6543
    },
    {
      "epoch": 3.51327136673266,
      "grad_norm": 0.71484375,
      "learning_rate": 1.0141879104629812e-05,
      "loss": 0.8432,
      "step": 6544
    },
    {
      "epoch": 3.513808261467736,
      "grad_norm": 0.63671875,
      "learning_rate": 1.0135088157822986e-05,
      "loss": 0.698,
      "step": 6545
    },
    {
      "epoch": 3.514345156202812,
      "grad_norm": 0.5859375,
      "learning_rate": 1.0128298907290853e-05,
      "loss": 0.7488,
      "step": 6546
    },
    {
      "epoch": 3.5148820509378877,
      "grad_norm": 0.72265625,
      "learning_rate": 1.0121511353808142e-05,
      "loss": 0.7864,
      "step": 6547
    },
    {
      "epoch": 3.515418945672964,
      "grad_norm": 0.734375,
      "learning_rate": 1.0114725498149402e-05,
      "loss": 0.8029,
      "step": 6548
    },
    {
      "epoch": 3.51595584040804,
      "grad_norm": 0.75390625,
      "learning_rate": 1.0107941341089003e-05,
      "loss": 0.9244,
      "step": 6549
    },
    {
      "epoch": 3.516492735143116,
      "grad_norm": 0.625,
      "learning_rate": 1.0101158883401077e-05,
      "loss": 0.7203,
      "step": 6550
    },
    {
      "epoch": 3.516492735143116,
      "eval_loss": 0.9030448794364929,
      "eval_runtime": 45.747,
      "eval_samples_per_second": 21.029,
      "eval_steps_per_second": 21.029,
      "step": 6550
    },
    {
      "epoch": 3.5170296298781922,
      "grad_norm": 0.671875,
      "learning_rate": 1.0094378125859602e-05,
      "loss": 0.7902,
      "step": 6551
    },
    {
      "epoch": 3.517566524613268,
      "grad_norm": 0.8046875,
      "learning_rate": 1.0087599069238357e-05,
      "loss": 0.9235,
      "step": 6552
    },
    {
      "epoch": 3.518103419348344,
      "grad_norm": 0.66796875,
      "learning_rate": 1.0080821714310908e-05,
      "loss": 0.8724,
      "step": 6553
    },
    {
      "epoch": 3.51864031408342,
      "grad_norm": 0.671875,
      "learning_rate": 1.0074046061850628e-05,
      "loss": 1.1629,
      "step": 6554
    },
    {
      "epoch": 3.519177208818496,
      "grad_norm": 0.76171875,
      "learning_rate": 1.0067272112630716e-05,
      "loss": 0.9107,
      "step": 6555
    },
    {
      "epoch": 3.519714103553572,
      "grad_norm": 0.640625,
      "learning_rate": 1.006049986742417e-05,
      "loss": 0.7408,
      "step": 6556
    },
    {
      "epoch": 3.520250998288648,
      "grad_norm": 0.71484375,
      "learning_rate": 1.0053729327003775e-05,
      "loss": 0.7296,
      "step": 6557
    },
    {
      "epoch": 3.520787893023724,
      "grad_norm": 0.65234375,
      "learning_rate": 1.0046960492142144e-05,
      "loss": 0.7931,
      "step": 6558
    },
    {
      "epoch": 3.5213247877588003,
      "grad_norm": 0.609375,
      "learning_rate": 1.0040193363611695e-05,
      "loss": 0.7758,
      "step": 6559
    },
    {
      "epoch": 3.521861682493876,
      "grad_norm": 0.67578125,
      "learning_rate": 1.0033427942184622e-05,
      "loss": 0.7793,
      "step": 6560
    },
    {
      "epoch": 3.522398577228952,
      "grad_norm": 0.83203125,
      "learning_rate": 1.0026664228632968e-05,
      "loss": 0.8087,
      "step": 6561
    },
    {
      "epoch": 3.5229354719640282,
      "grad_norm": 0.64453125,
      "learning_rate": 1.0019902223728548e-05,
      "loss": 0.7792,
      "step": 6562
    },
    {
      "epoch": 3.523472366699104,
      "grad_norm": 0.7265625,
      "learning_rate": 1.0013141928242981e-05,
      "loss": 0.7958,
      "step": 6563
    },
    {
      "epoch": 3.52400926143418,
      "grad_norm": 0.6328125,
      "learning_rate": 1.0006383342947714e-05,
      "loss": 0.7844,
      "step": 6564
    },
    {
      "epoch": 3.524546156169256,
      "grad_norm": 0.69921875,
      "learning_rate": 9.999626468613996e-06,
      "loss": 0.883,
      "step": 6565
    },
    {
      "epoch": 3.525083050904332,
      "grad_norm": 0.859375,
      "learning_rate": 9.992871306012852e-06,
      "loss": 1.0475,
      "step": 6566
    },
    {
      "epoch": 3.525619945639408,
      "grad_norm": 0.73828125,
      "learning_rate": 9.986117855915145e-06,
      "loss": 0.8462,
      "step": 6567
    },
    {
      "epoch": 3.526156840374484,
      "grad_norm": 0.78125,
      "learning_rate": 9.979366119091531e-06,
      "loss": 0.8852,
      "step": 6568
    },
    {
      "epoch": 3.52669373510956,
      "grad_norm": 0.62109375,
      "learning_rate": 9.972616096312456e-06,
      "loss": 0.7959,
      "step": 6569
    },
    {
      "epoch": 3.5272306298446363,
      "grad_norm": 0.60546875,
      "learning_rate": 9.965867788348201e-06,
      "loss": 0.9117,
      "step": 6570
    },
    {
      "epoch": 3.527767524579712,
      "grad_norm": 0.70703125,
      "learning_rate": 9.959121195968823e-06,
      "loss": 0.8907,
      "step": 6571
    },
    {
      "epoch": 3.528304419314788,
      "grad_norm": 0.6328125,
      "learning_rate": 9.952376319944184e-06,
      "loss": 0.7078,
      "step": 6572
    },
    {
      "epoch": 3.5288413140498642,
      "grad_norm": 0.73046875,
      "learning_rate": 9.94563316104397e-06,
      "loss": 0.9385,
      "step": 6573
    },
    {
      "epoch": 3.52937820878494,
      "grad_norm": 1.0390625,
      "learning_rate": 9.938891720037671e-06,
      "loss": 0.9329,
      "step": 6574
    },
    {
      "epoch": 3.529915103520016,
      "grad_norm": 0.640625,
      "learning_rate": 9.932151997694549e-06,
      "loss": 0.7381,
      "step": 6575
    },
    {
      "epoch": 3.530451998255092,
      "grad_norm": 0.63671875,
      "learning_rate": 9.925413994783703e-06,
      "loss": 0.8606,
      "step": 6576
    },
    {
      "epoch": 3.5309888929901683,
      "grad_norm": 0.6953125,
      "learning_rate": 9.918677712074035e-06,
      "loss": 0.9541,
      "step": 6577
    },
    {
      "epoch": 3.5315257877252444,
      "grad_norm": 0.671875,
      "learning_rate": 9.91194315033423e-06,
      "loss": 0.7419,
      "step": 6578
    },
    {
      "epoch": 3.53206268246032,
      "grad_norm": 0.671875,
      "learning_rate": 9.905210310332774e-06,
      "loss": 0.7282,
      "step": 6579
    },
    {
      "epoch": 3.532599577195396,
      "grad_norm": 0.60546875,
      "learning_rate": 9.898479192837983e-06,
      "loss": 0.8543,
      "step": 6580
    },
    {
      "epoch": 3.5331364719304723,
      "grad_norm": 0.72265625,
      "learning_rate": 9.891749798617972e-06,
      "loss": 0.8682,
      "step": 6581
    },
    {
      "epoch": 3.533673366665548,
      "grad_norm": 0.67578125,
      "learning_rate": 9.88502212844063e-06,
      "loss": 0.9144,
      "step": 6582
    },
    {
      "epoch": 3.534210261400624,
      "grad_norm": 0.72265625,
      "learning_rate": 9.878296183073678e-06,
      "loss": 0.8391,
      "step": 6583
    },
    {
      "epoch": 3.5347471561357002,
      "grad_norm": 0.66015625,
      "learning_rate": 9.871571963284643e-06,
      "loss": 0.7431,
      "step": 6584
    },
    {
      "epoch": 3.535284050870776,
      "grad_norm": 0.6171875,
      "learning_rate": 9.864849469840822e-06,
      "loss": 0.7317,
      "step": 6585
    },
    {
      "epoch": 3.535820945605852,
      "grad_norm": 0.625,
      "learning_rate": 9.858128703509348e-06,
      "loss": 0.7143,
      "step": 6586
    },
    {
      "epoch": 3.536357840340928,
      "grad_norm": 0.6484375,
      "learning_rate": 9.851409665057165e-06,
      "loss": 0.7855,
      "step": 6587
    },
    {
      "epoch": 3.5368947350760043,
      "grad_norm": 0.6875,
      "learning_rate": 9.844692355250962e-06,
      "loss": 1.0114,
      "step": 6588
    },
    {
      "epoch": 3.5374316298110804,
      "grad_norm": 0.671875,
      "learning_rate": 9.837976774857286e-06,
      "loss": 0.9623,
      "step": 6589
    },
    {
      "epoch": 3.537968524546156,
      "grad_norm": 0.72265625,
      "learning_rate": 9.831262924642481e-06,
      "loss": 0.9195,
      "step": 6590
    },
    {
      "epoch": 3.538505419281232,
      "grad_norm": 0.83203125,
      "learning_rate": 9.824550805372664e-06,
      "loss": 0.982,
      "step": 6591
    },
    {
      "epoch": 3.5390423140163083,
      "grad_norm": 0.72265625,
      "learning_rate": 9.817840417813784e-06,
      "loss": 0.8192,
      "step": 6592
    },
    {
      "epoch": 3.539579208751384,
      "grad_norm": 0.68359375,
      "learning_rate": 9.811131762731585e-06,
      "loss": 0.7611,
      "step": 6593
    },
    {
      "epoch": 3.54011610348646,
      "grad_norm": 0.6171875,
      "learning_rate": 9.804424840891594e-06,
      "loss": 0.821,
      "step": 6594
    },
    {
      "epoch": 3.5406529982215362,
      "grad_norm": 0.66015625,
      "learning_rate": 9.797719653059177e-06,
      "loss": 0.7946,
      "step": 6595
    },
    {
      "epoch": 3.5411898929566124,
      "grad_norm": 0.83203125,
      "learning_rate": 9.791016199999456e-06,
      "loss": 0.8176,
      "step": 6596
    },
    {
      "epoch": 3.5417267876916885,
      "grad_norm": 0.6875,
      "learning_rate": 9.784314482477406e-06,
      "loss": 0.7749,
      "step": 6597
    },
    {
      "epoch": 3.542263682426764,
      "grad_norm": 0.80078125,
      "learning_rate": 9.777614501257751e-06,
      "loss": 1.0866,
      "step": 6598
    },
    {
      "epoch": 3.5428005771618403,
      "grad_norm": 0.81640625,
      "learning_rate": 9.770916257105062e-06,
      "loss": 1.0468,
      "step": 6599
    },
    {
      "epoch": 3.5433374718969164,
      "grad_norm": 0.765625,
      "learning_rate": 9.764219750783693e-06,
      "loss": 0.8212,
      "step": 6600
    },
    {
      "epoch": 3.5433374718969164,
      "eval_loss": 0.9029698967933655,
      "eval_runtime": 45.9736,
      "eval_samples_per_second": 20.925,
      "eval_steps_per_second": 20.925,
      "step": 6600
    },
    {
      "epoch": 3.543874366631992,
      "grad_norm": 0.6640625,
      "learning_rate": 9.757524983057789e-06,
      "loss": 0.7546,
      "step": 6601
    },
    {
      "epoch": 3.544411261367068,
      "grad_norm": 0.6484375,
      "learning_rate": 9.750831954691326e-06,
      "loss": 0.9804,
      "step": 6602
    },
    {
      "epoch": 3.5449481561021443,
      "grad_norm": 0.8359375,
      "learning_rate": 9.74414066644804e-06,
      "loss": 0.882,
      "step": 6603
    },
    {
      "epoch": 3.54548505083722,
      "grad_norm": 0.75390625,
      "learning_rate": 9.737451119091515e-06,
      "loss": 0.7966,
      "step": 6604
    },
    {
      "epoch": 3.546021945572296,
      "grad_norm": 0.61328125,
      "learning_rate": 9.730763313385091e-06,
      "loss": 0.8061,
      "step": 6605
    },
    {
      "epoch": 3.5465588403073722,
      "grad_norm": 0.68359375,
      "learning_rate": 9.724077250091954e-06,
      "loss": 0.8407,
      "step": 6606
    },
    {
      "epoch": 3.5470957350424484,
      "grad_norm": 0.5859375,
      "learning_rate": 9.717392929975045e-06,
      "loss": 0.9106,
      "step": 6607
    },
    {
      "epoch": 3.5476326297775245,
      "grad_norm": 0.7890625,
      "learning_rate": 9.710710353797142e-06,
      "loss": 0.9718,
      "step": 6608
    },
    {
      "epoch": 3.5481695245126,
      "grad_norm": 0.7734375,
      "learning_rate": 9.70402952232082e-06,
      "loss": 0.8036,
      "step": 6609
    },
    {
      "epoch": 3.5487064192476763,
      "grad_norm": 0.8125,
      "learning_rate": 9.697350436308427e-06,
      "loss": 0.8215,
      "step": 6610
    },
    {
      "epoch": 3.5492433139827524,
      "grad_norm": 0.7578125,
      "learning_rate": 9.690673096522143e-06,
      "loss": 0.8967,
      "step": 6611
    },
    {
      "epoch": 3.549780208717828,
      "grad_norm": 0.63671875,
      "learning_rate": 9.683997503723954e-06,
      "loss": 0.857,
      "step": 6612
    },
    {
      "epoch": 3.550317103452904,
      "grad_norm": 0.6484375,
      "learning_rate": 9.677323658675594e-06,
      "loss": 0.7265,
      "step": 6613
    },
    {
      "epoch": 3.5508539981879803,
      "grad_norm": 0.63671875,
      "learning_rate": 9.67065156213865e-06,
      "loss": 0.6556,
      "step": 6614
    },
    {
      "epoch": 3.5513908929230564,
      "grad_norm": 0.76953125,
      "learning_rate": 9.6639812148745e-06,
      "loss": 0.9328,
      "step": 6615
    },
    {
      "epoch": 3.5519277876581326,
      "grad_norm": 0.6484375,
      "learning_rate": 9.657312617644302e-06,
      "loss": 0.819,
      "step": 6616
    },
    {
      "epoch": 3.5524646823932082,
      "grad_norm": 0.828125,
      "learning_rate": 9.650645771209033e-06,
      "loss": 0.807,
      "step": 6617
    },
    {
      "epoch": 3.5530015771282843,
      "grad_norm": 0.76171875,
      "learning_rate": 9.643980676329472e-06,
      "loss": 0.8549,
      "step": 6618
    },
    {
      "epoch": 3.5535384718633605,
      "grad_norm": 0.71484375,
      "learning_rate": 9.637317333766177e-06,
      "loss": 0.9689,
      "step": 6619
    },
    {
      "epoch": 3.554075366598436,
      "grad_norm": 0.75390625,
      "learning_rate": 9.630655744279537e-06,
      "loss": 0.7918,
      "step": 6620
    },
    {
      "epoch": 3.5546122613335123,
      "grad_norm": 0.96484375,
      "learning_rate": 9.623995908629704e-06,
      "loss": 1.0044,
      "step": 6621
    },
    {
      "epoch": 3.5551491560685884,
      "grad_norm": 0.96484375,
      "learning_rate": 9.617337827576666e-06,
      "loss": 0.9225,
      "step": 6622
    },
    {
      "epoch": 3.555686050803664,
      "grad_norm": 0.7109375,
      "learning_rate": 9.610681501880181e-06,
      "loss": 0.8674,
      "step": 6623
    },
    {
      "epoch": 3.55622294553874,
      "grad_norm": 0.6640625,
      "learning_rate": 9.604026932299825e-06,
      "loss": 0.7579,
      "step": 6624
    },
    {
      "epoch": 3.5567598402738163,
      "grad_norm": 0.64453125,
      "learning_rate": 9.597374119594982e-06,
      "loss": 0.6607,
      "step": 6625
    },
    {
      "epoch": 3.5572967350088924,
      "grad_norm": 0.70703125,
      "learning_rate": 9.590723064524799e-06,
      "loss": 0.7935,
      "step": 6626
    },
    {
      "epoch": 3.5578336297439686,
      "grad_norm": 0.65234375,
      "learning_rate": 9.584073767848257e-06,
      "loss": 0.8069,
      "step": 6627
    },
    {
      "epoch": 3.5583705244790442,
      "grad_norm": 0.61328125,
      "learning_rate": 9.577426230324135e-06,
      "loss": 0.7998,
      "step": 6628
    },
    {
      "epoch": 3.5589074192141203,
      "grad_norm": 0.63671875,
      "learning_rate": 9.570780452710989e-06,
      "loss": 0.642,
      "step": 6629
    },
    {
      "epoch": 3.5594443139491965,
      "grad_norm": 0.734375,
      "learning_rate": 9.564136435767181e-06,
      "loss": 0.9416,
      "step": 6630
    },
    {
      "epoch": 3.559981208684272,
      "grad_norm": 0.73828125,
      "learning_rate": 9.557494180250895e-06,
      "loss": 0.7998,
      "step": 6631
    },
    {
      "epoch": 3.5605181034193483,
      "grad_norm": 0.6796875,
      "learning_rate": 9.550853686920078e-06,
      "loss": 0.8869,
      "step": 6632
    },
    {
      "epoch": 3.5610549981544244,
      "grad_norm": 0.59765625,
      "learning_rate": 9.544214956532502e-06,
      "loss": 0.6988,
      "step": 6633
    },
    {
      "epoch": 3.5615918928895005,
      "grad_norm": 0.609375,
      "learning_rate": 9.537577989845738e-06,
      "loss": 0.719,
      "step": 6634
    },
    {
      "epoch": 3.5621287876245766,
      "grad_norm": 0.58984375,
      "learning_rate": 9.530942787617137e-06,
      "loss": 0.867,
      "step": 6635
    },
    {
      "epoch": 3.5626656823596523,
      "grad_norm": 0.7421875,
      "learning_rate": 9.52430935060386e-06,
      "loss": 0.7868,
      "step": 6636
    },
    {
      "epoch": 3.5632025770947284,
      "grad_norm": 0.65625,
      "learning_rate": 9.517677679562878e-06,
      "loss": 0.7565,
      "step": 6637
    },
    {
      "epoch": 3.5637394718298046,
      "grad_norm": 0.73046875,
      "learning_rate": 9.511047775250941e-06,
      "loss": 0.8856,
      "step": 6638
    },
    {
      "epoch": 3.5642763665648802,
      "grad_norm": 0.71875,
      "learning_rate": 9.5044196384246e-06,
      "loss": 0.8864,
      "step": 6639
    },
    {
      "epoch": 3.5648132612999563,
      "grad_norm": 0.625,
      "learning_rate": 9.497793269840211e-06,
      "loss": 0.6957,
      "step": 6640
    },
    {
      "epoch": 3.5653501560350325,
      "grad_norm": 0.62109375,
      "learning_rate": 9.491168670253936e-06,
      "loss": 0.7793,
      "step": 6641
    },
    {
      "epoch": 3.565887050770108,
      "grad_norm": 0.95703125,
      "learning_rate": 9.484545840421713e-06,
      "loss": 0.947,
      "step": 6642
    },
    {
      "epoch": 3.5664239455051843,
      "grad_norm": 0.65234375,
      "learning_rate": 9.477924781099296e-06,
      "loss": 0.7158,
      "step": 6643
    },
    {
      "epoch": 3.5669608402402604,
      "grad_norm": 0.6796875,
      "learning_rate": 9.471305493042243e-06,
      "loss": 0.805,
      "step": 6644
    },
    {
      "epoch": 3.5674977349753365,
      "grad_norm": 0.69921875,
      "learning_rate": 9.464687977005876e-06,
      "loss": 0.9085,
      "step": 6645
    },
    {
      "epoch": 3.5680346297104126,
      "grad_norm": 0.5703125,
      "learning_rate": 9.458072233745362e-06,
      "loss": 0.8781,
      "step": 6646
    },
    {
      "epoch": 3.5685715244454883,
      "grad_norm": 0.7578125,
      "learning_rate": 9.451458264015627e-06,
      "loss": 0.999,
      "step": 6647
    },
    {
      "epoch": 3.5691084191805644,
      "grad_norm": 0.66015625,
      "learning_rate": 9.444846068571404e-06,
      "loss": 1.0313,
      "step": 6648
    },
    {
      "epoch": 3.5696453139156405,
      "grad_norm": 0.84375,
      "learning_rate": 9.438235648167233e-06,
      "loss": 0.7287,
      "step": 6649
    },
    {
      "epoch": 3.5701822086507162,
      "grad_norm": 0.55859375,
      "learning_rate": 9.431627003557456e-06,
      "loss": 0.7335,
      "step": 6650
    },
    {
      "epoch": 3.5701822086507162,
      "eval_loss": 0.902814507484436,
      "eval_runtime": 46.0505,
      "eval_samples_per_second": 20.89,
      "eval_steps_per_second": 20.89,
      "step": 6650
    },
    {
      "epoch": 3.5707191033857923,
      "grad_norm": 0.76953125,
      "learning_rate": 9.425020135496185e-06,
      "loss": 0.9489,
      "step": 6651
    },
    {
      "epoch": 3.5712559981208685,
      "grad_norm": 0.703125,
      "learning_rate": 9.41841504473736e-06,
      "loss": 0.8986,
      "step": 6652
    },
    {
      "epoch": 3.5717928928559446,
      "grad_norm": 0.71484375,
      "learning_rate": 9.411811732034715e-06,
      "loss": 0.7683,
      "step": 6653
    },
    {
      "epoch": 3.5723297875910207,
      "grad_norm": 0.7578125,
      "learning_rate": 9.405210198141757e-06,
      "loss": 0.889,
      "step": 6654
    },
    {
      "epoch": 3.5728666823260964,
      "grad_norm": 1.0859375,
      "learning_rate": 9.398610443811797e-06,
      "loss": 0.7695,
      "step": 6655
    },
    {
      "epoch": 3.5734035770611725,
      "grad_norm": 0.70703125,
      "learning_rate": 9.39201246979797e-06,
      "loss": 0.8382,
      "step": 6656
    },
    {
      "epoch": 3.5739404717962486,
      "grad_norm": 0.8515625,
      "learning_rate": 9.385416276853173e-06,
      "loss": 0.8171,
      "step": 6657
    },
    {
      "epoch": 3.5744773665313243,
      "grad_norm": 0.765625,
      "learning_rate": 9.37882186573012e-06,
      "loss": 0.7891,
      "step": 6658
    },
    {
      "epoch": 3.5750142612664004,
      "grad_norm": 0.6796875,
      "learning_rate": 9.372229237181327e-06,
      "loss": 0.8629,
      "step": 6659
    },
    {
      "epoch": 3.5755511560014765,
      "grad_norm": 0.58203125,
      "learning_rate": 9.36563839195908e-06,
      "loss": 0.7094,
      "step": 6660
    },
    {
      "epoch": 3.5760880507365522,
      "grad_norm": 0.79296875,
      "learning_rate": 9.359049330815486e-06,
      "loss": 0.8081,
      "step": 6661
    },
    {
      "epoch": 3.5766249454716283,
      "grad_norm": 0.56640625,
      "learning_rate": 9.352462054502449e-06,
      "loss": 0.6752,
      "step": 6662
    },
    {
      "epoch": 3.5771618402067045,
      "grad_norm": 0.58203125,
      "learning_rate": 9.345876563771649e-06,
      "loss": 0.6144,
      "step": 6663
    },
    {
      "epoch": 3.5776987349417806,
      "grad_norm": 0.8671875,
      "learning_rate": 9.339292859374568e-06,
      "loss": 0.8694,
      "step": 6664
    },
    {
      "epoch": 3.5782356296768567,
      "grad_norm": 0.8359375,
      "learning_rate": 9.3327109420625e-06,
      "loss": 0.944,
      "step": 6665
    },
    {
      "epoch": 3.5787725244119324,
      "grad_norm": 0.72265625,
      "learning_rate": 9.326130812586534e-06,
      "loss": 0.9569,
      "step": 6666
    },
    {
      "epoch": 3.5793094191470085,
      "grad_norm": 0.75,
      "learning_rate": 9.319552471697524e-06,
      "loss": 1.0054,
      "step": 6667
    },
    {
      "epoch": 3.5798463138820846,
      "grad_norm": 0.68359375,
      "learning_rate": 9.312975920146155e-06,
      "loss": 0.749,
      "step": 6668
    },
    {
      "epoch": 3.5803832086171603,
      "grad_norm": 0.66015625,
      "learning_rate": 9.306401158682904e-06,
      "loss": 0.7842,
      "step": 6669
    },
    {
      "epoch": 3.5809201033522364,
      "grad_norm": 0.72265625,
      "learning_rate": 9.299828188058013e-06,
      "loss": 0.917,
      "step": 6670
    },
    {
      "epoch": 3.5814569980873125,
      "grad_norm": 0.63671875,
      "learning_rate": 9.29325700902156e-06,
      "loss": 0.7266,
      "step": 6671
    },
    {
      "epoch": 3.5819938928223887,
      "grad_norm": 0.703125,
      "learning_rate": 9.286687622323395e-06,
      "loss": 0.8447,
      "step": 6672
    },
    {
      "epoch": 3.582530787557465,
      "grad_norm": 0.64453125,
      "learning_rate": 9.280120028713154e-06,
      "loss": 0.9104,
      "step": 6673
    },
    {
      "epoch": 3.5830676822925405,
      "grad_norm": 0.703125,
      "learning_rate": 9.273554228940298e-06,
      "loss": 0.9141,
      "step": 6674
    },
    {
      "epoch": 3.5836045770276166,
      "grad_norm": 0.80078125,
      "learning_rate": 9.266990223754069e-06,
      "loss": 0.8476,
      "step": 6675
    },
    {
      "epoch": 3.5841414717626927,
      "grad_norm": 0.71875,
      "learning_rate": 9.26042801390349e-06,
      "loss": 0.8628,
      "step": 6676
    },
    {
      "epoch": 3.5846783664977684,
      "grad_norm": 0.5859375,
      "learning_rate": 9.2538676001374e-06,
      "loss": 0.7839,
      "step": 6677
    },
    {
      "epoch": 3.5852152612328445,
      "grad_norm": 0.70703125,
      "learning_rate": 9.247308983204433e-06,
      "loss": 0.8638,
      "step": 6678
    },
    {
      "epoch": 3.5857521559679206,
      "grad_norm": 0.66796875,
      "learning_rate": 9.240752163853006e-06,
      "loss": 0.7267,
      "step": 6679
    },
    {
      "epoch": 3.5862890507029963,
      "grad_norm": 0.66796875,
      "learning_rate": 9.234197142831321e-06,
      "loss": 0.7609,
      "step": 6680
    },
    {
      "epoch": 3.5868259454380724,
      "grad_norm": 0.8359375,
      "learning_rate": 9.227643920887399e-06,
      "loss": 0.6968,
      "step": 6681
    },
    {
      "epoch": 3.5873628401731485,
      "grad_norm": 0.7109375,
      "learning_rate": 9.221092498769058e-06,
      "loss": 0.7684,
      "step": 6682
    },
    {
      "epoch": 3.5878997349082247,
      "grad_norm": 0.75,
      "learning_rate": 9.214542877223876e-06,
      "loss": 0.959,
      "step": 6683
    },
    {
      "epoch": 3.588436629643301,
      "grad_norm": 0.76953125,
      "learning_rate": 9.207995056999258e-06,
      "loss": 0.8972,
      "step": 6684
    },
    {
      "epoch": 3.5889735243783765,
      "grad_norm": 0.63671875,
      "learning_rate": 9.201449038842402e-06,
      "loss": 0.7983,
      "step": 6685
    },
    {
      "epoch": 3.5895104191134526,
      "grad_norm": 0.58984375,
      "learning_rate": 9.194904823500275e-06,
      "loss": 0.7311,
      "step": 6686
    },
    {
      "epoch": 3.5900473138485287,
      "grad_norm": 0.625,
      "learning_rate": 9.188362411719672e-06,
      "loss": 0.7303,
      "step": 6687
    },
    {
      "epoch": 3.5905842085836044,
      "grad_norm": 0.7109375,
      "learning_rate": 9.181821804247156e-06,
      "loss": 0.8449,
      "step": 6688
    },
    {
      "epoch": 3.5911211033186805,
      "grad_norm": 0.64453125,
      "learning_rate": 9.175283001829083e-06,
      "loss": 0.6908,
      "step": 6689
    },
    {
      "epoch": 3.5916579980537566,
      "grad_norm": 0.6484375,
      "learning_rate": 9.168746005211626e-06,
      "loss": 0.7831,
      "step": 6690
    },
    {
      "epoch": 3.5921948927888327,
      "grad_norm": 0.75,
      "learning_rate": 9.162210815140743e-06,
      "loss": 0.9573,
      "step": 6691
    },
    {
      "epoch": 3.592731787523909,
      "grad_norm": 0.73046875,
      "learning_rate": 9.155677432362167e-06,
      "loss": 0.861,
      "step": 6692
    },
    {
      "epoch": 3.5932686822589845,
      "grad_norm": 0.71875,
      "learning_rate": 9.149145857621447e-06,
      "loss": 0.7768,
      "step": 6693
    },
    {
      "epoch": 3.5938055769940607,
      "grad_norm": 0.83203125,
      "learning_rate": 9.14261609166393e-06,
      "loss": 1.0587,
      "step": 6694
    },
    {
      "epoch": 3.594342471729137,
      "grad_norm": 0.69921875,
      "learning_rate": 9.136088135234727e-06,
      "loss": 0.8209,
      "step": 6695
    },
    {
      "epoch": 3.5948793664642125,
      "grad_norm": 0.671875,
      "learning_rate": 9.129561989078775e-06,
      "loss": 0.9221,
      "step": 6696
    },
    {
      "epoch": 3.5954162611992886,
      "grad_norm": 0.7109375,
      "learning_rate": 9.12303765394078e-06,
      "loss": 0.7608,
      "step": 6697
    },
    {
      "epoch": 3.5959531559343647,
      "grad_norm": 0.74609375,
      "learning_rate": 9.11651513056525e-06,
      "loss": 0.8028,
      "step": 6698
    },
    {
      "epoch": 3.5964900506694404,
      "grad_norm": 0.69140625,
      "learning_rate": 9.109994419696492e-06,
      "loss": 0.9619,
      "step": 6699
    },
    {
      "epoch": 3.5970269454045165,
      "grad_norm": 0.6328125,
      "learning_rate": 9.103475522078605e-06,
      "loss": 0.7196,
      "step": 6700
    },
    {
      "epoch": 3.5970269454045165,
      "eval_loss": 0.9028509259223938,
      "eval_runtime": 45.7648,
      "eval_samples_per_second": 21.021,
      "eval_steps_per_second": 21.021,
      "step": 6700
    },
    {
      "epoch": 3.5975638401395926,
      "grad_norm": 0.796875,
      "learning_rate": 9.09695843845547e-06,
      "loss": 0.8642,
      "step": 6701
    },
    {
      "epoch": 3.5981007348746687,
      "grad_norm": 0.7421875,
      "learning_rate": 9.09044316957077e-06,
      "loss": 0.8311,
      "step": 6702
    },
    {
      "epoch": 3.598637629609745,
      "grad_norm": 0.640625,
      "learning_rate": 9.083929716167994e-06,
      "loss": 0.9032,
      "step": 6703
    },
    {
      "epoch": 3.5991745243448205,
      "grad_norm": 0.6875,
      "learning_rate": 9.077418078990396e-06,
      "loss": 0.994,
      "step": 6704
    },
    {
      "epoch": 3.5997114190798967,
      "grad_norm": 0.62109375,
      "learning_rate": 9.07090825878103e-06,
      "loss": 0.9162,
      "step": 6705
    },
    {
      "epoch": 3.600248313814973,
      "grad_norm": 0.7109375,
      "learning_rate": 9.064400256282757e-06,
      "loss": 0.7996,
      "step": 6706
    },
    {
      "epoch": 3.6007852085500485,
      "grad_norm": 0.640625,
      "learning_rate": 9.057894072238233e-06,
      "loss": 0.9468,
      "step": 6707
    },
    {
      "epoch": 3.6013221032851246,
      "grad_norm": 0.8203125,
      "learning_rate": 9.051389707389877e-06,
      "loss": 0.7284,
      "step": 6708
    },
    {
      "epoch": 3.6018589980202007,
      "grad_norm": 0.6875,
      "learning_rate": 9.044887162479927e-06,
      "loss": 0.8251,
      "step": 6709
    },
    {
      "epoch": 3.602395892755277,
      "grad_norm": 0.7578125,
      "learning_rate": 9.038386438250415e-06,
      "loss": 0.7943,
      "step": 6710
    },
    {
      "epoch": 3.6029327874903525,
      "grad_norm": 0.671875,
      "learning_rate": 9.03188753544314e-06,
      "loss": 0.7361,
      "step": 6711
    },
    {
      "epoch": 3.6034696822254286,
      "grad_norm": 0.54296875,
      "learning_rate": 9.025390454799717e-06,
      "loss": 0.6254,
      "step": 6712
    },
    {
      "epoch": 3.6040065769605047,
      "grad_norm": 0.66015625,
      "learning_rate": 9.018895197061564e-06,
      "loss": 0.8518,
      "step": 6713
    },
    {
      "epoch": 3.604543471695581,
      "grad_norm": 0.609375,
      "learning_rate": 9.012401762969835e-06,
      "loss": 0.9286,
      "step": 6714
    },
    {
      "epoch": 3.6050803664306565,
      "grad_norm": 1.0390625,
      "learning_rate": 9.005910153265532e-06,
      "loss": 0.9985,
      "step": 6715
    },
    {
      "epoch": 3.6056172611657327,
      "grad_norm": 0.609375,
      "learning_rate": 8.999420368689434e-06,
      "loss": 0.8161,
      "step": 6716
    },
    {
      "epoch": 3.606154155900809,
      "grad_norm": 0.76171875,
      "learning_rate": 8.9929324099821e-06,
      "loss": 0.8656,
      "step": 6717
    },
    {
      "epoch": 3.6066910506358845,
      "grad_norm": 0.609375,
      "learning_rate": 8.986446277883887e-06,
      "loss": 0.831,
      "step": 6718
    },
    {
      "epoch": 3.6072279453709606,
      "grad_norm": 0.69140625,
      "learning_rate": 8.979961973134959e-06,
      "loss": 0.8525,
      "step": 6719
    },
    {
      "epoch": 3.6077648401060367,
      "grad_norm": 1.140625,
      "learning_rate": 8.97347949647524e-06,
      "loss": 0.9703,
      "step": 6720
    },
    {
      "epoch": 3.608301734841113,
      "grad_norm": 0.69140625,
      "learning_rate": 8.966998848644475e-06,
      "loss": 0.773,
      "step": 6721
    },
    {
      "epoch": 3.608838629576189,
      "grad_norm": 0.6171875,
      "learning_rate": 8.960520030382177e-06,
      "loss": 0.8466,
      "step": 6722
    },
    {
      "epoch": 3.6093755243112646,
      "grad_norm": 0.78125,
      "learning_rate": 8.954043042427676e-06,
      "loss": 0.7705,
      "step": 6723
    },
    {
      "epoch": 3.6099124190463407,
      "grad_norm": 0.73828125,
      "learning_rate": 8.947567885520061e-06,
      "loss": 0.8059,
      "step": 6724
    },
    {
      "epoch": 3.610449313781417,
      "grad_norm": 0.65625,
      "learning_rate": 8.941094560398236e-06,
      "loss": 0.8899,
      "step": 6725
    },
    {
      "epoch": 3.6109862085164925,
      "grad_norm": 0.68359375,
      "learning_rate": 8.934623067800902e-06,
      "loss": 0.7614,
      "step": 6726
    },
    {
      "epoch": 3.6115231032515687,
      "grad_norm": 0.58984375,
      "learning_rate": 8.928153408466516e-06,
      "loss": 0.7743,
      "step": 6727
    },
    {
      "epoch": 3.612059997986645,
      "grad_norm": 0.734375,
      "learning_rate": 8.921685583133372e-06,
      "loss": 0.8473,
      "step": 6728
    },
    {
      "epoch": 3.612596892721721,
      "grad_norm": 0.56640625,
      "learning_rate": 8.915219592539514e-06,
      "loss": 0.7977,
      "step": 6729
    },
    {
      "epoch": 3.6131337874567966,
      "grad_norm": 0.7109375,
      "learning_rate": 8.908755437422792e-06,
      "loss": 0.9503,
      "step": 6730
    },
    {
      "epoch": 3.6136706821918727,
      "grad_norm": 0.91015625,
      "learning_rate": 8.902293118520852e-06,
      "loss": 0.8522,
      "step": 6731
    },
    {
      "epoch": 3.614207576926949,
      "grad_norm": 0.60546875,
      "learning_rate": 8.895832636571138e-06,
      "loss": 0.7979,
      "step": 6732
    },
    {
      "epoch": 3.614744471662025,
      "grad_norm": 0.70703125,
      "learning_rate": 8.889373992310854e-06,
      "loss": 0.8807,
      "step": 6733
    },
    {
      "epoch": 3.6152813663971006,
      "grad_norm": 0.73046875,
      "learning_rate": 8.882917186477021e-06,
      "loss": 0.7953,
      "step": 6734
    },
    {
      "epoch": 3.6158182611321767,
      "grad_norm": 0.71875,
      "learning_rate": 8.876462219806456e-06,
      "loss": 0.7354,
      "step": 6735
    },
    {
      "epoch": 3.616355155867253,
      "grad_norm": 0.65234375,
      "learning_rate": 8.870009093035728e-06,
      "loss": 0.8158,
      "step": 6736
    },
    {
      "epoch": 3.6168920506023285,
      "grad_norm": 0.78515625,
      "learning_rate": 8.863557806901233e-06,
      "loss": 0.8887,
      "step": 6737
    },
    {
      "epoch": 3.6174289453374047,
      "grad_norm": 0.69921875,
      "learning_rate": 8.85710836213916e-06,
      "loss": 0.7492,
      "step": 6738
    },
    {
      "epoch": 3.617965840072481,
      "grad_norm": 0.78515625,
      "learning_rate": 8.850660759485438e-06,
      "loss": 0.8628,
      "step": 6739
    },
    {
      "epoch": 3.618502734807557,
      "grad_norm": 0.6328125,
      "learning_rate": 8.84421499967584e-06,
      "loss": 0.7729,
      "step": 6740
    },
    {
      "epoch": 3.619039629542633,
      "grad_norm": 0.7109375,
      "learning_rate": 8.837771083445918e-06,
      "loss": 0.8656,
      "step": 6741
    },
    {
      "epoch": 3.6195765242777087,
      "grad_norm": 0.7578125,
      "learning_rate": 8.83132901153098e-06,
      "loss": 0.9877,
      "step": 6742
    },
    {
      "epoch": 3.620113419012785,
      "grad_norm": 0.62890625,
      "learning_rate": 8.824888784666163e-06,
      "loss": 0.8382,
      "step": 6743
    },
    {
      "epoch": 3.620650313747861,
      "grad_norm": 0.94921875,
      "learning_rate": 8.818450403586387e-06,
      "loss": 1.0092,
      "step": 6744
    },
    {
      "epoch": 3.6211872084829366,
      "grad_norm": 0.80078125,
      "learning_rate": 8.812013869026334e-06,
      "loss": 0.9753,
      "step": 6745
    },
    {
      "epoch": 3.6217241032180127,
      "grad_norm": 0.703125,
      "learning_rate": 8.805579181720507e-06,
      "loss": 0.7735,
      "step": 6746
    },
    {
      "epoch": 3.622260997953089,
      "grad_norm": 0.65625,
      "learning_rate": 8.799146342403178e-06,
      "loss": 0.7595,
      "step": 6747
    },
    {
      "epoch": 3.622797892688165,
      "grad_norm": 0.625,
      "learning_rate": 8.792715351808425e-06,
      "loss": 0.8046,
      "step": 6748
    },
    {
      "epoch": 3.6233347874232407,
      "grad_norm": 0.74609375,
      "learning_rate": 8.786286210670092e-06,
      "loss": 0.8783,
      "step": 6749
    },
    {
      "epoch": 3.623871682158317,
      "grad_norm": 0.62109375,
      "learning_rate": 8.77985891972183e-06,
      "loss": 0.6572,
      "step": 6750
    },
    {
      "epoch": 3.623871682158317,
      "eval_loss": 0.9029676914215088,
      "eval_runtime": 46.3208,
      "eval_samples_per_second": 20.768,
      "eval_steps_per_second": 20.768,
      "step": 6750
    },
    {
      "epoch": 3.624408576893393,
      "grad_norm": 0.734375,
      "learning_rate": 8.773433479697086e-06,
      "loss": 0.8281,
      "step": 6751
    },
    {
      "epoch": 3.624945471628469,
      "grad_norm": 0.7421875,
      "learning_rate": 8.767009891329068e-06,
      "loss": 0.8847,
      "step": 6752
    },
    {
      "epoch": 3.6254823663635447,
      "grad_norm": 0.7109375,
      "learning_rate": 8.760588155350794e-06,
      "loss": 0.884,
      "step": 6753
    },
    {
      "epoch": 3.626019261098621,
      "grad_norm": 0.5703125,
      "learning_rate": 8.754168272495076e-06,
      "loss": 0.7977,
      "step": 6754
    },
    {
      "epoch": 3.626556155833697,
      "grad_norm": 0.6015625,
      "learning_rate": 8.747750243494492e-06,
      "loss": 0.7783,
      "step": 6755
    },
    {
      "epoch": 3.6270930505687726,
      "grad_norm": 0.65625,
      "learning_rate": 8.741334069081419e-06,
      "loss": 0.6992,
      "step": 6756
    },
    {
      "epoch": 3.6276299453038487,
      "grad_norm": 0.6640625,
      "learning_rate": 8.734919749988036e-06,
      "loss": 0.901,
      "step": 6757
    },
    {
      "epoch": 3.628166840038925,
      "grad_norm": 0.71484375,
      "learning_rate": 8.728507286946277e-06,
      "loss": 0.9386,
      "step": 6758
    },
    {
      "epoch": 3.628703734774001,
      "grad_norm": 0.703125,
      "learning_rate": 8.722096680687906e-06,
      "loss": 0.7702,
      "step": 6759
    },
    {
      "epoch": 3.629240629509077,
      "grad_norm": 0.74609375,
      "learning_rate": 8.715687931944449e-06,
      "loss": 0.8265,
      "step": 6760
    },
    {
      "epoch": 3.629777524244153,
      "grad_norm": 0.671875,
      "learning_rate": 8.709281041447217e-06,
      "loss": 0.9102,
      "step": 6761
    },
    {
      "epoch": 3.630314418979229,
      "grad_norm": 0.6484375,
      "learning_rate": 8.702876009927321e-06,
      "loss": 0.7795,
      "step": 6762
    },
    {
      "epoch": 3.630851313714305,
      "grad_norm": 0.83984375,
      "learning_rate": 8.696472838115671e-06,
      "loss": 1.0495,
      "step": 6763
    },
    {
      "epoch": 3.6313882084493807,
      "grad_norm": 0.625,
      "learning_rate": 8.690071526742935e-06,
      "loss": 0.7564,
      "step": 6764
    },
    {
      "epoch": 3.631925103184457,
      "grad_norm": 0.68359375,
      "learning_rate": 8.683672076539579e-06,
      "loss": 1.0103,
      "step": 6765
    },
    {
      "epoch": 3.632461997919533,
      "grad_norm": 0.7265625,
      "learning_rate": 8.67727448823587e-06,
      "loss": 0.6964,
      "step": 6766
    },
    {
      "epoch": 3.632998892654609,
      "grad_norm": 0.828125,
      "learning_rate": 8.670878762561857e-06,
      "loss": 0.9035,
      "step": 6767
    },
    {
      "epoch": 3.6335357873896847,
      "grad_norm": 0.69140625,
      "learning_rate": 8.664484900247363e-06,
      "loss": 0.8722,
      "step": 6768
    },
    {
      "epoch": 3.634072682124761,
      "grad_norm": 0.890625,
      "learning_rate": 8.658092902022017e-06,
      "loss": 0.8307,
      "step": 6769
    },
    {
      "epoch": 3.634609576859837,
      "grad_norm": 0.6953125,
      "learning_rate": 8.651702768615228e-06,
      "loss": 0.8419,
      "step": 6770
    },
    {
      "epoch": 3.635146471594913,
      "grad_norm": 0.70703125,
      "learning_rate": 8.645314500756188e-06,
      "loss": 0.6992,
      "step": 6771
    },
    {
      "epoch": 3.635683366329989,
      "grad_norm": 0.6015625,
      "learning_rate": 8.638928099173873e-06,
      "loss": 0.7571,
      "step": 6772
    },
    {
      "epoch": 3.636220261065065,
      "grad_norm": 0.65625,
      "learning_rate": 8.632543564597065e-06,
      "loss": 0.7757,
      "step": 6773
    },
    {
      "epoch": 3.636757155800141,
      "grad_norm": 0.68359375,
      "learning_rate": 8.626160897754306e-06,
      "loss": 1.0464,
      "step": 6774
    },
    {
      "epoch": 3.6372940505352167,
      "grad_norm": 0.84765625,
      "learning_rate": 8.619780099373945e-06,
      "loss": 0.9449,
      "step": 6775
    },
    {
      "epoch": 3.637830945270293,
      "grad_norm": 0.59765625,
      "learning_rate": 8.613401170184124e-06,
      "loss": 0.7641,
      "step": 6776
    },
    {
      "epoch": 3.638367840005369,
      "grad_norm": 0.671875,
      "learning_rate": 8.607024110912737e-06,
      "loss": 0.9775,
      "step": 6777
    },
    {
      "epoch": 3.638904734740445,
      "grad_norm": 0.546875,
      "learning_rate": 8.600648922287501e-06,
      "loss": 0.6239,
      "step": 6778
    },
    {
      "epoch": 3.639441629475521,
      "grad_norm": 0.7265625,
      "learning_rate": 8.594275605035915e-06,
      "loss": 0.7701,
      "step": 6779
    },
    {
      "epoch": 3.639978524210597,
      "grad_norm": 0.6328125,
      "learning_rate": 8.58790415988524e-06,
      "loss": 0.7851,
      "step": 6780
    },
    {
      "epoch": 3.640515418945673,
      "grad_norm": 0.65234375,
      "learning_rate": 8.581534587562538e-06,
      "loss": 0.8495,
      "step": 6781
    },
    {
      "epoch": 3.641052313680749,
      "grad_norm": 0.6484375,
      "learning_rate": 8.57516688879467e-06,
      "loss": 0.6921,
      "step": 6782
    },
    {
      "epoch": 3.641589208415825,
      "grad_norm": 0.83203125,
      "learning_rate": 8.568801064308252e-06,
      "loss": 1.1063,
      "step": 6783
    },
    {
      "epoch": 3.642126103150901,
      "grad_norm": 0.80078125,
      "learning_rate": 8.56243711482972e-06,
      "loss": 1.1216,
      "step": 6784
    },
    {
      "epoch": 3.642662997885977,
      "grad_norm": 0.69140625,
      "learning_rate": 8.556075041085285e-06,
      "loss": 0.9382,
      "step": 6785
    },
    {
      "epoch": 3.643199892621053,
      "grad_norm": 0.9296875,
      "learning_rate": 8.549714843800927e-06,
      "loss": 0.9093,
      "step": 6786
    },
    {
      "epoch": 3.643736787356129,
      "grad_norm": 0.7109375,
      "learning_rate": 8.543356523702428e-06,
      "loss": 0.7754,
      "step": 6787
    },
    {
      "epoch": 3.644273682091205,
      "grad_norm": 0.65234375,
      "learning_rate": 8.537000081515367e-06,
      "loss": 0.9546,
      "step": 6788
    },
    {
      "epoch": 3.644810576826281,
      "grad_norm": 0.62109375,
      "learning_rate": 8.530645517965083e-06,
      "loss": 0.7963,
      "step": 6789
    },
    {
      "epoch": 3.645347471561357,
      "grad_norm": 0.68359375,
      "learning_rate": 8.524292833776706e-06,
      "loss": 0.9665,
      "step": 6790
    },
    {
      "epoch": 3.645884366296433,
      "grad_norm": 0.703125,
      "learning_rate": 8.517942029675164e-06,
      "loss": 0.7845,
      "step": 6791
    },
    {
      "epoch": 3.646421261031509,
      "grad_norm": 0.54296875,
      "learning_rate": 8.51159310638517e-06,
      "loss": 0.687,
      "step": 6792
    },
    {
      "epoch": 3.646958155766585,
      "grad_norm": 0.703125,
      "learning_rate": 8.505246064631206e-06,
      "loss": 0.8961,
      "step": 6793
    },
    {
      "epoch": 3.647495050501661,
      "grad_norm": 0.75,
      "learning_rate": 8.498900905137555e-06,
      "loss": 0.828,
      "step": 6794
    },
    {
      "epoch": 3.648031945236737,
      "grad_norm": 0.72265625,
      "learning_rate": 8.49255762862829e-06,
      "loss": 0.8958,
      "step": 6795
    },
    {
      "epoch": 3.648568839971813,
      "grad_norm": 0.7734375,
      "learning_rate": 8.486216235827238e-06,
      "loss": 1.0279,
      "step": 6796
    },
    {
      "epoch": 3.649105734706889,
      "grad_norm": 0.74609375,
      "learning_rate": 8.479876727458052e-06,
      "loss": 0.8547,
      "step": 6797
    },
    {
      "epoch": 3.6496426294419653,
      "grad_norm": 0.63671875,
      "learning_rate": 8.473539104244146e-06,
      "loss": 0.9223,
      "step": 6798
    },
    {
      "epoch": 3.650179524177041,
      "grad_norm": 0.75390625,
      "learning_rate": 8.467203366908707e-06,
      "loss": 0.9157,
      "step": 6799
    },
    {
      "epoch": 3.650716418912117,
      "grad_norm": 0.6875,
      "learning_rate": 8.460869516174735e-06,
      "loss": 0.8649,
      "step": 6800
    },
    {
      "epoch": 3.650716418912117,
      "eval_loss": 0.9029330611228943,
      "eval_runtime": 46.3498,
      "eval_samples_per_second": 20.755,
      "eval_steps_per_second": 20.755,
      "step": 6800
    },
    {
      "epoch": 3.651253313647193,
      "grad_norm": 0.72265625,
      "learning_rate": 8.454537552765012e-06,
      "loss": 0.8005,
      "step": 6801
    },
    {
      "epoch": 3.651790208382269,
      "grad_norm": 0.66796875,
      "learning_rate": 8.448207477402076e-06,
      "loss": 0.8917,
      "step": 6802
    },
    {
      "epoch": 3.652327103117345,
      "grad_norm": 0.69140625,
      "learning_rate": 8.44187929080828e-06,
      "loss": 0.982,
      "step": 6803
    },
    {
      "epoch": 3.652863997852421,
      "grad_norm": 0.69921875,
      "learning_rate": 8.435552993705753e-06,
      "loss": 0.8059,
      "step": 6804
    },
    {
      "epoch": 3.6534008925874972,
      "grad_norm": 0.66015625,
      "learning_rate": 8.429228586816403e-06,
      "loss": 0.9227,
      "step": 6805
    },
    {
      "epoch": 3.653937787322573,
      "grad_norm": 0.671875,
      "learning_rate": 8.42290607086192e-06,
      "loss": 0.8203,
      "step": 6806
    },
    {
      "epoch": 3.654474682057649,
      "grad_norm": 0.890625,
      "learning_rate": 8.41658544656378e-06,
      "loss": 1.0071,
      "step": 6807
    },
    {
      "epoch": 3.655011576792725,
      "grad_norm": 0.6953125,
      "learning_rate": 8.410266714643261e-06,
      "loss": 0.7722,
      "step": 6808
    },
    {
      "epoch": 3.6555484715278013,
      "grad_norm": 0.7109375,
      "learning_rate": 8.403949875821397e-06,
      "loss": 1.0494,
      "step": 6809
    },
    {
      "epoch": 3.656085366262877,
      "grad_norm": 0.62890625,
      "learning_rate": 8.397634930819021e-06,
      "loss": 0.9386,
      "step": 6810
    },
    {
      "epoch": 3.656622260997953,
      "grad_norm": 0.66796875,
      "learning_rate": 8.391321880356761e-06,
      "loss": 0.8423,
      "step": 6811
    },
    {
      "epoch": 3.657159155733029,
      "grad_norm": 0.625,
      "learning_rate": 8.385010725154999e-06,
      "loss": 0.697,
      "step": 6812
    },
    {
      "epoch": 3.657696050468105,
      "grad_norm": 0.62890625,
      "learning_rate": 8.37870146593393e-06,
      "loss": 0.8463,
      "step": 6813
    },
    {
      "epoch": 3.658232945203181,
      "grad_norm": 0.6953125,
      "learning_rate": 8.372394103413519e-06,
      "loss": 0.8213,
      "step": 6814
    },
    {
      "epoch": 3.658769839938257,
      "grad_norm": 0.65625,
      "learning_rate": 8.366088638313504e-06,
      "loss": 0.6919,
      "step": 6815
    },
    {
      "epoch": 3.6593067346733332,
      "grad_norm": 0.6171875,
      "learning_rate": 8.359785071353423e-06,
      "loss": 0.6444,
      "step": 6816
    },
    {
      "epoch": 3.6598436294084093,
      "grad_norm": 0.65234375,
      "learning_rate": 8.35348340325261e-06,
      "loss": 0.7761,
      "step": 6817
    },
    {
      "epoch": 3.660380524143485,
      "grad_norm": 0.65234375,
      "learning_rate": 8.347183634730142e-06,
      "loss": 0.9443,
      "step": 6818
    },
    {
      "epoch": 3.660917418878561,
      "grad_norm": 0.65625,
      "learning_rate": 8.34088576650491e-06,
      "loss": 0.7114,
      "step": 6819
    },
    {
      "epoch": 3.6614543136136373,
      "grad_norm": 0.68359375,
      "learning_rate": 8.334589799295591e-06,
      "loss": 1.1432,
      "step": 6820
    },
    {
      "epoch": 3.661991208348713,
      "grad_norm": 0.66015625,
      "learning_rate": 8.32829573382062e-06,
      "loss": 0.7016,
      "step": 6821
    },
    {
      "epoch": 3.662528103083789,
      "grad_norm": 0.66015625,
      "learning_rate": 8.322003570798243e-06,
      "loss": 0.876,
      "step": 6822
    },
    {
      "epoch": 3.663064997818865,
      "grad_norm": 0.6875,
      "learning_rate": 8.315713310946469e-06,
      "loss": 0.7872,
      "step": 6823
    },
    {
      "epoch": 3.6636018925539413,
      "grad_norm": 0.65625,
      "learning_rate": 8.309424954983086e-06,
      "loss": 0.7113,
      "step": 6824
    },
    {
      "epoch": 3.664138787289017,
      "grad_norm": 0.70703125,
      "learning_rate": 8.303138503625688e-06,
      "loss": 0.8455,
      "step": 6825
    },
    {
      "epoch": 3.664675682024093,
      "grad_norm": 0.7890625,
      "learning_rate": 8.296853957591638e-06,
      "loss": 1.0172,
      "step": 6826
    },
    {
      "epoch": 3.6652125767591692,
      "grad_norm": 0.58203125,
      "learning_rate": 8.290571317598075e-06,
      "loss": 0.6483,
      "step": 6827
    },
    {
      "epoch": 3.6657494714942453,
      "grad_norm": 0.6796875,
      "learning_rate": 8.284290584361934e-06,
      "loss": 0.8555,
      "step": 6828
    },
    {
      "epoch": 3.666286366229321,
      "grad_norm": 0.8203125,
      "learning_rate": 8.278011758599929e-06,
      "loss": 0.8592,
      "step": 6829
    },
    {
      "epoch": 3.666823260964397,
      "grad_norm": 0.78125,
      "learning_rate": 8.271734841028553e-06,
      "loss": 0.7839,
      "step": 6830
    },
    {
      "epoch": 3.6673601556994733,
      "grad_norm": 0.63671875,
      "learning_rate": 8.265459832364066e-06,
      "loss": 0.8585,
      "step": 6831
    },
    {
      "epoch": 3.667897050434549,
      "grad_norm": 0.6953125,
      "learning_rate": 8.25918673332254e-06,
      "loss": 0.8377,
      "step": 6832
    },
    {
      "epoch": 3.668433945169625,
      "grad_norm": 0.70703125,
      "learning_rate": 8.252915544619821e-06,
      "loss": 0.8478,
      "step": 6833
    },
    {
      "epoch": 3.668970839904701,
      "grad_norm": 0.69921875,
      "learning_rate": 8.246646266971516e-06,
      "loss": 0.7837,
      "step": 6834
    },
    {
      "epoch": 3.6695077346397773,
      "grad_norm": 0.71875,
      "learning_rate": 8.240378901093034e-06,
      "loss": 0.8574,
      "step": 6835
    },
    {
      "epoch": 3.6700446293748534,
      "grad_norm": 0.6875,
      "learning_rate": 8.234113447699573e-06,
      "loss": 0.9252,
      "step": 6836
    },
    {
      "epoch": 3.670581524109929,
      "grad_norm": 0.65234375,
      "learning_rate": 8.22784990750608e-06,
      "loss": 0.8353,
      "step": 6837
    },
    {
      "epoch": 3.6711184188450052,
      "grad_norm": 0.7421875,
      "learning_rate": 8.221588281227315e-06,
      "loss": 0.9691,
      "step": 6838
    },
    {
      "epoch": 3.6716553135800813,
      "grad_norm": 0.68359375,
      "learning_rate": 8.215328569577826e-06,
      "loss": 0.7536,
      "step": 6839
    },
    {
      "epoch": 3.672192208315157,
      "grad_norm": 0.61328125,
      "learning_rate": 8.209070773271894e-06,
      "loss": 0.7848,
      "step": 6840
    },
    {
      "epoch": 3.672729103050233,
      "grad_norm": 0.734375,
      "learning_rate": 8.202814893023621e-06,
      "loss": 0.7424,
      "step": 6841
    },
    {
      "epoch": 3.6732659977853093,
      "grad_norm": 0.78515625,
      "learning_rate": 8.196560929546898e-06,
      "loss": 1.1235,
      "step": 6842
    },
    {
      "epoch": 3.6738028925203854,
      "grad_norm": 0.73828125,
      "learning_rate": 8.190308883555365e-06,
      "loss": 0.8499,
      "step": 6843
    },
    {
      "epoch": 3.674339787255461,
      "grad_norm": 0.6640625,
      "learning_rate": 8.184058755762466e-06,
      "loss": 0.8288,
      "step": 6844
    },
    {
      "epoch": 3.674876681990537,
      "grad_norm": 0.640625,
      "learning_rate": 8.177810546881426e-06,
      "loss": 0.7531,
      "step": 6845
    },
    {
      "epoch": 3.6754135767256133,
      "grad_norm": 0.6640625,
      "learning_rate": 8.17156425762523e-06,
      "loss": 0.7914,
      "step": 6846
    },
    {
      "epoch": 3.6759504714606894,
      "grad_norm": 0.6484375,
      "learning_rate": 8.165319888706677e-06,
      "loss": 0.8858,
      "step": 6847
    },
    {
      "epoch": 3.676487366195765,
      "grad_norm": 0.86328125,
      "learning_rate": 8.15907744083831e-06,
      "loss": 0.8575,
      "step": 6848
    },
    {
      "epoch": 3.6770242609308412,
      "grad_norm": 0.79296875,
      "learning_rate": 8.152836914732487e-06,
      "loss": 0.7521,
      "step": 6849
    },
    {
      "epoch": 3.6775611556659173,
      "grad_norm": 0.67578125,
      "learning_rate": 8.146598311101317e-06,
      "loss": 0.805,
      "step": 6850
    },
    {
      "epoch": 3.6775611556659173,
      "eval_loss": 0.9028520584106445,
      "eval_runtime": 46.2082,
      "eval_samples_per_second": 20.819,
      "eval_steps_per_second": 20.819,
      "step": 6850
    },
    {
      "epoch": 3.678098050400993,
      "grad_norm": 0.7578125,
      "learning_rate": 8.14036163065671e-06,
      "loss": 0.9158,
      "step": 6851
    },
    {
      "epoch": 3.678634945136069,
      "grad_norm": 0.67578125,
      "learning_rate": 8.13412687411036e-06,
      "loss": 0.7099,
      "step": 6852
    },
    {
      "epoch": 3.6791718398711453,
      "grad_norm": 0.6640625,
      "learning_rate": 8.127894042173714e-06,
      "loss": 0.7832,
      "step": 6853
    },
    {
      "epoch": 3.6797087346062214,
      "grad_norm": 0.73046875,
      "learning_rate": 8.121663135558032e-06,
      "loss": 0.8488,
      "step": 6854
    },
    {
      "epoch": 3.6802456293412975,
      "grad_norm": 0.703125,
      "learning_rate": 8.115434154974339e-06,
      "loss": 0.9325,
      "step": 6855
    },
    {
      "epoch": 3.680782524076373,
      "grad_norm": 0.71875,
      "learning_rate": 8.109207101133421e-06,
      "loss": 0.8746,
      "step": 6856
    },
    {
      "epoch": 3.6813194188114493,
      "grad_norm": 0.6015625,
      "learning_rate": 8.10298197474588e-06,
      "loss": 0.9481,
      "step": 6857
    },
    {
      "epoch": 3.6818563135465254,
      "grad_norm": 0.71875,
      "learning_rate": 8.096758776522085e-06,
      "loss": 1.0148,
      "step": 6858
    },
    {
      "epoch": 3.682393208281601,
      "grad_norm": 0.78515625,
      "learning_rate": 8.09053750717217e-06,
      "loss": 0.7556,
      "step": 6859
    },
    {
      "epoch": 3.6829301030166772,
      "grad_norm": 0.75,
      "learning_rate": 8.084318167406063e-06,
      "loss": 0.7653,
      "step": 6860
    },
    {
      "epoch": 3.6834669977517533,
      "grad_norm": 0.984375,
      "learning_rate": 8.078100757933485e-06,
      "loss": 0.8995,
      "step": 6861
    },
    {
      "epoch": 3.684003892486829,
      "grad_norm": 0.81640625,
      "learning_rate": 8.071885279463898e-06,
      "loss": 0.87,
      "step": 6862
    },
    {
      "epoch": 3.684540787221905,
      "grad_norm": 0.7265625,
      "learning_rate": 8.065671732706581e-06,
      "loss": 0.7983,
      "step": 6863
    },
    {
      "epoch": 3.6850776819569813,
      "grad_norm": 0.6640625,
      "learning_rate": 8.059460118370588e-06,
      "loss": 0.7964,
      "step": 6864
    },
    {
      "epoch": 3.6856145766920574,
      "grad_norm": 0.70703125,
      "learning_rate": 8.053250437164717e-06,
      "loss": 0.8647,
      "step": 6865
    },
    {
      "epoch": 3.6861514714271335,
      "grad_norm": 0.77734375,
      "learning_rate": 8.047042689797582e-06,
      "loss": 0.8245,
      "step": 6866
    },
    {
      "epoch": 3.686688366162209,
      "grad_norm": 0.69140625,
      "learning_rate": 8.040836876977576e-06,
      "loss": 0.8216,
      "step": 6867
    },
    {
      "epoch": 3.6872252608972853,
      "grad_norm": 0.78515625,
      "learning_rate": 8.034632999412846e-06,
      "loss": 0.8123,
      "step": 6868
    },
    {
      "epoch": 3.6877621556323614,
      "grad_norm": 0.79296875,
      "learning_rate": 8.02843105781134e-06,
      "loss": 1.2279,
      "step": 6869
    },
    {
      "epoch": 3.688299050367437,
      "grad_norm": 0.75390625,
      "learning_rate": 8.022231052880786e-06,
      "loss": 0.882,
      "step": 6870
    },
    {
      "epoch": 3.6888359451025132,
      "grad_norm": 0.6875,
      "learning_rate": 8.016032985328664e-06,
      "loss": 0.8362,
      "step": 6871
    },
    {
      "epoch": 3.6893728398375893,
      "grad_norm": 0.7578125,
      "learning_rate": 8.009836855862274e-06,
      "loss": 0.829,
      "step": 6872
    },
    {
      "epoch": 3.6899097345726655,
      "grad_norm": 0.7109375,
      "learning_rate": 8.003642665188652e-06,
      "loss": 0.7817,
      "step": 6873
    },
    {
      "epoch": 3.6904466293077416,
      "grad_norm": 1.203125,
      "learning_rate": 7.99745041401465e-06,
      "loss": 0.8557,
      "step": 6874
    },
    {
      "epoch": 3.6909835240428173,
      "grad_norm": 0.625,
      "learning_rate": 7.991260103046872e-06,
      "loss": 0.7789,
      "step": 6875
    },
    {
      "epoch": 3.6915204187778934,
      "grad_norm": 0.67578125,
      "learning_rate": 7.98507173299171e-06,
      "loss": 0.7369,
      "step": 6876
    },
    {
      "epoch": 3.6920573135129695,
      "grad_norm": 0.6875,
      "learning_rate": 7.978885304555353e-06,
      "loss": 1.0639,
      "step": 6877
    },
    {
      "epoch": 3.692594208248045,
      "grad_norm": 0.6953125,
      "learning_rate": 7.972700818443726e-06,
      "loss": 0.8837,
      "step": 6878
    },
    {
      "epoch": 3.6931311029831213,
      "grad_norm": 0.7421875,
      "learning_rate": 7.96651827536257e-06,
      "loss": 0.7951,
      "step": 6879
    },
    {
      "epoch": 3.6936679977181974,
      "grad_norm": 0.703125,
      "learning_rate": 7.960337676017395e-06,
      "loss": 0.7642,
      "step": 6880
    },
    {
      "epoch": 3.694204892453273,
      "grad_norm": 0.6953125,
      "learning_rate": 7.954159021113486e-06,
      "loss": 0.9309,
      "step": 6881
    },
    {
      "epoch": 3.6947417871883492,
      "grad_norm": 0.62890625,
      "learning_rate": 7.947982311355893e-06,
      "loss": 0.7926,
      "step": 6882
    },
    {
      "epoch": 3.6952786819234253,
      "grad_norm": 0.71484375,
      "learning_rate": 7.94180754744947e-06,
      "loss": 0.8497,
      "step": 6883
    },
    {
      "epoch": 3.6958155766585015,
      "grad_norm": 0.69140625,
      "learning_rate": 7.935634730098821e-06,
      "loss": 0.785,
      "step": 6884
    },
    {
      "epoch": 3.6963524713935776,
      "grad_norm": 0.640625,
      "learning_rate": 7.929463860008355e-06,
      "loss": 0.7934,
      "step": 6885
    },
    {
      "epoch": 3.6968893661286533,
      "grad_norm": 0.63671875,
      "learning_rate": 7.92329493788225e-06,
      "loss": 0.6657,
      "step": 6886
    },
    {
      "epoch": 3.6974262608637294,
      "grad_norm": 0.7421875,
      "learning_rate": 7.917127964424446e-06,
      "loss": 0.8921,
      "step": 6887
    },
    {
      "epoch": 3.6979631555988055,
      "grad_norm": 0.6640625,
      "learning_rate": 7.910962940338673e-06,
      "loss": 0.7757,
      "step": 6888
    },
    {
      "epoch": 3.698500050333881,
      "grad_norm": 0.6875,
      "learning_rate": 7.904799866328454e-06,
      "loss": 0.9523,
      "step": 6889
    },
    {
      "epoch": 3.6990369450689573,
      "grad_norm": 0.76171875,
      "learning_rate": 7.898638743097066e-06,
      "loss": 0.7095,
      "step": 6890
    },
    {
      "epoch": 3.6995738398040334,
      "grad_norm": 0.72265625,
      "learning_rate": 7.892479571347559e-06,
      "loss": 1.0438,
      "step": 6891
    },
    {
      "epoch": 3.7001107345391095,
      "grad_norm": 0.62890625,
      "learning_rate": 7.886322351782783e-06,
      "loss": 0.7061,
      "step": 6892
    },
    {
      "epoch": 3.7006476292741857,
      "grad_norm": 0.71875,
      "learning_rate": 7.880167085105362e-06,
      "loss": 0.7912,
      "step": 6893
    },
    {
      "epoch": 3.7011845240092613,
      "grad_norm": 0.71484375,
      "learning_rate": 7.874013772017672e-06,
      "loss": 0.7393,
      "step": 6894
    },
    {
      "epoch": 3.7017214187443375,
      "grad_norm": 0.9296875,
      "learning_rate": 7.867862413221893e-06,
      "loss": 0.9521,
      "step": 6895
    },
    {
      "epoch": 3.7022583134794136,
      "grad_norm": 0.7734375,
      "learning_rate": 7.861713009419982e-06,
      "loss": 0.9284,
      "step": 6896
    },
    {
      "epoch": 3.7027952082144893,
      "grad_norm": 0.71484375,
      "learning_rate": 7.855565561313658e-06,
      "loss": 0.7724,
      "step": 6897
    },
    {
      "epoch": 3.7033321029495654,
      "grad_norm": 0.6875,
      "learning_rate": 7.84942006960441e-06,
      "loss": 0.742,
      "step": 6898
    },
    {
      "epoch": 3.7038689976846415,
      "grad_norm": 0.796875,
      "learning_rate": 7.843276534993535e-06,
      "loss": 0.9588,
      "step": 6899
    },
    {
      "epoch": 3.704405892419717,
      "grad_norm": 0.6640625,
      "learning_rate": 7.837134958182072e-06,
      "loss": 0.8108,
      "step": 6900
    },
    {
      "epoch": 3.704405892419717,
      "eval_loss": 0.9026853442192078,
      "eval_runtime": 46.02,
      "eval_samples_per_second": 20.904,
      "eval_steps_per_second": 20.904,
      "step": 6900
    },
    {
      "epoch": 3.7049427871547933,
      "grad_norm": 0.73046875,
      "learning_rate": 7.830995339870858e-06,
      "loss": 0.9668,
      "step": 6901
    },
    {
      "epoch": 3.7054796818898694,
      "grad_norm": 0.671875,
      "learning_rate": 7.824857680760516e-06,
      "loss": 0.9215,
      "step": 6902
    },
    {
      "epoch": 3.7060165766249455,
      "grad_norm": 0.62109375,
      "learning_rate": 7.818721981551408e-06,
      "loss": 0.7267,
      "step": 6903
    },
    {
      "epoch": 3.7065534713600217,
      "grad_norm": 0.71875,
      "learning_rate": 7.812588242943706e-06,
      "loss": 0.9404,
      "step": 6904
    },
    {
      "epoch": 3.7070903660950973,
      "grad_norm": 0.62109375,
      "learning_rate": 7.806456465637352e-06,
      "loss": 0.7769,
      "step": 6905
    },
    {
      "epoch": 3.7076272608301735,
      "grad_norm": 0.734375,
      "learning_rate": 7.800326650332058e-06,
      "loss": 0.9568,
      "step": 6906
    },
    {
      "epoch": 3.7081641555652496,
      "grad_norm": 0.6796875,
      "learning_rate": 7.794198797727299e-06,
      "loss": 0.7476,
      "step": 6907
    },
    {
      "epoch": 3.7087010503003253,
      "grad_norm": 0.6875,
      "learning_rate": 7.788072908522356e-06,
      "loss": 0.8911,
      "step": 6908
    },
    {
      "epoch": 3.7092379450354014,
      "grad_norm": 0.64453125,
      "learning_rate": 7.781948983416262e-06,
      "loss": 0.8081,
      "step": 6909
    },
    {
      "epoch": 3.7097748397704775,
      "grad_norm": 0.734375,
      "learning_rate": 7.775827023107835e-06,
      "loss": 0.8794,
      "step": 6910
    },
    {
      "epoch": 3.7103117345055536,
      "grad_norm": 0.828125,
      "learning_rate": 7.769707028295681e-06,
      "loss": 0.8441,
      "step": 6911
    },
    {
      "epoch": 3.7108486292406297,
      "grad_norm": 0.6484375,
      "learning_rate": 7.76358899967815e-06,
      "loss": 0.9243,
      "step": 6912
    },
    {
      "epoch": 3.7113855239757054,
      "grad_norm": 0.7109375,
      "learning_rate": 7.757472937953398e-06,
      "loss": 0.7727,
      "step": 6913
    },
    {
      "epoch": 3.7119224187107815,
      "grad_norm": 0.71484375,
      "learning_rate": 7.751358843819348e-06,
      "loss": 0.7647,
      "step": 6914
    },
    {
      "epoch": 3.7124593134458577,
      "grad_norm": 0.69140625,
      "learning_rate": 7.745246717973692e-06,
      "loss": 0.6932,
      "step": 6915
    },
    {
      "epoch": 3.7129962081809333,
      "grad_norm": 0.703125,
      "learning_rate": 7.739136561113891e-06,
      "loss": 0.7426,
      "step": 6916
    },
    {
      "epoch": 3.7135331029160095,
      "grad_norm": 0.6328125,
      "learning_rate": 7.733028373937198e-06,
      "loss": 0.6847,
      "step": 6917
    },
    {
      "epoch": 3.7140699976510856,
      "grad_norm": 0.828125,
      "learning_rate": 7.726922157140645e-06,
      "loss": 0.9353,
      "step": 6918
    },
    {
      "epoch": 3.7146068923861613,
      "grad_norm": 0.671875,
      "learning_rate": 7.720817911421013e-06,
      "loss": 0.8364,
      "step": 6919
    },
    {
      "epoch": 3.7151437871212374,
      "grad_norm": 0.6953125,
      "learning_rate": 7.714715637474879e-06,
      "loss": 0.8889,
      "step": 6920
    },
    {
      "epoch": 3.7156806818563135,
      "grad_norm": 0.71875,
      "learning_rate": 7.7086153359986e-06,
      "loss": 0.9004,
      "step": 6921
    },
    {
      "epoch": 3.7162175765913896,
      "grad_norm": 0.6171875,
      "learning_rate": 7.702517007688279e-06,
      "loss": 0.6917,
      "step": 6922
    },
    {
      "epoch": 3.7167544713264657,
      "grad_norm": 0.6875,
      "learning_rate": 7.696420653239833e-06,
      "loss": 0.9131,
      "step": 6923
    },
    {
      "epoch": 3.7172913660615414,
      "grad_norm": 0.6796875,
      "learning_rate": 7.690326273348922e-06,
      "loss": 0.8444,
      "step": 6924
    },
    {
      "epoch": 3.7178282607966175,
      "grad_norm": 0.70703125,
      "learning_rate": 7.684233868710982e-06,
      "loss": 0.6928,
      "step": 6925
    },
    {
      "epoch": 3.7183651555316937,
      "grad_norm": 0.66015625,
      "learning_rate": 7.678143440021246e-06,
      "loss": 0.7861,
      "step": 6926
    },
    {
      "epoch": 3.7189020502667693,
      "grad_norm": 0.6640625,
      "learning_rate": 7.672054987974713e-06,
      "loss": 0.9232,
      "step": 6927
    },
    {
      "epoch": 3.7194389450018455,
      "grad_norm": 0.71875,
      "learning_rate": 7.665968513266142e-06,
      "loss": 0.7922,
      "step": 6928
    },
    {
      "epoch": 3.7199758397369216,
      "grad_norm": 0.7890625,
      "learning_rate": 7.65988401659008e-06,
      "loss": 0.9256,
      "step": 6929
    },
    {
      "epoch": 3.7205127344719977,
      "grad_norm": 0.66796875,
      "learning_rate": 7.653801498640853e-06,
      "loss": 0.8112,
      "step": 6930
    },
    {
      "epoch": 3.721049629207074,
      "grad_norm": 0.6875,
      "learning_rate": 7.64772096011255e-06,
      "loss": 0.7897,
      "step": 6931
    },
    {
      "epoch": 3.7215865239421495,
      "grad_norm": 0.66015625,
      "learning_rate": 7.641642401699022e-06,
      "loss": 0.8193,
      "step": 6932
    },
    {
      "epoch": 3.7221234186772256,
      "grad_norm": 0.74609375,
      "learning_rate": 7.635565824093926e-06,
      "loss": 0.8484,
      "step": 6933
    },
    {
      "epoch": 3.7226603134123017,
      "grad_norm": 0.6640625,
      "learning_rate": 7.629491227990679e-06,
      "loss": 0.8381,
      "step": 6934
    },
    {
      "epoch": 3.7231972081473774,
      "grad_norm": 0.578125,
      "learning_rate": 7.623418614082453e-06,
      "loss": 0.7133,
      "step": 6935
    },
    {
      "epoch": 3.7237341028824535,
      "grad_norm": 0.64453125,
      "learning_rate": 7.6173479830622205e-06,
      "loss": 0.7596,
      "step": 6936
    },
    {
      "epoch": 3.7242709976175297,
      "grad_norm": 0.65625,
      "learning_rate": 7.6112793356227255e-06,
      "loss": 0.7272,
      "step": 6937
    },
    {
      "epoch": 3.7248078923526053,
      "grad_norm": 0.6953125,
      "learning_rate": 7.605212672456461e-06,
      "loss": 1.0197,
      "step": 6938
    },
    {
      "epoch": 3.7253447870876815,
      "grad_norm": 0.7890625,
      "learning_rate": 7.599147994255723e-06,
      "loss": 0.9029,
      "step": 6939
    },
    {
      "epoch": 3.7258816818227576,
      "grad_norm": 0.9375,
      "learning_rate": 7.593085301712566e-06,
      "loss": 0.8767,
      "step": 6940
    },
    {
      "epoch": 3.7264185765578337,
      "grad_norm": 0.60546875,
      "learning_rate": 7.587024595518804e-06,
      "loss": 0.8026,
      "step": 6941
    },
    {
      "epoch": 3.72695547129291,
      "grad_norm": 0.6796875,
      "learning_rate": 7.580965876366058e-06,
      "loss": 0.7873,
      "step": 6942
    },
    {
      "epoch": 3.7274923660279855,
      "grad_norm": 0.69140625,
      "learning_rate": 7.5749091449457074e-06,
      "loss": 0.8802,
      "step": 6943
    },
    {
      "epoch": 3.7280292607630616,
      "grad_norm": 0.68359375,
      "learning_rate": 7.568854401948886e-06,
      "loss": 0.6855,
      "step": 6944
    },
    {
      "epoch": 3.7285661554981377,
      "grad_norm": 0.76171875,
      "learning_rate": 7.5628016480665225e-06,
      "loss": 0.7593,
      "step": 6945
    },
    {
      "epoch": 3.7291030502332134,
      "grad_norm": 0.71484375,
      "learning_rate": 7.556750883989324e-06,
      "loss": 0.9939,
      "step": 6946
    },
    {
      "epoch": 3.7296399449682895,
      "grad_norm": 0.68359375,
      "learning_rate": 7.550702110407743e-06,
      "loss": 0.9394,
      "step": 6947
    },
    {
      "epoch": 3.7301768397033657,
      "grad_norm": 0.609375,
      "learning_rate": 7.544655328012038e-06,
      "loss": 0.7396,
      "step": 6948
    },
    {
      "epoch": 3.730713734438442,
      "grad_norm": 0.71484375,
      "learning_rate": 7.538610537492213e-06,
      "loss": 0.8867,
      "step": 6949
    },
    {
      "epoch": 3.731250629173518,
      "grad_norm": 0.7578125,
      "learning_rate": 7.532567739538049e-06,
      "loss": 0.8756,
      "step": 6950
    },
    {
      "epoch": 3.731250629173518,
      "eval_loss": 0.9027646780014038,
      "eval_runtime": 45.7834,
      "eval_samples_per_second": 21.012,
      "eval_steps_per_second": 21.012,
      "step": 6950
    },
    {
      "epoch": 3.7317875239085936,
      "grad_norm": 0.76171875,
      "learning_rate": 7.526526934839112e-06,
      "loss": 0.8987,
      "step": 6951
    },
    {
      "epoch": 3.7323244186436697,
      "grad_norm": 0.859375,
      "learning_rate": 7.520488124084743e-06,
      "loss": 0.8377,
      "step": 6952
    },
    {
      "epoch": 3.732861313378746,
      "grad_norm": 0.66015625,
      "learning_rate": 7.514451307964032e-06,
      "loss": 0.6978,
      "step": 6953
    },
    {
      "epoch": 3.7333982081138215,
      "grad_norm": 0.6953125,
      "learning_rate": 7.508416487165862e-06,
      "loss": 0.7056,
      "step": 6954
    },
    {
      "epoch": 3.7339351028488976,
      "grad_norm": 0.71484375,
      "learning_rate": 7.502383662378895e-06,
      "loss": 0.8986,
      "step": 6955
    },
    {
      "epoch": 3.7344719975839737,
      "grad_norm": 0.6875,
      "learning_rate": 7.49635283429154e-06,
      "loss": 0.85,
      "step": 6956
    },
    {
      "epoch": 3.7350088923190494,
      "grad_norm": 0.7578125,
      "learning_rate": 7.490324003591984e-06,
      "loss": 0.843,
      "step": 6957
    },
    {
      "epoch": 3.7355457870541255,
      "grad_norm": 0.7421875,
      "learning_rate": 7.4842971709681995e-06,
      "loss": 0.7462,
      "step": 6958
    },
    {
      "epoch": 3.7360826817892017,
      "grad_norm": 0.68359375,
      "learning_rate": 7.478272337107936e-06,
      "loss": 0.8335,
      "step": 6959
    },
    {
      "epoch": 3.736619576524278,
      "grad_norm": 0.69140625,
      "learning_rate": 7.472249502698686e-06,
      "loss": 0.8435,
      "step": 6960
    },
    {
      "epoch": 3.737156471259354,
      "grad_norm": 0.62109375,
      "learning_rate": 7.466228668427738e-06,
      "loss": 0.8474,
      "step": 6961
    },
    {
      "epoch": 3.7376933659944296,
      "grad_norm": 0.66015625,
      "learning_rate": 7.460209834982154e-06,
      "loss": 0.9112,
      "step": 6962
    },
    {
      "epoch": 3.7382302607295057,
      "grad_norm": 0.68359375,
      "learning_rate": 7.454193003048743e-06,
      "loss": 0.6978,
      "step": 6963
    },
    {
      "epoch": 3.738767155464582,
      "grad_norm": 0.69921875,
      "learning_rate": 7.448178173314111e-06,
      "loss": 0.8502,
      "step": 6964
    },
    {
      "epoch": 3.7393040501996575,
      "grad_norm": 0.7109375,
      "learning_rate": 7.4421653464646416e-06,
      "loss": 0.8533,
      "step": 6965
    },
    {
      "epoch": 3.7398409449347336,
      "grad_norm": 0.72265625,
      "learning_rate": 7.4361545231864425e-06,
      "loss": 0.9364,
      "step": 6966
    },
    {
      "epoch": 3.7403778396698097,
      "grad_norm": 0.6328125,
      "learning_rate": 7.430145704165439e-06,
      "loss": 0.5971,
      "step": 6967
    },
    {
      "epoch": 3.740914734404886,
      "grad_norm": 0.765625,
      "learning_rate": 7.424138890087323e-06,
      "loss": 0.8235,
      "step": 6968
    },
    {
      "epoch": 3.741451629139962,
      "grad_norm": 0.75390625,
      "learning_rate": 7.418134081637532e-06,
      "loss": 0.8988,
      "step": 6969
    },
    {
      "epoch": 3.7419885238750377,
      "grad_norm": 0.77734375,
      "learning_rate": 7.412131279501297e-06,
      "loss": 0.9451,
      "step": 6970
    },
    {
      "epoch": 3.742525418610114,
      "grad_norm": 0.7109375,
      "learning_rate": 7.406130484363627e-06,
      "loss": 0.9615,
      "step": 6971
    },
    {
      "epoch": 3.74306231334519,
      "grad_norm": 0.59765625,
      "learning_rate": 7.400131696909266e-06,
      "loss": 0.9225,
      "step": 6972
    },
    {
      "epoch": 3.7435992080802656,
      "grad_norm": 0.71484375,
      "learning_rate": 7.3941349178227735e-06,
      "loss": 1.036,
      "step": 6973
    },
    {
      "epoch": 3.7441361028153417,
      "grad_norm": 0.62109375,
      "learning_rate": 7.388140147788436e-06,
      "loss": 0.7147,
      "step": 6974
    },
    {
      "epoch": 3.744672997550418,
      "grad_norm": 0.73046875,
      "learning_rate": 7.382147387490354e-06,
      "loss": 0.8636,
      "step": 6975
    },
    {
      "epoch": 3.7452098922854935,
      "grad_norm": 0.671875,
      "learning_rate": 7.376156637612358e-06,
      "loss": 0.7157,
      "step": 6976
    },
    {
      "epoch": 3.7457467870205696,
      "grad_norm": 0.734375,
      "learning_rate": 7.370167898838079e-06,
      "loss": 0.9783,
      "step": 6977
    },
    {
      "epoch": 3.7462836817556457,
      "grad_norm": 0.73046875,
      "learning_rate": 7.364181171850912e-06,
      "loss": 0.9377,
      "step": 6978
    },
    {
      "epoch": 3.746820576490722,
      "grad_norm": 0.66015625,
      "learning_rate": 7.358196457334007e-06,
      "loss": 0.8323,
      "step": 6979
    },
    {
      "epoch": 3.747357471225798,
      "grad_norm": 0.703125,
      "learning_rate": 7.352213755970302e-06,
      "loss": 0.9374,
      "step": 6980
    },
    {
      "epoch": 3.7478943659608737,
      "grad_norm": 0.8125,
      "learning_rate": 7.346233068442512e-06,
      "loss": 0.8611,
      "step": 6981
    },
    {
      "epoch": 3.74843126069595,
      "grad_norm": 0.7109375,
      "learning_rate": 7.3402543954330825e-06,
      "loss": 1.0845,
      "step": 6982
    },
    {
      "epoch": 3.748968155431026,
      "grad_norm": 0.65625,
      "learning_rate": 7.334277737624268e-06,
      "loss": 0.6241,
      "step": 6983
    },
    {
      "epoch": 3.7495050501661016,
      "grad_norm": 0.65234375,
      "learning_rate": 7.328303095698089e-06,
      "loss": 0.7232,
      "step": 6984
    },
    {
      "epoch": 3.7500419449011777,
      "grad_norm": 0.609375,
      "learning_rate": 7.3223304703363135e-06,
      "loss": 1.0548,
      "step": 6985
    },
    {
      "epoch": 3.750578839636254,
      "grad_norm": 0.8671875,
      "learning_rate": 7.316359862220501e-06,
      "loss": 1.0695,
      "step": 6986
    },
    {
      "epoch": 3.75111573437133,
      "grad_norm": 0.7109375,
      "learning_rate": 7.310391272031983e-06,
      "loss": 0.7739,
      "step": 6987
    },
    {
      "epoch": 3.751652629106406,
      "grad_norm": 0.703125,
      "learning_rate": 7.304424700451831e-06,
      "loss": 0.7856,
      "step": 6988
    },
    {
      "epoch": 3.7521895238414817,
      "grad_norm": 0.75,
      "learning_rate": 7.298460148160918e-06,
      "loss": 0.9983,
      "step": 6989
    },
    {
      "epoch": 3.752726418576558,
      "grad_norm": 0.7109375,
      "learning_rate": 7.292497615839889e-06,
      "loss": 0.7296,
      "step": 6990
    },
    {
      "epoch": 3.753263313311634,
      "grad_norm": 0.703125,
      "learning_rate": 7.286537104169114e-06,
      "loss": 0.8161,
      "step": 6991
    },
    {
      "epoch": 3.7538002080467097,
      "grad_norm": 0.671875,
      "learning_rate": 7.280578613828778e-06,
      "loss": 0.7749,
      "step": 6992
    },
    {
      "epoch": 3.754337102781786,
      "grad_norm": 0.703125,
      "learning_rate": 7.27462214549883e-06,
      "loss": 0.8895,
      "step": 6993
    },
    {
      "epoch": 3.754873997516862,
      "grad_norm": 0.81640625,
      "learning_rate": 7.268667699858961e-06,
      "loss": 0.9168,
      "step": 6994
    },
    {
      "epoch": 3.7554108922519376,
      "grad_norm": 0.7265625,
      "learning_rate": 7.262715277588655e-06,
      "loss": 0.745,
      "step": 6995
    },
    {
      "epoch": 3.7559477869870137,
      "grad_norm": 0.71484375,
      "learning_rate": 7.25676487936717e-06,
      "loss": 1.0201,
      "step": 6996
    },
    {
      "epoch": 3.75648468172209,
      "grad_norm": 0.765625,
      "learning_rate": 7.250816505873504e-06,
      "loss": 0.9542,
      "step": 6997
    },
    {
      "epoch": 3.757021576457166,
      "grad_norm": 0.68359375,
      "learning_rate": 7.244870157786454e-06,
      "loss": 0.8022,
      "step": 6998
    },
    {
      "epoch": 3.757558471192242,
      "grad_norm": 0.7265625,
      "learning_rate": 7.238925835784566e-06,
      "loss": 1.0832,
      "step": 6999
    },
    {
      "epoch": 3.7580953659273177,
      "grad_norm": 0.82421875,
      "learning_rate": 7.232983540546173e-06,
      "loss": 0.895,
      "step": 7000
    },
    {
      "epoch": 3.7580953659273177,
      "eval_loss": 0.9025610089302063,
      "eval_runtime": 46.3144,
      "eval_samples_per_second": 20.771,
      "eval_steps_per_second": 20.771,
      "step": 7000
    },
    {
      "epoch": 3.758632260662394,
      "grad_norm": 0.6953125,
      "learning_rate": 7.227043272749351e-06,
      "loss": 0.7568,
      "step": 7001
    },
    {
      "epoch": 3.75916915539747,
      "grad_norm": 0.71875,
      "learning_rate": 7.2211050330719674e-06,
      "loss": 0.9158,
      "step": 7002
    },
    {
      "epoch": 3.7597060501325457,
      "grad_norm": 0.65625,
      "learning_rate": 7.215168822191659e-06,
      "loss": 0.6484,
      "step": 7003
    },
    {
      "epoch": 3.760242944867622,
      "grad_norm": 0.80078125,
      "learning_rate": 7.209234640785806e-06,
      "loss": 1.013,
      "step": 7004
    },
    {
      "epoch": 3.760779839602698,
      "grad_norm": 0.6796875,
      "learning_rate": 7.203302489531583e-06,
      "loss": 0.6657,
      "step": 7005
    },
    {
      "epoch": 3.761316734337774,
      "grad_norm": 0.7265625,
      "learning_rate": 7.19737236910594e-06,
      "loss": 0.8123,
      "step": 7006
    },
    {
      "epoch": 3.76185362907285,
      "grad_norm": 0.73828125,
      "learning_rate": 7.191444280185544e-06,
      "loss": 0.8483,
      "step": 7007
    },
    {
      "epoch": 3.762390523807926,
      "grad_norm": 0.58203125,
      "learning_rate": 7.1855182234468825e-06,
      "loss": 0.9083,
      "step": 7008
    },
    {
      "epoch": 3.762927418543002,
      "grad_norm": 0.69140625,
      "learning_rate": 7.179594199566203e-06,
      "loss": 1.0029,
      "step": 7009
    },
    {
      "epoch": 3.763464313278078,
      "grad_norm": 0.859375,
      "learning_rate": 7.173672209219495e-06,
      "loss": 0.8219,
      "step": 7010
    },
    {
      "epoch": 3.7640012080131537,
      "grad_norm": 0.67578125,
      "learning_rate": 7.167752253082538e-06,
      "loss": 0.8656,
      "step": 7011
    },
    {
      "epoch": 3.76453810274823,
      "grad_norm": 0.74609375,
      "learning_rate": 7.161834331830886e-06,
      "loss": 0.7697,
      "step": 7012
    },
    {
      "epoch": 3.765074997483306,
      "grad_norm": 0.66015625,
      "learning_rate": 7.155918446139831e-06,
      "loss": 0.7004,
      "step": 7013
    },
    {
      "epoch": 3.7656118922183817,
      "grad_norm": 0.73828125,
      "learning_rate": 7.150004596684456e-06,
      "loss": 0.8632,
      "step": 7014
    },
    {
      "epoch": 3.766148786953458,
      "grad_norm": 0.7578125,
      "learning_rate": 7.144092784139619e-06,
      "loss": 0.856,
      "step": 7015
    },
    {
      "epoch": 3.766685681688534,
      "grad_norm": 0.703125,
      "learning_rate": 7.138183009179922e-06,
      "loss": 0.7863,
      "step": 7016
    },
    {
      "epoch": 3.76722257642361,
      "grad_norm": 0.7109375,
      "learning_rate": 7.132275272479735e-06,
      "loss": 0.8785,
      "step": 7017
    },
    {
      "epoch": 3.767759471158686,
      "grad_norm": 0.62890625,
      "learning_rate": 7.126369574713218e-06,
      "loss": 0.8522,
      "step": 7018
    },
    {
      "epoch": 3.768296365893762,
      "grad_norm": 0.734375,
      "learning_rate": 7.120465916554292e-06,
      "loss": 0.8985,
      "step": 7019
    },
    {
      "epoch": 3.768833260628838,
      "grad_norm": 0.73828125,
      "learning_rate": 7.114564298676621e-06,
      "loss": 0.8856,
      "step": 7020
    },
    {
      "epoch": 3.769370155363914,
      "grad_norm": 0.9296875,
      "learning_rate": 7.108664721753666e-06,
      "loss": 0.8923,
      "step": 7021
    },
    {
      "epoch": 3.7699070500989897,
      "grad_norm": 0.74609375,
      "learning_rate": 7.102767186458647e-06,
      "loss": 0.6679,
      "step": 7022
    },
    {
      "epoch": 3.770443944834066,
      "grad_norm": 0.6640625,
      "learning_rate": 7.096871693464544e-06,
      "loss": 0.7724,
      "step": 7023
    },
    {
      "epoch": 3.770980839569142,
      "grad_norm": 0.73046875,
      "learning_rate": 7.090978243444099e-06,
      "loss": 0.7654,
      "step": 7024
    },
    {
      "epoch": 3.771517734304218,
      "grad_norm": 0.66796875,
      "learning_rate": 7.08508683706984e-06,
      "loss": 0.7234,
      "step": 7025
    },
    {
      "epoch": 3.7720546290392942,
      "grad_norm": 0.734375,
      "learning_rate": 7.079197475014043e-06,
      "loss": 0.6981,
      "step": 7026
    },
    {
      "epoch": 3.77259152377437,
      "grad_norm": 0.70703125,
      "learning_rate": 7.073310157948759e-06,
      "loss": 0.7775,
      "step": 7027
    },
    {
      "epoch": 3.773128418509446,
      "grad_norm": 0.58984375,
      "learning_rate": 7.0674248865458206e-06,
      "loss": 0.6624,
      "step": 7028
    },
    {
      "epoch": 3.773665313244522,
      "grad_norm": 0.6640625,
      "learning_rate": 7.061541661476792e-06,
      "loss": 0.8177,
      "step": 7029
    },
    {
      "epoch": 3.774202207979598,
      "grad_norm": 0.74609375,
      "learning_rate": 7.055660483413029e-06,
      "loss": 0.9344,
      "step": 7030
    },
    {
      "epoch": 3.774739102714674,
      "grad_norm": 0.6875,
      "learning_rate": 7.049781353025661e-06,
      "loss": 0.7877,
      "step": 7031
    },
    {
      "epoch": 3.77527599744975,
      "grad_norm": 0.65234375,
      "learning_rate": 7.043904270985563e-06,
      "loss": 0.7633,
      "step": 7032
    },
    {
      "epoch": 3.7758128921848257,
      "grad_norm": 0.71875,
      "learning_rate": 7.038029237963376e-06,
      "loss": 1.0485,
      "step": 7033
    },
    {
      "epoch": 3.776349786919902,
      "grad_norm": 0.81640625,
      "learning_rate": 7.032156254629527e-06,
      "loss": 0.9159,
      "step": 7034
    },
    {
      "epoch": 3.776886681654978,
      "grad_norm": 0.55078125,
      "learning_rate": 7.0262853216541885e-06,
      "loss": 0.6439,
      "step": 7035
    },
    {
      "epoch": 3.777423576390054,
      "grad_norm": 0.671875,
      "learning_rate": 7.020416439707314e-06,
      "loss": 0.955,
      "step": 7036
    },
    {
      "epoch": 3.7779604711251302,
      "grad_norm": 0.74609375,
      "learning_rate": 7.014549609458624e-06,
      "loss": 1.1313,
      "step": 7037
    },
    {
      "epoch": 3.778497365860206,
      "grad_norm": 0.6484375,
      "learning_rate": 7.008684831577581e-06,
      "loss": 0.7717,
      "step": 7038
    },
    {
      "epoch": 3.779034260595282,
      "grad_norm": 0.69921875,
      "learning_rate": 7.002822106733442e-06,
      "loss": 0.7269,
      "step": 7039
    },
    {
      "epoch": 3.779571155330358,
      "grad_norm": 0.6328125,
      "learning_rate": 6.996961435595223e-06,
      "loss": 0.9413,
      "step": 7040
    },
    {
      "epoch": 3.780108050065434,
      "grad_norm": 0.76953125,
      "learning_rate": 6.991102818831696e-06,
      "loss": 0.9704,
      "step": 7041
    },
    {
      "epoch": 3.78064494480051,
      "grad_norm": 0.76953125,
      "learning_rate": 6.985246257111394e-06,
      "loss": 1.008,
      "step": 7042
    },
    {
      "epoch": 3.781181839535586,
      "grad_norm": 0.63671875,
      "learning_rate": 6.9793917511026315e-06,
      "loss": 0.8338,
      "step": 7043
    },
    {
      "epoch": 3.781718734270662,
      "grad_norm": 0.65625,
      "learning_rate": 6.973539301473489e-06,
      "loss": 0.7488,
      "step": 7044
    },
    {
      "epoch": 3.7822556290057383,
      "grad_norm": 0.73828125,
      "learning_rate": 6.967688908891792e-06,
      "loss": 0.8088,
      "step": 7045
    },
    {
      "epoch": 3.782792523740814,
      "grad_norm": 1.15625,
      "learning_rate": 6.961840574025153e-06,
      "loss": 1.0319,
      "step": 7046
    },
    {
      "epoch": 3.78332941847589,
      "grad_norm": 0.83984375,
      "learning_rate": 6.9559942975409465e-06,
      "loss": 0.8648,
      "step": 7047
    },
    {
      "epoch": 3.7838663132109662,
      "grad_norm": 0.58203125,
      "learning_rate": 6.950150080106299e-06,
      "loss": 0.6366,
      "step": 7048
    },
    {
      "epoch": 3.784403207946042,
      "grad_norm": 0.8359375,
      "learning_rate": 6.9443079223881015e-06,
      "loss": 0.8789,
      "step": 7049
    },
    {
      "epoch": 3.784940102681118,
      "grad_norm": 0.6796875,
      "learning_rate": 6.9384678250530355e-06,
      "loss": 0.8497,
      "step": 7050
    },
    {
      "epoch": 3.784940102681118,
      "eval_loss": 0.9027681946754456,
      "eval_runtime": 46.2488,
      "eval_samples_per_second": 20.801,
      "eval_steps_per_second": 20.801,
      "step": 7050
    },
    {
      "epoch": 3.785476997416194,
      "grad_norm": 1.046875,
      "learning_rate": 6.932629788767514e-06,
      "loss": 0.7524,
      "step": 7051
    },
    {
      "epoch": 3.78601389215127,
      "grad_norm": 0.7890625,
      "learning_rate": 6.926793814197741e-06,
      "loss": 0.913,
      "step": 7052
    },
    {
      "epoch": 3.786550786886346,
      "grad_norm": 0.66015625,
      "learning_rate": 6.920959902009675e-06,
      "loss": 0.8766,
      "step": 7053
    },
    {
      "epoch": 3.787087681621422,
      "grad_norm": 0.76953125,
      "learning_rate": 6.915128052869033e-06,
      "loss": 0.9065,
      "step": 7054
    },
    {
      "epoch": 3.787624576356498,
      "grad_norm": 0.65625,
      "learning_rate": 6.909298267441302e-06,
      "loss": 0.7086,
      "step": 7055
    },
    {
      "epoch": 3.7881614710915743,
      "grad_norm": 0.66796875,
      "learning_rate": 6.9034705463917525e-06,
      "loss": 0.8337,
      "step": 7056
    },
    {
      "epoch": 3.78869836582665,
      "grad_norm": 0.7265625,
      "learning_rate": 6.897644890385382e-06,
      "loss": 0.8837,
      "step": 7057
    },
    {
      "epoch": 3.789235260561726,
      "grad_norm": 0.8203125,
      "learning_rate": 6.891821300086973e-06,
      "loss": 0.8759,
      "step": 7058
    },
    {
      "epoch": 3.7897721552968022,
      "grad_norm": 0.70703125,
      "learning_rate": 6.885999776161076e-06,
      "loss": 1.0047,
      "step": 7059
    },
    {
      "epoch": 3.790309050031878,
      "grad_norm": 0.6171875,
      "learning_rate": 6.880180319272006e-06,
      "loss": 0.7059,
      "step": 7060
    },
    {
      "epoch": 3.790845944766954,
      "grad_norm": 0.68359375,
      "learning_rate": 6.874362930083822e-06,
      "loss": 0.8963,
      "step": 7061
    },
    {
      "epoch": 3.79138283950203,
      "grad_norm": 0.60546875,
      "learning_rate": 6.86854760926037e-06,
      "loss": 0.7088,
      "step": 7062
    },
    {
      "epoch": 3.7919197342371063,
      "grad_norm": 0.5703125,
      "learning_rate": 6.862734357465261e-06,
      "loss": 0.6758,
      "step": 7063
    },
    {
      "epoch": 3.7924566289721824,
      "grad_norm": 0.76953125,
      "learning_rate": 6.8569231753618426e-06,
      "loss": 0.9725,
      "step": 7064
    },
    {
      "epoch": 3.792993523707258,
      "grad_norm": 0.64453125,
      "learning_rate": 6.851114063613265e-06,
      "loss": 0.6546,
      "step": 7065
    },
    {
      "epoch": 3.793530418442334,
      "grad_norm": 0.6875,
      "learning_rate": 6.845307022882405e-06,
      "loss": 0.8853,
      "step": 7066
    },
    {
      "epoch": 3.7940673131774103,
      "grad_norm": 0.6640625,
      "learning_rate": 6.83950205383192e-06,
      "loss": 0.7502,
      "step": 7067
    },
    {
      "epoch": 3.794604207912486,
      "grad_norm": 0.64453125,
      "learning_rate": 6.833699157124232e-06,
      "loss": 1.0217,
      "step": 7068
    },
    {
      "epoch": 3.795141102647562,
      "grad_norm": 0.7265625,
      "learning_rate": 6.827898333421537e-06,
      "loss": 0.9204,
      "step": 7069
    },
    {
      "epoch": 3.7956779973826382,
      "grad_norm": 0.82421875,
      "learning_rate": 6.822099583385766e-06,
      "loss": 0.9674,
      "step": 7070
    },
    {
      "epoch": 3.796214892117714,
      "grad_norm": 0.75390625,
      "learning_rate": 6.816302907678638e-06,
      "loss": 0.9384,
      "step": 7071
    },
    {
      "epoch": 3.79675178685279,
      "grad_norm": 0.86328125,
      "learning_rate": 6.81050830696163e-06,
      "loss": 0.9101,
      "step": 7072
    },
    {
      "epoch": 3.797288681587866,
      "grad_norm": 0.6953125,
      "learning_rate": 6.804715781895971e-06,
      "loss": 0.8452,
      "step": 7073
    },
    {
      "epoch": 3.7978255763229423,
      "grad_norm": 0.6640625,
      "learning_rate": 6.798925333142672e-06,
      "loss": 0.8919,
      "step": 7074
    },
    {
      "epoch": 3.7983624710580184,
      "grad_norm": 0.7109375,
      "learning_rate": 6.793136961362489e-06,
      "loss": 0.8072,
      "step": 7075
    },
    {
      "epoch": 3.798899365793094,
      "grad_norm": 0.578125,
      "learning_rate": 6.787350667215947e-06,
      "loss": 0.7227,
      "step": 7076
    },
    {
      "epoch": 3.79943626052817,
      "grad_norm": 0.68359375,
      "learning_rate": 6.781566451363336e-06,
      "loss": 0.8098,
      "step": 7077
    },
    {
      "epoch": 3.7999731552632463,
      "grad_norm": 0.7265625,
      "learning_rate": 6.775784314464717e-06,
      "loss": 1.0118,
      "step": 7078
    },
    {
      "epoch": 3.800510049998322,
      "grad_norm": 1.09375,
      "learning_rate": 6.770004257179893e-06,
      "loss": 1.1284,
      "step": 7079
    },
    {
      "epoch": 3.801046944733398,
      "grad_norm": 0.6484375,
      "learning_rate": 6.764226280168445e-06,
      "loss": 0.7417,
      "step": 7080
    },
    {
      "epoch": 3.8015838394684742,
      "grad_norm": 0.63671875,
      "learning_rate": 6.758450384089726e-06,
      "loss": 0.8914,
      "step": 7081
    },
    {
      "epoch": 3.8021207342035503,
      "grad_norm": 0.73046875,
      "learning_rate": 6.752676569602828e-06,
      "loss": 0.8329,
      "step": 7082
    },
    {
      "epoch": 3.8026576289386265,
      "grad_norm": 0.671875,
      "learning_rate": 6.746904837366608e-06,
      "loss": 0.7742,
      "step": 7083
    },
    {
      "epoch": 3.803194523673702,
      "grad_norm": 0.625,
      "learning_rate": 6.7411351880397045e-06,
      "loss": 0.8232,
      "step": 7084
    },
    {
      "epoch": 3.8037314184087783,
      "grad_norm": 0.6875,
      "learning_rate": 6.735367622280514e-06,
      "loss": 0.9152,
      "step": 7085
    },
    {
      "epoch": 3.8042683131438544,
      "grad_norm": 0.6484375,
      "learning_rate": 6.729602140747171e-06,
      "loss": 0.7321,
      "step": 7086
    },
    {
      "epoch": 3.80480520787893,
      "grad_norm": 0.57421875,
      "learning_rate": 6.723838744097602e-06,
      "loss": 0.7819,
      "step": 7087
    },
    {
      "epoch": 3.805342102614006,
      "grad_norm": 0.6796875,
      "learning_rate": 6.7180774329894865e-06,
      "loss": 0.7493,
      "step": 7088
    },
    {
      "epoch": 3.8058789973490823,
      "grad_norm": 0.95703125,
      "learning_rate": 6.712318208080251e-06,
      "loss": 0.8014,
      "step": 7089
    },
    {
      "epoch": 3.806415892084158,
      "grad_norm": 0.70703125,
      "learning_rate": 6.706561070027109e-06,
      "loss": 1.0279,
      "step": 7090
    },
    {
      "epoch": 3.806952786819234,
      "grad_norm": 0.75,
      "learning_rate": 6.700806019487019e-06,
      "loss": 0.9602,
      "step": 7091
    },
    {
      "epoch": 3.8074896815543102,
      "grad_norm": 0.69921875,
      "learning_rate": 6.695053057116693e-06,
      "loss": 0.8572,
      "step": 7092
    },
    {
      "epoch": 3.8080265762893863,
      "grad_norm": 0.8203125,
      "learning_rate": 6.6893021835726295e-06,
      "loss": 0.8794,
      "step": 7093
    },
    {
      "epoch": 3.8085634710244625,
      "grad_norm": 0.640625,
      "learning_rate": 6.6835533995110775e-06,
      "loss": 0.7571,
      "step": 7094
    },
    {
      "epoch": 3.809100365759538,
      "grad_norm": 0.6484375,
      "learning_rate": 6.677806705588036e-06,
      "loss": 0.9336,
      "step": 7095
    },
    {
      "epoch": 3.8096372604946143,
      "grad_norm": 0.73046875,
      "learning_rate": 6.672062102459281e-06,
      "loss": 0.9442,
      "step": 7096
    },
    {
      "epoch": 3.8101741552296904,
      "grad_norm": 0.765625,
      "learning_rate": 6.666319590780351e-06,
      "loss": 0.643,
      "step": 7097
    },
    {
      "epoch": 3.810711049964766,
      "grad_norm": 0.6171875,
      "learning_rate": 6.660579171206527e-06,
      "loss": 0.7477,
      "step": 7098
    },
    {
      "epoch": 3.811247944699842,
      "grad_norm": 0.66796875,
      "learning_rate": 6.6548408443928765e-06,
      "loss": 0.9367,
      "step": 7099
    },
    {
      "epoch": 3.8117848394349183,
      "grad_norm": 0.7890625,
      "learning_rate": 6.6491046109942e-06,
      "loss": 0.9445,
      "step": 7100
    },
    {
      "epoch": 3.8117848394349183,
      "eval_loss": 0.9026427268981934,
      "eval_runtime": 46.0333,
      "eval_samples_per_second": 20.898,
      "eval_steps_per_second": 20.898,
      "step": 7100
    },
    {
      "epoch": 3.8123217341699944,
      "grad_norm": 0.65625,
      "learning_rate": 6.643370471665092e-06,
      "loss": 0.7945,
      "step": 7101
    },
    {
      "epoch": 3.8128586289050705,
      "grad_norm": 0.6796875,
      "learning_rate": 6.6376384270598695e-06,
      "loss": 0.7124,
      "step": 7102
    },
    {
      "epoch": 3.8133955236401462,
      "grad_norm": 0.6015625,
      "learning_rate": 6.631908477832646e-06,
      "loss": 0.6983,
      "step": 7103
    },
    {
      "epoch": 3.8139324183752223,
      "grad_norm": 0.68359375,
      "learning_rate": 6.6261806246372875e-06,
      "loss": 0.8853,
      "step": 7104
    },
    {
      "epoch": 3.8144693131102985,
      "grad_norm": 0.80078125,
      "learning_rate": 6.620454868127396e-06,
      "loss": 1.0888,
      "step": 7105
    },
    {
      "epoch": 3.815006207845374,
      "grad_norm": 0.69921875,
      "learning_rate": 6.6147312089563596e-06,
      "loss": 0.7172,
      "step": 7106
    },
    {
      "epoch": 3.8155431025804503,
      "grad_norm": 0.76953125,
      "learning_rate": 6.609009647777342e-06,
      "loss": 0.768,
      "step": 7107
    },
    {
      "epoch": 3.8160799973155264,
      "grad_norm": 0.67578125,
      "learning_rate": 6.6032901852432105e-06,
      "loss": 0.9998,
      "step": 7108
    },
    {
      "epoch": 3.816616892050602,
      "grad_norm": 0.7890625,
      "learning_rate": 6.5975728220066425e-06,
      "loss": 0.9766,
      "step": 7109
    },
    {
      "epoch": 3.817153786785678,
      "grad_norm": 1.2734375,
      "learning_rate": 6.591857558720071e-06,
      "loss": 1.0564,
      "step": 7110
    },
    {
      "epoch": 3.8176906815207543,
      "grad_norm": 0.62890625,
      "learning_rate": 6.586144396035668e-06,
      "loss": 0.6909,
      "step": 7111
    },
    {
      "epoch": 3.8182275762558304,
      "grad_norm": 0.640625,
      "learning_rate": 6.580433334605379e-06,
      "loss": 0.822,
      "step": 7112
    },
    {
      "epoch": 3.8187644709909065,
      "grad_norm": 0.7421875,
      "learning_rate": 6.57472437508092e-06,
      "loss": 1.0486,
      "step": 7113
    },
    {
      "epoch": 3.8193013657259822,
      "grad_norm": 0.68359375,
      "learning_rate": 6.56901751811374e-06,
      "loss": 0.7288,
      "step": 7114
    },
    {
      "epoch": 3.8198382604610583,
      "grad_norm": 0.78125,
      "learning_rate": 6.56331276435507e-06,
      "loss": 0.9105,
      "step": 7115
    },
    {
      "epoch": 3.8203751551961345,
      "grad_norm": 0.69921875,
      "learning_rate": 6.557610114455903e-06,
      "loss": 0.8362,
      "step": 7116
    },
    {
      "epoch": 3.82091204993121,
      "grad_norm": 0.82421875,
      "learning_rate": 6.551909569066974e-06,
      "loss": 0.9469,
      "step": 7117
    },
    {
      "epoch": 3.8214489446662863,
      "grad_norm": 0.75390625,
      "learning_rate": 6.546211128838786e-06,
      "loss": 0.9123,
      "step": 7118
    },
    {
      "epoch": 3.8219858394013624,
      "grad_norm": 0.671875,
      "learning_rate": 6.540514794421612e-06,
      "loss": 1.0064,
      "step": 7119
    },
    {
      "epoch": 3.8225227341364385,
      "grad_norm": 0.62890625,
      "learning_rate": 6.534820566465463e-06,
      "loss": 0.7518,
      "step": 7120
    },
    {
      "epoch": 3.8230596288715146,
      "grad_norm": 0.7265625,
      "learning_rate": 6.529128445620128e-06,
      "loss": 0.7842,
      "step": 7121
    },
    {
      "epoch": 3.8235965236065903,
      "grad_norm": 0.64453125,
      "learning_rate": 6.523438432535162e-06,
      "loss": 0.8895,
      "step": 7122
    },
    {
      "epoch": 3.8241334183416664,
      "grad_norm": 0.70703125,
      "learning_rate": 6.5177505278598506e-06,
      "loss": 0.8676,
      "step": 7123
    },
    {
      "epoch": 3.8246703130767425,
      "grad_norm": 0.7421875,
      "learning_rate": 6.512064732243267e-06,
      "loss": 0.856,
      "step": 7124
    },
    {
      "epoch": 3.825207207811818,
      "grad_norm": 0.6015625,
      "learning_rate": 6.506381046334223e-06,
      "loss": 0.6957,
      "step": 7125
    },
    {
      "epoch": 3.8257441025468943,
      "grad_norm": 0.6640625,
      "learning_rate": 6.50069947078131e-06,
      "loss": 0.8614,
      "step": 7126
    },
    {
      "epoch": 3.8262809972819705,
      "grad_norm": 0.6875,
      "learning_rate": 6.495020006232857e-06,
      "loss": 0.8632,
      "step": 7127
    },
    {
      "epoch": 3.826817892017046,
      "grad_norm": 0.671875,
      "learning_rate": 6.489342653336966e-06,
      "loss": 0.9827,
      "step": 7128
    },
    {
      "epoch": 3.8273547867521223,
      "grad_norm": 0.875,
      "learning_rate": 6.4836674127415064e-06,
      "loss": 0.8821,
      "step": 7129
    },
    {
      "epoch": 3.8278916814871984,
      "grad_norm": 0.63671875,
      "learning_rate": 6.477994285094077e-06,
      "loss": 0.8037,
      "step": 7130
    },
    {
      "epoch": 3.8284285762222745,
      "grad_norm": 0.58984375,
      "learning_rate": 6.47232327104206e-06,
      "loss": 0.8041,
      "step": 7131
    },
    {
      "epoch": 3.8289654709573506,
      "grad_norm": 0.7421875,
      "learning_rate": 6.466654371232608e-06,
      "loss": 0.7116,
      "step": 7132
    },
    {
      "epoch": 3.8295023656924263,
      "grad_norm": 0.6328125,
      "learning_rate": 6.460987586312587e-06,
      "loss": 0.7953,
      "step": 7133
    },
    {
      "epoch": 3.8300392604275024,
      "grad_norm": 0.734375,
      "learning_rate": 6.455322916928655e-06,
      "loss": 0.7814,
      "step": 7134
    },
    {
      "epoch": 3.8305761551625785,
      "grad_norm": 0.6328125,
      "learning_rate": 6.449660363727236e-06,
      "loss": 0.9174,
      "step": 7135
    },
    {
      "epoch": 3.831113049897654,
      "grad_norm": 0.703125,
      "learning_rate": 6.443999927354482e-06,
      "loss": 0.9298,
      "step": 7136
    },
    {
      "epoch": 3.8316499446327303,
      "grad_norm": 0.60546875,
      "learning_rate": 6.438341608456333e-06,
      "loss": 0.7381,
      "step": 7137
    },
    {
      "epoch": 3.8321868393678065,
      "grad_norm": 0.75,
      "learning_rate": 6.432685407678474e-06,
      "loss": 0.7947,
      "step": 7138
    },
    {
      "epoch": 3.8327237341028826,
      "grad_norm": 0.75,
      "learning_rate": 6.427031325666341e-06,
      "loss": 0.8098,
      "step": 7139
    },
    {
      "epoch": 3.8332606288379587,
      "grad_norm": 0.734375,
      "learning_rate": 6.421379363065142e-06,
      "loss": 0.9963,
      "step": 7140
    },
    {
      "epoch": 3.8337975235730344,
      "grad_norm": 0.6640625,
      "learning_rate": 6.415729520519842e-06,
      "loss": 0.7359,
      "step": 7141
    },
    {
      "epoch": 3.8343344183081105,
      "grad_norm": 0.65625,
      "learning_rate": 6.410081798675155e-06,
      "loss": 0.7311,
      "step": 7142
    },
    {
      "epoch": 3.8348713130431866,
      "grad_norm": 0.7734375,
      "learning_rate": 6.40443619817555e-06,
      "loss": 1.0181,
      "step": 7143
    },
    {
      "epoch": 3.8354082077782623,
      "grad_norm": 0.70703125,
      "learning_rate": 6.39879271966527e-06,
      "loss": 0.8381,
      "step": 7144
    },
    {
      "epoch": 3.8359451025133384,
      "grad_norm": 0.69921875,
      "learning_rate": 6.393151363788311e-06,
      "loss": 0.9254,
      "step": 7145
    },
    {
      "epoch": 3.8364819972484145,
      "grad_norm": 0.64453125,
      "learning_rate": 6.387512131188411e-06,
      "loss": 0.9559,
      "step": 7146
    },
    {
      "epoch": 3.83701889198349,
      "grad_norm": 0.5859375,
      "learning_rate": 6.381875022509087e-06,
      "loss": 0.5784,
      "step": 7147
    },
    {
      "epoch": 3.8375557867185663,
      "grad_norm": 0.609375,
      "learning_rate": 6.3762400383936075e-06,
      "loss": 0.8213,
      "step": 7148
    },
    {
      "epoch": 3.8380926814536425,
      "grad_norm": 0.71484375,
      "learning_rate": 6.370607179484991e-06,
      "loss": 0.9033,
      "step": 7149
    },
    {
      "epoch": 3.8386295761887186,
      "grad_norm": 0.640625,
      "learning_rate": 6.3649764464260105e-06,
      "loss": 0.7153,
      "step": 7150
    },
    {
      "epoch": 3.8386295761887186,
      "eval_loss": 0.9026381969451904,
      "eval_runtime": 47.0361,
      "eval_samples_per_second": 20.452,
      "eval_steps_per_second": 20.452,
      "step": 7150
    },
    {
      "epoch": 3.8391664709237947,
      "grad_norm": 0.6171875,
      "learning_rate": 6.35934783985922e-06,
      "loss": 0.7916,
      "step": 7151
    },
    {
      "epoch": 3.8397033656588704,
      "grad_norm": 0.734375,
      "learning_rate": 6.3537213604268955e-06,
      "loss": 1.0401,
      "step": 7152
    },
    {
      "epoch": 3.8402402603939465,
      "grad_norm": 0.75390625,
      "learning_rate": 6.3480970087711e-06,
      "loss": 0.8586,
      "step": 7153
    },
    {
      "epoch": 3.8407771551290226,
      "grad_norm": 0.61328125,
      "learning_rate": 6.342474785533653e-06,
      "loss": 0.6723,
      "step": 7154
    },
    {
      "epoch": 3.8413140498640983,
      "grad_norm": 0.640625,
      "learning_rate": 6.336854691356103e-06,
      "loss": 0.8799,
      "step": 7155
    },
    {
      "epoch": 3.8418509445991744,
      "grad_norm": 0.74609375,
      "learning_rate": 6.33123672687978e-06,
      "loss": 0.9772,
      "step": 7156
    },
    {
      "epoch": 3.8423878393342505,
      "grad_norm": 0.67578125,
      "learning_rate": 6.325620892745776e-06,
      "loss": 0.8332,
      "step": 7157
    },
    {
      "epoch": 3.8429247340693267,
      "grad_norm": 0.6015625,
      "learning_rate": 6.3200071895949206e-06,
      "loss": 0.8175,
      "step": 7158
    },
    {
      "epoch": 3.843461628804403,
      "grad_norm": 0.7578125,
      "learning_rate": 6.314395618067801e-06,
      "loss": 0.8394,
      "step": 7159
    },
    {
      "epoch": 3.8439985235394785,
      "grad_norm": 0.64453125,
      "learning_rate": 6.308786178804782e-06,
      "loss": 0.9289,
      "step": 7160
    },
    {
      "epoch": 3.8445354182745546,
      "grad_norm": 0.80078125,
      "learning_rate": 6.303178872445958e-06,
      "loss": 0.9521,
      "step": 7161
    },
    {
      "epoch": 3.8450723130096307,
      "grad_norm": 0.56640625,
      "learning_rate": 6.297573699631201e-06,
      "loss": 0.8274,
      "step": 7162
    },
    {
      "epoch": 3.8456092077447064,
      "grad_norm": 0.73828125,
      "learning_rate": 6.2919706610001375e-06,
      "loss": 0.947,
      "step": 7163
    },
    {
      "epoch": 3.8461461024797825,
      "grad_norm": 0.64453125,
      "learning_rate": 6.286369757192134e-06,
      "loss": 0.8731,
      "step": 7164
    },
    {
      "epoch": 3.8466829972148586,
      "grad_norm": 0.65234375,
      "learning_rate": 6.280770988846329e-06,
      "loss": 0.7913,
      "step": 7165
    },
    {
      "epoch": 3.8472198919499343,
      "grad_norm": 0.6640625,
      "learning_rate": 6.275174356601623e-06,
      "loss": 0.8454,
      "step": 7166
    },
    {
      "epoch": 3.8477567866850104,
      "grad_norm": 0.765625,
      "learning_rate": 6.269579861096653e-06,
      "loss": 0.814,
      "step": 7167
    },
    {
      "epoch": 3.8482936814200865,
      "grad_norm": 0.74609375,
      "learning_rate": 6.263987502969815e-06,
      "loss": 1.1018,
      "step": 7168
    },
    {
      "epoch": 3.8488305761551627,
      "grad_norm": 0.62890625,
      "learning_rate": 6.2583972828592744e-06,
      "loss": 0.7504,
      "step": 7169
    },
    {
      "epoch": 3.849367470890239,
      "grad_norm": 0.65625,
      "learning_rate": 6.252809201402953e-06,
      "loss": 0.7289,
      "step": 7170
    },
    {
      "epoch": 3.8499043656253145,
      "grad_norm": 0.6796875,
      "learning_rate": 6.247223259238511e-06,
      "loss": 0.7782,
      "step": 7171
    },
    {
      "epoch": 3.8504412603603906,
      "grad_norm": 0.6484375,
      "learning_rate": 6.241639457003379e-06,
      "loss": 0.7569,
      "step": 7172
    },
    {
      "epoch": 3.8509781550954667,
      "grad_norm": 0.7421875,
      "learning_rate": 6.236057795334749e-06,
      "loss": 0.8635,
      "step": 7173
    },
    {
      "epoch": 3.8515150498305424,
      "grad_norm": 0.609375,
      "learning_rate": 6.23047827486955e-06,
      "loss": 0.8061,
      "step": 7174
    },
    {
      "epoch": 3.8520519445656185,
      "grad_norm": 0.69140625,
      "learning_rate": 6.2249008962444674e-06,
      "loss": 0.9643,
      "step": 7175
    },
    {
      "epoch": 3.8525888393006946,
      "grad_norm": 0.6015625,
      "learning_rate": 6.219325660095971e-06,
      "loss": 0.6542,
      "step": 7176
    },
    {
      "epoch": 3.8531257340357707,
      "grad_norm": 0.609375,
      "learning_rate": 6.213752567060244e-06,
      "loss": 0.7527,
      "step": 7177
    },
    {
      "epoch": 3.853662628770847,
      "grad_norm": 0.71875,
      "learning_rate": 6.20818161777326e-06,
      "loss": 0.8118,
      "step": 7178
    },
    {
      "epoch": 3.8541995235059225,
      "grad_norm": 0.66796875,
      "learning_rate": 6.202612812870736e-06,
      "loss": 0.7214,
      "step": 7179
    },
    {
      "epoch": 3.8547364182409987,
      "grad_norm": 0.6796875,
      "learning_rate": 6.197046152988137e-06,
      "loss": 0.7413,
      "step": 7180
    },
    {
      "epoch": 3.855273312976075,
      "grad_norm": 0.83203125,
      "learning_rate": 6.19148163876069e-06,
      "loss": 0.8139,
      "step": 7181
    },
    {
      "epoch": 3.8558102077111505,
      "grad_norm": 0.61328125,
      "learning_rate": 6.185919270823384e-06,
      "loss": 0.8034,
      "step": 7182
    },
    {
      "epoch": 3.8563471024462266,
      "grad_norm": 0.609375,
      "learning_rate": 6.180359049810955e-06,
      "loss": 0.8297,
      "step": 7183
    },
    {
      "epoch": 3.8568839971813027,
      "grad_norm": 0.6171875,
      "learning_rate": 6.174800976357878e-06,
      "loss": 0.6517,
      "step": 7184
    },
    {
      "epoch": 3.8574208919163784,
      "grad_norm": 0.734375,
      "learning_rate": 6.169245051098413e-06,
      "loss": 0.9675,
      "step": 7185
    },
    {
      "epoch": 3.8579577866514545,
      "grad_norm": 0.68359375,
      "learning_rate": 6.1636912746665695e-06,
      "loss": 0.7807,
      "step": 7186
    },
    {
      "epoch": 3.8584946813865306,
      "grad_norm": 0.890625,
      "learning_rate": 6.158139647696085e-06,
      "loss": 1.0128,
      "step": 7187
    },
    {
      "epoch": 3.8590315761216067,
      "grad_norm": 0.75,
      "learning_rate": 6.15259017082048e-06,
      "loss": 0.8042,
      "step": 7188
    },
    {
      "epoch": 3.859568470856683,
      "grad_norm": 0.81640625,
      "learning_rate": 6.147042844673026e-06,
      "loss": 0.9115,
      "step": 7189
    },
    {
      "epoch": 3.8601053655917585,
      "grad_norm": 0.57421875,
      "learning_rate": 6.141497669886729e-06,
      "loss": 0.6325,
      "step": 7190
    },
    {
      "epoch": 3.8606422603268347,
      "grad_norm": 0.7109375,
      "learning_rate": 6.13595464709438e-06,
      "loss": 0.8846,
      "step": 7191
    },
    {
      "epoch": 3.861179155061911,
      "grad_norm": 0.76171875,
      "learning_rate": 6.130413776928498e-06,
      "loss": 0.8668,
      "step": 7192
    },
    {
      "epoch": 3.8617160497969865,
      "grad_norm": 0.59375,
      "learning_rate": 6.124875060021362e-06,
      "loss": 0.7045,
      "step": 7193
    },
    {
      "epoch": 3.8622529445320626,
      "grad_norm": 0.69921875,
      "learning_rate": 6.119338497005017e-06,
      "loss": 0.7714,
      "step": 7194
    },
    {
      "epoch": 3.8627898392671387,
      "grad_norm": 0.7109375,
      "learning_rate": 6.11380408851126e-06,
      "loss": 0.978,
      "step": 7195
    },
    {
      "epoch": 3.863326734002215,
      "grad_norm": 0.6640625,
      "learning_rate": 6.1082718351716225e-06,
      "loss": 0.8781,
      "step": 7196
    },
    {
      "epoch": 3.863863628737291,
      "grad_norm": 0.65234375,
      "learning_rate": 6.102741737617413e-06,
      "loss": 0.8758,
      "step": 7197
    },
    {
      "epoch": 3.8644005234723666,
      "grad_norm": 0.64453125,
      "learning_rate": 6.0972137964796946e-06,
      "loss": 0.9519,
      "step": 7198
    },
    {
      "epoch": 3.8649374182074427,
      "grad_norm": 0.70703125,
      "learning_rate": 6.091688012389257e-06,
      "loss": 0.9811,
      "step": 7199
    },
    {
      "epoch": 3.865474312942519,
      "grad_norm": 0.6328125,
      "learning_rate": 6.086164385976681e-06,
      "loss": 0.7897,
      "step": 7200
    },
    {
      "epoch": 3.865474312942519,
      "eval_loss": 0.902592658996582,
      "eval_runtime": 46.4544,
      "eval_samples_per_second": 20.708,
      "eval_steps_per_second": 20.708,
      "step": 7200
    },
    {
      "epoch": 3.8660112076775945,
      "grad_norm": 0.6875,
      "learning_rate": 6.080642917872273e-06,
      "loss": 0.8944,
      "step": 7201
    },
    {
      "epoch": 3.8665481024126707,
      "grad_norm": 0.6328125,
      "learning_rate": 6.075123608706093e-06,
      "loss": 0.7142,
      "step": 7202
    },
    {
      "epoch": 3.867084997147747,
      "grad_norm": 0.73828125,
      "learning_rate": 6.069606459107974e-06,
      "loss": 1.0613,
      "step": 7203
    },
    {
      "epoch": 3.8676218918828225,
      "grad_norm": 0.625,
      "learning_rate": 6.0640914697075e-06,
      "loss": 0.7551,
      "step": 7204
    },
    {
      "epoch": 3.8681587866178986,
      "grad_norm": 1.0234375,
      "learning_rate": 6.058578641133986e-06,
      "loss": 0.8401,
      "step": 7205
    },
    {
      "epoch": 3.8686956813529747,
      "grad_norm": 0.6953125,
      "learning_rate": 6.0530679740165245e-06,
      "loss": 0.9745,
      "step": 7206
    },
    {
      "epoch": 3.869232576088051,
      "grad_norm": 0.55859375,
      "learning_rate": 6.047559468983954e-06,
      "loss": 0.7397,
      "step": 7207
    },
    {
      "epoch": 3.869769470823127,
      "grad_norm": 0.60546875,
      "learning_rate": 6.042053126664862e-06,
      "loss": 0.8461,
      "step": 7208
    },
    {
      "epoch": 3.8703063655582026,
      "grad_norm": 0.71875,
      "learning_rate": 6.036548947687587e-06,
      "loss": 0.7573,
      "step": 7209
    },
    {
      "epoch": 3.8708432602932787,
      "grad_norm": 0.65625,
      "learning_rate": 6.031046932680229e-06,
      "loss": 0.8045,
      "step": 7210
    },
    {
      "epoch": 3.871380155028355,
      "grad_norm": 0.64453125,
      "learning_rate": 6.025547082270644e-06,
      "loss": 0.7573,
      "step": 7211
    },
    {
      "epoch": 3.8719170497634305,
      "grad_norm": 0.7265625,
      "learning_rate": 6.020049397086422e-06,
      "loss": 0.9924,
      "step": 7212
    },
    {
      "epoch": 3.8724539444985067,
      "grad_norm": 0.9375,
      "learning_rate": 6.014553877754925e-06,
      "loss": 0.8774,
      "step": 7213
    },
    {
      "epoch": 3.872990839233583,
      "grad_norm": 0.73828125,
      "learning_rate": 6.00906052490327e-06,
      "loss": 0.9509,
      "step": 7214
    },
    {
      "epoch": 3.873527733968659,
      "grad_norm": 0.8671875,
      "learning_rate": 6.003569339158302e-06,
      "loss": 0.8107,
      "step": 7215
    },
    {
      "epoch": 3.874064628703735,
      "grad_norm": 0.7734375,
      "learning_rate": 5.998080321146651e-06,
      "loss": 0.9037,
      "step": 7216
    },
    {
      "epoch": 3.8746015234388107,
      "grad_norm": 0.62890625,
      "learning_rate": 5.992593471494673e-06,
      "loss": 0.7999,
      "step": 7217
    },
    {
      "epoch": 3.875138418173887,
      "grad_norm": 0.6796875,
      "learning_rate": 5.987108790828486e-06,
      "loss": 0.8981,
      "step": 7218
    },
    {
      "epoch": 3.875675312908963,
      "grad_norm": 0.765625,
      "learning_rate": 5.981626279773961e-06,
      "loss": 0.9565,
      "step": 7219
    },
    {
      "epoch": 3.8762122076440386,
      "grad_norm": 0.64453125,
      "learning_rate": 5.9761459389567346e-06,
      "loss": 0.7808,
      "step": 7220
    },
    {
      "epoch": 3.8767491023791147,
      "grad_norm": 0.71875,
      "learning_rate": 5.9706677690021664e-06,
      "loss": 1.0312,
      "step": 7221
    },
    {
      "epoch": 3.877285997114191,
      "grad_norm": 0.6015625,
      "learning_rate": 5.965191770535391e-06,
      "loss": 0.712,
      "step": 7222
    },
    {
      "epoch": 3.8778228918492665,
      "grad_norm": 0.8828125,
      "learning_rate": 5.9597179441813e-06,
      "loss": 0.9597,
      "step": 7223
    },
    {
      "epoch": 3.8783597865843427,
      "grad_norm": 0.78125,
      "learning_rate": 5.954246290564511e-06,
      "loss": 0.983,
      "step": 7224
    },
    {
      "epoch": 3.878896681319419,
      "grad_norm": 0.6875,
      "learning_rate": 5.948776810309423e-06,
      "loss": 0.8047,
      "step": 7225
    },
    {
      "epoch": 3.879433576054495,
      "grad_norm": 0.7109375,
      "learning_rate": 5.943309504040154e-06,
      "loss": 0.8208,
      "step": 7226
    },
    {
      "epoch": 3.879970470789571,
      "grad_norm": 0.58984375,
      "learning_rate": 5.937844372380616e-06,
      "loss": 0.7954,
      "step": 7227
    },
    {
      "epoch": 3.8805073655246467,
      "grad_norm": 0.73046875,
      "learning_rate": 5.932381415954428e-06,
      "loss": 0.9025,
      "step": 7228
    },
    {
      "epoch": 3.881044260259723,
      "grad_norm": 0.640625,
      "learning_rate": 5.926920635384991e-06,
      "loss": 0.7858,
      "step": 7229
    },
    {
      "epoch": 3.881581154994799,
      "grad_norm": 0.6953125,
      "learning_rate": 5.92146203129546e-06,
      "loss": 0.7611,
      "step": 7230
    },
    {
      "epoch": 3.8821180497298746,
      "grad_norm": 0.7890625,
      "learning_rate": 5.916005604308714e-06,
      "loss": 0.8577,
      "step": 7231
    },
    {
      "epoch": 3.8826549444649507,
      "grad_norm": 0.90234375,
      "learning_rate": 5.9105513550474055e-06,
      "loss": 0.9093,
      "step": 7232
    },
    {
      "epoch": 3.883191839200027,
      "grad_norm": 0.6875,
      "learning_rate": 5.905099284133952e-06,
      "loss": 0.7829,
      "step": 7233
    },
    {
      "epoch": 3.883728733935103,
      "grad_norm": 0.875,
      "learning_rate": 5.899649392190473e-06,
      "loss": 0.8519,
      "step": 7234
    },
    {
      "epoch": 3.884265628670179,
      "grad_norm": 0.8203125,
      "learning_rate": 5.894201679838885e-06,
      "loss": 0.8464,
      "step": 7235
    },
    {
      "epoch": 3.884802523405255,
      "grad_norm": 0.703125,
      "learning_rate": 5.888756147700847e-06,
      "loss": 0.8117,
      "step": 7236
    },
    {
      "epoch": 3.885339418140331,
      "grad_norm": 0.67578125,
      "learning_rate": 5.88331279639775e-06,
      "loss": 0.8754,
      "step": 7237
    },
    {
      "epoch": 3.885876312875407,
      "grad_norm": 0.9140625,
      "learning_rate": 5.877871626550757e-06,
      "loss": 0.7996,
      "step": 7238
    },
    {
      "epoch": 3.8864132076104827,
      "grad_norm": 0.6171875,
      "learning_rate": 5.872432638780781e-06,
      "loss": 0.9268,
      "step": 7239
    },
    {
      "epoch": 3.886950102345559,
      "grad_norm": 0.65234375,
      "learning_rate": 5.866995833708464e-06,
      "loss": 0.7662,
      "step": 7240
    },
    {
      "epoch": 3.887486997080635,
      "grad_norm": 0.625,
      "learning_rate": 5.861561211954222e-06,
      "loss": 0.7394,
      "step": 7241
    },
    {
      "epoch": 3.8880238918157106,
      "grad_norm": 0.8046875,
      "learning_rate": 5.856128774138222e-06,
      "loss": 0.8664,
      "step": 7242
    },
    {
      "epoch": 3.8885607865507867,
      "grad_norm": 0.61328125,
      "learning_rate": 5.850698520880366e-06,
      "loss": 0.7173,
      "step": 7243
    },
    {
      "epoch": 3.889097681285863,
      "grad_norm": 0.6484375,
      "learning_rate": 5.845270452800309e-06,
      "loss": 0.8258,
      "step": 7244
    },
    {
      "epoch": 3.889634576020939,
      "grad_norm": 0.68359375,
      "learning_rate": 5.8398445705174745e-06,
      "loss": 0.9874,
      "step": 7245
    },
    {
      "epoch": 3.890171470756015,
      "grad_norm": 0.609375,
      "learning_rate": 5.8344208746510114e-06,
      "loss": 0.8062,
      "step": 7246
    },
    {
      "epoch": 3.890708365491091,
      "grad_norm": 0.75390625,
      "learning_rate": 5.828999365819837e-06,
      "loss": 0.8254,
      "step": 7247
    },
    {
      "epoch": 3.891245260226167,
      "grad_norm": 0.6640625,
      "learning_rate": 5.823580044642627e-06,
      "loss": 0.9911,
      "step": 7248
    },
    {
      "epoch": 3.891782154961243,
      "grad_norm": 0.6328125,
      "learning_rate": 5.818162911737774e-06,
      "loss": 0.7894,
      "step": 7249
    },
    {
      "epoch": 3.8923190496963187,
      "grad_norm": 0.6875,
      "learning_rate": 5.812747967723459e-06,
      "loss": 0.858,
      "step": 7250
    },
    {
      "epoch": 3.8923190496963187,
      "eval_loss": 0.9027091860771179,
      "eval_runtime": 45.9386,
      "eval_samples_per_second": 20.941,
      "eval_steps_per_second": 20.941,
      "step": 7250
    },
    {
      "epoch": 3.892855944431395,
      "grad_norm": 0.62890625,
      "learning_rate": 5.80733521321758e-06,
      "loss": 0.7453,
      "step": 7251
    },
    {
      "epoch": 3.893392839166471,
      "grad_norm": 0.72265625,
      "learning_rate": 5.801924648837815e-06,
      "loss": 0.8787,
      "step": 7252
    },
    {
      "epoch": 3.893929733901547,
      "grad_norm": 0.6875,
      "learning_rate": 5.796516275201566e-06,
      "loss": 0.9252,
      "step": 7253
    },
    {
      "epoch": 3.894466628636623,
      "grad_norm": 0.6875,
      "learning_rate": 5.791110092926e-06,
      "loss": 0.8662,
      "step": 7254
    },
    {
      "epoch": 3.895003523371699,
      "grad_norm": 0.66796875,
      "learning_rate": 5.785706102628044e-06,
      "loss": 0.7687,
      "step": 7255
    },
    {
      "epoch": 3.895540418106775,
      "grad_norm": 0.78515625,
      "learning_rate": 5.7803043049243415e-06,
      "loss": 1.0045,
      "step": 7256
    },
    {
      "epoch": 3.896077312841851,
      "grad_norm": 0.85546875,
      "learning_rate": 5.774904700431316e-06,
      "loss": 0.7697,
      "step": 7257
    },
    {
      "epoch": 3.896614207576927,
      "grad_norm": 0.60546875,
      "learning_rate": 5.769507289765144e-06,
      "loss": 0.9009,
      "step": 7258
    },
    {
      "epoch": 3.897151102312003,
      "grad_norm": 0.66015625,
      "learning_rate": 5.764112073541711e-06,
      "loss": 0.8913,
      "step": 7259
    },
    {
      "epoch": 3.897687997047079,
      "grad_norm": 0.5859375,
      "learning_rate": 5.758719052376693e-06,
      "loss": 0.6348,
      "step": 7260
    },
    {
      "epoch": 3.8982248917821547,
      "grad_norm": 0.60546875,
      "learning_rate": 5.753328226885513e-06,
      "loss": 0.744,
      "step": 7261
    },
    {
      "epoch": 3.898761786517231,
      "grad_norm": 0.75,
      "learning_rate": 5.747939597683311e-06,
      "loss": 0.9436,
      "step": 7262
    },
    {
      "epoch": 3.899298681252307,
      "grad_norm": 0.57421875,
      "learning_rate": 5.742553165385009e-06,
      "loss": 0.6824,
      "step": 7263
    },
    {
      "epoch": 3.899835575987383,
      "grad_norm": 0.6171875,
      "learning_rate": 5.737168930605272e-06,
      "loss": 0.9219,
      "step": 7264
    },
    {
      "epoch": 3.900372470722459,
      "grad_norm": 0.58203125,
      "learning_rate": 5.731786893958499e-06,
      "loss": 0.6955,
      "step": 7265
    },
    {
      "epoch": 3.900909365457535,
      "grad_norm": 0.69921875,
      "learning_rate": 5.726407056058852e-06,
      "loss": 0.7208,
      "step": 7266
    },
    {
      "epoch": 3.901446260192611,
      "grad_norm": 0.6640625,
      "learning_rate": 5.721029417520249e-06,
      "loss": 0.8116,
      "step": 7267
    },
    {
      "epoch": 3.901983154927687,
      "grad_norm": 0.67578125,
      "learning_rate": 5.71565397895634e-06,
      "loss": 0.8814,
      "step": 7268
    },
    {
      "epoch": 3.902520049662763,
      "grad_norm": 0.94921875,
      "learning_rate": 5.710280740980517e-06,
      "loss": 0.9441,
      "step": 7269
    },
    {
      "epoch": 3.903056944397839,
      "grad_norm": 0.72265625,
      "learning_rate": 5.704909704205949e-06,
      "loss": 0.9297,
      "step": 7270
    },
    {
      "epoch": 3.903593839132915,
      "grad_norm": 0.609375,
      "learning_rate": 5.69954086924554e-06,
      "loss": 0.8212,
      "step": 7271
    },
    {
      "epoch": 3.904130733867991,
      "grad_norm": 0.67578125,
      "learning_rate": 5.694174236711935e-06,
      "loss": 0.7893,
      "step": 7272
    },
    {
      "epoch": 3.9046676286030673,
      "grad_norm": 0.6875,
      "learning_rate": 5.688809807217535e-06,
      "loss": 0.861,
      "step": 7273
    },
    {
      "epoch": 3.905204523338143,
      "grad_norm": 0.64453125,
      "learning_rate": 5.6834475813745014e-06,
      "loss": 0.7909,
      "step": 7274
    },
    {
      "epoch": 3.905741418073219,
      "grad_norm": 0.76171875,
      "learning_rate": 5.678087559794723e-06,
      "loss": 0.8263,
      "step": 7275
    },
    {
      "epoch": 3.906278312808295,
      "grad_norm": 0.81640625,
      "learning_rate": 5.672729743089839e-06,
      "loss": 0.9041,
      "step": 7276
    },
    {
      "epoch": 3.906815207543371,
      "grad_norm": 0.578125,
      "learning_rate": 5.6673741318712585e-06,
      "loss": 0.697,
      "step": 7277
    },
    {
      "epoch": 3.907352102278447,
      "grad_norm": 0.76171875,
      "learning_rate": 5.662020726750111e-06,
      "loss": 1.0148,
      "step": 7278
    },
    {
      "epoch": 3.907888997013523,
      "grad_norm": 0.61328125,
      "learning_rate": 5.656669528337294e-06,
      "loss": 0.6864,
      "step": 7279
    },
    {
      "epoch": 3.908425891748599,
      "grad_norm": 0.796875,
      "learning_rate": 5.651320537243454e-06,
      "loss": 0.9782,
      "step": 7280
    },
    {
      "epoch": 3.908962786483675,
      "grad_norm": 0.6875,
      "learning_rate": 5.645973754078965e-06,
      "loss": 0.8902,
      "step": 7281
    },
    {
      "epoch": 3.909499681218751,
      "grad_norm": 0.66796875,
      "learning_rate": 5.640629179453968e-06,
      "loss": 0.8695,
      "step": 7282
    },
    {
      "epoch": 3.910036575953827,
      "grad_norm": 0.66015625,
      "learning_rate": 5.635286813978357e-06,
      "loss": 0.6493,
      "step": 7283
    },
    {
      "epoch": 3.9105734706889033,
      "grad_norm": 0.6171875,
      "learning_rate": 5.629946658261753e-06,
      "loss": 0.8964,
      "step": 7284
    },
    {
      "epoch": 3.911110365423979,
      "grad_norm": 0.73828125,
      "learning_rate": 5.6246087129135315e-06,
      "loss": 1.0302,
      "step": 7285
    },
    {
      "epoch": 3.911647260159055,
      "grad_norm": 0.66796875,
      "learning_rate": 5.6192729785428335e-06,
      "loss": 0.908,
      "step": 7286
    },
    {
      "epoch": 3.912184154894131,
      "grad_norm": 0.73046875,
      "learning_rate": 5.613939455758518e-06,
      "loss": 0.9243,
      "step": 7287
    },
    {
      "epoch": 3.912721049629207,
      "grad_norm": 0.7109375,
      "learning_rate": 5.608608145169214e-06,
      "loss": 0.8276,
      "step": 7288
    },
    {
      "epoch": 3.913257944364283,
      "grad_norm": 0.734375,
      "learning_rate": 5.6032790473833e-06,
      "loss": 0.9719,
      "step": 7289
    },
    {
      "epoch": 3.913794839099359,
      "grad_norm": 0.74609375,
      "learning_rate": 5.597952163008879e-06,
      "loss": 0.793,
      "step": 7290
    },
    {
      "epoch": 3.9143317338344352,
      "grad_norm": 0.6796875,
      "learning_rate": 5.592627492653826e-06,
      "loss": 0.8368,
      "step": 7291
    },
    {
      "epoch": 3.9148686285695113,
      "grad_norm": 0.6796875,
      "learning_rate": 5.587305036925755e-06,
      "loss": 0.9008,
      "step": 7292
    },
    {
      "epoch": 3.915405523304587,
      "grad_norm": 0.640625,
      "learning_rate": 5.581984796432022e-06,
      "loss": 0.7587,
      "step": 7293
    },
    {
      "epoch": 3.915942418039663,
      "grad_norm": 0.70703125,
      "learning_rate": 5.576666771779726e-06,
      "loss": 0.7886,
      "step": 7294
    },
    {
      "epoch": 3.9164793127747393,
      "grad_norm": 0.875,
      "learning_rate": 5.571350963575728e-06,
      "loss": 1.0384,
      "step": 7295
    },
    {
      "epoch": 3.917016207509815,
      "grad_norm": 0.6875,
      "learning_rate": 5.5660373724266354e-06,
      "loss": 0.7161,
      "step": 7296
    },
    {
      "epoch": 3.917553102244891,
      "grad_norm": 0.67578125,
      "learning_rate": 5.5607259989387815e-06,
      "loss": 0.7293,
      "step": 7297
    },
    {
      "epoch": 3.918089996979967,
      "grad_norm": 0.65625,
      "learning_rate": 5.5554168437182725e-06,
      "loss": 0.7379,
      "step": 7298
    },
    {
      "epoch": 3.918626891715043,
      "grad_norm": 0.609375,
      "learning_rate": 5.550109907370954e-06,
      "loss": 0.7603,
      "step": 7299
    },
    {
      "epoch": 3.919163786450119,
      "grad_norm": 0.796875,
      "learning_rate": 5.544805190502406e-06,
      "loss": 0.9963,
      "step": 7300
    },
    {
      "epoch": 3.919163786450119,
      "eval_loss": 0.9028392434120178,
      "eval_runtime": 45.9722,
      "eval_samples_per_second": 20.926,
      "eval_steps_per_second": 20.926,
      "step": 7300
    },
    {
      "epoch": 3.919700681185195,
      "grad_norm": 0.8515625,
      "learning_rate": 5.53950269371796e-06,
      "loss": 0.8868,
      "step": 7301
    },
    {
      "epoch": 3.9202375759202712,
      "grad_norm": 0.75390625,
      "learning_rate": 5.534202417622711e-06,
      "loss": 0.8868,
      "step": 7302
    },
    {
      "epoch": 3.9207744706553473,
      "grad_norm": 0.703125,
      "learning_rate": 5.528904362821475e-06,
      "loss": 0.9035,
      "step": 7303
    },
    {
      "epoch": 3.921311365390423,
      "grad_norm": 0.6171875,
      "learning_rate": 5.523608529918831e-06,
      "loss": 0.7714,
      "step": 7304
    },
    {
      "epoch": 3.921848260125499,
      "grad_norm": 0.59375,
      "learning_rate": 5.518314919519107e-06,
      "loss": 0.7216,
      "step": 7305
    },
    {
      "epoch": 3.9223851548605753,
      "grad_norm": 0.6484375,
      "learning_rate": 5.5130235322263634e-06,
      "loss": 0.8625,
      "step": 7306
    },
    {
      "epoch": 3.922922049595651,
      "grad_norm": 0.61328125,
      "learning_rate": 5.507734368644416e-06,
      "loss": 0.7752,
      "step": 7307
    },
    {
      "epoch": 3.923458944330727,
      "grad_norm": 0.6796875,
      "learning_rate": 5.502447429376833e-06,
      "loss": 0.84,
      "step": 7308
    },
    {
      "epoch": 3.923995839065803,
      "grad_norm": 0.69921875,
      "learning_rate": 5.497162715026915e-06,
      "loss": 0.7138,
      "step": 7309
    },
    {
      "epoch": 3.9245327338008793,
      "grad_norm": 0.640625,
      "learning_rate": 5.491880226197707e-06,
      "loss": 1.1213,
      "step": 7310
    },
    {
      "epoch": 3.9250696285359554,
      "grad_norm": 0.53125,
      "learning_rate": 5.486599963492017e-06,
      "loss": 0.6998,
      "step": 7311
    },
    {
      "epoch": 3.925606523271031,
      "grad_norm": 0.72265625,
      "learning_rate": 5.481321927512395e-06,
      "loss": 0.8277,
      "step": 7312
    },
    {
      "epoch": 3.926143418006107,
      "grad_norm": 0.65625,
      "learning_rate": 5.476046118861117e-06,
      "loss": 0.8081,
      "step": 7313
    },
    {
      "epoch": 3.9266803127411833,
      "grad_norm": 0.66015625,
      "learning_rate": 5.470772538140228e-06,
      "loss": 0.9648,
      "step": 7314
    },
    {
      "epoch": 3.927217207476259,
      "grad_norm": 0.65625,
      "learning_rate": 5.465501185951519e-06,
      "loss": 0.8677,
      "step": 7315
    },
    {
      "epoch": 3.927754102211335,
      "grad_norm": 0.640625,
      "learning_rate": 5.4602320628965e-06,
      "loss": 0.756,
      "step": 7316
    },
    {
      "epoch": 3.9282909969464113,
      "grad_norm": 0.671875,
      "learning_rate": 5.454965169576462e-06,
      "loss": 0.8091,
      "step": 7317
    },
    {
      "epoch": 3.928827891681487,
      "grad_norm": 0.76171875,
      "learning_rate": 5.4497005065924145e-06,
      "loss": 0.8709,
      "step": 7318
    },
    {
      "epoch": 3.929364786416563,
      "grad_norm": 0.70703125,
      "learning_rate": 5.444438074545119e-06,
      "loss": 0.8644,
      "step": 7319
    },
    {
      "epoch": 3.929901681151639,
      "grad_norm": 0.70703125,
      "learning_rate": 5.4391778740350896e-06,
      "loss": 0.9834,
      "step": 7320
    },
    {
      "epoch": 3.9304385758867153,
      "grad_norm": 0.6875,
      "learning_rate": 5.4339199056625876e-06,
      "loss": 0.8151,
      "step": 7321
    },
    {
      "epoch": 3.9309754706217914,
      "grad_norm": 0.76171875,
      "learning_rate": 5.428664170027603e-06,
      "loss": 0.8584,
      "step": 7322
    },
    {
      "epoch": 3.931512365356867,
      "grad_norm": 0.6953125,
      "learning_rate": 5.4234106677298894e-06,
      "loss": 0.8024,
      "step": 7323
    },
    {
      "epoch": 3.932049260091943,
      "grad_norm": 0.80859375,
      "learning_rate": 5.418159399368941e-06,
      "loss": 0.8314,
      "step": 7324
    },
    {
      "epoch": 3.9325861548270193,
      "grad_norm": 1.421875,
      "learning_rate": 5.412910365543988e-06,
      "loss": 0.7789,
      "step": 7325
    },
    {
      "epoch": 3.933123049562095,
      "grad_norm": 0.90234375,
      "learning_rate": 5.4076635668540075e-06,
      "loss": 0.967,
      "step": 7326
    },
    {
      "epoch": 3.933659944297171,
      "grad_norm": 0.7890625,
      "learning_rate": 5.402419003897738e-06,
      "loss": 0.8204,
      "step": 7327
    },
    {
      "epoch": 3.9341968390322473,
      "grad_norm": 0.68359375,
      "learning_rate": 5.3971766772736354e-06,
      "loss": 0.7443,
      "step": 7328
    },
    {
      "epoch": 3.9347337337673234,
      "grad_norm": 0.79296875,
      "learning_rate": 5.3919365875799235e-06,
      "loss": 0.8301,
      "step": 7329
    },
    {
      "epoch": 3.9352706285023995,
      "grad_norm": 0.65625,
      "learning_rate": 5.3866987354145724e-06,
      "loss": 0.7601,
      "step": 7330
    },
    {
      "epoch": 3.935807523237475,
      "grad_norm": 0.73828125,
      "learning_rate": 5.381463121375269e-06,
      "loss": 1.0165,
      "step": 7331
    },
    {
      "epoch": 3.9363444179725513,
      "grad_norm": 0.98828125,
      "learning_rate": 5.3762297460594714e-06,
      "loss": 0.8752,
      "step": 7332
    },
    {
      "epoch": 3.9368813127076274,
      "grad_norm": 0.796875,
      "learning_rate": 5.370998610064382e-06,
      "loss": 0.9076,
      "step": 7333
    },
    {
      "epoch": 3.937418207442703,
      "grad_norm": 0.76171875,
      "learning_rate": 5.365769713986935e-06,
      "loss": 0.9709,
      "step": 7334
    },
    {
      "epoch": 3.937955102177779,
      "grad_norm": 0.66796875,
      "learning_rate": 5.3605430584238e-06,
      "loss": 0.7881,
      "step": 7335
    },
    {
      "epoch": 3.9384919969128553,
      "grad_norm": 0.66796875,
      "learning_rate": 5.355318643971419e-06,
      "loss": 0.7656,
      "step": 7336
    },
    {
      "epoch": 3.939028891647931,
      "grad_norm": 0.63671875,
      "learning_rate": 5.350096471225966e-06,
      "loss": 0.7514,
      "step": 7337
    },
    {
      "epoch": 3.939565786383007,
      "grad_norm": 0.6953125,
      "learning_rate": 5.344876540783345e-06,
      "loss": 0.8952,
      "step": 7338
    },
    {
      "epoch": 3.9401026811180833,
      "grad_norm": 0.6328125,
      "learning_rate": 5.339658853239221e-06,
      "loss": 0.7225,
      "step": 7339
    },
    {
      "epoch": 3.9406395758531594,
      "grad_norm": 0.640625,
      "learning_rate": 5.334443409189011e-06,
      "loss": 0.8777,
      "step": 7340
    },
    {
      "epoch": 3.9411764705882355,
      "grad_norm": 0.5859375,
      "learning_rate": 5.3292302092278395e-06,
      "loss": 0.765,
      "step": 7341
    },
    {
      "epoch": 3.941713365323311,
      "grad_norm": 0.6953125,
      "learning_rate": 5.324019253950621e-06,
      "loss": 0.9015,
      "step": 7342
    },
    {
      "epoch": 3.9422502600583873,
      "grad_norm": 0.671875,
      "learning_rate": 5.3188105439519825e-06,
      "loss": 0.8368,
      "step": 7343
    },
    {
      "epoch": 3.9427871547934634,
      "grad_norm": 0.66796875,
      "learning_rate": 5.313604079826295e-06,
      "loss": 0.8367,
      "step": 7344
    },
    {
      "epoch": 3.943324049528539,
      "grad_norm": 0.6953125,
      "learning_rate": 5.308399862167693e-06,
      "loss": 0.9607,
      "step": 7345
    },
    {
      "epoch": 3.943860944263615,
      "grad_norm": 0.703125,
      "learning_rate": 5.303197891570044e-06,
      "loss": 0.811,
      "step": 7346
    },
    {
      "epoch": 3.9443978389986913,
      "grad_norm": 0.6328125,
      "learning_rate": 5.29799816862695e-06,
      "loss": 0.8934,
      "step": 7347
    },
    {
      "epoch": 3.9449347337337675,
      "grad_norm": 0.5546875,
      "learning_rate": 5.292800693931771e-06,
      "loss": 0.7024,
      "step": 7348
    },
    {
      "epoch": 3.9454716284688436,
      "grad_norm": 0.89453125,
      "learning_rate": 5.2876054680776085e-06,
      "loss": 0.9306,
      "step": 7349
    },
    {
      "epoch": 3.9460085232039193,
      "grad_norm": 0.73046875,
      "learning_rate": 5.282412491657296e-06,
      "loss": 0.7619,
      "step": 7350
    },
    {
      "epoch": 3.9460085232039193,
      "eval_loss": 0.902666449546814,
      "eval_runtime": 46.036,
      "eval_samples_per_second": 20.897,
      "eval_steps_per_second": 20.897,
      "step": 7350
    },
    {
      "epoch": 3.9465454179389954,
      "grad_norm": 1.1640625,
      "learning_rate": 5.2772217652634265e-06,
      "loss": 0.9595,
      "step": 7351
    },
    {
      "epoch": 3.9470823126740715,
      "grad_norm": 0.66796875,
      "learning_rate": 5.272033289488315e-06,
      "loss": 0.8536,
      "step": 7352
    },
    {
      "epoch": 3.947619207409147,
      "grad_norm": 0.9140625,
      "learning_rate": 5.266847064924047e-06,
      "loss": 0.7785,
      "step": 7353
    },
    {
      "epoch": 3.9481561021442233,
      "grad_norm": 0.640625,
      "learning_rate": 5.261663092162422e-06,
      "loss": 0.7009,
      "step": 7354
    },
    {
      "epoch": 3.9486929968792994,
      "grad_norm": 0.6484375,
      "learning_rate": 5.2564813717950055e-06,
      "loss": 0.8897,
      "step": 7355
    },
    {
      "epoch": 3.949229891614375,
      "grad_norm": 0.54296875,
      "learning_rate": 5.2513019044131e-06,
      "loss": 0.6483,
      "step": 7356
    },
    {
      "epoch": 3.949766786349451,
      "grad_norm": 0.6015625,
      "learning_rate": 5.24612469060774e-06,
      "loss": 0.7246,
      "step": 7357
    },
    {
      "epoch": 3.9503036810845273,
      "grad_norm": 0.7109375,
      "learning_rate": 5.240949730969713e-06,
      "loss": 0.9832,
      "step": 7358
    },
    {
      "epoch": 3.9508405758196035,
      "grad_norm": 0.66796875,
      "learning_rate": 5.235777026089566e-06,
      "loss": 0.7768,
      "step": 7359
    },
    {
      "epoch": 3.9513774705546796,
      "grad_norm": 0.78515625,
      "learning_rate": 5.23060657655754e-06,
      "loss": 0.8202,
      "step": 7360
    },
    {
      "epoch": 3.9519143652897553,
      "grad_norm": 0.671875,
      "learning_rate": 5.22543838296366e-06,
      "loss": 0.9459,
      "step": 7361
    },
    {
      "epoch": 3.9524512600248314,
      "grad_norm": 0.6953125,
      "learning_rate": 5.2202724458976945e-06,
      "loss": 0.8082,
      "step": 7362
    },
    {
      "epoch": 3.9529881547599075,
      "grad_norm": 0.6796875,
      "learning_rate": 5.215108765949126e-06,
      "loss": 0.7899,
      "step": 7363
    },
    {
      "epoch": 3.953525049494983,
      "grad_norm": 0.66015625,
      "learning_rate": 5.209947343707202e-06,
      "loss": 1.0224,
      "step": 7364
    },
    {
      "epoch": 3.9540619442300593,
      "grad_norm": 0.59375,
      "learning_rate": 5.2047881797609135e-06,
      "loss": 0.8479,
      "step": 7365
    },
    {
      "epoch": 3.9545988389651354,
      "grad_norm": 0.6796875,
      "learning_rate": 5.199631274698974e-06,
      "loss": 0.7561,
      "step": 7366
    },
    {
      "epoch": 3.9551357337002115,
      "grad_norm": 0.75,
      "learning_rate": 5.1944766291098625e-06,
      "loss": 0.7702,
      "step": 7367
    },
    {
      "epoch": 3.9556726284352877,
      "grad_norm": 0.75,
      "learning_rate": 5.189324243581778e-06,
      "loss": 0.9321,
      "step": 7368
    },
    {
      "epoch": 3.9562095231703633,
      "grad_norm": 0.66015625,
      "learning_rate": 5.1841741187026864e-06,
      "loss": 0.7873,
      "step": 7369
    },
    {
      "epoch": 3.9567464179054395,
      "grad_norm": 0.7109375,
      "learning_rate": 5.179026255060268e-06,
      "loss": 0.8677,
      "step": 7370
    },
    {
      "epoch": 3.9572833126405156,
      "grad_norm": 0.69921875,
      "learning_rate": 5.173880653241972e-06,
      "loss": 1.0095,
      "step": 7371
    },
    {
      "epoch": 3.9578202073755913,
      "grad_norm": 0.6796875,
      "learning_rate": 5.168737313834967e-06,
      "loss": 0.8071,
      "step": 7372
    },
    {
      "epoch": 3.9583571021106674,
      "grad_norm": 0.7421875,
      "learning_rate": 5.163596237426174e-06,
      "loss": 0.7922,
      "step": 7373
    },
    {
      "epoch": 3.9588939968457435,
      "grad_norm": 0.703125,
      "learning_rate": 5.158457424602267e-06,
      "loss": 0.776,
      "step": 7374
    },
    {
      "epoch": 3.959430891580819,
      "grad_norm": 0.61328125,
      "learning_rate": 5.153320875949632e-06,
      "loss": 0.9226,
      "step": 7375
    },
    {
      "epoch": 3.9599677863158953,
      "grad_norm": 0.67578125,
      "learning_rate": 5.148186592054433e-06,
      "loss": 0.9211,
      "step": 7376
    },
    {
      "epoch": 3.9605046810509714,
      "grad_norm": 0.65234375,
      "learning_rate": 5.143054573502537e-06,
      "loss": 0.746,
      "step": 7377
    },
    {
      "epoch": 3.9610415757860475,
      "grad_norm": 1.1484375,
      "learning_rate": 5.137924820879589e-06,
      "loss": 0.8041,
      "step": 7378
    },
    {
      "epoch": 3.9615784705211237,
      "grad_norm": 0.8203125,
      "learning_rate": 5.1327973347709474e-06,
      "loss": 0.8759,
      "step": 7379
    },
    {
      "epoch": 3.9621153652561993,
      "grad_norm": 0.69140625,
      "learning_rate": 5.1276721157617255e-06,
      "loss": 0.6991,
      "step": 7380
    },
    {
      "epoch": 3.9626522599912755,
      "grad_norm": 0.70703125,
      "learning_rate": 5.122549164436785e-06,
      "loss": 0.7255,
      "step": 7381
    },
    {
      "epoch": 3.9631891547263516,
      "grad_norm": 0.6640625,
      "learning_rate": 5.117428481380704e-06,
      "loss": 0.8164,
      "step": 7382
    },
    {
      "epoch": 3.9637260494614273,
      "grad_norm": 0.77734375,
      "learning_rate": 5.112310067177828e-06,
      "loss": 0.762,
      "step": 7383
    },
    {
      "epoch": 3.9642629441965034,
      "grad_norm": 0.73046875,
      "learning_rate": 5.107193922412243e-06,
      "loss": 0.8531,
      "step": 7384
    },
    {
      "epoch": 3.9647998389315795,
      "grad_norm": 0.64453125,
      "learning_rate": 5.10208004766774e-06,
      "loss": 0.8792,
      "step": 7385
    },
    {
      "epoch": 3.9653367336666556,
      "grad_norm": 0.59375,
      "learning_rate": 5.096968443527892e-06,
      "loss": 0.6783,
      "step": 7386
    },
    {
      "epoch": 3.9658736284017317,
      "grad_norm": 0.7421875,
      "learning_rate": 5.091859110576e-06,
      "loss": 0.7629,
      "step": 7387
    },
    {
      "epoch": 3.9664105231368074,
      "grad_norm": 0.6640625,
      "learning_rate": 5.086752049395094e-06,
      "loss": 0.8086,
      "step": 7388
    },
    {
      "epoch": 3.9669474178718835,
      "grad_norm": 0.6328125,
      "learning_rate": 5.081647260567962e-06,
      "loss": 0.8446,
      "step": 7389
    },
    {
      "epoch": 3.9674843126069597,
      "grad_norm": 0.75,
      "learning_rate": 5.076544744677128e-06,
      "loss": 0.877,
      "step": 7390
    },
    {
      "epoch": 3.9680212073420353,
      "grad_norm": 0.80078125,
      "learning_rate": 5.071444502304842e-06,
      "loss": 0.8079,
      "step": 7391
    },
    {
      "epoch": 3.9685581020771115,
      "grad_norm": 0.8125,
      "learning_rate": 5.0663465340331145e-06,
      "loss": 0.6302,
      "step": 7392
    },
    {
      "epoch": 3.9690949968121876,
      "grad_norm": 0.6875,
      "learning_rate": 5.06125084044369e-06,
      "loss": 0.9266,
      "step": 7393
    },
    {
      "epoch": 3.9696318915472633,
      "grad_norm": 0.921875,
      "learning_rate": 5.056157422118052e-06,
      "loss": 1.2568,
      "step": 7394
    },
    {
      "epoch": 3.9701687862823394,
      "grad_norm": 0.6796875,
      "learning_rate": 5.051066279637412e-06,
      "loss": 0.8446,
      "step": 7395
    },
    {
      "epoch": 3.9707056810174155,
      "grad_norm": 0.66796875,
      "learning_rate": 5.045977413582745e-06,
      "loss": 0.8624,
      "step": 7396
    },
    {
      "epoch": 3.9712425757524916,
      "grad_norm": 0.68359375,
      "learning_rate": 5.040890824534756e-06,
      "loss": 0.8896,
      "step": 7397
    },
    {
      "epoch": 3.9717794704875677,
      "grad_norm": 0.71484375,
      "learning_rate": 5.035806513073879e-06,
      "loss": 0.8596,
      "step": 7398
    },
    {
      "epoch": 3.9723163652226434,
      "grad_norm": 0.6953125,
      "learning_rate": 5.030724479780305e-06,
      "loss": 0.8962,
      "step": 7399
    },
    {
      "epoch": 3.9728532599577195,
      "grad_norm": 0.671875,
      "learning_rate": 5.025644725233966e-06,
      "loss": 0.8844,
      "step": 7400
    },
    {
      "epoch": 3.9728532599577195,
      "eval_loss": 0.9027516841888428,
      "eval_runtime": 46.1077,
      "eval_samples_per_second": 20.864,
      "eval_steps_per_second": 20.864,
      "step": 7400
    },
    {
      "epoch": 3.9733901546927957,
      "grad_norm": 0.796875,
      "learning_rate": 5.020567250014518e-06,
      "loss": 1.0648,
      "step": 7401
    },
    {
      "epoch": 3.9739270494278713,
      "grad_norm": 0.7578125,
      "learning_rate": 5.015492054701359e-06,
      "loss": 0.9729,
      "step": 7402
    },
    {
      "epoch": 3.9744639441629475,
      "grad_norm": 0.80859375,
      "learning_rate": 5.010419139873648e-06,
      "loss": 0.7904,
      "step": 7403
    },
    {
      "epoch": 3.9750008388980236,
      "grad_norm": 0.63671875,
      "learning_rate": 5.005348506110252e-06,
      "loss": 0.664,
      "step": 7404
    },
    {
      "epoch": 3.9755377336330997,
      "grad_norm": 0.7109375,
      "learning_rate": 5.000280153989806e-06,
      "loss": 0.7639,
      "step": 7405
    },
    {
      "epoch": 3.976074628368176,
      "grad_norm": 0.6328125,
      "learning_rate": 4.995214084090674e-06,
      "loss": 0.7974,
      "step": 7406
    },
    {
      "epoch": 3.9766115231032515,
      "grad_norm": 0.8203125,
      "learning_rate": 4.99015029699095e-06,
      "loss": 1.073,
      "step": 7407
    },
    {
      "epoch": 3.9771484178383276,
      "grad_norm": 0.60546875,
      "learning_rate": 4.98508879326848e-06,
      "loss": 0.73,
      "step": 7408
    },
    {
      "epoch": 3.9776853125734037,
      "grad_norm": 0.578125,
      "learning_rate": 4.980029573500855e-06,
      "loss": 0.6907,
      "step": 7409
    },
    {
      "epoch": 3.9782222073084794,
      "grad_norm": 0.67578125,
      "learning_rate": 4.9749726382653905e-06,
      "loss": 0.8197,
      "step": 7410
    },
    {
      "epoch": 3.9787591020435555,
      "grad_norm": 0.6953125,
      "learning_rate": 4.969917988139134e-06,
      "loss": 0.7788,
      "step": 7411
    },
    {
      "epoch": 3.9792959967786317,
      "grad_norm": 0.703125,
      "learning_rate": 4.964865623698903e-06,
      "loss": 0.8202,
      "step": 7412
    },
    {
      "epoch": 3.9798328915137073,
      "grad_norm": 0.6796875,
      "learning_rate": 4.959815545521224e-06,
      "loss": 0.9292,
      "step": 7413
    },
    {
      "epoch": 3.9803697862487835,
      "grad_norm": 0.6484375,
      "learning_rate": 4.9547677541823765e-06,
      "loss": 0.6873,
      "step": 7414
    },
    {
      "epoch": 3.9809066809838596,
      "grad_norm": 0.65234375,
      "learning_rate": 4.94972225025839e-06,
      "loss": 0.9967,
      "step": 7415
    },
    {
      "epoch": 3.9814435757189357,
      "grad_norm": 0.56640625,
      "learning_rate": 4.944679034325003e-06,
      "loss": 0.7426,
      "step": 7416
    },
    {
      "epoch": 3.981980470454012,
      "grad_norm": 0.59375,
      "learning_rate": 4.9396381069577165e-06,
      "loss": 0.9072,
      "step": 7417
    },
    {
      "epoch": 3.9825173651890875,
      "grad_norm": 0.79296875,
      "learning_rate": 4.934599468731777e-06,
      "loss": 0.9,
      "step": 7418
    },
    {
      "epoch": 3.9830542599241636,
      "grad_norm": 0.65234375,
      "learning_rate": 4.929563120222141e-06,
      "loss": 0.907,
      "step": 7419
    },
    {
      "epoch": 3.9835911546592397,
      "grad_norm": 0.58984375,
      "learning_rate": 4.924529062003522e-06,
      "loss": 1.0865,
      "step": 7420
    },
    {
      "epoch": 3.9841280493943154,
      "grad_norm": 0.6484375,
      "learning_rate": 4.919497294650369e-06,
      "loss": 0.8952,
      "step": 7421
    },
    {
      "epoch": 3.9846649441293915,
      "grad_norm": 0.65234375,
      "learning_rate": 4.9144678187368815e-06,
      "loss": 0.7385,
      "step": 7422
    },
    {
      "epoch": 3.9852018388644677,
      "grad_norm": 0.78125,
      "learning_rate": 4.909440634836973e-06,
      "loss": 0.9948,
      "step": 7423
    },
    {
      "epoch": 3.985738733599544,
      "grad_norm": 0.61328125,
      "learning_rate": 4.904415743524313e-06,
      "loss": 0.8738,
      "step": 7424
    },
    {
      "epoch": 3.98627562833462,
      "grad_norm": 0.72265625,
      "learning_rate": 4.899393145372313e-06,
      "loss": 0.9451,
      "step": 7425
    },
    {
      "epoch": 3.9868125230696956,
      "grad_norm": 0.84375,
      "learning_rate": 4.894372840954112e-06,
      "loss": 0.9289,
      "step": 7426
    },
    {
      "epoch": 3.9873494178047717,
      "grad_norm": 0.703125,
      "learning_rate": 4.889354830842577e-06,
      "loss": 0.9262,
      "step": 7427
    },
    {
      "epoch": 3.987886312539848,
      "grad_norm": 0.59765625,
      "learning_rate": 4.884339115610345e-06,
      "loss": 0.8329,
      "step": 7428
    },
    {
      "epoch": 3.9884232072749235,
      "grad_norm": 0.60546875,
      "learning_rate": 4.879325695829757e-06,
      "loss": 0.7719,
      "step": 7429
    },
    {
      "epoch": 3.9889601020099996,
      "grad_norm": 0.64453125,
      "learning_rate": 4.8743145720729134e-06,
      "loss": 0.677,
      "step": 7430
    },
    {
      "epoch": 3.9894969967450757,
      "grad_norm": 0.7421875,
      "learning_rate": 4.869305744911656e-06,
      "loss": 0.9912,
      "step": 7431
    },
    {
      "epoch": 3.9900338914801514,
      "grad_norm": 0.66796875,
      "learning_rate": 4.864299214917542e-06,
      "loss": 0.9451,
      "step": 7432
    },
    {
      "epoch": 3.9905707862152275,
      "grad_norm": 0.77734375,
      "learning_rate": 4.859294982661886e-06,
      "loss": 0.9915,
      "step": 7433
    },
    {
      "epoch": 3.9911076809503037,
      "grad_norm": 0.6875,
      "learning_rate": 4.854293048715738e-06,
      "loss": 0.7315,
      "step": 7434
    },
    {
      "epoch": 3.99164457568538,
      "grad_norm": 0.80859375,
      "learning_rate": 4.84929341364988e-06,
      "loss": 0.9712,
      "step": 7435
    },
    {
      "epoch": 3.992181470420456,
      "grad_norm": 0.83984375,
      "learning_rate": 4.844296078034824e-06,
      "loss": 0.9264,
      "step": 7436
    },
    {
      "epoch": 3.9927183651555316,
      "grad_norm": 0.80078125,
      "learning_rate": 4.8393010424408355e-06,
      "loss": 0.9062,
      "step": 7437
    },
    {
      "epoch": 3.9932552598906077,
      "grad_norm": 0.5859375,
      "learning_rate": 4.8343083074379205e-06,
      "loss": 0.5828,
      "step": 7438
    },
    {
      "epoch": 3.993792154625684,
      "grad_norm": 0.62890625,
      "learning_rate": 4.829317873595798e-06,
      "loss": 0.8912,
      "step": 7439
    },
    {
      "epoch": 3.9943290493607595,
      "grad_norm": 0.60546875,
      "learning_rate": 4.824329741483949e-06,
      "loss": 0.7507,
      "step": 7440
    },
    {
      "epoch": 3.9948659440958356,
      "grad_norm": 0.8359375,
      "learning_rate": 4.8193439116715865e-06,
      "loss": 0.9012,
      "step": 7441
    },
    {
      "epoch": 3.9954028388309117,
      "grad_norm": 0.6640625,
      "learning_rate": 4.814360384727646e-06,
      "loss": 0.8761,
      "step": 7442
    },
    {
      "epoch": 3.995939733565988,
      "grad_norm": 0.68359375,
      "learning_rate": 4.80937916122082e-06,
      "loss": 0.6817,
      "step": 7443
    },
    {
      "epoch": 3.996476628301064,
      "grad_norm": 0.69140625,
      "learning_rate": 4.804400241719525e-06,
      "loss": 0.752,
      "step": 7444
    },
    {
      "epoch": 3.9970135230361397,
      "grad_norm": 0.7265625,
      "learning_rate": 4.799423626791916e-06,
      "loss": 0.8484,
      "step": 7445
    },
    {
      "epoch": 3.997550417771216,
      "grad_norm": 0.77734375,
      "learning_rate": 4.794449317005886e-06,
      "loss": 0.9062,
      "step": 7446
    },
    {
      "epoch": 3.998087312506292,
      "grad_norm": 0.6640625,
      "learning_rate": 4.789477312929083e-06,
      "loss": 0.8681,
      "step": 7447
    },
    {
      "epoch": 3.9986242072413676,
      "grad_norm": 0.75390625,
      "learning_rate": 4.784507615128856e-06,
      "loss": 0.8729,
      "step": 7448
    },
    {
      "epoch": 3.9991611019764437,
      "grad_norm": 0.62890625,
      "learning_rate": 4.779540224172316e-06,
      "loss": 0.8511,
      "step": 7449
    },
    {
      "epoch": 3.99969799671152,
      "grad_norm": 0.609375,
      "learning_rate": 4.7745751406263165e-06,
      "loss": 0.8264,
      "step": 7450
    },
    {
      "epoch": 3.99969799671152,
      "eval_loss": 0.9027752876281738,
      "eval_runtime": 45.8409,
      "eval_samples_per_second": 20.986,
      "eval_steps_per_second": 20.986,
      "step": 7450
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.62109375,
      "learning_rate": 4.769612365057427e-06,
      "loss": 0.4663,
      "step": 7451
    },
    {
      "epoch": 4.000536894735076,
      "grad_norm": 0.6796875,
      "learning_rate": 4.764651898031958e-06,
      "loss": 0.8877,
      "step": 7452
    },
    {
      "epoch": 4.001073789470152,
      "grad_norm": 0.7578125,
      "learning_rate": 4.7596937401159725e-06,
      "loss": 0.9608,
      "step": 7453
    },
    {
      "epoch": 4.001610684205228,
      "grad_norm": 0.78515625,
      "learning_rate": 4.754737891875249e-06,
      "loss": 0.8652,
      "step": 7454
    },
    {
      "epoch": 4.002147578940304,
      "grad_norm": 0.7578125,
      "learning_rate": 4.749784353875316e-06,
      "loss": 0.9188,
      "step": 7455
    },
    {
      "epoch": 4.00268447367538,
      "grad_norm": 0.6484375,
      "learning_rate": 4.744833126681445e-06,
      "loss": 0.8245,
      "step": 7456
    },
    {
      "epoch": 4.003221368410456,
      "grad_norm": 0.734375,
      "learning_rate": 4.739884210858616e-06,
      "loss": 0.7953,
      "step": 7457
    },
    {
      "epoch": 4.003758263145532,
      "grad_norm": 0.76953125,
      "learning_rate": 4.734937606971576e-06,
      "loss": 0.9617,
      "step": 7458
    },
    {
      "epoch": 4.004295157880608,
      "grad_norm": 0.70703125,
      "learning_rate": 4.729993315584797e-06,
      "loss": 0.903,
      "step": 7459
    },
    {
      "epoch": 4.004832052615684,
      "grad_norm": 0.7265625,
      "learning_rate": 4.725051337262476e-06,
      "loss": 1.019,
      "step": 7460
    },
    {
      "epoch": 4.00536894735076,
      "grad_norm": 0.6171875,
      "learning_rate": 4.7201116725685555e-06,
      "loss": 0.9115,
      "step": 7461
    },
    {
      "epoch": 4.005905842085836,
      "grad_norm": 0.625,
      "learning_rate": 4.715174322066715e-06,
      "loss": 0.6203,
      "step": 7462
    },
    {
      "epoch": 4.006442736820912,
      "grad_norm": 0.73046875,
      "learning_rate": 4.710239286320378e-06,
      "loss": 0.9211,
      "step": 7463
    },
    {
      "epoch": 4.006979631555988,
      "grad_norm": 0.65625,
      "learning_rate": 4.7053065658926834e-06,
      "loss": 0.7785,
      "step": 7464
    },
    {
      "epoch": 4.007516526291064,
      "grad_norm": 0.62890625,
      "learning_rate": 4.700376161346517e-06,
      "loss": 0.9612,
      "step": 7465
    },
    {
      "epoch": 4.0080534210261405,
      "grad_norm": 0.66796875,
      "learning_rate": 4.695448073244516e-06,
      "loss": 0.9449,
      "step": 7466
    },
    {
      "epoch": 4.008590315761216,
      "grad_norm": 0.5859375,
      "learning_rate": 4.690522302149017e-06,
      "loss": 0.8617,
      "step": 7467
    },
    {
      "epoch": 4.009127210496292,
      "grad_norm": 0.6875,
      "learning_rate": 4.685598848622128e-06,
      "loss": 0.8689,
      "step": 7468
    },
    {
      "epoch": 4.009664105231368,
      "grad_norm": 0.5625,
      "learning_rate": 4.68067771322567e-06,
      "loss": 0.5935,
      "step": 7469
    },
    {
      "epoch": 4.010200999966444,
      "grad_norm": 0.80078125,
      "learning_rate": 4.675758896521204e-06,
      "loss": 0.9286,
      "step": 7470
    },
    {
      "epoch": 4.01073789470152,
      "grad_norm": 0.69140625,
      "learning_rate": 4.670842399070036e-06,
      "loss": 0.8348,
      "step": 7471
    },
    {
      "epoch": 4.011274789436596,
      "grad_norm": 0.69140625,
      "learning_rate": 4.665928221433203e-06,
      "loss": 0.7317,
      "step": 7472
    },
    {
      "epoch": 4.011811684171672,
      "grad_norm": 0.6796875,
      "learning_rate": 4.661016364171461e-06,
      "loss": 0.8828,
      "step": 7473
    },
    {
      "epoch": 4.012348578906748,
      "grad_norm": 0.64453125,
      "learning_rate": 4.656106827845325e-06,
      "loss": 0.8191,
      "step": 7474
    },
    {
      "epoch": 4.012885473641824,
      "grad_norm": 0.87890625,
      "learning_rate": 4.651199613015042e-06,
      "loss": 0.8596,
      "step": 7475
    },
    {
      "epoch": 4.0134223683769,
      "grad_norm": 0.640625,
      "learning_rate": 4.646294720240574e-06,
      "loss": 0.7695,
      "step": 7476
    },
    {
      "epoch": 4.0139592631119765,
      "grad_norm": 0.84375,
      "learning_rate": 4.6413921500816445e-06,
      "loss": 0.8243,
      "step": 7477
    },
    {
      "epoch": 4.014496157847052,
      "grad_norm": 0.7734375,
      "learning_rate": 4.636491903097681e-06,
      "loss": 0.9723,
      "step": 7478
    },
    {
      "epoch": 4.015033052582128,
      "grad_norm": 0.66796875,
      "learning_rate": 4.631593979847884e-06,
      "loss": 0.7722,
      "step": 7479
    },
    {
      "epoch": 4.015569947317204,
      "grad_norm": 0.73828125,
      "learning_rate": 4.626698380891154e-06,
      "loss": 0.8505,
      "step": 7480
    },
    {
      "epoch": 4.01610684205228,
      "grad_norm": 0.6484375,
      "learning_rate": 4.621805106786142e-06,
      "loss": 0.8458,
      "step": 7481
    },
    {
      "epoch": 4.016643736787356,
      "grad_norm": 0.703125,
      "learning_rate": 4.6169141580912455e-06,
      "loss": 0.8673,
      "step": 7482
    },
    {
      "epoch": 4.017180631522432,
      "grad_norm": 0.91796875,
      "learning_rate": 4.612025535364567e-06,
      "loss": 0.9364,
      "step": 7483
    },
    {
      "epoch": 4.017717526257508,
      "grad_norm": 0.64453125,
      "learning_rate": 4.607139239163969e-06,
      "loss": 0.8494,
      "step": 7484
    },
    {
      "epoch": 4.018254420992585,
      "grad_norm": 0.7421875,
      "learning_rate": 4.602255270047048e-06,
      "loss": 0.694,
      "step": 7485
    },
    {
      "epoch": 4.01879131572766,
      "grad_norm": 0.87890625,
      "learning_rate": 4.597373628571106e-06,
      "loss": 0.8932,
      "step": 7486
    },
    {
      "epoch": 4.019328210462736,
      "grad_norm": 0.65625,
      "learning_rate": 4.592494315293211e-06,
      "loss": 0.7776,
      "step": 7487
    },
    {
      "epoch": 4.0198651051978125,
      "grad_norm": 0.80078125,
      "learning_rate": 4.587617330770164e-06,
      "loss": 0.7525,
      "step": 7488
    },
    {
      "epoch": 4.020401999932888,
      "grad_norm": 0.7578125,
      "learning_rate": 4.582742675558471e-06,
      "loss": 0.9899,
      "step": 7489
    },
    {
      "epoch": 4.020938894667964,
      "grad_norm": 0.57421875,
      "learning_rate": 4.577870350214405e-06,
      "loss": 0.806,
      "step": 7490
    },
    {
      "epoch": 4.02147578940304,
      "grad_norm": 0.66796875,
      "learning_rate": 4.5730003552939634e-06,
      "loss": 0.7843,
      "step": 7491
    },
    {
      "epoch": 4.022012684138116,
      "grad_norm": 0.765625,
      "learning_rate": 4.568132691352861e-06,
      "loss": 0.8928,
      "step": 7492
    },
    {
      "epoch": 4.022549578873192,
      "grad_norm": 0.7734375,
      "learning_rate": 4.563267358946574e-06,
      "loss": 0.6967,
      "step": 7493
    },
    {
      "epoch": 4.023086473608268,
      "grad_norm": 0.87890625,
      "learning_rate": 4.558404358630286e-06,
      "loss": 0.734,
      "step": 7494
    },
    {
      "epoch": 4.023623368343344,
      "grad_norm": 0.671875,
      "learning_rate": 4.553543690958939e-06,
      "loss": 0.8173,
      "step": 7495
    },
    {
      "epoch": 4.024160263078421,
      "grad_norm": 0.6640625,
      "learning_rate": 4.548685356487184e-06,
      "loss": 0.794,
      "step": 7496
    },
    {
      "epoch": 4.024697157813496,
      "grad_norm": 0.703125,
      "learning_rate": 4.54382935576943e-06,
      "loss": 0.954,
      "step": 7497
    },
    {
      "epoch": 4.025234052548572,
      "grad_norm": 0.79296875,
      "learning_rate": 4.538975689359795e-06,
      "loss": 0.8888,
      "step": 7498
    },
    {
      "epoch": 4.0257709472836485,
      "grad_norm": 0.734375,
      "learning_rate": 4.534124357812156e-06,
      "loss": 0.9902,
      "step": 7499
    },
    {
      "epoch": 4.026307842018724,
      "grad_norm": 0.6953125,
      "learning_rate": 4.5292753616801106e-06,
      "loss": 0.9657,
      "step": 7500
    },
    {
      "epoch": 4.026307842018724,
      "eval_loss": 0.9026344418525696,
      "eval_runtime": 46.3739,
      "eval_samples_per_second": 20.744,
      "eval_steps_per_second": 20.744,
      "step": 7500
    },
    {
      "epoch": 4.0268447367538,
      "grad_norm": 0.65234375,
      "learning_rate": 4.524428701516981e-06,
      "loss": 0.7382,
      "step": 7501
    },
    {
      "epoch": 4.027381631488876,
      "grad_norm": 0.65625,
      "learning_rate": 4.519584377875846e-06,
      "loss": 1.0347,
      "step": 7502
    },
    {
      "epoch": 4.027918526223952,
      "grad_norm": 0.72265625,
      "learning_rate": 4.51474239130949e-06,
      "loss": 0.8116,
      "step": 7503
    },
    {
      "epoch": 4.028455420959029,
      "grad_norm": 0.69140625,
      "learning_rate": 4.509902742370459e-06,
      "loss": 0.89,
      "step": 7504
    },
    {
      "epoch": 4.028992315694104,
      "grad_norm": 0.890625,
      "learning_rate": 4.505065431611002e-06,
      "loss": 0.915,
      "step": 7505
    },
    {
      "epoch": 4.02952921042918,
      "grad_norm": 0.875,
      "learning_rate": 4.50023045958313e-06,
      "loss": 0.7722,
      "step": 7506
    },
    {
      "epoch": 4.030066105164257,
      "grad_norm": 0.73046875,
      "learning_rate": 4.495397826838577e-06,
      "loss": 0.6443,
      "step": 7507
    },
    {
      "epoch": 4.030602999899332,
      "grad_norm": 0.87890625,
      "learning_rate": 4.490567533928794e-06,
      "loss": 0.7557,
      "step": 7508
    },
    {
      "epoch": 4.031139894634408,
      "grad_norm": 0.7109375,
      "learning_rate": 4.485739581404988e-06,
      "loss": 0.9773,
      "step": 7509
    },
    {
      "epoch": 4.0316767893694845,
      "grad_norm": 0.671875,
      "learning_rate": 4.480913969818098e-06,
      "loss": 0.7444,
      "step": 7510
    },
    {
      "epoch": 4.03221368410456,
      "grad_norm": 0.765625,
      "learning_rate": 4.4760906997187655e-06,
      "loss": 0.8161,
      "step": 7511
    },
    {
      "epoch": 4.032750578839636,
      "grad_norm": 0.75390625,
      "learning_rate": 4.4712697716574e-06,
      "loss": 0.8348,
      "step": 7512
    },
    {
      "epoch": 4.033287473574712,
      "grad_norm": 0.65234375,
      "learning_rate": 4.466451186184134e-06,
      "loss": 0.7544,
      "step": 7513
    },
    {
      "epoch": 4.033824368309788,
      "grad_norm": 0.55859375,
      "learning_rate": 4.461634943848819e-06,
      "loss": 0.6289,
      "step": 7514
    },
    {
      "epoch": 4.034361263044865,
      "grad_norm": 0.703125,
      "learning_rate": 4.456821045201054e-06,
      "loss": 1.0013,
      "step": 7515
    },
    {
      "epoch": 4.03489815777994,
      "grad_norm": 0.7421875,
      "learning_rate": 4.452009490790171e-06,
      "loss": 0.9217,
      "step": 7516
    },
    {
      "epoch": 4.035435052515016,
      "grad_norm": 0.73046875,
      "learning_rate": 4.447200281165217e-06,
      "loss": 0.7167,
      "step": 7517
    },
    {
      "epoch": 4.035971947250093,
      "grad_norm": 0.96875,
      "learning_rate": 4.442393416875002e-06,
      "loss": 0.9273,
      "step": 7518
    },
    {
      "epoch": 4.036508841985168,
      "grad_norm": 0.68359375,
      "learning_rate": 4.437588898468029e-06,
      "loss": 0.8502,
      "step": 7519
    },
    {
      "epoch": 4.037045736720244,
      "grad_norm": 0.76953125,
      "learning_rate": 4.432786726492571e-06,
      "loss": 1.0083,
      "step": 7520
    },
    {
      "epoch": 4.0375826314553205,
      "grad_norm": 0.64453125,
      "learning_rate": 4.427986901496603e-06,
      "loss": 0.9081,
      "step": 7521
    },
    {
      "epoch": 4.038119526190396,
      "grad_norm": 0.9453125,
      "learning_rate": 4.423189424027857e-06,
      "loss": 0.8164,
      "step": 7522
    },
    {
      "epoch": 4.038656420925473,
      "grad_norm": 0.82421875,
      "learning_rate": 4.4183942946337845e-06,
      "loss": 0.9482,
      "step": 7523
    },
    {
      "epoch": 4.039193315660548,
      "grad_norm": 0.7265625,
      "learning_rate": 4.413601513861562e-06,
      "loss": 0.8399,
      "step": 7524
    },
    {
      "epoch": 4.039730210395624,
      "grad_norm": 0.8046875,
      "learning_rate": 4.408811082258116e-06,
      "loss": 0.7816,
      "step": 7525
    },
    {
      "epoch": 4.040267105130701,
      "grad_norm": 0.67578125,
      "learning_rate": 4.404023000370094e-06,
      "loss": 0.7986,
      "step": 7526
    },
    {
      "epoch": 4.040803999865776,
      "grad_norm": 0.65625,
      "learning_rate": 4.399237268743877e-06,
      "loss": 0.8299,
      "step": 7527
    },
    {
      "epoch": 4.041340894600852,
      "grad_norm": 0.703125,
      "learning_rate": 4.3944538879255674e-06,
      "loss": 0.8953,
      "step": 7528
    },
    {
      "epoch": 4.041877789335929,
      "grad_norm": 0.671875,
      "learning_rate": 4.389672858461025e-06,
      "loss": 0.6611,
      "step": 7529
    },
    {
      "epoch": 4.042414684071004,
      "grad_norm": 0.78515625,
      "learning_rate": 4.38489418089581e-06,
      "loss": 0.8634,
      "step": 7530
    },
    {
      "epoch": 4.04295157880608,
      "grad_norm": 0.7578125,
      "learning_rate": 4.38011785577524e-06,
      "loss": 0.7427,
      "step": 7531
    },
    {
      "epoch": 4.0434884735411565,
      "grad_norm": 0.671875,
      "learning_rate": 4.375343883644359e-06,
      "loss": 0.7341,
      "step": 7532
    },
    {
      "epoch": 4.044025368276232,
      "grad_norm": 0.703125,
      "learning_rate": 4.370572265047923e-06,
      "loss": 0.9833,
      "step": 7533
    },
    {
      "epoch": 4.044562263011309,
      "grad_norm": 1.03125,
      "learning_rate": 4.365803000530441e-06,
      "loss": 1.0359,
      "step": 7534
    },
    {
      "epoch": 4.045099157746384,
      "grad_norm": 0.734375,
      "learning_rate": 4.361036090636156e-06,
      "loss": 0.892,
      "step": 7535
    },
    {
      "epoch": 4.04563605248146,
      "grad_norm": 0.70703125,
      "learning_rate": 4.356271535909024e-06,
      "loss": 0.8823,
      "step": 7536
    },
    {
      "epoch": 4.046172947216537,
      "grad_norm": 0.58984375,
      "learning_rate": 4.351509336892737e-06,
      "loss": 0.7737,
      "step": 7537
    },
    {
      "epoch": 4.046709841951612,
      "grad_norm": 0.625,
      "learning_rate": 4.346749494130731e-06,
      "loss": 0.9118,
      "step": 7538
    },
    {
      "epoch": 4.047246736686688,
      "grad_norm": 0.92578125,
      "learning_rate": 4.341992008166151e-06,
      "loss": 0.7746,
      "step": 7539
    },
    {
      "epoch": 4.047783631421765,
      "grad_norm": 0.63671875,
      "learning_rate": 4.3372368795419006e-06,
      "loss": 0.845,
      "step": 7540
    },
    {
      "epoch": 4.04832052615684,
      "grad_norm": 0.671875,
      "learning_rate": 4.332484108800597e-06,
      "loss": 0.9345,
      "step": 7541
    },
    {
      "epoch": 4.048857420891917,
      "grad_norm": 0.765625,
      "learning_rate": 4.327733696484587e-06,
      "loss": 0.9044,
      "step": 7542
    },
    {
      "epoch": 4.0493943156269925,
      "grad_norm": 0.6484375,
      "learning_rate": 4.322985643135952e-06,
      "loss": 0.8434,
      "step": 7543
    },
    {
      "epoch": 4.049931210362068,
      "grad_norm": 0.75,
      "learning_rate": 4.318239949296515e-06,
      "loss": 0.7593,
      "step": 7544
    },
    {
      "epoch": 4.050468105097145,
      "grad_norm": 0.6484375,
      "learning_rate": 4.3134966155078144e-06,
      "loss": 0.7988,
      "step": 7545
    },
    {
      "epoch": 4.05100499983222,
      "grad_norm": 0.85546875,
      "learning_rate": 4.3087556423111186e-06,
      "loss": 1.0619,
      "step": 7546
    },
    {
      "epoch": 4.051541894567296,
      "grad_norm": 0.703125,
      "learning_rate": 4.304017030247434e-06,
      "loss": 1.0173,
      "step": 7547
    },
    {
      "epoch": 4.052078789302373,
      "grad_norm": 0.71875,
      "learning_rate": 4.299280779857506e-06,
      "loss": 0.855,
      "step": 7548
    },
    {
      "epoch": 4.052615684037448,
      "grad_norm": 0.7734375,
      "learning_rate": 4.294546891681791e-06,
      "loss": 0.8628,
      "step": 7549
    },
    {
      "epoch": 4.053152578772524,
      "grad_norm": 0.68359375,
      "learning_rate": 4.289815366260488e-06,
      "loss": 0.7688,
      "step": 7550
    },
    {
      "epoch": 4.053152578772524,
      "eval_loss": 0.9027820825576782,
      "eval_runtime": 46.7155,
      "eval_samples_per_second": 20.593,
      "eval_steps_per_second": 20.593,
      "step": 7550
    },
    {
      "epoch": 4.053689473507601,
      "grad_norm": 0.71484375,
      "learning_rate": 4.2850862041335316e-06,
      "loss": 0.9689,
      "step": 7551
    },
    {
      "epoch": 4.054226368242676,
      "grad_norm": 0.6875,
      "learning_rate": 4.28035940584057e-06,
      "loss": 0.8382,
      "step": 7552
    },
    {
      "epoch": 4.054763262977753,
      "grad_norm": 0.5703125,
      "learning_rate": 4.275634971920988e-06,
      "loss": 0.8201,
      "step": 7553
    },
    {
      "epoch": 4.0553001577128285,
      "grad_norm": 0.578125,
      "learning_rate": 4.270912902913918e-06,
      "loss": 0.9135,
      "step": 7554
    },
    {
      "epoch": 4.055837052447904,
      "grad_norm": 0.74609375,
      "learning_rate": 4.266193199358187e-06,
      "loss": 0.785,
      "step": 7555
    },
    {
      "epoch": 4.056373947182981,
      "grad_norm": 0.64453125,
      "learning_rate": 4.2614758617923885e-06,
      "loss": 0.7262,
      "step": 7556
    },
    {
      "epoch": 4.056910841918056,
      "grad_norm": 0.7421875,
      "learning_rate": 4.256760890754832e-06,
      "loss": 0.8601,
      "step": 7557
    },
    {
      "epoch": 4.057447736653132,
      "grad_norm": 0.64453125,
      "learning_rate": 4.252048286783541e-06,
      "loss": 0.7742,
      "step": 7558
    },
    {
      "epoch": 4.057984631388209,
      "grad_norm": 0.6875,
      "learning_rate": 4.247338050416294e-06,
      "loss": 0.7661,
      "step": 7559
    },
    {
      "epoch": 4.058521526123284,
      "grad_norm": 0.65234375,
      "learning_rate": 4.242630182190594e-06,
      "loss": 0.7695,
      "step": 7560
    },
    {
      "epoch": 4.059058420858361,
      "grad_norm": 0.8046875,
      "learning_rate": 4.237924682643657e-06,
      "loss": 0.8121,
      "step": 7561
    },
    {
      "epoch": 4.059595315593437,
      "grad_norm": 0.703125,
      "learning_rate": 4.2332215523124415e-06,
      "loss": 0.9495,
      "step": 7562
    },
    {
      "epoch": 4.060132210328512,
      "grad_norm": 0.70703125,
      "learning_rate": 4.228520791733636e-06,
      "loss": 0.8866,
      "step": 7563
    },
    {
      "epoch": 4.060669105063589,
      "grad_norm": 0.66015625,
      "learning_rate": 4.2238224014436635e-06,
      "loss": 0.7356,
      "step": 7564
    },
    {
      "epoch": 4.0612059997986645,
      "grad_norm": 0.68359375,
      "learning_rate": 4.219126381978658e-06,
      "loss": 0.8362,
      "step": 7565
    },
    {
      "epoch": 4.06174289453374,
      "grad_norm": 0.6953125,
      "learning_rate": 4.214432733874501e-06,
      "loss": 0.8295,
      "step": 7566
    },
    {
      "epoch": 4.062279789268817,
      "grad_norm": 0.546875,
      "learning_rate": 4.209741457666802e-06,
      "loss": 0.6949,
      "step": 7567
    },
    {
      "epoch": 4.062816684003892,
      "grad_norm": 0.703125,
      "learning_rate": 4.205052553890884e-06,
      "loss": 0.8235,
      "step": 7568
    },
    {
      "epoch": 4.063353578738968,
      "grad_norm": 0.59375,
      "learning_rate": 4.200366023081826e-06,
      "loss": 0.6671,
      "step": 7569
    },
    {
      "epoch": 4.063890473474045,
      "grad_norm": 0.66015625,
      "learning_rate": 4.195681865774406e-06,
      "loss": 0.7975,
      "step": 7570
    },
    {
      "epoch": 4.06442736820912,
      "grad_norm": 0.6953125,
      "learning_rate": 4.191000082503146e-06,
      "loss": 0.8361,
      "step": 7571
    },
    {
      "epoch": 4.064964262944197,
      "grad_norm": 0.68359375,
      "learning_rate": 4.186320673802302e-06,
      "loss": 0.7287,
      "step": 7572
    },
    {
      "epoch": 4.065501157679273,
      "grad_norm": 0.80859375,
      "learning_rate": 4.181643640205859e-06,
      "loss": 0.8191,
      "step": 7573
    },
    {
      "epoch": 4.066038052414348,
      "grad_norm": 0.60546875,
      "learning_rate": 4.176968982247514e-06,
      "loss": 0.8589,
      "step": 7574
    },
    {
      "epoch": 4.066574947149425,
      "grad_norm": 0.6953125,
      "learning_rate": 4.172296700460709e-06,
      "loss": 0.8424,
      "step": 7575
    },
    {
      "epoch": 4.0671118418845005,
      "grad_norm": 0.75,
      "learning_rate": 4.167626795378621e-06,
      "loss": 0.9578,
      "step": 7576
    },
    {
      "epoch": 4.067648736619576,
      "grad_norm": 0.6875,
      "learning_rate": 4.162959267534133e-06,
      "loss": 0.7207,
      "step": 7577
    },
    {
      "epoch": 4.068185631354653,
      "grad_norm": 0.59765625,
      "learning_rate": 4.158294117459868e-06,
      "loss": 0.7443,
      "step": 7578
    },
    {
      "epoch": 4.068722526089728,
      "grad_norm": 0.828125,
      "learning_rate": 4.153631345688183e-06,
      "loss": 0.8085,
      "step": 7579
    },
    {
      "epoch": 4.069259420824805,
      "grad_norm": 0.75390625,
      "learning_rate": 4.148970952751163e-06,
      "loss": 1.0337,
      "step": 7580
    },
    {
      "epoch": 4.069796315559881,
      "grad_norm": 0.671875,
      "learning_rate": 4.144312939180611e-06,
      "loss": 0.7783,
      "step": 7581
    },
    {
      "epoch": 4.070333210294956,
      "grad_norm": 0.7109375,
      "learning_rate": 4.139657305508074e-06,
      "loss": 0.879,
      "step": 7582
    },
    {
      "epoch": 4.070870105030033,
      "grad_norm": 0.7421875,
      "learning_rate": 4.135004052264807e-06,
      "loss": 0.7828,
      "step": 7583
    },
    {
      "epoch": 4.071406999765109,
      "grad_norm": 0.7109375,
      "learning_rate": 4.130353179981811e-06,
      "loss": 0.9673,
      "step": 7584
    },
    {
      "epoch": 4.071943894500184,
      "grad_norm": 0.7734375,
      "learning_rate": 4.125704689189819e-06,
      "loss": 0.8643,
      "step": 7585
    },
    {
      "epoch": 4.072480789235261,
      "grad_norm": 0.6171875,
      "learning_rate": 4.121058580419271e-06,
      "loss": 0.8345,
      "step": 7586
    },
    {
      "epoch": 4.0730176839703365,
      "grad_norm": 0.609375,
      "learning_rate": 4.116414854200343e-06,
      "loss": 0.7553,
      "step": 7587
    },
    {
      "epoch": 4.073554578705412,
      "grad_norm": 0.78125,
      "learning_rate": 4.111773511062949e-06,
      "loss": 0.909,
      "step": 7588
    },
    {
      "epoch": 4.074091473440489,
      "grad_norm": 0.73828125,
      "learning_rate": 4.1071345515367314e-06,
      "loss": 0.8026,
      "step": 7589
    },
    {
      "epoch": 4.074628368175564,
      "grad_norm": 0.6875,
      "learning_rate": 4.102497976151043e-06,
      "loss": 0.8353,
      "step": 7590
    },
    {
      "epoch": 4.075165262910641,
      "grad_norm": 0.65234375,
      "learning_rate": 4.09786378543498e-06,
      "loss": 0.9353,
      "step": 7591
    },
    {
      "epoch": 4.075702157645717,
      "grad_norm": 0.625,
      "learning_rate": 4.09323197991737e-06,
      "loss": 0.8854,
      "step": 7592
    },
    {
      "epoch": 4.076239052380792,
      "grad_norm": 0.84765625,
      "learning_rate": 4.0886025601267476e-06,
      "loss": 0.7122,
      "step": 7593
    },
    {
      "epoch": 4.076775947115869,
      "grad_norm": 0.73046875,
      "learning_rate": 4.083975526591402e-06,
      "loss": 0.8168,
      "step": 7594
    },
    {
      "epoch": 4.077312841850945,
      "grad_norm": 0.72265625,
      "learning_rate": 4.079350879839328e-06,
      "loss": 0.9631,
      "step": 7595
    },
    {
      "epoch": 4.07784973658602,
      "grad_norm": 0.59765625,
      "learning_rate": 4.07472862039825e-06,
      "loss": 0.8483,
      "step": 7596
    },
    {
      "epoch": 4.078386631321097,
      "grad_norm": 0.66796875,
      "learning_rate": 4.070108748795634e-06,
      "loss": 0.728,
      "step": 7597
    },
    {
      "epoch": 4.0789235260561725,
      "grad_norm": 0.6171875,
      "learning_rate": 4.065491265558671e-06,
      "loss": 0.666,
      "step": 7598
    },
    {
      "epoch": 4.079460420791249,
      "grad_norm": 0.640625,
      "learning_rate": 4.060876171214264e-06,
      "loss": 0.9069,
      "step": 7599
    },
    {
      "epoch": 4.079997315526325,
      "grad_norm": 0.7578125,
      "learning_rate": 4.056263466289059e-06,
      "loss": 0.9613,
      "step": 7600
    },
    {
      "epoch": 4.079997315526325,
      "eval_loss": 0.9027480483055115,
      "eval_runtime": 46.1054,
      "eval_samples_per_second": 20.865,
      "eval_steps_per_second": 20.865,
      "step": 7600
    },
    {
      "epoch": 4.0805342102614,
      "grad_norm": 0.6640625,
      "learning_rate": 4.0516531513094284e-06,
      "loss": 0.8165,
      "step": 7601
    },
    {
      "epoch": 4.081071104996477,
      "grad_norm": 0.6953125,
      "learning_rate": 4.047045226801463e-06,
      "loss": 0.9922,
      "step": 7602
    },
    {
      "epoch": 4.081607999731553,
      "grad_norm": 0.76171875,
      "learning_rate": 4.042439693290978e-06,
      "loss": 0.8781,
      "step": 7603
    },
    {
      "epoch": 4.082144894466628,
      "grad_norm": 0.796875,
      "learning_rate": 4.037836551303531e-06,
      "loss": 0.8919,
      "step": 7604
    },
    {
      "epoch": 4.082681789201705,
      "grad_norm": 0.76953125,
      "learning_rate": 4.0332358013644016e-06,
      "loss": 0.8596,
      "step": 7605
    },
    {
      "epoch": 4.083218683936781,
      "grad_norm": 0.55859375,
      "learning_rate": 4.028637443998587e-06,
      "loss": 0.731,
      "step": 7606
    },
    {
      "epoch": 4.083755578671856,
      "grad_norm": 0.75390625,
      "learning_rate": 4.024041479730819e-06,
      "loss": 0.7704,
      "step": 7607
    },
    {
      "epoch": 4.084292473406933,
      "grad_norm": 0.765625,
      "learning_rate": 4.019447909085564e-06,
      "loss": 0.8146,
      "step": 7608
    },
    {
      "epoch": 4.0848293681420085,
      "grad_norm": 1.03125,
      "learning_rate": 4.014856732586994e-06,
      "loss": 0.8638,
      "step": 7609
    },
    {
      "epoch": 4.085366262877085,
      "grad_norm": 0.6484375,
      "learning_rate": 4.010267950759025e-06,
      "loss": 0.7923,
      "step": 7610
    },
    {
      "epoch": 4.085903157612161,
      "grad_norm": 0.63671875,
      "learning_rate": 4.005681564125308e-06,
      "loss": 0.8261,
      "step": 7611
    },
    {
      "epoch": 4.086440052347236,
      "grad_norm": 0.74609375,
      "learning_rate": 4.001097573209187e-06,
      "loss": 0.8509,
      "step": 7612
    },
    {
      "epoch": 4.086976947082313,
      "grad_norm": 0.640625,
      "learning_rate": 3.996515978533758e-06,
      "loss": 0.7094,
      "step": 7613
    },
    {
      "epoch": 4.087513841817389,
      "grad_norm": 0.953125,
      "learning_rate": 3.9919367806218525e-06,
      "loss": 0.9203,
      "step": 7614
    },
    {
      "epoch": 4.088050736552464,
      "grad_norm": 0.62890625,
      "learning_rate": 3.987359979996e-06,
      "loss": 0.9241,
      "step": 7615
    },
    {
      "epoch": 4.088587631287541,
      "grad_norm": 0.6328125,
      "learning_rate": 3.982785577178477e-06,
      "loss": 0.8996,
      "step": 7616
    },
    {
      "epoch": 4.089124526022617,
      "grad_norm": 0.5625,
      "learning_rate": 3.9782135726912864e-06,
      "loss": 0.7987,
      "step": 7617
    },
    {
      "epoch": 4.089661420757693,
      "grad_norm": 0.6796875,
      "learning_rate": 3.97364396705614e-06,
      "loss": 0.8265,
      "step": 7618
    },
    {
      "epoch": 4.090198315492769,
      "grad_norm": 0.73046875,
      "learning_rate": 3.969076760794502e-06,
      "loss": 0.829,
      "step": 7619
    },
    {
      "epoch": 4.0907352102278445,
      "grad_norm": 0.6875,
      "learning_rate": 3.964511954427533e-06,
      "loss": 0.8665,
      "step": 7620
    },
    {
      "epoch": 4.091272104962921,
      "grad_norm": 0.71875,
      "learning_rate": 3.959949548476149e-06,
      "loss": 0.8798,
      "step": 7621
    },
    {
      "epoch": 4.091808999697997,
      "grad_norm": 0.69921875,
      "learning_rate": 3.955389543460969e-06,
      "loss": 0.7219,
      "step": 7622
    },
    {
      "epoch": 4.092345894433072,
      "grad_norm": 0.6015625,
      "learning_rate": 3.950831939902352e-06,
      "loss": 0.6613,
      "step": 7623
    },
    {
      "epoch": 4.092882789168149,
      "grad_norm": 0.83984375,
      "learning_rate": 3.946276738320373e-06,
      "loss": 0.906,
      "step": 7624
    },
    {
      "epoch": 4.093419683903225,
      "grad_norm": 0.71484375,
      "learning_rate": 3.941723939234843e-06,
      "loss": 0.7586,
      "step": 7625
    },
    {
      "epoch": 4.0939565786383,
      "grad_norm": 0.8515625,
      "learning_rate": 3.937173543165298e-06,
      "loss": 1.0252,
      "step": 7626
    },
    {
      "epoch": 4.094493473373377,
      "grad_norm": 0.69921875,
      "learning_rate": 3.932625550630986e-06,
      "loss": 1.0482,
      "step": 7627
    },
    {
      "epoch": 4.095030368108453,
      "grad_norm": 0.68359375,
      "learning_rate": 3.928079962150899e-06,
      "loss": 0.8752,
      "step": 7628
    },
    {
      "epoch": 4.095567262843529,
      "grad_norm": 0.7421875,
      "learning_rate": 3.923536778243739e-06,
      "loss": 0.921,
      "step": 7629
    },
    {
      "epoch": 4.096104157578605,
      "grad_norm": 0.73046875,
      "learning_rate": 3.918995999427949e-06,
      "loss": 0.8037,
      "step": 7630
    },
    {
      "epoch": 4.0966410523136805,
      "grad_norm": 0.75,
      "learning_rate": 3.914457626221679e-06,
      "loss": 0.7936,
      "step": 7631
    },
    {
      "epoch": 4.097177947048757,
      "grad_norm": 0.67578125,
      "learning_rate": 3.90992165914282e-06,
      "loss": 0.9538,
      "step": 7632
    },
    {
      "epoch": 4.097714841783833,
      "grad_norm": 0.69140625,
      "learning_rate": 3.9053880987089926e-06,
      "loss": 0.8611,
      "step": 7633
    },
    {
      "epoch": 4.098251736518908,
      "grad_norm": 0.6953125,
      "learning_rate": 3.900856945437517e-06,
      "loss": 0.87,
      "step": 7634
    },
    {
      "epoch": 4.098788631253985,
      "grad_norm": 0.6015625,
      "learning_rate": 3.8963281998454615e-06,
      "loss": 0.7026,
      "step": 7635
    },
    {
      "epoch": 4.099325525989061,
      "grad_norm": 0.640625,
      "learning_rate": 3.891801862449629e-06,
      "loss": 0.7265,
      "step": 7636
    },
    {
      "epoch": 4.099862420724137,
      "grad_norm": 0.9375,
      "learning_rate": 3.887277933766506e-06,
      "loss": 0.8535,
      "step": 7637
    },
    {
      "epoch": 4.100399315459213,
      "grad_norm": 0.76171875,
      "learning_rate": 3.882756414312342e-06,
      "loss": 0.837,
      "step": 7638
    },
    {
      "epoch": 4.1009362101942886,
      "grad_norm": 0.60546875,
      "learning_rate": 3.878237304603102e-06,
      "loss": 0.705,
      "step": 7639
    },
    {
      "epoch": 4.101473104929365,
      "grad_norm": 0.6875,
      "learning_rate": 3.873720605154468e-06,
      "loss": 1.023,
      "step": 7640
    },
    {
      "epoch": 4.102009999664441,
      "grad_norm": 0.71875,
      "learning_rate": 3.869206316481855e-06,
      "loss": 1.0272,
      "step": 7641
    },
    {
      "epoch": 4.1025468943995165,
      "grad_norm": 0.70703125,
      "learning_rate": 3.864694439100405e-06,
      "loss": 0.8219,
      "step": 7642
    },
    {
      "epoch": 4.103083789134593,
      "grad_norm": 0.7109375,
      "learning_rate": 3.860184973524972e-06,
      "loss": 0.7704,
      "step": 7643
    },
    {
      "epoch": 4.103620683869669,
      "grad_norm": 0.796875,
      "learning_rate": 3.855677920270154e-06,
      "loss": 0.9095,
      "step": 7644
    },
    {
      "epoch": 4.104157578604744,
      "grad_norm": 0.625,
      "learning_rate": 3.851173279850251e-06,
      "loss": 0.7077,
      "step": 7645
    },
    {
      "epoch": 4.104694473339821,
      "grad_norm": 0.76171875,
      "learning_rate": 3.8466710527793105e-06,
      "loss": 0.6791,
      "step": 7646
    },
    {
      "epoch": 4.105231368074897,
      "grad_norm": 0.765625,
      "learning_rate": 3.842171239571082e-06,
      "loss": 0.8216,
      "step": 7647
    },
    {
      "epoch": 4.105768262809973,
      "grad_norm": 0.6875,
      "learning_rate": 3.837673840739059e-06,
      "loss": 0.8582,
      "step": 7648
    },
    {
      "epoch": 4.106305157545049,
      "grad_norm": 0.6953125,
      "learning_rate": 3.833178856796452e-06,
      "loss": 0.8844,
      "step": 7649
    },
    {
      "epoch": 4.1068420522801246,
      "grad_norm": 0.7109375,
      "learning_rate": 3.8286862882561916e-06,
      "loss": 0.7074,
      "step": 7650
    },
    {
      "epoch": 4.1068420522801246,
      "eval_loss": 0.9025216698646545,
      "eval_runtime": 46.5886,
      "eval_samples_per_second": 20.649,
      "eval_steps_per_second": 20.649,
      "step": 7650
    },
    {
      "epoch": 4.107378947015201,
      "grad_norm": 0.6953125,
      "learning_rate": 3.824196135630936e-06,
      "loss": 0.6797,
      "step": 7651
    },
    {
      "epoch": 4.107915841750277,
      "grad_norm": 0.671875,
      "learning_rate": 3.819708399433078e-06,
      "loss": 0.8142,
      "step": 7652
    },
    {
      "epoch": 4.1084527364853525,
      "grad_norm": 0.671875,
      "learning_rate": 3.815223080174718e-06,
      "loss": 0.7858,
      "step": 7653
    },
    {
      "epoch": 4.108989631220429,
      "grad_norm": 0.765625,
      "learning_rate": 3.810740178367683e-06,
      "loss": 0.9258,
      "step": 7654
    },
    {
      "epoch": 4.109526525955505,
      "grad_norm": 0.64453125,
      "learning_rate": 3.8062596945235396e-06,
      "loss": 0.6984,
      "step": 7655
    },
    {
      "epoch": 4.110063420690581,
      "grad_norm": 0.6875,
      "learning_rate": 3.8017816291535556e-06,
      "loss": 0.8492,
      "step": 7656
    },
    {
      "epoch": 4.110600315425657,
      "grad_norm": 0.69921875,
      "learning_rate": 3.7973059827687425e-06,
      "loss": 0.8895,
      "step": 7657
    },
    {
      "epoch": 4.111137210160733,
      "grad_norm": 0.74609375,
      "learning_rate": 3.792832755879833e-06,
      "loss": 0.8753,
      "step": 7658
    },
    {
      "epoch": 4.111674104895809,
      "grad_norm": 0.63671875,
      "learning_rate": 3.7883619489972653e-06,
      "loss": 0.7956,
      "step": 7659
    },
    {
      "epoch": 4.112210999630885,
      "grad_norm": 0.6953125,
      "learning_rate": 3.7838935626312242e-06,
      "loss": 0.8918,
      "step": 7660
    },
    {
      "epoch": 4.1127478943659606,
      "grad_norm": 0.75,
      "learning_rate": 3.7794275972916125e-06,
      "loss": 0.9082,
      "step": 7661
    },
    {
      "epoch": 4.113284789101037,
      "grad_norm": 0.703125,
      "learning_rate": 3.7749640534880504e-06,
      "loss": 0.8224,
      "step": 7662
    },
    {
      "epoch": 4.113821683836113,
      "grad_norm": 0.89453125,
      "learning_rate": 3.770502931729877e-06,
      "loss": 0.9054,
      "step": 7663
    },
    {
      "epoch": 4.1143585785711885,
      "grad_norm": 0.6328125,
      "learning_rate": 3.766044232526175e-06,
      "loss": 0.7902,
      "step": 7664
    },
    {
      "epoch": 4.114895473306265,
      "grad_norm": 0.61328125,
      "learning_rate": 3.7615879563857257e-06,
      "loss": 0.8439,
      "step": 7665
    },
    {
      "epoch": 4.115432368041341,
      "grad_norm": 0.73046875,
      "learning_rate": 3.757134103817053e-06,
      "loss": 0.941,
      "step": 7666
    },
    {
      "epoch": 4.115969262776417,
      "grad_norm": 0.8515625,
      "learning_rate": 3.752682675328406e-06,
      "loss": 0.6705,
      "step": 7667
    },
    {
      "epoch": 4.116506157511493,
      "grad_norm": 0.703125,
      "learning_rate": 3.7482336714277337e-06,
      "loss": 0.8981,
      "step": 7668
    },
    {
      "epoch": 4.117043052246569,
      "grad_norm": 0.8203125,
      "learning_rate": 3.7437870926227304e-06,
      "loss": 0.885,
      "step": 7669
    },
    {
      "epoch": 4.117579946981645,
      "grad_norm": 0.8359375,
      "learning_rate": 3.739342939420815e-06,
      "loss": 0.8488,
      "step": 7670
    },
    {
      "epoch": 4.118116841716721,
      "grad_norm": 0.734375,
      "learning_rate": 3.7349012123291144e-06,
      "loss": 0.7188,
      "step": 7671
    },
    {
      "epoch": 4.1186537364517966,
      "grad_norm": 0.6875,
      "learning_rate": 3.7304619118544827e-06,
      "loss": 0.8741,
      "step": 7672
    },
    {
      "epoch": 4.119190631186873,
      "grad_norm": 0.69140625,
      "learning_rate": 3.7260250385035e-06,
      "loss": 0.855,
      "step": 7673
    },
    {
      "epoch": 4.119727525921949,
      "grad_norm": 0.640625,
      "learning_rate": 3.7215905927824812e-06,
      "loss": 0.8032,
      "step": 7674
    },
    {
      "epoch": 4.120264420657025,
      "grad_norm": 0.9609375,
      "learning_rate": 3.7171585751974408e-06,
      "loss": 0.8138,
      "step": 7675
    },
    {
      "epoch": 4.120801315392101,
      "grad_norm": 0.796875,
      "learning_rate": 3.7127289862541297e-06,
      "loss": 0.8487,
      "step": 7676
    },
    {
      "epoch": 4.121338210127177,
      "grad_norm": 0.71875,
      "learning_rate": 3.70830182645803e-06,
      "loss": 0.9556,
      "step": 7677
    },
    {
      "epoch": 4.121875104862253,
      "grad_norm": 0.671875,
      "learning_rate": 3.7038770963143286e-06,
      "loss": 0.718,
      "step": 7678
    },
    {
      "epoch": 4.122411999597329,
      "grad_norm": 0.8984375,
      "learning_rate": 3.6994547963279392e-06,
      "loss": 0.7813,
      "step": 7679
    },
    {
      "epoch": 4.122948894332405,
      "grad_norm": 0.61328125,
      "learning_rate": 3.695034927003513e-06,
      "loss": 0.9341,
      "step": 7680
    },
    {
      "epoch": 4.123485789067481,
      "grad_norm": 0.7890625,
      "learning_rate": 3.6906174888453992e-06,
      "loss": 0.692,
      "step": 7681
    },
    {
      "epoch": 4.124022683802557,
      "grad_norm": 1.140625,
      "learning_rate": 3.6862024823576926e-06,
      "loss": 0.8553,
      "step": 7682
    },
    {
      "epoch": 4.1245595785376326,
      "grad_norm": 0.63671875,
      "learning_rate": 3.6817899080442057e-06,
      "loss": 0.806,
      "step": 7683
    },
    {
      "epoch": 4.125096473272709,
      "grad_norm": 0.60546875,
      "learning_rate": 3.677379766408459e-06,
      "loss": 0.6916,
      "step": 7684
    },
    {
      "epoch": 4.125633368007785,
      "grad_norm": 0.71875,
      "learning_rate": 3.672972057953711e-06,
      "loss": 0.8773,
      "step": 7685
    },
    {
      "epoch": 4.126170262742861,
      "grad_norm": 0.7265625,
      "learning_rate": 3.668566783182939e-06,
      "loss": 0.8851,
      "step": 7686
    },
    {
      "epoch": 4.126707157477937,
      "grad_norm": 0.7421875,
      "learning_rate": 3.664163942598839e-06,
      "loss": 0.9127,
      "step": 7687
    },
    {
      "epoch": 4.127244052213013,
      "grad_norm": 0.62890625,
      "learning_rate": 3.659763536703825e-06,
      "loss": 0.7889,
      "step": 7688
    },
    {
      "epoch": 4.127780946948089,
      "grad_norm": 0.703125,
      "learning_rate": 3.6553655660000436e-06,
      "loss": 0.9252,
      "step": 7689
    },
    {
      "epoch": 4.128317841683165,
      "grad_norm": 0.8125,
      "learning_rate": 3.6509700309893618e-06,
      "loss": 0.8603,
      "step": 7690
    },
    {
      "epoch": 4.128854736418241,
      "grad_norm": 0.7265625,
      "learning_rate": 3.6465769321733594e-06,
      "loss": 0.7195,
      "step": 7691
    },
    {
      "epoch": 4.129391631153317,
      "grad_norm": 0.7265625,
      "learning_rate": 3.6421862700533495e-06,
      "loss": 0.9742,
      "step": 7692
    },
    {
      "epoch": 4.129928525888393,
      "grad_norm": 0.9765625,
      "learning_rate": 3.6377980451303666e-06,
      "loss": 0.8072,
      "step": 7693
    },
    {
      "epoch": 4.130465420623469,
      "grad_norm": 0.609375,
      "learning_rate": 3.63341225790515e-06,
      "loss": 0.9739,
      "step": 7694
    },
    {
      "epoch": 4.131002315358545,
      "grad_norm": 0.828125,
      "learning_rate": 3.6290289088781874e-06,
      "loss": 0.7622,
      "step": 7695
    },
    {
      "epoch": 4.131539210093621,
      "grad_norm": 0.5859375,
      "learning_rate": 3.6246479985496658e-06,
      "loss": 0.6696,
      "step": 7696
    },
    {
      "epoch": 4.132076104828697,
      "grad_norm": 0.75,
      "learning_rate": 3.6202695274195007e-06,
      "loss": 0.8104,
      "step": 7697
    },
    {
      "epoch": 4.132612999563773,
      "grad_norm": 0.66015625,
      "learning_rate": 3.6158934959873353e-06,
      "loss": 0.8761,
      "step": 7698
    },
    {
      "epoch": 4.133149894298849,
      "grad_norm": 0.72265625,
      "learning_rate": 3.611519904752533e-06,
      "loss": 0.8477,
      "step": 7699
    },
    {
      "epoch": 4.133686789033925,
      "grad_norm": 0.6796875,
      "learning_rate": 3.607148754214168e-06,
      "loss": 0.7589,
      "step": 7700
    },
    {
      "epoch": 4.133686789033925,
      "eval_loss": 0.9027608036994934,
      "eval_runtime": 46.5247,
      "eval_samples_per_second": 20.677,
      "eval_steps_per_second": 20.677,
      "step": 7700
    },
    {
      "epoch": 4.134223683769001,
      "grad_norm": 0.7890625,
      "learning_rate": 3.6027800448710515e-06,
      "loss": 0.857,
      "step": 7701
    },
    {
      "epoch": 4.134760578504077,
      "grad_norm": 0.6171875,
      "learning_rate": 3.5984137772217096e-06,
      "loss": 0.896,
      "step": 7702
    },
    {
      "epoch": 4.135297473239153,
      "grad_norm": 0.62109375,
      "learning_rate": 3.594049951764383e-06,
      "loss": 0.8503,
      "step": 7703
    },
    {
      "epoch": 4.135834367974229,
      "grad_norm": 0.76171875,
      "learning_rate": 3.5896885689970376e-06,
      "loss": 0.7745,
      "step": 7704
    },
    {
      "epoch": 4.136371262709305,
      "grad_norm": 0.609375,
      "learning_rate": 3.5853296294173666e-06,
      "loss": 0.6519,
      "step": 7705
    },
    {
      "epoch": 4.136908157444381,
      "grad_norm": 0.73046875,
      "learning_rate": 3.580973133522783e-06,
      "loss": 0.7922,
      "step": 7706
    },
    {
      "epoch": 4.137445052179457,
      "grad_norm": 0.640625,
      "learning_rate": 3.5766190818104118e-06,
      "loss": 0.7362,
      "step": 7707
    },
    {
      "epoch": 4.137981946914533,
      "grad_norm": 0.6953125,
      "learning_rate": 3.5722674747771134e-06,
      "loss": 0.7191,
      "step": 7708
    },
    {
      "epoch": 4.138518841649609,
      "grad_norm": 0.61328125,
      "learning_rate": 3.5679183129194493e-06,
      "loss": 0.7361,
      "step": 7709
    },
    {
      "epoch": 4.139055736384685,
      "grad_norm": 0.6953125,
      "learning_rate": 3.5635715967337223e-06,
      "loss": 0.7144,
      "step": 7710
    },
    {
      "epoch": 4.139592631119761,
      "grad_norm": 0.71484375,
      "learning_rate": 3.5592273267159525e-06,
      "loss": 0.8846,
      "step": 7711
    },
    {
      "epoch": 4.140129525854837,
      "grad_norm": 0.73828125,
      "learning_rate": 3.554885503361871e-06,
      "loss": 0.8658,
      "step": 7712
    },
    {
      "epoch": 4.140666420589913,
      "grad_norm": 0.74609375,
      "learning_rate": 3.550546127166926e-06,
      "loss": 0.8837,
      "step": 7713
    },
    {
      "epoch": 4.141203315324989,
      "grad_norm": 0.6640625,
      "learning_rate": 3.5462091986263048e-06,
      "loss": 0.7683,
      "step": 7714
    },
    {
      "epoch": 4.141740210060065,
      "grad_norm": 0.703125,
      "learning_rate": 3.5418747182349084e-06,
      "loss": 0.8164,
      "step": 7715
    },
    {
      "epoch": 4.142277104795141,
      "grad_norm": 0.67578125,
      "learning_rate": 3.5375426864873494e-06,
      "loss": 0.8122,
      "step": 7716
    },
    {
      "epoch": 4.142813999530217,
      "grad_norm": 0.796875,
      "learning_rate": 3.5332131038779663e-06,
      "loss": 0.8984,
      "step": 7717
    },
    {
      "epoch": 4.143350894265293,
      "grad_norm": 0.703125,
      "learning_rate": 3.5288859709008305e-06,
      "loss": 0.7758,
      "step": 7718
    },
    {
      "epoch": 4.143887789000369,
      "grad_norm": 0.90234375,
      "learning_rate": 3.52456128804971e-06,
      "loss": 0.9521,
      "step": 7719
    },
    {
      "epoch": 4.144424683735445,
      "grad_norm": 0.5703125,
      "learning_rate": 3.520239055818114e-06,
      "loss": 0.694,
      "step": 7720
    },
    {
      "epoch": 4.144961578470521,
      "grad_norm": 0.7109375,
      "learning_rate": 3.515919274699264e-06,
      "loss": 0.9689,
      "step": 7721
    },
    {
      "epoch": 4.145498473205597,
      "grad_norm": 0.609375,
      "learning_rate": 3.511601945186091e-06,
      "loss": 0.7541,
      "step": 7722
    },
    {
      "epoch": 4.146035367940673,
      "grad_norm": 0.75,
      "learning_rate": 3.5072870677712642e-06,
      "loss": 1.0244,
      "step": 7723
    },
    {
      "epoch": 4.1465722626757495,
      "grad_norm": 0.8125,
      "learning_rate": 3.5029746429471737e-06,
      "loss": 0.8331,
      "step": 7724
    },
    {
      "epoch": 4.147109157410825,
      "grad_norm": 0.75390625,
      "learning_rate": 3.4986646712059064e-06,
      "loss": 0.9679,
      "step": 7725
    },
    {
      "epoch": 4.147646052145901,
      "grad_norm": 0.69921875,
      "learning_rate": 3.494357153039296e-06,
      "loss": 0.9258,
      "step": 7726
    },
    {
      "epoch": 4.148182946880977,
      "grad_norm": 0.640625,
      "learning_rate": 3.490052088938883e-06,
      "loss": 0.7397,
      "step": 7727
    },
    {
      "epoch": 4.148719841616053,
      "grad_norm": 0.6484375,
      "learning_rate": 3.4857494793959307e-06,
      "loss": 0.7623,
      "step": 7728
    },
    {
      "epoch": 4.149256736351129,
      "grad_norm": 0.69140625,
      "learning_rate": 3.4814493249014116e-06,
      "loss": 0.7676,
      "step": 7729
    },
    {
      "epoch": 4.149793631086205,
      "grad_norm": 0.64453125,
      "learning_rate": 3.4771516259460347e-06,
      "loss": 0.7425,
      "step": 7730
    },
    {
      "epoch": 4.150330525821281,
      "grad_norm": 0.68359375,
      "learning_rate": 3.472856383020229e-06,
      "loss": 0.7885,
      "step": 7731
    },
    {
      "epoch": 4.150867420556358,
      "grad_norm": 0.85546875,
      "learning_rate": 3.4685635966141245e-06,
      "loss": 1.0595,
      "step": 7732
    },
    {
      "epoch": 4.151404315291433,
      "grad_norm": 0.80078125,
      "learning_rate": 3.4642732672175855e-06,
      "loss": 0.9652,
      "step": 7733
    },
    {
      "epoch": 4.151941210026509,
      "grad_norm": 0.65625,
      "learning_rate": 3.4599853953202004e-06,
      "loss": 0.8826,
      "step": 7734
    },
    {
      "epoch": 4.1524781047615855,
      "grad_norm": 1.203125,
      "learning_rate": 3.455699981411259e-06,
      "loss": 0.8383,
      "step": 7735
    },
    {
      "epoch": 4.153014999496661,
      "grad_norm": 0.70703125,
      "learning_rate": 3.4514170259797862e-06,
      "loss": 0.7869,
      "step": 7736
    },
    {
      "epoch": 4.153551894231737,
      "grad_norm": 0.734375,
      "learning_rate": 3.4471365295145304e-06,
      "loss": 0.9131,
      "step": 7737
    },
    {
      "epoch": 4.154088788966813,
      "grad_norm": 0.76171875,
      "learning_rate": 3.4428584925039337e-06,
      "loss": 0.7697,
      "step": 7738
    },
    {
      "epoch": 4.154625683701889,
      "grad_norm": 0.6875,
      "learning_rate": 3.4385829154361783e-06,
      "loss": 0.8784,
      "step": 7739
    },
    {
      "epoch": 4.155162578436965,
      "grad_norm": 0.61328125,
      "learning_rate": 3.4343097987991735e-06,
      "loss": 0.7595,
      "step": 7740
    },
    {
      "epoch": 4.155699473172041,
      "grad_norm": 0.65234375,
      "learning_rate": 3.430039143080524e-06,
      "loss": 0.7975,
      "step": 7741
    },
    {
      "epoch": 4.156236367907117,
      "grad_norm": 0.66796875,
      "learning_rate": 3.4257709487675677e-06,
      "loss": 0.8956,
      "step": 7742
    },
    {
      "epoch": 4.156773262642194,
      "grad_norm": 0.64453125,
      "learning_rate": 3.421505216347368e-06,
      "loss": 0.9228,
      "step": 7743
    },
    {
      "epoch": 4.157310157377269,
      "grad_norm": 0.66796875,
      "learning_rate": 3.4172419463066848e-06,
      "loss": 0.8605,
      "step": 7744
    },
    {
      "epoch": 4.157847052112345,
      "grad_norm": 0.63671875,
      "learning_rate": 3.4129811391320267e-06,
      "loss": 0.8328,
      "step": 7745
    },
    {
      "epoch": 4.1583839468474215,
      "grad_norm": 0.6484375,
      "learning_rate": 3.4087227953095934e-06,
      "loss": 0.765,
      "step": 7746
    },
    {
      "epoch": 4.158920841582497,
      "grad_norm": 0.67578125,
      "learning_rate": 3.4044669153253243e-06,
      "loss": 0.8829,
      "step": 7747
    },
    {
      "epoch": 4.159457736317573,
      "grad_norm": 0.75,
      "learning_rate": 3.4002134996648616e-06,
      "loss": 0.9027,
      "step": 7748
    },
    {
      "epoch": 4.159994631052649,
      "grad_norm": 0.64453125,
      "learning_rate": 3.3959625488135805e-06,
      "loss": 0.8091,
      "step": 7749
    },
    {
      "epoch": 4.160531525787725,
      "grad_norm": 0.65625,
      "learning_rate": 3.3917140632565624e-06,
      "loss": 0.8279,
      "step": 7750
    },
    {
      "epoch": 4.160531525787725,
      "eval_loss": 0.9027578234672546,
      "eval_runtime": 46.8161,
      "eval_samples_per_second": 20.548,
      "eval_steps_per_second": 20.548,
      "step": 7750
    },
    {
      "epoch": 4.161068420522801,
      "grad_norm": 0.71484375,
      "learning_rate": 3.3874680434786172e-06,
      "loss": 0.6601,
      "step": 7751
    },
    {
      "epoch": 4.161605315257877,
      "grad_norm": 0.734375,
      "learning_rate": 3.3832244899642706e-06,
      "loss": 0.9747,
      "step": 7752
    },
    {
      "epoch": 4.162142209992953,
      "grad_norm": 0.703125,
      "learning_rate": 3.378983403197761e-06,
      "loss": 0.7435,
      "step": 7753
    },
    {
      "epoch": 4.16267910472803,
      "grad_norm": 0.6796875,
      "learning_rate": 3.3747447836630587e-06,
      "loss": 0.836,
      "step": 7754
    },
    {
      "epoch": 4.163215999463105,
      "grad_norm": 0.6328125,
      "learning_rate": 3.3705086318438327e-06,
      "loss": 0.8198,
      "step": 7755
    },
    {
      "epoch": 4.163752894198181,
      "grad_norm": 0.65625,
      "learning_rate": 3.366274948223494e-06,
      "loss": 0.8979,
      "step": 7756
    },
    {
      "epoch": 4.1642897889332575,
      "grad_norm": 0.81640625,
      "learning_rate": 3.3620437332851446e-06,
      "loss": 0.7899,
      "step": 7757
    },
    {
      "epoch": 4.164826683668333,
      "grad_norm": 0.8671875,
      "learning_rate": 3.3578149875116288e-06,
      "loss": 1.0397,
      "step": 7758
    },
    {
      "epoch": 4.165363578403409,
      "grad_norm": 0.60546875,
      "learning_rate": 3.3535887113855052e-06,
      "loss": 0.8326,
      "step": 7759
    },
    {
      "epoch": 4.165900473138485,
      "grad_norm": 0.61328125,
      "learning_rate": 3.3493649053890326e-06,
      "loss": 0.713,
      "step": 7760
    },
    {
      "epoch": 4.166437367873561,
      "grad_norm": 0.71484375,
      "learning_rate": 3.3451435700042083e-06,
      "loss": 0.8868,
      "step": 7761
    },
    {
      "epoch": 4.166974262608638,
      "grad_norm": 0.70703125,
      "learning_rate": 3.34092470571275e-06,
      "loss": 0.9271,
      "step": 7762
    },
    {
      "epoch": 4.167511157343713,
      "grad_norm": 0.640625,
      "learning_rate": 3.336708312996062e-06,
      "loss": 0.8128,
      "step": 7763
    },
    {
      "epoch": 4.168048052078789,
      "grad_norm": 0.77734375,
      "learning_rate": 3.3324943923352974e-06,
      "loss": 0.7175,
      "step": 7764
    },
    {
      "epoch": 4.168584946813866,
      "grad_norm": 0.7734375,
      "learning_rate": 3.328282944211328e-06,
      "loss": 0.9151,
      "step": 7765
    },
    {
      "epoch": 4.169121841548941,
      "grad_norm": 0.59375,
      "learning_rate": 3.3240739691047157e-06,
      "loss": 0.6801,
      "step": 7766
    },
    {
      "epoch": 4.169658736284017,
      "grad_norm": 0.71484375,
      "learning_rate": 3.3198674674957684e-06,
      "loss": 0.6921,
      "step": 7767
    },
    {
      "epoch": 4.1701956310190935,
      "grad_norm": 0.80078125,
      "learning_rate": 3.315663439864508e-06,
      "loss": 0.8124,
      "step": 7768
    },
    {
      "epoch": 4.170732525754169,
      "grad_norm": 0.765625,
      "learning_rate": 3.3114618866906503e-06,
      "loss": 0.7704,
      "step": 7769
    },
    {
      "epoch": 4.171269420489246,
      "grad_norm": 0.66015625,
      "learning_rate": 3.3072628084536593e-06,
      "loss": 0.7589,
      "step": 7770
    },
    {
      "epoch": 4.171806315224321,
      "grad_norm": 0.75390625,
      "learning_rate": 3.3030662056326933e-06,
      "loss": 1.0278,
      "step": 7771
    },
    {
      "epoch": 4.172343209959397,
      "grad_norm": 0.68359375,
      "learning_rate": 3.2988720787066495e-06,
      "loss": 0.7851,
      "step": 7772
    },
    {
      "epoch": 4.172880104694474,
      "grad_norm": 0.66015625,
      "learning_rate": 3.2946804281541178e-06,
      "loss": 0.7814,
      "step": 7773
    },
    {
      "epoch": 4.173416999429549,
      "grad_norm": 0.75390625,
      "learning_rate": 3.290491254453426e-06,
      "loss": 0.8635,
      "step": 7774
    },
    {
      "epoch": 4.173953894164625,
      "grad_norm": 0.6640625,
      "learning_rate": 3.2863045580826145e-06,
      "loss": 0.8788,
      "step": 7775
    },
    {
      "epoch": 4.174490788899702,
      "grad_norm": 0.703125,
      "learning_rate": 3.282120339519429e-06,
      "loss": 0.8501,
      "step": 7776
    },
    {
      "epoch": 4.175027683634777,
      "grad_norm": 0.6640625,
      "learning_rate": 3.277938599241348e-06,
      "loss": 0.9007,
      "step": 7777
    },
    {
      "epoch": 4.175564578369853,
      "grad_norm": 0.56640625,
      "learning_rate": 3.273759337725568e-06,
      "loss": 0.7333,
      "step": 7778
    },
    {
      "epoch": 4.1761014731049295,
      "grad_norm": 0.75390625,
      "learning_rate": 3.2695825554489854e-06,
      "loss": 0.8948,
      "step": 7779
    },
    {
      "epoch": 4.176638367840005,
      "grad_norm": 0.7265625,
      "learning_rate": 3.265408252888222e-06,
      "loss": 0.7789,
      "step": 7780
    },
    {
      "epoch": 4.177175262575082,
      "grad_norm": 0.67578125,
      "learning_rate": 3.2612364305196293e-06,
      "loss": 0.8759,
      "step": 7781
    },
    {
      "epoch": 4.177712157310157,
      "grad_norm": 0.6796875,
      "learning_rate": 3.2570670888192547e-06,
      "loss": 0.8162,
      "step": 7782
    },
    {
      "epoch": 4.178249052045233,
      "grad_norm": 0.62109375,
      "learning_rate": 3.2529002282628756e-06,
      "loss": 0.9651,
      "step": 7783
    },
    {
      "epoch": 4.17878594678031,
      "grad_norm": 0.8359375,
      "learning_rate": 3.24873584932599e-06,
      "loss": 0.9906,
      "step": 7784
    },
    {
      "epoch": 4.179322841515385,
      "grad_norm": 0.640625,
      "learning_rate": 3.2445739524837953e-06,
      "loss": 0.8634,
      "step": 7785
    },
    {
      "epoch": 4.179859736250461,
      "grad_norm": 0.65234375,
      "learning_rate": 3.2404145382112254e-06,
      "loss": 0.8493,
      "step": 7786
    },
    {
      "epoch": 4.180396630985538,
      "grad_norm": 0.703125,
      "learning_rate": 3.2362576069829233e-06,
      "loss": 0.7157,
      "step": 7787
    },
    {
      "epoch": 4.180933525720613,
      "grad_norm": 0.61328125,
      "learning_rate": 3.232103159273245e-06,
      "loss": 0.9648,
      "step": 7788
    },
    {
      "epoch": 4.181470420455689,
      "grad_norm": 0.8359375,
      "learning_rate": 3.2279511955562588e-06,
      "loss": 0.9492,
      "step": 7789
    },
    {
      "epoch": 4.1820073151907655,
      "grad_norm": 0.62890625,
      "learning_rate": 3.223801716305766e-06,
      "loss": 0.8544,
      "step": 7790
    },
    {
      "epoch": 4.182544209925841,
      "grad_norm": 0.61328125,
      "learning_rate": 3.219654721995266e-06,
      "loss": 0.7633,
      "step": 7791
    },
    {
      "epoch": 4.183081104660918,
      "grad_norm": 0.75390625,
      "learning_rate": 3.2155102130979886e-06,
      "loss": 0.7513,
      "step": 7792
    },
    {
      "epoch": 4.183617999395993,
      "grad_norm": 0.65625,
      "learning_rate": 3.2113681900868775e-06,
      "loss": 0.8225,
      "step": 7793
    },
    {
      "epoch": 4.184154894131069,
      "grad_norm": 0.65625,
      "learning_rate": 3.2072286534345826e-06,
      "loss": 0.8405,
      "step": 7794
    },
    {
      "epoch": 4.184691788866146,
      "grad_norm": 0.61328125,
      "learning_rate": 3.2030916036134865e-06,
      "loss": 0.6495,
      "step": 7795
    },
    {
      "epoch": 4.185228683601221,
      "grad_norm": 0.59375,
      "learning_rate": 3.198957041095671e-06,
      "loss": 0.6269,
      "step": 7796
    },
    {
      "epoch": 4.185765578336297,
      "grad_norm": 0.71875,
      "learning_rate": 3.194824966352947e-06,
      "loss": 0.8509,
      "step": 7797
    },
    {
      "epoch": 4.186302473071374,
      "grad_norm": 0.76953125,
      "learning_rate": 3.190695379856831e-06,
      "loss": 0.7805,
      "step": 7798
    },
    {
      "epoch": 4.186839367806449,
      "grad_norm": 0.94140625,
      "learning_rate": 3.186568282078564e-06,
      "loss": 0.784,
      "step": 7799
    },
    {
      "epoch": 4.187376262541526,
      "grad_norm": 0.81640625,
      "learning_rate": 3.182443673489105e-06,
      "loss": 0.7417,
      "step": 7800
    },
    {
      "epoch": 4.187376262541526,
      "eval_loss": 0.902660608291626,
      "eval_runtime": 45.6911,
      "eval_samples_per_second": 21.054,
      "eval_steps_per_second": 21.054,
      "step": 7800
    },
    {
      "epoch": 4.1879131572766015,
      "grad_norm": 0.79296875,
      "learning_rate": 3.178321554559116e-06,
      "loss": 0.9144,
      "step": 7801
    },
    {
      "epoch": 4.188450052011677,
      "grad_norm": 0.70703125,
      "learning_rate": 3.174201925758988e-06,
      "loss": 0.7681,
      "step": 7802
    },
    {
      "epoch": 4.188986946746754,
      "grad_norm": 0.63671875,
      "learning_rate": 3.170084787558825e-06,
      "loss": 0.7775,
      "step": 7803
    },
    {
      "epoch": 4.189523841481829,
      "grad_norm": 0.68359375,
      "learning_rate": 3.1659701404284423e-06,
      "loss": 0.79,
      "step": 7804
    },
    {
      "epoch": 4.190060736216905,
      "grad_norm": 0.5859375,
      "learning_rate": 3.161857984837366e-06,
      "loss": 0.7043,
      "step": 7805
    },
    {
      "epoch": 4.190597630951982,
      "grad_norm": 0.76953125,
      "learning_rate": 3.157748321254858e-06,
      "loss": 0.8262,
      "step": 7806
    },
    {
      "epoch": 4.191134525687057,
      "grad_norm": 0.73046875,
      "learning_rate": 3.153641150149869e-06,
      "loss": 0.9911,
      "step": 7807
    },
    {
      "epoch": 4.191671420422134,
      "grad_norm": 0.73046875,
      "learning_rate": 3.149536471991088e-06,
      "loss": 0.8864,
      "step": 7808
    },
    {
      "epoch": 4.19220831515721,
      "grad_norm": 0.5703125,
      "learning_rate": 3.145434287246912e-06,
      "loss": 0.648,
      "step": 7809
    },
    {
      "epoch": 4.192745209892285,
      "grad_norm": 0.80078125,
      "learning_rate": 3.141334596385448e-06,
      "loss": 1.1176,
      "step": 7810
    },
    {
      "epoch": 4.193282104627362,
      "grad_norm": 0.828125,
      "learning_rate": 3.137237399874521e-06,
      "loss": 0.8613,
      "step": 7811
    },
    {
      "epoch": 4.1938189993624375,
      "grad_norm": 0.734375,
      "learning_rate": 3.1331426981816845e-06,
      "loss": 0.8342,
      "step": 7812
    },
    {
      "epoch": 4.194355894097513,
      "grad_norm": 0.79296875,
      "learning_rate": 3.1290504917741855e-06,
      "loss": 0.9059,
      "step": 7813
    },
    {
      "epoch": 4.19489278883259,
      "grad_norm": 0.6875,
      "learning_rate": 3.1249607811189934e-06,
      "loss": 1.1009,
      "step": 7814
    },
    {
      "epoch": 4.195429683567665,
      "grad_norm": 0.74609375,
      "learning_rate": 3.1208735666828027e-06,
      "loss": 0.9495,
      "step": 7815
    },
    {
      "epoch": 4.195966578302741,
      "grad_norm": 0.8984375,
      "learning_rate": 3.116788848932017e-06,
      "loss": 1.0348,
      "step": 7816
    },
    {
      "epoch": 4.196503473037818,
      "grad_norm": 0.62890625,
      "learning_rate": 3.112706628332748e-06,
      "loss": 0.6889,
      "step": 7817
    },
    {
      "epoch": 4.197040367772893,
      "grad_norm": 0.6484375,
      "learning_rate": 3.1086269053508337e-06,
      "loss": 0.8543,
      "step": 7818
    },
    {
      "epoch": 4.19757726250797,
      "grad_norm": 0.61328125,
      "learning_rate": 3.1045496804518263e-06,
      "loss": 0.6374,
      "step": 7819
    },
    {
      "epoch": 4.198114157243046,
      "grad_norm": 0.64453125,
      "learning_rate": 3.1004749541009792e-06,
      "loss": 0.8448,
      "step": 7820
    },
    {
      "epoch": 4.198651051978121,
      "grad_norm": 0.64453125,
      "learning_rate": 3.0964027267632784e-06,
      "loss": 1.0817,
      "step": 7821
    },
    {
      "epoch": 4.199187946713198,
      "grad_norm": 0.6953125,
      "learning_rate": 3.092332998903416e-06,
      "loss": 0.7765,
      "step": 7822
    },
    {
      "epoch": 4.1997248414482735,
      "grad_norm": 0.67578125,
      "learning_rate": 3.088265770985788e-06,
      "loss": 0.708,
      "step": 7823
    },
    {
      "epoch": 4.200261736183349,
      "grad_norm": 0.8203125,
      "learning_rate": 3.084201043474527e-06,
      "loss": 0.9858,
      "step": 7824
    },
    {
      "epoch": 4.200798630918426,
      "grad_norm": 0.57421875,
      "learning_rate": 3.080138816833472e-06,
      "loss": 0.8453,
      "step": 7825
    },
    {
      "epoch": 4.201335525653501,
      "grad_norm": 0.6484375,
      "learning_rate": 3.0760790915261685e-06,
      "loss": 0.8257,
      "step": 7826
    },
    {
      "epoch": 4.201872420388577,
      "grad_norm": 0.66796875,
      "learning_rate": 3.072021868015881e-06,
      "loss": 0.6764,
      "step": 7827
    },
    {
      "epoch": 4.202409315123654,
      "grad_norm": 0.73828125,
      "learning_rate": 3.067967146765602e-06,
      "loss": 0.8771,
      "step": 7828
    },
    {
      "epoch": 4.202946209858729,
      "grad_norm": 0.77734375,
      "learning_rate": 3.0639149282380154e-06,
      "loss": 0.8544,
      "step": 7829
    },
    {
      "epoch": 4.203483104593806,
      "grad_norm": 0.60546875,
      "learning_rate": 3.0598652128955295e-06,
      "loss": 0.7856,
      "step": 7830
    },
    {
      "epoch": 4.204019999328882,
      "grad_norm": 0.671875,
      "learning_rate": 3.0558180012002725e-06,
      "loss": 0.9165,
      "step": 7831
    },
    {
      "epoch": 4.204556894063957,
      "grad_norm": 0.63671875,
      "learning_rate": 3.051773293614088e-06,
      "loss": 0.7057,
      "step": 7832
    },
    {
      "epoch": 4.205093788799034,
      "grad_norm": 0.77734375,
      "learning_rate": 3.0477310905985156e-06,
      "loss": 0.971,
      "step": 7833
    },
    {
      "epoch": 4.2056306835341095,
      "grad_norm": 0.69140625,
      "learning_rate": 3.0436913926148336e-06,
      "loss": 0.8533,
      "step": 7834
    },
    {
      "epoch": 4.206167578269185,
      "grad_norm": 0.7578125,
      "learning_rate": 3.0396542001240147e-06,
      "loss": 0.8808,
      "step": 7835
    },
    {
      "epoch": 4.206704473004262,
      "grad_norm": 0.65234375,
      "learning_rate": 3.0356195135867572e-06,
      "loss": 0.7512,
      "step": 7836
    },
    {
      "epoch": 4.207241367739337,
      "grad_norm": 0.61328125,
      "learning_rate": 3.031587333463473e-06,
      "loss": 0.7196,
      "step": 7837
    },
    {
      "epoch": 4.207778262474414,
      "grad_norm": 0.8671875,
      "learning_rate": 3.027557660214281e-06,
      "loss": 1.1682,
      "step": 7838
    },
    {
      "epoch": 4.20831515720949,
      "grad_norm": 1.0390625,
      "learning_rate": 3.023530494299015e-06,
      "loss": 0.9743,
      "step": 7839
    },
    {
      "epoch": 4.208852051944565,
      "grad_norm": 1.234375,
      "learning_rate": 3.0195058361772277e-06,
      "loss": 0.856,
      "step": 7840
    },
    {
      "epoch": 4.209388946679642,
      "grad_norm": 0.62890625,
      "learning_rate": 3.0154836863081905e-06,
      "loss": 0.7588,
      "step": 7841
    },
    {
      "epoch": 4.209925841414718,
      "grad_norm": 0.72265625,
      "learning_rate": 3.011464045150872e-06,
      "loss": 0.8106,
      "step": 7842
    },
    {
      "epoch": 4.210462736149793,
      "grad_norm": 0.8046875,
      "learning_rate": 3.0074469131639665e-06,
      "loss": 0.8849,
      "step": 7843
    },
    {
      "epoch": 4.21099963088487,
      "grad_norm": 0.734375,
      "learning_rate": 3.0034322908058883e-06,
      "loss": 0.9444,
      "step": 7844
    },
    {
      "epoch": 4.2115365256199455,
      "grad_norm": 0.66796875,
      "learning_rate": 2.9994201785347427e-06,
      "loss": 0.8591,
      "step": 7845
    },
    {
      "epoch": 4.212073420355022,
      "grad_norm": 0.69140625,
      "learning_rate": 2.9954105768083773e-06,
      "loss": 0.9048,
      "step": 7846
    },
    {
      "epoch": 4.212610315090098,
      "grad_norm": 0.71875,
      "learning_rate": 2.9914034860843293e-06,
      "loss": 1.079,
      "step": 7847
    },
    {
      "epoch": 4.213147209825173,
      "grad_norm": 0.609375,
      "learning_rate": 2.987398906819852e-06,
      "loss": 0.6953,
      "step": 7848
    },
    {
      "epoch": 4.21368410456025,
      "grad_norm": 0.765625,
      "learning_rate": 2.98339683947193e-06,
      "loss": 0.9031,
      "step": 7849
    },
    {
      "epoch": 4.214220999295326,
      "grad_norm": 0.62109375,
      "learning_rate": 2.9793972844972512e-06,
      "loss": 0.8121,
      "step": 7850
    },
    {
      "epoch": 4.214220999295326,
      "eval_loss": 0.9026446342468262,
      "eval_runtime": 45.83,
      "eval_samples_per_second": 20.991,
      "eval_steps_per_second": 20.991,
      "step": 7850
    },
    {
      "epoch": 4.214757894030401,
      "grad_norm": 0.75390625,
      "learning_rate": 2.9754002423522024e-06,
      "loss": 0.7948,
      "step": 7851
    },
    {
      "epoch": 4.215294788765478,
      "grad_norm": 0.69921875,
      "learning_rate": 2.9714057134929087e-06,
      "loss": 0.7093,
      "step": 7852
    },
    {
      "epoch": 4.215831683500554,
      "grad_norm": 0.76171875,
      "learning_rate": 2.967413698375196e-06,
      "loss": 0.7571,
      "step": 7853
    },
    {
      "epoch": 4.216368578235629,
      "grad_norm": 0.6796875,
      "learning_rate": 2.9634241974546008e-06,
      "loss": 0.727,
      "step": 7854
    },
    {
      "epoch": 4.216905472970706,
      "grad_norm": 0.65625,
      "learning_rate": 2.9594372111863692e-06,
      "loss": 0.7647,
      "step": 7855
    },
    {
      "epoch": 4.2174423677057815,
      "grad_norm": 0.83984375,
      "learning_rate": 2.9554527400254745e-06,
      "loss": 0.899,
      "step": 7856
    },
    {
      "epoch": 4.217979262440858,
      "grad_norm": 0.69140625,
      "learning_rate": 2.951470784426599e-06,
      "loss": 0.8867,
      "step": 7857
    },
    {
      "epoch": 4.218516157175934,
      "grad_norm": 0.62890625,
      "learning_rate": 2.9474913448441222e-06,
      "loss": 0.704,
      "step": 7858
    },
    {
      "epoch": 4.219053051911009,
      "grad_norm": 0.6640625,
      "learning_rate": 2.9435144217321575e-06,
      "loss": 0.8729,
      "step": 7859
    },
    {
      "epoch": 4.219589946646086,
      "grad_norm": 0.6640625,
      "learning_rate": 2.939540015544523e-06,
      "loss": 0.8927,
      "step": 7860
    },
    {
      "epoch": 4.220126841381162,
      "grad_norm": 0.64453125,
      "learning_rate": 2.935568126734742e-06,
      "loss": 0.828,
      "step": 7861
    },
    {
      "epoch": 4.220663736116237,
      "grad_norm": 0.65234375,
      "learning_rate": 2.9315987557560627e-06,
      "loss": 0.749,
      "step": 7862
    },
    {
      "epoch": 4.221200630851314,
      "grad_norm": 0.796875,
      "learning_rate": 2.9276319030614454e-06,
      "loss": 0.8385,
      "step": 7863
    },
    {
      "epoch": 4.22173752558639,
      "grad_norm": 0.69140625,
      "learning_rate": 2.9236675691035443e-06,
      "loss": 0.7876,
      "step": 7864
    },
    {
      "epoch": 4.222274420321465,
      "grad_norm": 0.7265625,
      "learning_rate": 2.919705754334748e-06,
      "loss": 1.1208,
      "step": 7865
    },
    {
      "epoch": 4.222811315056542,
      "grad_norm": 0.609375,
      "learning_rate": 2.9157464592071525e-06,
      "loss": 0.638,
      "step": 7866
    },
    {
      "epoch": 4.2233482097916175,
      "grad_norm": 0.62109375,
      "learning_rate": 2.9117896841725585e-06,
      "loss": 1.0634,
      "step": 7867
    },
    {
      "epoch": 4.223885104526694,
      "grad_norm": 0.5546875,
      "learning_rate": 2.9078354296824816e-06,
      "loss": 0.6566,
      "step": 7868
    },
    {
      "epoch": 4.22442199926177,
      "grad_norm": 0.6171875,
      "learning_rate": 2.903883696188167e-06,
      "loss": 0.6774,
      "step": 7869
    },
    {
      "epoch": 4.224958893996845,
      "grad_norm": 0.76953125,
      "learning_rate": 2.8999344841405373e-06,
      "loss": 0.7448,
      "step": 7870
    },
    {
      "epoch": 4.225495788731922,
      "grad_norm": 0.72265625,
      "learning_rate": 2.8959877939902653e-06,
      "loss": 1.0266,
      "step": 7871
    },
    {
      "epoch": 4.226032683466998,
      "grad_norm": 0.59375,
      "learning_rate": 2.8920436261877072e-06,
      "loss": 0.7287,
      "step": 7872
    },
    {
      "epoch": 4.226569578202073,
      "grad_norm": 0.62109375,
      "learning_rate": 2.8881019811829475e-06,
      "loss": 0.9185,
      "step": 7873
    },
    {
      "epoch": 4.22710647293715,
      "grad_norm": 0.703125,
      "learning_rate": 2.8841628594257737e-06,
      "loss": 0.9315,
      "step": 7874
    },
    {
      "epoch": 4.227643367672226,
      "grad_norm": 0.765625,
      "learning_rate": 2.8802262613656983e-06,
      "loss": 0.9369,
      "step": 7875
    },
    {
      "epoch": 4.228180262407302,
      "grad_norm": 0.71484375,
      "learning_rate": 2.8762921874519234e-06,
      "loss": 0.9166,
      "step": 7876
    },
    {
      "epoch": 4.228717157142378,
      "grad_norm": 0.7109375,
      "learning_rate": 2.872360638133384e-06,
      "loss": 0.7692,
      "step": 7877
    },
    {
      "epoch": 4.2292540518774535,
      "grad_norm": 0.640625,
      "learning_rate": 2.8684316138587243e-06,
      "loss": 0.9973,
      "step": 7878
    },
    {
      "epoch": 4.22979094661253,
      "grad_norm": 0.65625,
      "learning_rate": 2.8645051150762934e-06,
      "loss": 0.6537,
      "step": 7879
    },
    {
      "epoch": 4.230327841347606,
      "grad_norm": 0.55859375,
      "learning_rate": 2.860581142234145e-06,
      "loss": 0.7868,
      "step": 7880
    },
    {
      "epoch": 4.230864736082681,
      "grad_norm": 0.703125,
      "learning_rate": 2.8566596957800616e-06,
      "loss": 0.7497,
      "step": 7881
    },
    {
      "epoch": 4.231401630817758,
      "grad_norm": 0.7109375,
      "learning_rate": 2.852740776161536e-06,
      "loss": 1.0806,
      "step": 7882
    },
    {
      "epoch": 4.231938525552834,
      "grad_norm": 0.66015625,
      "learning_rate": 2.8488243838257544e-06,
      "loss": 0.9128,
      "step": 7883
    },
    {
      "epoch": 4.23247542028791,
      "grad_norm": 0.74609375,
      "learning_rate": 2.8449105192196316e-06,
      "loss": 0.7724,
      "step": 7884
    },
    {
      "epoch": 4.233012315022986,
      "grad_norm": 0.75390625,
      "learning_rate": 2.840999182789797e-06,
      "loss": 0.928,
      "step": 7885
    },
    {
      "epoch": 4.233549209758062,
      "grad_norm": 0.6484375,
      "learning_rate": 2.837090374982568e-06,
      "loss": 0.7443,
      "step": 7886
    },
    {
      "epoch": 4.234086104493138,
      "grad_norm": 0.6484375,
      "learning_rate": 2.8331840962440015e-06,
      "loss": 0.8266,
      "step": 7887
    },
    {
      "epoch": 4.234622999228214,
      "grad_norm": 0.7265625,
      "learning_rate": 2.829280347019858e-06,
      "loss": 0.961,
      "step": 7888
    },
    {
      "epoch": 4.2351598939632895,
      "grad_norm": 0.765625,
      "learning_rate": 2.8253791277555867e-06,
      "loss": 0.9082,
      "step": 7889
    },
    {
      "epoch": 4.235696788698366,
      "grad_norm": 0.7578125,
      "learning_rate": 2.821480438896376e-06,
      "loss": 0.8474,
      "step": 7890
    },
    {
      "epoch": 4.236233683433442,
      "grad_norm": 0.82421875,
      "learning_rate": 2.81758428088712e-06,
      "loss": 0.9891,
      "step": 7891
    },
    {
      "epoch": 4.236770578168517,
      "grad_norm": 0.64453125,
      "learning_rate": 2.81369065417241e-06,
      "loss": 0.7695,
      "step": 7892
    },
    {
      "epoch": 4.237307472903594,
      "grad_norm": 0.73046875,
      "learning_rate": 2.809799559196566e-06,
      "loss": 0.9035,
      "step": 7893
    },
    {
      "epoch": 4.23784436763867,
      "grad_norm": 0.59765625,
      "learning_rate": 2.80591099640361e-06,
      "loss": 0.8582,
      "step": 7894
    },
    {
      "epoch": 4.238381262373746,
      "grad_norm": 0.74609375,
      "learning_rate": 2.802024966237271e-06,
      "loss": 0.9073,
      "step": 7895
    },
    {
      "epoch": 4.238918157108822,
      "grad_norm": 0.5703125,
      "learning_rate": 2.7981414691410025e-06,
      "loss": 0.8136,
      "step": 7896
    },
    {
      "epoch": 4.239455051843898,
      "grad_norm": 0.70703125,
      "learning_rate": 2.7942605055579523e-06,
      "loss": 0.6798,
      "step": 7897
    },
    {
      "epoch": 4.239991946578974,
      "grad_norm": 0.6328125,
      "learning_rate": 2.7903820759309945e-06,
      "loss": 0.9977,
      "step": 7898
    },
    {
      "epoch": 4.24052884131405,
      "grad_norm": 0.6875,
      "learning_rate": 2.7865061807026997e-06,
      "loss": 0.7733,
      "step": 7899
    },
    {
      "epoch": 4.2410657360491255,
      "grad_norm": 0.73828125,
      "learning_rate": 2.7826328203153617e-06,
      "loss": 0.877,
      "step": 7900
    },
    {
      "epoch": 4.2410657360491255,
      "eval_loss": 0.902637243270874,
      "eval_runtime": 46.3458,
      "eval_samples_per_second": 20.757,
      "eval_steps_per_second": 20.757,
      "step": 7900
    },
    {
      "epoch": 4.241602630784202,
      "grad_norm": 0.6640625,
      "learning_rate": 2.778761995210985e-06,
      "loss": 0.7297,
      "step": 7901
    },
    {
      "epoch": 4.242139525519278,
      "grad_norm": 0.625,
      "learning_rate": 2.774893705831266e-06,
      "loss": 0.7969,
      "step": 7902
    },
    {
      "epoch": 4.242676420254353,
      "grad_norm": 0.7578125,
      "learning_rate": 2.7710279526176357e-06,
      "loss": 0.9622,
      "step": 7903
    },
    {
      "epoch": 4.24321331498943,
      "grad_norm": 0.66015625,
      "learning_rate": 2.767164736011227e-06,
      "loss": 0.7954,
      "step": 7904
    },
    {
      "epoch": 4.243750209724506,
      "grad_norm": 0.6640625,
      "learning_rate": 2.763304056452878e-06,
      "loss": 0.8726,
      "step": 7905
    },
    {
      "epoch": 4.244287104459582,
      "grad_norm": 0.66796875,
      "learning_rate": 2.759445914383135e-06,
      "loss": 0.866,
      "step": 7906
    },
    {
      "epoch": 4.244823999194658,
      "grad_norm": 0.8515625,
      "learning_rate": 2.755590310242273e-06,
      "loss": 0.8803,
      "step": 7907
    },
    {
      "epoch": 4.245360893929734,
      "grad_norm": 0.6640625,
      "learning_rate": 2.7517372444702564e-06,
      "loss": 0.8575,
      "step": 7908
    },
    {
      "epoch": 4.24589778866481,
      "grad_norm": 0.56640625,
      "learning_rate": 2.7478867175067675e-06,
      "loss": 0.7142,
      "step": 7909
    },
    {
      "epoch": 4.246434683399886,
      "grad_norm": 0.62890625,
      "learning_rate": 2.7440387297912123e-06,
      "loss": 0.7729,
      "step": 7910
    },
    {
      "epoch": 4.2469715781349615,
      "grad_norm": 0.7109375,
      "learning_rate": 2.7401932817626817e-06,
      "loss": 0.8111,
      "step": 7911
    },
    {
      "epoch": 4.247508472870038,
      "grad_norm": 0.62890625,
      "learning_rate": 2.7363503738599943e-06,
      "loss": 0.699,
      "step": 7912
    },
    {
      "epoch": 4.248045367605114,
      "grad_norm": 0.65625,
      "learning_rate": 2.7325100065216823e-06,
      "loss": 0.7496,
      "step": 7913
    },
    {
      "epoch": 4.24858226234019,
      "grad_norm": 0.79296875,
      "learning_rate": 2.7286721801859734e-06,
      "loss": 0.7449,
      "step": 7914
    },
    {
      "epoch": 4.249119157075266,
      "grad_norm": 0.6953125,
      "learning_rate": 2.7248368952908053e-06,
      "loss": 0.7934,
      "step": 7915
    },
    {
      "epoch": 4.249656051810342,
      "grad_norm": 0.78125,
      "learning_rate": 2.7210041522738454e-06,
      "loss": 1.0268,
      "step": 7916
    },
    {
      "epoch": 4.250192946545418,
      "grad_norm": 0.61328125,
      "learning_rate": 2.7171739515724465e-06,
      "loss": 0.7507,
      "step": 7917
    },
    {
      "epoch": 4.250729841280494,
      "grad_norm": 0.703125,
      "learning_rate": 2.713346293623689e-06,
      "loss": 0.829,
      "step": 7918
    },
    {
      "epoch": 4.25126673601557,
      "grad_norm": 0.55078125,
      "learning_rate": 2.7095211788643633e-06,
      "loss": 0.701,
      "step": 7919
    },
    {
      "epoch": 4.251803630750646,
      "grad_norm": 0.66796875,
      "learning_rate": 2.7056986077309525e-06,
      "loss": 0.9173,
      "step": 7920
    },
    {
      "epoch": 4.252340525485722,
      "grad_norm": 0.640625,
      "learning_rate": 2.70187858065967e-06,
      "loss": 0.6739,
      "step": 7921
    },
    {
      "epoch": 4.252877420220798,
      "grad_norm": 0.65234375,
      "learning_rate": 2.6980610980864183e-06,
      "loss": 0.8878,
      "step": 7922
    },
    {
      "epoch": 4.253414314955874,
      "grad_norm": 0.77734375,
      "learning_rate": 2.694246160446831e-06,
      "loss": 0.9122,
      "step": 7923
    },
    {
      "epoch": 4.25395120969095,
      "grad_norm": 0.6171875,
      "learning_rate": 2.6904337681762343e-06,
      "loss": 0.8183,
      "step": 7924
    },
    {
      "epoch": 4.254488104426026,
      "grad_norm": 0.75390625,
      "learning_rate": 2.686623921709669e-06,
      "loss": 0.8204,
      "step": 7925
    },
    {
      "epoch": 4.255024999161102,
      "grad_norm": 0.6796875,
      "learning_rate": 2.6828166214818957e-06,
      "loss": 0.9076,
      "step": 7926
    },
    {
      "epoch": 4.255561893896178,
      "grad_norm": 0.81640625,
      "learning_rate": 2.679011867927367e-06,
      "loss": 0.8625,
      "step": 7927
    },
    {
      "epoch": 4.256098788631254,
      "grad_norm": 0.66015625,
      "learning_rate": 2.6752096614802546e-06,
      "loss": 0.9715,
      "step": 7928
    },
    {
      "epoch": 4.25663568336633,
      "grad_norm": 0.7109375,
      "learning_rate": 2.671410002574448e-06,
      "loss": 0.9036,
      "step": 7929
    },
    {
      "epoch": 4.257172578101406,
      "grad_norm": 0.6015625,
      "learning_rate": 2.6676128916435256e-06,
      "loss": 0.6897,
      "step": 7930
    },
    {
      "epoch": 4.257709472836482,
      "grad_norm": 0.640625,
      "learning_rate": 2.6638183291207874e-06,
      "loss": 0.8025,
      "step": 7931
    },
    {
      "epoch": 4.258246367571558,
      "grad_norm": 0.71484375,
      "learning_rate": 2.6600263154392457e-06,
      "loss": 0.7686,
      "step": 7932
    },
    {
      "epoch": 4.258783262306634,
      "grad_norm": 0.65234375,
      "learning_rate": 2.65623685103161e-06,
      "loss": 0.8541,
      "step": 7933
    },
    {
      "epoch": 4.25932015704171,
      "grad_norm": 0.6328125,
      "learning_rate": 2.652449936330309e-06,
      "loss": 0.7334,
      "step": 7934
    },
    {
      "epoch": 4.259857051776786,
      "grad_norm": 0.66015625,
      "learning_rate": 2.6486655717674863e-06,
      "loss": 0.697,
      "step": 7935
    },
    {
      "epoch": 4.260393946511862,
      "grad_norm": 0.76171875,
      "learning_rate": 2.6448837577749713e-06,
      "loss": 0.9147,
      "step": 7936
    },
    {
      "epoch": 4.260930841246938,
      "grad_norm": 0.64453125,
      "learning_rate": 2.641104494784327e-06,
      "loss": 0.7542,
      "step": 7937
    },
    {
      "epoch": 4.261467735982014,
      "grad_norm": 0.65625,
      "learning_rate": 2.6373277832268145e-06,
      "loss": 0.7827,
      "step": 7938
    },
    {
      "epoch": 4.26200463071709,
      "grad_norm": 0.7421875,
      "learning_rate": 2.633553623533405e-06,
      "loss": 0.7254,
      "step": 7939
    },
    {
      "epoch": 4.262541525452166,
      "grad_norm": 0.7421875,
      "learning_rate": 2.6297820161347686e-06,
      "loss": 0.9928,
      "step": 7940
    },
    {
      "epoch": 4.263078420187242,
      "grad_norm": 0.71484375,
      "learning_rate": 2.6260129614612995e-06,
      "loss": 0.8077,
      "step": 7941
    },
    {
      "epoch": 4.263615314922318,
      "grad_norm": 0.62890625,
      "learning_rate": 2.6222464599431006e-06,
      "loss": 0.7353,
      "step": 7942
    },
    {
      "epoch": 4.264152209657394,
      "grad_norm": 0.765625,
      "learning_rate": 2.6184825120099676e-06,
      "loss": 0.8158,
      "step": 7943
    },
    {
      "epoch": 4.26468910439247,
      "grad_norm": 0.6484375,
      "learning_rate": 2.6147211180914176e-06,
      "loss": 0.9351,
      "step": 7944
    },
    {
      "epoch": 4.265225999127546,
      "grad_norm": 0.6875,
      "learning_rate": 2.61096227861668e-06,
      "loss": 0.8199,
      "step": 7945
    },
    {
      "epoch": 4.265762893862622,
      "grad_norm": 0.73828125,
      "learning_rate": 2.6072059940146775e-06,
      "loss": 0.6854,
      "step": 7946
    },
    {
      "epoch": 4.266299788597698,
      "grad_norm": 0.6796875,
      "learning_rate": 2.603452264714057e-06,
      "loss": 0.8996,
      "step": 7947
    },
    {
      "epoch": 4.266836683332774,
      "grad_norm": 0.86328125,
      "learning_rate": 2.5997010911431614e-06,
      "loss": 0.8856,
      "step": 7948
    },
    {
      "epoch": 4.26737357806785,
      "grad_norm": 0.84375,
      "learning_rate": 2.595952473730043e-06,
      "loss": 1.1729,
      "step": 7949
    },
    {
      "epoch": 4.267910472802926,
      "grad_norm": 0.6171875,
      "learning_rate": 2.5922064129024733e-06,
      "loss": 0.7371,
      "step": 7950
    },
    {
      "epoch": 4.267910472802926,
      "eval_loss": 0.9027069211006165,
      "eval_runtime": 46.1566,
      "eval_samples_per_second": 20.842,
      "eval_steps_per_second": 20.842,
      "step": 7950
    },
    {
      "epoch": 4.268447367538002,
      "grad_norm": 0.78515625,
      "learning_rate": 2.588462909087927e-06,
      "loss": 0.9572,
      "step": 7951
    },
    {
      "epoch": 4.2689842622730785,
      "grad_norm": 0.7109375,
      "learning_rate": 2.5847219627135764e-06,
      "loss": 0.7605,
      "step": 7952
    },
    {
      "epoch": 4.269521157008154,
      "grad_norm": 0.7578125,
      "learning_rate": 2.5809835742063193e-06,
      "loss": 1.0415,
      "step": 7953
    },
    {
      "epoch": 4.27005805174323,
      "grad_norm": 0.73828125,
      "learning_rate": 2.577247743992753e-06,
      "loss": 1.1646,
      "step": 7954
    },
    {
      "epoch": 4.270594946478306,
      "grad_norm": 0.6875,
      "learning_rate": 2.5735144724991806e-06,
      "loss": 0.8315,
      "step": 7955
    },
    {
      "epoch": 4.271131841213382,
      "grad_norm": 0.6953125,
      "learning_rate": 2.5697837601516116e-06,
      "loss": 0.7845,
      "step": 7956
    },
    {
      "epoch": 4.271668735948458,
      "grad_norm": 0.87109375,
      "learning_rate": 2.5660556073757695e-06,
      "loss": 0.7312,
      "step": 7957
    },
    {
      "epoch": 4.272205630683534,
      "grad_norm": 0.63671875,
      "learning_rate": 2.562330014597092e-06,
      "loss": 0.9128,
      "step": 7958
    },
    {
      "epoch": 4.27274252541861,
      "grad_norm": 0.6796875,
      "learning_rate": 2.558606982240705e-06,
      "loss": 0.6866,
      "step": 7959
    },
    {
      "epoch": 4.273279420153687,
      "grad_norm": 0.98046875,
      "learning_rate": 2.5548865107314607e-06,
      "loss": 1.1446,
      "step": 7960
    },
    {
      "epoch": 4.273816314888762,
      "grad_norm": 0.7578125,
      "learning_rate": 2.5511686004939086e-06,
      "loss": 0.9014,
      "step": 7961
    },
    {
      "epoch": 4.274353209623838,
      "grad_norm": 0.6171875,
      "learning_rate": 2.5474532519523087e-06,
      "loss": 0.7237,
      "step": 7962
    },
    {
      "epoch": 4.2748901043589145,
      "grad_norm": 0.60546875,
      "learning_rate": 2.543740465530636e-06,
      "loss": 0.8437,
      "step": 7963
    },
    {
      "epoch": 4.27542699909399,
      "grad_norm": 0.8515625,
      "learning_rate": 2.5400302416525632e-06,
      "loss": 0.8711,
      "step": 7964
    },
    {
      "epoch": 4.275963893829066,
      "grad_norm": 0.84765625,
      "learning_rate": 2.5363225807414655e-06,
      "loss": 0.872,
      "step": 7965
    },
    {
      "epoch": 4.276500788564142,
      "grad_norm": 0.6484375,
      "learning_rate": 2.5326174832204426e-06,
      "loss": 0.7451,
      "step": 7966
    },
    {
      "epoch": 4.277037683299218,
      "grad_norm": 0.66015625,
      "learning_rate": 2.5289149495122956e-06,
      "loss": 0.7314,
      "step": 7967
    },
    {
      "epoch": 4.277574578034294,
      "grad_norm": 0.7109375,
      "learning_rate": 2.5252149800395197e-06,
      "loss": 0.9796,
      "step": 7968
    },
    {
      "epoch": 4.27811147276937,
      "grad_norm": 0.640625,
      "learning_rate": 2.5215175752243357e-06,
      "loss": 0.8763,
      "step": 7969
    },
    {
      "epoch": 4.278648367504446,
      "grad_norm": 1.1640625,
      "learning_rate": 2.51782273548867e-06,
      "loss": 0.9131,
      "step": 7970
    },
    {
      "epoch": 4.279185262239523,
      "grad_norm": 0.8125,
      "learning_rate": 2.5141304612541354e-06,
      "loss": 0.7431,
      "step": 7971
    },
    {
      "epoch": 4.279722156974598,
      "grad_norm": 0.69140625,
      "learning_rate": 2.5104407529420838e-06,
      "loss": 0.9591,
      "step": 7972
    },
    {
      "epoch": 4.280259051709674,
      "grad_norm": 0.578125,
      "learning_rate": 2.5067536109735474e-06,
      "loss": 0.5859,
      "step": 7973
    },
    {
      "epoch": 4.2807959464447505,
      "grad_norm": 0.98046875,
      "learning_rate": 2.503069035769276e-06,
      "loss": 1.0047,
      "step": 7974
    },
    {
      "epoch": 4.281332841179826,
      "grad_norm": 0.55859375,
      "learning_rate": 2.499387027749725e-06,
      "loss": 0.6877,
      "step": 7975
    },
    {
      "epoch": 4.281869735914902,
      "grad_norm": 0.73828125,
      "learning_rate": 2.4957075873350665e-06,
      "loss": 0.7373,
      "step": 7976
    },
    {
      "epoch": 4.282406630649978,
      "grad_norm": 0.80859375,
      "learning_rate": 2.492030714945162e-06,
      "loss": 1.0971,
      "step": 7977
    },
    {
      "epoch": 4.282943525385054,
      "grad_norm": 0.65625,
      "learning_rate": 2.4883564109995957e-06,
      "loss": 0.7235,
      "step": 7978
    },
    {
      "epoch": 4.28348042012013,
      "grad_norm": 0.71484375,
      "learning_rate": 2.4846846759176506e-06,
      "loss": 0.9833,
      "step": 7979
    },
    {
      "epoch": 4.284017314855206,
      "grad_norm": 0.79296875,
      "learning_rate": 2.48101551011832e-06,
      "loss": 0.8382,
      "step": 7980
    },
    {
      "epoch": 4.284554209590282,
      "grad_norm": 0.6796875,
      "learning_rate": 2.4773489140202942e-06,
      "loss": 1.1936,
      "step": 7981
    },
    {
      "epoch": 4.285091104325359,
      "grad_norm": 0.6328125,
      "learning_rate": 2.4736848880419827e-06,
      "loss": 1.0067,
      "step": 7982
    },
    {
      "epoch": 4.285627999060434,
      "grad_norm": 0.6015625,
      "learning_rate": 2.470023432601504e-06,
      "loss": 0.6354,
      "step": 7983
    },
    {
      "epoch": 4.28616489379551,
      "grad_norm": 0.6015625,
      "learning_rate": 2.4663645481166648e-06,
      "loss": 0.8343,
      "step": 7984
    },
    {
      "epoch": 4.2867017885305865,
      "grad_norm": 0.734375,
      "learning_rate": 2.462708235004996e-06,
      "loss": 0.9136,
      "step": 7985
    },
    {
      "epoch": 4.287238683265662,
      "grad_norm": 0.765625,
      "learning_rate": 2.459054493683735e-06,
      "loss": 0.9607,
      "step": 7986
    },
    {
      "epoch": 4.287775578000738,
      "grad_norm": 0.73828125,
      "learning_rate": 2.4554033245698105e-06,
      "loss": 0.7222,
      "step": 7987
    },
    {
      "epoch": 4.288312472735814,
      "grad_norm": 0.734375,
      "learning_rate": 2.4517547280798693e-06,
      "loss": 0.8734,
      "step": 7988
    },
    {
      "epoch": 4.28884936747089,
      "grad_norm": 0.81640625,
      "learning_rate": 2.4481087046302752e-06,
      "loss": 0.9613,
      "step": 7989
    },
    {
      "epoch": 4.289386262205967,
      "grad_norm": 0.7578125,
      "learning_rate": 2.444465254637063e-06,
      "loss": 0.7166,
      "step": 7990
    },
    {
      "epoch": 4.289923156941042,
      "grad_norm": 0.7265625,
      "learning_rate": 2.4408243785160075e-06,
      "loss": 0.8868,
      "step": 7991
    },
    {
      "epoch": 4.290460051676118,
      "grad_norm": 0.8046875,
      "learning_rate": 2.437186076682585e-06,
      "loss": 0.9208,
      "step": 7992
    },
    {
      "epoch": 4.290996946411195,
      "grad_norm": 0.69921875,
      "learning_rate": 2.4335503495519636e-06,
      "loss": 0.9259,
      "step": 7993
    },
    {
      "epoch": 4.29153384114627,
      "grad_norm": 0.62109375,
      "learning_rate": 2.429917197539025e-06,
      "loss": 0.7361,
      "step": 7994
    },
    {
      "epoch": 4.292070735881346,
      "grad_norm": 0.671875,
      "learning_rate": 2.4262866210583686e-06,
      "loss": 0.8867,
      "step": 7995
    },
    {
      "epoch": 4.2926076306164225,
      "grad_norm": 0.73046875,
      "learning_rate": 2.4226586205242735e-06,
      "loss": 0.7309,
      "step": 7996
    },
    {
      "epoch": 4.293144525351498,
      "grad_norm": 0.8203125,
      "learning_rate": 2.419033196350756e-06,
      "loss": 1.1732,
      "step": 7997
    },
    {
      "epoch": 4.293681420086575,
      "grad_norm": 0.8046875,
      "learning_rate": 2.4154103489515107e-06,
      "loss": 0.9048,
      "step": 7998
    },
    {
      "epoch": 4.29421831482165,
      "grad_norm": 0.8125,
      "learning_rate": 2.411790078739959e-06,
      "loss": 0.7071,
      "step": 7999
    },
    {
      "epoch": 4.294755209556726,
      "grad_norm": 0.6171875,
      "learning_rate": 2.408172386129212e-06,
      "loss": 0.8387,
      "step": 8000
    },
    {
      "epoch": 4.294755209556726,
      "eval_loss": 0.9026855230331421,
      "eval_runtime": 46.7378,
      "eval_samples_per_second": 20.583,
      "eval_steps_per_second": 20.583,
      "step": 8000
    },
    {
      "epoch": 4.295292104291803,
      "grad_norm": 0.6875,
      "learning_rate": 2.4045572715321014e-06,
      "loss": 0.971,
      "step": 8001
    },
    {
      "epoch": 4.295828999026878,
      "grad_norm": 0.6171875,
      "learning_rate": 2.400944735361152e-06,
      "loss": 0.8212,
      "step": 8002
    },
    {
      "epoch": 4.296365893761954,
      "grad_norm": 0.65625,
      "learning_rate": 2.397334778028601e-06,
      "loss": 0.8329,
      "step": 8003
    },
    {
      "epoch": 4.296902788497031,
      "grad_norm": 0.71484375,
      "learning_rate": 2.3937273999463967e-06,
      "loss": 0.8759,
      "step": 8004
    },
    {
      "epoch": 4.297439683232106,
      "grad_norm": 0.62890625,
      "learning_rate": 2.390122601526179e-06,
      "loss": 0.8097,
      "step": 8005
    },
    {
      "epoch": 4.297976577967182,
      "grad_norm": 0.71484375,
      "learning_rate": 2.3865203831793e-06,
      "loss": 1.0514,
      "step": 8006
    },
    {
      "epoch": 4.2985134727022585,
      "grad_norm": 0.6015625,
      "learning_rate": 2.382920745316822e-06,
      "loss": 0.8337,
      "step": 8007
    },
    {
      "epoch": 4.299050367437334,
      "grad_norm": 0.72265625,
      "learning_rate": 2.379323688349516e-06,
      "loss": 0.9042,
      "step": 8008
    },
    {
      "epoch": 4.299587262172411,
      "grad_norm": 0.76171875,
      "learning_rate": 2.3757292126878382e-06,
      "loss": 0.8833,
      "step": 8009
    },
    {
      "epoch": 4.300124156907486,
      "grad_norm": 0.703125,
      "learning_rate": 2.372137318741968e-06,
      "loss": 0.7728,
      "step": 8010
    },
    {
      "epoch": 4.300661051642562,
      "grad_norm": 0.75,
      "learning_rate": 2.3685480069217973e-06,
      "loss": 0.8913,
      "step": 8011
    },
    {
      "epoch": 4.301197946377639,
      "grad_norm": 0.7109375,
      "learning_rate": 2.3649612776368958e-06,
      "loss": 0.7886,
      "step": 8012
    },
    {
      "epoch": 4.301734841112714,
      "grad_norm": 0.66015625,
      "learning_rate": 2.3613771312965633e-06,
      "loss": 0.8771,
      "step": 8013
    },
    {
      "epoch": 4.30227173584779,
      "grad_norm": 0.6875,
      "learning_rate": 2.3577955683098037e-06,
      "loss": 0.9066,
      "step": 8014
    },
    {
      "epoch": 4.3028086305828666,
      "grad_norm": 0.73046875,
      "learning_rate": 2.3542165890853006e-06,
      "loss": 0.9207,
      "step": 8015
    },
    {
      "epoch": 4.303345525317942,
      "grad_norm": 0.78515625,
      "learning_rate": 2.3506401940314727e-06,
      "loss": 0.909,
      "step": 8016
    },
    {
      "epoch": 4.303882420053018,
      "grad_norm": 0.6875,
      "learning_rate": 2.3470663835564317e-06,
      "loss": 1.0307,
      "step": 8017
    },
    {
      "epoch": 4.3044193147880945,
      "grad_norm": 0.671875,
      "learning_rate": 2.3434951580679876e-06,
      "loss": 0.9595,
      "step": 8018
    },
    {
      "epoch": 4.30495620952317,
      "grad_norm": 0.64453125,
      "learning_rate": 2.3399265179736675e-06,
      "loss": 0.6826,
      "step": 8019
    },
    {
      "epoch": 4.305493104258247,
      "grad_norm": 0.62109375,
      "learning_rate": 2.336360463680706e-06,
      "loss": 0.7527,
      "step": 8020
    },
    {
      "epoch": 4.306029998993322,
      "grad_norm": 0.7421875,
      "learning_rate": 2.3327969955960204e-06,
      "loss": 0.9664,
      "step": 8021
    },
    {
      "epoch": 4.306566893728398,
      "grad_norm": 0.83984375,
      "learning_rate": 2.329236114126257e-06,
      "loss": 0.9296,
      "step": 8022
    },
    {
      "epoch": 4.307103788463475,
      "grad_norm": 0.6484375,
      "learning_rate": 2.3256778196777544e-06,
      "loss": 0.7358,
      "step": 8023
    },
    {
      "epoch": 4.30764068319855,
      "grad_norm": 0.57421875,
      "learning_rate": 2.3221221126565606e-06,
      "loss": 0.752,
      "step": 8024
    },
    {
      "epoch": 4.308177577933626,
      "grad_norm": 0.6015625,
      "learning_rate": 2.3185689934684255e-06,
      "loss": 0.7511,
      "step": 8025
    },
    {
      "epoch": 4.3087144726687026,
      "grad_norm": 0.61328125,
      "learning_rate": 2.315018462518803e-06,
      "loss": 0.8922,
      "step": 8026
    },
    {
      "epoch": 4.309251367403778,
      "grad_norm": 0.59375,
      "learning_rate": 2.3114705202128607e-06,
      "loss": 0.6395,
      "step": 8027
    },
    {
      "epoch": 4.309788262138855,
      "grad_norm": 0.5859375,
      "learning_rate": 2.307925166955452e-06,
      "loss": 0.7146,
      "step": 8028
    },
    {
      "epoch": 4.3103251568739305,
      "grad_norm": 0.75,
      "learning_rate": 2.304382403151156e-06,
      "loss": 0.9844,
      "step": 8029
    },
    {
      "epoch": 4.310862051609006,
      "grad_norm": 0.9375,
      "learning_rate": 2.300842229204253e-06,
      "loss": 0.7643,
      "step": 8030
    },
    {
      "epoch": 4.311398946344083,
      "grad_norm": 0.66015625,
      "learning_rate": 2.2973046455187054e-06,
      "loss": 0.7227,
      "step": 8031
    },
    {
      "epoch": 4.311935841079158,
      "grad_norm": 0.6484375,
      "learning_rate": 2.293769652498204e-06,
      "loss": 0.7683,
      "step": 8032
    },
    {
      "epoch": 4.312472735814234,
      "grad_norm": 0.66796875,
      "learning_rate": 2.29023725054614e-06,
      "loss": 0.8554,
      "step": 8033
    },
    {
      "epoch": 4.313009630549311,
      "grad_norm": 0.70703125,
      "learning_rate": 2.2867074400655967e-06,
      "loss": 0.83,
      "step": 8034
    },
    {
      "epoch": 4.313546525284386,
      "grad_norm": 0.640625,
      "learning_rate": 2.2831802214593775e-06,
      "loss": 0.7588,
      "step": 8035
    },
    {
      "epoch": 4.314083420019463,
      "grad_norm": 0.63671875,
      "learning_rate": 2.2796555951299814e-06,
      "loss": 0.8654,
      "step": 8036
    },
    {
      "epoch": 4.3146203147545386,
      "grad_norm": 0.67578125,
      "learning_rate": 2.2761335614796097e-06,
      "loss": 1.0302,
      "step": 8037
    },
    {
      "epoch": 4.315157209489614,
      "grad_norm": 0.6875,
      "learning_rate": 2.2726141209101712e-06,
      "loss": 0.8198,
      "step": 8038
    },
    {
      "epoch": 4.315694104224691,
      "grad_norm": 0.59765625,
      "learning_rate": 2.269097273823287e-06,
      "loss": 0.7993,
      "step": 8039
    },
    {
      "epoch": 4.3162309989597665,
      "grad_norm": 0.71484375,
      "learning_rate": 2.2655830206202655e-06,
      "loss": 1.0365,
      "step": 8040
    },
    {
      "epoch": 4.316767893694842,
      "grad_norm": 0.6875,
      "learning_rate": 2.262071361702128e-06,
      "loss": 0.811,
      "step": 8041
    },
    {
      "epoch": 4.317304788429919,
      "grad_norm": 0.81640625,
      "learning_rate": 2.258562297469599e-06,
      "loss": 0.9286,
      "step": 8042
    },
    {
      "epoch": 4.317841683164994,
      "grad_norm": 0.7109375,
      "learning_rate": 2.255055828323113e-06,
      "loss": 0.7901,
      "step": 8043
    },
    {
      "epoch": 4.31837857790007,
      "grad_norm": 0.59765625,
      "learning_rate": 2.2515519546627924e-06,
      "loss": 0.7912,
      "step": 8044
    },
    {
      "epoch": 4.318915472635147,
      "grad_norm": 0.60546875,
      "learning_rate": 2.248050676888486e-06,
      "loss": 0.7476,
      "step": 8045
    },
    {
      "epoch": 4.319452367370222,
      "grad_norm": 0.78125,
      "learning_rate": 2.244551995399724e-06,
      "loss": 1.039,
      "step": 8046
    },
    {
      "epoch": 4.319989262105299,
      "grad_norm": 0.66796875,
      "learning_rate": 2.241055910595757e-06,
      "loss": 0.7083,
      "step": 8047
    },
    {
      "epoch": 4.3205261568403746,
      "grad_norm": 0.58203125,
      "learning_rate": 2.237562422875525e-06,
      "loss": 0.8362,
      "step": 8048
    },
    {
      "epoch": 4.32106305157545,
      "grad_norm": 0.6796875,
      "learning_rate": 2.2340715326376854e-06,
      "loss": 0.9089,
      "step": 8049
    },
    {
      "epoch": 4.321599946310527,
      "grad_norm": 0.72265625,
      "learning_rate": 2.23058324028059e-06,
      "loss": 0.8789,
      "step": 8050
    },
    {
      "epoch": 4.321599946310527,
      "eval_loss": 0.9027815461158752,
      "eval_runtime": 46.3347,
      "eval_samples_per_second": 20.762,
      "eval_steps_per_second": 20.762,
      "step": 8050
    },
    {
      "epoch": 4.3221368410456025,
      "grad_norm": 0.57421875,
      "learning_rate": 2.2270975462022973e-06,
      "loss": 0.7919,
      "step": 8051
    },
    {
      "epoch": 4.322673735780678,
      "grad_norm": 0.70703125,
      "learning_rate": 2.2236144508005735e-06,
      "loss": 0.7806,
      "step": 8052
    },
    {
      "epoch": 4.323210630515755,
      "grad_norm": 0.66015625,
      "learning_rate": 2.2201339544728776e-06,
      "loss": 0.9323,
      "step": 8053
    },
    {
      "epoch": 4.32374752525083,
      "grad_norm": 0.65234375,
      "learning_rate": 2.2166560576163813e-06,
      "loss": 1.1064,
      "step": 8054
    },
    {
      "epoch": 4.324284419985906,
      "grad_norm": 0.62109375,
      "learning_rate": 2.213180760627959e-06,
      "loss": 0.7244,
      "step": 8055
    },
    {
      "epoch": 4.324821314720983,
      "grad_norm": 0.71484375,
      "learning_rate": 2.209708063904184e-06,
      "loss": 1.0234,
      "step": 8056
    },
    {
      "epoch": 4.325358209456058,
      "grad_norm": 0.62890625,
      "learning_rate": 2.206237967841332e-06,
      "loss": 0.7779,
      "step": 8057
    },
    {
      "epoch": 4.325895104191135,
      "grad_norm": 0.7109375,
      "learning_rate": 2.2027704728353917e-06,
      "loss": 0.8836,
      "step": 8058
    },
    {
      "epoch": 4.3264319989262106,
      "grad_norm": 0.70703125,
      "learning_rate": 2.1993055792820377e-06,
      "loss": 0.988,
      "step": 8059
    },
    {
      "epoch": 4.326968893661286,
      "grad_norm": 0.640625,
      "learning_rate": 2.1958432875766653e-06,
      "loss": 0.7226,
      "step": 8060
    },
    {
      "epoch": 4.327505788396363,
      "grad_norm": 0.87109375,
      "learning_rate": 2.1923835981143697e-06,
      "loss": 0.9372,
      "step": 8061
    },
    {
      "epoch": 4.3280426831314385,
      "grad_norm": 0.6328125,
      "learning_rate": 2.1889265112899377e-06,
      "loss": 0.8695,
      "step": 8062
    },
    {
      "epoch": 4.328579577866514,
      "grad_norm": 0.6328125,
      "learning_rate": 2.1854720274978673e-06,
      "loss": 0.8805,
      "step": 8063
    },
    {
      "epoch": 4.329116472601591,
      "grad_norm": 0.7734375,
      "learning_rate": 2.182020147132366e-06,
      "loss": 0.9173,
      "step": 8064
    },
    {
      "epoch": 4.329653367336666,
      "grad_norm": 0.609375,
      "learning_rate": 2.178570870587335e-06,
      "loss": 0.7687,
      "step": 8065
    },
    {
      "epoch": 4.330190262071743,
      "grad_norm": 0.72265625,
      "learning_rate": 2.1751241982563702e-06,
      "loss": 0.9901,
      "step": 8066
    },
    {
      "epoch": 4.330727156806819,
      "grad_norm": 0.7734375,
      "learning_rate": 2.1716801305327884e-06,
      "loss": 0.9185,
      "step": 8067
    },
    {
      "epoch": 4.331264051541894,
      "grad_norm": 0.65234375,
      "learning_rate": 2.168238667809605e-06,
      "loss": 0.6998,
      "step": 8068
    },
    {
      "epoch": 4.331800946276971,
      "grad_norm": 0.78125,
      "learning_rate": 2.1647998104795286e-06,
      "loss": 0.8205,
      "step": 8069
    },
    {
      "epoch": 4.3323378410120466,
      "grad_norm": 0.734375,
      "learning_rate": 2.1613635589349756e-06,
      "loss": 0.9229,
      "step": 8070
    },
    {
      "epoch": 4.332874735747122,
      "grad_norm": 0.734375,
      "learning_rate": 2.157929913568074e-06,
      "loss": 0.7987,
      "step": 8071
    },
    {
      "epoch": 4.333411630482199,
      "grad_norm": 0.6953125,
      "learning_rate": 2.1544988747706414e-06,
      "loss": 0.7715,
      "step": 8072
    },
    {
      "epoch": 4.3339485252172745,
      "grad_norm": 0.78515625,
      "learning_rate": 2.151070442934197e-06,
      "loss": 0.907,
      "step": 8073
    },
    {
      "epoch": 4.334485419952351,
      "grad_norm": 0.8046875,
      "learning_rate": 2.147644618449976e-06,
      "loss": 1.0381,
      "step": 8074
    },
    {
      "epoch": 4.335022314687427,
      "grad_norm": 0.7109375,
      "learning_rate": 2.144221401708904e-06,
      "loss": 0.8121,
      "step": 8075
    },
    {
      "epoch": 4.335559209422502,
      "grad_norm": 0.7421875,
      "learning_rate": 2.140800793101613e-06,
      "loss": 0.8282,
      "step": 8076
    },
    {
      "epoch": 4.336096104157579,
      "grad_norm": 0.6484375,
      "learning_rate": 2.1373827930184444e-06,
      "loss": 0.7177,
      "step": 8077
    },
    {
      "epoch": 4.336632998892655,
      "grad_norm": 0.52734375,
      "learning_rate": 2.1339674018494292e-06,
      "loss": 0.7439,
      "step": 8078
    },
    {
      "epoch": 4.33716989362773,
      "grad_norm": 0.6171875,
      "learning_rate": 2.130554619984304e-06,
      "loss": 0.9383,
      "step": 8079
    },
    {
      "epoch": 4.337706788362807,
      "grad_norm": 0.62890625,
      "learning_rate": 2.1271444478125235e-06,
      "loss": 0.7015,
      "step": 8080
    },
    {
      "epoch": 4.3382436830978826,
      "grad_norm": 0.66015625,
      "learning_rate": 2.123736885723221e-06,
      "loss": 0.8099,
      "step": 8081
    },
    {
      "epoch": 4.338780577832958,
      "grad_norm": 0.75390625,
      "learning_rate": 2.120331934105238e-06,
      "loss": 0.9487,
      "step": 8082
    },
    {
      "epoch": 4.339317472568035,
      "grad_norm": 0.71875,
      "learning_rate": 2.1169295933471307e-06,
      "loss": 0.8674,
      "step": 8083
    },
    {
      "epoch": 4.3398543673031105,
      "grad_norm": 0.671875,
      "learning_rate": 2.113529863837152e-06,
      "loss": 0.9466,
      "step": 8084
    },
    {
      "epoch": 4.340391262038187,
      "grad_norm": 0.55859375,
      "learning_rate": 2.1101327459632446e-06,
      "loss": 0.7537,
      "step": 8085
    },
    {
      "epoch": 4.340928156773263,
      "grad_norm": 0.625,
      "learning_rate": 2.1067382401130707e-06,
      "loss": 0.8198,
      "step": 8086
    },
    {
      "epoch": 4.341465051508338,
      "grad_norm": 0.69921875,
      "learning_rate": 2.1033463466739816e-06,
      "loss": 0.9313,
      "step": 8087
    },
    {
      "epoch": 4.342001946243415,
      "grad_norm": 0.6640625,
      "learning_rate": 2.0999570660330344e-06,
      "loss": 0.8846,
      "step": 8088
    },
    {
      "epoch": 4.342538840978491,
      "grad_norm": 0.67578125,
      "learning_rate": 2.0965703985769996e-06,
      "loss": 0.7537,
      "step": 8089
    },
    {
      "epoch": 4.343075735713566,
      "grad_norm": 0.69921875,
      "learning_rate": 2.093186344692327e-06,
      "loss": 0.8586,
      "step": 8090
    },
    {
      "epoch": 4.343612630448643,
      "grad_norm": 0.62109375,
      "learning_rate": 2.0898049047651802e-06,
      "loss": 0.7498,
      "step": 8091
    },
    {
      "epoch": 4.3441495251837186,
      "grad_norm": 0.58203125,
      "learning_rate": 2.0864260791814305e-06,
      "loss": 0.6604,
      "step": 8092
    },
    {
      "epoch": 4.344686419918794,
      "grad_norm": 0.70703125,
      "learning_rate": 2.0830498683266446e-06,
      "loss": 0.8944,
      "step": 8093
    },
    {
      "epoch": 4.345223314653871,
      "grad_norm": 0.71484375,
      "learning_rate": 2.079676272586084e-06,
      "loss": 0.8687,
      "step": 8094
    },
    {
      "epoch": 4.3457602093889465,
      "grad_norm": 0.609375,
      "learning_rate": 2.0763052923447213e-06,
      "loss": 0.7794,
      "step": 8095
    },
    {
      "epoch": 4.346297104124023,
      "grad_norm": 0.6484375,
      "learning_rate": 2.0729369279872375e-06,
      "loss": 0.7759,
      "step": 8096
    },
    {
      "epoch": 4.346833998859099,
      "grad_norm": 0.61328125,
      "learning_rate": 2.0695711798979915e-06,
      "loss": 0.7231,
      "step": 8097
    },
    {
      "epoch": 4.347370893594174,
      "grad_norm": 0.63671875,
      "learning_rate": 2.0662080484610686e-06,
      "loss": 0.8852,
      "step": 8098
    },
    {
      "epoch": 4.347907788329251,
      "grad_norm": 0.68359375,
      "learning_rate": 2.0628475340602416e-06,
      "loss": 0.7639,
      "step": 8099
    },
    {
      "epoch": 4.348444683064327,
      "grad_norm": 2.5,
      "learning_rate": 2.059489637078982e-06,
      "loss": 1.0297,
      "step": 8100
    },
    {
      "epoch": 4.348444683064327,
      "eval_loss": 0.9026857018470764,
      "eval_runtime": 46.6968,
      "eval_samples_per_second": 20.601,
      "eval_steps_per_second": 20.601,
      "step": 8100
    },
    {
      "epoch": 4.348981577799402,
      "grad_norm": 0.69140625,
      "learning_rate": 2.0561343579004715e-06,
      "loss": 0.8775,
      "step": 8101
    },
    {
      "epoch": 4.349518472534479,
      "grad_norm": 0.66015625,
      "learning_rate": 2.0527816969075964e-06,
      "loss": 0.849,
      "step": 8102
    },
    {
      "epoch": 4.3500553672695546,
      "grad_norm": 0.875,
      "learning_rate": 2.049431654482928e-06,
      "loss": 0.9483,
      "step": 8103
    },
    {
      "epoch": 4.350592262004631,
      "grad_norm": 0.69921875,
      "learning_rate": 2.0460842310087548e-06,
      "loss": 0.912,
      "step": 8104
    },
    {
      "epoch": 4.351129156739707,
      "grad_norm": 0.69140625,
      "learning_rate": 2.042739426867063e-06,
      "loss": 0.9216,
      "step": 8105
    },
    {
      "epoch": 4.3516660514747825,
      "grad_norm": 0.6953125,
      "learning_rate": 2.0393972424395308e-06,
      "loss": 0.83,
      "step": 8106
    },
    {
      "epoch": 4.352202946209859,
      "grad_norm": 0.6484375,
      "learning_rate": 2.0360576781075417e-06,
      "loss": 0.882,
      "step": 8107
    },
    {
      "epoch": 4.352739840944935,
      "grad_norm": 0.765625,
      "learning_rate": 2.0327207342521885e-06,
      "loss": 0.848,
      "step": 8108
    },
    {
      "epoch": 4.35327673568001,
      "grad_norm": 0.74609375,
      "learning_rate": 2.029386411254258e-06,
      "loss": 0.9986,
      "step": 8109
    },
    {
      "epoch": 4.353813630415087,
      "grad_norm": 0.671875,
      "learning_rate": 2.026054709494235e-06,
      "loss": 0.7709,
      "step": 8110
    },
    {
      "epoch": 4.354350525150163,
      "grad_norm": 0.62109375,
      "learning_rate": 2.0227256293523096e-06,
      "loss": 0.9553,
      "step": 8111
    },
    {
      "epoch": 4.354887419885239,
      "grad_norm": 0.7265625,
      "learning_rate": 2.0193991712083804e-06,
      "loss": 1.0477,
      "step": 8112
    },
    {
      "epoch": 4.355424314620315,
      "grad_norm": 0.63671875,
      "learning_rate": 2.0160753354420246e-06,
      "loss": 0.7884,
      "step": 8113
    },
    {
      "epoch": 4.3559612093553906,
      "grad_norm": 0.5703125,
      "learning_rate": 2.0127541224325474e-06,
      "loss": 0.827,
      "step": 8114
    },
    {
      "epoch": 4.356498104090467,
      "grad_norm": 0.8046875,
      "learning_rate": 2.0094355325589315e-06,
      "loss": 0.9843,
      "step": 8115
    },
    {
      "epoch": 4.357034998825543,
      "grad_norm": 0.67578125,
      "learning_rate": 2.006119566199871e-06,
      "loss": 0.7487,
      "step": 8116
    },
    {
      "epoch": 4.3575718935606185,
      "grad_norm": 0.5859375,
      "learning_rate": 2.0028062237337634e-06,
      "loss": 0.7653,
      "step": 8117
    },
    {
      "epoch": 4.358108788295695,
      "grad_norm": 0.7890625,
      "learning_rate": 1.999495505538704e-06,
      "loss": 0.9481,
      "step": 8118
    },
    {
      "epoch": 4.358645683030771,
      "grad_norm": 0.671875,
      "learning_rate": 1.996187411992481e-06,
      "loss": 0.7662,
      "step": 8119
    },
    {
      "epoch": 4.359182577765846,
      "grad_norm": 0.7265625,
      "learning_rate": 1.992881943472594e-06,
      "loss": 0.8063,
      "step": 8120
    },
    {
      "epoch": 4.359719472500923,
      "grad_norm": 0.62890625,
      "learning_rate": 1.9895791003562425e-06,
      "loss": 0.7327,
      "step": 8121
    },
    {
      "epoch": 4.360256367235999,
      "grad_norm": 0.80859375,
      "learning_rate": 1.986278883020318e-06,
      "loss": 0.9624,
      "step": 8122
    },
    {
      "epoch": 4.360793261971075,
      "grad_norm": 0.64453125,
      "learning_rate": 1.9829812918414183e-06,
      "loss": 0.7113,
      "step": 8123
    },
    {
      "epoch": 4.361330156706151,
      "grad_norm": 0.7578125,
      "learning_rate": 1.9796863271958373e-06,
      "loss": 0.6861,
      "step": 8124
    },
    {
      "epoch": 4.3618670514412266,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9763939894595797e-06,
      "loss": 0.865,
      "step": 8125
    },
    {
      "epoch": 4.362403946176303,
      "grad_norm": 0.73828125,
      "learning_rate": 1.9731042790083367e-06,
      "loss": 0.7453,
      "step": 8126
    },
    {
      "epoch": 4.362940840911379,
      "grad_norm": 0.79296875,
      "learning_rate": 1.9698171962175078e-06,
      "loss": 0.9922,
      "step": 8127
    },
    {
      "epoch": 4.3634777356464545,
      "grad_norm": 0.6328125,
      "learning_rate": 1.9665327414621876e-06,
      "loss": 0.8148,
      "step": 8128
    },
    {
      "epoch": 4.364014630381531,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9632509151171763e-06,
      "loss": 0.9316,
      "step": 8129
    },
    {
      "epoch": 4.364551525116607,
      "grad_norm": 0.76171875,
      "learning_rate": 1.959971717556977e-06,
      "loss": 1.1203,
      "step": 8130
    },
    {
      "epoch": 4.365088419851682,
      "grad_norm": 0.67578125,
      "learning_rate": 1.956695149155785e-06,
      "loss": 0.7563,
      "step": 8131
    },
    {
      "epoch": 4.365625314586759,
      "grad_norm": 0.69921875,
      "learning_rate": 1.95342121028749e-06,
      "loss": 0.7891,
      "step": 8132
    },
    {
      "epoch": 4.366162209321835,
      "grad_norm": 0.67578125,
      "learning_rate": 1.950149901325696e-06,
      "loss": 0.8395,
      "step": 8133
    },
    {
      "epoch": 4.366699104056911,
      "grad_norm": 0.63671875,
      "learning_rate": 1.9468812226437067e-06,
      "loss": 0.6971,
      "step": 8134
    },
    {
      "epoch": 4.367235998791987,
      "grad_norm": 0.77734375,
      "learning_rate": 1.9436151746145077e-06,
      "loss": 0.8547,
      "step": 8135
    },
    {
      "epoch": 4.3677728935270625,
      "grad_norm": 0.6328125,
      "learning_rate": 1.9403517576108034e-06,
      "loss": 0.6845,
      "step": 8136
    },
    {
      "epoch": 4.368309788262139,
      "grad_norm": 0.640625,
      "learning_rate": 1.9370909720049957e-06,
      "loss": 0.995,
      "step": 8137
    },
    {
      "epoch": 4.368846682997215,
      "grad_norm": 0.6875,
      "learning_rate": 1.9338328181691705e-06,
      "loss": 0.9313,
      "step": 8138
    },
    {
      "epoch": 4.3693835777322905,
      "grad_norm": 0.5859375,
      "learning_rate": 1.9305772964751304e-06,
      "loss": 0.7913,
      "step": 8139
    },
    {
      "epoch": 4.369920472467367,
      "grad_norm": 0.75390625,
      "learning_rate": 1.9273244072943815e-06,
      "loss": 0.9748,
      "step": 8140
    },
    {
      "epoch": 4.370457367202443,
      "grad_norm": 0.6953125,
      "learning_rate": 1.924074150998101e-06,
      "loss": 0.7767,
      "step": 8141
    },
    {
      "epoch": 4.370994261937519,
      "grad_norm": 0.59765625,
      "learning_rate": 1.9208265279571926e-06,
      "loss": 0.8383,
      "step": 8142
    },
    {
      "epoch": 4.371531156672595,
      "grad_norm": 0.5625,
      "learning_rate": 1.917581538542254e-06,
      "loss": 0.6568,
      "step": 8143
    },
    {
      "epoch": 4.372068051407671,
      "grad_norm": 0.5625,
      "learning_rate": 1.9143391831235728e-06,
      "loss": 0.6376,
      "step": 8144
    },
    {
      "epoch": 4.372604946142747,
      "grad_norm": 0.6484375,
      "learning_rate": 1.9110994620711476e-06,
      "loss": 0.7222,
      "step": 8145
    },
    {
      "epoch": 4.373141840877823,
      "grad_norm": 0.6640625,
      "learning_rate": 1.9078623757546767e-06,
      "loss": 0.8181,
      "step": 8146
    },
    {
      "epoch": 4.3736787356128985,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9046279245435421e-06,
      "loss": 0.775,
      "step": 8147
    },
    {
      "epoch": 4.374215630347975,
      "grad_norm": 0.71875,
      "learning_rate": 1.9013961088068438e-06,
      "loss": 0.9851,
      "step": 8148
    },
    {
      "epoch": 4.374752525083051,
      "grad_norm": 0.75,
      "learning_rate": 1.8981669289133668e-06,
      "loss": 0.728,
      "step": 8149
    },
    {
      "epoch": 4.375289419818127,
      "grad_norm": 0.59765625,
      "learning_rate": 1.8949403852316084e-06,
      "loss": 0.7222,
      "step": 8150
    },
    {
      "epoch": 4.375289419818127,
      "eval_loss": 0.9027720093727112,
      "eval_runtime": 46.7202,
      "eval_samples_per_second": 20.591,
      "eval_steps_per_second": 20.591,
      "step": 8150
    },
    {
      "epoch": 4.375826314553203,
      "grad_norm": 0.62890625,
      "learning_rate": 1.8917164781297486e-06,
      "loss": 0.7624,
      "step": 8151
    },
    {
      "epoch": 4.376363209288279,
      "grad_norm": 0.703125,
      "learning_rate": 1.8884952079756796e-06,
      "loss": 0.8883,
      "step": 8152
    },
    {
      "epoch": 4.376900104023355,
      "grad_norm": 0.640625,
      "learning_rate": 1.885276575136999e-06,
      "loss": 0.8894,
      "step": 8153
    },
    {
      "epoch": 4.377436998758431,
      "grad_norm": 0.63671875,
      "learning_rate": 1.8820605799809793e-06,
      "loss": 0.7652,
      "step": 8154
    },
    {
      "epoch": 4.377973893493507,
      "grad_norm": 0.6484375,
      "learning_rate": 1.8788472228746135e-06,
      "loss": 0.8017,
      "step": 8155
    },
    {
      "epoch": 4.378510788228583,
      "grad_norm": 0.76953125,
      "learning_rate": 1.8756365041845913e-06,
      "loss": 0.8047,
      "step": 8156
    },
    {
      "epoch": 4.379047682963659,
      "grad_norm": 0.734375,
      "learning_rate": 1.872428424277284e-06,
      "loss": 0.859,
      "step": 8157
    },
    {
      "epoch": 4.3795845776987345,
      "grad_norm": 0.7265625,
      "learning_rate": 1.8692229835187813e-06,
      "loss": 0.8247,
      "step": 8158
    },
    {
      "epoch": 4.380121472433811,
      "grad_norm": 0.69921875,
      "learning_rate": 1.8660201822748664e-06,
      "loss": 0.6967,
      "step": 8159
    },
    {
      "epoch": 4.380658367168887,
      "grad_norm": 0.87109375,
      "learning_rate": 1.8628200209110131e-06,
      "loss": 1.0077,
      "step": 8160
    },
    {
      "epoch": 4.381195261903963,
      "grad_norm": 0.60546875,
      "learning_rate": 1.8596224997924043e-06,
      "loss": 1.0272,
      "step": 8161
    },
    {
      "epoch": 4.381732156639039,
      "grad_norm": 0.7109375,
      "learning_rate": 1.8564276192839208e-06,
      "loss": 0.7763,
      "step": 8162
    },
    {
      "epoch": 4.382269051374115,
      "grad_norm": 0.61328125,
      "learning_rate": 1.8532353797501318e-06,
      "loss": 0.8049,
      "step": 8163
    },
    {
      "epoch": 4.382805946109191,
      "grad_norm": 0.7421875,
      "learning_rate": 1.8500457815553152e-06,
      "loss": 0.9138,
      "step": 8164
    },
    {
      "epoch": 4.383342840844267,
      "grad_norm": 0.7578125,
      "learning_rate": 1.8468588250634494e-06,
      "loss": 0.9941,
      "step": 8165
    },
    {
      "epoch": 4.383879735579343,
      "grad_norm": 0.73828125,
      "learning_rate": 1.8436745106382047e-06,
      "loss": 0.9625,
      "step": 8166
    },
    {
      "epoch": 4.384416630314419,
      "grad_norm": 0.671875,
      "learning_rate": 1.8404928386429432e-06,
      "loss": 0.7265,
      "step": 8167
    },
    {
      "epoch": 4.384953525049495,
      "grad_norm": 0.77734375,
      "learning_rate": 1.837313809440741e-06,
      "loss": 0.8467,
      "step": 8168
    },
    {
      "epoch": 4.3854904197845705,
      "grad_norm": 0.703125,
      "learning_rate": 1.8341374233943693e-06,
      "loss": 0.8626,
      "step": 8169
    },
    {
      "epoch": 4.386027314519647,
      "grad_norm": 0.63671875,
      "learning_rate": 1.830963680866285e-06,
      "loss": 0.9047,
      "step": 8170
    },
    {
      "epoch": 4.386564209254723,
      "grad_norm": 0.69140625,
      "learning_rate": 1.8277925822186625e-06,
      "loss": 0.949,
      "step": 8171
    },
    {
      "epoch": 4.387101103989799,
      "grad_norm": 1.0234375,
      "learning_rate": 1.824624127813354e-06,
      "loss": 0.7457,
      "step": 8172
    },
    {
      "epoch": 4.387637998724875,
      "grad_norm": 0.7109375,
      "learning_rate": 1.8214583180119288e-06,
      "loss": 0.7518,
      "step": 8173
    },
    {
      "epoch": 4.388174893459951,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8182951531756364e-06,
      "loss": 0.8073,
      "step": 8174
    },
    {
      "epoch": 4.388711788195027,
      "grad_norm": 0.7578125,
      "learning_rate": 1.815134633665444e-06,
      "loss": 0.6695,
      "step": 8175
    },
    {
      "epoch": 4.389248682930103,
      "grad_norm": 0.61328125,
      "learning_rate": 1.811976759841999e-06,
      "loss": 0.7827,
      "step": 8176
    },
    {
      "epoch": 4.389785577665179,
      "grad_norm": 0.953125,
      "learning_rate": 1.8088215320656577e-06,
      "loss": 0.9644,
      "step": 8177
    },
    {
      "epoch": 4.390322472400255,
      "grad_norm": 0.74609375,
      "learning_rate": 1.8056689506964764e-06,
      "loss": 0.9412,
      "step": 8178
    },
    {
      "epoch": 4.390859367135331,
      "grad_norm": 1.015625,
      "learning_rate": 1.802519016094198e-06,
      "loss": 0.9344,
      "step": 8179
    },
    {
      "epoch": 4.391396261870407,
      "grad_norm": 0.73046875,
      "learning_rate": 1.799371728618271e-06,
      "loss": 0.8584,
      "step": 8180
    },
    {
      "epoch": 4.391933156605483,
      "grad_norm": 0.7578125,
      "learning_rate": 1.7962270886278442e-06,
      "loss": 0.9523,
      "step": 8181
    },
    {
      "epoch": 4.392470051340559,
      "grad_norm": 0.6875,
      "learning_rate": 1.7930850964817614e-06,
      "loss": 0.9462,
      "step": 8182
    },
    {
      "epoch": 4.393006946075635,
      "grad_norm": 0.7734375,
      "learning_rate": 1.789945752538555e-06,
      "loss": 0.852,
      "step": 8183
    },
    {
      "epoch": 4.393543840810711,
      "grad_norm": 0.65625,
      "learning_rate": 1.7868090571564744e-06,
      "loss": 0.8787,
      "step": 8184
    },
    {
      "epoch": 4.394080735545787,
      "grad_norm": 0.61328125,
      "learning_rate": 1.7836750106934474e-06,
      "loss": 0.8575,
      "step": 8185
    },
    {
      "epoch": 4.394617630280863,
      "grad_norm": 0.71875,
      "learning_rate": 1.780543613507113e-06,
      "loss": 0.8147,
      "step": 8186
    },
    {
      "epoch": 4.395154525015939,
      "grad_norm": 0.75,
      "learning_rate": 1.7774148659548072e-06,
      "loss": 1.0441,
      "step": 8187
    },
    {
      "epoch": 4.3956914197510155,
      "grad_norm": 0.68359375,
      "learning_rate": 1.7742887683935506e-06,
      "loss": 1.0124,
      "step": 8188
    },
    {
      "epoch": 4.396228314486091,
      "grad_norm": 0.69921875,
      "learning_rate": 1.7711653211800739e-06,
      "loss": 0.7514,
      "step": 8189
    },
    {
      "epoch": 4.396765209221167,
      "grad_norm": 0.625,
      "learning_rate": 1.7680445246708115e-06,
      "loss": 0.6354,
      "step": 8190
    },
    {
      "epoch": 4.397302103956243,
      "grad_norm": 0.8046875,
      "learning_rate": 1.7649263792218734e-06,
      "loss": 1.0701,
      "step": 8191
    },
    {
      "epoch": 4.397838998691319,
      "grad_norm": 1.1328125,
      "learning_rate": 1.761810885189083e-06,
      "loss": 0.9761,
      "step": 8192
    },
    {
      "epoch": 4.398375893426395,
      "grad_norm": 0.67578125,
      "learning_rate": 1.7586980429279588e-06,
      "loss": 0.865,
      "step": 8193
    },
    {
      "epoch": 4.398912788161471,
      "grad_norm": 0.703125,
      "learning_rate": 1.7555878527937164e-06,
      "loss": 0.7017,
      "step": 8194
    },
    {
      "epoch": 4.399449682896547,
      "grad_norm": 0.8046875,
      "learning_rate": 1.752480315141264e-06,
      "loss": 0.8551,
      "step": 8195
    },
    {
      "epoch": 4.399986577631623,
      "grad_norm": 0.62890625,
      "learning_rate": 1.7493754303252147e-06,
      "loss": 0.7346,
      "step": 8196
    },
    {
      "epoch": 4.400523472366699,
      "grad_norm": 0.640625,
      "learning_rate": 1.746273198699877e-06,
      "loss": 0.7483,
      "step": 8197
    },
    {
      "epoch": 4.401060367101775,
      "grad_norm": 0.58984375,
      "learning_rate": 1.743173620619251e-06,
      "loss": 0.8668,
      "step": 8198
    },
    {
      "epoch": 4.4015972618368515,
      "grad_norm": 0.58984375,
      "learning_rate": 1.7400766964370375e-06,
      "loss": 0.6748,
      "step": 8199
    },
    {
      "epoch": 4.402134156571927,
      "grad_norm": 0.609375,
      "learning_rate": 1.7369824265066365e-06,
      "loss": 0.8673,
      "step": 8200
    },
    {
      "epoch": 4.402134156571927,
      "eval_loss": 0.9027100205421448,
      "eval_runtime": 46.0007,
      "eval_samples_per_second": 20.913,
      "eval_steps_per_second": 20.913,
      "step": 8200
    },
    {
      "epoch": 4.402671051307003,
      "grad_norm": 0.63671875,
      "learning_rate": 1.7338908111811408e-06,
      "loss": 0.9569,
      "step": 8201
    },
    {
      "epoch": 4.403207946042079,
      "grad_norm": 0.734375,
      "learning_rate": 1.7308018508133462e-06,
      "loss": 0.8345,
      "step": 8202
    },
    {
      "epoch": 4.403744840777155,
      "grad_norm": 0.62109375,
      "learning_rate": 1.7277155457557426e-06,
      "loss": 0.8133,
      "step": 8203
    },
    {
      "epoch": 4.404281735512231,
      "grad_norm": 0.66796875,
      "learning_rate": 1.7246318963605123e-06,
      "loss": 0.9958,
      "step": 8204
    },
    {
      "epoch": 4.404818630247307,
      "grad_norm": 0.625,
      "learning_rate": 1.7215509029795436e-06,
      "loss": 0.8414,
      "step": 8205
    },
    {
      "epoch": 4.405355524982383,
      "grad_norm": 0.5625,
      "learning_rate": 1.7184725659644157e-06,
      "loss": 0.7352,
      "step": 8206
    },
    {
      "epoch": 4.405892419717459,
      "grad_norm": 0.68359375,
      "learning_rate": 1.7153968856664038e-06,
      "loss": 0.8369,
      "step": 8207
    },
    {
      "epoch": 4.406429314452535,
      "grad_norm": 0.765625,
      "learning_rate": 1.7123238624364795e-06,
      "loss": 0.8539,
      "step": 8208
    },
    {
      "epoch": 4.406966209187611,
      "grad_norm": 0.68359375,
      "learning_rate": 1.7092534966253182e-06,
      "loss": 0.7729,
      "step": 8209
    },
    {
      "epoch": 4.4075031039226875,
      "grad_norm": 0.76171875,
      "learning_rate": 1.7061857885832893e-06,
      "loss": 0.8306,
      "step": 8210
    },
    {
      "epoch": 4.408039998657763,
      "grad_norm": 0.61328125,
      "learning_rate": 1.703120738660452e-06,
      "loss": 0.729,
      "step": 8211
    },
    {
      "epoch": 4.408576893392839,
      "grad_norm": 0.76171875,
      "learning_rate": 1.7000583472065734e-06,
      "loss": 0.8055,
      "step": 8212
    },
    {
      "epoch": 4.409113788127915,
      "grad_norm": 0.66796875,
      "learning_rate": 1.696998614571102e-06,
      "loss": 0.8735,
      "step": 8213
    },
    {
      "epoch": 4.409650682862991,
      "grad_norm": 0.65625,
      "learning_rate": 1.6939415411031972e-06,
      "loss": 0.9641,
      "step": 8214
    },
    {
      "epoch": 4.410187577598067,
      "grad_norm": 0.7734375,
      "learning_rate": 1.6908871271517135e-06,
      "loss": 0.849,
      "step": 8215
    },
    {
      "epoch": 4.410724472333143,
      "grad_norm": 0.64453125,
      "learning_rate": 1.687835373065194e-06,
      "loss": 0.9493,
      "step": 8216
    },
    {
      "epoch": 4.411261367068219,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6847862791918829e-06,
      "loss": 0.8929,
      "step": 8217
    },
    {
      "epoch": 4.411798261803296,
      "grad_norm": 0.8125,
      "learning_rate": 1.681739845879718e-06,
      "loss": 0.8998,
      "step": 8218
    },
    {
      "epoch": 4.412335156538371,
      "grad_norm": 0.671875,
      "learning_rate": 1.6786960734763408e-06,
      "loss": 0.7558,
      "step": 8219
    },
    {
      "epoch": 4.412872051273447,
      "grad_norm": 0.66796875,
      "learning_rate": 1.6756549623290818e-06,
      "loss": 0.8733,
      "step": 8220
    },
    {
      "epoch": 4.4134089460085235,
      "grad_norm": 0.7734375,
      "learning_rate": 1.6726165127849692e-06,
      "loss": 0.7853,
      "step": 8221
    },
    {
      "epoch": 4.413945840743599,
      "grad_norm": 0.73046875,
      "learning_rate": 1.6695807251907335e-06,
      "loss": 0.9125,
      "step": 8222
    },
    {
      "epoch": 4.414482735478675,
      "grad_norm": 0.97265625,
      "learning_rate": 1.6665475998927872e-06,
      "loss": 1.0584,
      "step": 8223
    },
    {
      "epoch": 4.415019630213751,
      "grad_norm": 0.703125,
      "learning_rate": 1.663517137237261e-06,
      "loss": 0.8236,
      "step": 8224
    },
    {
      "epoch": 4.415556524948827,
      "grad_norm": 0.7421875,
      "learning_rate": 1.6604893375699594e-06,
      "loss": 0.867,
      "step": 8225
    },
    {
      "epoch": 4.416093419683904,
      "grad_norm": 0.73828125,
      "learning_rate": 1.6574642012363945e-06,
      "loss": 0.8422,
      "step": 8226
    },
    {
      "epoch": 4.416630314418979,
      "grad_norm": 0.765625,
      "learning_rate": 1.654441728581771e-06,
      "loss": 0.8131,
      "step": 8227
    },
    {
      "epoch": 4.417167209154055,
      "grad_norm": 0.65625,
      "learning_rate": 1.651421919950999e-06,
      "loss": 0.8059,
      "step": 8228
    },
    {
      "epoch": 4.417704103889132,
      "grad_norm": 0.7734375,
      "learning_rate": 1.6484047756886666e-06,
      "loss": 0.7703,
      "step": 8229
    },
    {
      "epoch": 4.418240998624207,
      "grad_norm": 0.73046875,
      "learning_rate": 1.6453902961390738e-06,
      "loss": 0.9109,
      "step": 8230
    },
    {
      "epoch": 4.418777893359283,
      "grad_norm": 0.6796875,
      "learning_rate": 1.6423784816462145e-06,
      "loss": 0.789,
      "step": 8231
    },
    {
      "epoch": 4.4193147880943595,
      "grad_norm": 0.69921875,
      "learning_rate": 1.6393693325537724e-06,
      "loss": 0.8911,
      "step": 8232
    },
    {
      "epoch": 4.419851682829435,
      "grad_norm": 0.74609375,
      "learning_rate": 1.6363628492051225e-06,
      "loss": 0.8648,
      "step": 8233
    },
    {
      "epoch": 4.420388577564511,
      "grad_norm": 0.6328125,
      "learning_rate": 1.6333590319433461e-06,
      "loss": 0.7889,
      "step": 8234
    },
    {
      "epoch": 4.420925472299587,
      "grad_norm": 0.61328125,
      "learning_rate": 1.6303578811112246e-06,
      "loss": 0.7976,
      "step": 8235
    },
    {
      "epoch": 4.421462367034663,
      "grad_norm": 0.80859375,
      "learning_rate": 1.6273593970512174e-06,
      "loss": 0.8777,
      "step": 8236
    },
    {
      "epoch": 4.42199926176974,
      "grad_norm": 0.671875,
      "learning_rate": 1.624363580105495e-06,
      "loss": 0.8699,
      "step": 8237
    },
    {
      "epoch": 4.422536156504815,
      "grad_norm": 0.80859375,
      "learning_rate": 1.6213704306159178e-06,
      "loss": 0.7813,
      "step": 8238
    },
    {
      "epoch": 4.423073051239891,
      "grad_norm": 0.625,
      "learning_rate": 1.6183799489240398e-06,
      "loss": 0.7055,
      "step": 8239
    },
    {
      "epoch": 4.423609945974968,
      "grad_norm": 0.78125,
      "learning_rate": 1.615392135371116e-06,
      "loss": 0.8254,
      "step": 8240
    },
    {
      "epoch": 4.424146840710043,
      "grad_norm": 0.76953125,
      "learning_rate": 1.612406990298093e-06,
      "loss": 0.9572,
      "step": 8241
    },
    {
      "epoch": 4.424683735445119,
      "grad_norm": 0.6875,
      "learning_rate": 1.6094245140456072e-06,
      "loss": 0.8455,
      "step": 8242
    },
    {
      "epoch": 4.4252206301801955,
      "grad_norm": 0.66015625,
      "learning_rate": 1.6064447069540024e-06,
      "loss": 0.762,
      "step": 8243
    },
    {
      "epoch": 4.425757524915271,
      "grad_norm": 0.6328125,
      "learning_rate": 1.603467569363315e-06,
      "loss": 0.8894,
      "step": 8244
    },
    {
      "epoch": 4.426294419650347,
      "grad_norm": 0.66015625,
      "learning_rate": 1.600493101613268e-06,
      "loss": 0.8646,
      "step": 8245
    },
    {
      "epoch": 4.426831314385423,
      "grad_norm": 0.63671875,
      "learning_rate": 1.5975213040432897e-06,
      "loss": 0.9088,
      "step": 8246
    },
    {
      "epoch": 4.427368209120499,
      "grad_norm": 0.7421875,
      "learning_rate": 1.5945521769925004e-06,
      "loss": 0.912,
      "step": 8247
    },
    {
      "epoch": 4.427905103855576,
      "grad_norm": 0.66015625,
      "learning_rate": 1.5915857207997126e-06,
      "loss": 0.9382,
      "step": 8248
    },
    {
      "epoch": 4.428441998590651,
      "grad_norm": 0.64453125,
      "learning_rate": 1.5886219358034387e-06,
      "loss": 0.7741,
      "step": 8249
    },
    {
      "epoch": 4.428978893325727,
      "grad_norm": 0.64453125,
      "learning_rate": 1.5856608223418806e-06,
      "loss": 0.7866,
      "step": 8250
    },
    {
      "epoch": 4.428978893325727,
      "eval_loss": 0.9027072787284851,
      "eval_runtime": 46.2751,
      "eval_samples_per_second": 20.789,
      "eval_steps_per_second": 20.789,
      "step": 8250
    },
    {
      "epoch": 4.429515788060804,
      "grad_norm": 0.70703125,
      "learning_rate": 1.5827023807529456e-06,
      "loss": 0.8154,
      "step": 8251
    },
    {
      "epoch": 4.430052682795879,
      "grad_norm": 0.6171875,
      "learning_rate": 1.579746611374222e-06,
      "loss": 0.8866,
      "step": 8252
    },
    {
      "epoch": 4.430589577530955,
      "grad_norm": 0.62890625,
      "learning_rate": 1.5767935145430068e-06,
      "loss": 0.7472,
      "step": 8253
    },
    {
      "epoch": 4.4311264722660315,
      "grad_norm": 0.6953125,
      "learning_rate": 1.5738430905962776e-06,
      "loss": 0.9132,
      "step": 8254
    },
    {
      "epoch": 4.431663367001107,
      "grad_norm": 0.6484375,
      "learning_rate": 1.5708953398707204e-06,
      "loss": 0.8426,
      "step": 8255
    },
    {
      "epoch": 4.432200261736184,
      "grad_norm": 0.84765625,
      "learning_rate": 1.5679502627027136e-06,
      "loss": 0.9216,
      "step": 8256
    },
    {
      "epoch": 4.432737156471259,
      "grad_norm": 0.6484375,
      "learning_rate": 1.5650078594283275e-06,
      "loss": 0.6307,
      "step": 8257
    },
    {
      "epoch": 4.433274051206335,
      "grad_norm": 0.8359375,
      "learning_rate": 1.5620681303833178e-06,
      "loss": 0.8897,
      "step": 8258
    },
    {
      "epoch": 4.433810945941412,
      "grad_norm": 0.671875,
      "learning_rate": 1.559131075903153e-06,
      "loss": 0.7865,
      "step": 8259
    },
    {
      "epoch": 4.434347840676487,
      "grad_norm": 0.70703125,
      "learning_rate": 1.5561966963229924e-06,
      "loss": 0.8214,
      "step": 8260
    },
    {
      "epoch": 4.434884735411563,
      "grad_norm": 0.63671875,
      "learning_rate": 1.5532649919776738e-06,
      "loss": 0.8391,
      "step": 8261
    },
    {
      "epoch": 4.43542163014664,
      "grad_norm": 0.8125,
      "learning_rate": 1.550335963201749e-06,
      "loss": 0.6785,
      "step": 8262
    },
    {
      "epoch": 4.435958524881715,
      "grad_norm": 0.73828125,
      "learning_rate": 1.5474096103294616e-06,
      "loss": 0.8154,
      "step": 8263
    },
    {
      "epoch": 4.436495419616792,
      "grad_norm": 0.5859375,
      "learning_rate": 1.5444859336947393e-06,
      "loss": 0.6973,
      "step": 8264
    },
    {
      "epoch": 4.4370323143518675,
      "grad_norm": 0.7109375,
      "learning_rate": 1.5415649336312093e-06,
      "loss": 0.8809,
      "step": 8265
    },
    {
      "epoch": 4.437569209086943,
      "grad_norm": 0.57421875,
      "learning_rate": 1.5386466104722053e-06,
      "loss": 0.983,
      "step": 8266
    },
    {
      "epoch": 4.43810610382202,
      "grad_norm": 0.7109375,
      "learning_rate": 1.5357309645507273e-06,
      "loss": 0.6873,
      "step": 8267
    },
    {
      "epoch": 4.438642998557095,
      "grad_norm": 0.6796875,
      "learning_rate": 1.532817996199501e-06,
      "loss": 0.8135,
      "step": 8268
    },
    {
      "epoch": 4.439179893292171,
      "grad_norm": 0.6796875,
      "learning_rate": 1.5299077057509297e-06,
      "loss": 0.7436,
      "step": 8269
    },
    {
      "epoch": 4.439716788027248,
      "grad_norm": 0.70703125,
      "learning_rate": 1.527000093537112e-06,
      "loss": 0.8359,
      "step": 8270
    },
    {
      "epoch": 4.440253682762323,
      "grad_norm": 0.70703125,
      "learning_rate": 1.5240951598898463e-06,
      "loss": 0.9547,
      "step": 8271
    },
    {
      "epoch": 4.440790577497399,
      "grad_norm": 0.88671875,
      "learning_rate": 1.5211929051406226e-06,
      "loss": 0.9057,
      "step": 8272
    },
    {
      "epoch": 4.441327472232476,
      "grad_norm": 0.7421875,
      "learning_rate": 1.5182933296206208e-06,
      "loss": 0.9338,
      "step": 8273
    },
    {
      "epoch": 4.441864366967551,
      "grad_norm": 0.78515625,
      "learning_rate": 1.5153964336607234e-06,
      "loss": 0.9568,
      "step": 8274
    },
    {
      "epoch": 4.442401261702628,
      "grad_norm": 0.62890625,
      "learning_rate": 1.5125022175914994e-06,
      "loss": 0.6984,
      "step": 8275
    },
    {
      "epoch": 4.4429381564377035,
      "grad_norm": 0.6640625,
      "learning_rate": 1.5096106817432231e-06,
      "loss": 0.7784,
      "step": 8276
    },
    {
      "epoch": 4.443475051172779,
      "grad_norm": 0.79296875,
      "learning_rate": 1.5067218264458422e-06,
      "loss": 0.7825,
      "step": 8277
    },
    {
      "epoch": 4.444011945907856,
      "grad_norm": 0.78125,
      "learning_rate": 1.503835652029023e-06,
      "loss": 0.9397,
      "step": 8278
    },
    {
      "epoch": 4.444548840642931,
      "grad_norm": 0.67578125,
      "learning_rate": 1.500952158822111e-06,
      "loss": 0.7453,
      "step": 8279
    },
    {
      "epoch": 4.445085735378007,
      "grad_norm": 0.63671875,
      "learning_rate": 1.498071347154148e-06,
      "loss": 0.7393,
      "step": 8280
    },
    {
      "epoch": 4.445622630113084,
      "grad_norm": 0.70703125,
      "learning_rate": 1.4951932173538713e-06,
      "loss": 0.8314,
      "step": 8281
    },
    {
      "epoch": 4.446159524848159,
      "grad_norm": 0.640625,
      "learning_rate": 1.492317769749721e-06,
      "loss": 0.9092,
      "step": 8282
    },
    {
      "epoch": 4.446696419583235,
      "grad_norm": 0.5859375,
      "learning_rate": 1.4894450046698039e-06,
      "loss": 0.6895,
      "step": 8283
    },
    {
      "epoch": 4.447233314318312,
      "grad_norm": 0.7421875,
      "learning_rate": 1.4865749224419496e-06,
      "loss": 0.7358,
      "step": 8284
    },
    {
      "epoch": 4.447770209053387,
      "grad_norm": 0.6796875,
      "learning_rate": 1.4837075233936765e-06,
      "loss": 1.0611,
      "step": 8285
    },
    {
      "epoch": 4.448307103788464,
      "grad_norm": 0.84765625,
      "learning_rate": 1.4808428078521785e-06,
      "loss": 0.731,
      "step": 8286
    },
    {
      "epoch": 4.4488439985235395,
      "grad_norm": 0.67578125,
      "learning_rate": 1.4779807761443636e-06,
      "loss": 0.9906,
      "step": 8287
    },
    {
      "epoch": 4.449380893258615,
      "grad_norm": 0.625,
      "learning_rate": 1.4751214285968263e-06,
      "loss": 0.8957,
      "step": 8288
    },
    {
      "epoch": 4.449917787993692,
      "grad_norm": 0.8828125,
      "learning_rate": 1.4722647655358496e-06,
      "loss": 0.8662,
      "step": 8289
    },
    {
      "epoch": 4.450454682728767,
      "grad_norm": 0.55859375,
      "learning_rate": 1.4694107872874175e-06,
      "loss": 0.8454,
      "step": 8290
    },
    {
      "epoch": 4.450991577463843,
      "grad_norm": 0.61328125,
      "learning_rate": 1.466559494177211e-06,
      "loss": 0.7678,
      "step": 8291
    },
    {
      "epoch": 4.45152847219892,
      "grad_norm": 0.62890625,
      "learning_rate": 1.463710886530592e-06,
      "loss": 0.8827,
      "step": 8292
    },
    {
      "epoch": 4.452065366933995,
      "grad_norm": 0.66796875,
      "learning_rate": 1.4608649646726202e-06,
      "loss": 0.8198,
      "step": 8293
    },
    {
      "epoch": 4.452602261669072,
      "grad_norm": 0.84375,
      "learning_rate": 1.4580217289280524e-06,
      "loss": 1.0586,
      "step": 8294
    },
    {
      "epoch": 4.453139156404148,
      "grad_norm": 0.71484375,
      "learning_rate": 1.4551811796213483e-06,
      "loss": 0.8704,
      "step": 8295
    },
    {
      "epoch": 4.453676051139223,
      "grad_norm": 0.796875,
      "learning_rate": 1.4523433170766348e-06,
      "loss": 0.7902,
      "step": 8296
    },
    {
      "epoch": 4.4542129458743,
      "grad_norm": 0.8984375,
      "learning_rate": 1.4495081416177613e-06,
      "loss": 0.8432,
      "step": 8297
    },
    {
      "epoch": 4.4547498406093755,
      "grad_norm": 0.62109375,
      "learning_rate": 1.4466756535682496e-06,
      "loss": 0.8071,
      "step": 8298
    },
    {
      "epoch": 4.455286735344451,
      "grad_norm": 0.69140625,
      "learning_rate": 1.4438458532513272e-06,
      "loss": 0.8832,
      "step": 8299
    },
    {
      "epoch": 4.455823630079528,
      "grad_norm": 0.6171875,
      "learning_rate": 1.4410187409899023e-06,
      "loss": 0.7187,
      "step": 8300
    },
    {
      "epoch": 4.455823630079528,
      "eval_loss": 0.9027251601219177,
      "eval_runtime": 46.0364,
      "eval_samples_per_second": 20.896,
      "eval_steps_per_second": 20.896,
      "step": 8300
    },
    {
      "epoch": 4.456360524814603,
      "grad_norm": 0.640625,
      "learning_rate": 1.4381943171065947e-06,
      "loss": 0.7582,
      "step": 8301
    },
    {
      "epoch": 4.45689741954968,
      "grad_norm": 0.76171875,
      "learning_rate": 1.4353725819236969e-06,
      "loss": 0.8503,
      "step": 8302
    },
    {
      "epoch": 4.457434314284756,
      "grad_norm": 0.66015625,
      "learning_rate": 1.432553535763212e-06,
      "loss": 0.7626,
      "step": 8303
    },
    {
      "epoch": 4.457971209019831,
      "grad_norm": 0.61328125,
      "learning_rate": 1.4297371789468272e-06,
      "loss": 0.8649,
      "step": 8304
    },
    {
      "epoch": 4.458508103754908,
      "grad_norm": 1.015625,
      "learning_rate": 1.4269235117959218e-06,
      "loss": 0.93,
      "step": 8305
    },
    {
      "epoch": 4.459044998489984,
      "grad_norm": 1.859375,
      "learning_rate": 1.4241125346315691e-06,
      "loss": 0.9761,
      "step": 8306
    },
    {
      "epoch": 4.459581893225059,
      "grad_norm": 0.69921875,
      "learning_rate": 1.421304247774552e-06,
      "loss": 0.7589,
      "step": 8307
    },
    {
      "epoch": 4.460118787960136,
      "grad_norm": 0.671875,
      "learning_rate": 1.4184986515453108e-06,
      "loss": 0.6836,
      "step": 8308
    },
    {
      "epoch": 4.4606556826952115,
      "grad_norm": 0.6875,
      "learning_rate": 1.4156957462640092e-06,
      "loss": 0.8432,
      "step": 8309
    },
    {
      "epoch": 4.461192577430287,
      "grad_norm": 0.56640625,
      "learning_rate": 1.4128955322504966e-06,
      "loss": 0.6673,
      "step": 8310
    },
    {
      "epoch": 4.461729472165364,
      "grad_norm": 0.77734375,
      "learning_rate": 1.4100980098243093e-06,
      "loss": 0.9891,
      "step": 8311
    },
    {
      "epoch": 4.462266366900439,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4073031793046782e-06,
      "loss": 0.8597,
      "step": 8312
    },
    {
      "epoch": 4.462803261635516,
      "grad_norm": 0.7265625,
      "learning_rate": 1.4045110410105367e-06,
      "loss": 1.0593,
      "step": 8313
    },
    {
      "epoch": 4.463340156370592,
      "grad_norm": 0.65625,
      "learning_rate": 1.401721595260494e-06,
      "loss": 0.8861,
      "step": 8314
    },
    {
      "epoch": 4.463877051105667,
      "grad_norm": 0.76171875,
      "learning_rate": 1.3989348423728678e-06,
      "loss": 0.6185,
      "step": 8315
    },
    {
      "epoch": 4.464413945840744,
      "grad_norm": 0.74609375,
      "learning_rate": 1.3961507826656589e-06,
      "loss": 0.9124,
      "step": 8316
    },
    {
      "epoch": 4.46495084057582,
      "grad_norm": 0.6171875,
      "learning_rate": 1.3933694164565663e-06,
      "loss": 0.8167,
      "step": 8317
    },
    {
      "epoch": 4.465487735310895,
      "grad_norm": 0.6171875,
      "learning_rate": 1.3905907440629752e-06,
      "loss": 0.7692,
      "step": 8318
    },
    {
      "epoch": 4.466024630045972,
      "grad_norm": 0.64453125,
      "learning_rate": 1.3878147658019674e-06,
      "loss": 0.6746,
      "step": 8319
    },
    {
      "epoch": 4.4665615247810475,
      "grad_norm": 0.7734375,
      "learning_rate": 1.3850414819903235e-06,
      "loss": 0.8528,
      "step": 8320
    },
    {
      "epoch": 4.467098419516123,
      "grad_norm": 0.58203125,
      "learning_rate": 1.382270892944501e-06,
      "loss": 0.7957,
      "step": 8321
    },
    {
      "epoch": 4.4676353142512,
      "grad_norm": 0.6875,
      "learning_rate": 1.379502998980664e-06,
      "loss": 0.8642,
      "step": 8322
    },
    {
      "epoch": 4.468172208986275,
      "grad_norm": 0.6796875,
      "learning_rate": 1.3767378004146681e-06,
      "loss": 0.7861,
      "step": 8323
    },
    {
      "epoch": 4.468709103721352,
      "grad_norm": 0.75390625,
      "learning_rate": 1.3739752975620523e-06,
      "loss": 0.8906,
      "step": 8324
    },
    {
      "epoch": 4.469245998456428,
      "grad_norm": 0.7109375,
      "learning_rate": 1.3712154907380537e-06,
      "loss": 0.7889,
      "step": 8325
    },
    {
      "epoch": 4.469782893191503,
      "grad_norm": 1.2578125,
      "learning_rate": 1.3684583802576034e-06,
      "loss": 0.8172,
      "step": 8326
    },
    {
      "epoch": 4.47031978792658,
      "grad_norm": 0.6875,
      "learning_rate": 1.3657039664353166e-06,
      "loss": 0.9455,
      "step": 8327
    },
    {
      "epoch": 4.470856682661656,
      "grad_norm": 0.671875,
      "learning_rate": 1.362952249585514e-06,
      "loss": 0.8061,
      "step": 8328
    },
    {
      "epoch": 4.471393577396731,
      "grad_norm": 0.70703125,
      "learning_rate": 1.3602032300222e-06,
      "loss": 0.8529,
      "step": 8329
    },
    {
      "epoch": 4.471930472131808,
      "grad_norm": 0.83203125,
      "learning_rate": 1.357456908059071e-06,
      "loss": 0.7752,
      "step": 8330
    },
    {
      "epoch": 4.4724673668668835,
      "grad_norm": 0.78125,
      "learning_rate": 1.354713284009515e-06,
      "loss": 0.8533,
      "step": 8331
    },
    {
      "epoch": 4.47300426160196,
      "grad_norm": 0.58984375,
      "learning_rate": 1.3519723581866207e-06,
      "loss": 0.6661,
      "step": 8332
    },
    {
      "epoch": 4.473541156337036,
      "grad_norm": 0.60546875,
      "learning_rate": 1.349234130903157e-06,
      "loss": 0.7504,
      "step": 8333
    },
    {
      "epoch": 4.474078051072111,
      "grad_norm": 0.8203125,
      "learning_rate": 1.3464986024715908e-06,
      "loss": 0.9392,
      "step": 8334
    },
    {
      "epoch": 4.474614945807188,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3437657732040782e-06,
      "loss": 0.8801,
      "step": 8335
    },
    {
      "epoch": 4.475151840542264,
      "grad_norm": 0.74609375,
      "learning_rate": 1.3410356434124777e-06,
      "loss": 0.8965,
      "step": 8336
    },
    {
      "epoch": 4.475688735277339,
      "grad_norm": 0.66796875,
      "learning_rate": 1.3383082134083237e-06,
      "loss": 0.8219,
      "step": 8337
    },
    {
      "epoch": 4.476225630012416,
      "grad_norm": 0.7421875,
      "learning_rate": 1.3355834835028586e-06,
      "loss": 0.8903,
      "step": 8338
    },
    {
      "epoch": 4.476762524747492,
      "grad_norm": 0.69921875,
      "learning_rate": 1.3328614540069979e-06,
      "loss": 0.8317,
      "step": 8339
    },
    {
      "epoch": 4.477299419482568,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3301421252313651e-06,
      "loss": 0.8192,
      "step": 8340
    },
    {
      "epoch": 4.477836314217644,
      "grad_norm": 0.65234375,
      "learning_rate": 1.327425497486276e-06,
      "loss": 0.7166,
      "step": 8341
    },
    {
      "epoch": 4.4783732089527195,
      "grad_norm": 0.703125,
      "learning_rate": 1.3247115710817242e-06,
      "loss": 0.9011,
      "step": 8342
    },
    {
      "epoch": 4.478910103687796,
      "grad_norm": 0.625,
      "learning_rate": 1.3220003463274038e-06,
      "loss": 0.78,
      "step": 8343
    },
    {
      "epoch": 4.479446998422872,
      "grad_norm": 0.74609375,
      "learning_rate": 1.3192918235327006e-06,
      "loss": 0.8576,
      "step": 8344
    },
    {
      "epoch": 4.479983893157947,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3165860030066951e-06,
      "loss": 0.9406,
      "step": 8345
    },
    {
      "epoch": 4.480520787893024,
      "grad_norm": 0.609375,
      "learning_rate": 1.3138828850581515e-06,
      "loss": 0.8202,
      "step": 8346
    },
    {
      "epoch": 4.4810576826281,
      "grad_norm": 0.79296875,
      "learning_rate": 1.3111824699955283e-06,
      "loss": 0.899,
      "step": 8347
    },
    {
      "epoch": 4.481594577363175,
      "grad_norm": 0.59375,
      "learning_rate": 1.3084847581269849e-06,
      "loss": 0.7537,
      "step": 8348
    },
    {
      "epoch": 4.482131472098252,
      "grad_norm": 0.64453125,
      "learning_rate": 1.305789749760361e-06,
      "loss": 0.6815,
      "step": 8349
    },
    {
      "epoch": 4.482668366833328,
      "grad_norm": 0.796875,
      "learning_rate": 1.303097445203183e-06,
      "loss": 0.8237,
      "step": 8350
    },
    {
      "epoch": 4.482668366833328,
      "eval_loss": 0.9026736617088318,
      "eval_runtime": 46.1503,
      "eval_samples_per_second": 20.845,
      "eval_steps_per_second": 20.845,
      "step": 8350
    },
    {
      "epoch": 4.483205261568404,
      "grad_norm": 0.671875,
      "learning_rate": 1.3004078447626883e-06,
      "loss": 0.9488,
      "step": 8351
    },
    {
      "epoch": 4.48374215630348,
      "grad_norm": 0.70703125,
      "learning_rate": 1.297720948745787e-06,
      "loss": 1.0053,
      "step": 8352
    },
    {
      "epoch": 4.4842790510385555,
      "grad_norm": 0.69921875,
      "learning_rate": 1.2950367574590894e-06,
      "loss": 0.7899,
      "step": 8353
    },
    {
      "epoch": 4.484815945773632,
      "grad_norm": 0.75,
      "learning_rate": 1.2923552712089005e-06,
      "loss": 1.1165,
      "step": 8354
    },
    {
      "epoch": 4.485352840508708,
      "grad_norm": 0.7578125,
      "learning_rate": 1.289676490301206e-06,
      "loss": 0.9944,
      "step": 8355
    },
    {
      "epoch": 4.485889735243783,
      "grad_norm": 0.625,
      "learning_rate": 1.2870004150416893e-06,
      "loss": 0.7521,
      "step": 8356
    },
    {
      "epoch": 4.48642662997886,
      "grad_norm": 0.68359375,
      "learning_rate": 1.2843270457357283e-06,
      "loss": 0.8393,
      "step": 8357
    },
    {
      "epoch": 4.486963524713936,
      "grad_norm": 0.76953125,
      "learning_rate": 1.28165638268839e-06,
      "loss": 0.889,
      "step": 8358
    },
    {
      "epoch": 4.487500419449011,
      "grad_norm": 0.78125,
      "learning_rate": 1.2789884262044194e-06,
      "loss": 0.9361,
      "step": 8359
    },
    {
      "epoch": 4.488037314184088,
      "grad_norm": 0.71875,
      "learning_rate": 1.2763231765882732e-06,
      "loss": 0.8338,
      "step": 8360
    },
    {
      "epoch": 4.488574208919164,
      "grad_norm": 0.73828125,
      "learning_rate": 1.2736606341440916e-06,
      "loss": 0.8465,
      "step": 8361
    },
    {
      "epoch": 4.48911110365424,
      "grad_norm": 0.7578125,
      "learning_rate": 1.2710007991757005e-06,
      "loss": 0.9843,
      "step": 8362
    },
    {
      "epoch": 4.489647998389316,
      "grad_norm": 0.62890625,
      "learning_rate": 1.2683436719866188e-06,
      "loss": 0.6926,
      "step": 8363
    },
    {
      "epoch": 4.4901848931243915,
      "grad_norm": 0.6953125,
      "learning_rate": 1.2656892528800674e-06,
      "loss": 0.8333,
      "step": 8364
    },
    {
      "epoch": 4.490721787859468,
      "grad_norm": 0.6328125,
      "learning_rate": 1.2630375421589375e-06,
      "loss": 0.8713,
      "step": 8365
    },
    {
      "epoch": 4.491258682594544,
      "grad_norm": 0.71875,
      "learning_rate": 1.2603885401258314e-06,
      "loss": 0.8221,
      "step": 8366
    },
    {
      "epoch": 4.491795577329619,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2577422470830324e-06,
      "loss": 1.1343,
      "step": 8367
    },
    {
      "epoch": 4.492332472064696,
      "grad_norm": 0.90625,
      "learning_rate": 1.2550986633325102e-06,
      "loss": 0.8528,
      "step": 8368
    },
    {
      "epoch": 4.492869366799772,
      "grad_norm": 0.609375,
      "learning_rate": 1.2524577891759342e-06,
      "loss": 0.8208,
      "step": 8369
    },
    {
      "epoch": 4.493406261534848,
      "grad_norm": 0.8203125,
      "learning_rate": 1.2498196249146665e-06,
      "loss": 0.9491,
      "step": 8370
    },
    {
      "epoch": 4.493943156269924,
      "grad_norm": 1.0,
      "learning_rate": 1.2471841708497494e-06,
      "loss": 0.9982,
      "step": 8371
    },
    {
      "epoch": 4.494480051005,
      "grad_norm": 0.6171875,
      "learning_rate": 1.244551427281923e-06,
      "loss": 0.5941,
      "step": 8372
    },
    {
      "epoch": 4.495016945740076,
      "grad_norm": 0.61328125,
      "learning_rate": 1.2419213945116194e-06,
      "loss": 0.8029,
      "step": 8373
    },
    {
      "epoch": 4.495553840475152,
      "grad_norm": 0.765625,
      "learning_rate": 1.2392940728389536e-06,
      "loss": 0.7634,
      "step": 8374
    },
    {
      "epoch": 4.4960907352102275,
      "grad_norm": 0.6796875,
      "learning_rate": 1.2366694625637442e-06,
      "loss": 0.8302,
      "step": 8375
    },
    {
      "epoch": 4.496627629945304,
      "grad_norm": 0.6484375,
      "learning_rate": 1.2340475639854826e-06,
      "loss": 0.8304,
      "step": 8376
    },
    {
      "epoch": 4.49716452468038,
      "grad_norm": 0.76171875,
      "learning_rate": 1.231428377403368e-06,
      "loss": 1.0635,
      "step": 8377
    },
    {
      "epoch": 4.497701419415456,
      "grad_norm": 0.81640625,
      "learning_rate": 1.2288119031162782e-06,
      "loss": 0.828,
      "step": 8378
    },
    {
      "epoch": 4.498238314150532,
      "grad_norm": 0.69921875,
      "learning_rate": 1.2261981414227852e-06,
      "loss": 0.9418,
      "step": 8379
    },
    {
      "epoch": 4.498775208885608,
      "grad_norm": 0.6640625,
      "learning_rate": 1.2235870926211619e-06,
      "loss": 0.8015,
      "step": 8380
    },
    {
      "epoch": 4.499312103620684,
      "grad_norm": 0.734375,
      "learning_rate": 1.2209787570093501e-06,
      "loss": 0.783,
      "step": 8381
    },
    {
      "epoch": 4.49984899835576,
      "grad_norm": 0.6640625,
      "learning_rate": 1.2183731348850035e-06,
      "loss": 0.8503,
      "step": 8382
    },
    {
      "epoch": 4.500385893090836,
      "grad_norm": 0.76171875,
      "learning_rate": 1.2157702265454512e-06,
      "loss": 0.9928,
      "step": 8383
    },
    {
      "epoch": 4.500922787825912,
      "grad_norm": 0.6171875,
      "learning_rate": 1.2131700322877166e-06,
      "loss": 0.7411,
      "step": 8384
    },
    {
      "epoch": 4.501459682560988,
      "grad_norm": 0.65234375,
      "learning_rate": 1.210572552408515e-06,
      "loss": 0.8119,
      "step": 8385
    },
    {
      "epoch": 4.5019965772960635,
      "grad_norm": 0.734375,
      "learning_rate": 1.2079777872042592e-06,
      "loss": 1.1473,
      "step": 8386
    },
    {
      "epoch": 4.50253347203114,
      "grad_norm": 0.65234375,
      "learning_rate": 1.2053857369710348e-06,
      "loss": 0.831,
      "step": 8387
    },
    {
      "epoch": 4.503070366766216,
      "grad_norm": 0.67578125,
      "learning_rate": 1.2027964020046329e-06,
      "loss": 0.7662,
      "step": 8388
    },
    {
      "epoch": 4.503607261501292,
      "grad_norm": 0.6640625,
      "learning_rate": 1.2002097826005337e-06,
      "loss": 0.9336,
      "step": 8389
    },
    {
      "epoch": 4.504144156236368,
      "grad_norm": 0.734375,
      "learning_rate": 1.197625879053893e-06,
      "loss": 0.9547,
      "step": 8390
    },
    {
      "epoch": 4.504681050971444,
      "grad_norm": 0.703125,
      "learning_rate": 1.1950446916595748e-06,
      "loss": 0.979,
      "step": 8391
    },
    {
      "epoch": 4.50521794570652,
      "grad_norm": 0.80859375,
      "learning_rate": 1.192466220712124e-06,
      "loss": 0.7748,
      "step": 8392
    },
    {
      "epoch": 4.505754840441596,
      "grad_norm": 0.53515625,
      "learning_rate": 1.189890466505772e-06,
      "loss": 0.7529,
      "step": 8393
    },
    {
      "epoch": 4.506291735176672,
      "grad_norm": 0.73828125,
      "learning_rate": 1.1873174293344474e-06,
      "loss": 0.9212,
      "step": 8394
    },
    {
      "epoch": 4.506828629911748,
      "grad_norm": 0.6484375,
      "learning_rate": 1.1847471094917712e-06,
      "loss": 0.8835,
      "step": 8395
    },
    {
      "epoch": 4.507365524646824,
      "grad_norm": 0.75390625,
      "learning_rate": 1.1821795072710446e-06,
      "loss": 0.7913,
      "step": 8396
    },
    {
      "epoch": 4.5079024193818995,
      "grad_norm": 0.640625,
      "learning_rate": 1.179614622965261e-06,
      "loss": 0.7279,
      "step": 8397
    },
    {
      "epoch": 4.508439314116976,
      "grad_norm": 0.75,
      "learning_rate": 1.1770524568671142e-06,
      "loss": 0.8137,
      "step": 8398
    },
    {
      "epoch": 4.508976208852052,
      "grad_norm": 0.69921875,
      "learning_rate": 1.1744930092689733e-06,
      "loss": 0.9252,
      "step": 8399
    },
    {
      "epoch": 4.509513103587128,
      "grad_norm": 0.59765625,
      "learning_rate": 1.171936280462907e-06,
      "loss": 0.8223,
      "step": 8400
    },
    {
      "epoch": 4.509513103587128,
      "eval_loss": 0.9027235507965088,
      "eval_runtime": 46.0731,
      "eval_samples_per_second": 20.88,
      "eval_steps_per_second": 20.88,
      "step": 8400
    },
    {
      "epoch": 4.510049998322204,
      "grad_norm": 0.71875,
      "learning_rate": 1.169382270740668e-06,
      "loss": 0.8002,
      "step": 8401
    },
    {
      "epoch": 4.51058689305728,
      "grad_norm": 0.73046875,
      "learning_rate": 1.1668309803937044e-06,
      "loss": 0.9601,
      "step": 8402
    },
    {
      "epoch": 4.511123787792356,
      "grad_norm": 0.59375,
      "learning_rate": 1.1642824097131438e-06,
      "loss": 0.6579,
      "step": 8403
    },
    {
      "epoch": 4.511660682527432,
      "grad_norm": 0.7890625,
      "learning_rate": 1.1617365589898178e-06,
      "loss": 0.8987,
      "step": 8404
    },
    {
      "epoch": 4.512197577262508,
      "grad_norm": 0.91015625,
      "learning_rate": 1.1591934285142408e-06,
      "loss": 0.9232,
      "step": 8405
    },
    {
      "epoch": 4.512734471997584,
      "grad_norm": 0.8671875,
      "learning_rate": 1.156653018576609e-06,
      "loss": 0.8704,
      "step": 8406
    },
    {
      "epoch": 4.51327136673266,
      "grad_norm": 0.69140625,
      "learning_rate": 1.1541153294668178e-06,
      "loss": 0.8116,
      "step": 8407
    },
    {
      "epoch": 4.5138082614677355,
      "grad_norm": 0.71484375,
      "learning_rate": 1.151580361474458e-06,
      "loss": 0.9706,
      "step": 8408
    },
    {
      "epoch": 4.514345156202812,
      "grad_norm": 0.62890625,
      "learning_rate": 1.14904811488879e-06,
      "loss": 0.8165,
      "step": 8409
    },
    {
      "epoch": 4.514882050937888,
      "grad_norm": 0.78515625,
      "learning_rate": 1.1465185899987797e-06,
      "loss": 0.9603,
      "step": 8410
    },
    {
      "epoch": 4.515418945672964,
      "grad_norm": 0.6484375,
      "learning_rate": 1.1439917870930793e-06,
      "loss": 0.8335,
      "step": 8411
    },
    {
      "epoch": 4.51595584040804,
      "grad_norm": 0.60546875,
      "learning_rate": 1.1414677064600282e-06,
      "loss": 0.7845,
      "step": 8412
    },
    {
      "epoch": 4.516492735143116,
      "grad_norm": 0.68359375,
      "learning_rate": 1.138946348387654e-06,
      "loss": 0.7134,
      "step": 8413
    },
    {
      "epoch": 4.517029629878192,
      "grad_norm": 0.796875,
      "learning_rate": 1.136427713163679e-06,
      "loss": 0.918,
      "step": 8414
    },
    {
      "epoch": 4.517566524613268,
      "grad_norm": 0.60546875,
      "learning_rate": 1.1339118010755073e-06,
      "loss": 0.826,
      "step": 8415
    },
    {
      "epoch": 4.5181034193483445,
      "grad_norm": 0.62890625,
      "learning_rate": 1.1313986124102394e-06,
      "loss": 0.8624,
      "step": 8416
    },
    {
      "epoch": 4.51864031408342,
      "grad_norm": 0.65625,
      "learning_rate": 1.1288881474546625e-06,
      "loss": 0.7929,
      "step": 8417
    },
    {
      "epoch": 4.519177208818496,
      "grad_norm": 0.953125,
      "learning_rate": 1.1263804064952532e-06,
      "loss": 0.8775,
      "step": 8418
    },
    {
      "epoch": 4.519714103553572,
      "grad_norm": 0.65234375,
      "learning_rate": 1.1238753898181715e-06,
      "loss": 0.8291,
      "step": 8419
    },
    {
      "epoch": 4.520250998288648,
      "grad_norm": 0.72265625,
      "learning_rate": 1.1213730977092746e-06,
      "loss": 1.0208,
      "step": 8420
    },
    {
      "epoch": 4.520787893023724,
      "grad_norm": 0.65234375,
      "learning_rate": 1.1188735304541065e-06,
      "loss": 0.8897,
      "step": 8421
    },
    {
      "epoch": 4.5213247877588,
      "grad_norm": 0.66015625,
      "learning_rate": 1.1163766883378973e-06,
      "loss": 0.877,
      "step": 8422
    },
    {
      "epoch": 4.521861682493876,
      "grad_norm": 1.1640625,
      "learning_rate": 1.113882571645572e-06,
      "loss": 0.8493,
      "step": 8423
    },
    {
      "epoch": 4.522398577228952,
      "grad_norm": 0.80859375,
      "learning_rate": 1.111391180661736e-06,
      "loss": 0.9505,
      "step": 8424
    },
    {
      "epoch": 4.522935471964028,
      "grad_norm": 0.63671875,
      "learning_rate": 1.1089025156706928e-06,
      "loss": 0.8568,
      "step": 8425
    },
    {
      "epoch": 4.523472366699104,
      "grad_norm": 0.63671875,
      "learning_rate": 1.106416576956426e-06,
      "loss": 0.9714,
      "step": 8426
    },
    {
      "epoch": 4.5240092614341805,
      "grad_norm": 0.6484375,
      "learning_rate": 1.10393336480262e-06,
      "loss": 0.8765,
      "step": 8427
    },
    {
      "epoch": 4.524546156169256,
      "grad_norm": 0.65234375,
      "learning_rate": 1.1014528794926316e-06,
      "loss": 0.7755,
      "step": 8428
    },
    {
      "epoch": 4.525083050904332,
      "grad_norm": 0.68359375,
      "learning_rate": 1.09897512130952e-06,
      "loss": 0.798,
      "step": 8429
    },
    {
      "epoch": 4.525619945639408,
      "grad_norm": 0.66796875,
      "learning_rate": 1.0965000905360345e-06,
      "loss": 1.0487,
      "step": 8430
    },
    {
      "epoch": 4.526156840374484,
      "grad_norm": 0.60546875,
      "learning_rate": 1.0940277874545962e-06,
      "loss": 0.6858,
      "step": 8431
    },
    {
      "epoch": 4.52669373510956,
      "grad_norm": 0.8046875,
      "learning_rate": 1.091558212347335e-06,
      "loss": 0.9631,
      "step": 8432
    },
    {
      "epoch": 4.527230629844636,
      "grad_norm": 0.7265625,
      "learning_rate": 1.0890913654960617e-06,
      "loss": 0.8211,
      "step": 8433
    },
    {
      "epoch": 4.527767524579712,
      "grad_norm": 0.69140625,
      "learning_rate": 1.0866272471822647e-06,
      "loss": 0.8105,
      "step": 8434
    },
    {
      "epoch": 4.528304419314788,
      "grad_norm": 0.5703125,
      "learning_rate": 1.0841658576871388e-06,
      "loss": 0.7098,
      "step": 8435
    },
    {
      "epoch": 4.528841314049864,
      "grad_norm": 0.68359375,
      "learning_rate": 1.0817071972915592e-06,
      "loss": 0.7504,
      "step": 8436
    },
    {
      "epoch": 4.52937820878494,
      "grad_norm": 0.734375,
      "learning_rate": 1.0792512662760879e-06,
      "loss": 0.9925,
      "step": 8437
    },
    {
      "epoch": 4.5299151035200165,
      "grad_norm": 0.671875,
      "learning_rate": 1.076798064920978e-06,
      "loss": 0.9186,
      "step": 8438
    },
    {
      "epoch": 4.530451998255092,
      "grad_norm": 0.75,
      "learning_rate": 1.0743475935061725e-06,
      "loss": 0.8319,
      "step": 8439
    },
    {
      "epoch": 4.530988892990168,
      "grad_norm": 0.6953125,
      "learning_rate": 1.0718998523113004e-06,
      "loss": 1.0014,
      "step": 8440
    },
    {
      "epoch": 4.531525787725244,
      "grad_norm": 0.56640625,
      "learning_rate": 1.0694548416156774e-06,
      "loss": 0.7704,
      "step": 8441
    },
    {
      "epoch": 4.53206268246032,
      "grad_norm": 0.6484375,
      "learning_rate": 1.067012561698319e-06,
      "loss": 0.9283,
      "step": 8442
    },
    {
      "epoch": 4.532599577195396,
      "grad_norm": 0.734375,
      "learning_rate": 1.0645730128379132e-06,
      "loss": 1.0053,
      "step": 8443
    },
    {
      "epoch": 4.533136471930472,
      "grad_norm": 0.67578125,
      "learning_rate": 1.0621361953128407e-06,
      "loss": 0.809,
      "step": 8444
    },
    {
      "epoch": 4.533673366665548,
      "grad_norm": 0.66015625,
      "learning_rate": 1.0597021094011755e-06,
      "loss": 0.7751,
      "step": 8445
    },
    {
      "epoch": 4.534210261400624,
      "grad_norm": 0.7265625,
      "learning_rate": 1.057270755380685e-06,
      "loss": 1.0098,
      "step": 8446
    },
    {
      "epoch": 4.5347471561357,
      "grad_norm": 0.8125,
      "learning_rate": 1.054842133528805e-06,
      "loss": 0.886,
      "step": 8447
    },
    {
      "epoch": 4.535284050870776,
      "grad_norm": 0.65234375,
      "learning_rate": 1.052416244122681e-06,
      "loss": 0.7425,
      "step": 8448
    },
    {
      "epoch": 4.5358209456058525,
      "grad_norm": 0.68359375,
      "learning_rate": 1.0499930874391357e-06,
      "loss": 0.7849,
      "step": 8449
    },
    {
      "epoch": 4.536357840340928,
      "grad_norm": 0.75,
      "learning_rate": 1.047572663754684e-06,
      "loss": 0.8093,
      "step": 8450
    },
    {
      "epoch": 4.536357840340928,
      "eval_loss": 0.9026585817337036,
      "eval_runtime": 46.8877,
      "eval_samples_per_second": 20.517,
      "eval_steps_per_second": 20.517,
      "step": 8450
    },
    {
      "epoch": 4.536894735076004,
      "grad_norm": 0.703125,
      "learning_rate": 1.0451549733455219e-06,
      "loss": 0.8279,
      "step": 8451
    },
    {
      "epoch": 4.53743162981108,
      "grad_norm": 0.7421875,
      "learning_rate": 1.0427400164875423e-06,
      "loss": 0.8743,
      "step": 8452
    },
    {
      "epoch": 4.537968524546156,
      "grad_norm": 0.70703125,
      "learning_rate": 1.0403277934563165e-06,
      "loss": 0.8063,
      "step": 8453
    },
    {
      "epoch": 4.538505419281233,
      "grad_norm": 0.70703125,
      "learning_rate": 1.0379183045271162e-06,
      "loss": 0.7471,
      "step": 8454
    },
    {
      "epoch": 4.539042314016308,
      "grad_norm": 0.6875,
      "learning_rate": 1.0355115499748936e-06,
      "loss": 0.8042,
      "step": 8455
    },
    {
      "epoch": 4.539579208751384,
      "grad_norm": 0.78125,
      "learning_rate": 1.0331075300742898e-06,
      "loss": 0.6991,
      "step": 8456
    },
    {
      "epoch": 4.5401161034864606,
      "grad_norm": 0.62109375,
      "learning_rate": 1.0307062450996302e-06,
      "loss": 0.6688,
      "step": 8457
    },
    {
      "epoch": 4.540652998221536,
      "grad_norm": 0.69140625,
      "learning_rate": 1.0283076953249371e-06,
      "loss": 0.7197,
      "step": 8458
    },
    {
      "epoch": 4.541189892956612,
      "grad_norm": 0.66015625,
      "learning_rate": 1.0259118810239139e-06,
      "loss": 0.9387,
      "step": 8459
    },
    {
      "epoch": 4.5417267876916885,
      "grad_norm": 0.5703125,
      "learning_rate": 1.023518802469947e-06,
      "loss": 0.7205,
      "step": 8460
    },
    {
      "epoch": 4.542263682426764,
      "grad_norm": 0.640625,
      "learning_rate": 1.0211284599361271e-06,
      "loss": 0.7472,
      "step": 8461
    },
    {
      "epoch": 4.54280057716184,
      "grad_norm": 0.6953125,
      "learning_rate": 1.0187408536952158e-06,
      "loss": 0.7683,
      "step": 8462
    },
    {
      "epoch": 4.543337471896916,
      "grad_norm": 0.640625,
      "learning_rate": 1.0163559840196708e-06,
      "loss": 0.7384,
      "step": 8463
    },
    {
      "epoch": 4.543874366631992,
      "grad_norm": 0.6875,
      "learning_rate": 1.0139738511816404e-06,
      "loss": 0.8512,
      "step": 8464
    },
    {
      "epoch": 4.544411261367069,
      "grad_norm": 0.74609375,
      "learning_rate": 1.0115944554529467e-06,
      "loss": 1.043,
      "step": 8465
    },
    {
      "epoch": 4.544948156102144,
      "grad_norm": 0.71484375,
      "learning_rate": 1.0092177971051137e-06,
      "loss": 0.7697,
      "step": 8466
    },
    {
      "epoch": 4.54548505083722,
      "grad_norm": 0.69921875,
      "learning_rate": 1.006843876409355e-06,
      "loss": 0.9354,
      "step": 8467
    },
    {
      "epoch": 4.5460219455722966,
      "grad_norm": 0.6875,
      "learning_rate": 1.0044726936365574e-06,
      "loss": 0.8374,
      "step": 8468
    },
    {
      "epoch": 4.546558840307372,
      "grad_norm": 0.59375,
      "learning_rate": 1.002104249057298e-06,
      "loss": 0.6824,
      "step": 8469
    },
    {
      "epoch": 4.547095735042448,
      "grad_norm": 0.859375,
      "learning_rate": 9.997385429418555e-07,
      "loss": 0.64,
      "step": 8470
    },
    {
      "epoch": 4.5476326297775245,
      "grad_norm": 0.6640625,
      "learning_rate": 9.973755755601888e-07,
      "loss": 0.8754,
      "step": 8471
    },
    {
      "epoch": 4.5481695245126,
      "grad_norm": 0.74609375,
      "learning_rate": 9.950153471819324e-07,
      "loss": 0.9445,
      "step": 8472
    },
    {
      "epoch": 4.548706419247676,
      "grad_norm": 0.66796875,
      "learning_rate": 9.926578580764234e-07,
      "loss": 0.8091,
      "step": 8473
    },
    {
      "epoch": 4.549243313982752,
      "grad_norm": 0.75390625,
      "learning_rate": 9.90303108512683e-07,
      "loss": 0.8357,
      "step": 8474
    },
    {
      "epoch": 4.549780208717828,
      "grad_norm": 0.6953125,
      "learning_rate": 9.879510987594182e-07,
      "loss": 0.7445,
      "step": 8475
    },
    {
      "epoch": 4.550317103452905,
      "grad_norm": 0.7578125,
      "learning_rate": 9.856018290850172e-07,
      "loss": 0.791,
      "step": 8476
    },
    {
      "epoch": 4.55085399818798,
      "grad_norm": 0.75390625,
      "learning_rate": 9.832552997575684e-07,
      "loss": 0.9048,
      "step": 8477
    },
    {
      "epoch": 4.551390892923056,
      "grad_norm": 0.6328125,
      "learning_rate": 9.809115110448325e-07,
      "loss": 0.7591,
      "step": 8478
    },
    {
      "epoch": 4.5519277876581326,
      "grad_norm": 0.7421875,
      "learning_rate": 9.785704632142734e-07,
      "loss": 0.8141,
      "step": 8479
    },
    {
      "epoch": 4.552464682393208,
      "grad_norm": 0.66015625,
      "learning_rate": 9.76232156533033e-07,
      "loss": 1.0441,
      "step": 8480
    },
    {
      "epoch": 4.553001577128284,
      "grad_norm": 0.83984375,
      "learning_rate": 9.738965912679366e-07,
      "loss": 0.8971,
      "step": 8481
    },
    {
      "epoch": 4.5535384718633605,
      "grad_norm": 0.78515625,
      "learning_rate": 9.715637676855043e-07,
      "loss": 0.7573,
      "step": 8482
    },
    {
      "epoch": 4.554075366598436,
      "grad_norm": 0.69921875,
      "learning_rate": 9.692336860519458e-07,
      "loss": 0.7689,
      "step": 8483
    },
    {
      "epoch": 4.554612261333512,
      "grad_norm": 0.6796875,
      "learning_rate": 9.66906346633148e-07,
      "loss": 0.8638,
      "step": 8484
    },
    {
      "epoch": 4.555149156068588,
      "grad_norm": 0.63671875,
      "learning_rate": 9.645817496946903e-07,
      "loss": 0.9896,
      "step": 8485
    },
    {
      "epoch": 4.555686050803664,
      "grad_norm": 0.69921875,
      "learning_rate": 9.622598955018352e-07,
      "loss": 0.8124,
      "step": 8486
    },
    {
      "epoch": 4.556222945538741,
      "grad_norm": 0.59765625,
      "learning_rate": 9.599407843195434e-07,
      "loss": 0.7511,
      "step": 8487
    },
    {
      "epoch": 4.556759840273816,
      "grad_norm": 0.62109375,
      "learning_rate": 9.576244164124504e-07,
      "loss": 0.6803,
      "step": 8488
    },
    {
      "epoch": 4.557296735008892,
      "grad_norm": 0.66796875,
      "learning_rate": 9.553107920448806e-07,
      "loss": 0.7879,
      "step": 8489
    },
    {
      "epoch": 4.5578336297439686,
      "grad_norm": 0.6015625,
      "learning_rate": 9.529999114808535e-07,
      "loss": 0.7233,
      "step": 8490
    },
    {
      "epoch": 4.558370524479044,
      "grad_norm": 0.63671875,
      "learning_rate": 9.506917749840665e-07,
      "loss": 0.7202,
      "step": 8491
    },
    {
      "epoch": 4.558907419214121,
      "grad_norm": 0.67578125,
      "learning_rate": 9.48386382817909e-07,
      "loss": 0.8645,
      "step": 8492
    },
    {
      "epoch": 4.5594443139491965,
      "grad_norm": 0.6953125,
      "learning_rate": 9.460837352454566e-07,
      "loss": 0.735,
      "step": 8493
    },
    {
      "epoch": 4.559981208684272,
      "grad_norm": 0.73046875,
      "learning_rate": 9.437838325294629e-07,
      "loss": 0.8704,
      "step": 8494
    },
    {
      "epoch": 4.560518103419349,
      "grad_norm": 0.68359375,
      "learning_rate": 9.414866749323819e-07,
      "loss": 0.9574,
      "step": 8495
    },
    {
      "epoch": 4.561054998154424,
      "grad_norm": 0.7578125,
      "learning_rate": 9.391922627163513e-07,
      "loss": 0.8052,
      "step": 8496
    },
    {
      "epoch": 4.5615918928895,
      "grad_norm": 0.7890625,
      "learning_rate": 9.369005961431837e-07,
      "loss": 0.8902,
      "step": 8497
    },
    {
      "epoch": 4.562128787624577,
      "grad_norm": 0.734375,
      "learning_rate": 9.34611675474395e-07,
      "loss": 0.8239,
      "step": 8498
    },
    {
      "epoch": 4.562665682359652,
      "grad_norm": 0.60546875,
      "learning_rate": 9.32325500971179e-07,
      "loss": 0.7137,
      "step": 8499
    },
    {
      "epoch": 4.563202577094728,
      "grad_norm": 0.71484375,
      "learning_rate": 9.300420728944131e-07,
      "loss": 0.815,
      "step": 8500
    },
    {
      "epoch": 4.563202577094728,
      "eval_loss": 0.902619481086731,
      "eval_runtime": 45.7874,
      "eval_samples_per_second": 21.01,
      "eval_steps_per_second": 21.01,
      "step": 8500
    },
    {
      "epoch": 4.5637394718298046,
      "grad_norm": 1.0859375,
      "learning_rate": 9.277613915046723e-07,
      "loss": 0.7932,
      "step": 8501
    },
    {
      "epoch": 4.56427636656488,
      "grad_norm": 0.6953125,
      "learning_rate": 9.25483457062204e-07,
      "loss": 0.9043,
      "step": 8502
    },
    {
      "epoch": 4.564813261299957,
      "grad_norm": 0.6640625,
      "learning_rate": 9.232082698269557e-07,
      "loss": 0.6617,
      "step": 8503
    },
    {
      "epoch": 4.5653501560350325,
      "grad_norm": 0.75,
      "learning_rate": 9.209358300585474e-07,
      "loss": 0.7606,
      "step": 8504
    },
    {
      "epoch": 4.565887050770108,
      "grad_norm": 0.6875,
      "learning_rate": 9.18666138016297e-07,
      "loss": 0.7473,
      "step": 8505
    },
    {
      "epoch": 4.566423945505185,
      "grad_norm": 0.58203125,
      "learning_rate": 9.163991939592081e-07,
      "loss": 0.7939,
      "step": 8506
    },
    {
      "epoch": 4.56696084024026,
      "grad_norm": 0.65625,
      "learning_rate": 9.141349981459657e-07,
      "loss": 0.8223,
      "step": 8507
    },
    {
      "epoch": 4.567497734975336,
      "grad_norm": 0.73828125,
      "learning_rate": 9.118735508349435e-07,
      "loss": 0.9082,
      "step": 8508
    },
    {
      "epoch": 4.568034629710413,
      "grad_norm": 0.71875,
      "learning_rate": 9.096148522841991e-07,
      "loss": 0.817,
      "step": 8509
    },
    {
      "epoch": 4.568571524445488,
      "grad_norm": 0.6328125,
      "learning_rate": 9.073589027514789e-07,
      "loss": 0.8953,
      "step": 8510
    },
    {
      "epoch": 4.569108419180564,
      "grad_norm": 0.69921875,
      "learning_rate": 9.051057024942189e-07,
      "loss": 0.8384,
      "step": 8511
    },
    {
      "epoch": 4.5696453139156405,
      "grad_norm": 0.73828125,
      "learning_rate": 9.028552517695354e-07,
      "loss": 0.7219,
      "step": 8512
    },
    {
      "epoch": 4.570182208650716,
      "grad_norm": 0.63671875,
      "learning_rate": 9.006075508342287e-07,
      "loss": 0.7437,
      "step": 8513
    },
    {
      "epoch": 4.570719103385793,
      "grad_norm": 0.66015625,
      "learning_rate": 8.983625999447992e-07,
      "loss": 0.6465,
      "step": 8514
    },
    {
      "epoch": 4.5712559981208685,
      "grad_norm": 0.84765625,
      "learning_rate": 8.961203993574196e-07,
      "loss": 0.9275,
      "step": 8515
    },
    {
      "epoch": 4.571792892855944,
      "grad_norm": 0.79296875,
      "learning_rate": 8.938809493279493e-07,
      "loss": 0.8312,
      "step": 8516
    },
    {
      "epoch": 4.572329787591021,
      "grad_norm": 0.62109375,
      "learning_rate": 8.916442501119477e-07,
      "loss": 0.8756,
      "step": 8517
    },
    {
      "epoch": 4.572866682326096,
      "grad_norm": 0.74609375,
      "learning_rate": 8.894103019646438e-07,
      "loss": 0.888,
      "step": 8518
    },
    {
      "epoch": 4.573403577061172,
      "grad_norm": 0.63671875,
      "learning_rate": 8.871791051409562e-07,
      "loss": 0.6117,
      "step": 8519
    },
    {
      "epoch": 4.573940471796249,
      "grad_norm": 0.671875,
      "learning_rate": 8.849506598954977e-07,
      "loss": 0.8381,
      "step": 8520
    },
    {
      "epoch": 4.574477366531324,
      "grad_norm": 0.74609375,
      "learning_rate": 8.827249664825649e-07,
      "loss": 0.8809,
      "step": 8521
    },
    {
      "epoch": 4.5750142612664,
      "grad_norm": 0.671875,
      "learning_rate": 8.805020251561324e-07,
      "loss": 0.9112,
      "step": 8522
    },
    {
      "epoch": 4.5755511560014765,
      "grad_norm": 0.5625,
      "learning_rate": 8.782818361698642e-07,
      "loss": 0.6419,
      "step": 8523
    },
    {
      "epoch": 4.576088050736552,
      "grad_norm": 0.6953125,
      "learning_rate": 8.760643997771212e-07,
      "loss": 0.7746,
      "step": 8524
    },
    {
      "epoch": 4.576624945471629,
      "grad_norm": 0.60546875,
      "learning_rate": 8.73849716230929e-07,
      "loss": 0.8607,
      "step": 8525
    },
    {
      "epoch": 4.5771618402067045,
      "grad_norm": 0.71484375,
      "learning_rate": 8.716377857840241e-07,
      "loss": 0.8451,
      "step": 8526
    },
    {
      "epoch": 4.57769873494178,
      "grad_norm": 0.55078125,
      "learning_rate": 8.69428608688802e-07,
      "loss": 0.6915,
      "step": 8527
    },
    {
      "epoch": 4.578235629676857,
      "grad_norm": 0.84765625,
      "learning_rate": 8.672221851973694e-07,
      "loss": 0.8643,
      "step": 8528
    },
    {
      "epoch": 4.578772524411932,
      "grad_norm": 0.73046875,
      "learning_rate": 8.650185155614998e-07,
      "loss": 0.8649,
      "step": 8529
    },
    {
      "epoch": 4.579309419147009,
      "grad_norm": 0.7890625,
      "learning_rate": 8.628176000326615e-07,
      "loss": 0.7384,
      "step": 8530
    },
    {
      "epoch": 4.579846313882085,
      "grad_norm": 0.703125,
      "learning_rate": 8.60619438862012e-07,
      "loss": 0.866,
      "step": 8531
    },
    {
      "epoch": 4.58038320861716,
      "grad_norm": 0.65625,
      "learning_rate": 8.584240323003811e-07,
      "loss": 0.8837,
      "step": 8532
    },
    {
      "epoch": 4.580920103352237,
      "grad_norm": 0.6796875,
      "learning_rate": 8.562313805982963e-07,
      "loss": 0.7141,
      "step": 8533
    },
    {
      "epoch": 4.5814569980873125,
      "grad_norm": 0.71875,
      "learning_rate": 8.54041484005974e-07,
      "loss": 0.9274,
      "step": 8534
    },
    {
      "epoch": 4.581993892822388,
      "grad_norm": 0.59765625,
      "learning_rate": 8.51854342773295e-07,
      "loss": 0.7147,
      "step": 8535
    },
    {
      "epoch": 4.582530787557465,
      "grad_norm": 0.65625,
      "learning_rate": 8.496699571498485e-07,
      "loss": 0.8153,
      "step": 8536
    },
    {
      "epoch": 4.5830676822925405,
      "grad_norm": 0.85546875,
      "learning_rate": 8.474883273848988e-07,
      "loss": 0.7827,
      "step": 8537
    },
    {
      "epoch": 4.583604577027616,
      "grad_norm": 0.58203125,
      "learning_rate": 8.45309453727397e-07,
      "loss": 0.8565,
      "step": 8538
    },
    {
      "epoch": 4.584141471762693,
      "grad_norm": 0.71875,
      "learning_rate": 8.431333364259831e-07,
      "loss": 1.0511,
      "step": 8539
    },
    {
      "epoch": 4.584678366497768,
      "grad_norm": 0.83984375,
      "learning_rate": 8.409599757289777e-07,
      "loss": 0.9118,
      "step": 8540
    },
    {
      "epoch": 4.585215261232845,
      "grad_norm": 0.63671875,
      "learning_rate": 8.387893718843853e-07,
      "loss": 0.7262,
      "step": 8541
    },
    {
      "epoch": 4.585752155967921,
      "grad_norm": 0.72265625,
      "learning_rate": 8.366215251399078e-07,
      "loss": 0.8432,
      "step": 8542
    },
    {
      "epoch": 4.586289050702996,
      "grad_norm": 0.6875,
      "learning_rate": 8.344564357429141e-07,
      "loss": 0.8498,
      "step": 8543
    },
    {
      "epoch": 4.586825945438073,
      "grad_norm": 0.6875,
      "learning_rate": 8.322941039404758e-07,
      "loss": 0.8212,
      "step": 8544
    },
    {
      "epoch": 4.5873628401731485,
      "grad_norm": 0.6640625,
      "learning_rate": 8.301345299793373e-07,
      "loss": 0.774,
      "step": 8545
    },
    {
      "epoch": 4.587899734908224,
      "grad_norm": 0.6015625,
      "learning_rate": 8.279777141059347e-07,
      "loss": 0.8208,
      "step": 8546
    },
    {
      "epoch": 4.588436629643301,
      "grad_norm": 0.73046875,
      "learning_rate": 8.258236565663908e-07,
      "loss": 0.7831,
      "step": 8547
    },
    {
      "epoch": 4.5889735243783765,
      "grad_norm": 0.71484375,
      "learning_rate": 8.236723576065087e-07,
      "loss": 1.0149,
      "step": 8548
    },
    {
      "epoch": 4.589510419113452,
      "grad_norm": 0.796875,
      "learning_rate": 8.215238174717809e-07,
      "loss": 0.695,
      "step": 8549
    },
    {
      "epoch": 4.590047313848529,
      "grad_norm": 0.5859375,
      "learning_rate": 8.193780364073755e-07,
      "loss": 0.7278,
      "step": 8550
    },
    {
      "epoch": 4.590047313848529,
      "eval_loss": 0.9027642011642456,
      "eval_runtime": 46.3462,
      "eval_samples_per_second": 20.757,
      "eval_steps_per_second": 20.757,
      "step": 8550
    },
    {
      "epoch": 4.590584208583604,
      "grad_norm": 0.6875,
      "learning_rate": 8.172350146581658e-07,
      "loss": 0.8991,
      "step": 8551
    },
    {
      "epoch": 4.591121103318681,
      "grad_norm": 0.5625,
      "learning_rate": 8.150947524686869e-07,
      "loss": 0.7414,
      "step": 8552
    },
    {
      "epoch": 4.591657998053757,
      "grad_norm": 0.71875,
      "learning_rate": 8.129572500831739e-07,
      "loss": 0.775,
      "step": 8553
    },
    {
      "epoch": 4.592194892788832,
      "grad_norm": 0.65234375,
      "learning_rate": 8.108225077455428e-07,
      "loss": 0.7875,
      "step": 8554
    },
    {
      "epoch": 4.592731787523909,
      "grad_norm": 0.68359375,
      "learning_rate": 8.08690525699396e-07,
      "loss": 0.8853,
      "step": 8555
    },
    {
      "epoch": 4.5932686822589845,
      "grad_norm": 0.66015625,
      "learning_rate": 8.065613041880194e-07,
      "loss": 0.9799,
      "step": 8556
    },
    {
      "epoch": 4.59380557699406,
      "grad_norm": 0.890625,
      "learning_rate": 8.044348434543797e-07,
      "loss": 1.0202,
      "step": 8557
    },
    {
      "epoch": 4.594342471729137,
      "grad_norm": 0.69140625,
      "learning_rate": 8.023111437411357e-07,
      "loss": 1.0828,
      "step": 8558
    },
    {
      "epoch": 4.5948793664642125,
      "grad_norm": 0.65625,
      "learning_rate": 8.001902052906351e-07,
      "loss": 1.018,
      "step": 8559
    },
    {
      "epoch": 4.595416261199288,
      "grad_norm": 0.859375,
      "learning_rate": 7.980720283448956e-07,
      "loss": 0.8757,
      "step": 8560
    },
    {
      "epoch": 4.595953155934365,
      "grad_norm": 0.72265625,
      "learning_rate": 7.959566131456264e-07,
      "loss": 0.772,
      "step": 8561
    },
    {
      "epoch": 4.59649005066944,
      "grad_norm": 0.70703125,
      "learning_rate": 7.938439599342317e-07,
      "loss": 0.9219,
      "step": 8562
    },
    {
      "epoch": 4.597026945404517,
      "grad_norm": 0.75,
      "learning_rate": 7.917340689517828e-07,
      "loss": 1.0212,
      "step": 8563
    },
    {
      "epoch": 4.597563840139593,
      "grad_norm": 0.7421875,
      "learning_rate": 7.896269404390505e-07,
      "loss": 1.0342,
      "step": 8564
    },
    {
      "epoch": 4.598100734874668,
      "grad_norm": 0.7421875,
      "learning_rate": 7.875225746364873e-07,
      "loss": 1.0311,
      "step": 8565
    },
    {
      "epoch": 4.598637629609745,
      "grad_norm": 0.65234375,
      "learning_rate": 7.854209717842231e-07,
      "loss": 0.8156,
      "step": 8566
    },
    {
      "epoch": 4.5991745243448205,
      "grad_norm": 0.6875,
      "learning_rate": 7.833221321220774e-07,
      "loss": 0.7297,
      "step": 8567
    },
    {
      "epoch": 4.599711419079897,
      "grad_norm": 0.6796875,
      "learning_rate": 7.812260558895584e-07,
      "loss": 0.8428,
      "step": 8568
    },
    {
      "epoch": 4.600248313814973,
      "grad_norm": 0.671875,
      "learning_rate": 7.791327433258499e-07,
      "loss": 0.7613,
      "step": 8569
    },
    {
      "epoch": 4.6007852085500485,
      "grad_norm": 0.67578125,
      "learning_rate": 7.770421946698275e-07,
      "loss": 0.817,
      "step": 8570
    },
    {
      "epoch": 4.601322103285125,
      "grad_norm": 0.7265625,
      "learning_rate": 7.749544101600504e-07,
      "loss": 1.1847,
      "step": 8571
    },
    {
      "epoch": 4.601858998020201,
      "grad_norm": 0.5859375,
      "learning_rate": 7.728693900347639e-07,
      "loss": 0.6922,
      "step": 8572
    },
    {
      "epoch": 4.602395892755276,
      "grad_norm": 0.60546875,
      "learning_rate": 7.707871345318862e-07,
      "loss": 0.6023,
      "step": 8573
    },
    {
      "epoch": 4.602932787490353,
      "grad_norm": 0.765625,
      "learning_rate": 7.687076438890356e-07,
      "loss": 0.8528,
      "step": 8574
    },
    {
      "epoch": 4.603469682225429,
      "grad_norm": 0.78125,
      "learning_rate": 7.666309183435111e-07,
      "loss": 0.9262,
      "step": 8575
    },
    {
      "epoch": 4.604006576960504,
      "grad_norm": 0.7109375,
      "learning_rate": 7.645569581322898e-07,
      "loss": 1.0498,
      "step": 8576
    },
    {
      "epoch": 4.604543471695581,
      "grad_norm": 0.703125,
      "learning_rate": 7.624857634920324e-07,
      "loss": 0.8385,
      "step": 8577
    },
    {
      "epoch": 4.6050803664306565,
      "grad_norm": 0.83203125,
      "learning_rate": 7.60417334659097e-07,
      "loss": 1.0223,
      "step": 8578
    },
    {
      "epoch": 4.605617261165733,
      "grad_norm": 0.609375,
      "learning_rate": 7.583516718695088e-07,
      "loss": 0.6642,
      "step": 8579
    },
    {
      "epoch": 4.606154155900809,
      "grad_norm": 0.72265625,
      "learning_rate": 7.562887753589904e-07,
      "loss": 0.9018,
      "step": 8580
    },
    {
      "epoch": 4.6066910506358845,
      "grad_norm": 0.62109375,
      "learning_rate": 7.542286453629505e-07,
      "loss": 0.8002,
      "step": 8581
    },
    {
      "epoch": 4.607227945370961,
      "grad_norm": 0.7109375,
      "learning_rate": 7.52171282116465e-07,
      "loss": 0.9679,
      "step": 8582
    },
    {
      "epoch": 4.607764840106037,
      "grad_norm": 0.703125,
      "learning_rate": 7.501166858543101e-07,
      "loss": 0.9165,
      "step": 8583
    },
    {
      "epoch": 4.608301734841112,
      "grad_norm": 0.56640625,
      "learning_rate": 7.480648568109428e-07,
      "loss": 0.7012,
      "step": 8584
    },
    {
      "epoch": 4.608838629576189,
      "grad_norm": 0.6171875,
      "learning_rate": 7.460157952205033e-07,
      "loss": 0.7311,
      "step": 8585
    },
    {
      "epoch": 4.609375524311265,
      "grad_norm": 0.58984375,
      "learning_rate": 7.439695013168075e-07,
      "loss": 0.7021,
      "step": 8586
    },
    {
      "epoch": 4.60991241904634,
      "grad_norm": 0.7578125,
      "learning_rate": 7.41925975333374e-07,
      "loss": 0.9652,
      "step": 8587
    },
    {
      "epoch": 4.610449313781417,
      "grad_norm": 0.66015625,
      "learning_rate": 7.398852175033888e-07,
      "loss": 0.7665,
      "step": 8588
    },
    {
      "epoch": 4.6109862085164925,
      "grad_norm": 0.6953125,
      "learning_rate": 7.378472280597292e-07,
      "loss": 0.7301,
      "step": 8589
    },
    {
      "epoch": 4.611523103251569,
      "grad_norm": 0.6484375,
      "learning_rate": 7.358120072349595e-07,
      "loss": 0.814,
      "step": 8590
    },
    {
      "epoch": 4.612059997986645,
      "grad_norm": 0.65234375,
      "learning_rate": 7.337795552613158e-07,
      "loss": 0.7413,
      "step": 8591
    },
    {
      "epoch": 4.6125968927217205,
      "grad_norm": 0.74609375,
      "learning_rate": 7.31749872370735e-07,
      "loss": 0.8707,
      "step": 8592
    },
    {
      "epoch": 4.613133787456797,
      "grad_norm": 0.66015625,
      "learning_rate": 7.297229587948262e-07,
      "loss": 0.7414,
      "step": 8593
    },
    {
      "epoch": 4.613670682191873,
      "grad_norm": 0.703125,
      "learning_rate": 7.276988147648878e-07,
      "loss": 0.8463,
      "step": 8594
    },
    {
      "epoch": 4.614207576926948,
      "grad_norm": 0.68359375,
      "learning_rate": 7.256774405118932e-07,
      "loss": 0.7516,
      "step": 8595
    },
    {
      "epoch": 4.614744471662025,
      "grad_norm": 0.6328125,
      "learning_rate": 7.236588362665136e-07,
      "loss": 0.8484,
      "step": 8596
    },
    {
      "epoch": 4.615281366397101,
      "grad_norm": 0.765625,
      "learning_rate": 7.216430022591008e-07,
      "loss": 0.8635,
      "step": 8597
    },
    {
      "epoch": 4.615818261132176,
      "grad_norm": 0.734375,
      "learning_rate": 7.196299387196764e-07,
      "loss": 0.7139,
      "step": 8598
    },
    {
      "epoch": 4.616355155867253,
      "grad_norm": 0.62109375,
      "learning_rate": 7.176196458779622e-07,
      "loss": 0.6986,
      "step": 8599
    },
    {
      "epoch": 4.6168920506023285,
      "grad_norm": 0.61328125,
      "learning_rate": 7.156121239633607e-07,
      "loss": 0.7515,
      "step": 8600
    },
    {
      "epoch": 4.6168920506023285,
      "eval_loss": 0.9026636481285095,
      "eval_runtime": 46.4391,
      "eval_samples_per_second": 20.715,
      "eval_steps_per_second": 20.715,
      "step": 8600
    },
    {
      "epoch": 4.617428945337405,
      "grad_norm": 0.6875,
      "learning_rate": 7.136073732049553e-07,
      "loss": 0.6794,
      "step": 8601
    },
    {
      "epoch": 4.617965840072481,
      "grad_norm": 0.78515625,
      "learning_rate": 7.116053938315048e-07,
      "loss": 1.0429,
      "step": 8602
    },
    {
      "epoch": 4.6185027348075565,
      "grad_norm": 0.73046875,
      "learning_rate": 7.096061860714709e-07,
      "loss": 0.8808,
      "step": 8603
    },
    {
      "epoch": 4.619039629542633,
      "grad_norm": 0.73046875,
      "learning_rate": 7.076097501529794e-07,
      "loss": 0.9149,
      "step": 8604
    },
    {
      "epoch": 4.619576524277709,
      "grad_norm": 0.6953125,
      "learning_rate": 7.056160863038563e-07,
      "loss": 0.8187,
      "step": 8605
    },
    {
      "epoch": 4.620113419012785,
      "grad_norm": 0.7578125,
      "learning_rate": 7.036251947516032e-07,
      "loss": 0.7818,
      "step": 8606
    },
    {
      "epoch": 4.620650313747861,
      "grad_norm": 0.9453125,
      "learning_rate": 7.016370757233992e-07,
      "loss": 0.7576,
      "step": 8607
    },
    {
      "epoch": 4.621187208482937,
      "grad_norm": 0.671875,
      "learning_rate": 6.996517294461213e-07,
      "loss": 0.9365,
      "step": 8608
    },
    {
      "epoch": 4.621724103218013,
      "grad_norm": 0.6015625,
      "learning_rate": 6.976691561463217e-07,
      "loss": 0.7382,
      "step": 8609
    },
    {
      "epoch": 4.622260997953089,
      "grad_norm": 0.6328125,
      "learning_rate": 6.956893560502359e-07,
      "loss": 0.7634,
      "step": 8610
    },
    {
      "epoch": 4.6227978926881645,
      "grad_norm": 0.6015625,
      "learning_rate": 6.937123293837805e-07,
      "loss": 0.8106,
      "step": 8611
    },
    {
      "epoch": 4.623334787423241,
      "grad_norm": 0.76953125,
      "learning_rate": 6.91738076372564e-07,
      "loss": 0.7068,
      "step": 8612
    },
    {
      "epoch": 4.623871682158317,
      "grad_norm": 0.6875,
      "learning_rate": 6.897665972418755e-07,
      "loss": 0.7453,
      "step": 8613
    },
    {
      "epoch": 4.6244085768933925,
      "grad_norm": 0.65234375,
      "learning_rate": 6.877978922166767e-07,
      "loss": 0.7838,
      "step": 8614
    },
    {
      "epoch": 4.624945471628469,
      "grad_norm": 0.72265625,
      "learning_rate": 6.858319615216296e-07,
      "loss": 0.8509,
      "step": 8615
    },
    {
      "epoch": 4.625482366363545,
      "grad_norm": 0.73828125,
      "learning_rate": 6.838688053810716e-07,
      "loss": 0.8256,
      "step": 8616
    },
    {
      "epoch": 4.626019261098621,
      "grad_norm": 0.703125,
      "learning_rate": 6.819084240190205e-07,
      "loss": 0.7821,
      "step": 8617
    },
    {
      "epoch": 4.626556155833697,
      "grad_norm": 0.66796875,
      "learning_rate": 6.799508176591834e-07,
      "loss": 0.9111,
      "step": 8618
    },
    {
      "epoch": 4.627093050568773,
      "grad_norm": 0.65234375,
      "learning_rate": 6.779959865249458e-07,
      "loss": 0.7661,
      "step": 8619
    },
    {
      "epoch": 4.627629945303849,
      "grad_norm": 0.6953125,
      "learning_rate": 6.760439308393763e-07,
      "loss": 0.6803,
      "step": 8620
    },
    {
      "epoch": 4.628166840038925,
      "grad_norm": 0.75,
      "learning_rate": 6.740946508252355e-07,
      "loss": 0.9843,
      "step": 8621
    },
    {
      "epoch": 4.6287037347740005,
      "grad_norm": 0.6015625,
      "learning_rate": 6.721481467049568e-07,
      "loss": 0.7459,
      "step": 8622
    },
    {
      "epoch": 4.629240629509077,
      "grad_norm": 0.8515625,
      "learning_rate": 6.702044187006623e-07,
      "loss": 0.7871,
      "step": 8623
    },
    {
      "epoch": 4.629777524244153,
      "grad_norm": 0.71484375,
      "learning_rate": 6.682634670341525e-07,
      "loss": 0.7931,
      "step": 8624
    },
    {
      "epoch": 4.6303144189792285,
      "grad_norm": 0.828125,
      "learning_rate": 6.663252919269225e-07,
      "loss": 0.9139,
      "step": 8625
    },
    {
      "epoch": 4.630851313714305,
      "grad_norm": 0.61328125,
      "learning_rate": 6.643898936001369e-07,
      "loss": 0.8128,
      "step": 8626
    },
    {
      "epoch": 4.631388208449381,
      "grad_norm": 0.6953125,
      "learning_rate": 6.62457272274647e-07,
      "loss": 0.8223,
      "step": 8627
    },
    {
      "epoch": 4.631925103184457,
      "grad_norm": 0.69140625,
      "learning_rate": 6.605274281709928e-07,
      "loss": 0.9057,
      "step": 8628
    },
    {
      "epoch": 4.632461997919533,
      "grad_norm": 0.6328125,
      "learning_rate": 6.586003615093955e-07,
      "loss": 0.8061,
      "step": 8629
    },
    {
      "epoch": 4.632998892654609,
      "grad_norm": 0.9453125,
      "learning_rate": 6.566760725097543e-07,
      "loss": 0.9027,
      "step": 8630
    },
    {
      "epoch": 4.633535787389685,
      "grad_norm": 0.94140625,
      "learning_rate": 6.547545613916572e-07,
      "loss": 0.6901,
      "step": 8631
    },
    {
      "epoch": 4.634072682124761,
      "grad_norm": 0.71484375,
      "learning_rate": 6.528358283743735e-07,
      "loss": 0.9554,
      "step": 8632
    },
    {
      "epoch": 4.6346095768598365,
      "grad_norm": 0.64453125,
      "learning_rate": 6.509198736768529e-07,
      "loss": 0.7403,
      "step": 8633
    },
    {
      "epoch": 4.635146471594913,
      "grad_norm": 0.625,
      "learning_rate": 6.490066975177317e-07,
      "loss": 0.7575,
      "step": 8634
    },
    {
      "epoch": 4.635683366329989,
      "grad_norm": 0.62109375,
      "learning_rate": 6.470963001153269e-07,
      "loss": 0.8484,
      "step": 8635
    },
    {
      "epoch": 4.6362202610650645,
      "grad_norm": 0.625,
      "learning_rate": 6.451886816876363e-07,
      "loss": 0.6832,
      "step": 8636
    },
    {
      "epoch": 4.636757155800141,
      "grad_norm": 0.734375,
      "learning_rate": 6.432838424523441e-07,
      "loss": 0.9626,
      "step": 8637
    },
    {
      "epoch": 4.637294050535217,
      "grad_norm": 0.68359375,
      "learning_rate": 6.413817826268236e-07,
      "loss": 0.8905,
      "step": 8638
    },
    {
      "epoch": 4.637830945270293,
      "grad_norm": 0.70703125,
      "learning_rate": 6.394825024281154e-07,
      "loss": 0.7792,
      "step": 8639
    },
    {
      "epoch": 4.638367840005369,
      "grad_norm": 0.7578125,
      "learning_rate": 6.375860020729541e-07,
      "loss": 0.9051,
      "step": 8640
    },
    {
      "epoch": 4.638904734740445,
      "grad_norm": 0.70703125,
      "learning_rate": 6.356922817777583e-07,
      "loss": 0.7692,
      "step": 8641
    },
    {
      "epoch": 4.639441629475521,
      "grad_norm": 0.83203125,
      "learning_rate": 6.338013417586192e-07,
      "loss": 0.8862,
      "step": 8642
    },
    {
      "epoch": 4.639978524210597,
      "grad_norm": 0.62109375,
      "learning_rate": 6.319131822313224e-07,
      "loss": 0.7468,
      "step": 8643
    },
    {
      "epoch": 4.640515418945673,
      "grad_norm": 0.63671875,
      "learning_rate": 6.30027803411326e-07,
      "loss": 0.8482,
      "step": 8644
    },
    {
      "epoch": 4.641052313680749,
      "grad_norm": 0.8515625,
      "learning_rate": 6.281452055137804e-07,
      "loss": 1.0957,
      "step": 8645
    },
    {
      "epoch": 4.641589208415825,
      "grad_norm": 0.59765625,
      "learning_rate": 6.262653887535081e-07,
      "loss": 0.7458,
      "step": 8646
    },
    {
      "epoch": 4.642126103150901,
      "grad_norm": 0.7421875,
      "learning_rate": 6.243883533450262e-07,
      "loss": 0.7512,
      "step": 8647
    },
    {
      "epoch": 4.642662997885977,
      "grad_norm": 0.77734375,
      "learning_rate": 6.22514099502522e-07,
      "loss": 0.9586,
      "step": 8648
    },
    {
      "epoch": 4.643199892621053,
      "grad_norm": 0.68359375,
      "learning_rate": 6.206426274398769e-07,
      "loss": 0.8132,
      "step": 8649
    },
    {
      "epoch": 4.643736787356129,
      "grad_norm": 0.6875,
      "learning_rate": 6.187739373706508e-07,
      "loss": 0.9041,
      "step": 8650
    },
    {
      "epoch": 4.643736787356129,
      "eval_loss": 0.9026292562484741,
      "eval_runtime": 46.1503,
      "eval_samples_per_second": 20.845,
      "eval_steps_per_second": 20.845,
      "step": 8650
    },
    {
      "epoch": 4.644273682091205,
      "grad_norm": 0.71875,
      "learning_rate": 6.16908029508076e-07,
      "loss": 0.6977,
      "step": 8651
    },
    {
      "epoch": 4.644810576826281,
      "grad_norm": 0.63671875,
      "learning_rate": 6.150449040650874e-07,
      "loss": 0.7941,
      "step": 8652
    },
    {
      "epoch": 4.645347471561357,
      "grad_norm": 0.6953125,
      "learning_rate": 6.13184561254282e-07,
      "loss": 0.7809,
      "step": 8653
    },
    {
      "epoch": 4.645884366296433,
      "grad_norm": 0.6328125,
      "learning_rate": 6.113270012879535e-07,
      "loss": 0.725,
      "step": 8654
    },
    {
      "epoch": 4.646421261031509,
      "grad_norm": 0.68359375,
      "learning_rate": 6.094722243780714e-07,
      "loss": 0.8718,
      "step": 8655
    },
    {
      "epoch": 4.646958155766585,
      "grad_norm": 0.7421875,
      "learning_rate": 6.076202307362888e-07,
      "loss": 0.7824,
      "step": 8656
    },
    {
      "epoch": 4.647495050501661,
      "grad_norm": 0.703125,
      "learning_rate": 6.05771020573942e-07,
      "loss": 0.8179,
      "step": 8657
    },
    {
      "epoch": 4.648031945236737,
      "grad_norm": 0.90234375,
      "learning_rate": 6.039245941020511e-07,
      "loss": 0.7794,
      "step": 8658
    },
    {
      "epoch": 4.648568839971813,
      "grad_norm": 0.63671875,
      "learning_rate": 6.020809515313142e-07,
      "loss": 0.9527,
      "step": 8659
    },
    {
      "epoch": 4.649105734706889,
      "grad_norm": 0.609375,
      "learning_rate": 6.002400930721186e-07,
      "loss": 0.671,
      "step": 8660
    },
    {
      "epoch": 4.649642629441965,
      "grad_norm": 0.703125,
      "learning_rate": 5.984020189345241e-07,
      "loss": 0.8642,
      "step": 8661
    },
    {
      "epoch": 4.650179524177041,
      "grad_norm": 0.625,
      "learning_rate": 5.965667293282795e-07,
      "loss": 0.8086,
      "step": 8662
    },
    {
      "epoch": 4.650716418912117,
      "grad_norm": 0.5625,
      "learning_rate": 5.947342244628174e-07,
      "loss": 0.7407,
      "step": 8663
    },
    {
      "epoch": 4.651253313647193,
      "grad_norm": 0.671875,
      "learning_rate": 5.929045045472453e-07,
      "loss": 0.865,
      "step": 8664
    },
    {
      "epoch": 4.651790208382269,
      "grad_norm": 0.73828125,
      "learning_rate": 5.910775697903603e-07,
      "loss": 0.7146,
      "step": 8665
    },
    {
      "epoch": 4.652327103117345,
      "grad_norm": 0.71875,
      "learning_rate": 5.892534204006428e-07,
      "loss": 0.9263,
      "step": 8666
    },
    {
      "epoch": 4.652863997852421,
      "grad_norm": 0.64453125,
      "learning_rate": 5.874320565862429e-07,
      "loss": 0.9011,
      "step": 8667
    },
    {
      "epoch": 4.653400892587497,
      "grad_norm": 0.63671875,
      "learning_rate": 5.856134785550082e-07,
      "loss": 0.8471,
      "step": 8668
    },
    {
      "epoch": 4.653937787322573,
      "grad_norm": 0.60546875,
      "learning_rate": 5.837976865144563e-07,
      "loss": 0.6945,
      "step": 8669
    },
    {
      "epoch": 4.654474682057649,
      "grad_norm": 0.6875,
      "learning_rate": 5.81984680671796e-07,
      "loss": 0.9044,
      "step": 8670
    },
    {
      "epoch": 4.655011576792725,
      "grad_norm": 0.7109375,
      "learning_rate": 5.801744612339121e-07,
      "loss": 0.7981,
      "step": 8671
    },
    {
      "epoch": 4.655548471527801,
      "grad_norm": 0.6953125,
      "learning_rate": 5.783670284073727e-07,
      "loss": 0.7813,
      "step": 8672
    },
    {
      "epoch": 4.656085366262877,
      "grad_norm": 0.79296875,
      "learning_rate": 5.765623823984351e-07,
      "loss": 0.8944,
      "step": 8673
    },
    {
      "epoch": 4.656622260997953,
      "grad_norm": 0.8203125,
      "learning_rate": 5.747605234130233e-07,
      "loss": 0.937,
      "step": 8674
    },
    {
      "epoch": 4.657159155733029,
      "grad_norm": 0.6640625,
      "learning_rate": 5.729614516567588e-07,
      "loss": 0.6755,
      "step": 8675
    },
    {
      "epoch": 4.657696050468105,
      "grad_norm": 0.69921875,
      "learning_rate": 5.711651673349333e-07,
      "loss": 0.9269,
      "step": 8676
    },
    {
      "epoch": 4.658232945203181,
      "grad_norm": 0.703125,
      "learning_rate": 5.693716706525326e-07,
      "loss": 0.7199,
      "step": 8677
    },
    {
      "epoch": 4.658769839938257,
      "grad_norm": 0.7265625,
      "learning_rate": 5.675809618142097e-07,
      "loss": 0.8776,
      "step": 8678
    },
    {
      "epoch": 4.659306734673333,
      "grad_norm": 0.64453125,
      "learning_rate": 5.657930410243123e-07,
      "loss": 1.0098,
      "step": 8679
    },
    {
      "epoch": 4.659843629408409,
      "grad_norm": 0.7578125,
      "learning_rate": 5.640079084868632e-07,
      "loss": 0.9221,
      "step": 8680
    },
    {
      "epoch": 4.660380524143485,
      "grad_norm": 0.66015625,
      "learning_rate": 5.622255644055663e-07,
      "loss": 0.7009,
      "step": 8681
    },
    {
      "epoch": 4.660917418878562,
      "grad_norm": 0.796875,
      "learning_rate": 5.604460089838143e-07,
      "loss": 0.9555,
      "step": 8682
    },
    {
      "epoch": 4.661454313613637,
      "grad_norm": 0.70703125,
      "learning_rate": 5.586692424246725e-07,
      "loss": 0.8738,
      "step": 8683
    },
    {
      "epoch": 4.661991208348713,
      "grad_norm": 0.703125,
      "learning_rate": 5.568952649308957e-07,
      "loss": 1.0114,
      "step": 8684
    },
    {
      "epoch": 4.6625281030837895,
      "grad_norm": 0.62890625,
      "learning_rate": 5.551240767049187e-07,
      "loss": 0.8329,
      "step": 8685
    },
    {
      "epoch": 4.663064997818865,
      "grad_norm": 0.65625,
      "learning_rate": 5.533556779488497e-07,
      "loss": 0.7449,
      "step": 8686
    },
    {
      "epoch": 4.663601892553941,
      "grad_norm": 0.5546875,
      "learning_rate": 5.515900688644909e-07,
      "loss": 0.7247,
      "step": 8687
    },
    {
      "epoch": 4.664138787289017,
      "grad_norm": 0.84375,
      "learning_rate": 5.498272496533202e-07,
      "loss": 1.0778,
      "step": 8688
    },
    {
      "epoch": 4.664675682024093,
      "grad_norm": 0.703125,
      "learning_rate": 5.480672205164961e-07,
      "loss": 1.026,
      "step": 8689
    },
    {
      "epoch": 4.665212576759169,
      "grad_norm": 0.70703125,
      "learning_rate": 5.463099816548579e-07,
      "loss": 0.8403,
      "step": 8690
    },
    {
      "epoch": 4.665749471494245,
      "grad_norm": 0.6484375,
      "learning_rate": 5.445555332689367e-07,
      "loss": 0.8892,
      "step": 8691
    },
    {
      "epoch": 4.666286366229321,
      "grad_norm": 0.703125,
      "learning_rate": 5.428038755589282e-07,
      "loss": 0.7186,
      "step": 8692
    },
    {
      "epoch": 4.666823260964398,
      "grad_norm": 0.64453125,
      "learning_rate": 5.410550087247251e-07,
      "loss": 0.7183,
      "step": 8693
    },
    {
      "epoch": 4.667360155699473,
      "grad_norm": 0.56640625,
      "learning_rate": 5.393089329658929e-07,
      "loss": 0.7167,
      "step": 8694
    },
    {
      "epoch": 4.667897050434549,
      "grad_norm": 0.64453125,
      "learning_rate": 5.375656484816832e-07,
      "loss": 0.8619,
      "step": 8695
    },
    {
      "epoch": 4.6684339451696255,
      "grad_norm": 0.7109375,
      "learning_rate": 5.358251554710203e-07,
      "loss": 0.9259,
      "step": 8696
    },
    {
      "epoch": 4.668970839904701,
      "grad_norm": 0.62890625,
      "learning_rate": 5.340874541325231e-07,
      "loss": 0.7198,
      "step": 8697
    },
    {
      "epoch": 4.669507734639777,
      "grad_norm": 0.73046875,
      "learning_rate": 5.323525446644828e-07,
      "loss": 1.0486,
      "step": 8698
    },
    {
      "epoch": 4.670044629374853,
      "grad_norm": 0.69140625,
      "learning_rate": 5.306204272648747e-07,
      "loss": 0.8277,
      "step": 8699
    },
    {
      "epoch": 4.670581524109929,
      "grad_norm": 0.7890625,
      "learning_rate": 5.288911021313542e-07,
      "loss": 0.7683,
      "step": 8700
    },
    {
      "epoch": 4.670581524109929,
      "eval_loss": 0.9025928974151611,
      "eval_runtime": 46.0125,
      "eval_samples_per_second": 20.907,
      "eval_steps_per_second": 20.907,
      "step": 8700
    },
    {
      "epoch": 4.671118418845005,
      "grad_norm": 0.6640625,
      "learning_rate": 5.271645694612637e-07,
      "loss": 0.7914,
      "step": 8701
    },
    {
      "epoch": 4.671655313580081,
      "grad_norm": 0.6796875,
      "learning_rate": 5.254408294516178e-07,
      "loss": 0.7356,
      "step": 8702
    },
    {
      "epoch": 4.672192208315157,
      "grad_norm": 1.078125,
      "learning_rate": 5.237198822991174e-07,
      "loss": 0.8037,
      "step": 8703
    },
    {
      "epoch": 4.672729103050234,
      "grad_norm": 0.6953125,
      "learning_rate": 5.220017282001472e-07,
      "loss": 0.7648,
      "step": 8704
    },
    {
      "epoch": 4.673265997785309,
      "grad_norm": 0.84375,
      "learning_rate": 5.202863673507668e-07,
      "loss": 0.9845,
      "step": 8705
    },
    {
      "epoch": 4.673802892520385,
      "grad_norm": 0.6640625,
      "learning_rate": 5.185737999467227e-07,
      "loss": 0.8392,
      "step": 8706
    },
    {
      "epoch": 4.6743397872554615,
      "grad_norm": 0.8515625,
      "learning_rate": 5.168640261834417e-07,
      "loss": 0.8451,
      "step": 8707
    },
    {
      "epoch": 4.674876681990537,
      "grad_norm": 0.69921875,
      "learning_rate": 5.151570462560262e-07,
      "loss": 0.8909,
      "step": 8708
    },
    {
      "epoch": 4.675413576725613,
      "grad_norm": 0.71875,
      "learning_rate": 5.134528603592676e-07,
      "loss": 0.9123,
      "step": 8709
    },
    {
      "epoch": 4.675950471460689,
      "grad_norm": 0.76171875,
      "learning_rate": 5.117514686876379e-07,
      "loss": 0.8133,
      "step": 8710
    },
    {
      "epoch": 4.676487366195765,
      "grad_norm": 0.72265625,
      "learning_rate": 5.100528714352848e-07,
      "loss": 0.8935,
      "step": 8711
    },
    {
      "epoch": 4.677024260930841,
      "grad_norm": 0.75390625,
      "learning_rate": 5.083570687960337e-07,
      "loss": 1.0045,
      "step": 8712
    },
    {
      "epoch": 4.677561155665917,
      "grad_norm": 0.70703125,
      "learning_rate": 5.066640609634049e-07,
      "loss": 0.9306,
      "step": 8713
    },
    {
      "epoch": 4.678098050400993,
      "grad_norm": 0.734375,
      "learning_rate": 5.049738481305938e-07,
      "loss": 0.9433,
      "step": 8714
    },
    {
      "epoch": 4.67863494513607,
      "grad_norm": 0.671875,
      "learning_rate": 5.032864304904656e-07,
      "loss": 0.8312,
      "step": 8715
    },
    {
      "epoch": 4.679171839871145,
      "grad_norm": 0.59375,
      "learning_rate": 5.016018082355856e-07,
      "loss": 0.741,
      "step": 8716
    },
    {
      "epoch": 4.679708734606221,
      "grad_norm": 0.62890625,
      "learning_rate": 4.999199815581834e-07,
      "loss": 0.775,
      "step": 8717
    },
    {
      "epoch": 4.6802456293412975,
      "grad_norm": 0.73046875,
      "learning_rate": 4.982409506501779e-07,
      "loss": 0.9109,
      "step": 8718
    },
    {
      "epoch": 4.680782524076373,
      "grad_norm": 0.640625,
      "learning_rate": 4.965647157031739e-07,
      "loss": 0.8084,
      "step": 8719
    },
    {
      "epoch": 4.68131941881145,
      "grad_norm": 0.64453125,
      "learning_rate": 4.948912769084463e-07,
      "loss": 0.877,
      "step": 8720
    },
    {
      "epoch": 4.681856313546525,
      "grad_norm": 0.66796875,
      "learning_rate": 4.932206344569562e-07,
      "loss": 0.7337,
      "step": 8721
    },
    {
      "epoch": 4.682393208281601,
      "grad_norm": 0.6484375,
      "learning_rate": 4.915527885393429e-07,
      "loss": 0.8027,
      "step": 8722
    },
    {
      "epoch": 4.682930103016678,
      "grad_norm": 0.76171875,
      "learning_rate": 4.898877393459317e-07,
      "loss": 0.9704,
      "step": 8723
    },
    {
      "epoch": 4.683466997751753,
      "grad_norm": 0.69140625,
      "learning_rate": 4.882254870667236e-07,
      "loss": 1.0239,
      "step": 8724
    },
    {
      "epoch": 4.684003892486829,
      "grad_norm": 0.62109375,
      "learning_rate": 4.865660318914056e-07,
      "loss": 0.7066,
      "step": 8725
    },
    {
      "epoch": 4.684540787221906,
      "grad_norm": 0.78125,
      "learning_rate": 4.849093740093403e-07,
      "loss": 1.0652,
      "step": 8726
    },
    {
      "epoch": 4.685077681956981,
      "grad_norm": 0.61328125,
      "learning_rate": 4.832555136095762e-07,
      "loss": 0.7992,
      "step": 8727
    },
    {
      "epoch": 4.685614576692057,
      "grad_norm": 0.64453125,
      "learning_rate": 4.816044508808349e-07,
      "loss": 1.0047,
      "step": 8728
    },
    {
      "epoch": 4.6861514714271335,
      "grad_norm": 1.0234375,
      "learning_rate": 4.799561860115265e-07,
      "loss": 0.9498,
      "step": 8729
    },
    {
      "epoch": 4.686688366162209,
      "grad_norm": 0.8203125,
      "learning_rate": 4.783107191897368e-07,
      "loss": 0.9027,
      "step": 8730
    },
    {
      "epoch": 4.687225260897286,
      "grad_norm": 0.70703125,
      "learning_rate": 4.766680506032378e-07,
      "loss": 0.8182,
      "step": 8731
    },
    {
      "epoch": 4.687762155632361,
      "grad_norm": 0.6875,
      "learning_rate": 4.750281804394796e-07,
      "loss": 0.7541,
      "step": 8732
    },
    {
      "epoch": 4.688299050367437,
      "grad_norm": 0.59765625,
      "learning_rate": 4.733911088855847e-07,
      "loss": 0.823,
      "step": 8733
    },
    {
      "epoch": 4.688835945102514,
      "grad_norm": 0.80078125,
      "learning_rate": 4.7175683612837027e-07,
      "loss": 0.8659,
      "step": 8734
    },
    {
      "epoch": 4.689372839837589,
      "grad_norm": 0.72265625,
      "learning_rate": 4.701253623543289e-07,
      "loss": 0.7827,
      "step": 8735
    },
    {
      "epoch": 4.689909734572665,
      "grad_norm": 0.7265625,
      "learning_rate": 4.684966877496283e-07,
      "loss": 0.9438,
      "step": 8736
    },
    {
      "epoch": 4.690446629307742,
      "grad_norm": 0.74609375,
      "learning_rate": 4.668708125001198e-07,
      "loss": 0.9088,
      "step": 8737
    },
    {
      "epoch": 4.690983524042817,
      "grad_norm": 0.74609375,
      "learning_rate": 4.65247736791341e-07,
      "loss": 0.9442,
      "step": 8738
    },
    {
      "epoch": 4.691520418777893,
      "grad_norm": 0.8984375,
      "learning_rate": 4.636274608085023e-07,
      "loss": 0.9731,
      "step": 8739
    },
    {
      "epoch": 4.6920573135129695,
      "grad_norm": 0.71875,
      "learning_rate": 4.6200998473649725e-07,
      "loss": 0.792,
      "step": 8740
    },
    {
      "epoch": 4.692594208248045,
      "grad_norm": 0.66015625,
      "learning_rate": 4.6039530875990057e-07,
      "loss": 0.8352,
      "step": 8741
    },
    {
      "epoch": 4.693131102983122,
      "grad_norm": 0.6875,
      "learning_rate": 4.5878343306297033e-07,
      "loss": 0.7705,
      "step": 8742
    },
    {
      "epoch": 4.693667997718197,
      "grad_norm": 0.8125,
      "learning_rate": 4.5717435782963726e-07,
      "loss": 0.9759,
      "step": 8743
    },
    {
      "epoch": 4.694204892453273,
      "grad_norm": 0.66015625,
      "learning_rate": 4.5556808324352386e-07,
      "loss": 0.807,
      "step": 8744
    },
    {
      "epoch": 4.69474178718835,
      "grad_norm": 0.76171875,
      "learning_rate": 4.5396460948791963e-07,
      "loss": 1.0424,
      "step": 8745
    },
    {
      "epoch": 4.695278681923425,
      "grad_norm": 0.625,
      "learning_rate": 4.5236393674580324e-07,
      "loss": 1.0559,
      "step": 8746
    },
    {
      "epoch": 4.695815576658501,
      "grad_norm": 0.60546875,
      "learning_rate": 4.5076606519983135e-07,
      "loss": 0.8719,
      "step": 8747
    },
    {
      "epoch": 4.696352471393578,
      "grad_norm": 0.82421875,
      "learning_rate": 4.491709950323469e-07,
      "loss": 0.7993,
      "step": 8748
    },
    {
      "epoch": 4.696889366128653,
      "grad_norm": 0.66796875,
      "learning_rate": 4.4757872642536e-07,
      "loss": 0.8002,
      "step": 8749
    },
    {
      "epoch": 4.697426260863729,
      "grad_norm": 0.6875,
      "learning_rate": 4.459892595605697e-07,
      "loss": 0.8538,
      "step": 8750
    },
    {
      "epoch": 4.697426260863729,
      "eval_loss": 0.9027050137519836,
      "eval_runtime": 45.879,
      "eval_samples_per_second": 20.968,
      "eval_steps_per_second": 20.968,
      "step": 8750
    },
    {
      "epoch": 4.6979631555988055,
      "grad_norm": 0.67578125,
      "learning_rate": 4.444025946193586e-07,
      "loss": 0.9229,
      "step": 8751
    },
    {
      "epoch": 4.698500050333881,
      "grad_norm": 0.7109375,
      "learning_rate": 4.4281873178278475e-07,
      "loss": 0.9207,
      "step": 8752
    },
    {
      "epoch": 4.699036945068958,
      "grad_norm": 0.609375,
      "learning_rate": 4.412376712315813e-07,
      "loss": 0.7659,
      "step": 8753
    },
    {
      "epoch": 4.699573839804033,
      "grad_norm": 0.6015625,
      "learning_rate": 4.396594131461734e-07,
      "loss": 0.725,
      "step": 8754
    },
    {
      "epoch": 4.700110734539109,
      "grad_norm": 0.609375,
      "learning_rate": 4.3808395770665865e-07,
      "loss": 0.6612,
      "step": 8755
    },
    {
      "epoch": 4.700647629274186,
      "grad_norm": 0.78125,
      "learning_rate": 4.365113050928127e-07,
      "loss": 0.9153,
      "step": 8756
    },
    {
      "epoch": 4.701184524009261,
      "grad_norm": 0.71484375,
      "learning_rate": 4.349414554841003e-07,
      "loss": 0.8003,
      "step": 8757
    },
    {
      "epoch": 4.701721418744338,
      "grad_norm": 0.5625,
      "learning_rate": 4.3337440905966143e-07,
      "loss": 0.6134,
      "step": 8758
    },
    {
      "epoch": 4.702258313479414,
      "grad_norm": 0.640625,
      "learning_rate": 4.318101659983087e-07,
      "loss": 0.8941,
      "step": 8759
    },
    {
      "epoch": 4.702795208214489,
      "grad_norm": 0.6328125,
      "learning_rate": 4.302487264785521e-07,
      "loss": 0.6993,
      "step": 8760
    },
    {
      "epoch": 4.703332102949566,
      "grad_norm": 0.7421875,
      "learning_rate": 4.2869009067856293e-07,
      "loss": 0.8569,
      "step": 8761
    },
    {
      "epoch": 4.7038689976846415,
      "grad_norm": 0.8203125,
      "learning_rate": 4.2713425877620726e-07,
      "loss": 0.9141,
      "step": 8762
    },
    {
      "epoch": 4.704405892419717,
      "grad_norm": 0.703125,
      "learning_rate": 4.255812309490209e-07,
      "loss": 0.9555,
      "step": 8763
    },
    {
      "epoch": 4.704942787154794,
      "grad_norm": 0.640625,
      "learning_rate": 4.2403100737422595e-07,
      "loss": 0.8367,
      "step": 8764
    },
    {
      "epoch": 4.705479681889869,
      "grad_norm": 0.59765625,
      "learning_rate": 4.2248358822872256e-07,
      "loss": 0.6338,
      "step": 8765
    },
    {
      "epoch": 4.706016576624945,
      "grad_norm": 0.75390625,
      "learning_rate": 4.2093897368909173e-07,
      "loss": 1.0227,
      "step": 8766
    },
    {
      "epoch": 4.706553471360022,
      "grad_norm": 0.68359375,
      "learning_rate": 4.1939716393159245e-07,
      "loss": 0.7389,
      "step": 8767
    },
    {
      "epoch": 4.707090366095097,
      "grad_norm": 0.9140625,
      "learning_rate": 4.178581591321645e-07,
      "loss": 0.7469,
      "step": 8768
    },
    {
      "epoch": 4.707627260830174,
      "grad_norm": 0.6484375,
      "learning_rate": 4.163219594664286e-07,
      "loss": 0.5844,
      "step": 8769
    },
    {
      "epoch": 4.70816415556525,
      "grad_norm": 0.78515625,
      "learning_rate": 4.147885651096861e-07,
      "loss": 0.8751,
      "step": 8770
    },
    {
      "epoch": 4.708701050300325,
      "grad_norm": 0.609375,
      "learning_rate": 4.13257976236911e-07,
      "loss": 0.8312,
      "step": 8771
    },
    {
      "epoch": 4.709237945035402,
      "grad_norm": 0.7109375,
      "learning_rate": 4.1173019302276905e-07,
      "loss": 0.8552,
      "step": 8772
    },
    {
      "epoch": 4.7097748397704775,
      "grad_norm": 0.8359375,
      "learning_rate": 4.1020521564159863e-07,
      "loss": 0.8906,
      "step": 8773
    },
    {
      "epoch": 4.710311734505553,
      "grad_norm": 0.7578125,
      "learning_rate": 4.0868304426741334e-07,
      "loss": 0.8281,
      "step": 8774
    },
    {
      "epoch": 4.71084862924063,
      "grad_norm": 0.68359375,
      "learning_rate": 4.071636790739186e-07,
      "loss": 0.5785,
      "step": 8775
    },
    {
      "epoch": 4.711385523975705,
      "grad_norm": 0.9140625,
      "learning_rate": 4.0564712023449246e-07,
      "loss": 0.7485,
      "step": 8776
    },
    {
      "epoch": 4.711922418710781,
      "grad_norm": 0.71484375,
      "learning_rate": 4.041333679221909e-07,
      "loss": 1.0854,
      "step": 8777
    },
    {
      "epoch": 4.712459313445858,
      "grad_norm": 0.5703125,
      "learning_rate": 4.0262242230975353e-07,
      "loss": 0.717,
      "step": 8778
    },
    {
      "epoch": 4.712996208180933,
      "grad_norm": 0.80078125,
      "learning_rate": 4.01114283569598e-07,
      "loss": 0.8729,
      "step": 8779
    },
    {
      "epoch": 4.71353310291601,
      "grad_norm": 0.70703125,
      "learning_rate": 3.996089518738255e-07,
      "loss": 0.8346,
      "step": 8780
    },
    {
      "epoch": 4.714069997651086,
      "grad_norm": 0.71875,
      "learning_rate": 3.98106427394207e-07,
      "loss": 0.7998,
      "step": 8781
    },
    {
      "epoch": 4.714606892386161,
      "grad_norm": 0.65234375,
      "learning_rate": 3.9660671030220254e-07,
      "loss": 0.8557,
      "step": 8782
    },
    {
      "epoch": 4.715143787121238,
      "grad_norm": 0.69921875,
      "learning_rate": 3.95109800768953e-07,
      "loss": 0.8293,
      "step": 8783
    },
    {
      "epoch": 4.7156806818563135,
      "grad_norm": 0.6953125,
      "learning_rate": 3.9361569896526616e-07,
      "loss": 0.9219,
      "step": 8784
    },
    {
      "epoch": 4.716217576591389,
      "grad_norm": 0.7109375,
      "learning_rate": 3.921244050616446e-07,
      "loss": 0.9588,
      "step": 8785
    },
    {
      "epoch": 4.716754471326466,
      "grad_norm": 0.796875,
      "learning_rate": 3.9063591922826604e-07,
      "loss": 0.9811,
      "step": 8786
    },
    {
      "epoch": 4.717291366061541,
      "grad_norm": 0.69140625,
      "learning_rate": 3.891502416349779e-07,
      "loss": 0.696,
      "step": 8787
    },
    {
      "epoch": 4.717828260796617,
      "grad_norm": 0.76953125,
      "learning_rate": 3.876673724513169e-07,
      "loss": 0.8588,
      "step": 8788
    },
    {
      "epoch": 4.718365155531694,
      "grad_norm": 0.64453125,
      "learning_rate": 3.8618731184650035e-07,
      "loss": 0.9651,
      "step": 8789
    },
    {
      "epoch": 4.718902050266769,
      "grad_norm": 0.69140625,
      "learning_rate": 3.84710059989421e-07,
      "loss": 0.8654,
      "step": 8790
    },
    {
      "epoch": 4.719438945001846,
      "grad_norm": 0.703125,
      "learning_rate": 3.8323561704864676e-07,
      "loss": 0.8077,
      "step": 8791
    },
    {
      "epoch": 4.719975839736922,
      "grad_norm": 0.78125,
      "learning_rate": 3.817639831924402e-07,
      "loss": 0.804,
      "step": 8792
    },
    {
      "epoch": 4.720512734471997,
      "grad_norm": 0.765625,
      "learning_rate": 3.8029515858872543e-07,
      "loss": 0.9968,
      "step": 8793
    },
    {
      "epoch": 4.721049629207074,
      "grad_norm": 0.65234375,
      "learning_rate": 3.788291434051183e-07,
      "loss": 0.9055,
      "step": 8794
    },
    {
      "epoch": 4.7215865239421495,
      "grad_norm": 0.77734375,
      "learning_rate": 3.773659378089045e-07,
      "loss": 0.9227,
      "step": 8795
    },
    {
      "epoch": 4.722123418677226,
      "grad_norm": 0.69921875,
      "learning_rate": 3.759055419670615e-07,
      "loss": 0.8571,
      "step": 8796
    },
    {
      "epoch": 4.722660313412302,
      "grad_norm": 0.71484375,
      "learning_rate": 3.744479560462311e-07,
      "loss": 0.898,
      "step": 8797
    },
    {
      "epoch": 4.723197208147377,
      "grad_norm": 0.65234375,
      "learning_rate": 3.7299318021274953e-07,
      "loss": 0.7944,
      "step": 8798
    },
    {
      "epoch": 4.723734102882454,
      "grad_norm": 0.703125,
      "learning_rate": 3.715412146326203e-07,
      "loss": 1.0103,
      "step": 8799
    },
    {
      "epoch": 4.72427099761753,
      "grad_norm": 0.63671875,
      "learning_rate": 3.700920594715329e-07,
      "loss": 0.837,
      "step": 8800
    },
    {
      "epoch": 4.72427099761753,
      "eval_loss": 0.9027426838874817,
      "eval_runtime": 46.2574,
      "eval_samples_per_second": 20.797,
      "eval_steps_per_second": 20.797,
      "step": 8800
    },
    {
      "epoch": 4.724807892352605,
      "grad_norm": 0.6640625,
      "learning_rate": 3.6864571489485524e-07,
      "loss": 0.7167,
      "step": 8801
    },
    {
      "epoch": 4.725344787087682,
      "grad_norm": 0.67578125,
      "learning_rate": 3.6720218106763016e-07,
      "loss": 0.9992,
      "step": 8802
    },
    {
      "epoch": 4.725881681822758,
      "grad_norm": 0.76171875,
      "learning_rate": 3.6576145815458707e-07,
      "loss": 0.8699,
      "step": 8803
    },
    {
      "epoch": 4.726418576557833,
      "grad_norm": 0.73828125,
      "learning_rate": 3.6432354632012776e-07,
      "loss": 0.9403,
      "step": 8804
    },
    {
      "epoch": 4.72695547129291,
      "grad_norm": 0.6171875,
      "learning_rate": 3.6288844572834046e-07,
      "loss": 0.8786,
      "step": 8805
    },
    {
      "epoch": 4.7274923660279855,
      "grad_norm": 0.671875,
      "learning_rate": 3.6145615654298024e-07,
      "loss": 0.7672,
      "step": 8806
    },
    {
      "epoch": 4.728029260763062,
      "grad_norm": 0.69921875,
      "learning_rate": 3.60026678927497e-07,
      "loss": 0.7765,
      "step": 8807
    },
    {
      "epoch": 4.728566155498138,
      "grad_norm": 0.66015625,
      "learning_rate": 3.586000130450101e-07,
      "loss": 0.8297,
      "step": 8808
    },
    {
      "epoch": 4.729103050233213,
      "grad_norm": 0.69140625,
      "learning_rate": 3.5717615905832003e-07,
      "loss": 0.8162,
      "step": 8809
    },
    {
      "epoch": 4.72963994496829,
      "grad_norm": 0.71875,
      "learning_rate": 3.557551171299051e-07,
      "loss": 0.8945,
      "step": 8810
    },
    {
      "epoch": 4.730176839703366,
      "grad_norm": 0.69140625,
      "learning_rate": 3.5433688742192995e-07,
      "loss": 0.8245,
      "step": 8811
    },
    {
      "epoch": 4.730713734438441,
      "grad_norm": 0.6875,
      "learning_rate": 3.529214700962236e-07,
      "loss": 0.8519,
      "step": 8812
    },
    {
      "epoch": 4.731250629173518,
      "grad_norm": 0.58984375,
      "learning_rate": 3.5150886531430673e-07,
      "loss": 0.6998,
      "step": 8813
    },
    {
      "epoch": 4.731787523908594,
      "grad_norm": 0.7265625,
      "learning_rate": 3.5009907323737825e-07,
      "loss": 0.7962,
      "step": 8814
    },
    {
      "epoch": 4.732324418643669,
      "grad_norm": 0.8125,
      "learning_rate": 3.4869209402630943e-07,
      "loss": 0.8786,
      "step": 8815
    },
    {
      "epoch": 4.732861313378746,
      "grad_norm": 0.640625,
      "learning_rate": 3.4728792784165797e-07,
      "loss": 0.8427,
      "step": 8816
    },
    {
      "epoch": 4.7333982081138215,
      "grad_norm": 0.78515625,
      "learning_rate": 3.458865748436541e-07,
      "loss": 0.8291,
      "step": 8817
    },
    {
      "epoch": 4.733935102848898,
      "grad_norm": 0.77734375,
      "learning_rate": 3.444880351922114e-07,
      "loss": 0.9613,
      "step": 8818
    },
    {
      "epoch": 4.734471997583974,
      "grad_norm": 0.70703125,
      "learning_rate": 3.430923090469218e-07,
      "loss": 0.8607,
      "step": 8819
    },
    {
      "epoch": 4.735008892319049,
      "grad_norm": 0.76171875,
      "learning_rate": 3.416993965670523e-07,
      "loss": 0.7448,
      "step": 8820
    },
    {
      "epoch": 4.735545787054126,
      "grad_norm": 0.90234375,
      "learning_rate": 3.403092979115563e-07,
      "loss": 0.9397,
      "step": 8821
    },
    {
      "epoch": 4.736082681789202,
      "grad_norm": 0.6640625,
      "learning_rate": 3.389220132390569e-07,
      "loss": 0.8316,
      "step": 8822
    },
    {
      "epoch": 4.736619576524277,
      "grad_norm": 0.6796875,
      "learning_rate": 3.3753754270786365e-07,
      "loss": 0.879,
      "step": 8823
    },
    {
      "epoch": 4.737156471259354,
      "grad_norm": 0.78125,
      "learning_rate": 3.36155886475964e-07,
      "loss": 0.6894,
      "step": 8824
    },
    {
      "epoch": 4.73769336599443,
      "grad_norm": 0.90234375,
      "learning_rate": 3.3477704470101787e-07,
      "loss": 1.0246,
      "step": 8825
    },
    {
      "epoch": 4.738230260729505,
      "grad_norm": 0.6796875,
      "learning_rate": 3.33401017540369e-07,
      "loss": 0.9192,
      "step": 8826
    },
    {
      "epoch": 4.738767155464582,
      "grad_norm": 0.609375,
      "learning_rate": 3.3202780515104447e-07,
      "loss": 0.9538,
      "step": 8827
    },
    {
      "epoch": 4.7393040501996575,
      "grad_norm": 0.6328125,
      "learning_rate": 3.306574076897412e-07,
      "loss": 0.7781,
      "step": 8828
    },
    {
      "epoch": 4.739840944934734,
      "grad_norm": 0.671875,
      "learning_rate": 3.292898253128368e-07,
      "loss": 0.8118,
      "step": 8829
    },
    {
      "epoch": 4.74037783966981,
      "grad_norm": 0.66796875,
      "learning_rate": 3.279250581763982e-07,
      "loss": 1.0324,
      "step": 8830
    },
    {
      "epoch": 4.740914734404885,
      "grad_norm": 0.83203125,
      "learning_rate": 3.2656310643615074e-07,
      "loss": 0.9676,
      "step": 8831
    },
    {
      "epoch": 4.741451629139962,
      "grad_norm": 0.703125,
      "learning_rate": 3.2520397024752016e-07,
      "loss": 0.9083,
      "step": 8832
    },
    {
      "epoch": 4.741988523875038,
      "grad_norm": 0.5078125,
      "learning_rate": 3.238476497655962e-07,
      "loss": 0.5256,
      "step": 8833
    },
    {
      "epoch": 4.742525418610114,
      "grad_norm": 0.796875,
      "learning_rate": 3.224941451451552e-07,
      "loss": 0.9408,
      "step": 8834
    },
    {
      "epoch": 4.74306231334519,
      "grad_norm": 0.69921875,
      "learning_rate": 3.211434565406457e-07,
      "loss": 0.9741,
      "step": 8835
    },
    {
      "epoch": 4.743599208080266,
      "grad_norm": 0.94140625,
      "learning_rate": 3.197955841062028e-07,
      "loss": 0.7964,
      "step": 8836
    },
    {
      "epoch": 4.744136102815342,
      "grad_norm": 0.74609375,
      "learning_rate": 3.184505279956312e-07,
      "loss": 0.7008,
      "step": 8837
    },
    {
      "epoch": 4.744672997550418,
      "grad_norm": 0.70703125,
      "learning_rate": 3.171082883624221e-07,
      "loss": 0.9395,
      "step": 8838
    },
    {
      "epoch": 4.7452098922854935,
      "grad_norm": 0.7734375,
      "learning_rate": 3.1576886535973906e-07,
      "loss": 0.8211,
      "step": 8839
    },
    {
      "epoch": 4.74574678702057,
      "grad_norm": 0.63671875,
      "learning_rate": 3.144322591404292e-07,
      "loss": 0.9074,
      "step": 8840
    },
    {
      "epoch": 4.746283681755646,
      "grad_norm": 0.65234375,
      "learning_rate": 3.1309846985701774e-07,
      "loss": 0.8083,
      "step": 8841
    },
    {
      "epoch": 4.746820576490721,
      "grad_norm": 0.6640625,
      "learning_rate": 3.117674976617024e-07,
      "loss": 0.9155,
      "step": 8842
    },
    {
      "epoch": 4.747357471225798,
      "grad_norm": 0.80859375,
      "learning_rate": 3.104393427063701e-07,
      "loss": 0.8939,
      "step": 8843
    },
    {
      "epoch": 4.747894365960874,
      "grad_norm": 0.66796875,
      "learning_rate": 3.091140051425745e-07,
      "loss": 0.6923,
      "step": 8844
    },
    {
      "epoch": 4.74843126069595,
      "grad_norm": 0.72265625,
      "learning_rate": 3.077914851215585e-07,
      "loss": 0.8931,
      "step": 8845
    },
    {
      "epoch": 4.748968155431026,
      "grad_norm": 0.67578125,
      "learning_rate": 3.0647178279423206e-07,
      "loss": 0.9833,
      "step": 8846
    },
    {
      "epoch": 4.749505050166102,
      "grad_norm": 0.6953125,
      "learning_rate": 3.051548983111968e-07,
      "loss": 0.8665,
      "step": 8847
    },
    {
      "epoch": 4.750041944901178,
      "grad_norm": 0.65234375,
      "learning_rate": 3.038408318227187e-07,
      "loss": 0.8933,
      "step": 8848
    },
    {
      "epoch": 4.750578839636254,
      "grad_norm": 0.71875,
      "learning_rate": 3.025295834787556e-07,
      "loss": 0.9279,
      "step": 8849
    },
    {
      "epoch": 4.7511157343713295,
      "grad_norm": 0.5234375,
      "learning_rate": 3.01221153428935e-07,
      "loss": 0.7077,
      "step": 8850
    },
    {
      "epoch": 4.7511157343713295,
      "eval_loss": 0.9026684761047363,
      "eval_runtime": 47.3564,
      "eval_samples_per_second": 20.314,
      "eval_steps_per_second": 20.314,
      "step": 8850
    },
    {
      "epoch": 4.751652629106406,
      "grad_norm": 0.5703125,
      "learning_rate": 2.999155418225652e-07,
      "loss": 0.6956,
      "step": 8851
    },
    {
      "epoch": 4.752189523841482,
      "grad_norm": 0.70703125,
      "learning_rate": 2.9861274880863267e-07,
      "loss": 0.8107,
      "step": 8852
    },
    {
      "epoch": 4.752726418576557,
      "grad_norm": 0.640625,
      "learning_rate": 2.973127745358045e-07,
      "loss": 0.9229,
      "step": 8853
    },
    {
      "epoch": 4.753263313311634,
      "grad_norm": 0.69140625,
      "learning_rate": 2.960156191524233e-07,
      "loss": 0.9259,
      "step": 8854
    },
    {
      "epoch": 4.75380020804671,
      "grad_norm": 0.6796875,
      "learning_rate": 2.9472128280650933e-07,
      "loss": 0.8425,
      "step": 8855
    },
    {
      "epoch": 4.754337102781786,
      "grad_norm": 0.609375,
      "learning_rate": 2.934297656457641e-07,
      "loss": 0.739,
      "step": 8856
    },
    {
      "epoch": 4.754873997516862,
      "grad_norm": 0.8671875,
      "learning_rate": 2.9214106781756676e-07,
      "loss": 0.9329,
      "step": 8857
    },
    {
      "epoch": 4.755410892251938,
      "grad_norm": 0.6875,
      "learning_rate": 2.9085518946897484e-07,
      "loss": 0.7387,
      "step": 8858
    },
    {
      "epoch": 4.755947786987014,
      "grad_norm": 0.6328125,
      "learning_rate": 2.8957213074671806e-07,
      "loss": 0.7765,
      "step": 8859
    },
    {
      "epoch": 4.75648468172209,
      "grad_norm": 0.67578125,
      "learning_rate": 2.8829189179721547e-07,
      "loss": 0.7699,
      "step": 8860
    },
    {
      "epoch": 4.7570215764571655,
      "grad_norm": 0.6796875,
      "learning_rate": 2.870144727665558e-07,
      "loss": 0.8795,
      "step": 8861
    },
    {
      "epoch": 4.757558471192242,
      "grad_norm": 0.87109375,
      "learning_rate": 2.857398738005113e-07,
      "loss": 1.0587,
      "step": 8862
    },
    {
      "epoch": 4.758095365927318,
      "grad_norm": 0.609375,
      "learning_rate": 2.8446809504452665e-07,
      "loss": 0.8542,
      "step": 8863
    },
    {
      "epoch": 4.758632260662393,
      "grad_norm": 0.8203125,
      "learning_rate": 2.8319913664372755e-07,
      "loss": 0.8493,
      "step": 8864
    },
    {
      "epoch": 4.75916915539747,
      "grad_norm": 0.70703125,
      "learning_rate": 2.8193299874292313e-07,
      "loss": 0.8772,
      "step": 8865
    },
    {
      "epoch": 4.759706050132546,
      "grad_norm": 0.7421875,
      "learning_rate": 2.8066968148658945e-07,
      "loss": 0.8548,
      "step": 8866
    },
    {
      "epoch": 4.760242944867622,
      "grad_norm": 0.734375,
      "learning_rate": 2.794091850188918e-07,
      "loss": 0.9739,
      "step": 8867
    },
    {
      "epoch": 4.760779839602698,
      "grad_norm": 0.7578125,
      "learning_rate": 2.7815150948366784e-07,
      "loss": 1.0512,
      "step": 8868
    },
    {
      "epoch": 4.761316734337774,
      "grad_norm": 0.82421875,
      "learning_rate": 2.7689665502443064e-07,
      "loss": 0.9138,
      "step": 8869
    },
    {
      "epoch": 4.76185362907285,
      "grad_norm": 0.7421875,
      "learning_rate": 2.7564462178438224e-07,
      "loss": 0.903,
      "step": 8870
    },
    {
      "epoch": 4.762390523807926,
      "grad_norm": 0.59765625,
      "learning_rate": 2.743954099063889e-07,
      "loss": 0.7088,
      "step": 8871
    },
    {
      "epoch": 4.762927418543002,
      "grad_norm": 0.60546875,
      "learning_rate": 2.7314901953300334e-07,
      "loss": 0.71,
      "step": 8872
    },
    {
      "epoch": 4.763464313278078,
      "grad_norm": 0.58984375,
      "learning_rate": 2.7190545080645336e-07,
      "loss": 0.7832,
      "step": 8873
    },
    {
      "epoch": 4.764001208013154,
      "grad_norm": 0.61328125,
      "learning_rate": 2.7066470386865053e-07,
      "loss": 0.7738,
      "step": 8874
    },
    {
      "epoch": 4.76453810274823,
      "grad_norm": 0.64453125,
      "learning_rate": 2.6942677886117597e-07,
      "loss": 0.9703,
      "step": 8875
    },
    {
      "epoch": 4.765074997483306,
      "grad_norm": 0.609375,
      "learning_rate": 2.681916759252917e-07,
      "loss": 0.759,
      "step": 8876
    },
    {
      "epoch": 4.765611892218382,
      "grad_norm": 0.703125,
      "learning_rate": 2.669593952019406e-07,
      "loss": 0.9378,
      "step": 8877
    },
    {
      "epoch": 4.766148786953458,
      "grad_norm": 0.7109375,
      "learning_rate": 2.657299368317434e-07,
      "loss": 0.8146,
      "step": 8878
    },
    {
      "epoch": 4.766685681688534,
      "grad_norm": 0.77734375,
      "learning_rate": 2.645033009549935e-07,
      "loss": 0.743,
      "step": 8879
    },
    {
      "epoch": 4.76722257642361,
      "grad_norm": 0.953125,
      "learning_rate": 2.632794877116651e-07,
      "loss": 0.8351,
      "step": 8880
    },
    {
      "epoch": 4.767759471158686,
      "grad_norm": 0.7265625,
      "learning_rate": 2.620584972414131e-07,
      "loss": 0.8081,
      "step": 8881
    },
    {
      "epoch": 4.768296365893762,
      "grad_norm": 0.68359375,
      "learning_rate": 2.6084032968356763e-07,
      "loss": 0.9437,
      "step": 8882
    },
    {
      "epoch": 4.768833260628838,
      "grad_norm": 0.5859375,
      "learning_rate": 2.5962498517713706e-07,
      "loss": 0.8131,
      "step": 8883
    },
    {
      "epoch": 4.769370155363914,
      "grad_norm": 0.77734375,
      "learning_rate": 2.5841246386080765e-07,
      "loss": 0.7389,
      "step": 8884
    },
    {
      "epoch": 4.76990705009899,
      "grad_norm": 0.59375,
      "learning_rate": 2.572027658729409e-07,
      "loss": 0.7616,
      "step": 8885
    },
    {
      "epoch": 4.770443944834066,
      "grad_norm": 0.6484375,
      "learning_rate": 2.55995891351582e-07,
      "loss": 0.8065,
      "step": 8886
    },
    {
      "epoch": 4.770980839569142,
      "grad_norm": 0.6015625,
      "learning_rate": 2.5479184043445127e-07,
      "loss": 0.8393,
      "step": 8887
    },
    {
      "epoch": 4.771517734304218,
      "grad_norm": 0.73828125,
      "learning_rate": 2.535906132589416e-07,
      "loss": 0.8836,
      "step": 8888
    },
    {
      "epoch": 4.772054629039294,
      "grad_norm": 0.796875,
      "learning_rate": 2.5239220996212955e-07,
      "loss": 0.999,
      "step": 8889
    },
    {
      "epoch": 4.77259152377437,
      "grad_norm": 0.66015625,
      "learning_rate": 2.5119663068077227e-07,
      "loss": 0.7627,
      "step": 8890
    },
    {
      "epoch": 4.773128418509446,
      "grad_norm": 0.7421875,
      "learning_rate": 2.500038755512968e-07,
      "loss": 0.922,
      "step": 8891
    },
    {
      "epoch": 4.773665313244522,
      "grad_norm": 0.6171875,
      "learning_rate": 2.488139447098109e-07,
      "loss": 0.7346,
      "step": 8892
    },
    {
      "epoch": 4.774202207979598,
      "grad_norm": 0.76171875,
      "learning_rate": 2.47626838292106e-07,
      "loss": 0.9338,
      "step": 8893
    },
    {
      "epoch": 4.774739102714674,
      "grad_norm": 0.83984375,
      "learning_rate": 2.4644255643363755e-07,
      "loss": 0.9669,
      "step": 8894
    },
    {
      "epoch": 4.77527599744975,
      "grad_norm": 0.59765625,
      "learning_rate": 2.452610992695559e-07,
      "loss": 0.7975,
      "step": 8895
    },
    {
      "epoch": 4.775812892184826,
      "grad_norm": 0.625,
      "learning_rate": 2.440824669346753e-07,
      "loss": 0.7544,
      "step": 8896
    },
    {
      "epoch": 4.776349786919902,
      "grad_norm": 0.74609375,
      "learning_rate": 2.429066595634938e-07,
      "loss": 0.9778,
      "step": 8897
    },
    {
      "epoch": 4.776886681654978,
      "grad_norm": 0.66796875,
      "learning_rate": 2.4173367729018457e-07,
      "loss": 0.7701,
      "step": 8898
    },
    {
      "epoch": 4.777423576390054,
      "grad_norm": 0.73046875,
      "learning_rate": 2.405635202486017e-07,
      "loss": 1.0011,
      "step": 8899
    },
    {
      "epoch": 4.77796047112513,
      "grad_norm": 0.6171875,
      "learning_rate": 2.3939618857227174e-07,
      "loss": 0.8734,
      "step": 8900
    },
    {
      "epoch": 4.77796047112513,
      "eval_loss": 0.9026810526847839,
      "eval_runtime": 46.4544,
      "eval_samples_per_second": 20.709,
      "eval_steps_per_second": 20.709,
      "step": 8900
    },
    {
      "epoch": 4.778497365860206,
      "grad_norm": 0.60546875,
      "learning_rate": 2.3823168239440752e-07,
      "loss": 0.8939,
      "step": 8901
    },
    {
      "epoch": 4.779034260595282,
      "grad_norm": 1.0390625,
      "learning_rate": 2.3707000184788887e-07,
      "loss": 0.9509,
      "step": 8902
    },
    {
      "epoch": 4.779571155330358,
      "grad_norm": 0.58984375,
      "learning_rate": 2.3591114706528195e-07,
      "loss": 0.7246,
      "step": 8903
    },
    {
      "epoch": 4.780108050065434,
      "grad_norm": 0.796875,
      "learning_rate": 2.3475511817882268e-07,
      "loss": 0.8044,
      "step": 8904
    },
    {
      "epoch": 4.78064494480051,
      "grad_norm": 0.55078125,
      "learning_rate": 2.336019153204305e-07,
      "loss": 0.646,
      "step": 8905
    },
    {
      "epoch": 4.781181839535586,
      "grad_norm": 0.6796875,
      "learning_rate": 2.3245153862170298e-07,
      "loss": 0.888,
      "step": 8906
    },
    {
      "epoch": 4.781718734270662,
      "grad_norm": 0.66796875,
      "learning_rate": 2.3130398821391007e-07,
      "loss": 0.9328,
      "step": 8907
    },
    {
      "epoch": 4.782255629005738,
      "grad_norm": 0.6328125,
      "learning_rate": 2.301592642280026e-07,
      "loss": 0.7437,
      "step": 8908
    },
    {
      "epoch": 4.782792523740814,
      "grad_norm": 0.70703125,
      "learning_rate": 2.290173667946066e-07,
      "loss": 0.8947,
      "step": 8909
    },
    {
      "epoch": 4.7833294184758905,
      "grad_norm": 0.78125,
      "learning_rate": 2.27878296044029e-07,
      "loss": 0.7926,
      "step": 8910
    },
    {
      "epoch": 4.783866313210966,
      "grad_norm": 0.62109375,
      "learning_rate": 2.267420521062491e-07,
      "loss": 0.919,
      "step": 8911
    },
    {
      "epoch": 4.784403207946042,
      "grad_norm": 0.68359375,
      "learning_rate": 2.2560863511093267e-07,
      "loss": 0.9469,
      "step": 8912
    },
    {
      "epoch": 4.7849401026811185,
      "grad_norm": 0.7578125,
      "learning_rate": 2.2447804518741232e-07,
      "loss": 0.8512,
      "step": 8913
    },
    {
      "epoch": 4.785476997416194,
      "grad_norm": 0.78515625,
      "learning_rate": 2.2335028246470159e-07,
      "loss": 0.8271,
      "step": 8914
    },
    {
      "epoch": 4.78601389215127,
      "grad_norm": 0.6953125,
      "learning_rate": 2.2222534707149745e-07,
      "loss": 0.9416,
      "step": 8915
    },
    {
      "epoch": 4.786550786886346,
      "grad_norm": 0.7578125,
      "learning_rate": 2.2110323913616392e-07,
      "loss": 0.8162,
      "step": 8916
    },
    {
      "epoch": 4.787087681621422,
      "grad_norm": 1.1875,
      "learning_rate": 2.199839587867486e-07,
      "loss": 0.9047,
      "step": 8917
    },
    {
      "epoch": 4.787624576356498,
      "grad_norm": 0.78125,
      "learning_rate": 2.1886750615097984e-07,
      "loss": 0.8363,
      "step": 8918
    },
    {
      "epoch": 4.788161471091574,
      "grad_norm": 0.67578125,
      "learning_rate": 2.1775388135625573e-07,
      "loss": 0.817,
      "step": 8919
    },
    {
      "epoch": 4.78869836582665,
      "grad_norm": 0.71484375,
      "learning_rate": 2.16643084529658e-07,
      "loss": 0.9826,
      "step": 8920
    },
    {
      "epoch": 4.7892352605617265,
      "grad_norm": 0.75,
      "learning_rate": 2.1553511579793518e-07,
      "loss": 0.8506,
      "step": 8921
    },
    {
      "epoch": 4.789772155296802,
      "grad_norm": 0.7578125,
      "learning_rate": 2.144299752875306e-07,
      "loss": 0.8597,
      "step": 8922
    },
    {
      "epoch": 4.790309050031878,
      "grad_norm": 0.8671875,
      "learning_rate": 2.1332766312454623e-07,
      "loss": 1.0634,
      "step": 8923
    },
    {
      "epoch": 4.7908459447669545,
      "grad_norm": 0.6484375,
      "learning_rate": 2.12228179434773e-07,
      "loss": 0.7643,
      "step": 8924
    },
    {
      "epoch": 4.79138283950203,
      "grad_norm": 0.61328125,
      "learning_rate": 2.1113152434367733e-07,
      "loss": 0.8133,
      "step": 8925
    },
    {
      "epoch": 4.791919734237106,
      "grad_norm": 0.6328125,
      "learning_rate": 2.1003769797640072e-07,
      "loss": 0.8601,
      "step": 8926
    },
    {
      "epoch": 4.792456628972182,
      "grad_norm": 0.83984375,
      "learning_rate": 2.0894670045776278e-07,
      "loss": 0.8927,
      "step": 8927
    },
    {
      "epoch": 4.792993523707258,
      "grad_norm": 0.72265625,
      "learning_rate": 2.0785853191225835e-07,
      "loss": 0.811,
      "step": 8928
    },
    {
      "epoch": 4.793530418442334,
      "grad_norm": 0.71484375,
      "learning_rate": 2.0677319246406312e-07,
      "loss": 0.7321,
      "step": 8929
    },
    {
      "epoch": 4.79406731317741,
      "grad_norm": 0.66015625,
      "learning_rate": 2.0569068223702803e-07,
      "loss": 0.7338,
      "step": 8930
    },
    {
      "epoch": 4.794604207912486,
      "grad_norm": 0.6875,
      "learning_rate": 2.0461100135467925e-07,
      "loss": 0.8936,
      "step": 8931
    },
    {
      "epoch": 4.7951411026475625,
      "grad_norm": 0.62109375,
      "learning_rate": 2.035341499402238e-07,
      "loss": 0.7176,
      "step": 8932
    },
    {
      "epoch": 4.795677997382638,
      "grad_norm": 0.671875,
      "learning_rate": 2.0246012811654391e-07,
      "loss": 0.9077,
      "step": 8933
    },
    {
      "epoch": 4.796214892117714,
      "grad_norm": 0.71484375,
      "learning_rate": 2.013889360061999e-07,
      "loss": 0.8842,
      "step": 8934
    },
    {
      "epoch": 4.7967517868527905,
      "grad_norm": 0.64453125,
      "learning_rate": 2.0032057373142454e-07,
      "loss": 0.943,
      "step": 8935
    },
    {
      "epoch": 4.797288681587866,
      "grad_norm": 0.640625,
      "learning_rate": 1.9925504141413697e-07,
      "loss": 0.7171,
      "step": 8936
    },
    {
      "epoch": 4.797825576322942,
      "grad_norm": 0.5703125,
      "learning_rate": 1.9819233917592606e-07,
      "loss": 0.8325,
      "step": 8937
    },
    {
      "epoch": 4.798362471058018,
      "grad_norm": 0.73046875,
      "learning_rate": 1.9713246713805588e-07,
      "loss": 0.7394,
      "step": 8938
    },
    {
      "epoch": 4.798899365793094,
      "grad_norm": 0.64453125,
      "learning_rate": 1.9607542542147416e-07,
      "loss": 0.957,
      "step": 8939
    },
    {
      "epoch": 4.79943626052817,
      "grad_norm": 0.77734375,
      "learning_rate": 1.9502121414680385e-07,
      "loss": 0.9808,
      "step": 8940
    },
    {
      "epoch": 4.799973155263246,
      "grad_norm": 0.9921875,
      "learning_rate": 1.9396983343434315e-07,
      "loss": 0.9053,
      "step": 8941
    },
    {
      "epoch": 4.800510049998322,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9292128340406556e-07,
      "loss": 1.0395,
      "step": 8942
    },
    {
      "epoch": 4.8010469447333985,
      "grad_norm": 0.65625,
      "learning_rate": 1.9187556417562812e-07,
      "loss": 0.7515,
      "step": 8943
    },
    {
      "epoch": 4.801583839468474,
      "grad_norm": 0.734375,
      "learning_rate": 1.9083267586835484e-07,
      "loss": 0.9611,
      "step": 8944
    },
    {
      "epoch": 4.80212073420355,
      "grad_norm": 0.65625,
      "learning_rate": 1.8979261860125885e-07,
      "loss": 0.7925,
      "step": 8945
    },
    {
      "epoch": 4.8026576289386265,
      "grad_norm": 0.78125,
      "learning_rate": 1.887553924930202e-07,
      "loss": 0.9188,
      "step": 8946
    },
    {
      "epoch": 4.803194523673702,
      "grad_norm": 0.75390625,
      "learning_rate": 1.877209976619998e-07,
      "loss": 0.8435,
      "step": 8947
    },
    {
      "epoch": 4.803731418408779,
      "grad_norm": 0.734375,
      "learning_rate": 1.8668943422623652e-07,
      "loss": 1.0408,
      "step": 8948
    },
    {
      "epoch": 4.804268313143854,
      "grad_norm": 0.65625,
      "learning_rate": 1.8566070230344179e-07,
      "loss": 0.8568,
      "step": 8949
    },
    {
      "epoch": 4.80480520787893,
      "grad_norm": 0.671875,
      "learning_rate": 1.8463480201101334e-07,
      "loss": 0.8391,
      "step": 8950
    },
    {
      "epoch": 4.80480520787893,
      "eval_loss": 0.9026857018470764,
      "eval_runtime": 46.5931,
      "eval_samples_per_second": 20.647,
      "eval_steps_per_second": 20.647,
      "step": 8950
    },
    {
      "epoch": 4.805342102614007,
      "grad_norm": 0.60546875,
      "learning_rate": 1.8361173346601034e-07,
      "loss": 0.6363,
      "step": 8951
    },
    {
      "epoch": 4.805878997349082,
      "grad_norm": 0.6796875,
      "learning_rate": 1.8259149678518662e-07,
      "loss": 0.6233,
      "step": 8952
    },
    {
      "epoch": 4.806415892084158,
      "grad_norm": 0.74609375,
      "learning_rate": 1.8157409208495735e-07,
      "loss": 0.7178,
      "step": 8953
    },
    {
      "epoch": 4.8069527868192345,
      "grad_norm": 0.7265625,
      "learning_rate": 1.805595194814269e-07,
      "loss": 0.8452,
      "step": 8954
    },
    {
      "epoch": 4.80748968155431,
      "grad_norm": 0.6875,
      "learning_rate": 1.7954777909036656e-07,
      "loss": 0.9099,
      "step": 8955
    },
    {
      "epoch": 4.808026576289386,
      "grad_norm": 0.71484375,
      "learning_rate": 1.7853887102723123e-07,
      "loss": 0.8106,
      "step": 8956
    },
    {
      "epoch": 4.8085634710244625,
      "grad_norm": 0.62890625,
      "learning_rate": 1.7753279540714817e-07,
      "loss": 0.7842,
      "step": 8957
    },
    {
      "epoch": 4.809100365759538,
      "grad_norm": 0.88671875,
      "learning_rate": 1.7652955234492285e-07,
      "loss": 1.0033,
      "step": 8958
    },
    {
      "epoch": 4.809637260494615,
      "grad_norm": 0.6796875,
      "learning_rate": 1.7552914195504145e-07,
      "loss": 0.7133,
      "step": 8959
    },
    {
      "epoch": 4.81017415522969,
      "grad_norm": 0.69921875,
      "learning_rate": 1.7453156435165986e-07,
      "loss": 0.8815,
      "step": 8960
    },
    {
      "epoch": 4.810711049964766,
      "grad_norm": 0.58984375,
      "learning_rate": 1.735368196486148e-07,
      "loss": 0.7701,
      "step": 8961
    },
    {
      "epoch": 4.811247944699843,
      "grad_norm": 0.6875,
      "learning_rate": 1.7254490795942378e-07,
      "loss": 0.8473,
      "step": 8962
    },
    {
      "epoch": 4.811784839434918,
      "grad_norm": 0.73046875,
      "learning_rate": 1.7155582939727134e-07,
      "loss": 0.7896,
      "step": 8963
    },
    {
      "epoch": 4.812321734169994,
      "grad_norm": 0.66015625,
      "learning_rate": 1.7056958407502544e-07,
      "loss": 0.8496,
      "step": 8964
    },
    {
      "epoch": 4.8128586289050705,
      "grad_norm": 0.6875,
      "learning_rate": 1.6958617210522943e-07,
      "loss": 0.6718,
      "step": 8965
    },
    {
      "epoch": 4.813395523640146,
      "grad_norm": 0.59375,
      "learning_rate": 1.6860559360010463e-07,
      "loss": 0.8748,
      "step": 8966
    },
    {
      "epoch": 4.813932418375222,
      "grad_norm": 0.78125,
      "learning_rate": 1.6762784867154212e-07,
      "loss": 0.8733,
      "step": 8967
    },
    {
      "epoch": 4.8144693131102985,
      "grad_norm": 0.62890625,
      "learning_rate": 1.666529374311221e-07,
      "loss": 0.8356,
      "step": 8968
    },
    {
      "epoch": 4.815006207845374,
      "grad_norm": 0.65234375,
      "learning_rate": 1.6568085999008888e-07,
      "loss": 0.6574,
      "step": 8969
    },
    {
      "epoch": 4.815543102580451,
      "grad_norm": 0.66796875,
      "learning_rate": 1.6471161645937038e-07,
      "loss": 0.76,
      "step": 8970
    },
    {
      "epoch": 4.816079997315526,
      "grad_norm": 0.64453125,
      "learning_rate": 1.6374520694957263e-07,
      "loss": 0.7994,
      "step": 8971
    },
    {
      "epoch": 4.816616892050602,
      "grad_norm": 0.7734375,
      "learning_rate": 1.6278163157097127e-07,
      "loss": 0.9175,
      "step": 8972
    },
    {
      "epoch": 4.817153786785679,
      "grad_norm": 0.7890625,
      "learning_rate": 1.6182089043352277e-07,
      "loss": 0.7615,
      "step": 8973
    },
    {
      "epoch": 4.817690681520754,
      "grad_norm": 0.6953125,
      "learning_rate": 1.608629836468617e-07,
      "loss": 0.8836,
      "step": 8974
    },
    {
      "epoch": 4.81822757625583,
      "grad_norm": 0.68359375,
      "learning_rate": 1.5990791132029503e-07,
      "loss": 0.7681,
      "step": 8975
    },
    {
      "epoch": 4.8187644709909065,
      "grad_norm": 0.94140625,
      "learning_rate": 1.589556735628106e-07,
      "loss": 0.9287,
      "step": 8976
    },
    {
      "epoch": 4.819301365725982,
      "grad_norm": 0.68359375,
      "learning_rate": 1.5800627048307148e-07,
      "loss": 0.9093,
      "step": 8977
    },
    {
      "epoch": 4.819838260461058,
      "grad_norm": 0.89453125,
      "learning_rate": 1.5705970218941323e-07,
      "loss": 0.9605,
      "step": 8978
    },
    {
      "epoch": 4.8203751551961345,
      "grad_norm": 0.6015625,
      "learning_rate": 1.56115968789855e-07,
      "loss": 0.812,
      "step": 8979
    },
    {
      "epoch": 4.82091204993121,
      "grad_norm": 0.5859375,
      "learning_rate": 1.5517507039208568e-07,
      "loss": 0.6377,
      "step": 8980
    },
    {
      "epoch": 4.821448944666287,
      "grad_norm": 0.6796875,
      "learning_rate": 1.5423700710347765e-07,
      "loss": 0.6437,
      "step": 8981
    },
    {
      "epoch": 4.821985839401362,
      "grad_norm": 0.7421875,
      "learning_rate": 1.5330177903107036e-07,
      "loss": 0.8401,
      "step": 8982
    },
    {
      "epoch": 4.822522734136438,
      "grad_norm": 0.78125,
      "learning_rate": 1.523693862815895e-07,
      "loss": 0.9717,
      "step": 8983
    },
    {
      "epoch": 4.823059628871515,
      "grad_norm": 0.80859375,
      "learning_rate": 1.514398289614305e-07,
      "loss": 0.8353,
      "step": 8984
    },
    {
      "epoch": 4.82359652360659,
      "grad_norm": 0.671875,
      "learning_rate": 1.5051310717666967e-07,
      "loss": 0.8644,
      "step": 8985
    },
    {
      "epoch": 4.824133418341667,
      "grad_norm": 0.63671875,
      "learning_rate": 1.4958922103305574e-07,
      "loss": 0.8548,
      "step": 8986
    },
    {
      "epoch": 4.8246703130767425,
      "grad_norm": 0.75390625,
      "learning_rate": 1.4866817063601547e-07,
      "loss": 0.7597,
      "step": 8987
    },
    {
      "epoch": 4.825207207811818,
      "grad_norm": 0.73828125,
      "learning_rate": 1.4774995609065646e-07,
      "loss": 1.0039,
      "step": 8988
    },
    {
      "epoch": 4.825744102546895,
      "grad_norm": 0.70703125,
      "learning_rate": 1.4683457750175323e-07,
      "loss": 0.7785,
      "step": 8989
    },
    {
      "epoch": 4.8262809972819705,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4592203497376667e-07,
      "loss": 0.9638,
      "step": 8990
    },
    {
      "epoch": 4.826817892017046,
      "grad_norm": 0.65234375,
      "learning_rate": 1.4501232861082735e-07,
      "loss": 0.708,
      "step": 8991
    },
    {
      "epoch": 4.827354786752123,
      "grad_norm": 0.76953125,
      "learning_rate": 1.4410545851674118e-07,
      "loss": 0.9362,
      "step": 8992
    },
    {
      "epoch": 4.827891681487198,
      "grad_norm": 0.6953125,
      "learning_rate": 1.432014247950003e-07,
      "loss": 0.7802,
      "step": 8993
    },
    {
      "epoch": 4.828428576222274,
      "grad_norm": 0.71484375,
      "learning_rate": 1.4230022754876115e-07,
      "loss": 0.8584,
      "step": 8994
    },
    {
      "epoch": 4.828965470957351,
      "grad_norm": 0.74609375,
      "learning_rate": 1.4140186688086366e-07,
      "loss": 0.7625,
      "step": 8995
    },
    {
      "epoch": 4.829502365692426,
      "grad_norm": 0.796875,
      "learning_rate": 1.4050634289382302e-07,
      "loss": 0.8922,
      "step": 8996
    },
    {
      "epoch": 4.830039260427503,
      "grad_norm": 0.6640625,
      "learning_rate": 1.3961365568982698e-07,
      "loss": 0.7938,
      "step": 8997
    },
    {
      "epoch": 4.8305761551625785,
      "grad_norm": 0.68359375,
      "learning_rate": 1.3872380537074404e-07,
      "loss": 0.8929,
      "step": 8998
    },
    {
      "epoch": 4.831113049897654,
      "grad_norm": 0.64453125,
      "learning_rate": 1.3783679203811794e-07,
      "loss": 0.7465,
      "step": 8999
    },
    {
      "epoch": 4.831649944632731,
      "grad_norm": 0.64453125,
      "learning_rate": 1.3695261579316777e-07,
      "loss": 0.7243,
      "step": 9000
    },
    {
      "epoch": 4.831649944632731,
      "eval_loss": 0.9027743935585022,
      "eval_runtime": 46.2143,
      "eval_samples_per_second": 20.816,
      "eval_steps_per_second": 20.816,
      "step": 9000
    },
    {
      "epoch": 4.8321868393678065,
      "grad_norm": 0.7265625,
      "learning_rate": 1.3607127673678777e-07,
      "loss": 0.936,
      "step": 9001
    },
    {
      "epoch": 4.832723734102882,
      "grad_norm": 0.6484375,
      "learning_rate": 1.3519277496955307e-07,
      "loss": 0.8228,
      "step": 9002
    },
    {
      "epoch": 4.833260628837959,
      "grad_norm": 0.7890625,
      "learning_rate": 1.343171105917085e-07,
      "loss": 0.9724,
      "step": 9003
    },
    {
      "epoch": 4.833797523573034,
      "grad_norm": 0.55859375,
      "learning_rate": 1.334442837031824e-07,
      "loss": 0.5936,
      "step": 9004
    },
    {
      "epoch": 4.83433441830811,
      "grad_norm": 0.71484375,
      "learning_rate": 1.3257429440357016e-07,
      "loss": 0.7406,
      "step": 9005
    },
    {
      "epoch": 4.834871313043187,
      "grad_norm": 0.74609375,
      "learning_rate": 1.3170714279215346e-07,
      "loss": 0.7096,
      "step": 9006
    },
    {
      "epoch": 4.835408207778262,
      "grad_norm": 0.6953125,
      "learning_rate": 1.3084282896788368e-07,
      "loss": 0.814,
      "step": 9007
    },
    {
      "epoch": 4.835945102513339,
      "grad_norm": 0.6484375,
      "learning_rate": 1.2998135302938753e-07,
      "loss": 0.8512,
      "step": 9008
    },
    {
      "epoch": 4.8364819972484145,
      "grad_norm": 1.0546875,
      "learning_rate": 1.2912271507497243e-07,
      "loss": 0.9884,
      "step": 9009
    },
    {
      "epoch": 4.83701889198349,
      "grad_norm": 0.80859375,
      "learning_rate": 1.2826691520262114e-07,
      "loss": 0.9556,
      "step": 9010
    },
    {
      "epoch": 4.837555786718567,
      "grad_norm": 0.68359375,
      "learning_rate": 1.2741395350998887e-07,
      "loss": 0.8672,
      "step": 9011
    },
    {
      "epoch": 4.8380926814536425,
      "grad_norm": 0.63671875,
      "learning_rate": 1.2656383009441163e-07,
      "loss": 0.6835,
      "step": 9012
    },
    {
      "epoch": 4.838629576188718,
      "grad_norm": 0.72265625,
      "learning_rate": 1.2571654505289798e-07,
      "loss": 0.8848,
      "step": 9013
    },
    {
      "epoch": 4.839166470923795,
      "grad_norm": 0.63671875,
      "learning_rate": 1.2487209848213165e-07,
      "loss": 0.8183,
      "step": 9014
    },
    {
      "epoch": 4.83970336565887,
      "grad_norm": 0.6640625,
      "learning_rate": 1.2403049047847725e-07,
      "loss": 0.9652,
      "step": 9015
    },
    {
      "epoch": 4.840240260393946,
      "grad_norm": 0.828125,
      "learning_rate": 1.231917211379746e-07,
      "loss": 1.0092,
      "step": 9016
    },
    {
      "epoch": 4.840777155129023,
      "grad_norm": 0.65234375,
      "learning_rate": 1.2235579055633327e-07,
      "loss": 0.7694,
      "step": 9017
    },
    {
      "epoch": 4.841314049864098,
      "grad_norm": 0.69921875,
      "learning_rate": 1.2152269882894918e-07,
      "loss": 0.7765,
      "step": 9018
    },
    {
      "epoch": 4.841850944599175,
      "grad_norm": 0.73046875,
      "learning_rate": 1.2069244605088236e-07,
      "loss": 0.8609,
      "step": 9019
    },
    {
      "epoch": 4.8423878393342505,
      "grad_norm": 0.66015625,
      "learning_rate": 1.1986503231688207e-07,
      "loss": 0.8848,
      "step": 9020
    },
    {
      "epoch": 4.842924734069326,
      "grad_norm": 0.6953125,
      "learning_rate": 1.1904045772135886e-07,
      "loss": 0.6962,
      "step": 9021
    },
    {
      "epoch": 4.843461628804403,
      "grad_norm": 0.7578125,
      "learning_rate": 1.1821872235841525e-07,
      "loss": 0.9984,
      "step": 9022
    },
    {
      "epoch": 4.8439985235394785,
      "grad_norm": 0.7265625,
      "learning_rate": 1.1739982632181512e-07,
      "loss": 0.8243,
      "step": 9023
    },
    {
      "epoch": 4.844535418274555,
      "grad_norm": 0.55078125,
      "learning_rate": 1.1658376970500873e-07,
      "loss": 0.6503,
      "step": 9024
    },
    {
      "epoch": 4.845072313009631,
      "grad_norm": 0.73046875,
      "learning_rate": 1.1577055260111602e-07,
      "loss": 0.6767,
      "step": 9025
    },
    {
      "epoch": 4.845609207744706,
      "grad_norm": 0.64453125,
      "learning_rate": 1.1496017510293777e-07,
      "loss": 0.7621,
      "step": 9026
    },
    {
      "epoch": 4.846146102479782,
      "grad_norm": 0.609375,
      "learning_rate": 1.1415263730294445e-07,
      "loss": 0.7268,
      "step": 9027
    },
    {
      "epoch": 4.846682997214859,
      "grad_norm": 0.58984375,
      "learning_rate": 1.1334793929329291e-07,
      "loss": 0.7942,
      "step": 9028
    },
    {
      "epoch": 4.847219891949934,
      "grad_norm": 0.73828125,
      "learning_rate": 1.1254608116580412e-07,
      "loss": 0.7029,
      "step": 9029
    },
    {
      "epoch": 4.847756786685011,
      "grad_norm": 0.7109375,
      "learning_rate": 1.117470630119799e-07,
      "loss": 0.785,
      "step": 9030
    },
    {
      "epoch": 4.8482936814200865,
      "grad_norm": 0.78515625,
      "learning_rate": 1.109508849230001e-07,
      "loss": 0.8514,
      "step": 9031
    },
    {
      "epoch": 4.848830576155162,
      "grad_norm": 0.62890625,
      "learning_rate": 1.1015754698971981e-07,
      "loss": 0.7724,
      "step": 9032
    },
    {
      "epoch": 4.849367470890239,
      "grad_norm": 0.6640625,
      "learning_rate": 1.0936704930266661e-07,
      "loss": 0.8121,
      "step": 9033
    },
    {
      "epoch": 4.8499043656253145,
      "grad_norm": 0.7109375,
      "learning_rate": 1.0857939195204891e-07,
      "loss": 0.7787,
      "step": 9034
    },
    {
      "epoch": 4.850441260360391,
      "grad_norm": 0.6484375,
      "learning_rate": 1.0779457502774482e-07,
      "loss": 0.755,
      "step": 9035
    },
    {
      "epoch": 4.850978155095467,
      "grad_norm": 0.6875,
      "learning_rate": 1.07012598619316e-07,
      "loss": 0.9678,
      "step": 9036
    },
    {
      "epoch": 4.851515049830542,
      "grad_norm": 0.71484375,
      "learning_rate": 1.0623346281599389e-07,
      "loss": 0.6666,
      "step": 9037
    },
    {
      "epoch": 4.852051944565619,
      "grad_norm": 0.69921875,
      "learning_rate": 1.054571677066879e-07,
      "loss": 0.8457,
      "step": 9038
    },
    {
      "epoch": 4.852588839300695,
      "grad_norm": 0.93359375,
      "learning_rate": 1.0468371337998273e-07,
      "loss": 0.8623,
      "step": 9039
    },
    {
      "epoch": 4.85312573403577,
      "grad_norm": 0.703125,
      "learning_rate": 1.0391309992413833e-07,
      "loss": 1.0475,
      "step": 9040
    },
    {
      "epoch": 4.853662628770847,
      "grad_norm": 0.6640625,
      "learning_rate": 1.0314532742709271e-07,
      "loss": 0.8704,
      "step": 9041
    },
    {
      "epoch": 4.8541995235059225,
      "grad_norm": 0.6640625,
      "learning_rate": 1.023803959764591e-07,
      "loss": 0.7975,
      "step": 9042
    },
    {
      "epoch": 4.854736418240998,
      "grad_norm": 0.76171875,
      "learning_rate": 1.0161830565952602e-07,
      "loss": 0.8559,
      "step": 9043
    },
    {
      "epoch": 4.855273312976075,
      "grad_norm": 0.671875,
      "learning_rate": 1.0085905656325445e-07,
      "loss": 0.743,
      "step": 9044
    },
    {
      "epoch": 4.8558102077111505,
      "grad_norm": 0.546875,
      "learning_rate": 1.0010264877428899e-07,
      "loss": 1.0218,
      "step": 9045
    },
    {
      "epoch": 4.856347102446227,
      "grad_norm": 0.6015625,
      "learning_rate": 9.934908237894114e-08,
      "loss": 0.8144,
      "step": 9046
    },
    {
      "epoch": 4.856883997181303,
      "grad_norm": 0.6953125,
      "learning_rate": 9.859835746320601e-08,
      "loss": 0.8618,
      "step": 9047
    },
    {
      "epoch": 4.857420891916378,
      "grad_norm": 0.72265625,
      "learning_rate": 9.785047411274839e-08,
      "loss": 0.9914,
      "step": 9048
    },
    {
      "epoch": 4.857957786651455,
      "grad_norm": 0.609375,
      "learning_rate": 9.710543241291115e-08,
      "loss": 0.7195,
      "step": 9049
    },
    {
      "epoch": 4.858494681386531,
      "grad_norm": 0.67578125,
      "learning_rate": 9.636323244871514e-08,
      "loss": 0.6905,
      "step": 9050
    },
    {
      "epoch": 4.858494681386531,
      "eval_loss": 0.9026276469230652,
      "eval_runtime": 45.9643,
      "eval_samples_per_second": 20.929,
      "eval_steps_per_second": 20.929,
      "step": 9050
    },
    {
      "epoch": 4.859031576121606,
      "grad_norm": 0.671875,
      "learning_rate": 9.562387430485099e-08,
      "loss": 0.8008,
      "step": 9051
    },
    {
      "epoch": 4.859568470856683,
      "grad_norm": 0.671875,
      "learning_rate": 9.488735806569283e-08,
      "loss": 1.2051,
      "step": 9052
    },
    {
      "epoch": 4.8601053655917585,
      "grad_norm": 0.6796875,
      "learning_rate": 9.415368381528455e-08,
      "loss": 0.7499,
      "step": 9053
    },
    {
      "epoch": 4.860642260326834,
      "grad_norm": 0.703125,
      "learning_rate": 9.342285163734809e-08,
      "loss": 0.8332,
      "step": 9054
    },
    {
      "epoch": 4.861179155061911,
      "grad_norm": 0.68359375,
      "learning_rate": 9.269486161528063e-08,
      "loss": 0.8453,
      "step": 9055
    },
    {
      "epoch": 4.8617160497969865,
      "grad_norm": 0.6953125,
      "learning_rate": 9.196971383215458e-08,
      "loss": 0.8733,
      "step": 9056
    },
    {
      "epoch": 4.862252944532063,
      "grad_norm": 0.78515625,
      "learning_rate": 9.124740837072043e-08,
      "loss": 0.9948,
      "step": 9057
    },
    {
      "epoch": 4.862789839267139,
      "grad_norm": 0.80078125,
      "learning_rate": 9.052794531339837e-08,
      "loss": 0.8759,
      "step": 9058
    },
    {
      "epoch": 4.863326734002214,
      "grad_norm": 0.67578125,
      "learning_rate": 8.981132474229214e-08,
      "loss": 0.9871,
      "step": 9059
    },
    {
      "epoch": 4.863863628737291,
      "grad_norm": 0.7421875,
      "learning_rate": 8.909754673917525e-08,
      "loss": 1.0342,
      "step": 9060
    },
    {
      "epoch": 4.864400523472367,
      "grad_norm": 0.72265625,
      "learning_rate": 8.83866113854992e-08,
      "loss": 0.7909,
      "step": 9061
    },
    {
      "epoch": 4.864937418207443,
      "grad_norm": 0.765625,
      "learning_rate": 8.767851876239074e-08,
      "loss": 1.0984,
      "step": 9062
    },
    {
      "epoch": 4.865474312942519,
      "grad_norm": 0.73046875,
      "learning_rate": 8.697326895065472e-08,
      "loss": 0.8747,
      "step": 9063
    },
    {
      "epoch": 4.8660112076775945,
      "grad_norm": 0.60546875,
      "learning_rate": 8.627086203076284e-08,
      "loss": 0.8663,
      "step": 9064
    },
    {
      "epoch": 4.86654810241267,
      "grad_norm": 0.71875,
      "learning_rate": 8.557129808287601e-08,
      "loss": 1.0312,
      "step": 9065
    },
    {
      "epoch": 4.867084997147747,
      "grad_norm": 0.65625,
      "learning_rate": 8.487457718681924e-08,
      "loss": 0.7214,
      "step": 9066
    },
    {
      "epoch": 4.8676218918828225,
      "grad_norm": 0.69140625,
      "learning_rate": 8.418069942209561e-08,
      "loss": 0.8832,
      "step": 9067
    },
    {
      "epoch": 4.868158786617899,
      "grad_norm": 0.61328125,
      "learning_rate": 8.348966486788622e-08,
      "loss": 0.7356,
      "step": 9068
    },
    {
      "epoch": 4.868695681352975,
      "grad_norm": 0.76171875,
      "learning_rate": 8.280147360305302e-08,
      "loss": 0.8824,
      "step": 9069
    },
    {
      "epoch": 4.86923257608805,
      "grad_norm": 0.64453125,
      "learning_rate": 8.211612570611926e-08,
      "loss": 0.9492,
      "step": 9070
    },
    {
      "epoch": 4.869769470823127,
      "grad_norm": 0.65234375,
      "learning_rate": 8.143362125529463e-08,
      "loss": 0.7919,
      "step": 9071
    },
    {
      "epoch": 4.870306365558203,
      "grad_norm": 0.76953125,
      "learning_rate": 8.075396032846405e-08,
      "loss": 0.8273,
      "step": 9072
    },
    {
      "epoch": 4.870843260293279,
      "grad_norm": 0.7578125,
      "learning_rate": 8.007714300318214e-08,
      "loss": 0.7849,
      "step": 9073
    },
    {
      "epoch": 4.871380155028355,
      "grad_norm": 0.734375,
      "learning_rate": 7.940316935668435e-08,
      "loss": 0.8282,
      "step": 9074
    },
    {
      "epoch": 4.8719170497634305,
      "grad_norm": 0.69921875,
      "learning_rate": 7.873203946587581e-08,
      "loss": 0.7447,
      "step": 9075
    },
    {
      "epoch": 4.872453944498507,
      "grad_norm": 0.6953125,
      "learning_rate": 7.806375340734806e-08,
      "loss": 0.8362,
      "step": 9076
    },
    {
      "epoch": 4.872990839233583,
      "grad_norm": 0.77734375,
      "learning_rate": 7.739831125735397e-08,
      "loss": 1.0521,
      "step": 9077
    },
    {
      "epoch": 4.8735277339686585,
      "grad_norm": 0.87109375,
      "learning_rate": 7.673571309183281e-08,
      "loss": 0.8534,
      "step": 9078
    },
    {
      "epoch": 4.874064628703735,
      "grad_norm": 0.796875,
      "learning_rate": 7.607595898639352e-08,
      "loss": 0.8739,
      "step": 9079
    },
    {
      "epoch": 4.874601523438811,
      "grad_norm": 0.69140625,
      "learning_rate": 7.54190490163259e-08,
      "loss": 0.8838,
      "step": 9080
    },
    {
      "epoch": 4.875138418173886,
      "grad_norm": 0.55859375,
      "learning_rate": 7.476498325658665e-08,
      "loss": 0.6999,
      "step": 9081
    },
    {
      "epoch": 4.875675312908963,
      "grad_norm": 0.70703125,
      "learning_rate": 7.411376178181606e-08,
      "loss": 0.8838,
      "step": 9082
    },
    {
      "epoch": 4.876212207644039,
      "grad_norm": 0.94921875,
      "learning_rate": 7.346538466632414e-08,
      "loss": 0.9781,
      "step": 9083
    },
    {
      "epoch": 4.876749102379115,
      "grad_norm": 0.7109375,
      "learning_rate": 7.281985198410169e-08,
      "loss": 0.8947,
      "step": 9084
    },
    {
      "epoch": 4.877285997114191,
      "grad_norm": 0.65234375,
      "learning_rate": 7.217716380881479e-08,
      "loss": 0.8,
      "step": 9085
    },
    {
      "epoch": 4.8778228918492665,
      "grad_norm": 0.7265625,
      "learning_rate": 7.153732021379367e-08,
      "loss": 0.9429,
      "step": 9086
    },
    {
      "epoch": 4.878359786584343,
      "grad_norm": 0.62890625,
      "learning_rate": 7.090032127206048e-08,
      "loss": 0.7188,
      "step": 9087
    },
    {
      "epoch": 4.878896681319419,
      "grad_norm": 0.8203125,
      "learning_rate": 7.026616705630429e-08,
      "loss": 0.8974,
      "step": 9088
    },
    {
      "epoch": 4.8794335760544945,
      "grad_norm": 0.76171875,
      "learning_rate": 6.963485763888389e-08,
      "loss": 0.9295,
      "step": 9089
    },
    {
      "epoch": 4.879970470789571,
      "grad_norm": 0.94140625,
      "learning_rate": 6.90063930918472e-08,
      "loss": 1.2854,
      "step": 9090
    },
    {
      "epoch": 4.880507365524647,
      "grad_norm": 0.69140625,
      "learning_rate": 6.838077348690353e-08,
      "loss": 0.8567,
      "step": 9091
    },
    {
      "epoch": 4.881044260259722,
      "grad_norm": 0.7421875,
      "learning_rate": 6.775799889544854e-08,
      "loss": 0.8553,
      "step": 9092
    },
    {
      "epoch": 4.881581154994799,
      "grad_norm": 0.71875,
      "learning_rate": 6.71380693885476e-08,
      "loss": 0.8453,
      "step": 9093
    },
    {
      "epoch": 4.882118049729875,
      "grad_norm": 0.76171875,
      "learning_rate": 6.652098503694137e-08,
      "loss": 0.7966,
      "step": 9094
    },
    {
      "epoch": 4.882654944464951,
      "grad_norm": 0.5859375,
      "learning_rate": 6.59067459110485e-08,
      "loss": 0.7641,
      "step": 9095
    },
    {
      "epoch": 4.883191839200027,
      "grad_norm": 0.703125,
      "learning_rate": 6.529535208096016e-08,
      "loss": 1.0088,
      "step": 9096
    },
    {
      "epoch": 4.8837287339351025,
      "grad_norm": 0.76171875,
      "learning_rate": 6.468680361644275e-08,
      "loss": 0.7856,
      "step": 9097
    },
    {
      "epoch": 4.884265628670179,
      "grad_norm": 0.67578125,
      "learning_rate": 6.40811005869435e-08,
      "loss": 0.7847,
      "step": 9098
    },
    {
      "epoch": 4.884802523405255,
      "grad_norm": 0.66015625,
      "learning_rate": 6.347824306157935e-08,
      "loss": 0.6684,
      "step": 9099
    },
    {
      "epoch": 4.885339418140331,
      "grad_norm": 0.6796875,
      "learning_rate": 6.287823110914248e-08,
      "loss": 0.8787,
      "step": 9100
    },
    {
      "epoch": 4.885339418140331,
      "eval_loss": 0.90264892578125,
      "eval_runtime": 46.047,
      "eval_samples_per_second": 20.892,
      "eval_steps_per_second": 20.892,
      "step": 9100
    },
    {
      "epoch": 4.885876312875407,
      "grad_norm": 0.96875,
      "learning_rate": 6.228106479810036e-08,
      "loss": 1.0519,
      "step": 9101
    },
    {
      "epoch": 4.886413207610483,
      "grad_norm": 0.671875,
      "learning_rate": 6.168674419660126e-08,
      "loss": 0.9016,
      "step": 9102
    },
    {
      "epoch": 4.886950102345558,
      "grad_norm": 0.59375,
      "learning_rate": 6.109526937246313e-08,
      "loss": 0.6995,
      "step": 9103
    },
    {
      "epoch": 4.887486997080635,
      "grad_norm": 0.67578125,
      "learning_rate": 6.050664039317921e-08,
      "loss": 0.794,
      "step": 9104
    },
    {
      "epoch": 4.888023891815711,
      "grad_norm": 0.69140625,
      "learning_rate": 5.992085732592079e-08,
      "loss": 0.8779,
      "step": 9105
    },
    {
      "epoch": 4.888560786550787,
      "grad_norm": 0.70703125,
      "learning_rate": 5.9337920237534375e-08,
      "loss": 0.9777,
      "step": 9106
    },
    {
      "epoch": 4.889097681285863,
      "grad_norm": 0.75390625,
      "learning_rate": 5.875782919453898e-08,
      "loss": 0.7092,
      "step": 9107
    },
    {
      "epoch": 4.8896345760209385,
      "grad_norm": 0.70703125,
      "learning_rate": 5.818058426312889e-08,
      "loss": 0.769,
      "step": 9108
    },
    {
      "epoch": 4.890171470756015,
      "grad_norm": 0.78125,
      "learning_rate": 5.760618550917918e-08,
      "loss": 0.8875,
      "step": 9109
    },
    {
      "epoch": 4.890708365491091,
      "grad_norm": 0.68359375,
      "learning_rate": 5.7034632998231865e-08,
      "loss": 0.9944,
      "step": 9110
    },
    {
      "epoch": 4.891245260226167,
      "grad_norm": 0.75,
      "learning_rate": 5.646592679550977e-08,
      "loss": 0.8484,
      "step": 9111
    },
    {
      "epoch": 4.891782154961243,
      "grad_norm": 0.765625,
      "learning_rate": 5.590006696591099e-08,
      "loss": 0.8101,
      "step": 9112
    },
    {
      "epoch": 4.892319049696319,
      "grad_norm": 0.703125,
      "learning_rate": 5.533705357400609e-08,
      "loss": 0.9721,
      "step": 9113
    },
    {
      "epoch": 4.892855944431395,
      "grad_norm": 0.609375,
      "learning_rate": 5.477688668404091e-08,
      "loss": 0.8035,
      "step": 9114
    },
    {
      "epoch": 4.893392839166471,
      "grad_norm": 0.69140625,
      "learning_rate": 5.42195663599393e-08,
      "loss": 0.9419,
      "step": 9115
    },
    {
      "epoch": 4.893929733901547,
      "grad_norm": 0.69921875,
      "learning_rate": 5.366509266529762e-08,
      "loss": 0.8662,
      "step": 9116
    },
    {
      "epoch": 4.894466628636623,
      "grad_norm": 0.703125,
      "learning_rate": 5.311346566339026e-08,
      "loss": 0.8157,
      "step": 9117
    },
    {
      "epoch": 4.895003523371699,
      "grad_norm": 0.75,
      "learning_rate": 5.256468541716131e-08,
      "loss": 0.8522,
      "step": 9118
    },
    {
      "epoch": 4.8955404181067745,
      "grad_norm": 0.66796875,
      "learning_rate": 5.201875198923567e-08,
      "loss": 0.8041,
      "step": 9119
    },
    {
      "epoch": 4.896077312841851,
      "grad_norm": 0.6328125,
      "learning_rate": 5.1475665441910733e-08,
      "loss": 0.7414,
      "step": 9120
    },
    {
      "epoch": 4.896614207576927,
      "grad_norm": 0.625,
      "learning_rate": 5.0935425837159156e-08,
      "loss": 0.8075,
      "step": 9121
    },
    {
      "epoch": 4.897151102312003,
      "grad_norm": 0.734375,
      "learning_rate": 5.039803323663162e-08,
      "loss": 0.8519,
      "step": 9122
    },
    {
      "epoch": 4.897687997047079,
      "grad_norm": 0.67578125,
      "learning_rate": 4.986348770165128e-08,
      "loss": 0.8335,
      "step": 9123
    },
    {
      "epoch": 4.898224891782155,
      "grad_norm": 0.6484375,
      "learning_rate": 4.9331789293211026e-08,
      "loss": 0.679,
      "step": 9124
    },
    {
      "epoch": 4.898761786517231,
      "grad_norm": 0.68359375,
      "learning_rate": 4.88029380719901e-08,
      "loss": 0.7697,
      "step": 9125
    },
    {
      "epoch": 4.899298681252307,
      "grad_norm": 0.7421875,
      "learning_rate": 4.827693409833467e-08,
      "loss": 0.7572,
      "step": 9126
    },
    {
      "epoch": 4.899835575987383,
      "grad_norm": 0.66796875,
      "learning_rate": 4.775377743226894e-08,
      "loss": 0.8181,
      "step": 9127
    },
    {
      "epoch": 4.900372470722459,
      "grad_norm": 0.765625,
      "learning_rate": 4.7233468133492386e-08,
      "loss": 0.8985,
      "step": 9128
    },
    {
      "epoch": 4.900909365457535,
      "grad_norm": 0.6484375,
      "learning_rate": 4.671600626137973e-08,
      "loss": 0.7619,
      "step": 9129
    },
    {
      "epoch": 4.9014462601926105,
      "grad_norm": 0.703125,
      "learning_rate": 4.6201391874978186e-08,
      "loss": 0.8286,
      "step": 9130
    },
    {
      "epoch": 4.901983154927687,
      "grad_norm": 0.78125,
      "learning_rate": 4.568962503301022e-08,
      "loss": 0.9733,
      "step": 9131
    },
    {
      "epoch": 4.902520049662763,
      "grad_norm": 0.5859375,
      "learning_rate": 4.5180705793876344e-08,
      "loss": 0.7693,
      "step": 9132
    },
    {
      "epoch": 4.903056944397839,
      "grad_norm": 0.6015625,
      "learning_rate": 4.4674634215655095e-08,
      "loss": 0.8485,
      "step": 9133
    },
    {
      "epoch": 4.903593839132915,
      "grad_norm": 0.6484375,
      "learning_rate": 4.417141035608641e-08,
      "loss": 0.7162,
      "step": 9134
    },
    {
      "epoch": 4.904130733867991,
      "grad_norm": 0.81640625,
      "learning_rate": 4.367103427260211e-08,
      "loss": 0.9069,
      "step": 9135
    },
    {
      "epoch": 4.904667628603067,
      "grad_norm": 0.58203125,
      "learning_rate": 4.317350602230097e-08,
      "loss": 0.7779,
      "step": 9136
    },
    {
      "epoch": 4.905204523338143,
      "grad_norm": 0.82421875,
      "learning_rate": 4.2678825661954245e-08,
      "loss": 0.9377,
      "step": 9137
    },
    {
      "epoch": 4.9057414180732195,
      "grad_norm": 0.69921875,
      "learning_rate": 4.218699324801123e-08,
      "loss": 0.9546,
      "step": 9138
    },
    {
      "epoch": 4.906278312808295,
      "grad_norm": 0.75,
      "learning_rate": 4.1698008836599245e-08,
      "loss": 0.9133,
      "step": 9139
    },
    {
      "epoch": 4.906815207543371,
      "grad_norm": 0.68359375,
      "learning_rate": 4.121187248351532e-08,
      "loss": 0.837,
      "step": 9140
    },
    {
      "epoch": 4.9073521022784465,
      "grad_norm": 0.6015625,
      "learning_rate": 4.072858424423454e-08,
      "loss": 0.7976,
      "step": 9141
    },
    {
      "epoch": 4.907888997013523,
      "grad_norm": 0.7109375,
      "learning_rate": 4.024814417390443e-08,
      "loss": 0.8258,
      "step": 9142
    },
    {
      "epoch": 4.908425891748599,
      "grad_norm": 0.7578125,
      "learning_rate": 3.977055232735061e-08,
      "loss": 0.8559,
      "step": 9143
    },
    {
      "epoch": 4.908962786483675,
      "grad_norm": 1.0234375,
      "learning_rate": 3.9295808759073904e-08,
      "loss": 0.9074,
      "step": 9144
    },
    {
      "epoch": 4.909499681218751,
      "grad_norm": 0.8046875,
      "learning_rate": 3.8823913523247654e-08,
      "loss": 0.9378,
      "step": 9145
    },
    {
      "epoch": 4.910036575953827,
      "grad_norm": 0.63671875,
      "learning_rate": 3.835486667371768e-08,
      "loss": 0.8543,
      "step": 9146
    },
    {
      "epoch": 4.910573470688903,
      "grad_norm": 0.65625,
      "learning_rate": 3.788866826401338e-08,
      "loss": 0.7796,
      "step": 9147
    },
    {
      "epoch": 4.911110365423979,
      "grad_norm": 0.765625,
      "learning_rate": 3.7425318347331096e-08,
      "loss": 0.9348,
      "step": 9148
    },
    {
      "epoch": 4.9116472601590555,
      "grad_norm": 0.76953125,
      "learning_rate": 3.696481697654519e-08,
      "loss": 0.7457,
      "step": 9149
    },
    {
      "epoch": 4.912184154894131,
      "grad_norm": 0.7109375,
      "learning_rate": 3.65071642042053e-08,
      "loss": 0.9105,
      "step": 9150
    },
    {
      "epoch": 4.912184154894131,
      "eval_loss": 0.9025958776473999,
      "eval_runtime": 46.1145,
      "eval_samples_per_second": 20.861,
      "eval_steps_per_second": 20.861,
      "step": 9150
    },
    {
      "epoch": 4.912721049629207,
      "grad_norm": 0.66796875,
      "learning_rate": 3.605236008253632e-08,
      "loss": 0.8419,
      "step": 9151
    },
    {
      "epoch": 4.913257944364283,
      "grad_norm": 0.828125,
      "learning_rate": 3.5600404663435615e-08,
      "loss": 0.8061,
      "step": 9152
    },
    {
      "epoch": 4.913794839099359,
      "grad_norm": 0.6796875,
      "learning_rate": 3.515129799847583e-08,
      "loss": 0.8222,
      "step": 9153
    },
    {
      "epoch": 4.914331733834435,
      "grad_norm": 0.68359375,
      "learning_rate": 3.470504013890763e-08,
      "loss": 0.8127,
      "step": 9154
    },
    {
      "epoch": 4.914868628569511,
      "grad_norm": 0.6328125,
      "learning_rate": 3.426163113565417e-08,
      "loss": 0.8199,
      "step": 9155
    },
    {
      "epoch": 4.915405523304587,
      "grad_norm": 0.6484375,
      "learning_rate": 3.3821071039313865e-08,
      "loss": 0.7451,
      "step": 9156
    },
    {
      "epoch": 4.915942418039663,
      "grad_norm": 0.6171875,
      "learning_rate": 3.3383359900160396e-08,
      "loss": 0.8019,
      "step": 9157
    },
    {
      "epoch": 4.916479312774739,
      "grad_norm": 0.69921875,
      "learning_rate": 3.294849776814268e-08,
      "loss": 0.7966,
      "step": 9158
    },
    {
      "epoch": 4.917016207509815,
      "grad_norm": 0.76171875,
      "learning_rate": 3.2516484692882156e-08,
      "loss": 0.697,
      "step": 9159
    },
    {
      "epoch": 4.9175531022448915,
      "grad_norm": 0.609375,
      "learning_rate": 3.208732072368104e-08,
      "loss": 0.87,
      "step": 9160
    },
    {
      "epoch": 4.918089996979967,
      "grad_norm": 0.6953125,
      "learning_rate": 3.1661005909505716e-08,
      "loss": 0.8946,
      "step": 9161
    },
    {
      "epoch": 4.918626891715043,
      "grad_norm": 0.6796875,
      "learning_rate": 3.123754029901171e-08,
      "loss": 0.9249,
      "step": 9162
    },
    {
      "epoch": 4.919163786450119,
      "grad_norm": 0.5546875,
      "learning_rate": 3.081692394051594e-08,
      "loss": 0.7447,
      "step": 9163
    },
    {
      "epoch": 4.919700681185195,
      "grad_norm": 0.67578125,
      "learning_rate": 3.0399156882021665e-08,
      "loss": 0.6506,
      "step": 9164
    },
    {
      "epoch": 4.920237575920271,
      "grad_norm": 0.75390625,
      "learning_rate": 2.998423917119353e-08,
      "loss": 0.8776,
      "step": 9165
    },
    {
      "epoch": 4.920774470655347,
      "grad_norm": 0.7578125,
      "learning_rate": 2.9572170855385327e-08,
      "loss": 0.8856,
      "step": 9166
    },
    {
      "epoch": 4.921311365390423,
      "grad_norm": 0.6953125,
      "learning_rate": 2.916295198161778e-08,
      "loss": 0.9974,
      "step": 9167
    },
    {
      "epoch": 4.921848260125499,
      "grad_norm": 0.609375,
      "learning_rate": 2.8756582596586867e-08,
      "loss": 0.8075,
      "step": 9168
    },
    {
      "epoch": 4.922385154860575,
      "grad_norm": 0.73828125,
      "learning_rate": 2.8353062746663827e-08,
      "loss": 0.8311,
      "step": 9169
    },
    {
      "epoch": 4.922922049595651,
      "grad_norm": 0.70703125,
      "learning_rate": 2.7952392477900714e-08,
      "loss": 0.8493,
      "step": 9170
    },
    {
      "epoch": 4.9234589443307275,
      "grad_norm": 0.6796875,
      "learning_rate": 2.7554571836010957e-08,
      "loss": 0.9469,
      "step": 9171
    },
    {
      "epoch": 4.923995839065803,
      "grad_norm": 0.70703125,
      "learning_rate": 2.7159600866397127e-08,
      "loss": 0.9322,
      "step": 9172
    },
    {
      "epoch": 4.924532733800879,
      "grad_norm": 0.73046875,
      "learning_rate": 2.6767479614125957e-08,
      "loss": 0.9732,
      "step": 9173
    },
    {
      "epoch": 4.925069628535955,
      "grad_norm": 0.7109375,
      "learning_rate": 2.6378208123947757e-08,
      "loss": 0.8026,
      "step": 9174
    },
    {
      "epoch": 4.925606523271031,
      "grad_norm": 0.58203125,
      "learning_rate": 2.599178644028255e-08,
      "loss": 0.6899,
      "step": 9175
    },
    {
      "epoch": 4.926143418006108,
      "grad_norm": 0.796875,
      "learning_rate": 2.5608214607222847e-08,
      "loss": 0.7736,
      "step": 9176
    },
    {
      "epoch": 4.926680312741183,
      "grad_norm": 1.0625,
      "learning_rate": 2.5227492668541962e-08,
      "loss": 0.8652,
      "step": 9177
    },
    {
      "epoch": 4.927217207476259,
      "grad_norm": 0.59765625,
      "learning_rate": 2.4849620667682928e-08,
      "loss": 0.8428,
      "step": 9178
    },
    {
      "epoch": 4.927754102211335,
      "grad_norm": 0.7421875,
      "learning_rate": 2.44745986477668e-08,
      "loss": 0.7495,
      "step": 9179
    },
    {
      "epoch": 4.928290996946411,
      "grad_norm": 0.734375,
      "learning_rate": 2.4102426651589905e-08,
      "loss": 0.9329,
      "step": 9180
    },
    {
      "epoch": 4.928827891681487,
      "grad_norm": 0.6875,
      "learning_rate": 2.3733104721618273e-08,
      "loss": 0.8752,
      "step": 9181
    },
    {
      "epoch": 4.9293647864165635,
      "grad_norm": 0.6328125,
      "learning_rate": 2.3366632899998743e-08,
      "loss": 0.7079,
      "step": 9182
    },
    {
      "epoch": 4.929901681151639,
      "grad_norm": 0.6640625,
      "learning_rate": 2.3003011228550643e-08,
      "loss": 0.7176,
      "step": 9183
    },
    {
      "epoch": 4.930438575886715,
      "grad_norm": 0.62109375,
      "learning_rate": 2.2642239748765782e-08,
      "loss": 0.7696,
      "step": 9184
    },
    {
      "epoch": 4.930975470621791,
      "grad_norm": 0.77734375,
      "learning_rate": 2.228431850181678e-08,
      "loss": 0.887,
      "step": 9185
    },
    {
      "epoch": 4.931512365356867,
      "grad_norm": 0.65625,
      "learning_rate": 2.192924752854042e-08,
      "loss": 0.7757,
      "step": 9186
    },
    {
      "epoch": 4.932049260091944,
      "grad_norm": 0.64453125,
      "learning_rate": 2.157702686946261e-08,
      "loss": 0.9643,
      "step": 9187
    },
    {
      "epoch": 4.932586154827019,
      "grad_norm": 0.62109375,
      "learning_rate": 2.1227656564767883e-08,
      "loss": 0.891,
      "step": 9188
    },
    {
      "epoch": 4.933123049562095,
      "grad_norm": 0.83984375,
      "learning_rate": 2.0881136654332668e-08,
      "loss": 0.8697,
      "step": 9189
    },
    {
      "epoch": 4.933659944297172,
      "grad_norm": 0.71484375,
      "learning_rate": 2.0537467177692005e-08,
      "loss": 0.8539,
      "step": 9190
    },
    {
      "epoch": 4.934196839032247,
      "grad_norm": 0.70703125,
      "learning_rate": 2.0196648174064525e-08,
      "loss": 0.6448,
      "step": 9191
    },
    {
      "epoch": 4.934733733767323,
      "grad_norm": 0.72265625,
      "learning_rate": 1.9858679682346892e-08,
      "loss": 0.7918,
      "step": 9192
    },
    {
      "epoch": 4.9352706285023995,
      "grad_norm": 0.70703125,
      "learning_rate": 1.9523561741097153e-08,
      "loss": 0.7458,
      "step": 9193
    },
    {
      "epoch": 4.935807523237475,
      "grad_norm": 0.74609375,
      "learning_rate": 1.919129438856526e-08,
      "loss": 1.1106,
      "step": 9194
    },
    {
      "epoch": 4.936344417972551,
      "grad_norm": 0.8515625,
      "learning_rate": 1.886187766265979e-08,
      "loss": 0.8313,
      "step": 9195
    },
    {
      "epoch": 4.936881312707627,
      "grad_norm": 0.75,
      "learning_rate": 1.8535311600975656e-08,
      "loss": 1.0673,
      "step": 9196
    },
    {
      "epoch": 4.937418207442703,
      "grad_norm": 0.66015625,
      "learning_rate": 1.8211596240777508e-08,
      "loss": 0.8802,
      "step": 9197
    },
    {
      "epoch": 4.93795510217778,
      "grad_norm": 0.6875,
      "learning_rate": 1.789073161900523e-08,
      "loss": 0.8451,
      "step": 9198
    },
    {
      "epoch": 4.938491996912855,
      "grad_norm": 0.60546875,
      "learning_rate": 1.7572717772273982e-08,
      "loss": 0.699,
      "step": 9199
    },
    {
      "epoch": 4.939028891647931,
      "grad_norm": 0.79296875,
      "learning_rate": 1.7257554736871407e-08,
      "loss": 0.9295,
      "step": 9200
    },
    {
      "epoch": 4.939028891647931,
      "eval_loss": 0.9025484323501587,
      "eval_runtime": 46.1447,
      "eval_samples_per_second": 20.847,
      "eval_steps_per_second": 20.847,
      "step": 9200
    },
    {
      "epoch": 4.939565786383008,
      "grad_norm": 0.65625,
      "learning_rate": 1.6945242548763174e-08,
      "loss": 0.8636,
      "step": 9201
    },
    {
      "epoch": 4.940102681118083,
      "grad_norm": 0.65625,
      "learning_rate": 1.6635781243587446e-08,
      "loss": 0.7182,
      "step": 9202
    },
    {
      "epoch": 4.940639575853159,
      "grad_norm": 0.7265625,
      "learning_rate": 1.632917085665764e-08,
      "loss": 0.7932,
      "step": 9203
    },
    {
      "epoch": 4.9411764705882355,
      "grad_norm": 0.6875,
      "learning_rate": 1.6025411422962433e-08,
      "loss": 0.9285,
      "step": 9204
    },
    {
      "epoch": 4.941713365323311,
      "grad_norm": 0.5625,
      "learning_rate": 1.5724502977162993e-08,
      "loss": 0.6713,
      "step": 9205
    },
    {
      "epoch": 4.942250260058387,
      "grad_norm": 0.66796875,
      "learning_rate": 1.542644555359851e-08,
      "loss": 0.8203,
      "step": 9206
    },
    {
      "epoch": 4.942787154793463,
      "grad_norm": 0.76171875,
      "learning_rate": 1.5131239186280675e-08,
      "loss": 1.0477,
      "step": 9207
    },
    {
      "epoch": 4.943324049528539,
      "grad_norm": 0.6640625,
      "learning_rate": 1.4838883908896428e-08,
      "loss": 0.8604,
      "step": 9208
    },
    {
      "epoch": 4.943860944263616,
      "grad_norm": 0.75390625,
      "learning_rate": 1.4549379754807969e-08,
      "loss": 0.9156,
      "step": 9209
    },
    {
      "epoch": 4.944397838998691,
      "grad_norm": 0.7421875,
      "learning_rate": 1.4262726757049982e-08,
      "loss": 0.7788,
      "step": 9210
    },
    {
      "epoch": 4.944934733733767,
      "grad_norm": 0.796875,
      "learning_rate": 1.3978924948332416e-08,
      "loss": 0.8569,
      "step": 9211
    },
    {
      "epoch": 4.945471628468844,
      "grad_norm": 1.15625,
      "learning_rate": 1.3697974361043254e-08,
      "loss": 0.8195,
      "step": 9212
    },
    {
      "epoch": 4.946008523203919,
      "grad_norm": 0.671875,
      "learning_rate": 1.3419875027240181e-08,
      "loss": 0.6782,
      "step": 9213
    },
    {
      "epoch": 4.946545417938996,
      "grad_norm": 0.62890625,
      "learning_rate": 1.3144626978658925e-08,
      "loss": 0.9173,
      "step": 9214
    },
    {
      "epoch": 4.9470823126740715,
      "grad_norm": 0.640625,
      "learning_rate": 1.2872230246707695e-08,
      "loss": 0.8079,
      "step": 9215
    },
    {
      "epoch": 4.947619207409147,
      "grad_norm": 0.671875,
      "learning_rate": 1.2602684862472736e-08,
      "loss": 0.9495,
      "step": 9216
    },
    {
      "epoch": 4.948156102144223,
      "grad_norm": 0.55859375,
      "learning_rate": 1.233599085671e-08,
      "loss": 0.7799,
      "step": 9217
    },
    {
      "epoch": 4.948692996879299,
      "grad_norm": 0.6875,
      "learning_rate": 1.2072148259856253e-08,
      "loss": 0.7358,
      "step": 9218
    },
    {
      "epoch": 4.949229891614375,
      "grad_norm": 0.69140625,
      "learning_rate": 1.1811157102015192e-08,
      "loss": 0.9001,
      "step": 9219
    },
    {
      "epoch": 4.949766786349452,
      "grad_norm": 0.62890625,
      "learning_rate": 1.1553017412971323e-08,
      "loss": 0.7197,
      "step": 9220
    },
    {
      "epoch": 4.950303681084527,
      "grad_norm": 0.75390625,
      "learning_rate": 1.1297729222181642e-08,
      "loss": 0.8497,
      "step": 9221
    },
    {
      "epoch": 4.950840575819603,
      "grad_norm": 0.66015625,
      "learning_rate": 1.1045292558775622e-08,
      "loss": 0.7473,
      "step": 9222
    },
    {
      "epoch": 4.95137747055468,
      "grad_norm": 0.65234375,
      "learning_rate": 1.0795707451563553e-08,
      "loss": 0.8857,
      "step": 9223
    },
    {
      "epoch": 4.951914365289755,
      "grad_norm": 0.8671875,
      "learning_rate": 1.0548973929022655e-08,
      "loss": 0.9471,
      "step": 9224
    },
    {
      "epoch": 4.952451260024832,
      "grad_norm": 0.77734375,
      "learning_rate": 1.0305092019308182e-08,
      "loss": 0.8383,
      "step": 9225
    },
    {
      "epoch": 4.9529881547599075,
      "grad_norm": 0.78515625,
      "learning_rate": 1.0064061750253429e-08,
      "loss": 0.9087,
      "step": 9226
    },
    {
      "epoch": 4.953525049494983,
      "grad_norm": 0.6796875,
      "learning_rate": 9.825883149361392e-09,
      "loss": 0.7706,
      "step": 9227
    },
    {
      "epoch": 4.95406194423006,
      "grad_norm": 0.765625,
      "learning_rate": 9.590556243810333e-09,
      "loss": 0.8166,
      "step": 9228
    },
    {
      "epoch": 4.954598838965135,
      "grad_norm": 0.6484375,
      "learning_rate": 9.358081060456548e-09,
      "loss": 0.8657,
      "step": 9229
    },
    {
      "epoch": 4.955135733700211,
      "grad_norm": 0.640625,
      "learning_rate": 9.128457625823261e-09,
      "loss": 0.8099,
      "step": 9230
    },
    {
      "epoch": 4.955672628435288,
      "grad_norm": 0.61328125,
      "learning_rate": 8.901685966120065e-09,
      "loss": 0.7184,
      "step": 9231
    },
    {
      "epoch": 4.956209523170363,
      "grad_norm": 0.69140625,
      "learning_rate": 8.677766107217933e-09,
      "loss": 0.9066,
      "step": 9232
    },
    {
      "epoch": 4.956746417905439,
      "grad_norm": 0.76953125,
      "learning_rate": 8.456698074674197e-09,
      "loss": 0.9604,
      "step": 9233
    },
    {
      "epoch": 4.957283312640516,
      "grad_norm": 0.73828125,
      "learning_rate": 8.238481893713123e-09,
      "loss": 0.9137,
      "step": 9234
    },
    {
      "epoch": 4.957820207375591,
      "grad_norm": 0.77734375,
      "learning_rate": 8.023117589237017e-09,
      "loss": 0.8297,
      "step": 9235
    },
    {
      "epoch": 4.958357102110668,
      "grad_norm": 0.6015625,
      "learning_rate": 7.810605185820663e-09,
      "loss": 0.6425,
      "step": 9236
    },
    {
      "epoch": 4.9588939968457435,
      "grad_norm": 0.609375,
      "learning_rate": 7.600944707714108e-09,
      "loss": 0.7196,
      "step": 9237
    },
    {
      "epoch": 4.959430891580819,
      "grad_norm": 0.71484375,
      "learning_rate": 7.394136178842659e-09,
      "loss": 1.0134,
      "step": 9238
    },
    {
      "epoch": 4.959967786315896,
      "grad_norm": 0.71875,
      "learning_rate": 7.190179622806881e-09,
      "loss": 0.8386,
      "step": 9239
    },
    {
      "epoch": 4.960504681050971,
      "grad_norm": 0.62890625,
      "learning_rate": 6.989075062879824e-09,
      "loss": 0.6899,
      "step": 9240
    },
    {
      "epoch": 4.961041575786047,
      "grad_norm": 0.6796875,
      "learning_rate": 6.790822522009799e-09,
      "loss": 0.9326,
      "step": 9241
    },
    {
      "epoch": 4.961578470521124,
      "grad_norm": 0.6171875,
      "learning_rate": 6.595422022820375e-09,
      "loss": 0.8257,
      "step": 9242
    },
    {
      "epoch": 4.962115365256199,
      "grad_norm": 0.609375,
      "learning_rate": 6.402873587610381e-09,
      "loss": 0.9685,
      "step": 9243
    },
    {
      "epoch": 4.962652259991275,
      "grad_norm": 0.6796875,
      "learning_rate": 6.213177238348356e-09,
      "loss": 0.8452,
      "step": 9244
    },
    {
      "epoch": 4.963189154726352,
      "grad_norm": 0.79296875,
      "learning_rate": 6.0263329966836476e-09,
      "loss": 0.7881,
      "step": 9245
    },
    {
      "epoch": 4.963726049461427,
      "grad_norm": 0.68359375,
      "learning_rate": 5.842340883940866e-09,
      "loss": 0.8278,
      "step": 9246
    },
    {
      "epoch": 4.964262944196504,
      "grad_norm": 0.609375,
      "learning_rate": 5.661200921108778e-09,
      "loss": 0.7414,
      "step": 9247
    },
    {
      "epoch": 4.9647998389315795,
      "grad_norm": 0.68359375,
      "learning_rate": 5.48291312886251e-09,
      "loss": 0.6948,
      "step": 9248
    },
    {
      "epoch": 4.965336733666655,
      "grad_norm": 0.71484375,
      "learning_rate": 5.3074775275469e-09,
      "loss": 0.9773,
      "step": 9249
    },
    {
      "epoch": 4.965873628401732,
      "grad_norm": 0.87890625,
      "learning_rate": 5.134894137179269e-09,
      "loss": 1.0437,
      "step": 9250
    },
    {
      "epoch": 4.965873628401732,
      "eval_loss": 0.9026386737823486,
      "eval_runtime": 45.7381,
      "eval_samples_per_second": 21.033,
      "eval_steps_per_second": 21.033,
      "step": 9250
    },
    {
      "epoch": 4.966410523136807,
      "grad_norm": 0.7109375,
      "learning_rate": 4.965162977454973e-09,
      "loss": 0.7639,
      "step": 9251
    },
    {
      "epoch": 4.966947417871884,
      "grad_norm": 0.58203125,
      "learning_rate": 4.798284067741854e-09,
      "loss": 0.6897,
      "step": 9252
    },
    {
      "epoch": 4.96748431260696,
      "grad_norm": 0.75,
      "learning_rate": 4.6342574270830104e-09,
      "loss": 0.9268,
      "step": 9253
    },
    {
      "epoch": 4.968021207342035,
      "grad_norm": 0.69921875,
      "learning_rate": 4.4730830741940285e-09,
      "loss": 0.9176,
      "step": 9254
    },
    {
      "epoch": 4.968558102077111,
      "grad_norm": 0.80078125,
      "learning_rate": 4.3147610274713034e-09,
      "loss": 0.8222,
      "step": 9255
    },
    {
      "epoch": 4.969094996812188,
      "grad_norm": 0.75,
      "learning_rate": 4.159291304980939e-09,
      "loss": 0.7937,
      "step": 9256
    },
    {
      "epoch": 4.969631891547263,
      "grad_norm": 0.8046875,
      "learning_rate": 4.006673924458748e-09,
      "loss": 0.9635,
      "step": 9257
    },
    {
      "epoch": 4.97016878628234,
      "grad_norm": 0.69140625,
      "learning_rate": 3.8569089033241305e-09,
      "loss": 0.8897,
      "step": 9258
    },
    {
      "epoch": 4.9707056810174155,
      "grad_norm": 0.68359375,
      "learning_rate": 3.7099962586689685e-09,
      "loss": 1.0692,
      "step": 9259
    },
    {
      "epoch": 4.971242575752491,
      "grad_norm": 0.55078125,
      "learning_rate": 3.565936007254855e-09,
      "loss": 0.7195,
      "step": 9260
    },
    {
      "epoch": 4.971779470487568,
      "grad_norm": 0.71484375,
      "learning_rate": 3.424728165521418e-09,
      "loss": 0.8123,
      "step": 9261
    },
    {
      "epoch": 4.972316365222643,
      "grad_norm": 0.58984375,
      "learning_rate": 3.2863727495807685e-09,
      "loss": 0.6917,
      "step": 9262
    },
    {
      "epoch": 4.97285325995772,
      "grad_norm": 0.6640625,
      "learning_rate": 3.150869775225829e-09,
      "loss": 0.8037,
      "step": 9263
    },
    {
      "epoch": 4.973390154692796,
      "grad_norm": 0.66796875,
      "learning_rate": 3.0182192579136792e-09,
      "loss": 0.9472,
      "step": 9264
    },
    {
      "epoch": 4.973927049427871,
      "grad_norm": 0.69921875,
      "learning_rate": 2.8884212127849865e-09,
      "loss": 0.7353,
      "step": 9265
    },
    {
      "epoch": 4.974463944162948,
      "grad_norm": 0.62890625,
      "learning_rate": 2.7614756546501254e-09,
      "loss": 0.8754,
      "step": 9266
    },
    {
      "epoch": 4.975000838898024,
      "grad_norm": 0.703125,
      "learning_rate": 2.637382597994731e-09,
      "loss": 0.8499,
      "step": 9267
    },
    {
      "epoch": 4.975537733633099,
      "grad_norm": 0.625,
      "learning_rate": 2.5161420569796977e-09,
      "loss": 0.8057,
      "step": 9268
    },
    {
      "epoch": 4.976074628368176,
      "grad_norm": 0.66015625,
      "learning_rate": 2.3977540454411805e-09,
      "loss": 0.7986,
      "step": 9269
    },
    {
      "epoch": 4.9766115231032515,
      "grad_norm": 0.671875,
      "learning_rate": 2.282218576887818e-09,
      "loss": 0.7206,
      "step": 9270
    },
    {
      "epoch": 4.977148417838327,
      "grad_norm": 0.765625,
      "learning_rate": 2.169535664503508e-09,
      "loss": 0.9028,
      "step": 9271
    },
    {
      "epoch": 4.977685312573404,
      "grad_norm": 0.69140625,
      "learning_rate": 2.0597053211474095e-09,
      "loss": 0.9991,
      "step": 9272
    },
    {
      "epoch": 4.978222207308479,
      "grad_norm": 0.734375,
      "learning_rate": 1.9527275593539397e-09,
      "loss": 0.7342,
      "step": 9273
    },
    {
      "epoch": 4.978759102043556,
      "grad_norm": 0.72265625,
      "learning_rate": 1.8486023913272255e-09,
      "loss": 0.8345,
      "step": 9274
    },
    {
      "epoch": 4.979295996778632,
      "grad_norm": 0.58203125,
      "learning_rate": 1.7473298289494288e-09,
      "loss": 0.6468,
      "step": 9275
    },
    {
      "epoch": 4.979832891513707,
      "grad_norm": 0.6328125,
      "learning_rate": 1.648909883780747e-09,
      "loss": 0.6854,
      "step": 9276
    },
    {
      "epoch": 4.980369786248784,
      "grad_norm": 0.70703125,
      "learning_rate": 1.553342567051086e-09,
      "loss": 0.9647,
      "step": 9277
    },
    {
      "epoch": 4.98090668098386,
      "grad_norm": 0.62890625,
      "learning_rate": 1.4606278896628357e-09,
      "loss": 0.8783,
      "step": 9278
    },
    {
      "epoch": 4.981443575718935,
      "grad_norm": 0.6015625,
      "learning_rate": 1.3707658621964215e-09,
      "loss": 0.7503,
      "step": 9279
    },
    {
      "epoch": 4.981980470454012,
      "grad_norm": 0.8125,
      "learning_rate": 1.2837564949103042e-09,
      "loss": 0.9091,
      "step": 9280
    },
    {
      "epoch": 4.9825173651890875,
      "grad_norm": 0.8125,
      "learning_rate": 1.1995997977298779e-09,
      "loss": 0.876,
      "step": 9281
    },
    {
      "epoch": 4.983054259924163,
      "grad_norm": 0.68359375,
      "learning_rate": 1.1182957802585715e-09,
      "loss": 0.7893,
      "step": 9282
    },
    {
      "epoch": 4.98359115465924,
      "grad_norm": 0.62890625,
      "learning_rate": 1.039844451775074e-09,
      "loss": 0.6618,
      "step": 9283
    },
    {
      "epoch": 4.984128049394315,
      "grad_norm": 0.71875,
      "learning_rate": 9.642458212305584e-10,
      "loss": 0.7565,
      "step": 9284
    },
    {
      "epoch": 4.984664944129392,
      "grad_norm": 0.71875,
      "learning_rate": 8.914998972542332e-10,
      "loss": 0.8786,
      "step": 9285
    },
    {
      "epoch": 4.985201838864468,
      "grad_norm": 0.68359375,
      "learning_rate": 8.216066881450157e-10,
      "loss": 0.8447,
      "step": 9286
    },
    {
      "epoch": 4.985738733599543,
      "grad_norm": 0.578125,
      "learning_rate": 7.545662018826339e-10,
      "loss": 0.843,
      "step": 9287
    },
    {
      "epoch": 4.98627562833462,
      "grad_norm": 0.828125,
      "learning_rate": 6.90378446110973e-10,
      "loss": 1.0274,
      "step": 9288
    },
    {
      "epoch": 4.986812523069696,
      "grad_norm": 0.8515625,
      "learning_rate": 6.290434281575052e-10,
      "loss": 0.8809,
      "step": 9289
    },
    {
      "epoch": 4.987349417804772,
      "grad_norm": 0.8046875,
      "learning_rate": 5.705611550221868e-10,
      "loss": 0.8501,
      "step": 9290
    },
    {
      "epoch": 4.987886312539848,
      "grad_norm": 0.625,
      "learning_rate": 5.149316333802334e-10,
      "loss": 0.7096,
      "step": 9291
    },
    {
      "epoch": 4.9884232072749235,
      "grad_norm": 0.765625,
      "learning_rate": 4.621548695737943e-10,
      "loss": 0.8204,
      "step": 9292
    },
    {
      "epoch": 4.988960102009999,
      "grad_norm": 0.62109375,
      "learning_rate": 4.122308696313804e-10,
      "loss": 0.7726,
      "step": 9293
    },
    {
      "epoch": 4.989496996745076,
      "grad_norm": 0.671875,
      "learning_rate": 3.651596392456602e-10,
      "loss": 0.9576,
      "step": 9294
    },
    {
      "epoch": 4.990033891480151,
      "grad_norm": 0.73828125,
      "learning_rate": 3.2094118379288885e-10,
      "loss": 0.7951,
      "step": 9295
    },
    {
      "epoch": 4.990570786215228,
      "grad_norm": 0.828125,
      "learning_rate": 2.795755083134788e-10,
      "loss": 1.0182,
      "step": 9296
    },
    {
      "epoch": 4.991107680950304,
      "grad_norm": 0.765625,
      "learning_rate": 2.41062617531429e-10,
      "loss": 0.8384,
      "step": 9297
    },
    {
      "epoch": 4.991644575685379,
      "grad_norm": 0.8671875,
      "learning_rate": 2.054025158404471e-10,
      "loss": 0.8835,
      "step": 9298
    },
    {
      "epoch": 4.992181470420456,
      "grad_norm": 0.71484375,
      "learning_rate": 1.7259520730950052e-10,
      "loss": 0.8248,
      "step": 9299
    },
    {
      "epoch": 4.992718365155532,
      "grad_norm": 0.77734375,
      "learning_rate": 1.4264069568281635e-10,
      "loss": 0.9296,
      "step": 9300
    },
    {
      "epoch": 4.992718365155532,
      "eval_loss": 0.9026424288749695,
      "eval_runtime": 45.729,
      "eval_samples_per_second": 21.037,
      "eval_steps_per_second": 21.037,
      "step": 9300
    },
    {
      "epoch": 4.993255259890608,
      "grad_norm": 0.76171875,
      "learning_rate": 1.1553898437988154e-10,
      "loss": 1.0127,
      "step": 9301
    },
    {
      "epoch": 4.993792154625684,
      "grad_norm": 0.65234375,
      "learning_rate": 9.129007648989162e-11,
      "loss": 0.8544,
      "step": 9302
    },
    {
      "epoch": 4.9943290493607595,
      "grad_norm": 0.6953125,
      "learning_rate": 6.989397478285309e-11,
      "loss": 0.8514,
      "step": 9303
    },
    {
      "epoch": 4.994865944095836,
      "grad_norm": 0.86328125,
      "learning_rate": 5.1350681698481004e-11,
      "loss": 0.818,
      "step": 9304
    },
    {
      "epoch": 4.995402838830912,
      "grad_norm": 0.6640625,
      "learning_rate": 3.5660199354525806e-11,
      "loss": 0.734,
      "step": 9305
    },
    {
      "epoch": 4.995939733565987,
      "grad_norm": 0.64453125,
      "learning_rate": 2.2822529538446545e-11,
      "loss": 0.8164,
      "step": 9306
    },
    {
      "epoch": 4.996476628301064,
      "grad_norm": 0.6484375,
      "learning_rate": 1.2837673721288746e-11,
      "loss": 0.6163,
      "step": 9307
    },
    {
      "epoch": 4.99701352303614,
      "grad_norm": 0.62109375,
      "learning_rate": 5.705633035479885e-12,
      "loss": 0.7307,
      "step": 9308
    },
    {
      "epoch": 4.997550417771215,
      "grad_norm": 0.6953125,
      "learning_rate": 1.4264082998094451e-12,
      "loss": 0.786,
      "step": 9309
    },
    {
      "epoch": 4.998087312506292,
      "grad_norm": 0.75,
      "learning_rate": 0.0,
      "loss": 0.8572,
      "step": 9310
    }
  ],
  "logging_steps": 1,
  "max_steps": 9310,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 50,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5.079528727663903e+17,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}