{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 975,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.003078699249567058,
      "grad_norm": 12.21335978513995,
      "learning_rate": 0.0,
      "loss": 1.742,
      "step": 1
    },
    {
      "epoch": 0.006157398499134116,
      "grad_norm": 11.487754752029703,
      "learning_rate": 2.0408163265306121e-07,
      "loss": 1.6701,
      "step": 2
    },
    {
      "epoch": 0.009236097748701174,
      "grad_norm": 10.985667831752409,
      "learning_rate": 4.0816326530612243e-07,
      "loss": 1.6399,
      "step": 3
    },
    {
      "epoch": 0.012314796998268231,
      "grad_norm": 11.070868134463685,
      "learning_rate": 6.122448979591837e-07,
      "loss": 1.6554,
      "step": 4
    },
    {
      "epoch": 0.01539349624783529,
      "grad_norm": 10.020903811348754,
      "learning_rate": 8.163265306122449e-07,
      "loss": 1.6089,
      "step": 5
    },
    {
      "epoch": 0.01847219549740235,
      "grad_norm": 11.060664091957227,
      "learning_rate": 1.0204081632653063e-06,
      "loss": 1.6911,
      "step": 6
    },
    {
      "epoch": 0.021550894746969404,
      "grad_norm": 10.662749151275332,
      "learning_rate": 1.2244897959183673e-06,
      "loss": 1.6171,
      "step": 7
    },
    {
      "epoch": 0.024629593996536463,
      "grad_norm": 9.121885958913245,
      "learning_rate": 1.4285714285714286e-06,
      "loss": 1.5977,
      "step": 8
    },
    {
      "epoch": 0.02770829324610352,
      "grad_norm": 9.109818225412159,
      "learning_rate": 1.6326530612244897e-06,
      "loss": 1.6257,
      "step": 9
    },
    {
      "epoch": 0.03078699249567058,
      "grad_norm": 8.31708197448478,
      "learning_rate": 1.8367346938775512e-06,
      "loss": 1.563,
      "step": 10
    },
    {
      "epoch": 0.03386569174523764,
      "grad_norm": 6.191045733418002,
      "learning_rate": 2.0408163265306125e-06,
      "loss": 1.5794,
      "step": 11
    },
    {
      "epoch": 0.0369443909948047,
      "grad_norm": 5.6206186940093925,
      "learning_rate": 2.244897959183674e-06,
      "loss": 1.5264,
      "step": 12
    },
    {
      "epoch": 0.040023090244371756,
      "grad_norm": 5.126427975001733,
      "learning_rate": 2.4489795918367347e-06,
      "loss": 1.494,
      "step": 13
    },
    {
      "epoch": 0.04310178949393881,
      "grad_norm": 2.1767312289915077,
      "learning_rate": 2.6530612244897964e-06,
      "loss": 1.3521,
      "step": 14
    },
    {
      "epoch": 0.04618048874350587,
      "grad_norm": 2.029725595705277,
      "learning_rate": 2.8571428571428573e-06,
      "loss": 1.4186,
      "step": 15
    },
    {
      "epoch": 0.049259187993072925,
      "grad_norm": 1.8575122400264834,
      "learning_rate": 3.0612244897959185e-06,
      "loss": 1.4132,
      "step": 16
    },
    {
      "epoch": 0.052337887242639984,
      "grad_norm": 1.6538312300538953,
      "learning_rate": 3.2653061224489794e-06,
      "loss": 1.3887,
      "step": 17
    },
    {
      "epoch": 0.05541658649220704,
      "grad_norm": 1.4827160294661688,
      "learning_rate": 3.469387755102041e-06,
      "loss": 1.3592,
      "step": 18
    },
    {
      "epoch": 0.0584952857417741,
      "grad_norm": 1.8455684854938696,
      "learning_rate": 3.6734693877551024e-06,
      "loss": 1.3569,
      "step": 19
    },
    {
      "epoch": 0.06157398499134116,
      "grad_norm": 2.042472442283914,
      "learning_rate": 3.877551020408164e-06,
      "loss": 1.2896,
      "step": 20
    },
    {
      "epoch": 0.06465268424090821,
      "grad_norm": 2.069392948986389,
      "learning_rate": 4.081632653061225e-06,
      "loss": 1.3175,
      "step": 21
    },
    {
      "epoch": 0.06773138349047528,
      "grad_norm": 1.9510125615848375,
      "learning_rate": 4.2857142857142855e-06,
      "loss": 1.3455,
      "step": 22
    },
    {
      "epoch": 0.07081008274004233,
      "grad_norm": 1.5028946825375884,
      "learning_rate": 4.489795918367348e-06,
      "loss": 1.2891,
      "step": 23
    },
    {
      "epoch": 0.0738887819896094,
      "grad_norm": 1.2813404851651375,
      "learning_rate": 4.693877551020409e-06,
      "loss": 1.3129,
      "step": 24
    },
    {
      "epoch": 0.07696748123917645,
      "grad_norm": 1.2931702488069334,
      "learning_rate": 4.897959183673469e-06,
      "loss": 1.2935,
      "step": 25
    },
    {
      "epoch": 0.08004618048874351,
      "grad_norm": 0.7526395622678241,
      "learning_rate": 5.1020408163265315e-06,
      "loss": 1.2579,
      "step": 26
    },
    {
      "epoch": 0.08312487973831056,
      "grad_norm": 0.7721573799461998,
      "learning_rate": 5.306122448979593e-06,
      "loss": 1.2001,
      "step": 27
    },
    {
      "epoch": 0.08620357898787762,
      "grad_norm": 0.8570440392543626,
      "learning_rate": 5.510204081632653e-06,
      "loss": 1.2091,
      "step": 28
    },
    {
      "epoch": 0.08928227823744468,
      "grad_norm": 0.8320279856338966,
      "learning_rate": 5.7142857142857145e-06,
      "loss": 1.1586,
      "step": 29
    },
    {
      "epoch": 0.09236097748701173,
      "grad_norm": 0.7136367487242015,
      "learning_rate": 5.918367346938776e-06,
      "loss": 1.171,
      "step": 30
    },
    {
      "epoch": 0.0954396767365788,
      "grad_norm": 0.6596186291083772,
      "learning_rate": 6.122448979591837e-06,
      "loss": 1.1825,
      "step": 31
    },
    {
      "epoch": 0.09851837598614585,
      "grad_norm": 0.5457237344846158,
      "learning_rate": 6.326530612244899e-06,
      "loss": 1.1424,
      "step": 32
    },
    {
      "epoch": 0.10159707523571292,
      "grad_norm": 0.513191911660899,
      "learning_rate": 6.530612244897959e-06,
      "loss": 1.1316,
      "step": 33
    },
    {
      "epoch": 0.10467577448527997,
      "grad_norm": 0.5176786732904974,
      "learning_rate": 6.734693877551021e-06,
      "loss": 1.1343,
      "step": 34
    },
    {
      "epoch": 0.10775447373484703,
      "grad_norm": 0.5276566764778993,
      "learning_rate": 6.938775510204082e-06,
      "loss": 1.1526,
      "step": 35
    },
    {
      "epoch": 0.11083317298441409,
      "grad_norm": 0.5047836589862809,
      "learning_rate": 7.1428571428571436e-06,
      "loss": 1.1388,
      "step": 36
    },
    {
      "epoch": 0.11391187223398114,
      "grad_norm": 0.6153214921651887,
      "learning_rate": 7.346938775510205e-06,
      "loss": 1.103,
      "step": 37
    },
    {
      "epoch": 0.1169905714835482,
      "grad_norm": 0.4276045007808924,
      "learning_rate": 7.551020408163265e-06,
      "loss": 1.103,
      "step": 38
    },
    {
      "epoch": 0.12006927073311525,
      "grad_norm": 0.414768944541662,
      "learning_rate": 7.755102040816327e-06,
      "loss": 1.1076,
      "step": 39
    },
    {
      "epoch": 0.12314796998268232,
      "grad_norm": 0.428396745469851,
      "learning_rate": 7.959183673469388e-06,
      "loss": 1.0977,
      "step": 40
    },
    {
      "epoch": 0.12622666923224937,
      "grad_norm": 0.42945055407862726,
      "learning_rate": 8.16326530612245e-06,
      "loss": 1.1299,
      "step": 41
    },
    {
      "epoch": 0.12930536848181642,
      "grad_norm": 0.4546450834675262,
      "learning_rate": 8.36734693877551e-06,
      "loss": 1.0616,
      "step": 42
    },
    {
      "epoch": 0.1323840677313835,
      "grad_norm": 0.4127182777105827,
      "learning_rate": 8.571428571428571e-06,
      "loss": 1.1026,
      "step": 43
    },
    {
      "epoch": 0.13546276698095056,
      "grad_norm": 0.39650154404932086,
      "learning_rate": 8.775510204081633e-06,
      "loss": 1.1123,
      "step": 44
    },
    {
      "epoch": 0.1385414662305176,
      "grad_norm": 0.37809595942229435,
      "learning_rate": 8.979591836734695e-06,
      "loss": 1.1088,
      "step": 45
    },
    {
      "epoch": 0.14162016548008466,
      "grad_norm": 0.37819360072348407,
      "learning_rate": 9.183673469387756e-06,
      "loss": 1.119,
      "step": 46
    },
    {
      "epoch": 0.1446988647296517,
      "grad_norm": 0.34738091590119174,
      "learning_rate": 9.387755102040818e-06,
      "loss": 1.0717,
      "step": 47
    },
    {
      "epoch": 0.1477775639792188,
      "grad_norm": 0.35687807645113984,
      "learning_rate": 9.591836734693878e-06,
      "loss": 1.0717,
      "step": 48
    },
    {
      "epoch": 0.15085626322878584,
      "grad_norm": 0.3437835723834522,
      "learning_rate": 9.795918367346939e-06,
      "loss": 1.0548,
      "step": 49
    },
    {
      "epoch": 0.1539349624783529,
      "grad_norm": 0.3575384834131811,
      "learning_rate": 1e-05,
      "loss": 1.1032,
      "step": 50
    },
    {
      "epoch": 0.15701366172791995,
      "grad_norm": 0.38465285158381735,
      "learning_rate": 1.0204081632653063e-05,
      "loss": 1.0226,
      "step": 51
    },
    {
      "epoch": 0.16009236097748702,
      "grad_norm": 0.36157755266782327,
      "learning_rate": 1.0408163265306123e-05,
      "loss": 1.1004,
      "step": 52
    },
    {
      "epoch": 0.16317106022705408,
      "grad_norm": 0.3425589474781937,
      "learning_rate": 1.0612244897959186e-05,
      "loss": 1.1298,
      "step": 53
    },
    {
      "epoch": 0.16624975947662113,
      "grad_norm": 0.3771875566021622,
      "learning_rate": 1.0816326530612246e-05,
      "loss": 1.0878,
      "step": 54
    },
    {
      "epoch": 0.16932845872618818,
      "grad_norm": 0.3465546601916469,
      "learning_rate": 1.1020408163265306e-05,
      "loss": 1.078,
      "step": 55
    },
    {
      "epoch": 0.17240715797575523,
      "grad_norm": 0.33832375827267613,
      "learning_rate": 1.1224489795918367e-05,
      "loss": 1.0577,
      "step": 56
    },
    {
      "epoch": 0.1754858572253223,
      "grad_norm": 0.35564682947195225,
      "learning_rate": 1.1428571428571429e-05,
      "loss": 1.0382,
      "step": 57
    },
    {
      "epoch": 0.17856455647488936,
      "grad_norm": 0.3303690639963582,
      "learning_rate": 1.1632653061224491e-05,
      "loss": 1.0543,
      "step": 58
    },
    {
      "epoch": 0.18164325572445641,
      "grad_norm": 0.3761166378029424,
      "learning_rate": 1.1836734693877552e-05,
      "loss": 1.0341,
      "step": 59
    },
    {
      "epoch": 0.18472195497402347,
      "grad_norm": 0.3374489204260131,
      "learning_rate": 1.2040816326530614e-05,
      "loss": 1.0833,
      "step": 60
    },
    {
      "epoch": 0.18780065422359055,
      "grad_norm": 0.3092112888573424,
      "learning_rate": 1.2244897959183674e-05,
      "loss": 1.0424,
      "step": 61
    },
    {
      "epoch": 0.1908793534731576,
      "grad_norm": 0.3260065569325197,
      "learning_rate": 1.2448979591836736e-05,
      "loss": 1.0345,
      "step": 62
    },
    {
      "epoch": 0.19395805272272465,
      "grad_norm": 0.5818102389477204,
      "learning_rate": 1.2653061224489798e-05,
      "loss": 1.0365,
      "step": 63
    },
    {
      "epoch": 0.1970367519722917,
      "grad_norm": 0.32665928843927566,
      "learning_rate": 1.2857142857142859e-05,
      "loss": 1.0198,
      "step": 64
    },
    {
      "epoch": 0.20011545122185875,
      "grad_norm": 0.3307786056400255,
      "learning_rate": 1.3061224489795918e-05,
      "loss": 0.985,
      "step": 65
    },
    {
      "epoch": 0.20319415047142583,
      "grad_norm": 0.33227834487200975,
      "learning_rate": 1.326530612244898e-05,
      "loss": 1.0143,
      "step": 66
    },
    {
      "epoch": 0.20627284972099288,
      "grad_norm": 0.3181868348939409,
      "learning_rate": 1.3469387755102042e-05,
      "loss": 1.0422,
      "step": 67
    },
    {
      "epoch": 0.20935154897055994,
      "grad_norm": 0.33251459358403723,
      "learning_rate": 1.3673469387755102e-05,
      "loss": 1.0539,
      "step": 68
    },
    {
      "epoch": 0.212430248220127,
      "grad_norm": 0.322302267670571,
      "learning_rate": 1.3877551020408165e-05,
      "loss": 0.99,
      "step": 69
    },
    {
      "epoch": 0.21550894746969407,
      "grad_norm": 0.33698378548397045,
      "learning_rate": 1.4081632653061225e-05,
      "loss": 1.0656,
      "step": 70
    },
    {
      "epoch": 0.21858764671926112,
      "grad_norm": 0.3136413744451391,
      "learning_rate": 1.4285714285714287e-05,
      "loss": 1.0022,
      "step": 71
    },
    {
      "epoch": 0.22166634596882817,
      "grad_norm": 0.31845872407754233,
      "learning_rate": 1.448979591836735e-05,
      "loss": 1.0052,
      "step": 72
    },
    {
      "epoch": 0.22474504521839522,
      "grad_norm": 0.3430141418679699,
      "learning_rate": 1.469387755102041e-05,
      "loss": 1.0515,
      "step": 73
    },
    {
      "epoch": 0.22782374446796227,
      "grad_norm": 0.3293185169767395,
      "learning_rate": 1.4897959183673472e-05,
      "loss": 1.0167,
      "step": 74
    },
    {
      "epoch": 0.23090244371752935,
      "grad_norm": 0.3377340815166731,
      "learning_rate": 1.510204081632653e-05,
      "loss": 1.0351,
      "step": 75
    },
    {
      "epoch": 0.2339811429670964,
      "grad_norm": 0.34199207352026373,
      "learning_rate": 1.530612244897959e-05,
      "loss": 1.0224,
      "step": 76
    },
    {
      "epoch": 0.23705984221666346,
      "grad_norm": 0.33648222342610135,
      "learning_rate": 1.5510204081632655e-05,
      "loss": 1.0301,
      "step": 77
    },
    {
      "epoch": 0.2401385414662305,
      "grad_norm": 0.3375077367750986,
      "learning_rate": 1.5714285714285715e-05,
      "loss": 0.9986,
      "step": 78
    },
    {
      "epoch": 0.2432172407157976,
      "grad_norm": 0.35884531384675983,
      "learning_rate": 1.5918367346938776e-05,
      "loss": 1.0351,
      "step": 79
    },
    {
      "epoch": 0.24629593996536464,
      "grad_norm": 0.3246959610386908,
      "learning_rate": 1.612244897959184e-05,
      "loss": 1.0152,
      "step": 80
    },
    {
      "epoch": 0.2493746392149317,
      "grad_norm": 0.3342862615303175,
      "learning_rate": 1.63265306122449e-05,
      "loss": 0.9859,
      "step": 81
    },
    {
      "epoch": 0.25245333846449874,
      "grad_norm": 0.3341410986256379,
      "learning_rate": 1.653061224489796e-05,
      "loss": 1.0239,
      "step": 82
    },
    {
      "epoch": 0.2555320377140658,
      "grad_norm": 0.33290200606516884,
      "learning_rate": 1.673469387755102e-05,
      "loss": 1.0223,
      "step": 83
    },
    {
      "epoch": 0.25861073696363285,
      "grad_norm": 0.33164027510097127,
      "learning_rate": 1.6938775510204085e-05,
      "loss": 1.023,
      "step": 84
    },
    {
      "epoch": 0.2616894362131999,
      "grad_norm": 0.3297175698049479,
      "learning_rate": 1.7142857142857142e-05,
      "loss": 1.0094,
      "step": 85
    },
    {
      "epoch": 0.264768135462767,
      "grad_norm": 0.32959398888036584,
      "learning_rate": 1.7346938775510206e-05,
      "loss": 0.9919,
      "step": 86
    },
    {
      "epoch": 0.26784683471233406,
      "grad_norm": 0.33581141821980925,
      "learning_rate": 1.7551020408163266e-05,
      "loss": 0.9917,
      "step": 87
    },
    {
      "epoch": 0.2709255339619011,
      "grad_norm": 0.3310474384138704,
      "learning_rate": 1.7755102040816327e-05,
      "loss": 1.02,
      "step": 88
    },
    {
      "epoch": 0.27400423321146816,
      "grad_norm": 0.3483681310567162,
      "learning_rate": 1.795918367346939e-05,
      "loss": 1.026,
      "step": 89
    },
    {
      "epoch": 0.2770829324610352,
      "grad_norm": 0.3230473360097209,
      "learning_rate": 1.816326530612245e-05,
      "loss": 0.9815,
      "step": 90
    },
    {
      "epoch": 0.28016163171060227,
      "grad_norm": 0.35360057167090125,
      "learning_rate": 1.836734693877551e-05,
      "loss": 1.0201,
      "step": 91
    },
    {
      "epoch": 0.2832403309601693,
      "grad_norm": 0.3262450584358491,
      "learning_rate": 1.8571428571428575e-05,
      "loss": 0.9889,
      "step": 92
    },
    {
      "epoch": 0.28631903020973637,
      "grad_norm": 0.3467496142607115,
      "learning_rate": 1.8775510204081636e-05,
      "loss": 1.0123,
      "step": 93
    },
    {
      "epoch": 0.2893977294593034,
      "grad_norm": 0.3503238310844547,
      "learning_rate": 1.8979591836734696e-05,
      "loss": 1.0132,
      "step": 94
    },
    {
      "epoch": 0.29247642870887053,
      "grad_norm": 0.3478178973594713,
      "learning_rate": 1.9183673469387756e-05,
      "loss": 0.9742,
      "step": 95
    },
    {
      "epoch": 0.2955551279584376,
      "grad_norm": 0.38303095818781363,
      "learning_rate": 1.9387755102040817e-05,
      "loss": 0.9535,
      "step": 96
    },
    {
      "epoch": 0.29863382720800463,
      "grad_norm": 0.4200945363234874,
      "learning_rate": 1.9591836734693877e-05,
      "loss": 1.0235,
      "step": 97
    },
    {
      "epoch": 0.3017125264575717,
      "grad_norm": 0.3355677653267444,
      "learning_rate": 1.979591836734694e-05,
      "loss": 0.9718,
      "step": 98
    },
    {
      "epoch": 0.30479122570713874,
      "grad_norm": 0.34650608014923445,
      "learning_rate": 2e-05,
      "loss": 1.0429,
      "step": 99
    },
    {
      "epoch": 0.3078699249567058,
      "grad_norm": 0.37036783747791796,
      "learning_rate": 1.9999935839151513e-05,
      "loss": 1.0482,
      "step": 100
    },
    {
      "epoch": 0.31094862420627284,
      "grad_norm": 0.3386846939054312,
      "learning_rate": 1.999974335742938e-05,
      "loss": 0.9951,
      "step": 101
    },
    {
      "epoch": 0.3140273234558399,
      "grad_norm": 0.33829797518246174,
      "learning_rate": 1.9999422557303553e-05,
      "loss": 0.973,
      "step": 102
    },
    {
      "epoch": 0.31710602270540694,
      "grad_norm": 0.3463319150701838,
      "learning_rate": 1.99989734428906e-05,
      "loss": 0.9935,
      "step": 103
    },
    {
      "epoch": 0.32018472195497405,
      "grad_norm": 0.329294144684054,
      "learning_rate": 1.9998396019953627e-05,
      "loss": 0.9831,
      "step": 104
    },
    {
      "epoch": 0.3232634212045411,
      "grad_norm": 0.3654390033358469,
      "learning_rate": 1.9997690295902225e-05,
      "loss": 0.96,
      "step": 105
    },
    {
      "epoch": 0.32634212045410815,
      "grad_norm": 0.3453128729243265,
      "learning_rate": 1.9996856279792368e-05,
      "loss": 1.0098,
      "step": 106
    },
    {
      "epoch": 0.3294208197036752,
      "grad_norm": 0.3375953207038091,
      "learning_rate": 1.9995893982326288e-05,
      "loss": 0.9629,
      "step": 107
    },
    {
      "epoch": 0.33249951895324226,
      "grad_norm": 0.34579836946401155,
      "learning_rate": 1.9994803415852352e-05,
      "loss": 0.9722,
      "step": 108
    },
    {
      "epoch": 0.3355782182028093,
      "grad_norm": 0.36907117226806796,
      "learning_rate": 1.9993584594364893e-05,
      "loss": 1.0272,
      "step": 109
    },
    {
      "epoch": 0.33865691745237636,
      "grad_norm": 0.36188004391911355,
      "learning_rate": 1.9992237533504037e-05,
      "loss": 0.9657,
      "step": 110
    },
    {
      "epoch": 0.3417356167019434,
      "grad_norm": 0.36508432395585566,
      "learning_rate": 1.9990762250555495e-05,
      "loss": 0.9919,
      "step": 111
    },
    {
      "epoch": 0.34481431595151046,
      "grad_norm": 0.335720924174464,
      "learning_rate": 1.9989158764450352e-05,
      "loss": 0.9439,
      "step": 112
    },
    {
      "epoch": 0.34789301520107757,
      "grad_norm": 0.36557291734926056,
      "learning_rate": 1.9987427095764808e-05,
      "loss": 0.9641,
      "step": 113
    },
    {
      "epoch": 0.3509717144506446,
      "grad_norm": 0.3304611778772956,
      "learning_rate": 1.9985567266719935e-05,
      "loss": 0.9784,
      "step": 114
    },
    {
      "epoch": 0.3540504137002117,
      "grad_norm": 0.3455095590721512,
      "learning_rate": 1.9983579301181374e-05,
      "loss": 0.9818,
      "step": 115
    },
    {
      "epoch": 0.3571291129497787,
      "grad_norm": 0.3358951815209104,
      "learning_rate": 1.9981463224659034e-05,
      "loss": 1.0008,
      "step": 116
    },
    {
      "epoch": 0.3602078121993458,
      "grad_norm": 0.3262977559593158,
      "learning_rate": 1.9979219064306764e-05,
      "loss": 0.9734,
      "step": 117
    },
    {
      "epoch": 0.36328651144891283,
      "grad_norm": 0.35678765436333343,
      "learning_rate": 1.9976846848922018e-05,
      "loss": 0.9809,
      "step": 118
    },
    {
      "epoch": 0.3663652106984799,
      "grad_norm": 0.3443393589809869,
      "learning_rate": 1.9974346608945466e-05,
      "loss": 0.9514,
      "step": 119
    },
    {
      "epoch": 0.36944390994804693,
      "grad_norm": 0.34490042723445796,
      "learning_rate": 1.9971718376460605e-05,
      "loss": 0.9665,
      "step": 120
    },
    {
      "epoch": 0.372522609197614,
      "grad_norm": 0.35223949033287877,
      "learning_rate": 1.9968962185193367e-05,
      "loss": 1.0105,
      "step": 121
    },
    {
      "epoch": 0.3756013084471811,
      "grad_norm": 0.34453558306444904,
      "learning_rate": 1.9966078070511663e-05,
      "loss": 0.943,
      "step": 122
    },
    {
      "epoch": 0.37868000769674814,
      "grad_norm": 0.3374675056941535,
      "learning_rate": 1.9963066069424944e-05,
      "loss": 0.9443,
      "step": 123
    },
    {
      "epoch": 0.3817587069463152,
      "grad_norm": 0.34242762321779807,
      "learning_rate": 1.9959926220583713e-05,
      "loss": 0.9575,
      "step": 124
    },
    {
      "epoch": 0.38483740619588225,
      "grad_norm": 0.3449443618024186,
      "learning_rate": 1.995665856427905e-05,
      "loss": 0.9726,
      "step": 125
    },
    {
      "epoch": 0.3879161054454493,
      "grad_norm": 0.34486894327315176,
      "learning_rate": 1.9953263142442078e-05,
      "loss": 0.9883,
      "step": 126
    },
    {
      "epoch": 0.39099480469501635,
      "grad_norm": 0.3496265922220688,
      "learning_rate": 1.9949739998643415e-05,
      "loss": 1.0179,
      "step": 127
    },
    {
      "epoch": 0.3940735039445834,
      "grad_norm": 0.328111871914487,
      "learning_rate": 1.994608917809265e-05,
      "loss": 0.9683,
      "step": 128
    },
    {
      "epoch": 0.39715220319415045,
      "grad_norm": 0.3692713615995683,
      "learning_rate": 1.9942310727637725e-05,
      "loss": 0.9869,
      "step": 129
    },
    {
      "epoch": 0.4002309024437175,
      "grad_norm": 0.32990543192027705,
      "learning_rate": 1.9938404695764362e-05,
      "loss": 0.9473,
      "step": 130
    },
    {
      "epoch": 0.4033096016932846,
      "grad_norm": 0.33640315535500387,
      "learning_rate": 1.9934371132595426e-05,
      "loss": 0.9186,
      "step": 131
    },
    {
      "epoch": 0.40638830094285167,
      "grad_norm": 0.5706199509902835,
      "learning_rate": 1.9930210089890276e-05,
      "loss": 0.9641,
      "step": 132
    },
    {
      "epoch": 0.4094670001924187,
      "grad_norm": 0.3611669123633751,
      "learning_rate": 1.992592162104413e-05,
      "loss": 0.9696,
      "step": 133
    },
    {
      "epoch": 0.41254569944198577,
      "grad_norm": 0.34723353803054974,
      "learning_rate": 1.9921505781087335e-05,
      "loss": 0.9798,
      "step": 134
    },
    {
      "epoch": 0.4156243986915528,
      "grad_norm": 0.3506959175663828,
      "learning_rate": 1.9916962626684707e-05,
      "loss": 0.9792,
      "step": 135
    },
    {
      "epoch": 0.4187030979411199,
      "grad_norm": 0.34634241710205943,
      "learning_rate": 1.9912292216134775e-05,
      "loss": 0.9738,
      "step": 136
    },
    {
      "epoch": 0.4217817971906869,
      "grad_norm": 0.34697995488537503,
      "learning_rate": 1.9907494609369037e-05,
      "loss": 0.9404,
      "step": 137
    },
    {
      "epoch": 0.424860496440254,
      "grad_norm": 0.35421106326145807,
      "learning_rate": 1.9902569867951195e-05,
      "loss": 1.0471,
      "step": 138
    },
    {
      "epoch": 0.42793919568982103,
      "grad_norm": 0.33596127901805994,
      "learning_rate": 1.9897518055076372e-05,
      "loss": 0.958,
      "step": 139
    },
    {
      "epoch": 0.43101789493938814,
      "grad_norm": 0.379813761830078,
      "learning_rate": 1.9892339235570284e-05,
      "loss": 0.9877,
      "step": 140
    },
    {
      "epoch": 0.4340965941889552,
      "grad_norm": 0.33390757430450047,
      "learning_rate": 1.988703347588842e-05,
      "loss": 0.9648,
      "step": 141
    },
    {
      "epoch": 0.43717529343852224,
      "grad_norm": 0.3503814308635588,
      "learning_rate": 1.9881600844115193e-05,
      "loss": 0.9602,
      "step": 142
    },
    {
      "epoch": 0.4402539926880893,
      "grad_norm": 0.3171186799112764,
      "learning_rate": 1.9876041409963055e-05,
      "loss": 0.9382,
      "step": 143
    },
    {
      "epoch": 0.44333269193765634,
      "grad_norm": 0.3850355954449746,
      "learning_rate": 1.987035524477161e-05,
      "loss": 0.9532,
      "step": 144
    },
    {
      "epoch": 0.4464113911872234,
      "grad_norm": 0.34269185461875096,
      "learning_rate": 1.986454242150669e-05,
      "loss": 0.9445,
      "step": 145
    },
    {
      "epoch": 0.44949009043679045,
      "grad_norm": 0.34433819119728326,
      "learning_rate": 1.985860301475943e-05,
      "loss": 0.9511,
      "step": 146
    },
    {
      "epoch": 0.4525687896863575,
      "grad_norm": 0.35163464443256537,
      "learning_rate": 1.9852537100745307e-05,
      "loss": 0.9823,
      "step": 147
    },
    {
      "epoch": 0.45564748893592455,
      "grad_norm": 0.3446603644019156,
      "learning_rate": 1.9846344757303163e-05,
      "loss": 0.9249,
      "step": 148
    },
    {
      "epoch": 0.45872618818549166,
      "grad_norm": 0.3696522189978251,
      "learning_rate": 1.9840026063894193e-05,
      "loss": 0.9712,
      "step": 149
    },
    {
      "epoch": 0.4618048874350587,
      "grad_norm": 0.3782282458104307,
      "learning_rate": 1.983358110160095e-05,
      "loss": 1.0061,
      "step": 150
    },
    {
      "epoch": 0.46488358668462576,
      "grad_norm": 0.36319167528571833,
      "learning_rate": 1.9827009953126277e-05,
      "loss": 0.939,
      "step": 151
    },
    {
      "epoch": 0.4679622859341928,
      "grad_norm": 0.3375057228291175,
      "learning_rate": 1.982031270279227e-05,
      "loss": 0.9074,
      "step": 152
    },
    {
      "epoch": 0.47104098518375986,
      "grad_norm": 0.43941398797936343,
      "learning_rate": 1.981348943653918e-05,
      "loss": 0.9972,
      "step": 153
    },
    {
      "epoch": 0.4741196844333269,
      "grad_norm": 0.3524907849516785,
      "learning_rate": 1.9806540241924316e-05,
      "loss": 0.9696,
      "step": 154
    },
    {
      "epoch": 0.47719838368289397,
      "grad_norm": 0.4206305610214785,
      "learning_rate": 1.9799465208120928e-05,
      "loss": 0.9688,
      "step": 155
    },
    {
      "epoch": 0.480277082932461,
      "grad_norm": 0.33618660105650827,
      "learning_rate": 1.9792264425917048e-05,
      "loss": 0.9746,
      "step": 156
    },
    {
      "epoch": 0.48335578218202807,
      "grad_norm": 0.34450850841040015,
      "learning_rate": 1.9784937987714334e-05,
      "loss": 0.9604,
      "step": 157
    },
    {
      "epoch": 0.4864344814315952,
      "grad_norm": 0.383550809099091,
      "learning_rate": 1.9777485987526884e-05,
      "loss": 0.9545,
      "step": 158
    },
    {
      "epoch": 0.48951318068116223,
      "grad_norm": 0.36475409691044036,
      "learning_rate": 1.9769908520980036e-05,
      "loss": 0.9617,
      "step": 159
    },
    {
      "epoch": 0.4925918799307293,
      "grad_norm": 0.3430306429998958,
      "learning_rate": 1.9762205685309115e-05,
      "loss": 0.9602,
      "step": 160
    },
    {
      "epoch": 0.49567057918029633,
      "grad_norm": 0.37315227056019334,
      "learning_rate": 1.9754377579358222e-05,
      "loss": 0.947,
      "step": 161
    },
    {
      "epoch": 0.4987492784298634,
      "grad_norm": 0.33802191729774367,
      "learning_rate": 1.9746424303578936e-05,
      "loss": 0.9702,
      "step": 162
    },
    {
      "epoch": 0.5018279776794304,
      "grad_norm": 0.36273101333792734,
      "learning_rate": 1.973834596002905e-05,
      "loss": 0.8744,
      "step": 163
    },
    {
      "epoch": 0.5049066769289975,
      "grad_norm": 0.35934510029142486,
      "learning_rate": 1.9730142652371235e-05,
      "loss": 0.9275,
      "step": 164
    },
    {
      "epoch": 0.5079853761785645,
      "grad_norm": 0.3695947546609777,
      "learning_rate": 1.9721814485871725e-05,
      "loss": 0.9663,
      "step": 165
    },
    {
      "epoch": 0.5110640754281316,
      "grad_norm": 0.3519806320827514,
      "learning_rate": 1.971336156739897e-05,
      "loss": 0.9813,
      "step": 166
    },
    {
      "epoch": 0.5141427746776986,
      "grad_norm": 0.3964320108939583,
      "learning_rate": 1.9704784005422252e-05,
      "loss": 0.9459,
      "step": 167
    },
    {
      "epoch": 0.5172214739272657,
      "grad_norm": 0.3755499757058549,
      "learning_rate": 1.9696081910010298e-05,
      "loss": 1.001,
      "step": 168
    },
    {
      "epoch": 0.5203001731768327,
      "grad_norm": 0.34443735281701665,
      "learning_rate": 1.9687255392829877e-05,
      "loss": 0.962,
      "step": 169
    },
    {
      "epoch": 0.5233788724263998,
      "grad_norm": 0.39955497194325645,
      "learning_rate": 1.9678304567144356e-05,
      "loss": 0.9666,
      "step": 170
    },
    {
      "epoch": 0.5264575716759669,
      "grad_norm": 0.3460094995738535,
      "learning_rate": 1.966922954781225e-05,
      "loss": 1.0106,
      "step": 171
    },
    {
      "epoch": 0.529536270925534,
      "grad_norm": 0.3683405855814382,
      "learning_rate": 1.9660030451285744e-05,
      "loss": 0.9081,
      "step": 172
    },
    {
      "epoch": 0.5326149701751011,
      "grad_norm": 0.3624738077044505,
      "learning_rate": 1.9650707395609205e-05,
      "loss": 0.9463,
      "step": 173
    },
    {
      "epoch": 0.5356936694246681,
      "grad_norm": 0.329489042964862,
      "learning_rate": 1.9641260500417672e-05,
      "loss": 0.9407,
      "step": 174
    },
    {
      "epoch": 0.5387723686742352,
      "grad_norm": 0.3747798974208803,
      "learning_rate": 1.9631689886935298e-05,
      "loss": 0.9358,
      "step": 175
    },
    {
      "epoch": 0.5418510679238022,
      "grad_norm": 0.35738013181658707,
      "learning_rate": 1.9621995677973827e-05,
      "loss": 0.9747,
      "step": 176
    },
    {
      "epoch": 0.5449297671733693,
      "grad_norm": 0.33419135407909323,
      "learning_rate": 1.9612177997930988e-05,
      "loss": 0.943,
      "step": 177
    },
    {
      "epoch": 0.5480084664229363,
      "grad_norm": 0.39868007086165486,
      "learning_rate": 1.9602236972788922e-05,
      "loss": 0.9325,
      "step": 178
    },
    {
      "epoch": 0.5510871656725034,
      "grad_norm": 0.3088641565304442,
      "learning_rate": 1.9592172730112543e-05,
      "loss": 0.9626,
      "step": 179
    },
    {
      "epoch": 0.5541658649220704,
      "grad_norm": 0.35068340170327067,
      "learning_rate": 1.9581985399047934e-05,
      "loss": 0.9498,
      "step": 180
    },
    {
      "epoch": 0.5572445641716375,
      "grad_norm": 0.31208001387569334,
      "learning_rate": 1.9571675110320643e-05,
      "loss": 0.9414,
      "step": 181
    },
    {
      "epoch": 0.5603232634212045,
      "grad_norm": 0.3318397906478814,
      "learning_rate": 1.956124199623405e-05,
      "loss": 0.9542,
      "step": 182
    },
    {
      "epoch": 0.5634019626707716,
      "grad_norm": 0.3438267781878893,
      "learning_rate": 1.955068619066765e-05,
      "loss": 0.9928,
      "step": 183
    },
    {
      "epoch": 0.5664806619203386,
      "grad_norm": 0.3508203349788369,
      "learning_rate": 1.9540007829075322e-05,
      "loss": 0.9319,
      "step": 184
    },
    {
      "epoch": 0.5695593611699057,
      "grad_norm": 0.34965277525798527,
      "learning_rate": 1.952920704848362e-05,
      "loss": 0.957,
      "step": 185
    },
    {
      "epoch": 0.5726380604194727,
      "grad_norm": 0.3366949351990667,
      "learning_rate": 1.951828398748999e-05,
      "loss": 0.9481,
      "step": 186
    },
    {
      "epoch": 0.5757167596690398,
      "grad_norm": 0.327802882679232,
      "learning_rate": 1.950723878626101e-05,
      "loss": 0.9346,
      "step": 187
    },
    {
      "epoch": 0.5787954589186068,
      "grad_norm": 0.3085761935123904,
      "learning_rate": 1.949607158653057e-05,
      "loss": 0.9774,
      "step": 188
    },
    {
      "epoch": 0.5818741581681739,
      "grad_norm": 0.32472718520443367,
      "learning_rate": 1.9484782531598075e-05,
      "loss": 0.9297,
      "step": 189
    },
    {
      "epoch": 0.5849528574177411,
      "grad_norm": 0.3306405011537101,
      "learning_rate": 1.9473371766326594e-05,
      "loss": 0.947,
      "step": 190
    },
    {
      "epoch": 0.5880315566673081,
      "grad_norm": 0.31872496949433493,
      "learning_rate": 1.9461839437141003e-05,
      "loss": 0.9311,
      "step": 191
    },
    {
      "epoch": 0.5911102559168752,
      "grad_norm": 0.33751219506858604,
      "learning_rate": 1.9450185692026105e-05,
      "loss": 0.9247,
      "step": 192
    },
    {
      "epoch": 0.5941889551664422,
      "grad_norm": 0.3461470937416127,
      "learning_rate": 1.943841068052474e-05,
      "loss": 0.9779,
      "step": 193
    },
    {
      "epoch": 0.5972676544160093,
      "grad_norm": 0.32982349951466394,
      "learning_rate": 1.942651455373585e-05,
      "loss": 0.9236,
      "step": 194
    },
    {
      "epoch": 0.6003463536655763,
      "grad_norm": 0.36648455271670227,
      "learning_rate": 1.941449746431255e-05,
      "loss": 0.9528,
      "step": 195
    },
    {
      "epoch": 0.6034250529151434,
      "grad_norm": 0.3107713613696933,
      "learning_rate": 1.9402359566460175e-05,
      "loss": 0.9425,
      "step": 196
    },
    {
      "epoch": 0.6065037521647104,
      "grad_norm": 0.35149969170290607,
      "learning_rate": 1.939010101593429e-05,
      "loss": 0.9385,
      "step": 197
    },
    {
      "epoch": 0.6095824514142775,
      "grad_norm": 0.3354666654924034,
      "learning_rate": 1.9377721970038696e-05,
      "loss": 0.9675,
      "step": 198
    },
    {
      "epoch": 0.6126611506638445,
      "grad_norm": 0.3329554853428688,
      "learning_rate": 1.9365222587623407e-05,
      "loss": 0.9536,
      "step": 199
    },
    {
      "epoch": 0.6157398499134116,
      "grad_norm": 0.34061080263632726,
      "learning_rate": 1.9352603029082622e-05,
      "loss": 0.9522,
      "step": 200
    },
    {
      "epoch": 0.6188185491629786,
      "grad_norm": 0.3382720934063105,
      "learning_rate": 1.9339863456352658e-05,
      "loss": 0.9309,
      "step": 201
    },
    {
      "epoch": 0.6218972484125457,
      "grad_norm": 0.3520371439983956,
      "learning_rate": 1.9327004032909873e-05,
      "loss": 0.9175,
      "step": 202
    },
    {
      "epoch": 0.6249759476621127,
      "grad_norm": 0.3970256772326755,
      "learning_rate": 1.931402492376857e-05,
      "loss": 0.9462,
      "step": 203
    },
    {
      "epoch": 0.6280546469116798,
      "grad_norm": 0.3735141819511392,
      "learning_rate": 1.9300926295478883e-05,
      "loss": 0.9619,
      "step": 204
    },
    {
      "epoch": 0.6311333461612468,
      "grad_norm": 0.3294788194623726,
      "learning_rate": 1.928770831612463e-05,
      "loss": 0.8685,
      "step": 205
    },
    {
      "epoch": 0.6342120454108139,
      "grad_norm": 0.3553207131584365,
      "learning_rate": 1.9274371155321167e-05,
      "loss": 0.9589,
      "step": 206
    },
    {
      "epoch": 0.6372907446603809,
      "grad_norm": 0.33809252417404695,
      "learning_rate": 1.9260914984213205e-05,
      "loss": 0.9575,
      "step": 207
    },
    {
      "epoch": 0.6403694439099481,
      "grad_norm": 0.3439452589077563,
      "learning_rate": 1.924733997547261e-05,
      "loss": 0.9424,
      "step": 208
    },
    {
      "epoch": 0.6434481431595152,
      "grad_norm": 0.32453465719566726,
      "learning_rate": 1.9233646303296204e-05,
      "loss": 0.957,
      "step": 209
    },
    {
      "epoch": 0.6465268424090822,
      "grad_norm": 0.3353072864242234,
      "learning_rate": 1.921983414340351e-05,
      "loss": 0.9578,
      "step": 210
    },
    {
      "epoch": 0.6496055416586493,
      "grad_norm": 0.2986195728173464,
      "learning_rate": 1.920590367303451e-05,
      "loss": 0.914,
      "step": 211
    },
    {
      "epoch": 0.6526842409082163,
      "grad_norm": 0.3170162015297293,
      "learning_rate": 1.919185507094736e-05,
      "loss": 0.9474,
      "step": 212
    },
    {
      "epoch": 0.6557629401577834,
      "grad_norm": 0.3408938972206461,
      "learning_rate": 1.9177688517416105e-05,
      "loss": 0.9571,
      "step": 213
    },
    {
      "epoch": 0.6588416394073504,
      "grad_norm": 0.32232522680119735,
      "learning_rate": 1.916340419422837e-05,
      "loss": 0.9935,
      "step": 214
    },
    {
      "epoch": 0.6619203386569175,
      "grad_norm": 0.3574222323210756,
      "learning_rate": 1.9149002284683008e-05,
      "loss": 0.9287,
      "step": 215
    },
    {
      "epoch": 0.6649990379064845,
      "grad_norm": 0.33446630337671696,
      "learning_rate": 1.9134482973587773e-05,
      "loss": 0.9701,
      "step": 216
    },
    {
      "epoch": 0.6680777371560516,
      "grad_norm": 0.3346785403226774,
      "learning_rate": 1.911984644725692e-05,
      "loss": 0.9443,
      "step": 217
    },
    {
      "epoch": 0.6711564364056186,
      "grad_norm": 0.3180178038648051,
      "learning_rate": 1.9105092893508845e-05,
      "loss": 0.9537,
      "step": 218
    },
    {
      "epoch": 0.6742351356551857,
      "grad_norm": 0.32090192657151384,
      "learning_rate": 1.9090222501663648e-05,
      "loss": 0.9275,
      "step": 219
    },
    {
      "epoch": 0.6773138349047527,
      "grad_norm": 0.33073360863899837,
      "learning_rate": 1.907523546254073e-05,
      "loss": 0.9029,
      "step": 220
    },
    {
      "epoch": 0.6803925341543198,
      "grad_norm": 0.356977804533973,
      "learning_rate": 1.906013196845631e-05,
      "loss": 0.9776,
      "step": 221
    },
    {
      "epoch": 0.6834712334038868,
      "grad_norm": 0.3240040600578929,
      "learning_rate": 1.9044912213220996e-05,
      "loss": 0.9445,
      "step": 222
    },
    {
      "epoch": 0.6865499326534539,
      "grad_norm": 0.3210704034968986,
      "learning_rate": 1.9029576392137263e-05,
      "loss": 0.8871,
      "step": 223
    },
    {
      "epoch": 0.6896286319030209,
      "grad_norm": 0.3106518690023471,
      "learning_rate": 1.9014124701996973e-05,
      "loss": 0.8951,
      "step": 224
    },
    {
      "epoch": 0.692707331152588,
      "grad_norm": 0.32628190575598676,
      "learning_rate": 1.8998557341078837e-05,
      "loss": 0.9307,
      "step": 225
    },
    {
      "epoch": 0.6957860304021551,
      "grad_norm": 0.32268161908950876,
      "learning_rate": 1.8982874509145866e-05,
      "loss": 0.9315,
      "step": 226
    },
    {
      "epoch": 0.6988647296517222,
      "grad_norm": 0.3711292396862262,
      "learning_rate": 1.896707640744283e-05,
      "loss": 0.9824,
      "step": 227
    },
    {
      "epoch": 0.7019434289012892,
      "grad_norm": 0.303785022271297,
      "learning_rate": 1.895116323869364e-05,
      "loss": 0.9291,
      "step": 228
    },
    {
      "epoch": 0.7050221281508563,
      "grad_norm": 0.33003343275586033,
      "learning_rate": 1.8935135207098783e-05,
      "loss": 0.9357,
      "step": 229
    },
    {
      "epoch": 0.7081008274004233,
      "grad_norm": 0.3188349201869865,
      "learning_rate": 1.8918992518332686e-05,
      "loss": 0.9668,
      "step": 230
    },
    {
      "epoch": 0.7111795266499904,
      "grad_norm": 0.31792909466236696,
      "learning_rate": 1.8902735379541064e-05,
      "loss": 0.9384,
      "step": 231
    },
    {
      "epoch": 0.7142582258995575,
      "grad_norm": 0.3150969530965864,
      "learning_rate": 1.8886363999338285e-05,
      "loss": 0.9406,
      "step": 232
    },
    {
      "epoch": 0.7173369251491245,
      "grad_norm": 0.3357190316990474,
      "learning_rate": 1.886987858780467e-05,
      "loss": 0.9349,
      "step": 233
    },
    {
      "epoch": 0.7204156243986916,
      "grad_norm": 0.31601828253547465,
      "learning_rate": 1.8853279356483828e-05,
      "loss": 0.9372,
      "step": 234
    },
    {
      "epoch": 0.7234943236482586,
      "grad_norm": 0.3210995374116869,
      "learning_rate": 1.88365665183799e-05,
      "loss": 0.9262,
      "step": 235
    },
    {
      "epoch": 0.7265730228978257,
      "grad_norm": 0.309922937349771,
      "learning_rate": 1.8819740287954876e-05,
      "loss": 0.9177,
      "step": 236
    },
    {
      "epoch": 0.7296517221473927,
      "grad_norm": 0.3392427346432449,
      "learning_rate": 1.8802800881125785e-05,
      "loss": 0.9438,
      "step": 237
    },
    {
      "epoch": 0.7327304213969598,
      "grad_norm": 0.29875010645220373,
      "learning_rate": 1.8785748515261977e-05,
      "loss": 0.926,
      "step": 238
    },
    {
      "epoch": 0.7358091206465268,
      "grad_norm": 0.3468502318241443,
      "learning_rate": 1.8768583409182307e-05,
      "loss": 0.9409,
      "step": 239
    },
    {
      "epoch": 0.7388878198960939,
      "grad_norm": 0.31982979911374104,
      "learning_rate": 1.8751305783152326e-05,
      "loss": 0.9195,
      "step": 240
    },
    {
      "epoch": 0.7419665191456609,
      "grad_norm": 0.3350297622230112,
      "learning_rate": 1.8733915858881462e-05,
      "loss": 0.9682,
      "step": 241
    },
    {
      "epoch": 0.745045218395228,
      "grad_norm": 0.3507725781350706,
      "learning_rate": 1.8716413859520176e-05,
      "loss": 0.9549,
      "step": 242
    },
    {
      "epoch": 0.748123917644795,
      "grad_norm": 0.33477437516104136,
      "learning_rate": 1.8698800009657095e-05,
      "loss": 0.9633,
      "step": 243
    },
    {
      "epoch": 0.7512026168943622,
      "grad_norm": 0.3299583593004973,
      "learning_rate": 1.8681074535316127e-05,
      "loss": 0.929,
      "step": 244
    },
    {
      "epoch": 0.7542813161439292,
      "grad_norm": 0.3272584764297615,
      "learning_rate": 1.8663237663953565e-05,
      "loss": 0.9299,
      "step": 245
    },
    {
      "epoch": 0.7573600153934963,
      "grad_norm": 0.539785976930736,
      "learning_rate": 1.8645289624455175e-05,
      "loss": 0.9718,
      "step": 246
    },
    {
      "epoch": 0.7604387146430633,
      "grad_norm": 0.3526411587574386,
      "learning_rate": 1.8627230647133242e-05,
      "loss": 0.9189,
      "step": 247
    },
    {
      "epoch": 0.7635174138926304,
      "grad_norm": 0.33476055108444863,
      "learning_rate": 1.8609060963723626e-05,
      "loss": 0.9445,
      "step": 248
    },
    {
      "epoch": 0.7665961131421974,
      "grad_norm": 0.34869291212815673,
      "learning_rate": 1.859078080738279e-05,
      "loss": 0.9324,
      "step": 249
    },
    {
      "epoch": 0.7696748123917645,
      "grad_norm": 0.3649810793217048,
      "learning_rate": 1.8572390412684807e-05,
      "loss": 0.9489,
      "step": 250
    },
    {
      "epoch": 0.7727535116413315,
      "grad_norm": 0.3357218099291148,
      "learning_rate": 1.8553890015618333e-05,
      "loss": 0.9302,
      "step": 251
    },
    {
      "epoch": 0.7758322108908986,
      "grad_norm": 0.34493272284601867,
      "learning_rate": 1.853527985358361e-05,
      "loss": 0.9331,
      "step": 252
    },
    {
      "epoch": 0.7789109101404657,
      "grad_norm": 0.3619906059499835,
      "learning_rate": 1.851656016538939e-05,
      "loss": 0.9264,
      "step": 253
    },
    {
      "epoch": 0.7819896093900327,
      "grad_norm": 0.34798362652882314,
      "learning_rate": 1.8497731191249894e-05,
      "loss": 0.9675,
      "step": 254
    },
    {
      "epoch": 0.7850683086395998,
      "grad_norm": 0.3312540751742641,
      "learning_rate": 1.8478793172781708e-05,
      "loss": 0.9204,
      "step": 255
    },
    {
      "epoch": 0.7881470078891668,
      "grad_norm": 0.33463700481954023,
      "learning_rate": 1.8459746353000704e-05,
      "loss": 0.9468,
      "step": 256
    },
    {
      "epoch": 0.7912257071387339,
      "grad_norm": 0.38548449068572227,
      "learning_rate": 1.84405909763189e-05,
      "loss": 0.9834,
      "step": 257
    },
    {
      "epoch": 0.7943044063883009,
      "grad_norm": 0.3372945659300871,
      "learning_rate": 1.8421327288541343e-05,
      "loss": 0.9439,
      "step": 258
    },
    {
      "epoch": 0.797383105637868,
      "grad_norm": 0.3582869526923825,
      "learning_rate": 1.8401955536862948e-05,
      "loss": 0.9868,
      "step": 259
    },
    {
      "epoch": 0.800461804887435,
      "grad_norm": 0.3228400913444128,
      "learning_rate": 1.8382475969865312e-05,
      "loss": 0.9451,
      "step": 260
    },
    {
      "epoch": 0.8035405041370021,
      "grad_norm": 0.3469219573602684,
      "learning_rate": 1.8362888837513548e-05,
      "loss": 0.9287,
      "step": 261
    },
    {
      "epoch": 0.8066192033865692,
      "grad_norm": 0.33585399343411004,
      "learning_rate": 1.8343194391153064e-05,
      "loss": 0.9466,
      "step": 262
    },
    {
      "epoch": 0.8096979026361363,
      "grad_norm": 0.32463474870094694,
      "learning_rate": 1.8323392883506335e-05,
      "loss": 0.9286,
      "step": 263
    },
    {
      "epoch": 0.8127766018857033,
      "grad_norm": 0.3358329129766793,
      "learning_rate": 1.830348456866967e-05,
      "loss": 0.9344,
      "step": 264
    },
    {
      "epoch": 0.8158553011352704,
      "grad_norm": 0.34155550235000554,
      "learning_rate": 1.8283469702109937e-05,
      "loss": 0.9313,
      "step": 265
    },
    {
      "epoch": 0.8189340003848374,
      "grad_norm": 0.34551453649829983,
      "learning_rate": 1.8263348540661306e-05,
      "loss": 0.8916,
      "step": 266
    },
    {
      "epoch": 0.8220126996344045,
      "grad_norm": 0.3522340051738529,
      "learning_rate": 1.8243121342521935e-05,
      "loss": 0.9437,
      "step": 267
    },
    {
      "epoch": 0.8250913988839715,
      "grad_norm": 0.34053719153257295,
      "learning_rate": 1.8222788367250665e-05,
      "loss": 0.9357,
      "step": 268
    },
    {
      "epoch": 0.8281700981335386,
      "grad_norm": 0.3390029570728729,
      "learning_rate": 1.8202349875763682e-05,
      "loss": 0.9369,
      "step": 269
    },
    {
      "epoch": 0.8312487973831056,
      "grad_norm": 0.34915883960306154,
      "learning_rate": 1.8181806130331173e-05,
      "loss": 0.9281,
      "step": 270
    },
    {
      "epoch": 0.8343274966326727,
      "grad_norm": 0.3226519382364744,
      "learning_rate": 1.816115739457397e-05,
      "loss": 0.9416,
      "step": 271
    },
    {
      "epoch": 0.8374061958822397,
      "grad_norm": 0.3330518874203431,
      "learning_rate": 1.814040393346016e-05,
      "loss": 0.9302,
      "step": 272
    },
    {
      "epoch": 0.8404848951318068,
      "grad_norm": 0.3503792075436629,
      "learning_rate": 1.8119546013301664e-05,
      "loss": 0.9522,
      "step": 273
    },
    {
      "epoch": 0.8435635943813738,
      "grad_norm": 0.33547740615130206,
      "learning_rate": 1.809858390175087e-05,
      "loss": 0.9238,
      "step": 274
    },
    {
      "epoch": 0.8466422936309409,
      "grad_norm": 0.3527045502319622,
      "learning_rate": 1.807751786779713e-05,
      "loss": 0.9767,
      "step": 275
    },
    {
      "epoch": 0.849720992880508,
      "grad_norm": 0.3282450280938344,
      "learning_rate": 1.8056348181763387e-05,
      "loss": 0.9002,
      "step": 276
    },
    {
      "epoch": 0.852799692130075,
      "grad_norm": 0.3219892021752841,
      "learning_rate": 1.8035075115302632e-05,
      "loss": 0.9289,
      "step": 277
    },
    {
      "epoch": 0.8558783913796421,
      "grad_norm": 0.34957525690544,
      "learning_rate": 1.8013698941394472e-05,
      "loss": 0.9728,
      "step": 278
    },
    {
      "epoch": 0.8589570906292091,
      "grad_norm": 0.3497198288104489,
      "learning_rate": 1.7992219934341592e-05,
      "loss": 0.8799,
      "step": 279
    },
    {
      "epoch": 0.8620357898787763,
      "grad_norm": 0.32841962062091323,
      "learning_rate": 1.7970638369766254e-05,
      "loss": 0.8829,
      "step": 280
    },
    {
      "epoch": 0.8651144891283433,
      "grad_norm": 0.3211078469298204,
      "learning_rate": 1.7948954524606764e-05,
      "loss": 0.9098,
      "step": 281
    },
    {
      "epoch": 0.8681931883779104,
      "grad_norm": 0.30391645437343673,
      "learning_rate": 1.7927168677113895e-05,
      "loss": 0.8933,
      "step": 282
    },
    {
      "epoch": 0.8712718876274774,
      "grad_norm": 0.3241976131726522,
      "learning_rate": 1.7905281106847346e-05,
      "loss": 0.9553,
      "step": 283
    },
    {
      "epoch": 0.8743505868770445,
      "grad_norm": 0.33449695787062905,
      "learning_rate": 1.7883292094672127e-05,
      "loss": 0.9574,
      "step": 284
    },
    {
      "epoch": 0.8774292861266115,
      "grad_norm": 0.3186236939297094,
      "learning_rate": 1.786120192275498e-05,
      "loss": 0.9239,
      "step": 285
    },
    {
      "epoch": 0.8805079853761786,
      "grad_norm": 0.32188688837359464,
      "learning_rate": 1.7839010874560732e-05,
      "loss": 0.9086,
      "step": 286
    },
    {
      "epoch": 0.8835866846257456,
      "grad_norm": 0.3266147921500336,
      "learning_rate": 1.781671923484869e-05,
      "loss": 0.9055,
      "step": 287
    },
    {
      "epoch": 0.8866653838753127,
      "grad_norm": 0.328654943221853,
      "learning_rate": 1.7794327289668945e-05,
      "loss": 0.9559,
      "step": 288
    },
    {
      "epoch": 0.8897440831248797,
      "grad_norm": 0.29295157740717115,
      "learning_rate": 1.7771835326358742e-05,
      "loss": 0.9249,
      "step": 289
    },
    {
      "epoch": 0.8928227823744468,
      "grad_norm": 0.3532373486933444,
      "learning_rate": 1.7749243633538778e-05,
      "loss": 0.9462,
      "step": 290
    },
    {
      "epoch": 0.8959014816240138,
      "grad_norm": 0.3256732207476102,
      "learning_rate": 1.772655250110948e-05,
      "loss": 0.9332,
      "step": 291
    },
    {
      "epoch": 0.8989801808735809,
      "grad_norm": 0.32651069875672634,
      "learning_rate": 1.7703762220247314e-05,
      "loss": 0.9634,
      "step": 292
    },
    {
      "epoch": 0.9020588801231479,
      "grad_norm": 0.3187920181572499,
      "learning_rate": 1.768087308340103e-05,
      "loss": 0.9596,
      "step": 293
    },
    {
      "epoch": 0.905137579372715,
      "grad_norm": 0.33811383571626685,
      "learning_rate": 1.765788538428792e-05,
      "loss": 0.9296,
      "step": 294
    },
    {
      "epoch": 0.908216278622282,
      "grad_norm": 0.3108811485408006,
      "learning_rate": 1.7634799417890037e-05,
      "loss": 0.9182,
      "step": 295
    },
    {
      "epoch": 0.9112949778718491,
      "grad_norm": 0.33615096135984524,
      "learning_rate": 1.7611615480450413e-05,
      "loss": 0.948,
      "step": 296
    },
    {
      "epoch": 0.9143736771214162,
      "grad_norm": 0.31804146075858497,
      "learning_rate": 1.758833386946928e-05,
      "loss": 0.859,
      "step": 297
    },
    {
      "epoch": 0.9174523763709833,
      "grad_norm": 0.3180959030566039,
      "learning_rate": 1.7564954883700214e-05,
      "loss": 0.9662,
      "step": 298
    },
    {
      "epoch": 0.9205310756205504,
      "grad_norm": 0.30130139696674757,
      "learning_rate": 1.7541478823146328e-05,
      "loss": 0.9533,
      "step": 299
    },
    {
      "epoch": 0.9236097748701174,
      "grad_norm": 0.31810708781359087,
      "learning_rate": 1.7517905989056416e-05,
      "loss": 0.8938,
      "step": 300
    },
    {
      "epoch": 0.9266884741196845,
      "grad_norm": 0.3175985825459941,
      "learning_rate": 1.7494236683921084e-05,
      "loss": 0.9531,
      "step": 301
    },
    {
      "epoch": 0.9297671733692515,
      "grad_norm": 0.31058485813592024,
      "learning_rate": 1.7470471211468872e-05,
      "loss": 0.9336,
      "step": 302
    },
    {
      "epoch": 0.9328458726188186,
      "grad_norm": 0.3187712431688849,
      "learning_rate": 1.7446609876662358e-05,
      "loss": 0.9522,
      "step": 303
    },
    {
      "epoch": 0.9359245718683856,
      "grad_norm": 0.3046669309865387,
      "learning_rate": 1.7422652985694237e-05,
      "loss": 0.9228,
      "step": 304
    },
    {
      "epoch": 0.9390032711179527,
      "grad_norm": 0.29351923190587265,
      "learning_rate": 1.7398600845983402e-05,
      "loss": 0.9518,
      "step": 305
    },
    {
      "epoch": 0.9420819703675197,
      "grad_norm": 0.3195326416534455,
      "learning_rate": 1.7374453766170987e-05,
      "loss": 0.9309,
      "step": 306
    },
    {
      "epoch": 0.9451606696170868,
      "grad_norm": 0.31717844770250414,
      "learning_rate": 1.735021205611642e-05,
      "loss": 0.9126,
      "step": 307
    },
    {
      "epoch": 0.9482393688666538,
      "grad_norm": 0.3529174896894431,
      "learning_rate": 1.732587602689344e-05,
      "loss": 0.9476,
      "step": 308
    },
    {
      "epoch": 0.9513180681162209,
      "grad_norm": 0.32692231079387907,
      "learning_rate": 1.73014459907861e-05,
      "loss": 0.9174,
      "step": 309
    },
    {
      "epoch": 0.9543967673657879,
      "grad_norm": 0.34031475301680525,
      "learning_rate": 1.7276922261284772e-05,
      "loss": 0.9421,
      "step": 310
    },
    {
      "epoch": 0.957475466615355,
      "grad_norm": 0.34832574529691984,
      "learning_rate": 1.7252305153082118e-05,
      "loss": 0.9459,
      "step": 311
    },
    {
      "epoch": 0.960554165864922,
      "grad_norm": 0.28585806717602336,
      "learning_rate": 1.7227594982069038e-05,
      "loss": 0.8616,
      "step": 312
    },
    {
      "epoch": 0.9636328651144891,
      "grad_norm": 0.33893214247386305,
      "learning_rate": 1.7202792065330647e-05,
      "loss": 0.939,
      "step": 313
    },
    {
      "epoch": 0.9667115643640561,
      "grad_norm": 0.30729369627315095,
      "learning_rate": 1.717789672114218e-05,
      "loss": 0.9199,
      "step": 314
    },
    {
      "epoch": 0.9697902636136232,
      "grad_norm": 0.33906122913802444,
      "learning_rate": 1.7152909268964915e-05,
      "loss": 0.9611,
      "step": 315
    },
    {
      "epoch": 0.9728689628631904,
      "grad_norm": 0.33426448364456823,
      "learning_rate": 1.712783002944209e-05,
      "loss": 0.9315,
      "step": 316
    },
    {
      "epoch": 0.9759476621127574,
      "grad_norm": 0.3629670869669532,
      "learning_rate": 1.710265932439475e-05,
      "loss": 0.929,
      "step": 317
    },
    {
      "epoch": 0.9790263613623245,
      "grad_norm": 0.3199756324906867,
      "learning_rate": 1.7077397476817657e-05,
      "loss": 0.8936,
      "step": 318
    },
    {
      "epoch": 0.9821050606118915,
      "grad_norm": 0.34092680210251797,
      "learning_rate": 1.705204481087513e-05,
      "loss": 0.947,
      "step": 319
    },
    {
      "epoch": 0.9851837598614586,
      "grad_norm": 0.3948165528726888,
      "learning_rate": 1.702660165189687e-05,
      "loss": 0.9313,
      "step": 320
    },
    {
      "epoch": 0.9882624591110256,
      "grad_norm": 0.306576582362445,
      "learning_rate": 1.7001068326373827e-05,
      "loss": 0.9138,
      "step": 321
    },
    {
      "epoch": 0.9913411583605927,
      "grad_norm": 0.36139272904680547,
      "learning_rate": 1.6975445161953953e-05,
      "loss": 0.9326,
      "step": 322
    },
    {
      "epoch": 0.9944198576101597,
      "grad_norm": 0.3419274330213101,
      "learning_rate": 1.6949732487438048e-05,
      "loss": 0.9451,
      "step": 323
    },
    {
      "epoch": 0.9974985568597268,
      "grad_norm": 0.34090117874723225,
      "learning_rate": 1.6923930632775517e-05,
      "loss": 0.9676,
      "step": 324
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.34090117874723225,
      "learning_rate": 1.689803992906013e-05,
      "loss": 0.8989,
      "step": 325
    },
    {
      "epoch": 1.003078699249567,
      "grad_norm": 0.41071041357443766,
      "learning_rate": 1.68720607085258e-05,
      "loss": 0.9419,
      "step": 326
    },
    {
      "epoch": 1.006157398499134,
      "grad_norm": 0.40791985654884316,
      "learning_rate": 1.6845993304542285e-05,
      "loss": 0.9106,
      "step": 327
    },
    {
      "epoch": 1.0092360977487012,
      "grad_norm": 0.3693406349922746,
      "learning_rate": 1.6819838051610942e-05,
      "loss": 0.9347,
      "step": 328
    },
    {
      "epoch": 1.0123147969982682,
      "grad_norm": 0.3911010266181506,
      "learning_rate": 1.679359528536041e-05,
      "loss": 0.9049,
      "step": 329
    },
    {
      "epoch": 1.0153934962478353,
      "grad_norm": 0.3549570336907906,
      "learning_rate": 1.6767265342542325e-05,
      "loss": 0.8773,
      "step": 330
    },
    {
      "epoch": 1.0184721954974023,
      "grad_norm": 0.3752840129468656,
      "learning_rate": 1.674084856102698e-05,
      "loss": 0.8595,
      "step": 331
    },
    {
      "epoch": 1.0215508947469694,
      "grad_norm": 0.3416893264748966,
      "learning_rate": 1.6714345279798994e-05,
      "loss": 0.8216,
      "step": 332
    },
    {
      "epoch": 1.0246295939965364,
      "grad_norm": 0.33575226595733065,
      "learning_rate": 1.668775583895297e-05,
      "loss": 0.866,
      "step": 333
    },
    {
      "epoch": 1.0277082932461035,
      "grad_norm": 0.3648883024581598,
      "learning_rate": 1.6661080579689133e-05,
      "loss": 0.8654,
      "step": 334
    },
    {
      "epoch": 1.0307869924956705,
      "grad_norm": 0.32646027577376685,
      "learning_rate": 1.6634319844308926e-05,
      "loss": 0.9148,
      "step": 335
    },
    {
      "epoch": 1.0338656917452376,
      "grad_norm": 0.37042041985365864,
      "learning_rate": 1.660747397621065e-05,
      "loss": 0.8971,
      "step": 336
    },
    {
      "epoch": 1.0369443909948046,
      "grad_norm": 0.31799470595697027,
      "learning_rate": 1.6580543319885048e-05,
      "loss": 0.8784,
      "step": 337
    },
    {
      "epoch": 1.0400230902443717,
      "grad_norm": 0.37178111808501296,
      "learning_rate": 1.655352822091086e-05,
      "loss": 0.9303,
      "step": 338
    },
    {
      "epoch": 1.0431017894939387,
      "grad_norm": 0.31283196939778596,
      "learning_rate": 1.6526429025950427e-05,
      "loss": 0.873,
      "step": 339
    },
    {
      "epoch": 1.0461804887435058,
      "grad_norm": 0.3228591627553886,
      "learning_rate": 1.6499246082745214e-05,
      "loss": 0.8626,
      "step": 340
    },
    {
      "epoch": 1.0492591879930728,
      "grad_norm": 0.33728028419250156,
      "learning_rate": 1.647197974011137e-05,
      "loss": 0.857,
      "step": 341
    },
    {
      "epoch": 1.0523378872426399,
      "grad_norm": 0.34314585643300666,
      "learning_rate": 1.6444630347935216e-05,
      "loss": 0.8776,
      "step": 342
    },
    {
      "epoch": 1.055416586492207,
      "grad_norm": 0.3118893202141781,
      "learning_rate": 1.6417198257168805e-05,
      "loss": 0.8744,
      "step": 343
    },
    {
      "epoch": 1.0584952857417742,
      "grad_norm": 0.34591012900525225,
      "learning_rate": 1.638968381982538e-05,
      "loss": 0.8796,
      "step": 344
    },
    {
      "epoch": 1.0615739849913413,
      "grad_norm": 0.3110701222694401,
      "learning_rate": 1.6362087388974865e-05,
      "loss": 0.8751,
      "step": 345
    },
    {
      "epoch": 1.0646526842409083,
      "grad_norm": 0.32598062981146453,
      "learning_rate": 1.6334409318739344e-05,
      "loss": 0.8709,
      "step": 346
    },
    {
      "epoch": 1.0677313834904754,
      "grad_norm": 0.31583007905401905,
      "learning_rate": 1.6306649964288516e-05,
      "loss": 0.8819,
      "step": 347
    },
    {
      "epoch": 1.0708100827400424,
      "grad_norm": 0.3111204398727088,
      "learning_rate": 1.6278809681835124e-05,
      "loss": 0.8814,
      "step": 348
    },
    {
      "epoch": 1.0738887819896095,
      "grad_norm": 0.3331557887436647,
      "learning_rate": 1.62508888286304e-05,
      "loss": 0.893,
      "step": 349
    },
    {
      "epoch": 1.0769674812391765,
      "grad_norm": 0.3079985576657154,
      "learning_rate": 1.6222887762959467e-05,
      "loss": 0.891,
      "step": 350
    },
    {
      "epoch": 1.0800461804887436,
      "grad_norm": 0.3212924340010561,
      "learning_rate": 1.6194806844136755e-05,
      "loss": 0.8546,
      "step": 351
    },
    {
      "epoch": 1.0831248797383106,
      "grad_norm": 0.3150986832096223,
      "learning_rate": 1.6166646432501377e-05,
      "loss": 0.8209,
      "step": 352
    },
    {
      "epoch": 1.0862035789878777,
      "grad_norm": 0.3071705578928538,
      "learning_rate": 1.6138406889412514e-05,
      "loss": 0.8776,
      "step": 353
    },
    {
      "epoch": 1.0892822782374447,
      "grad_norm": 0.3296429660192889,
      "learning_rate": 1.6110088577244772e-05,
      "loss": 0.8383,
      "step": 354
    },
    {
      "epoch": 1.0923609774870118,
      "grad_norm": 0.31958447032569326,
      "learning_rate": 1.6081691859383546e-05,
      "loss": 0.8863,
      "step": 355
    },
    {
      "epoch": 1.0954396767365788,
      "grad_norm": 0.3330740818545088,
      "learning_rate": 1.6053217100220332e-05,
      "loss": 0.8655,
      "step": 356
    },
    {
      "epoch": 1.0985183759861459,
      "grad_norm": 0.31990724303663104,
      "learning_rate": 1.6024664665148076e-05,
      "loss": 0.9089,
      "step": 357
    },
    {
      "epoch": 1.101597075235713,
      "grad_norm": 0.3058782893528889,
      "learning_rate": 1.5996034920556468e-05,
      "loss": 0.8824,
      "step": 358
    },
    {
      "epoch": 1.10467577448528,
      "grad_norm": 0.3455313091893555,
      "learning_rate": 1.596732823382725e-05,
      "loss": 0.8966,
      "step": 359
    },
    {
      "epoch": 1.107754473734847,
      "grad_norm": 0.3171305924320794,
      "learning_rate": 1.5938544973329494e-05,
      "loss": 0.853,
      "step": 360
    },
    {
      "epoch": 1.110833172984414,
      "grad_norm": 0.32300909338220946,
      "learning_rate": 1.5909685508414884e-05,
      "loss": 0.8813,
      "step": 361
    },
    {
      "epoch": 1.1139118722339811,
      "grad_norm": 0.342964799028595,
      "learning_rate": 1.5880750209412977e-05,
      "loss": 0.8616,
      "step": 362
    },
    {
      "epoch": 1.1169905714835482,
      "grad_norm": 0.2967872284446462,
      "learning_rate": 1.5851739447626436e-05,
      "loss": 0.8701,
      "step": 363
    },
    {
      "epoch": 1.1200692707331152,
      "grad_norm": 0.3082189636602768,
      "learning_rate": 1.5822653595326277e-05,
      "loss": 0.8355,
      "step": 364
    },
    {
      "epoch": 1.1231479699826823,
      "grad_norm": 0.32543576086943427,
      "learning_rate": 1.5793493025747092e-05,
      "loss": 0.8651,
      "step": 365
    },
    {
      "epoch": 1.1262266692322493,
      "grad_norm": 0.3049211504593774,
      "learning_rate": 1.5764258113082266e-05,
      "loss": 0.9022,
      "step": 366
    },
    {
      "epoch": 1.1293053684818164,
      "grad_norm": 0.3339177262841496,
      "learning_rate": 1.5734949232479154e-05,
      "loss": 0.8984,
      "step": 367
    },
    {
      "epoch": 1.1323840677313834,
      "grad_norm": 0.32693550793684595,
      "learning_rate": 1.5705566760034283e-05,
      "loss": 0.8863,
      "step": 368
    },
    {
      "epoch": 1.1354627669809505,
      "grad_norm": 0.32728701986644293,
      "learning_rate": 1.567611107278853e-05,
      "loss": 0.8609,
      "step": 369
    },
    {
      "epoch": 1.1385414662305176,
      "grad_norm": 0.32617397828457784,
      "learning_rate": 1.5646582548722273e-05,
      "loss": 0.8883,
      "step": 370
    },
    {
      "epoch": 1.1416201654800846,
      "grad_norm": 0.33423169883030784,
      "learning_rate": 1.561698156675054e-05,
      "loss": 0.8892,
      "step": 371
    },
    {
      "epoch": 1.1446988647296517,
      "grad_norm": 0.3335256225192547,
      "learning_rate": 1.558730850671815e-05,
      "loss": 0.8822,
      "step": 372
    },
    {
      "epoch": 1.1477775639792187,
      "grad_norm": 0.32295476150294805,
      "learning_rate": 1.555756374939486e-05,
      "loss": 0.8622,
      "step": 373
    },
    {
      "epoch": 1.1508562632287858,
      "grad_norm": 0.3457038489904611,
      "learning_rate": 1.552774767647043e-05,
      "loss": 0.8516,
      "step": 374
    },
    {
      "epoch": 1.1539349624783528,
      "grad_norm": 0.3211578768686955,
      "learning_rate": 1.5497860670549774e-05,
      "loss": 0.9013,
      "step": 375
    },
    {
      "epoch": 1.1570136617279199,
      "grad_norm": 0.35008907414862833,
      "learning_rate": 1.5467903115148023e-05,
      "loss": 0.8458,
      "step": 376
    },
    {
      "epoch": 1.1600923609774871,
      "grad_norm": 0.3120136402358896,
      "learning_rate": 1.5437875394685605e-05,
      "loss": 0.853,
      "step": 377
    },
    {
      "epoch": 1.1631710602270542,
      "grad_norm": 0.302248854171887,
      "learning_rate": 1.5407777894483335e-05,
      "loss": 0.896,
      "step": 378
    },
    {
      "epoch": 1.1662497594766212,
      "grad_norm": 0.30520575629146834,
      "learning_rate": 1.537761100075744e-05,
      "loss": 0.8695,
      "step": 379
    },
    {
      "epoch": 1.1693284587261883,
      "grad_norm": 0.3406543242526408,
      "learning_rate": 1.5347375100614615e-05,
      "loss": 0.8492,
      "step": 380
    },
    {
      "epoch": 1.1724071579757553,
      "grad_norm": 0.3092979764652377,
      "learning_rate": 1.5317070582047066e-05,
      "loss": 0.8734,
      "step": 381
    },
    {
      "epoch": 1.1754858572253224,
      "grad_norm": 0.33276042439642484,
      "learning_rate": 1.528669783392751e-05,
      "loss": 0.8941,
      "step": 382
    },
    {
      "epoch": 1.1785645564748894,
      "grad_norm": 0.32408381659266244,
      "learning_rate": 1.5256257246004218e-05,
      "loss": 0.8922,
      "step": 383
    },
    {
      "epoch": 1.1816432557244565,
      "grad_norm": 0.33175590937325034,
      "learning_rate": 1.522574920889597e-05,
      "loss": 0.899,
      "step": 384
    },
    {
      "epoch": 1.1847219549740236,
      "grad_norm": 0.3102861278490506,
      "learning_rate": 1.519517411408708e-05,
      "loss": 0.8893,
      "step": 385
    },
    {
      "epoch": 1.1878006542235906,
      "grad_norm": 0.29528130790514806,
      "learning_rate": 1.516453235392235e-05,
      "loss": 0.9224,
      "step": 386
    },
    {
      "epoch": 1.1908793534731577,
      "grad_norm": 0.31535910787344396,
      "learning_rate": 1.5133824321602047e-05,
      "loss": 0.8307,
      "step": 387
    },
    {
      "epoch": 1.1939580527227247,
      "grad_norm": 0.2895600466593424,
      "learning_rate": 1.5103050411176855e-05,
      "loss": 0.9203,
      "step": 388
    },
    {
      "epoch": 1.1970367519722918,
      "grad_norm": 0.33207372959424775,
      "learning_rate": 1.5072211017542813e-05,
      "loss": 0.8361,
      "step": 389
    },
    {
      "epoch": 1.2001154512218588,
      "grad_norm": 0.3197169440612259,
      "learning_rate": 1.5041306536436256e-05,
      "loss": 0.8773,
      "step": 390
    },
    {
      "epoch": 1.2031941504714259,
      "grad_norm": 0.31358133771660507,
      "learning_rate": 1.5010337364428723e-05,
      "loss": 0.8502,
      "step": 391
    },
    {
      "epoch": 1.206272849720993,
      "grad_norm": 0.2962333612520629,
      "learning_rate": 1.4979303898921891e-05,
      "loss": 0.8992,
      "step": 392
    },
    {
      "epoch": 1.20935154897056,
      "grad_norm": 0.32383848218585004,
      "learning_rate": 1.4948206538142457e-05,
      "loss": 0.8741,
      "step": 393
    },
    {
      "epoch": 1.212430248220127,
      "grad_norm": 0.3267602152518158,
      "learning_rate": 1.4917045681137028e-05,
      "loss": 0.8351,
      "step": 394
    },
    {
      "epoch": 1.215508947469694,
      "grad_norm": 0.31003983519618994,
      "learning_rate": 1.4885821727767007e-05,
      "loss": 0.9118,
      "step": 395
    },
    {
      "epoch": 1.2185876467192611,
      "grad_norm": 0.35703216014985906,
      "learning_rate": 1.4854535078703466e-05,
      "loss": 0.8543,
      "step": 396
    },
    {
      "epoch": 1.2216663459688282,
      "grad_norm": 0.30514486220734727,
      "learning_rate": 1.4823186135421994e-05,
      "loss": 0.8712,
      "step": 397
    },
    {
      "epoch": 1.2247450452183952,
      "grad_norm": 0.3240737336288631,
      "learning_rate": 1.4791775300197553e-05,
      "loss": 0.8589,
      "step": 398
    },
    {
      "epoch": 1.2278237444679623,
      "grad_norm": 0.33380364086342496,
      "learning_rate": 1.4760302976099305e-05,
      "loss": 0.8527,
      "step": 399
    },
    {
      "epoch": 1.2309024437175293,
      "grad_norm": 0.3044027306966491,
      "learning_rate": 1.4728769566985461e-05,
      "loss": 0.8934,
      "step": 400
    },
    {
      "epoch": 1.2339811429670964,
      "grad_norm": 0.3276187691959482,
      "learning_rate": 1.4697175477498074e-05,
      "loss": 0.8988,
      "step": 401
    },
    {
      "epoch": 1.2370598422166634,
      "grad_norm": 0.30428244973743845,
      "learning_rate": 1.4665521113057861e-05,
      "loss": 0.8747,
      "step": 402
    },
    {
      "epoch": 1.2401385414662305,
      "grad_norm": 0.3123985645740693,
      "learning_rate": 1.4633806879859001e-05,
      "loss": 0.8578,
      "step": 403
    },
    {
      "epoch": 1.2432172407157975,
      "grad_norm": 0.31437718916639906,
      "learning_rate": 1.4602033184863913e-05,
      "loss": 0.8594,
      "step": 404
    },
    {
      "epoch": 1.2462959399653646,
      "grad_norm": 0.3077957145741549,
      "learning_rate": 1.4570200435798045e-05,
      "loss": 0.8454,
      "step": 405
    },
    {
      "epoch": 1.2493746392149316,
      "grad_norm": 0.31778242937333095,
      "learning_rate": 1.4538309041144636e-05,
      "loss": 0.8643,
      "step": 406
    },
    {
      "epoch": 1.2524533384644987,
      "grad_norm": 0.30709405339722806,
      "learning_rate": 1.450635941013947e-05,
      "loss": 0.869,
      "step": 407
    },
    {
      "epoch": 1.2555320377140657,
      "grad_norm": 0.30781992372288985,
      "learning_rate": 1.447435195276564e-05,
      "loss": 0.8649,
      "step": 408
    },
    {
      "epoch": 1.2586107369636328,
      "grad_norm": 0.3199591135856051,
      "learning_rate": 1.4442287079748262e-05,
      "loss": 0.8394,
      "step": 409
    },
    {
      "epoch": 1.2616894362131998,
      "grad_norm": 0.3122228301050136,
      "learning_rate": 1.441016520254924e-05,
      "loss": 0.8777,
      "step": 410
    },
    {
      "epoch": 1.264768135462767,
      "grad_norm": 0.32240257825829544,
      "learning_rate": 1.437798673336194e-05,
      "loss": 0.8736,
      "step": 411
    },
    {
      "epoch": 1.267846834712334,
      "grad_norm": 0.3147532362386699,
      "learning_rate": 1.4345752085105948e-05,
      "loss": 0.8358,
      "step": 412
    },
    {
      "epoch": 1.270925533961901,
      "grad_norm": 0.3205388075876308,
      "learning_rate": 1.4313461671421736e-05,
      "loss": 0.8713,
      "step": 413
    },
    {
      "epoch": 1.274004233211468,
      "grad_norm": 0.32939551261098976,
      "learning_rate": 1.4281115906665375e-05,
      "loss": 0.8683,
      "step": 414
    },
    {
      "epoch": 1.277082932461035,
      "grad_norm": 0.2981486408074813,
      "learning_rate": 1.4248715205903205e-05,
      "loss": 0.8581,
      "step": 415
    },
    {
      "epoch": 1.2801616317106022,
      "grad_norm": 0.3226726685046096,
      "learning_rate": 1.4216259984906522e-05,
      "loss": 0.8874,
      "step": 416
    },
    {
      "epoch": 1.2832403309601692,
      "grad_norm": 0.3367900812919343,
      "learning_rate": 1.418375066014622e-05,
      "loss": 0.8767,
      "step": 417
    },
    {
      "epoch": 1.2863190302097363,
      "grad_norm": 0.3151756021170849,
      "learning_rate": 1.4151187648787482e-05,
      "loss": 0.913,
      "step": 418
    },
    {
      "epoch": 1.2893977294593033,
      "grad_norm": 0.32926853433520553,
      "learning_rate": 1.4118571368684383e-05,
      "loss": 0.8677,
      "step": 419
    },
    {
      "epoch": 1.2924764287088706,
      "grad_norm": 0.3465732699292231,
      "learning_rate": 1.4085902238374576e-05,
      "loss": 0.8498,
      "step": 420
    },
    {
      "epoch": 1.2955551279584376,
      "grad_norm": 0.31441754185891374,
      "learning_rate": 1.4053180677073877e-05,
      "loss": 0.8646,
      "step": 421
    },
    {
      "epoch": 1.2986338272080047,
      "grad_norm": 0.3256212283397938,
      "learning_rate": 1.4020407104670917e-05,
      "loss": 0.877,
      "step": 422
    },
    {
      "epoch": 1.3017125264575717,
      "grad_norm": 0.34144711296421126,
      "learning_rate": 1.398758194172174e-05,
      "loss": 0.9049,
      "step": 423
    },
    {
      "epoch": 1.3047912257071388,
      "grad_norm": 0.32769293479974987,
      "learning_rate": 1.3954705609444403e-05,
      "loss": 0.8535,
      "step": 424
    },
    {
      "epoch": 1.3078699249567058,
      "grad_norm": 0.2932609686753501,
      "learning_rate": 1.3921778529713583e-05,
      "loss": 0.8269,
      "step": 425
    },
    {
      "epoch": 1.310948624206273,
      "grad_norm": 0.3528672754837672,
      "learning_rate": 1.3888801125055156e-05,
      "loss": 0.8873,
      "step": 426
    },
    {
      "epoch": 1.31402732345584,
      "grad_norm": 0.311598618664693,
      "learning_rate": 1.3855773818640774e-05,
      "loss": 0.8856,
      "step": 427
    },
    {
      "epoch": 1.317106022705407,
      "grad_norm": 0.32956706146783693,
      "learning_rate": 1.3822697034282435e-05,
      "loss": 0.8661,
      "step": 428
    },
    {
      "epoch": 1.320184721954974,
      "grad_norm": 0.32514939875336313,
      "learning_rate": 1.3789571196427054e-05,
      "loss": 0.8808,
      "step": 429
    },
    {
      "epoch": 1.323263421204541,
      "grad_norm": 0.33888688027064223,
      "learning_rate": 1.3756396730150998e-05,
      "loss": 0.8946,
      "step": 430
    },
    {
      "epoch": 1.3263421204541082,
      "grad_norm": 0.3095587350783383,
      "learning_rate": 1.3723174061154652e-05,
      "loss": 0.8443,
      "step": 431
    },
    {
      "epoch": 1.3294208197036752,
      "grad_norm": 0.31193130766365595,
      "learning_rate": 1.3689903615756941e-05,
      "loss": 0.8793,
      "step": 432
    },
    {
      "epoch": 1.3324995189532423,
      "grad_norm": 0.3170840414920221,
      "learning_rate": 1.3656585820889868e-05,
      "loss": 0.857,
      "step": 433
    },
    {
      "epoch": 1.3355782182028093,
      "grad_norm": 0.3230813661396788,
      "learning_rate": 1.3623221104093025e-05,
      "loss": 0.8731,
      "step": 434
    },
    {
      "epoch": 1.3386569174523764,
      "grad_norm": 0.3255983812711661,
      "learning_rate": 1.3589809893508128e-05,
      "loss": 0.8477,
      "step": 435
    },
    {
      "epoch": 1.3417356167019434,
      "grad_norm": 0.3022244037149595,
      "learning_rate": 1.3556352617873492e-05,
      "loss": 0.8766,
      "step": 436
    },
    {
      "epoch": 1.3448143159515105,
      "grad_norm": 0.32831756582394367,
      "learning_rate": 1.3522849706518566e-05,
      "loss": 0.8861,
      "step": 437
    },
    {
      "epoch": 1.3478930152010775,
      "grad_norm": 0.3355154446903635,
      "learning_rate": 1.3489301589358385e-05,
      "loss": 0.8489,
      "step": 438
    },
    {
      "epoch": 1.3509717144506446,
      "grad_norm": 0.3125685639116911,
      "learning_rate": 1.3455708696888086e-05,
      "loss": 0.8519,
      "step": 439
    },
    {
      "epoch": 1.3540504137002116,
      "grad_norm": 0.30815057891030156,
      "learning_rate": 1.3422071460177361e-05,
      "loss": 0.8721,
      "step": 440
    },
    {
      "epoch": 1.3571291129497787,
      "grad_norm": 0.30324383855673887,
      "learning_rate": 1.3388390310864945e-05,
      "loss": 0.8429,
      "step": 441
    },
    {
      "epoch": 1.3602078121993457,
      "grad_norm": 0.3019168880249637,
      "learning_rate": 1.3354665681153061e-05,
      "loss": 0.8955,
      "step": 442
    },
    {
      "epoch": 1.3632865114489128,
      "grad_norm": 0.3233304476988152,
      "learning_rate": 1.3320898003801878e-05,
      "loss": 0.9209,
      "step": 443
    },
    {
      "epoch": 1.3663652106984798,
      "grad_norm": 0.33307045190706225,
      "learning_rate": 1.3287087712123963e-05,
      "loss": 0.8418,
      "step": 444
    },
    {
      "epoch": 1.3694439099480469,
      "grad_norm": 0.29878285384131975,
      "learning_rate": 1.3253235239978716e-05,
      "loss": 0.8769,
      "step": 445
    },
    {
      "epoch": 1.372522609197614,
      "grad_norm": 0.3181518276566564,
      "learning_rate": 1.3219341021766803e-05,
      "loss": 0.8699,
      "step": 446
    },
    {
      "epoch": 1.3756013084471812,
      "grad_norm": 0.321250923717796,
      "learning_rate": 1.318540549242459e-05,
      "loss": 0.899,
      "step": 447
    },
    {
      "epoch": 1.3786800076967483,
      "grad_norm": 0.3170563313084402,
      "learning_rate": 1.3151429087418542e-05,
      "loss": 0.9267,
      "step": 448
    },
    {
      "epoch": 1.3817587069463153,
      "grad_norm": 0.33418913025096886,
      "learning_rate": 1.3117412242739655e-05,
      "loss": 0.9033,
      "step": 449
    },
    {
      "epoch": 1.3848374061958824,
      "grad_norm": 0.32401292815551974,
      "learning_rate": 1.3083355394897852e-05,
      "loss": 0.8998,
      "step": 450
    },
    {
      "epoch": 1.3879161054454494,
      "grad_norm": 0.3126335371280449,
      "learning_rate": 1.3049258980916387e-05,
      "loss": 0.8413,
      "step": 451
    },
    {
      "epoch": 1.3909948046950165,
      "grad_norm": 0.30682296672433884,
      "learning_rate": 1.3015123438326223e-05,
      "loss": 0.9014,
      "step": 452
    },
    {
      "epoch": 1.3940735039445835,
      "grad_norm": 0.3213865171795104,
      "learning_rate": 1.2980949205160447e-05,
      "loss": 0.8468,
      "step": 453
    },
    {
      "epoch": 1.3971522031941506,
      "grad_norm": 0.3145610487499757,
      "learning_rate": 1.2946736719948607e-05,
      "loss": 0.8599,
      "step": 454
    },
    {
      "epoch": 1.4002309024437176,
      "grad_norm": 0.32053666828416755,
      "learning_rate": 1.2912486421711128e-05,
      "loss": 0.8574,
      "step": 455
    },
    {
      "epoch": 1.4033096016932847,
      "grad_norm": 0.31408127509345524,
      "learning_rate": 1.2878198749953642e-05,
      "loss": 0.8813,
      "step": 456
    },
    {
      "epoch": 1.4063883009428517,
      "grad_norm": 0.31771593612737364,
      "learning_rate": 1.2843874144661372e-05,
      "loss": 0.9023,
      "step": 457
    },
    {
      "epoch": 1.4094670001924188,
      "grad_norm": 0.3224416006849034,
      "learning_rate": 1.2809513046293485e-05,
      "loss": 0.87,
      "step": 458
    },
    {
      "epoch": 1.4125456994419858,
      "grad_norm": 0.3035429513747013,
      "learning_rate": 1.2775115895777417e-05,
      "loss": 0.85,
      "step": 459
    },
    {
      "epoch": 1.4156243986915529,
      "grad_norm": 0.30304921193330653,
      "learning_rate": 1.2740683134503248e-05,
      "loss": 0.8297,
      "step": 460
    },
    {
      "epoch": 1.41870309794112,
      "grad_norm": 0.3080332293930409,
      "learning_rate": 1.270621520431801e-05,
      "loss": 0.8727,
      "step": 461
    },
    {
      "epoch": 1.421781797190687,
      "grad_norm": 0.30779650164992445,
      "learning_rate": 1.2671712547520033e-05,
      "loss": 0.8806,
      "step": 462
    },
    {
      "epoch": 1.424860496440254,
      "grad_norm": 0.3106302008055908,
      "learning_rate": 1.2637175606853265e-05,
      "loss": 0.8798,
      "step": 463
    },
    {
      "epoch": 1.427939195689821,
      "grad_norm": 0.3122426723405643,
      "learning_rate": 1.2602604825501588e-05,
      "loss": 0.8734,
      "step": 464
    },
    {
      "epoch": 1.4310178949393881,
      "grad_norm": 0.30312477877743715,
      "learning_rate": 1.2568000647083131e-05,
      "loss": 0.9195,
      "step": 465
    },
    {
      "epoch": 1.4340965941889552,
      "grad_norm": 0.3202594779944576,
      "learning_rate": 1.2533363515644595e-05,
      "loss": 0.8842,
      "step": 466
    },
    {
      "epoch": 1.4371752934385222,
      "grad_norm": 0.32225481373732046,
      "learning_rate": 1.2498693875655517e-05,
      "loss": 0.9175,
      "step": 467
    },
    {
      "epoch": 1.4402539926880893,
      "grad_norm": 0.3148776367548769,
      "learning_rate": 1.2463992172002606e-05,
      "loss": 0.8817,
      "step": 468
    },
    {
      "epoch": 1.4433326919376563,
      "grad_norm": 0.3081495243525174,
      "learning_rate": 1.2429258849984013e-05,
      "loss": 0.8807,
      "step": 469
    },
    {
      "epoch": 1.4464113911872234,
      "grad_norm": 0.32039782590382654,
      "learning_rate": 1.2394494355303616e-05,
      "loss": 0.8602,
      "step": 470
    },
    {
      "epoch": 1.4494900904367904,
      "grad_norm": 0.30779622644928506,
      "learning_rate": 1.2359699134065316e-05,
      "loss": 0.8909,
      "step": 471
    },
    {
      "epoch": 1.4525687896863575,
      "grad_norm": 0.3139040893384437,
      "learning_rate": 1.2324873632767292e-05,
      "loss": 0.8851,
      "step": 472
    },
    {
      "epoch": 1.4556474889359245,
      "grad_norm": 0.2993601115366549,
      "learning_rate": 1.2290018298296285e-05,
      "loss": 0.8958,
      "step": 473
    },
    {
      "epoch": 1.4587261881854916,
      "grad_norm": 0.32124966786683795,
      "learning_rate": 1.225513357792187e-05,
      "loss": 0.8672,
      "step": 474
    },
    {
      "epoch": 1.4618048874350587,
      "grad_norm": 0.31648247030247395,
      "learning_rate": 1.2220219919290688e-05,
      "loss": 0.9023,
      "step": 475
    },
    {
      "epoch": 1.4648835866846257,
      "grad_norm": 0.30861179129568306,
      "learning_rate": 1.2185277770420739e-05,
      "loss": 0.8646,
      "step": 476
    },
    {
      "epoch": 1.4679622859341928,
      "grad_norm": 0.31428436576394503,
      "learning_rate": 1.2150307579695602e-05,
      "loss": 0.8736,
      "step": 477
    },
    {
      "epoch": 1.4710409851837598,
      "grad_norm": 0.28552571951126965,
      "learning_rate": 1.2115309795858705e-05,
      "loss": 0.875,
      "step": 478
    },
    {
      "epoch": 1.4741196844333269,
      "grad_norm": 0.3154082716893484,
      "learning_rate": 1.2080284868007541e-05,
      "loss": 0.8863,
      "step": 479
    },
    {
      "epoch": 1.477198383682894,
      "grad_norm": 0.3199971135554803,
      "learning_rate": 1.2045233245587938e-05,
      "loss": 0.8397,
      "step": 480
    },
    {
      "epoch": 1.480277082932461,
      "grad_norm": 0.3067319371344443,
      "learning_rate": 1.2010155378388253e-05,
      "loss": 0.8304,
      "step": 481
    },
    {
      "epoch": 1.483355782182028,
      "grad_norm": 0.309132437682419,
      "learning_rate": 1.1975051716533636e-05,
      "loss": 0.876,
      "step": 482
    },
    {
      "epoch": 1.486434481431595,
      "grad_norm": 0.3083854116132613,
      "learning_rate": 1.1939922710480229e-05,
      "loss": 0.8818,
      "step": 483
    },
    {
      "epoch": 1.4895131806811621,
      "grad_norm": 0.32913256788649264,
      "learning_rate": 1.1904768811009406e-05,
      "loss": 0.897,
      "step": 484
    },
    {
      "epoch": 1.4925918799307292,
      "grad_norm": 0.3093107562410849,
      "learning_rate": 1.1869590469221965e-05,
      "loss": 0.8983,
      "step": 485
    },
    {
      "epoch": 1.4956705791802962,
      "grad_norm": 0.3294462460497043,
      "learning_rate": 1.1834388136532358e-05,
      "loss": 0.8675,
      "step": 486
    },
    {
      "epoch": 1.4987492784298633,
      "grad_norm": 0.30654975739245144,
      "learning_rate": 1.1799162264662891e-05,
      "loss": 0.8729,
      "step": 487
    },
    {
      "epoch": 1.5018279776794303,
      "grad_norm": 0.3102363720158079,
      "learning_rate": 1.1763913305637932e-05,
      "loss": 0.84,
      "step": 488
    },
    {
      "epoch": 1.5049066769289974,
      "grad_norm": 0.28698818483688415,
      "learning_rate": 1.1728641711778103e-05,
      "loss": 0.8961,
      "step": 489
    },
    {
      "epoch": 1.5079853761785644,
      "grad_norm": 0.3300283420988258,
      "learning_rate": 1.1693347935694485e-05,
      "loss": 0.8903,
      "step": 490
    },
    {
      "epoch": 1.5110640754281315,
      "grad_norm": 0.3162371178804905,
      "learning_rate": 1.16580324302828e-05,
      "loss": 0.8707,
      "step": 491
    },
    {
      "epoch": 1.5141427746776985,
      "grad_norm": 0.31346523981617114,
      "learning_rate": 1.1622695648717604e-05,
      "loss": 0.9063,
      "step": 492
    },
    {
      "epoch": 1.5172214739272656,
      "grad_norm": 0.30673883160717863,
      "learning_rate": 1.1587338044446477e-05,
      "loss": 0.8743,
      "step": 493
    },
    {
      "epoch": 1.5203001731768326,
      "grad_norm": 0.32312493902434325,
      "learning_rate": 1.1551960071184196e-05,
      "loss": 0.9046,
      "step": 494
    },
    {
      "epoch": 1.5233788724263997,
      "grad_norm": 0.31238223058346687,
      "learning_rate": 1.1516562182906923e-05,
      "loss": 0.8928,
      "step": 495
    },
    {
      "epoch": 1.5264575716759667,
      "grad_norm": 0.2931603490561117,
      "learning_rate": 1.1481144833846358e-05,
      "loss": 0.853,
      "step": 496
    },
    {
      "epoch": 1.529536270925534,
      "grad_norm": 0.3032859629126185,
      "learning_rate": 1.1445708478483941e-05,
      "loss": 0.9083,
      "step": 497
    },
    {
      "epoch": 1.532614970175101,
      "grad_norm": 0.2940257638122967,
      "learning_rate": 1.1410253571544993e-05,
      "loss": 0.8397,
      "step": 498
    },
    {
      "epoch": 1.5356936694246681,
      "grad_norm": 0.3122112103197989,
      "learning_rate": 1.13747805679929e-05,
      "loss": 0.7894,
      "step": 499
    },
    {
      "epoch": 1.5387723686742352,
      "grad_norm": 0.2973699105741789,
      "learning_rate": 1.1339289923023263e-05,
      "loss": 0.8842,
      "step": 500
    },
    {
      "epoch": 1.5418510679238022,
      "grad_norm": 0.3188287086528435,
      "learning_rate": 1.1303782092058062e-05,
      "loss": 0.8619,
      "step": 501
    },
    {
      "epoch": 1.5449297671733693,
      "grad_norm": 0.31773876297130843,
      "learning_rate": 1.1268257530739804e-05,
      "loss": 0.8658,
      "step": 502
    },
    {
      "epoch": 1.5480084664229363,
      "grad_norm": 0.3725291259254551,
      "learning_rate": 1.1232716694925693e-05,
      "loss": 0.913,
      "step": 503
    },
    {
      "epoch": 1.5510871656725034,
      "grad_norm": 0.3150460039324213,
      "learning_rate": 1.1197160040681763e-05,
      "loss": 0.8812,
      "step": 504
    },
    {
      "epoch": 1.5541658649220704,
      "grad_norm": 0.30553118471185003,
      "learning_rate": 1.1161588024277037e-05,
      "loss": 0.867,
      "step": 505
    },
    {
      "epoch": 1.5572445641716375,
      "grad_norm": 0.3094992932239147,
      "learning_rate": 1.1126001102177667e-05,
      "loss": 0.8792,
      "step": 506
    },
    {
      "epoch": 1.5603232634212045,
      "grad_norm": 0.3192601991916358,
      "learning_rate": 1.1090399731041072e-05,
      "loss": 0.8822,
      "step": 507
    },
    {
      "epoch": 1.5634019626707716,
      "grad_norm": 0.3120073723786125,
      "learning_rate": 1.1054784367710092e-05,
      "loss": 0.8697,
      "step": 508
    },
    {
      "epoch": 1.5664806619203386,
      "grad_norm": 0.3106728762159597,
      "learning_rate": 1.101915546920711e-05,
      "loss": 0.8907,
      "step": 509
    },
    {
      "epoch": 1.5695593611699057,
      "grad_norm": 0.32453650840743337,
      "learning_rate": 1.09835134927282e-05,
      "loss": 0.8665,
      "step": 510
    },
    {
      "epoch": 1.5726380604194727,
      "grad_norm": 0.28641039718691014,
      "learning_rate": 1.0947858895637255e-05,
      "loss": 0.8378,
      "step": 511
    },
    {
      "epoch": 1.5757167596690398,
      "grad_norm": 0.28784784711548145,
      "learning_rate": 1.091219213546011e-05,
      "loss": 0.8302,
      "step": 512
    },
    {
      "epoch": 1.5787954589186068,
      "grad_norm": 0.2938812712413679,
      "learning_rate": 1.0876513669878683e-05,
      "loss": 0.8569,
      "step": 513
    },
    {
      "epoch": 1.581874158168174,
      "grad_norm": 0.2991003577210808,
      "learning_rate": 1.0840823956725104e-05,
      "loss": 0.8257,
      "step": 514
    },
    {
      "epoch": 1.5849528574177412,
      "grad_norm": 0.2891216592413506,
      "learning_rate": 1.080512345397583e-05,
      "loss": 0.8017,
      "step": 515
    },
    {
      "epoch": 1.5880315566673082,
      "grad_norm": 0.2900795042276432,
      "learning_rate": 1.0769412619745762e-05,
      "loss": 0.9126,
      "step": 516
    },
    {
      "epoch": 1.5911102559168753,
      "grad_norm": 0.29547567136153025,
      "learning_rate": 1.0733691912282396e-05,
      "loss": 0.8303,
      "step": 517
    },
    {
      "epoch": 1.5941889551664423,
      "grad_norm": 0.2946801263686893,
      "learning_rate": 1.0697961789959903e-05,
      "loss": 0.8866,
      "step": 518
    },
    {
      "epoch": 1.5972676544160094,
      "grad_norm": 0.30460701651473854,
      "learning_rate": 1.0662222711273281e-05,
      "loss": 0.8689,
      "step": 519
    },
    {
      "epoch": 1.6003463536655764,
      "grad_norm": 0.2945313672992684,
      "learning_rate": 1.0626475134832448e-05,
      "loss": 0.8813,
      "step": 520
    },
    {
      "epoch": 1.6034250529151435,
      "grad_norm": 0.295373198270631,
      "learning_rate": 1.0590719519356373e-05,
      "loss": 0.9093,
      "step": 521
    },
    {
      "epoch": 1.6065037521647105,
      "grad_norm": 0.3028324040021512,
      "learning_rate": 1.0554956323667187e-05,
      "loss": 0.8721,
      "step": 522
    },
    {
      "epoch": 1.6095824514142776,
      "grad_norm": 0.304832559388564,
      "learning_rate": 1.0519186006684277e-05,
      "loss": 0.8871,
      "step": 523
    },
    {
      "epoch": 1.6126611506638446,
      "grad_norm": 0.3042551047057534,
      "learning_rate": 1.0483409027418426e-05,
      "loss": 0.8892,
      "step": 524
    },
    {
      "epoch": 1.6157398499134117,
      "grad_norm": 0.2824266662554397,
      "learning_rate": 1.04476258449659e-05,
      "loss": 0.9112,
      "step": 525
    },
    {
      "epoch": 1.6188185491629787,
      "grad_norm": 0.3023139375560548,
      "learning_rate": 1.0411836918502573e-05,
      "loss": 0.8366,
      "step": 526
    },
    {
      "epoch": 1.6218972484125458,
      "grad_norm": 0.3036693814531651,
      "learning_rate": 1.037604270727802e-05,
      "loss": 0.8892,
      "step": 527
    },
    {
      "epoch": 1.6249759476621128,
      "grad_norm": 0.3042992557555934,
      "learning_rate": 1.0340243670609636e-05,
      "loss": 0.8246,
      "step": 528
    },
    {
      "epoch": 1.62805464691168,
      "grad_norm": 0.2906266203671198,
      "learning_rate": 1.0304440267876728e-05,
      "loss": 0.8786,
      "step": 529
    },
    {
      "epoch": 1.631133346161247,
      "grad_norm": 0.31874610280968646,
      "learning_rate": 1.0268632958514641e-05,
      "loss": 0.9065,
      "step": 530
    },
    {
      "epoch": 1.634212045410814,
      "grad_norm": 0.31790497114177657,
      "learning_rate": 1.0232822202008845e-05,
      "loss": 0.8417,
      "step": 531
    },
    {
      "epoch": 1.637290744660381,
      "grad_norm": 0.3158827547312656,
      "learning_rate": 1.0197008457889045e-05,
      "loss": 0.8816,
      "step": 532
    },
    {
      "epoch": 1.640369443909948,
      "grad_norm": 0.3066682258603489,
      "learning_rate": 1.0161192185723282e-05,
      "loss": 0.8935,
      "step": 533
    },
    {
      "epoch": 1.6434481431595152,
      "grad_norm": 0.30740918668387035,
      "learning_rate": 1.0125373845112035e-05,
      "loss": 0.8492,
      "step": 534
    },
    {
      "epoch": 1.6465268424090822,
      "grad_norm": 0.30824464370641036,
      "learning_rate": 1.0089553895682331e-05,
      "loss": 0.9003,
      "step": 535
    },
    {
      "epoch": 1.6496055416586493,
      "grad_norm": 0.3030695407147231,
      "learning_rate": 1.0053732797081843e-05,
      "loss": 0.8715,
      "step": 536
    },
    {
      "epoch": 1.6526842409082163,
      "grad_norm": 0.3090578298233797,
      "learning_rate": 1.0017911008972982e-05,
      "loss": 0.8557,
      "step": 537
    },
    {
      "epoch": 1.6557629401577834,
      "grad_norm": 0.3022067180751946,
      "learning_rate": 9.98208899102702e-06,
      "loss": 0.8617,
      "step": 538
    },
    {
      "epoch": 1.6588416394073504,
      "grad_norm": 0.28585613041700164,
      "learning_rate": 9.946267202918157e-06,
      "loss": 0.9085,
      "step": 539
    },
    {
      "epoch": 1.6619203386569175,
      "grad_norm": 0.31601244383841565,
      "learning_rate": 9.91044610431767e-06,
      "loss": 0.8356,
      "step": 540
    },
    {
      "epoch": 1.6649990379064845,
      "grad_norm": 0.294533971117609,
      "learning_rate": 9.87462615488797e-06,
      "loss": 0.9026,
      "step": 541
    },
    {
      "epoch": 1.6680777371560516,
      "grad_norm": 0.32858362488357395,
      "learning_rate": 9.838807814276723e-06,
      "loss": 0.8471,
      "step": 542
    },
    {
      "epoch": 1.6711564364056186,
      "grad_norm": 0.3126508657036644,
      "learning_rate": 9.802991542110959e-06,
      "loss": 0.8586,
      "step": 543
    },
    {
      "epoch": 1.6742351356551857,
      "grad_norm": 0.305212114850939,
      "learning_rate": 9.767177797991155e-06,
      "loss": 0.8693,
      "step": 544
    },
    {
      "epoch": 1.6773138349047527,
      "grad_norm": 0.30238063464176523,
      "learning_rate": 9.73136704148536e-06,
      "loss": 0.8347,
      "step": 545
    },
    {
      "epoch": 1.6803925341543198,
      "grad_norm": 0.2879528202820905,
      "learning_rate": 9.695559732123275e-06,
      "loss": 0.8558,
      "step": 546
    },
    {
      "epoch": 1.6834712334038868,
      "grad_norm": 0.36354481203148437,
      "learning_rate": 9.659756329390368e-06,
      "loss": 0.8341,
      "step": 547
    },
    {
      "epoch": 1.6865499326534539,
      "grad_norm": 0.29198844358892345,
      "learning_rate": 9.623957292721981e-06,
      "loss": 0.89,
      "step": 548
    },
    {
      "epoch": 1.689628631903021,
      "grad_norm": 0.31401879023688667,
      "learning_rate": 9.588163081497427e-06,
      "loss": 0.885,
      "step": 549
    },
    {
      "epoch": 1.692707331152588,
      "grad_norm": 0.30572889910141404,
      "learning_rate": 9.552374155034101e-06,
      "loss": 0.9078,
      "step": 550
    },
    {
      "epoch": 1.695786030402155,
      "grad_norm": 0.300557244996185,
      "learning_rate": 9.516590972581579e-06,
      "loss": 0.8661,
      "step": 551
    },
    {
      "epoch": 1.698864729651722,
      "grad_norm": 0.28835248944539604,
      "learning_rate": 9.480813993315727e-06,
      "loss": 0.8467,
      "step": 552
    },
    {
      "epoch": 1.7019434289012891,
      "grad_norm": 0.30045767852839034,
      "learning_rate": 9.44504367633282e-06,
      "loss": 0.9156,
      "step": 553
    },
    {
      "epoch": 1.7050221281508562,
      "grad_norm": 0.29583657388045703,
      "learning_rate": 9.409280480643628e-06,
      "loss": 0.8883,
      "step": 554
    },
    {
      "epoch": 1.7081008274004232,
      "grad_norm": 0.2958140058131901,
      "learning_rate": 9.373524865167557e-06,
      "loss": 0.821,
      "step": 555
    },
    {
      "epoch": 1.7111795266499903,
      "grad_norm": 0.29140622959073614,
      "learning_rate": 9.337777288726722e-06,
      "loss": 0.8613,
      "step": 556
    },
    {
      "epoch": 1.7142582258995573,
      "grad_norm": 0.2973485503104455,
      "learning_rate": 9.302038210040099e-06,
      "loss": 0.8631,
      "step": 557
    },
    {
      "epoch": 1.7173369251491244,
      "grad_norm": 0.2998669194936975,
      "learning_rate": 9.266308087717609e-06,
      "loss": 0.8654,
      "step": 558
    },
    {
      "epoch": 1.7204156243986914,
      "grad_norm": 0.2892936770745875,
      "learning_rate": 9.230587380254238e-06,
      "loss": 0.8978,
      "step": 559
    },
    {
      "epoch": 1.7234943236482585,
      "grad_norm": 0.30206285829577895,
      "learning_rate": 9.194876546024174e-06,
      "loss": 0.8852,
      "step": 560
    },
    {
      "epoch": 1.7265730228978255,
      "grad_norm": 0.2858317744825158,
      "learning_rate": 9.159176043274896e-06,
      "loss": 0.8537,
      "step": 561
    },
    {
      "epoch": 1.7296517221473926,
      "grad_norm": 0.305130027988899,
      "learning_rate": 9.123486330121318e-06,
      "loss": 0.8721,
      "step": 562
    },
    {
      "epoch": 1.7327304213969597,
      "grad_norm": 0.3021117368569213,
      "learning_rate": 9.087807864539897e-06,
      "loss": 0.8778,
      "step": 563
    },
    {
      "epoch": 1.7358091206465267,
      "grad_norm": 0.2876839937732219,
      "learning_rate": 9.052141104362748e-06,
      "loss": 0.877,
      "step": 564
    },
    {
      "epoch": 1.7388878198960938,
      "grad_norm": 0.28260740291239217,
      "learning_rate": 9.016486507271804e-06,
      "loss": 0.8531,
      "step": 565
    },
    {
      "epoch": 1.7419665191456608,
      "grad_norm": 0.2962333027934723,
      "learning_rate": 8.980844530792889e-06,
      "loss": 0.8969,
      "step": 566
    },
    {
      "epoch": 1.7450452183952279,
      "grad_norm": 0.30024435181377557,
      "learning_rate": 8.945215632289912e-06,
      "loss": 0.8449,
      "step": 567
    },
    {
      "epoch": 1.748123917644795,
      "grad_norm": 0.2942824119343657,
      "learning_rate": 8.909600268958933e-06,
      "loss": 0.8214,
      "step": 568
    },
    {
      "epoch": 1.7512026168943622,
      "grad_norm": 0.2761462277196388,
      "learning_rate": 8.873998897822337e-06,
      "loss": 0.8629,
      "step": 569
    },
    {
      "epoch": 1.7542813161439292,
      "grad_norm": 0.29115231077412185,
      "learning_rate": 8.838411975722966e-06,
      "loss": 0.8361,
      "step": 570
    },
    {
      "epoch": 1.7573600153934963,
      "grad_norm": 0.2881144082844479,
      "learning_rate": 8.802839959318238e-06,
      "loss": 0.8366,
      "step": 571
    },
    {
      "epoch": 1.7604387146430633,
      "grad_norm": 0.2978578093438894,
      "learning_rate": 8.76728330507431e-06,
      "loss": 0.9007,
      "step": 572
    },
    {
      "epoch": 1.7635174138926304,
      "grad_norm": 0.31701373474962163,
      "learning_rate": 8.731742469260201e-06,
      "loss": 0.8635,
      "step": 573
    },
    {
      "epoch": 1.7665961131421974,
      "grad_norm": 0.2849825755871595,
      "learning_rate": 8.696217907941941e-06,
      "loss": 0.8504,
      "step": 574
    },
    {
      "epoch": 1.7696748123917645,
      "grad_norm": 0.28601131597700175,
      "learning_rate": 8.66071007697674e-06,
      "loss": 0.8637,
      "step": 575
    },
    {
      "epoch": 1.7727535116413315,
      "grad_norm": 0.30262289832991585,
      "learning_rate": 8.6252194320071e-06,
      "loss": 0.8564,
      "step": 576
    },
    {
      "epoch": 1.7758322108908986,
      "grad_norm": 0.2899453740238269,
      "learning_rate": 8.58974642845501e-06,
      "loss": 0.8825,
      "step": 577
    },
    {
      "epoch": 1.7789109101404657,
      "grad_norm": 0.30361400703583946,
      "learning_rate": 8.554291521516064e-06,
      "loss": 0.8786,
      "step": 578
    },
    {
      "epoch": 1.7819896093900327,
      "grad_norm": 0.3049770256191141,
      "learning_rate": 8.518855166153645e-06,
      "loss": 0.8732,
      "step": 579
    },
    {
      "epoch": 1.7850683086395998,
      "grad_norm": 0.28063072021335805,
      "learning_rate": 8.483437817093082e-06,
      "loss": 0.891,
      "step": 580
    },
    {
      "epoch": 1.7881470078891668,
      "grad_norm": 0.28924540819724887,
      "learning_rate": 8.448039928815804e-06,
      "loss": 0.8284,
      "step": 581
    },
    {
      "epoch": 1.7912257071387339,
      "grad_norm": 0.28626742312264597,
      "learning_rate": 8.412661955553524e-06,
      "loss": 0.8544,
      "step": 582
    },
    {
      "epoch": 1.794304406388301,
      "grad_norm": 0.28803633337269535,
      "learning_rate": 8.3773043512824e-06,
      "loss": 0.8472,
      "step": 583
    },
    {
      "epoch": 1.797383105637868,
      "grad_norm": 0.4060779108058977,
      "learning_rate": 8.341967569717203e-06,
      "loss": 0.8702,
      "step": 584
    },
    {
      "epoch": 1.800461804887435,
      "grad_norm": 0.302289242398114,
      "learning_rate": 8.306652064305517e-06,
      "loss": 0.8962,
      "step": 585
    },
    {
      "epoch": 1.803540504137002,
      "grad_norm": 0.29192617728890813,
      "learning_rate": 8.271358288221897e-06,
      "loss": 0.855,
      "step": 586
    },
    {
      "epoch": 1.8066192033865693,
      "grad_norm": 0.28580088678781107,
      "learning_rate": 8.23608669436207e-06,
      "loss": 0.8679,
      "step": 587
    },
    {
      "epoch": 1.8096979026361364,
      "grad_norm": 0.2970404247783746,
      "learning_rate": 8.200837735337114e-06,
      "loss": 0.8743,
      "step": 588
    },
    {
      "epoch": 1.8127766018857034,
      "grad_norm": 0.28989129719435497,
      "learning_rate": 8.165611863467645e-06,
      "loss": 0.8644,
      "step": 589
    },
    {
      "epoch": 1.8158553011352705,
      "grad_norm": 0.3020112959495261,
      "learning_rate": 8.13040953077804e-06,
      "loss": 0.8965,
      "step": 590
    },
    {
      "epoch": 1.8189340003848375,
      "grad_norm": 0.2939848161836744,
      "learning_rate": 8.095231188990597e-06,
      "loss": 0.8818,
      "step": 591
    },
    {
      "epoch": 1.8220126996344046,
      "grad_norm": 0.3227796278800699,
      "learning_rate": 8.060077289519773e-06,
      "loss": 0.8904,
      "step": 592
    },
    {
      "epoch": 1.8250913988839716,
      "grad_norm": 0.2832074112814755,
      "learning_rate": 8.024948283466367e-06,
      "loss": 0.8461,
      "step": 593
    },
    {
      "epoch": 1.8281700981335387,
      "grad_norm": 0.29965049703850605,
      "learning_rate": 7.98984462161175e-06,
      "loss": 0.8863,
      "step": 594
    },
    {
      "epoch": 1.8312487973831058,
      "grad_norm": 0.2869783010432895,
      "learning_rate": 7.954766754412067e-06,
      "loss": 0.8961,
      "step": 595
    },
    {
      "epoch": 1.8343274966326728,
      "grad_norm": 0.299876848195199,
      "learning_rate": 7.919715131992459e-06,
      "loss": 0.8687,
      "step": 596
    },
    {
      "epoch": 1.8374061958822399,
      "grad_norm": 0.29459400534329894,
      "learning_rate": 7.884690204141298e-06,
      "loss": 0.9044,
      "step": 597
    },
    {
      "epoch": 1.840484895131807,
      "grad_norm": 0.29972256141999637,
      "learning_rate": 7.849692420304398e-06,
      "loss": 0.8526,
      "step": 598
    },
    {
      "epoch": 1.843563594381374,
      "grad_norm": 0.2864776156670731,
      "learning_rate": 7.814722229579263e-06,
      "loss": 0.8598,
      "step": 599
    },
    {
      "epoch": 1.846642293630941,
      "grad_norm": 0.2869249790305462,
      "learning_rate": 7.779780080709317e-06,
      "loss": 0.8893,
      "step": 600
    },
    {
      "epoch": 1.849720992880508,
      "grad_norm": 0.2950908959781418,
      "learning_rate": 7.744866422078133e-06,
      "loss": 0.8512,
      "step": 601
    },
    {
      "epoch": 1.8527996921300751,
      "grad_norm": 0.29873784813612286,
      "learning_rate": 7.709981701703719e-06,
      "loss": 0.851,
      "step": 602
    },
    {
      "epoch": 1.8558783913796422,
      "grad_norm": 0.29577638493873193,
      "learning_rate": 7.67512636723271e-06,
      "loss": 0.8487,
      "step": 603
    },
    {
      "epoch": 1.8589570906292092,
      "grad_norm": 0.29717100333736324,
      "learning_rate": 7.640300865934688e-06,
      "loss": 0.8493,
      "step": 604
    },
    {
      "epoch": 1.8620357898787763,
      "grad_norm": 0.27936018482047204,
      "learning_rate": 7.605505644696387e-06,
      "loss": 0.8646,
      "step": 605
    },
    {
      "epoch": 1.8651144891283433,
      "grad_norm": 0.3003787289879184,
      "learning_rate": 7.57074115001599e-06,
      "loss": 0.8856,
      "step": 606
    },
    {
      "epoch": 1.8681931883779104,
      "grad_norm": 0.31296251426359023,
      "learning_rate": 7.536007827997397e-06,
      "loss": 0.8131,
      "step": 607
    },
    {
      "epoch": 1.8712718876274774,
      "grad_norm": 0.2835084924989488,
      "learning_rate": 7.501306124344485e-06,
      "loss": 0.8644,
      "step": 608
    },
    {
      "epoch": 1.8743505868770445,
      "grad_norm": 0.2880445633552251,
      "learning_rate": 7.4666364843554095e-06,
      "loss": 0.8497,
      "step": 609
    },
    {
      "epoch": 1.8774292861266115,
      "grad_norm": 0.30170698934479745,
      "learning_rate": 7.43199935291687e-06,
      "loss": 0.8793,
      "step": 610
    },
    {
      "epoch": 1.8805079853761786,
      "grad_norm": 0.30144285088838757,
      "learning_rate": 7.397395174498416e-06,
      "loss": 0.8779,
      "step": 611
    },
    {
      "epoch": 1.8835866846257456,
      "grad_norm": 0.291682942551406,
      "learning_rate": 7.362824393146739e-06,
      "loss": 0.8468,
      "step": 612
    },
    {
      "epoch": 1.8866653838753127,
      "grad_norm": 0.297371165603849,
      "learning_rate": 7.328287452479968e-06,
      "loss": 0.8799,
      "step": 613
    },
    {
      "epoch": 1.8897440831248797,
      "grad_norm": 0.3148503050800278,
      "learning_rate": 7.293784795681994e-06,
      "loss": 0.8557,
      "step": 614
    },
    {
      "epoch": 1.8928227823744468,
      "grad_norm": 0.29911717977430063,
      "learning_rate": 7.2593168654967575e-06,
      "loss": 0.8705,
      "step": 615
    },
    {
      "epoch": 1.8959014816240138,
      "grad_norm": 0.2753300107700702,
      "learning_rate": 7.224884104222585e-06,
      "loss": 0.8732,
      "step": 616
    },
    {
      "epoch": 1.898980180873581,
      "grad_norm": 0.2967906721911725,
      "learning_rate": 7.19048695370652e-06,
      "loss": 0.8385,
      "step": 617
    },
    {
      "epoch": 1.902058880123148,
      "grad_norm": 0.30199987444272564,
      "learning_rate": 7.15612585533863e-06,
      "loss": 0.8935,
      "step": 618
    },
    {
      "epoch": 1.905137579372715,
      "grad_norm": 0.31454143166500004,
      "learning_rate": 7.121801250046363e-06,
      "loss": 0.8597,
      "step": 619
    },
    {
      "epoch": 1.908216278622282,
      "grad_norm": 0.2848603829649693,
      "learning_rate": 7.087513578288878e-06,
      "loss": 0.8696,
      "step": 620
    },
    {
      "epoch": 1.911294977871849,
      "grad_norm": 0.31187715381246334,
      "learning_rate": 7.053263280051394e-06,
      "loss": 0.863,
      "step": 621
    },
    {
      "epoch": 1.9143736771214162,
      "grad_norm": 0.29913327401657797,
      "learning_rate": 7.019050794839557e-06,
      "loss": 0.8753,
      "step": 622
    },
    {
      "epoch": 1.9174523763709832,
      "grad_norm": 0.2979235662627265,
      "learning_rate": 6.984876561673776e-06,
      "loss": 0.859,
      "step": 623
    },
    {
      "epoch": 1.9205310756205503,
      "grad_norm": 0.2891669972098585,
      "learning_rate": 6.9507410190836174e-06,
      "loss": 0.8482,
      "step": 624
    },
    {
      "epoch": 1.9236097748701173,
      "grad_norm": 0.2816793796175202,
      "learning_rate": 6.9166446051021495e-06,
      "loss": 0.8633,
      "step": 625
    },
    {
      "epoch": 1.9266884741196844,
      "grad_norm": 0.28649513282889255,
      "learning_rate": 6.882587757260349e-06,
      "loss": 0.8462,
      "step": 626
    },
    {
      "epoch": 1.9297671733692514,
      "grad_norm": 0.28926096192888073,
      "learning_rate": 6.848570912581464e-06,
      "loss": 0.8142,
      "step": 627
    },
    {
      "epoch": 1.9328458726188185,
      "grad_norm": 0.2853168938909875,
      "learning_rate": 6.814594507575412e-06,
      "loss": 0.8395,
      "step": 628
    },
    {
      "epoch": 1.9359245718683855,
      "grad_norm": 0.2887361261434629,
      "learning_rate": 6.7806589782331985e-06,
      "loss": 0.901,
      "step": 629
    },
    {
      "epoch": 1.9390032711179526,
      "grad_norm": 0.32960267359210593,
      "learning_rate": 6.746764760021287e-06,
      "loss": 0.8674,
      "step": 630
    },
    {
      "epoch": 1.9420819703675196,
      "grad_norm": 0.2852642721618185,
      "learning_rate": 6.712912287876041e-06,
      "loss": 0.8992,
      "step": 631
    },
    {
      "epoch": 1.9451606696170867,
      "grad_norm": 0.2891421143873303,
      "learning_rate": 6.679101996198126e-06,
      "loss": 0.8719,
      "step": 632
    },
    {
      "epoch": 1.9482393688666537,
      "grad_norm": 0.30434740981455694,
      "learning_rate": 6.645334318846941e-06,
      "loss": 0.8457,
      "step": 633
    },
    {
      "epoch": 1.9513180681162208,
      "grad_norm": 0.28638065802450147,
      "learning_rate": 6.611609689135057e-06,
      "loss": 0.8733,
      "step": 634
    },
    {
      "epoch": 1.9543967673657878,
      "grad_norm": 0.29657313867012225,
      "learning_rate": 6.57792853982264e-06,
      "loss": 0.8293,
      "step": 635
    },
    {
      "epoch": 1.9574754666153549,
      "grad_norm": 0.29060575252564846,
      "learning_rate": 6.544291303111918e-06,
      "loss": 0.8909,
      "step": 636
    },
    {
      "epoch": 1.960554165864922,
      "grad_norm": 0.2912800610260448,
      "learning_rate": 6.510698410641621e-06,
      "loss": 0.872,
      "step": 637
    },
    {
      "epoch": 1.963632865114489,
      "grad_norm": 0.289541100021876,
      "learning_rate": 6.477150293481435e-06,
      "loss": 0.864,
      "step": 638
    },
    {
      "epoch": 1.966711564364056,
      "grad_norm": 0.2981027550677072,
      "learning_rate": 6.44364738212651e-06,
      "loss": 0.8601,
      "step": 639
    },
    {
      "epoch": 1.969790263613623,
      "grad_norm": 0.5095873767791187,
      "learning_rate": 6.410190106491874e-06,
      "loss": 0.8561,
      "step": 640
    },
    {
      "epoch": 1.9728689628631904,
      "grad_norm": 0.303957493116429,
      "learning_rate": 6.3767788959069765e-06,
      "loss": 0.8707,
      "step": 641
    },
    {
      "epoch": 1.9759476621127574,
      "grad_norm": 0.29522839267330486,
      "learning_rate": 6.343414179110136e-06,
      "loss": 0.8589,
      "step": 642
    },
    {
      "epoch": 1.9790263613623245,
      "grad_norm": 0.27418226060019785,
      "learning_rate": 6.310096384243061e-06,
      "loss": 0.864,
      "step": 643
    },
    {
      "epoch": 1.9821050606118915,
      "grad_norm": 0.2821263623187557,
      "learning_rate": 6.27682593884535e-06,
      "loss": 0.8931,
      "step": 644
    },
    {
      "epoch": 1.9851837598614586,
      "grad_norm": 0.28337448527838904,
      "learning_rate": 6.243603269849004e-06,
      "loss": 0.8786,
      "step": 645
    },
    {
      "epoch": 1.9882624591110256,
      "grad_norm": 0.2761426036428095,
      "learning_rate": 6.210428803572949e-06,
      "loss": 0.8249,
      "step": 646
    },
    {
      "epoch": 1.9913411583605927,
      "grad_norm": 0.27754458868905746,
      "learning_rate": 6.177302965717566e-06,
      "loss": 0.8117,
      "step": 647
    },
    {
      "epoch": 1.9944198576101597,
      "grad_norm": 0.27280227244225397,
      "learning_rate": 6.144226181359227e-06,
      "loss": 0.8801,
      "step": 648
    },
    {
      "epoch": 1.9974985568597268,
      "grad_norm": 0.28447360096898316,
      "learning_rate": 6.111198874944846e-06,
      "loss": 0.886,
      "step": 649
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.31856808824442095,
      "learning_rate": 6.078221470286417e-06,
      "loss": 0.8202,
      "step": 650
    },
    {
      "epoch": 2.003078699249567,
      "grad_norm": 0.34215937182253053,
      "learning_rate": 6.045294390555598e-06,
      "loss": 0.793,
      "step": 651
    },
    {
      "epoch": 2.006157398499134,
      "grad_norm": 0.3619384154906207,
      "learning_rate": 6.012418058278265e-06,
      "loss": 0.8283,
      "step": 652
    },
    {
      "epoch": 2.009236097748701,
      "grad_norm": 0.3254485652094408,
      "learning_rate": 5.979592895329085e-06,
      "loss": 0.8403,
      "step": 653
    },
    {
      "epoch": 2.012314796998268,
      "grad_norm": 0.30225650913014335,
      "learning_rate": 5.946819322926127e-06,
      "loss": 0.822,
      "step": 654
    },
    {
      "epoch": 2.0153934962478353,
      "grad_norm": 0.29390661506701493,
      "learning_rate": 5.9140977616254285e-06,
      "loss": 0.8393,
      "step": 655
    },
    {
      "epoch": 2.0184721954974023,
      "grad_norm": 0.299697414704573,
      "learning_rate": 5.88142863131562e-06,
      "loss": 0.8111,
      "step": 656
    },
    {
      "epoch": 2.0215508947469694,
      "grad_norm": 0.3423490458575073,
      "learning_rate": 5.8488123512125225e-06,
      "loss": 0.8352,
      "step": 657
    },
    {
      "epoch": 2.0246295939965364,
      "grad_norm": 0.3050493635906226,
      "learning_rate": 5.816249339853781e-06,
      "loss": 0.8168,
      "step": 658
    },
    {
      "epoch": 2.0277082932461035,
      "grad_norm": 0.2996652722528695,
      "learning_rate": 5.783740015093484e-06,
      "loss": 0.8339,
      "step": 659
    },
    {
      "epoch": 2.0307869924956705,
      "grad_norm": 0.29615726047575064,
      "learning_rate": 5.751284794096793e-06,
      "loss": 0.8313,
      "step": 660
    },
    {
      "epoch": 2.0338656917452376,
      "grad_norm": 0.30123217504039823,
      "learning_rate": 5.7188840933346265e-06,
      "loss": 0.8041,
      "step": 661
    },
    {
      "epoch": 2.0369443909948046,
      "grad_norm": 0.2943075195993166,
      "learning_rate": 5.686538328578266e-06,
      "loss": 0.7894,
      "step": 662
    },
    {
      "epoch": 2.0400230902443717,
      "grad_norm": 0.30251811918134236,
      "learning_rate": 5.654247914894058e-06,
      "loss": 0.8257,
      "step": 663
    },
    {
      "epoch": 2.0431017894939387,
      "grad_norm": 0.30695132672297026,
      "learning_rate": 5.622013266638063e-06,
      "loss": 0.8378,
      "step": 664
    },
    {
      "epoch": 2.0461804887435058,
      "grad_norm": 0.3112435739945259,
      "learning_rate": 5.589834797450764e-06,
      "loss": 0.8125,
      "step": 665
    },
    {
      "epoch": 2.049259187993073,
      "grad_norm": 0.27385134484668433,
      "learning_rate": 5.557712920251741e-06,
      "loss": 0.8087,
      "step": 666
    },
    {
      "epoch": 2.05233788724264,
      "grad_norm": 0.2911103749626962,
      "learning_rate": 5.525648047234364e-06,
      "loss": 0.8196,
      "step": 667
    },
    {
      "epoch": 2.055416586492207,
      "grad_norm": 0.2944043023252442,
      "learning_rate": 5.49364058986053e-06,
      "loss": 0.7952,
      "step": 668
    },
    {
      "epoch": 2.058495285741774,
      "grad_norm": 0.29313936919689215,
      "learning_rate": 5.461690958855368e-06,
      "loss": 0.8104,
      "step": 669
    },
    {
      "epoch": 2.061573984991341,
      "grad_norm": 0.2943558768922095,
      "learning_rate": 5.429799564201956e-06,
      "loss": 0.818,
      "step": 670
    },
    {
      "epoch": 2.064652684240908,
      "grad_norm": 0.312335296560079,
      "learning_rate": 5.3979668151360905e-06,
      "loss": 0.8023,
      "step": 671
    },
    {
      "epoch": 2.067731383490475,
      "grad_norm": 0.2989981258310999,
      "learning_rate": 5.366193120140999e-06,
      "loss": 0.825,
      "step": 672
    },
    {
      "epoch": 2.070810082740042,
      "grad_norm": 0.28937109358133983,
      "learning_rate": 5.3344788869421405e-06,
      "loss": 0.8432,
      "step": 673
    },
    {
      "epoch": 2.0738887819896092,
      "grad_norm": 0.286718259292366,
      "learning_rate": 5.302824522501931e-06,
      "loss": 0.8527,
      "step": 674
    },
    {
      "epoch": 2.0769674812391763,
      "grad_norm": 0.3074591659129627,
      "learning_rate": 5.271230433014543e-06,
      "loss": 0.8223,
      "step": 675
    },
    {
      "epoch": 2.0800461804887433,
      "grad_norm": 0.2915351505160932,
      "learning_rate": 5.239697023900696e-06,
      "loss": 0.8314,
      "step": 676
    },
    {
      "epoch": 2.0831248797383104,
      "grad_norm": 0.27929702926202116,
      "learning_rate": 5.208224699802449e-06,
      "loss": 0.8793,
      "step": 677
    },
    {
      "epoch": 2.0862035789878774,
      "grad_norm": 0.2812626861842994,
      "learning_rate": 5.17681386457801e-06,
      "loss": 0.8459,
      "step": 678
    },
    {
      "epoch": 2.0892822782374445,
      "grad_norm": 0.2943219897605081,
      "learning_rate": 5.145464921296537e-06,
      "loss": 0.864,
      "step": 679
    },
    {
      "epoch": 2.0923609774870116,
      "grad_norm": 0.3094467993093644,
      "learning_rate": 5.114178272232995e-06,
      "loss": 0.8452,
      "step": 680
    },
    {
      "epoch": 2.0954396767365786,
      "grad_norm": 0.2918549674564614,
      "learning_rate": 5.082954318862978e-06,
      "loss": 0.8136,
      "step": 681
    },
    {
      "epoch": 2.0985183759861457,
      "grad_norm": 0.27528972659653544,
      "learning_rate": 5.051793461857546e-06,
      "loss": 0.8571,
      "step": 682
    },
    {
      "epoch": 2.1015970752357127,
      "grad_norm": 0.30432618550555474,
      "learning_rate": 5.0206961010781085e-06,
      "loss": 0.8137,
      "step": 683
    },
    {
      "epoch": 2.1046757744852798,
      "grad_norm": 0.2755280521450484,
      "learning_rate": 4.9896626355712805e-06,
      "loss": 0.8829,
      "step": 684
    },
    {
      "epoch": 2.107754473734847,
      "grad_norm": 0.2866259928077301,
      "learning_rate": 4.958693463563749e-06,
      "loss": 0.7983,
      "step": 685
    },
    {
      "epoch": 2.110833172984414,
      "grad_norm": 0.2801726212427764,
      "learning_rate": 4.9277889824571925e-06,
      "loss": 0.8245,
      "step": 686
    },
    {
      "epoch": 2.113911872233981,
      "grad_norm": 0.2819272459173822,
      "learning_rate": 4.896949588823149e-06,
      "loss": 0.8426,
      "step": 687
    },
    {
      "epoch": 2.1169905714835484,
      "grad_norm": 0.29085299353306987,
      "learning_rate": 4.866175678397955e-06,
      "loss": 0.8507,
      "step": 688
    },
    {
      "epoch": 2.1200692707331155,
      "grad_norm": 0.28243823185433353,
      "learning_rate": 4.835467646077656e-06,
      "loss": 0.8111,
      "step": 689
    },
    {
      "epoch": 2.1231479699826825,
      "grad_norm": 0.2903642622752368,
      "learning_rate": 4.804825885912925e-06,
      "loss": 0.8297,
      "step": 690
    },
    {
      "epoch": 2.1262266692322496,
      "grad_norm": 0.28126903878184484,
      "learning_rate": 4.774250791104033e-06,
      "loss": 0.7966,
      "step": 691
    },
    {
      "epoch": 2.1293053684818166,
      "grad_norm": 0.2826131536649894,
      "learning_rate": 4.743742753995785e-06,
      "loss": 0.8035,
      "step": 692
    },
    {
      "epoch": 2.1323840677313837,
      "grad_norm": 0.28427851874380133,
      "learning_rate": 4.713302166072492e-06,
      "loss": 0.8265,
      "step": 693
    },
    {
      "epoch": 2.1354627669809507,
      "grad_norm": 0.28430271404968804,
      "learning_rate": 4.682929417952939e-06,
      "loss": 0.8036,
      "step": 694
    },
    {
      "epoch": 2.1385414662305178,
      "grad_norm": 0.28471177394746744,
      "learning_rate": 4.652624899385387e-06,
      "loss": 0.8422,
      "step": 695
    },
    {
      "epoch": 2.141620165480085,
      "grad_norm": 0.2790223034916252,
      "learning_rate": 4.622388999242564e-06,
      "loss": 0.8154,
      "step": 696
    },
    {
      "epoch": 2.144698864729652,
      "grad_norm": 0.29844292663643224,
      "learning_rate": 4.592222105516666e-06,
      "loss": 0.8458,
      "step": 697
    },
    {
      "epoch": 2.147777563979219,
      "grad_norm": 0.2967856737028742,
      "learning_rate": 4.5621246053143976e-06,
      "loss": 0.8068,
      "step": 698
    },
    {
      "epoch": 2.150856263228786,
      "grad_norm": 0.2747193168472799,
      "learning_rate": 4.532096884851978e-06,
      "loss": 0.8132,
      "step": 699
    },
    {
      "epoch": 2.153934962478353,
      "grad_norm": 0.27926384089212486,
      "learning_rate": 4.502139329450228e-06,
      "loss": 0.7867,
      "step": 700
    },
    {
      "epoch": 2.15701366172792,
      "grad_norm": 0.28986616555793143,
      "learning_rate": 4.472252323529575e-06,
      "loss": 0.8422,
      "step": 701
    },
    {
      "epoch": 2.160092360977487,
      "grad_norm": 0.2948839943111143,
      "learning_rate": 4.442436250605144e-06,
      "loss": 0.8316,
      "step": 702
    },
    {
      "epoch": 2.163171060227054,
      "grad_norm": 0.2902592698703501,
      "learning_rate": 4.41269149328185e-06,
      "loss": 0.8035,
      "step": 703
    },
    {
      "epoch": 2.1662497594766212,
      "grad_norm": 0.28410141533609207,
      "learning_rate": 4.383018433249464e-06,
      "loss": 0.8337,
      "step": 704
    },
    {
      "epoch": 2.1693284587261883,
      "grad_norm": 0.29746268796461406,
      "learning_rate": 4.353417451277732e-06,
      "loss": 0.7823,
      "step": 705
    },
    {
      "epoch": 2.1724071579757553,
      "grad_norm": 0.29719456990498205,
      "learning_rate": 4.323888927211472e-06,
      "loss": 0.8021,
      "step": 706
    },
    {
      "epoch": 2.1754858572253224,
      "grad_norm": 0.2923839158586591,
      "learning_rate": 4.294433239965718e-06,
      "loss": 0.8282,
      "step": 707
    },
    {
      "epoch": 2.1785645564748894,
      "grad_norm": 0.28616242808642994,
      "learning_rate": 4.265050767520851e-06,
      "loss": 0.8516,
      "step": 708
    },
    {
      "epoch": 2.1816432557244565,
      "grad_norm": 0.2957973317162664,
      "learning_rate": 4.235741886917736e-06,
      "loss": 0.8041,
      "step": 709
    },
    {
      "epoch": 2.1847219549740236,
      "grad_norm": 0.27305442159552007,
      "learning_rate": 4.206506974252911e-06,
      "loss": 0.8077,
      "step": 710
    },
    {
      "epoch": 2.1878006542235906,
      "grad_norm": 0.2725662607446614,
      "learning_rate": 4.1773464046737275e-06,
      "loss": 0.8856,
      "step": 711
    },
    {
      "epoch": 2.1908793534731577,
      "grad_norm": 0.3000324424898719,
      "learning_rate": 4.1482605523735676e-06,
      "loss": 0.8413,
      "step": 712
    },
    {
      "epoch": 2.1939580527227247,
      "grad_norm": 0.30334566863851975,
      "learning_rate": 4.119249790587028e-06,
      "loss": 0.8085,
      "step": 713
    },
    {
      "epoch": 2.1970367519722918,
      "grad_norm": 0.28954509405923956,
      "learning_rate": 4.0903144915851174e-06,
      "loss": 0.832,
      "step": 714
    },
    {
      "epoch": 2.200115451221859,
      "grad_norm": 0.2943172854036903,
      "learning_rate": 4.061455026670509e-06,
      "loss": 0.8704,
      "step": 715
    },
    {
      "epoch": 2.203194150471426,
      "grad_norm": 0.3099194345524093,
      "learning_rate": 4.032671766172756e-06,
      "loss": 0.8577,
      "step": 716
    },
    {
      "epoch": 2.206272849720993,
      "grad_norm": 0.28720392777054404,
      "learning_rate": 4.003965079443535e-06,
      "loss": 0.7789,
      "step": 717
    },
    {
      "epoch": 2.20935154897056,
      "grad_norm": 0.28676341486239026,
      "learning_rate": 3.975335334851925e-06,
      "loss": 0.8165,
      "step": 718
    },
    {
      "epoch": 2.212430248220127,
      "grad_norm": 0.2748803500992399,
      "learning_rate": 3.9467828997796675e-06,
      "loss": 0.8635,
      "step": 719
    },
    {
      "epoch": 2.215508947469694,
      "grad_norm": 0.29157857541732224,
      "learning_rate": 3.918308140616457e-06,
      "loss": 0.8561,
      "step": 720
    },
    {
      "epoch": 2.218587646719261,
      "grad_norm": 0.3025656123551583,
      "learning_rate": 3.8899114227552315e-06,
      "loss": 0.8026,
      "step": 721
    },
    {
      "epoch": 2.221666345968828,
      "grad_norm": 0.28485563854137397,
      "learning_rate": 3.861593110587488e-06,
      "loss": 0.8384,
      "step": 722
    },
    {
      "epoch": 2.2247450452183952,
      "grad_norm": 0.28354929325901884,
      "learning_rate": 3.833353567498627e-06,
      "loss": 0.8332,
      "step": 723
    },
    {
      "epoch": 2.2278237444679623,
      "grad_norm": 0.3052575478173997,
      "learning_rate": 3.805193155863247e-06,
      "loss": 0.8308,
      "step": 724
    },
    {
      "epoch": 2.2309024437175293,
      "grad_norm": 0.29111313475346245,
      "learning_rate": 3.7771122370405365e-06,
      "loss": 0.841,
      "step": 725
    },
    {
      "epoch": 2.2339811429670964,
      "grad_norm": 0.27735380789802716,
      "learning_rate": 3.7491111713696026e-06,
      "loss": 0.8603,
      "step": 726
    },
    {
      "epoch": 2.2370598422166634,
      "grad_norm": 0.29604767950002,
      "learning_rate": 3.721190318164877e-06,
      "loss": 0.8471,
      "step": 727
    },
    {
      "epoch": 2.2401385414662305,
      "grad_norm": 0.27686343055004564,
      "learning_rate": 3.6933500357114884e-06,
      "loss": 0.8187,
      "step": 728
    },
    {
      "epoch": 2.2432172407157975,
      "grad_norm": 0.27125262380873255,
      "learning_rate": 3.665590681260658e-06,
      "loss": 0.7872,
      "step": 729
    },
    {
      "epoch": 2.2462959399653646,
      "grad_norm": 0.27551052809894294,
      "learning_rate": 3.6379126110251383e-06,
      "loss": 0.8264,
      "step": 730
    },
    {
      "epoch": 2.2493746392149316,
      "grad_norm": 0.29288423496444244,
      "learning_rate": 3.6103161801746224e-06,
      "loss": 0.764,
      "step": 731
    },
    {
      "epoch": 2.2524533384644987,
      "grad_norm": 0.26962894568543544,
      "learning_rate": 3.582801742831197e-06,
      "loss": 0.7866,
      "step": 732
    },
    {
      "epoch": 2.2555320377140657,
      "grad_norm": 0.28224510451243046,
      "learning_rate": 3.5553696520647873e-06,
      "loss": 0.8824,
      "step": 733
    },
    {
      "epoch": 2.258610736963633,
      "grad_norm": 0.2761135656149013,
      "learning_rate": 3.5280202598886327e-06,
      "loss": 0.8461,
      "step": 734
    },
    {
      "epoch": 2.2616894362132,
      "grad_norm": 0.2721306393161773,
      "learning_rate": 3.5007539172547867e-06,
      "loss": 0.861,
      "step": 735
    },
    {
      "epoch": 2.264768135462767,
      "grad_norm": 0.2908594667012971,
      "learning_rate": 3.4735709740495748e-06,
      "loss": 0.8286,
      "step": 736
    },
    {
      "epoch": 2.267846834712334,
      "grad_norm": 0.2775445081655492,
      "learning_rate": 3.4464717790891445e-06,
      "loss": 0.8034,
      "step": 737
    },
    {
      "epoch": 2.270925533961901,
      "grad_norm": 0.2882421306216314,
      "learning_rate": 3.4194566801149564e-06,
      "loss": 0.8614,
      "step": 738
    },
    {
      "epoch": 2.274004233211468,
      "grad_norm": 0.28191666007027566,
      "learning_rate": 3.3925260237893485e-06,
      "loss": 0.8236,
      "step": 739
    },
    {
      "epoch": 2.277082932461035,
      "grad_norm": 0.2983487805977567,
      "learning_rate": 3.365680155691078e-06,
      "loss": 0.8311,
      "step": 740
    },
    {
      "epoch": 2.280161631710602,
      "grad_norm": 0.2799614157121612,
      "learning_rate": 3.338919420310871e-06,
      "loss": 0.8382,
      "step": 741
    },
    {
      "epoch": 2.283240330960169,
      "grad_norm": 0.29499277753759784,
      "learning_rate": 3.312244161047029e-06,
      "loss": 0.841,
      "step": 742
    },
    {
      "epoch": 2.2863190302097363,
      "grad_norm": 0.28192422705576375,
      "learning_rate": 3.2856547202010105e-06,
      "loss": 0.8455,
      "step": 743
    },
    {
      "epoch": 2.2893977294593033,
      "grad_norm": 0.29598516140386033,
      "learning_rate": 3.259151438973024e-06,
      "loss": 0.8409,
      "step": 744
    },
    {
      "epoch": 2.2924764287088704,
      "grad_norm": 0.2775301448691774,
      "learning_rate": 3.232734657457676e-06,
      "loss": 0.8436,
      "step": 745
    },
    {
      "epoch": 2.2955551279584374,
      "grad_norm": 0.2953219515791992,
      "learning_rate": 3.2064047146395894e-06,
      "loss": 0.8341,
      "step": 746
    },
    {
      "epoch": 2.2986338272080045,
      "grad_norm": 0.2824401479873467,
      "learning_rate": 3.1801619483890622e-06,
      "loss": 0.7941,
      "step": 747
    },
    {
      "epoch": 2.3017125264575715,
      "grad_norm": 0.2809848187316755,
      "learning_rate": 3.1540066954577197e-06,
      "loss": 0.8249,
      "step": 748
    },
    {
      "epoch": 2.3047912257071386,
      "grad_norm": 0.2888230774314398,
      "learning_rate": 3.1279392914742047e-06,
      "loss": 0.8212,
      "step": 749
    },
    {
      "epoch": 2.3078699249567056,
      "grad_norm": 0.2745993030301653,
      "learning_rate": 3.101960070939871e-06,
      "loss": 0.8217,
      "step": 750
    },
    {
      "epoch": 2.3109486242062727,
      "grad_norm": 0.280384892048276,
      "learning_rate": 3.0760693672244858e-06,
      "loss": 0.8304,
      "step": 751
    },
    {
      "epoch": 2.3140273234558397,
      "grad_norm": 0.2794382644484308,
      "learning_rate": 3.050267512561954e-06,
      "loss": 0.8579,
      "step": 752
    },
    {
      "epoch": 2.3171060227054068,
      "grad_norm": 0.28007746456626664,
      "learning_rate": 3.0245548380460486e-06,
      "loss": 0.8664,
      "step": 753
    },
    {
      "epoch": 2.3201847219549743,
      "grad_norm": 0.2811263419334688,
      "learning_rate": 2.9989316736261753e-06,
      "loss": 0.8194,
      "step": 754
    },
    {
      "epoch": 2.3232634212045413,
      "grad_norm": 0.26952550897713806,
      "learning_rate": 2.97339834810313e-06,
      "loss": 0.7975,
      "step": 755
    },
    {
      "epoch": 2.3263421204541084,
      "grad_norm": 0.28067501580665116,
      "learning_rate": 2.9479551891248746e-06,
      "loss": 0.8098,
      "step": 756
    },
    {
      "epoch": 2.3294208197036754,
      "grad_norm": 0.2797980052594121,
      "learning_rate": 2.922602523182344e-06,
      "loss": 0.7944,
      "step": 757
    },
    {
      "epoch": 2.3324995189532425,
      "grad_norm": 0.276895824133422,
      "learning_rate": 2.8973406756052512e-06,
      "loss": 0.8192,
      "step": 758
    },
    {
      "epoch": 2.3355782182028095,
      "grad_norm": 0.28092292652289896,
      "learning_rate": 2.872169970557913e-06,
      "loss": 0.8001,
      "step": 759
    },
    {
      "epoch": 2.3386569174523766,
      "grad_norm": 0.2723446846081883,
      "learning_rate": 2.847090731035086e-06,
      "loss": 0.7862,
      "step": 760
    },
    {
      "epoch": 2.3417356167019436,
      "grad_norm": 0.27491911223733717,
      "learning_rate": 2.8221032788578206e-06,
      "loss": 0.8572,
      "step": 761
    },
    {
      "epoch": 2.3448143159515107,
      "grad_norm": 0.27721790406865726,
      "learning_rate": 2.7972079346693558e-06,
      "loss": 0.81,
      "step": 762
    },
    {
      "epoch": 2.3478930152010777,
      "grad_norm": 0.28061334487764106,
      "learning_rate": 2.772405017930965e-06,
      "loss": 0.8766,
      "step": 763
    },
    {
      "epoch": 2.350971714450645,
      "grad_norm": 0.2861454387092297,
      "learning_rate": 2.7476948469178886e-06,
      "loss": 0.8175,
      "step": 764
    },
    {
      "epoch": 2.354050413700212,
      "grad_norm": 0.29594331961161097,
      "learning_rate": 2.7230777387152297e-06,
      "loss": 0.8582,
      "step": 765
    },
    {
      "epoch": 2.357129112949779,
      "grad_norm": 0.2875146035999326,
      "learning_rate": 2.698554009213902e-06,
      "loss": 0.8405,
      "step": 766
    },
    {
      "epoch": 2.360207812199346,
      "grad_norm": 0.27109813967039786,
      "learning_rate": 2.6741239731065648e-06,
      "loss": 0.7687,
      "step": 767
    },
    {
      "epoch": 2.363286511448913,
      "grad_norm": 0.7879905238961213,
      "learning_rate": 2.649787943883583e-06,
      "loss": 0.8088,
      "step": 768
    },
    {
      "epoch": 2.36636521069848,
      "grad_norm": 0.2736743283427718,
      "learning_rate": 2.625546233829016e-06,
      "loss": 0.7826,
      "step": 769
    },
    {
      "epoch": 2.369443909948047,
      "grad_norm": 0.2937003279907383,
      "learning_rate": 2.601399154016603e-06,
      "loss": 0.8259,
      "step": 770
    },
    {
      "epoch": 2.372522609197614,
      "grad_norm": 0.28406779989742936,
      "learning_rate": 2.5773470143057657e-06,
      "loss": 0.8101,
      "step": 771
    },
    {
      "epoch": 2.375601308447181,
      "grad_norm": 0.2711412331124363,
      "learning_rate": 2.5533901233376466e-06,
      "loss": 0.825,
      "step": 772
    },
    {
      "epoch": 2.3786800076967483,
      "grad_norm": 0.27081877588642744,
      "learning_rate": 2.529528788531128e-06,
      "loss": 0.7938,
      "step": 773
    },
    {
      "epoch": 2.3817587069463153,
      "grad_norm": 0.2689332576956399,
      "learning_rate": 2.5057633160789186e-06,
      "loss": 0.7981,
      "step": 774
    },
    {
      "epoch": 2.3848374061958824,
      "grad_norm": 0.29981086360523423,
      "learning_rate": 2.4820940109435888e-06,
      "loss": 0.8007,
      "step": 775
    },
    {
      "epoch": 2.3879161054454494,
      "grad_norm": 0.28113493041010784,
      "learning_rate": 2.4585211768536754e-06,
      "loss": 0.8292,
      "step": 776
    },
    {
      "epoch": 2.3909948046950165,
      "grad_norm": 0.27439512014998807,
      "learning_rate": 2.435045116299788e-06,
      "loss": 0.8607,
      "step": 777
    },
    {
      "epoch": 2.3940735039445835,
      "grad_norm": 0.29710248181756843,
      "learning_rate": 2.411666130530721e-06,
      "loss": 0.8139,
      "step": 778
    },
    {
      "epoch": 2.3971522031941506,
      "grad_norm": 0.2758499884549454,
      "learning_rate": 2.388384519549588e-06,
      "loss": 0.8359,
      "step": 779
    },
    {
      "epoch": 2.4002309024437176,
      "grad_norm": 0.2753177457083272,
      "learning_rate": 2.3652005821099678e-06,
      "loss": 0.7929,
      "step": 780
    },
    {
      "epoch": 2.4033096016932847,
      "grad_norm": 0.27773393634843235,
      "learning_rate": 2.3421146157120813e-06,
      "loss": 0.8603,
      "step": 781
    },
    {
      "epoch": 2.4063883009428517,
      "grad_norm": 0.276254464492646,
      "learning_rate": 2.3191269165989715e-06,
      "loss": 0.7593,
      "step": 782
    },
    {
      "epoch": 2.4094670001924188,
      "grad_norm": 0.26182946397325757,
      "learning_rate": 2.296237779752687e-06,
      "loss": 0.8023,
      "step": 783
    },
    {
      "epoch": 2.412545699441986,
      "grad_norm": 0.282602819264514,
      "learning_rate": 2.2734474988905207e-06,
      "loss": 0.8381,
      "step": 784
    },
    {
      "epoch": 2.415624398691553,
      "grad_norm": 0.2749548773493264,
      "learning_rate": 2.250756366461225e-06,
      "loss": 0.8045,
      "step": 785
    },
    {
      "epoch": 2.41870309794112,
      "grad_norm": 0.28139009506990537,
      "learning_rate": 2.2281646736412575e-06,
      "loss": 0.7846,
      "step": 786
    },
    {
      "epoch": 2.421781797190687,
      "grad_norm": 0.2821460650208842,
      "learning_rate": 2.2056727103310592e-06,
      "loss": 0.8268,
      "step": 787
    },
    {
      "epoch": 2.424860496440254,
      "grad_norm": 0.33007047972302583,
      "learning_rate": 2.1832807651513156e-06,
      "loss": 0.8267,
      "step": 788
    },
    {
      "epoch": 2.427939195689821,
      "grad_norm": 0.2650571357504714,
      "learning_rate": 2.160989125439268e-06,
      "loss": 0.8062,
      "step": 789
    },
    {
      "epoch": 2.431017894939388,
      "grad_norm": 0.2851853642288212,
      "learning_rate": 2.1387980772450236e-06,
      "loss": 0.8263,
      "step": 790
    },
    {
      "epoch": 2.434096594188955,
      "grad_norm": 0.27578799469628795,
      "learning_rate": 2.116707905327874e-06,
      "loss": 0.8479,
      "step": 791
    },
    {
      "epoch": 2.4371752934385222,
      "grad_norm": 0.2885814603331029,
      "learning_rate": 2.0947188931526564e-06,
      "loss": 0.8077,
      "step": 792
    },
    {
      "epoch": 2.4402539926880893,
      "grad_norm": 0.32576729073954375,
      "learning_rate": 2.072831322886105e-06,
      "loss": 0.8384,
      "step": 793
    },
    {
      "epoch": 2.4433326919376563,
      "grad_norm": 0.27700306111762474,
      "learning_rate": 2.0510454753932395e-06,
      "loss": 0.7906,
      "step": 794
    },
    {
      "epoch": 2.4464113911872234,
      "grad_norm": 0.29314136943515867,
      "learning_rate": 2.029361630233747e-06,
      "loss": 0.7988,
      "step": 795
    },
    {
      "epoch": 2.4494900904367904,
      "grad_norm": 0.26588343761293465,
      "learning_rate": 2.0077800656584102e-06,
      "loss": 0.8093,
      "step": 796
    },
    {
      "epoch": 2.4525687896863575,
      "grad_norm": 0.2768704535338434,
      "learning_rate": 1.986301058605531e-06,
      "loss": 0.8052,
      "step": 797
    },
    {
      "epoch": 2.4556474889359245,
      "grad_norm": 0.2837725644857283,
      "learning_rate": 1.9649248846973665e-06,
      "loss": 0.8137,
      "step": 798
    },
    {
      "epoch": 2.4587261881854916,
      "grad_norm": 0.27438068811967264,
      "learning_rate": 1.943651818236616e-06,
      "loss": 0.8815,
      "step": 799
    },
    {
      "epoch": 2.4618048874350587,
      "grad_norm": 0.2757328205720069,
      "learning_rate": 1.9224821322028676e-06,
      "loss": 0.8225,
      "step": 800
    },
    {
      "epoch": 2.4648835866846257,
      "grad_norm": 0.26815165925457274,
      "learning_rate": 1.901416098249136e-06,
      "loss": 0.8201,
      "step": 801
    },
    {
      "epoch": 2.4679622859341928,
      "grad_norm": 0.2847320094509794,
      "learning_rate": 1.8804539866983384e-06,
      "loss": 0.8271,
      "step": 802
    },
    {
      "epoch": 2.47104098518376,
      "grad_norm": 0.2772920703083176,
      "learning_rate": 1.859596066539846e-06,
      "loss": 0.7605,
      "step": 803
    },
    {
      "epoch": 2.474119684433327,
      "grad_norm": 0.26650123355437666,
      "learning_rate": 1.838842605426031e-06,
      "loss": 0.8091,
      "step": 804
    },
    {
      "epoch": 2.477198383682894,
      "grad_norm": 0.28552294476867696,
      "learning_rate": 1.8181938696688294e-06,
      "loss": 0.8441,
      "step": 805
    },
    {
      "epoch": 2.480277082932461,
      "grad_norm": 0.2728276771522539,
      "learning_rate": 1.7976501242363242e-06,
      "loss": 0.7991,
      "step": 806
    },
    {
      "epoch": 2.483355782182028,
      "grad_norm": 0.27178588739117615,
      "learning_rate": 1.7772116327493372e-06,
      "loss": 0.8043,
      "step": 807
    },
    {
      "epoch": 2.486434481431595,
      "grad_norm": 0.2911558986489875,
      "learning_rate": 1.756878657478064e-06,
      "loss": 0.8033,
      "step": 808
    },
    {
      "epoch": 2.489513180681162,
      "grad_norm": 0.27778268777488235,
      "learning_rate": 1.7366514593386953e-06,
      "loss": 0.858,
      "step": 809
    },
    {
      "epoch": 2.492591879930729,
      "grad_norm": 0.2769032299034109,
      "learning_rate": 1.7165302978900655e-06,
      "loss": 0.7946,
      "step": 810
    },
    {
      "epoch": 2.4956705791802962,
      "grad_norm": 0.2755011795428173,
      "learning_rate": 1.6965154313303367e-06,
      "loss": 0.8308,
      "step": 811
    },
    {
      "epoch": 2.4987492784298633,
      "grad_norm": 0.28601652795993476,
      "learning_rate": 1.6766071164936681e-06,
      "loss": 0.8647,
      "step": 812
    },
    {
      "epoch": 2.5018279776794303,
      "grad_norm": 0.2901617804507704,
      "learning_rate": 1.6568056088469387e-06,
      "loss": 0.844,
      "step": 813
    },
    {
      "epoch": 2.5049066769289974,
      "grad_norm": 0.261856671165456,
      "learning_rate": 1.6371111624864544e-06,
      "loss": 0.8376,
      "step": 814
    },
    {
      "epoch": 2.5079853761785644,
      "grad_norm": 0.2671683408589042,
      "learning_rate": 1.6175240301346906e-06,
      "loss": 0.7809,
      "step": 815
    },
    {
      "epoch": 2.5110640754281315,
      "grad_norm": 0.2658169846384481,
      "learning_rate": 1.598044463137054e-06,
      "loss": 0.8149,
      "step": 816
    },
    {
      "epoch": 2.5141427746776985,
      "grad_norm": 0.28070929560068214,
      "learning_rate": 1.5786727114586586e-06,
      "loss": 0.8641,
      "step": 817
    },
    {
      "epoch": 2.5172214739272656,
      "grad_norm": 0.28456142928576605,
      "learning_rate": 1.5594090236811022e-06,
      "loss": 0.795,
      "step": 818
    },
    {
      "epoch": 2.5203001731768326,
      "grad_norm": 0.27789722921558807,
      "learning_rate": 1.540253646999299e-06,
      "loss": 0.809,
      "step": 819
    },
    {
      "epoch": 2.5233788724263997,
      "grad_norm": 0.26929278412282603,
      "learning_rate": 1.5212068272182923e-06,
      "loss": 0.8347,
      "step": 820
    },
    {
      "epoch": 2.5264575716759667,
      "grad_norm": 0.26840612505547773,
      "learning_rate": 1.5022688087501092e-06,
      "loss": 0.8272,
      "step": 821
    },
    {
      "epoch": 2.529536270925534,
      "grad_norm": 0.2807957662907676,
      "learning_rate": 1.4834398346106126e-06,
      "loss": 0.8659,
      "step": 822
    },
    {
      "epoch": 2.532614970175101,
      "grad_norm": 0.27978539683559006,
      "learning_rate": 1.4647201464163908e-06,
      "loss": 0.8051,
      "step": 823
    },
    {
      "epoch": 2.535693669424668,
      "grad_norm": 0.27632085713351334,
      "learning_rate": 1.4461099843816684e-06,
      "loss": 0.8029,
      "step": 824
    },
    {
      "epoch": 2.538772368674235,
      "grad_norm": 0.2717605243016332,
      "learning_rate": 1.4276095873151952e-06,
      "loss": 0.8168,
      "step": 825
    },
    {
      "epoch": 2.541851067923802,
      "grad_norm": 0.28568643400148436,
      "learning_rate": 1.4092191926172106e-06,
      "loss": 0.8071,
      "step": 826
    },
    {
      "epoch": 2.544929767173369,
      "grad_norm": 0.27680003547563325,
      "learning_rate": 1.3909390362763752e-06,
      "loss": 0.8376,
      "step": 827
    },
    {
      "epoch": 2.548008466422936,
      "grad_norm": 0.27361830205323207,
      "learning_rate": 1.3727693528667595e-06,
      "loss": 0.8116,
      "step": 828
    },
    {
      "epoch": 2.551087165672503,
      "grad_norm": 0.27485407337184703,
      "learning_rate": 1.3547103755448287e-06,
      "loss": 0.8394,
      "step": 829
    },
    {
      "epoch": 2.55416586492207,
      "grad_norm": 0.2798509396151052,
      "learning_rate": 1.3367623360464365e-06,
      "loss": 0.78,
      "step": 830
    },
    {
      "epoch": 2.5572445641716373,
      "grad_norm": 0.26708169012911964,
      "learning_rate": 1.3189254646838766e-06,
      "loss": 0.8575,
      "step": 831
    },
    {
      "epoch": 2.5603232634212043,
      "grad_norm": 0.2816682890264548,
      "learning_rate": 1.3011999903429074e-06,
      "loss": 0.8687,
      "step": 832
    },
    {
      "epoch": 2.5634019626707714,
      "grad_norm": 0.26698976714048817,
      "learning_rate": 1.2835861404798266e-06,
      "loss": 0.8296,
      "step": 833
    },
    {
      "epoch": 2.5664806619203384,
      "grad_norm": 0.27553049349529934,
      "learning_rate": 1.266084141118542e-06,
      "loss": 0.8342,
      "step": 834
    },
    {
      "epoch": 2.5695593611699055,
      "grad_norm": 0.2804172974218874,
      "learning_rate": 1.2486942168476756e-06,
      "loss": 0.8524,
      "step": 835
    },
    {
      "epoch": 2.5726380604194725,
      "grad_norm": 0.42447044678925017,
      "learning_rate": 1.2314165908176956e-06,
      "loss": 0.7738,
      "step": 836
    },
    {
      "epoch": 2.5757167596690396,
      "grad_norm": 0.26766276960782454,
      "learning_rate": 1.2142514847380238e-06,
      "loss": 0.849,
      "step": 837
    },
    {
      "epoch": 2.5787954589186066,
      "grad_norm": 0.27267141763725705,
      "learning_rate": 1.1971991188742182e-06,
      "loss": 0.805,
      "step": 838
    },
    {
      "epoch": 2.5818741581681737,
      "grad_norm": 0.26861682445697344,
      "learning_rate": 1.1802597120451288e-06,
      "loss": 0.842,
      "step": 839
    },
    {
      "epoch": 2.584952857417741,
      "grad_norm": 0.27632193328344473,
      "learning_rate": 1.1634334816200999e-06,
      "loss": 0.8086,
      "step": 840
    },
    {
      "epoch": 2.588031556667308,
      "grad_norm": 0.2707726611578219,
      "learning_rate": 1.146720643516177e-06,
      "loss": 0.8396,
      "step": 841
    },
    {
      "epoch": 2.5911102559168753,
      "grad_norm": 0.26853876369003304,
      "learning_rate": 1.1301214121953319e-06,
      "loss": 0.8191,
      "step": 842
    },
    {
      "epoch": 2.5941889551664423,
      "grad_norm": 0.26613715118500036,
      "learning_rate": 1.1136360006617186e-06,
      "loss": 0.8279,
      "step": 843
    },
    {
      "epoch": 2.5972676544160094,
      "grad_norm": 0.2798307089457197,
      "learning_rate": 1.0972646204589377e-06,
      "loss": 0.8706,
      "step": 844
    },
    {
      "epoch": 2.6003463536655764,
      "grad_norm": 0.26715825844705327,
      "learning_rate": 1.0810074816673155e-06,
      "loss": 0.8554,
      "step": 845
    },
    {
      "epoch": 2.6034250529151435,
      "grad_norm": 0.2676366387033547,
      "learning_rate": 1.0648647929012157e-06,
      "loss": 0.8477,
      "step": 846
    },
    {
      "epoch": 2.6065037521647105,
      "grad_norm": 0.2751107320379091,
      "learning_rate": 1.048836761306361e-06,
      "loss": 0.7911,
      "step": 847
    },
    {
      "epoch": 2.6095824514142776,
      "grad_norm": 0.2937315423846763,
      "learning_rate": 1.0329235925571756e-06,
      "loss": 0.7932,
      "step": 848
    },
    {
      "epoch": 2.6126611506638446,
      "grad_norm": 0.2590213075174649,
      "learning_rate": 1.0171254908541373e-06,
      "loss": 0.8198,
      "step": 849
    },
    {
      "epoch": 2.6157398499134117,
      "grad_norm": 0.26023864908886096,
      "learning_rate": 1.0014426589211678e-06,
      "loss": 0.8179,
      "step": 850
    },
    {
      "epoch": 2.6188185491629787,
      "grad_norm": 0.3222246922785001,
      "learning_rate": 9.858752980030295e-07,
      "loss": 0.8492,
      "step": 851
    },
    {
      "epoch": 2.621897248412546,
      "grad_norm": 0.2692818771806659,
      "learning_rate": 9.704236078627382e-07,
      "loss": 0.7775,
      "step": 852
    },
    {
      "epoch": 2.624975947662113,
      "grad_norm": 0.267841695761463,
      "learning_rate": 9.550877867790064e-07,
      "loss": 0.8318,
      "step": 853
    },
    {
      "epoch": 2.62805464691168,
      "grad_norm": 0.275773780017956,
      "learning_rate": 9.398680315436903e-07,
      "loss": 0.8362,
      "step": 854
    },
    {
      "epoch": 2.631133346161247,
      "grad_norm": 0.27198472460024337,
      "learning_rate": 9.247645374592718e-07,
      "loss": 0.8287,
      "step": 855
    },
    {
      "epoch": 2.634212045410814,
      "grad_norm": 0.2675419948358205,
      "learning_rate": 9.097774983363527e-07,
      "loss": 0.8625,
      "step": 856
    },
    {
      "epoch": 2.637290744660381,
      "grad_norm": 0.2711432248254979,
      "learning_rate": 8.949071064911586e-07,
      "loss": 0.8165,
      "step": 857
    },
    {
      "epoch": 2.640369443909948,
      "grad_norm": 0.28538270676942096,
      "learning_rate": 8.801535527430827e-07,
      "loss": 0.8053,
      "step": 858
    },
    {
      "epoch": 2.643448143159515,
      "grad_norm": 0.27402019247215187,
      "learning_rate": 8.655170264122303e-07,
      "loss": 0.7972,
      "step": 859
    },
    {
      "epoch": 2.646526842409082,
      "grad_norm": 0.27284601557791,
      "learning_rate": 8.509977153169935e-07,
      "loss": 0.8515,
      "step": 860
    },
    {
      "epoch": 2.6496055416586493,
      "grad_norm": 0.2608561359947752,
      "learning_rate": 8.365958057716339e-07,
      "loss": 0.8481,
      "step": 861
    },
    {
      "epoch": 2.6526842409082163,
      "grad_norm": 0.27266028245792845,
      "learning_rate": 8.223114825838952e-07,
      "loss": 0.8111,
      "step": 862
    },
    {
      "epoch": 2.6557629401577834,
      "grad_norm": 0.2732381651814444,
      "learning_rate": 8.081449290526433e-07,
      "loss": 0.8162,
      "step": 863
    },
    {
      "epoch": 2.6588416394073504,
      "grad_norm": 0.2739957733242725,
      "learning_rate": 7.940963269654923e-07,
      "loss": 0.7987,
      "step": 864
    },
    {
      "epoch": 2.6619203386569175,
      "grad_norm": 0.26471374952563903,
      "learning_rate": 7.80165856596492e-07,
      "loss": 0.7846,
      "step": 865
    },
    {
      "epoch": 2.6649990379064845,
      "grad_norm": 0.2610835627553217,
      "learning_rate": 7.663536967037977e-07,
      "loss": 0.8246,
      "step": 866
    },
    {
      "epoch": 2.6680777371560516,
      "grad_norm": 0.2633767453678502,
      "learning_rate": 7.526600245273919e-07,
      "loss": 0.8,
      "step": 867
    },
    {
      "epoch": 2.6711564364056186,
      "grad_norm": 0.25905949344954676,
      "learning_rate": 7.390850157868001e-07,
      "loss": 0.7805,
      "step": 868
    },
    {
      "epoch": 2.6742351356551857,
      "grad_norm": 0.2621650855611167,
      "learning_rate": 7.256288446788362e-07,
      "loss": 0.8161,
      "step": 869
    },
    {
      "epoch": 2.6773138349047527,
      "grad_norm": 0.25521330996359914,
      "learning_rate": 7.122916838753724e-07,
      "loss": 0.8363,
      "step": 870
    },
    {
      "epoch": 2.6803925341543198,
      "grad_norm": 0.27117952572954535,
      "learning_rate": 6.990737045211204e-07,
      "loss": 0.8209,
      "step": 871
    },
    {
      "epoch": 2.683471233403887,
      "grad_norm": 0.2638314988191108,
      "learning_rate": 6.859750762314321e-07,
      "loss": 0.8084,
      "step": 872
    },
    {
      "epoch": 2.686549932653454,
      "grad_norm": 0.2744140019820073,
      "learning_rate": 6.729959670901309e-07,
      "loss": 0.868,
      "step": 873
    },
    {
      "epoch": 2.689628631903021,
      "grad_norm": 0.30050858869664904,
      "learning_rate": 6.601365436473439e-07,
      "loss": 0.7888,
      "step": 874
    },
    {
      "epoch": 2.692707331152588,
      "grad_norm": 0.36090539355041923,
      "learning_rate": 6.473969709173799e-07,
      "loss": 0.848,
      "step": 875
    },
    {
      "epoch": 2.695786030402155,
      "grad_norm": 0.27394809641317425,
      "learning_rate": 6.34777412376596e-07,
      "loss": 0.8144,
      "step": 876
    },
    {
      "epoch": 2.698864729651722,
      "grad_norm": 0.2685305528887961,
      "learning_rate": 6.222780299613074e-07,
      "loss": 0.769,
      "step": 877
    },
    {
      "epoch": 2.701943428901289,
      "grad_norm": 0.2618795095946147,
      "learning_rate": 6.09898984065711e-07,
      "loss": 0.8097,
      "step": 878
    },
    {
      "epoch": 2.705022128150856,
      "grad_norm": 0.27693092860333407,
      "learning_rate": 5.976404335398256e-07,
      "loss": 0.8199,
      "step": 879
    },
    {
      "epoch": 2.7081008274004232,
      "grad_norm": 0.30204111261233374,
      "learning_rate": 5.855025356874533e-07,
      "loss": 0.8056,
      "step": 880
    },
    {
      "epoch": 2.7111795266499903,
      "grad_norm": 0.2767470206660611,
      "learning_rate": 5.734854462641548e-07,
      "loss": 0.8396,
      "step": 881
    },
    {
      "epoch": 2.7142582258995573,
      "grad_norm": 0.2713018253179607,
      "learning_rate": 5.61589319475263e-07,
      "loss": 0.8624,
      "step": 882
    },
    {
      "epoch": 2.7173369251491244,
      "grad_norm": 0.27600999136568044,
      "learning_rate": 5.498143079738971e-07,
      "loss": 0.8418,
      "step": 883
    },
    {
      "epoch": 2.7204156243986914,
      "grad_norm": 0.27493654753547714,
      "learning_rate": 5.381605628590003e-07,
      "loss": 0.8467,
      "step": 884
    },
    {
      "epoch": 2.7234943236482585,
      "grad_norm": 0.27209541755366307,
      "learning_rate": 5.266282336734086e-07,
      "loss": 0.8216,
      "step": 885
    },
    {
      "epoch": 2.7265730228978255,
      "grad_norm": 0.2673770663716259,
      "learning_rate": 5.152174684019285e-07,
      "loss": 0.7926,
      "step": 886
    },
    {
      "epoch": 2.7296517221473926,
      "grad_norm": 0.27318679803501866,
      "learning_rate": 5.039284134694333e-07,
      "loss": 0.8263,
      "step": 887
    },
    {
      "epoch": 2.7327304213969597,
      "grad_norm": 0.27528795903837844,
      "learning_rate": 4.927612137389947e-07,
      "loss": 0.8627,
      "step": 888
    },
    {
      "epoch": 2.7358091206465267,
      "grad_norm": 0.27413441133510624,
      "learning_rate": 4.817160125100106e-07,
      "loss": 0.8319,
      "step": 889
    },
    {
      "epoch": 2.7388878198960938,
      "grad_norm": 0.2714845185513935,
      "learning_rate": 4.7079295151638337e-07,
      "loss": 0.8227,
      "step": 890
    },
    {
      "epoch": 2.741966519145661,
      "grad_norm": 0.2629699074778668,
      "learning_rate": 4.5999217092468127e-07,
      "loss": 0.7831,
      "step": 891
    },
    {
      "epoch": 2.745045218395228,
      "grad_norm": 0.36069532568747825,
      "learning_rate": 4.493138093323546e-07,
      "loss": 0.8535,
      "step": 892
    },
    {
      "epoch": 2.748123917644795,
      "grad_norm": 0.26766876789451727,
      "learning_rate": 4.3875800376595003e-07,
      "loss": 0.8166,
      "step": 893
    },
    {
      "epoch": 2.7512026168943624,
      "grad_norm": 0.2748241934872642,
      "learning_rate": 4.283248896793579e-07,
      "loss": 0.8433,
      "step": 894
    },
    {
      "epoch": 2.7542813161439295,
      "grad_norm": 0.2654186325026999,
      "learning_rate": 4.1801460095207025e-07,
      "loss": 0.7772,
      "step": 895
    },
    {
      "epoch": 2.7573600153934965,
      "grad_norm": 0.2661384870877932,
      "learning_rate": 4.0782726988745634e-07,
      "loss": 0.8332,
      "step": 896
    },
    {
      "epoch": 2.7604387146430636,
      "grad_norm": 0.27408294183957815,
      "learning_rate": 3.9776302721108104e-07,
      "loss": 0.8857,
      "step": 897
    },
    {
      "epoch": 2.7635174138926306,
      "grad_norm": 0.2792482201391334,
      "learning_rate": 3.8782200206901487e-07,
      "loss": 0.8363,
      "step": 898
    },
    {
      "epoch": 2.7665961131421977,
      "grad_norm": 0.26535317287787613,
      "learning_rate": 3.780043220261764e-07,
      "loss": 0.8023,
      "step": 899
    },
    {
      "epoch": 2.7696748123917647,
      "grad_norm": 0.26380407774598097,
      "learning_rate": 3.683101130647049e-07,
      "loss": 0.8225,
      "step": 900
    },
    {
      "epoch": 2.7727535116413318,
      "grad_norm": 0.26283606268327003,
      "learning_rate": 3.587394995823301e-07,
      "loss": 0.8169,
      "step": 901
    },
    {
      "epoch": 2.775832210890899,
      "grad_norm": 0.27588585822759004,
      "learning_rate": 3.4929260439079516e-07,
      "loss": 0.8301,
      "step": 902
    },
    {
      "epoch": 2.778910910140466,
      "grad_norm": 0.2589378077573019,
      "learning_rate": 3.3996954871425846e-07,
      "loss": 0.8105,
      "step": 903
    },
    {
      "epoch": 2.781989609390033,
      "grad_norm": 0.2775648332219003,
      "learning_rate": 3.3077045218775194e-07,
      "loss": 0.8157,
      "step": 904
    },
    {
      "epoch": 2.7850683086396,
      "grad_norm": 0.2553373294381815,
      "learning_rate": 3.216954328556443e-07,
      "loss": 0.8192,
      "step": 905
    },
    {
      "epoch": 2.788147007889167,
      "grad_norm": 0.28249549915341615,
      "learning_rate": 3.1274460717012346e-07,
      "loss": 0.8085,
      "step": 906
    },
    {
      "epoch": 2.791225707138734,
      "grad_norm": 0.27502135092643304,
      "learning_rate": 3.039180899897043e-07,
      "loss": 0.8251,
      "step": 907
    },
    {
      "epoch": 2.794304406388301,
      "grad_norm": 0.2601364826982241,
      "learning_rate": 2.9521599457775105e-07,
      "loss": 0.7936,
      "step": 908
    },
    {
      "epoch": 2.797383105637868,
      "grad_norm": 0.2879096974710736,
      "learning_rate": 2.8663843260103076e-07,
      "loss": 0.8571,
      "step": 909
    },
    {
      "epoch": 2.8004618048874352,
      "grad_norm": 0.2654160915370888,
      "learning_rate": 2.781855141282752e-07,
      "loss": 0.7899,
      "step": 910
    },
    {
      "epoch": 2.8035405041370023,
      "grad_norm": 0.27419285708042607,
      "learning_rate": 2.698573476287658e-07,
      "loss": 0.802,
      "step": 911
    },
    {
      "epoch": 2.8066192033865693,
      "grad_norm": 0.2606082564794992,
      "learning_rate": 2.6165403997094995e-07,
      "loss": 0.835,
      "step": 912
    },
    {
      "epoch": 2.8096979026361364,
      "grad_norm": 0.26816238926108155,
      "learning_rate": 2.535756964210634e-07,
      "loss": 0.8576,
      "step": 913
    },
    {
      "epoch": 2.8127766018857034,
      "grad_norm": 0.26371628038897765,
      "learning_rate": 2.4562242064178124e-07,
      "loss": 0.8013,
      "step": 914
    },
    {
      "epoch": 2.8158553011352705,
      "grad_norm": 0.2733375479680703,
      "learning_rate": 2.3779431469088898e-07,
      "loss": 0.8134,
      "step": 915
    },
    {
      "epoch": 2.8189340003848375,
      "grad_norm": 0.2614770536121384,
      "learning_rate": 2.300914790199682e-07,
      "loss": 0.8275,
      "step": 916
    },
    {
      "epoch": 2.8220126996344046,
      "grad_norm": 0.2612954027453335,
      "learning_rate": 2.2251401247311512e-07,
      "loss": 0.791,
      "step": 917
    },
    {
      "epoch": 2.8250913988839716,
      "grad_norm": 0.2663579144764531,
      "learning_rate": 2.1506201228566725e-07,
      "loss": 0.8193,
      "step": 918
    },
    {
      "epoch": 2.8281700981335387,
      "grad_norm": 0.2545779263870759,
      "learning_rate": 2.0773557408295342e-07,
      "loss": 0.7996,
      "step": 919
    },
    {
      "epoch": 2.8312487973831058,
      "grad_norm": 0.2637821604639863,
      "learning_rate": 2.005347918790723e-07,
      "loss": 0.8485,
      "step": 920
    },
    {
      "epoch": 2.834327496632673,
      "grad_norm": 0.26414601756837885,
      "learning_rate": 1.9345975807568473e-07,
      "loss": 0.8291,
      "step": 921
    },
    {
      "epoch": 2.83740619588224,
      "grad_norm": 0.2773154519170929,
      "learning_rate": 1.865105634608244e-07,
      "loss": 0.847,
      "step": 922
    },
    {
      "epoch": 2.840484895131807,
      "grad_norm": 0.26330890432587517,
      "learning_rate": 1.796872972077346e-07,
      "loss": 0.8016,
      "step": 923
    },
    {
      "epoch": 2.843563594381374,
      "grad_norm": 0.27263636879051406,
      "learning_rate": 1.7299004687372666e-07,
      "loss": 0.7996,
      "step": 924
    },
    {
      "epoch": 2.846642293630941,
      "grad_norm": 0.2571883141129233,
      "learning_rate": 1.6641889839905446e-07,
      "loss": 0.8137,
      "step": 925
    },
    {
      "epoch": 2.849720992880508,
      "grad_norm": 0.26710348662161776,
      "learning_rate": 1.5997393610580837e-07,
      "loss": 0.8126,
      "step": 926
    },
    {
      "epoch": 2.852799692130075,
      "grad_norm": 0.2685533204491046,
      "learning_rate": 1.5365524269683962e-07,
      "loss": 0.8662,
      "step": 927
    },
    {
      "epoch": 2.855878391379642,
      "grad_norm": 0.27357728161246847,
      "learning_rate": 1.4746289925469225e-07,
      "loss": 0.7947,
      "step": 928
    },
    {
      "epoch": 2.858957090629209,
      "grad_norm": 0.25992410644484126,
      "learning_rate": 1.4139698524057165e-07,
      "loss": 0.8075,
      "step": 929
    },
    {
      "epoch": 2.8620357898787763,
      "grad_norm": 0.2595955452694006,
      "learning_rate": 1.3545757849331432e-07,
      "loss": 0.798,
      "step": 930
    },
    {
      "epoch": 2.8651144891283433,
      "grad_norm": 0.2685455459484489,
      "learning_rate": 1.2964475522839304e-07,
      "loss": 0.8184,
      "step": 931
    },
    {
      "epoch": 2.8681931883779104,
      "grad_norm": 0.2729426270519641,
      "learning_rate": 1.2395859003694554e-07,
      "loss": 0.7825,
      "step": 932
    },
    {
      "epoch": 2.8712718876274774,
      "grad_norm": 0.2711027038349177,
      "learning_rate": 1.1839915588480743e-07,
      "loss": 0.7906,
      "step": 933
    },
    {
      "epoch": 2.8743505868770445,
      "grad_norm": 0.26786692453204086,
      "learning_rate": 1.1296652411158182e-07,
      "loss": 0.8352,
      "step": 934
    },
    {
      "epoch": 2.8774292861266115,
      "grad_norm": 0.2694821661804626,
      "learning_rate": 1.0766076442971896e-07,
      "loss": 0.8098,
      "step": 935
    },
    {
      "epoch": 2.8805079853761786,
      "grad_norm": 0.28262631416274897,
      "learning_rate": 1.0248194492363028e-07,
      "loss": 0.8821,
      "step": 936
    },
    {
      "epoch": 2.8835866846257456,
      "grad_norm": 0.2721206185247127,
      "learning_rate": 9.743013204880581e-08,
      "loss": 0.839,
      "step": 937
    },
    {
      "epoch": 2.8866653838753127,
      "grad_norm": 0.27134973267220935,
      "learning_rate": 9.250539063096586e-08,
      "loss": 0.8173,
      "step": 938
    },
    {
      "epoch": 2.8897440831248797,
      "grad_norm": 0.26410284006098733,
      "learning_rate": 8.770778386522628e-08,
      "loss": 0.7948,
      "step": 939
    },
    {
      "epoch": 2.892822782374447,
      "grad_norm": 0.2628449394105977,
      "learning_rate": 8.303737331529338e-08,
      "loss": 0.7999,
      "step": 940
    },
    {
      "epoch": 2.895901481624014,
      "grad_norm": 0.2645656380061273,
      "learning_rate": 7.849421891266585e-08,
      "loss": 0.822,
      "step": 941
    },
    {
      "epoch": 2.898980180873581,
      "grad_norm": 0.2643011674855393,
      "learning_rate": 7.407837895587522e-08,
      "loss": 0.8031,
      "step": 942
    },
    {
      "epoch": 2.902058880123148,
      "grad_norm": 0.26935457883681724,
      "learning_rate": 6.978991010972546e-08,
      "loss": 0.8305,
      "step": 943
    },
    {
      "epoch": 2.905137579372715,
      "grad_norm": 0.26251688301566717,
      "learning_rate": 6.562886740457796e-08,
      "loss": 0.8306,
      "step": 944
    },
    {
      "epoch": 2.908216278622282,
      "grad_norm": 0.2609214156568824,
      "learning_rate": 6.159530423563986e-08,
      "loss": 0.8295,
      "step": 945
    },
    {
      "epoch": 2.911294977871849,
      "grad_norm": 0.2680017273218864,
      "learning_rate": 5.768927236227684e-08,
      "loss": 0.7836,
      "step": 946
    },
    {
      "epoch": 2.914373677121416,
      "grad_norm": 0.2792289113023973,
      "learning_rate": 5.391082190735253e-08,
      "loss": 0.8209,
      "step": 947
    },
    {
      "epoch": 2.917452376370983,
      "grad_norm": 0.2640318807966613,
      "learning_rate": 5.026000135658571e-08,
      "loss": 0.8255,
      "step": 948
    },
    {
      "epoch": 2.9205310756205503,
      "grad_norm": 0.27342283658490174,
      "learning_rate": 4.673685755792523e-08,
      "loss": 0.8285,
      "step": 949
    },
    {
      "epoch": 2.9236097748701173,
      "grad_norm": 0.27227271678458953,
      "learning_rate": 4.33414357209494e-08,
      "loss": 0.7834,
      "step": 950
    },
    {
      "epoch": 2.9266884741196844,
      "grad_norm": 0.2708846715327822,
      "learning_rate": 4.007377941628754e-08,
      "loss": 0.8149,
      "step": 951
    },
    {
      "epoch": 2.9297671733692514,
      "grad_norm": 0.2750457356964072,
      "learning_rate": 3.693393057505934e-08,
      "loss": 0.8111,
      "step": 952
    },
    {
      "epoch": 2.9328458726188185,
      "grad_norm": 0.27599376462981645,
      "learning_rate": 3.392192948833861e-08,
      "loss": 0.8167,
      "step": 953
    },
    {
      "epoch": 2.9359245718683855,
      "grad_norm": 0.2702471648231062,
      "learning_rate": 3.103781480663481e-08,
      "loss": 0.8387,
      "step": 954
    },
    {
      "epoch": 2.9390032711179526,
      "grad_norm": 0.264442955831359,
      "learning_rate": 2.8281623539396785e-08,
      "loss": 0.819,
      "step": 955
    },
    {
      "epoch": 2.9420819703675196,
      "grad_norm": 0.25074630662843894,
      "learning_rate": 2.565339105453757e-08,
      "loss": 0.7793,
      "step": 956
    },
    {
      "epoch": 2.9451606696170867,
      "grad_norm": 0.25988492784110945,
      "learning_rate": 2.315315107798366e-08,
      "loss": 0.7917,
      "step": 957
    },
    {
      "epoch": 2.9482393688666537,
      "grad_norm": 0.2641106448623674,
      "learning_rate": 2.078093569323758e-08,
      "loss": 0.8427,
      "step": 958
    },
    {
      "epoch": 2.9513180681162208,
      "grad_norm": 0.27618105925178915,
      "learning_rate": 1.8536775340970426e-08,
      "loss": 0.7992,
      "step": 959
    },
    {
      "epoch": 2.954396767365788,
      "grad_norm": 0.277075879992233,
      "learning_rate": 1.6420698818628844e-08,
      "loss": 0.8238,
      "step": 960
    },
    {
      "epoch": 2.957475466615355,
      "grad_norm": 0.27143988843400674,
      "learning_rate": 1.4432733280065336e-08,
      "loss": 0.8226,
      "step": 961
    },
    {
      "epoch": 2.960554165864922,
      "grad_norm": 0.2647021910381104,
      "learning_rate": 1.2572904235191864e-08,
      "loss": 0.8449,
      "step": 962
    },
    {
      "epoch": 2.963632865114489,
      "grad_norm": 0.25969014831275705,
      "learning_rate": 1.0841235549648998e-08,
      "loss": 0.7646,
      "step": 963
    },
    {
      "epoch": 2.966711564364056,
      "grad_norm": 0.2652825849612162,
      "learning_rate": 9.237749444505062e-09,
      "loss": 0.7871,
      "step": 964
    },
    {
      "epoch": 2.969790263613623,
      "grad_norm": 0.26590935264240734,
      "learning_rate": 7.762466495964126e-09,
      "loss": 0.8531,
      "step": 965
    },
    {
      "epoch": 2.97286896286319,
      "grad_norm": 0.26695821933600367,
      "learning_rate": 6.415405635107341e-09,
      "loss": 0.8004,
      "step": 966
    },
    {
      "epoch": 2.975947662112757,
      "grad_norm": 0.271351590249374,
      "learning_rate": 5.1965841476497854e-09,
      "loss": 0.8598,
      "step": 967
    },
    {
      "epoch": 2.9790263613623242,
      "grad_norm": 0.26987680286923094,
      "learning_rate": 4.106017673713991e-09,
      "loss": 0.7626,
      "step": 968
    },
    {
      "epoch": 2.9821050606118913,
      "grad_norm": 0.26290469935227,
      "learning_rate": 3.1437202076356476e-09,
      "loss": 0.8391,
      "step": 969
    },
    {
      "epoch": 2.9851837598614583,
      "grad_norm": 0.2764115019808121,
      "learning_rate": 2.309704097777088e-09,
      "loss": 0.8276,
      "step": 970
    },
    {
      "epoch": 2.9882624591110254,
      "grad_norm": 0.2729432904988962,
      "learning_rate": 1.6039800463762966e-09,
      "loss": 0.8252,
      "step": 971
    },
    {
      "epoch": 2.9913411583605924,
      "grad_norm": 0.2804501968646771,
      "learning_rate": 1.0265571094036919e-09,
      "loss": 0.8375,
      "step": 972
    },
    {
      "epoch": 2.9944198576101595,
      "grad_norm": 0.2734482777764422,
      "learning_rate": 5.77442696447772e-10,
      "loss": 0.8155,
      "step": 973
    },
    {
      "epoch": 2.9974985568597265,
      "grad_norm": 0.27263442991230996,
      "learning_rate": 2.5664257062185673e-10,
      "loss": 0.8456,
      "step": 974
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.2834188501674651,
      "learning_rate": 6.416084848637205e-11,
      "loss": 0.8002,
      "step": 975
    }
  ],
  "logging_steps": 1,
  "max_steps": 975,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1202115041034240.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}