{
  "best_metric": 0.026174582540988922,
  "best_model_checkpoint": "models/BKAI/checkpoint-24000",
  "epoch": 6.974716652136007,
  "eval_steps": 300,
  "global_step": 24000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02906131938390003,
      "grad_norm": 6.751401901245117,
      "learning_rate": 1.816695431010991e-07,
      "loss": 0.202,
      "step": 100
    },
    {
      "epoch": 0.05812263876780006,
      "grad_norm": 7.665974140167236,
      "learning_rate": 3.633390862021982e-07,
      "loss": 0.1929,
      "step": 200
    },
    {
      "epoch": 0.08718395815170009,
      "grad_norm": 5.795717239379883,
      "learning_rate": 5.450086293032973e-07,
      "loss": 0.1751,
      "step": 300
    },
    {
      "epoch": 0.08718395815170009,
      "eval_loss": 0.14298047125339508,
      "eval_runtime": 27.3707,
      "eval_samples_per_second": 446.937,
      "eval_steps_per_second": 13.993,
      "step": 300
    },
    {
      "epoch": 0.11624527753560011,
      "grad_norm": 2.549802303314209,
      "learning_rate": 7.266781724043964e-07,
      "loss": 0.1245,
      "step": 400
    },
    {
      "epoch": 0.14530659691950015,
      "grad_norm": 4.094095230102539,
      "learning_rate": 9.083477155054955e-07,
      "loss": 0.1347,
      "step": 500
    },
    {
      "epoch": 0.17436791630340018,
      "grad_norm": 3.1329522132873535,
      "learning_rate": 1.0900172586065947e-06,
      "loss": 0.105,
      "step": 600
    },
    {
      "epoch": 0.17436791630340018,
      "eval_loss": 0.09839651733636856,
      "eval_runtime": 26.3248,
      "eval_samples_per_second": 464.695,
      "eval_steps_per_second": 14.549,
      "step": 600
    },
    {
      "epoch": 0.2034292356873002,
      "grad_norm": 4.288038730621338,
      "learning_rate": 1.2716868017076938e-06,
      "loss": 0.0873,
      "step": 700
    },
    {
      "epoch": 0.23249055507120023,
      "grad_norm": 5.486178398132324,
      "learning_rate": 1.4533563448087928e-06,
      "loss": 0.0916,
      "step": 800
    },
    {
      "epoch": 0.26155187445510025,
      "grad_norm": 3.6196625232696533,
      "learning_rate": 1.6350258879098921e-06,
      "loss": 0.0882,
      "step": 900
    },
    {
      "epoch": 0.26155187445510025,
      "eval_loss": 0.08222991973161697,
      "eval_runtime": 27.0353,
      "eval_samples_per_second": 452.483,
      "eval_steps_per_second": 14.167,
      "step": 900
    },
    {
      "epoch": 0.2906131938390003,
      "grad_norm": 5.516987323760986,
      "learning_rate": 1.816695431010991e-06,
      "loss": 0.079,
      "step": 1000
    },
    {
      "epoch": 0.3196745132229003,
      "grad_norm": 0.9596192836761475,
      "learning_rate": 1.9983649741120904e-06,
      "loss": 0.071,
      "step": 1100
    },
    {
      "epoch": 0.34873583260680036,
      "grad_norm": 5.4598069190979,
      "learning_rate": 2.1800345172131893e-06,
      "loss": 0.0817,
      "step": 1200
    },
    {
      "epoch": 0.34873583260680036,
      "eval_loss": 0.07388558238744736,
      "eval_runtime": 26.4246,
      "eval_samples_per_second": 462.94,
      "eval_steps_per_second": 14.494,
      "step": 1200
    },
    {
      "epoch": 0.37779715199070035,
      "grad_norm": 3.363698959350586,
      "learning_rate": 2.3617040603142887e-06,
      "loss": 0.063,
      "step": 1300
    },
    {
      "epoch": 0.4068584713746004,
      "grad_norm": 4.268736839294434,
      "learning_rate": 2.5433736034153876e-06,
      "loss": 0.0726,
      "step": 1400
    },
    {
      "epoch": 0.43591979075850046,
      "grad_norm": 0.6281489133834839,
      "learning_rate": 2.7250431465164866e-06,
      "loss": 0.0767,
      "step": 1500
    },
    {
      "epoch": 0.43591979075850046,
      "eval_loss": 0.06805469840765,
      "eval_runtime": 27.0745,
      "eval_samples_per_second": 451.827,
      "eval_steps_per_second": 14.146,
      "step": 1500
    },
    {
      "epoch": 0.46498111014240046,
      "grad_norm": 1.1631907224655151,
      "learning_rate": 2.9067126896175855e-06,
      "loss": 0.0753,
      "step": 1600
    },
    {
      "epoch": 0.4940424295263005,
      "grad_norm": 1.6831640005111694,
      "learning_rate": 3.0883822327186853e-06,
      "loss": 0.0647,
      "step": 1700
    },
    {
      "epoch": 0.5231037489102005,
      "grad_norm": 0.39547181129455566,
      "learning_rate": 3.268235080388773e-06,
      "loss": 0.0538,
      "step": 1800
    },
    {
      "epoch": 0.5231037489102005,
      "eval_loss": 0.06306594610214233,
      "eval_runtime": 26.3785,
      "eval_samples_per_second": 463.748,
      "eval_steps_per_second": 14.519,
      "step": 1800
    },
    {
      "epoch": 0.5521650682941005,
      "grad_norm": 2.629936933517456,
      "learning_rate": 3.449904623489872e-06,
      "loss": 0.057,
      "step": 1900
    },
    {
      "epoch": 0.5812263876780006,
      "grad_norm": 0.44680893421173096,
      "learning_rate": 3.6315741665909717e-06,
      "loss": 0.0592,
      "step": 2000
    },
    {
      "epoch": 0.6102877070619006,
      "grad_norm": 6.168762683868408,
      "learning_rate": 3.8132437096920706e-06,
      "loss": 0.06,
      "step": 2100
    },
    {
      "epoch": 0.6102877070619006,
      "eval_loss": 0.058808207511901855,
      "eval_runtime": 27.1053,
      "eval_samples_per_second": 451.314,
      "eval_steps_per_second": 14.13,
      "step": 2100
    },
    {
      "epoch": 0.6393490264458006,
      "grad_norm": 2.6968770027160645,
      "learning_rate": 3.994913252793169e-06,
      "loss": 0.0626,
      "step": 2200
    },
    {
      "epoch": 0.6684103458297007,
      "grad_norm": 1.397504210472107,
      "learning_rate": 4.176582795894268e-06,
      "loss": 0.0562,
      "step": 2300
    },
    {
      "epoch": 0.6974716652136007,
      "grad_norm": 3.1529994010925293,
      "learning_rate": 4.358252338995368e-06,
      "loss": 0.0589,
      "step": 2400
    },
    {
      "epoch": 0.6974716652136007,
      "eval_loss": 0.05554657801985741,
      "eval_runtime": 26.2164,
      "eval_samples_per_second": 466.617,
      "eval_steps_per_second": 14.609,
      "step": 2400
    },
    {
      "epoch": 0.7265329845975007,
      "grad_norm": 3.10916805267334,
      "learning_rate": 4.539921882096467e-06,
      "loss": 0.0515,
      "step": 2500
    },
    {
      "epoch": 0.7555943039814007,
      "grad_norm": 1.4427434206008911,
      "learning_rate": 4.721591425197566e-06,
      "loss": 0.0581,
      "step": 2600
    },
    {
      "epoch": 0.7846556233653008,
      "grad_norm": 2.437389850616455,
      "learning_rate": 4.9032609682986655e-06,
      "loss": 0.0498,
      "step": 2700
    },
    {
      "epoch": 0.7846556233653008,
      "eval_loss": 0.053471144288778305,
      "eval_runtime": 26.9156,
      "eval_samples_per_second": 454.495,
      "eval_steps_per_second": 14.23,
      "step": 2700
    },
    {
      "epoch": 0.8137169427492008,
      "grad_norm": 1.7395578622817993,
      "learning_rate": 5.084930511399764e-06,
      "loss": 0.0503,
      "step": 2800
    },
    {
      "epoch": 0.8427782621331008,
      "grad_norm": 0.2161996066570282,
      "learning_rate": 5.266600054500863e-06,
      "loss": 0.047,
      "step": 2900
    },
    {
      "epoch": 0.8718395815170009,
      "grad_norm": 0.2165631204843521,
      "learning_rate": 5.448269597601963e-06,
      "loss": 0.0454,
      "step": 3000
    },
    {
      "epoch": 0.8718395815170009,
      "eval_loss": 0.052133820950984955,
      "eval_runtime": 26.4629,
      "eval_samples_per_second": 462.27,
      "eval_steps_per_second": 14.473,
      "step": 3000
    },
    {
      "epoch": 0.9009009009009009,
      "grad_norm": 0.16847111284732819,
      "learning_rate": 5.629939140703061e-06,
      "loss": 0.0497,
      "step": 3100
    },
    {
      "epoch": 0.9299622202848009,
      "grad_norm": 4.040170669555664,
      "learning_rate": 5.811608683804161e-06,
      "loss": 0.0524,
      "step": 3200
    },
    {
      "epoch": 0.9590235396687009,
      "grad_norm": 3.6626970767974854,
      "learning_rate": 5.99327822690526e-06,
      "loss": 0.0423,
      "step": 3300
    },
    {
      "epoch": 0.9590235396687009,
      "eval_loss": 0.04921015352010727,
      "eval_runtime": 27.2421,
      "eval_samples_per_second": 449.048,
      "eval_steps_per_second": 14.059,
      "step": 3300
    },
    {
      "epoch": 0.988084859052601,
      "grad_norm": 0.33825862407684326,
      "learning_rate": 6.174947770006359e-06,
      "loss": 0.0468,
      "step": 3400
    },
    {
      "epoch": 1.017146178436501,
      "grad_norm": 2.7132785320281982,
      "learning_rate": 6.356617313107458e-06,
      "loss": 0.0578,
      "step": 3500
    },
    {
      "epoch": 1.046207497820401,
      "grad_norm": 1.4403324127197266,
      "learning_rate": 6.538286856208557e-06,
      "loss": 0.0428,
      "step": 3600
    },
    {
      "epoch": 1.046207497820401,
      "eval_loss": 0.047407593578100204,
      "eval_runtime": 26.5867,
      "eval_samples_per_second": 460.117,
      "eval_steps_per_second": 14.406,
      "step": 3600
    },
    {
      "epoch": 1.075268817204301,
      "grad_norm": 3.5505058765411377,
      "learning_rate": 6.7199563993096566e-06,
      "loss": 0.0458,
      "step": 3700
    },
    {
      "epoch": 1.104330136588201,
      "grad_norm": 1.7245237827301025,
      "learning_rate": 6.9016259424107555e-06,
      "loss": 0.0377,
      "step": 3800
    },
    {
      "epoch": 1.1333914559721012,
      "grad_norm": 1.81324303150177,
      "learning_rate": 7.0832954855118544e-06,
      "loss": 0.0432,
      "step": 3900
    },
    {
      "epoch": 1.1333914559721012,
      "eval_loss": 0.04528222978115082,
      "eval_runtime": 26.7505,
      "eval_samples_per_second": 457.299,
      "eval_steps_per_second": 14.317,
      "step": 3900
    },
    {
      "epoch": 1.1624527753560012,
      "grad_norm": 0.5516623854637146,
      "learning_rate": 7.264965028612953e-06,
      "loss": 0.0402,
      "step": 4000
    },
    {
      "epoch": 1.1915140947399012,
      "grad_norm": 3.600156545639038,
      "learning_rate": 7.446634571714053e-06,
      "loss": 0.0403,
      "step": 4100
    },
    {
      "epoch": 1.2205754141238012,
      "grad_norm": 0.5710099935531616,
      "learning_rate": 7.628304114815151e-06,
      "loss": 0.0401,
      "step": 4200
    },
    {
      "epoch": 1.2205754141238012,
      "eval_loss": 0.04419328644871712,
      "eval_runtime": 26.7562,
      "eval_samples_per_second": 457.203,
      "eval_steps_per_second": 14.314,
      "step": 4200
    },
    {
      "epoch": 1.2496367335077012,
      "grad_norm": 0.9024702310562134,
      "learning_rate": 7.809973657916251e-06,
      "loss": 0.0441,
      "step": 4300
    },
    {
      "epoch": 1.2786980528916012,
      "grad_norm": 0.7580955028533936,
      "learning_rate": 7.99164320101735e-06,
      "loss": 0.0373,
      "step": 4400
    },
    {
      "epoch": 1.3077593722755014,
      "grad_norm": 0.7472540140151978,
      "learning_rate": 8.173312744118449e-06,
      "loss": 0.043,
      "step": 4500
    },
    {
      "epoch": 1.3077593722755014,
      "eval_loss": 0.0427699089050293,
      "eval_runtime": 26.7025,
      "eval_samples_per_second": 458.123,
      "eval_steps_per_second": 14.343,
      "step": 4500
    },
    {
      "epoch": 1.3368206916594012,
      "grad_norm": 5.362630367279053,
      "learning_rate": 8.354982287219548e-06,
      "loss": 0.0432,
      "step": 4600
    },
    {
      "epoch": 1.3658820110433014,
      "grad_norm": 2.7061116695404053,
      "learning_rate": 8.536651830320647e-06,
      "loss": 0.0386,
      "step": 4700
    },
    {
      "epoch": 1.3949433304272014,
      "grad_norm": 1.1733832359313965,
      "learning_rate": 8.718321373421746e-06,
      "loss": 0.0352,
      "step": 4800
    },
    {
      "epoch": 1.3949433304272014,
      "eval_loss": 0.041440799832344055,
      "eval_runtime": 26.9106,
      "eval_samples_per_second": 454.58,
      "eval_steps_per_second": 14.232,
      "step": 4800
    },
    {
      "epoch": 1.4240046498111014,
      "grad_norm": 2.7326438426971436,
      "learning_rate": 8.899990916522846e-06,
      "loss": 0.0389,
      "step": 4900
    },
    {
      "epoch": 1.4530659691950014,
      "grad_norm": 0.20390258729457855,
      "learning_rate": 9.081660459623944e-06,
      "loss": 0.04,
      "step": 5000
    },
    {
      "epoch": 1.4821272885789014,
      "grad_norm": 4.956657886505127,
      "learning_rate": 9.263330002725044e-06,
      "loss": 0.0394,
      "step": 5100
    },
    {
      "epoch": 1.4821272885789014,
      "eval_loss": 0.042791176587343216,
      "eval_runtime": 26.6527,
      "eval_samples_per_second": 458.977,
      "eval_steps_per_second": 14.37,
      "step": 5100
    },
    {
      "epoch": 1.5111886079628016,
      "grad_norm": 0.23265118896961212,
      "learning_rate": 9.444999545826143e-06,
      "loss": 0.0342,
      "step": 5200
    },
    {
      "epoch": 1.5402499273467014,
      "grad_norm": 2.009842872619629,
      "learning_rate": 9.626669088927242e-06,
      "loss": 0.0462,
      "step": 5300
    },
    {
      "epoch": 1.5693112467306016,
      "grad_norm": 0.5204980969429016,
      "learning_rate": 9.808338632028341e-06,
      "loss": 0.0412,
      "step": 5400
    },
    {
      "epoch": 1.5693112467306016,
      "eval_loss": 0.04058554396033287,
      "eval_runtime": 27.0549,
      "eval_samples_per_second": 452.154,
      "eval_steps_per_second": 14.156,
      "step": 5400
    },
    {
      "epoch": 1.5983725661145016,
      "grad_norm": 1.264519214630127,
      "learning_rate": 9.990008175129442e-06,
      "loss": 0.0352,
      "step": 5500
    },
    {
      "epoch": 1.6274338854984016,
      "grad_norm": 1.4044185876846313,
      "learning_rate": 1.0171677718230539e-05,
      "loss": 0.0363,
      "step": 5600
    },
    {
      "epoch": 1.6564952048823016,
      "grad_norm": 0.773714542388916,
      "learning_rate": 1.035334726133164e-05,
      "loss": 0.0416,
      "step": 5700
    },
    {
      "epoch": 1.6564952048823016,
      "eval_loss": 0.039202187210321426,
      "eval_runtime": 26.6017,
      "eval_samples_per_second": 459.858,
      "eval_steps_per_second": 14.398,
      "step": 5700
    },
    {
      "epoch": 1.6855565242662016,
      "grad_norm": 0.1811431348323822,
      "learning_rate": 1.0535016804432738e-05,
      "loss": 0.0287,
      "step": 5800
    },
    {
      "epoch": 1.7146178436501018,
      "grad_norm": 3.958439350128174,
      "learning_rate": 1.0716686347533837e-05,
      "loss": 0.0325,
      "step": 5900
    },
    {
      "epoch": 1.7436791630340016,
      "grad_norm": 0.5695263147354126,
      "learning_rate": 1.0898355890634935e-05,
      "loss": 0.0331,
      "step": 6000
    },
    {
      "epoch": 1.7436791630340016,
      "eval_loss": 0.037496764212846756,
      "eval_runtime": 26.9127,
      "eval_samples_per_second": 454.543,
      "eval_steps_per_second": 14.231,
      "step": 6000
    },
    {
      "epoch": 1.7727404824179018,
      "grad_norm": 3.3903560638427734,
      "learning_rate": 1.1080025433736035e-05,
      "loss": 0.0361,
      "step": 6100
    },
    {
      "epoch": 1.8018018018018018,
      "grad_norm": 2.8720123767852783,
      "learning_rate": 1.1261694976837134e-05,
      "loss": 0.0468,
      "step": 6200
    },
    {
      "epoch": 1.8308631211857018,
      "grad_norm": 1.0204403400421143,
      "learning_rate": 1.1443364519938233e-05,
      "loss": 0.0343,
      "step": 6300
    },
    {
      "epoch": 1.8308631211857018,
      "eval_loss": 0.037053827196359634,
      "eval_runtime": 26.3034,
      "eval_samples_per_second": 465.074,
      "eval_steps_per_second": 14.561,
      "step": 6300
    },
    {
      "epoch": 1.8599244405696018,
      "grad_norm": 2.0732407569885254,
      "learning_rate": 1.1625034063039334e-05,
      "loss": 0.0401,
      "step": 6400
    },
    {
      "epoch": 1.8889857599535018,
      "grad_norm": 3.727902412414551,
      "learning_rate": 1.1806703606140431e-05,
      "loss": 0.0425,
      "step": 6500
    },
    {
      "epoch": 1.918047079337402,
      "grad_norm": 1.8206995725631714,
      "learning_rate": 1.198837314924153e-05,
      "loss": 0.0342,
      "step": 6600
    },
    {
      "epoch": 1.918047079337402,
      "eval_loss": 0.038177795708179474,
      "eval_runtime": 27.1924,
      "eval_samples_per_second": 449.869,
      "eval_steps_per_second": 14.085,
      "step": 6600
    },
    {
      "epoch": 1.9471083987213018,
      "grad_norm": 0.9479349255561829,
      "learning_rate": 1.2170042692342629e-05,
      "loss": 0.0411,
      "step": 6700
    },
    {
      "epoch": 1.976169718105202,
      "grad_norm": 1.6102315187454224,
      "learning_rate": 1.235171223544373e-05,
      "loss": 0.0393,
      "step": 6800
    },
    {
      "epoch": 2.005231037489102,
      "grad_norm": 0.39504683017730713,
      "learning_rate": 1.2533381778544828e-05,
      "loss": 0.0379,
      "step": 6900
    },
    {
      "epoch": 2.005231037489102,
      "eval_loss": 0.037571169435977936,
      "eval_runtime": 26.9177,
      "eval_samples_per_second": 454.46,
      "eval_steps_per_second": 14.229,
      "step": 6900
    },
    {
      "epoch": 2.034292356873002,
      "grad_norm": 0.09198792278766632,
      "learning_rate": 1.2715051321645927e-05,
      "loss": 0.0226,
      "step": 7000
    },
    {
      "epoch": 2.0633536762569022,
      "grad_norm": 0.5564383864402771,
      "learning_rate": 1.2896720864747025e-05,
      "loss": 0.0251,
      "step": 7100
    },
    {
      "epoch": 2.092414995640802,
      "grad_norm": 1.9805023670196533,
      "learning_rate": 1.3078390407848125e-05,
      "loss": 0.0232,
      "step": 7200
    },
    {
      "epoch": 2.092414995640802,
      "eval_loss": 0.03626781702041626,
      "eval_runtime": 26.9532,
      "eval_samples_per_second": 453.86,
      "eval_steps_per_second": 14.21,
      "step": 7200
    },
    {
      "epoch": 2.1214763150247022,
      "grad_norm": 2.2151503562927246,
      "learning_rate": 1.3260059950949224e-05,
      "loss": 0.0311,
      "step": 7300
    },
    {
      "epoch": 2.150537634408602,
      "grad_norm": 3.625196695327759,
      "learning_rate": 1.3441729494050325e-05,
      "loss": 0.025,
      "step": 7400
    },
    {
      "epoch": 2.1795989537925022,
      "grad_norm": 3.1274352073669434,
      "learning_rate": 1.3623399037151424e-05,
      "loss": 0.0245,
      "step": 7500
    },
    {
      "epoch": 2.1795989537925022,
      "eval_loss": 0.036362066864967346,
      "eval_runtime": 26.6185,
      "eval_samples_per_second": 459.567,
      "eval_steps_per_second": 14.388,
      "step": 7500
    },
    {
      "epoch": 2.208660273176402,
      "grad_norm": 0.11472488194704056,
      "learning_rate": 1.3805068580252521e-05,
      "loss": 0.0291,
      "step": 7600
    },
    {
      "epoch": 2.2377215925603022,
      "grad_norm": 1.0436164140701294,
      "learning_rate": 1.398673812335362e-05,
      "loss": 0.03,
      "step": 7700
    },
    {
      "epoch": 2.2667829119442024,
      "grad_norm": 1.9227488040924072,
      "learning_rate": 1.416840766645472e-05,
      "loss": 0.0277,
      "step": 7800
    },
    {
      "epoch": 2.2667829119442024,
      "eval_loss": 0.03540974110364914,
      "eval_runtime": 26.7128,
      "eval_samples_per_second": 457.944,
      "eval_steps_per_second": 14.338,
      "step": 7800
    },
    {
      "epoch": 2.2958442313281022,
      "grad_norm": 4.112268447875977,
      "learning_rate": 1.4348260514124808e-05,
      "loss": 0.0242,
      "step": 7900
    },
    {
      "epoch": 2.3249055507120024,
      "grad_norm": 1.2703102827072144,
      "learning_rate": 1.4529930057225907e-05,
      "loss": 0.03,
      "step": 8000
    },
    {
      "epoch": 2.353966870095902,
      "grad_norm": 2.46053409576416,
      "learning_rate": 1.4711599600327007e-05,
      "loss": 0.0301,
      "step": 8100
    },
    {
      "epoch": 2.353966870095902,
      "eval_loss": 0.035567089915275574,
      "eval_runtime": 26.7341,
      "eval_samples_per_second": 457.58,
      "eval_steps_per_second": 14.326,
      "step": 8100
    },
    {
      "epoch": 2.3830281894798024,
      "grad_norm": 1.049849271774292,
      "learning_rate": 1.4893269143428106e-05,
      "loss": 0.0288,
      "step": 8200
    },
    {
      "epoch": 2.412089508863702,
      "grad_norm": 2.9456753730773926,
      "learning_rate": 1.5074938686529204e-05,
      "loss": 0.0197,
      "step": 8300
    },
    {
      "epoch": 2.4411508282476024,
      "grad_norm": 0.44164979457855225,
      "learning_rate": 1.5254791534199294e-05,
      "loss": 0.0351,
      "step": 8400
    },
    {
      "epoch": 2.4411508282476024,
      "eval_loss": 0.03602970018982887,
      "eval_runtime": 26.2631,
      "eval_samples_per_second": 465.787,
      "eval_steps_per_second": 14.583,
      "step": 8400
    },
    {
      "epoch": 2.4702121476315027,
      "grad_norm": 0.2530701160430908,
      "learning_rate": 1.543646107730039e-05,
      "loss": 0.0215,
      "step": 8500
    },
    {
      "epoch": 2.4992734670154024,
      "grad_norm": 0.1154685765504837,
      "learning_rate": 1.561813062040149e-05,
      "loss": 0.0359,
      "step": 8600
    },
    {
      "epoch": 2.5283347863993026,
      "grad_norm": 0.16534653306007385,
      "learning_rate": 1.579980016350259e-05,
      "loss": 0.0257,
      "step": 8700
    },
    {
      "epoch": 2.5283347863993026,
      "eval_loss": 0.03711829334497452,
      "eval_runtime": 26.9334,
      "eval_samples_per_second": 454.194,
      "eval_steps_per_second": 14.22,
      "step": 8700
    },
    {
      "epoch": 2.5573961057832024,
      "grad_norm": 4.247994899749756,
      "learning_rate": 1.5981469706603688e-05,
      "loss": 0.025,
      "step": 8800
    },
    {
      "epoch": 2.5864574251671026,
      "grad_norm": 3.4336764812469482,
      "learning_rate": 1.616313924970479e-05,
      "loss": 0.0337,
      "step": 8900
    },
    {
      "epoch": 2.615518744551003,
      "grad_norm": 3.021991014480591,
      "learning_rate": 1.634480879280589e-05,
      "loss": 0.0236,
      "step": 9000
    },
    {
      "epoch": 2.615518744551003,
      "eval_loss": 0.03501100093126297,
      "eval_runtime": 26.2977,
      "eval_samples_per_second": 465.175,
      "eval_steps_per_second": 14.564,
      "step": 9000
    },
    {
      "epoch": 2.6445800639349026,
      "grad_norm": 4.529074192047119,
      "learning_rate": 1.6526478335906987e-05,
      "loss": 0.0245,
      "step": 9100
    },
    {
      "epoch": 2.6736413833188024,
      "grad_norm": 2.828688859939575,
      "learning_rate": 1.6708147879008084e-05,
      "loss": 0.0293,
      "step": 9200
    },
    {
      "epoch": 2.7027027027027026,
      "grad_norm": 0.3687061071395874,
      "learning_rate": 1.6889817422109185e-05,
      "loss": 0.0291,
      "step": 9300
    },
    {
      "epoch": 2.7027027027027026,
      "eval_loss": 0.03628409281373024,
      "eval_runtime": 26.8142,
      "eval_samples_per_second": 456.214,
      "eval_steps_per_second": 14.283,
      "step": 9300
    },
    {
      "epoch": 2.731764022086603,
      "grad_norm": 0.35936692357063293,
      "learning_rate": 1.7071486965210285e-05,
      "loss": 0.0294,
      "step": 9400
    },
    {
      "epoch": 2.7608253414705026,
      "grad_norm": 0.48872271180152893,
      "learning_rate": 1.7253156508311382e-05,
      "loss": 0.0273,
      "step": 9500
    },
    {
      "epoch": 2.789886660854403,
      "grad_norm": 0.6060481071472168,
      "learning_rate": 1.743482605141248e-05,
      "loss": 0.0358,
      "step": 9600
    },
    {
      "epoch": 2.789886660854403,
      "eval_loss": 0.037452585995197296,
      "eval_runtime": 26.8066,
      "eval_samples_per_second": 456.344,
      "eval_steps_per_second": 14.288,
      "step": 9600
    },
    {
      "epoch": 2.8189479802383026,
      "grad_norm": 0.19470466673374176,
      "learning_rate": 1.761649559451358e-05,
      "loss": 0.0251,
      "step": 9700
    },
    {
      "epoch": 2.848009299622203,
      "grad_norm": 0.7011229991912842,
      "learning_rate": 1.779816513761468e-05,
      "loss": 0.0352,
      "step": 9800
    },
    {
      "epoch": 2.877070619006103,
      "grad_norm": 1.1960341930389404,
      "learning_rate": 1.7979834680715778e-05,
      "loss": 0.0289,
      "step": 9900
    },
    {
      "epoch": 2.877070619006103,
      "eval_loss": 0.036693498492240906,
      "eval_runtime": 28.4942,
      "eval_samples_per_second": 429.315,
      "eval_steps_per_second": 13.441,
      "step": 9900
    },
    {
      "epoch": 2.906131938390003,
      "grad_norm": 0.120386503636837,
      "learning_rate": 1.816150422381688e-05,
      "loss": 0.0306,
      "step": 10000
    },
    {
      "epoch": 2.935193257773903,
      "grad_norm": 0.15321113169193268,
      "learning_rate": 1.834317376691798e-05,
      "loss": 0.0249,
      "step": 10100
    },
    {
      "epoch": 2.964254577157803,
      "grad_norm": 0.7859700918197632,
      "learning_rate": 1.8524843310019077e-05,
      "loss": 0.0257,
      "step": 10200
    },
    {
      "epoch": 2.964254577157803,
      "eval_loss": 0.036162618547677994,
      "eval_runtime": 30.9741,
      "eval_samples_per_second": 394.944,
      "eval_steps_per_second": 12.365,
      "step": 10200
    },
    {
      "epoch": 2.993315896541703,
      "grad_norm": 0.7350950837135315,
      "learning_rate": 1.8706512853120174e-05,
      "loss": 0.0332,
      "step": 10300
    },
    {
      "epoch": 3.022377215925603,
      "grad_norm": 1.5788044929504395,
      "learning_rate": 1.8888182396221275e-05,
      "loss": 0.0208,
      "step": 10400
    },
    {
      "epoch": 3.051438535309503,
      "grad_norm": 1.8751883506774902,
      "learning_rate": 1.9069851939322375e-05,
      "loss": 0.0231,
      "step": 10500
    },
    {
      "epoch": 3.051438535309503,
      "eval_loss": 0.03594466298818588,
      "eval_runtime": 29.5347,
      "eval_samples_per_second": 414.191,
      "eval_steps_per_second": 12.968,
      "step": 10500
    },
    {
      "epoch": 3.0804998546934033,
      "grad_norm": 2.0853710174560547,
      "learning_rate": 1.9251521482423476e-05,
      "loss": 0.0216,
      "step": 10600
    },
    {
      "epoch": 3.109561174077303,
      "grad_norm": 0.48473218083381653,
      "learning_rate": 1.9433191025524573e-05,
      "loss": 0.0193,
      "step": 10700
    },
    {
      "epoch": 3.1386224934612033,
      "grad_norm": 0.1810705065727234,
      "learning_rate": 1.961486056862567e-05,
      "loss": 0.0175,
      "step": 10800
    },
    {
      "epoch": 3.1386224934612033,
      "eval_loss": 0.03672377020120621,
      "eval_runtime": 32.0584,
      "eval_samples_per_second": 381.584,
      "eval_steps_per_second": 11.947,
      "step": 10800
    },
    {
      "epoch": 3.167683812845103,
      "grad_norm": 0.8471315503120422,
      "learning_rate": 1.979653011172677e-05,
      "loss": 0.0219,
      "step": 10900
    },
    {
      "epoch": 3.1967451322290033,
      "grad_norm": 0.23426327109336853,
      "learning_rate": 1.997819965482787e-05,
      "loss": 0.0188,
      "step": 11000
    },
    {
      "epoch": 3.225806451612903,
      "grad_norm": 0.2814389765262604,
      "learning_rate": 1.9865350776528193e-05,
      "loss": 0.0188,
      "step": 11100
    },
    {
      "epoch": 3.225806451612903,
      "eval_loss": 0.034333910793066025,
      "eval_runtime": 27.346,
      "eval_samples_per_second": 447.341,
      "eval_steps_per_second": 14.006,
      "step": 11100
    },
    {
      "epoch": 3.2548677709968032,
      "grad_norm": 2.401367425918579,
      "learning_rate": 1.971234029531023e-05,
      "loss": 0.0265,
      "step": 11200
    },
    {
      "epoch": 3.2839290903807035,
      "grad_norm": 0.25739622116088867,
      "learning_rate": 1.9559329814092266e-05,
      "loss": 0.0218,
      "step": 11300
    },
    {
      "epoch": 3.3129904097646032,
      "grad_norm": 0.7228596210479736,
      "learning_rate": 1.9406319332874305e-05,
      "loss": 0.0208,
      "step": 11400
    },
    {
      "epoch": 3.3129904097646032,
      "eval_loss": 0.03501081466674805,
      "eval_runtime": 31.5795,
      "eval_samples_per_second": 387.372,
      "eval_steps_per_second": 12.128,
      "step": 11400
    },
    {
      "epoch": 3.3420517291485035,
      "grad_norm": 0.4601193964481354,
      "learning_rate": 1.925330885165634e-05,
      "loss": 0.0184,
      "step": 11500
    },
    {
      "epoch": 3.3711130485324032,
      "grad_norm": 0.033944468945264816,
      "learning_rate": 1.9100298370438376e-05,
      "loss": 0.0232,
      "step": 11600
    },
    {
      "epoch": 3.4001743679163035,
      "grad_norm": 5.643810749053955,
      "learning_rate": 1.8947287889220414e-05,
      "loss": 0.0193,
      "step": 11700
    },
    {
      "epoch": 3.4001743679163035,
      "eval_loss": 0.03553704172372818,
      "eval_runtime": 27.8198,
      "eval_samples_per_second": 439.723,
      "eval_steps_per_second": 13.767,
      "step": 11700
    },
    {
      "epoch": 3.4292356873002037,
      "grad_norm": 0.33075031638145447,
      "learning_rate": 1.879427740800245e-05,
      "loss": 0.0147,
      "step": 11800
    },
    {
      "epoch": 3.4582970066841034,
      "grad_norm": 0.36583471298217773,
      "learning_rate": 1.8641266926784488e-05,
      "loss": 0.0209,
      "step": 11900
    },
    {
      "epoch": 3.4873583260680037,
      "grad_norm": 0.2965396046638489,
      "learning_rate": 1.8488256445566524e-05,
      "loss": 0.028,
      "step": 12000
    },
    {
      "epoch": 3.4873583260680037,
      "eval_loss": 0.03441368043422699,
      "eval_runtime": 31.5965,
      "eval_samples_per_second": 387.163,
      "eval_steps_per_second": 12.122,
      "step": 12000
    },
    {
      "epoch": 3.5164196454519034,
      "grad_norm": 0.10767892003059387,
      "learning_rate": 1.833524596434856e-05,
      "loss": 0.0203,
      "step": 12100
    },
    {
      "epoch": 3.5454809648358037,
      "grad_norm": 0.4897698163986206,
      "learning_rate": 1.8182235483130597e-05,
      "loss": 0.0186,
      "step": 12200
    },
    {
      "epoch": 3.5745422842197034,
      "grad_norm": 0.7638312578201294,
      "learning_rate": 1.8029225001912633e-05,
      "loss": 0.0233,
      "step": 12300
    },
    {
      "epoch": 3.5745422842197034,
      "eval_loss": 0.0343441404402256,
      "eval_runtime": 27.0272,
      "eval_samples_per_second": 452.618,
      "eval_steps_per_second": 14.171,
      "step": 12300
    },
    {
      "epoch": 3.6036036036036037,
      "grad_norm": 2.574690341949463,
      "learning_rate": 1.7876214520694668e-05,
      "loss": 0.0231,
      "step": 12400
    },
    {
      "epoch": 3.6326649229875034,
      "grad_norm": 0.528499960899353,
      "learning_rate": 1.7723204039476703e-05,
      "loss": 0.022,
      "step": 12500
    },
    {
      "epoch": 3.6617262423714037,
      "grad_norm": 0.08761309087276459,
      "learning_rate": 1.7570193558258742e-05,
      "loss": 0.0232,
      "step": 12600
    },
    {
      "epoch": 3.6617262423714037,
      "eval_loss": 0.03451988101005554,
      "eval_runtime": 32.6592,
      "eval_samples_per_second": 374.565,
      "eval_steps_per_second": 11.727,
      "step": 12600
    },
    {
      "epoch": 3.690787561755304,
      "grad_norm": 0.10693353414535522,
      "learning_rate": 1.7417183077040777e-05,
      "loss": 0.0249,
      "step": 12700
    },
    {
      "epoch": 3.7198488811392036,
      "grad_norm": 0.23877334594726562,
      "learning_rate": 1.7264172595822816e-05,
      "loss": 0.0241,
      "step": 12800
    },
    {
      "epoch": 3.748910200523104,
      "grad_norm": 0.376647025346756,
      "learning_rate": 1.711116211460485e-05,
      "loss": 0.025,
      "step": 12900
    },
    {
      "epoch": 3.748910200523104,
      "eval_loss": 0.033673714846372604,
      "eval_runtime": 27.1988,
      "eval_samples_per_second": 449.763,
      "eval_steps_per_second": 14.082,
      "step": 12900
    },
    {
      "epoch": 3.7779715199070036,
      "grad_norm": 0.7081992030143738,
      "learning_rate": 1.6958151633386886e-05,
      "loss": 0.0182,
      "step": 13000
    },
    {
      "epoch": 3.807032839290904,
      "grad_norm": 0.9505396485328674,
      "learning_rate": 1.6805141152168925e-05,
      "loss": 0.0197,
      "step": 13100
    },
    {
      "epoch": 3.836094158674804,
      "grad_norm": 0.3097359836101532,
      "learning_rate": 1.665213067095096e-05,
      "loss": 0.0187,
      "step": 13200
    },
    {
      "epoch": 3.836094158674804,
      "eval_loss": 0.03151945769786835,
      "eval_runtime": 31.0646,
      "eval_samples_per_second": 393.793,
      "eval_steps_per_second": 12.329,
      "step": 13200
    },
    {
      "epoch": 3.865155478058704,
      "grad_norm": 2.482163667678833,
      "learning_rate": 1.6499120189733e-05,
      "loss": 0.0168,
      "step": 13300
    },
    {
      "epoch": 3.8942167974426036,
      "grad_norm": 0.2182529866695404,
      "learning_rate": 1.6346109708515034e-05,
      "loss": 0.0244,
      "step": 13400
    },
    {
      "epoch": 3.923278116826504,
      "grad_norm": 0.4611717462539673,
      "learning_rate": 1.619309922729707e-05,
      "loss": 0.0179,
      "step": 13500
    },
    {
      "epoch": 3.923278116826504,
      "eval_loss": 0.03170377016067505,
      "eval_runtime": 28.0835,
      "eval_samples_per_second": 435.594,
      "eval_steps_per_second": 13.638,
      "step": 13500
    },
    {
      "epoch": 3.952339436210404,
      "grad_norm": 2.3013105392456055,
      "learning_rate": 1.6040088746079108e-05,
      "loss": 0.019,
      "step": 13600
    },
    {
      "epoch": 3.981400755594304,
      "grad_norm": 1.6527272462844849,
      "learning_rate": 1.5887078264861143e-05,
      "loss": 0.0196,
      "step": 13700
    },
    {
      "epoch": 4.010462074978204,
      "grad_norm": 0.042673151940107346,
      "learning_rate": 1.5734067783643182e-05,
      "loss": 0.0162,
      "step": 13800
    },
    {
      "epoch": 4.010462074978204,
      "eval_loss": 0.03321225941181183,
      "eval_runtime": 30.968,
      "eval_samples_per_second": 395.021,
      "eval_steps_per_second": 12.368,
      "step": 13800
    },
    {
      "epoch": 4.039523394362104,
      "grad_norm": 1.3607178926467896,
      "learning_rate": 1.5581057302425217e-05,
      "loss": 0.0141,
      "step": 13900
    },
    {
      "epoch": 4.068584713746004,
      "grad_norm": 0.053625594824552536,
      "learning_rate": 1.5428046821207253e-05,
      "loss": 0.0172,
      "step": 14000
    },
    {
      "epoch": 4.097646033129904,
      "grad_norm": 1.0355161428451538,
      "learning_rate": 1.527503633998929e-05,
      "loss": 0.0173,
      "step": 14100
    },
    {
      "epoch": 4.097646033129904,
      "eval_loss": 0.032125264406204224,
      "eval_runtime": 29.8018,
      "eval_samples_per_second": 410.478,
      "eval_steps_per_second": 12.852,
      "step": 14100
    },
    {
      "epoch": 4.1267073525138045,
      "grad_norm": 0.1364974081516266,
      "learning_rate": 1.5122025858771328e-05,
      "loss": 0.0126,
      "step": 14200
    },
    {
      "epoch": 4.155768671897704,
      "grad_norm": 0.20887862145900726,
      "learning_rate": 1.4969015377553364e-05,
      "loss": 0.0113,
      "step": 14300
    },
    {
      "epoch": 4.184829991281604,
      "grad_norm": 0.16281986236572266,
      "learning_rate": 1.48160048963354e-05,
      "loss": 0.017,
      "step": 14400
    },
    {
      "epoch": 4.184829991281604,
      "eval_loss": 0.03163045644760132,
      "eval_runtime": 30.8438,
      "eval_samples_per_second": 396.611,
      "eval_steps_per_second": 12.417,
      "step": 14400
    },
    {
      "epoch": 4.213891310665504,
      "grad_norm": 0.06442072987556458,
      "learning_rate": 1.4662994415117438e-05,
      "loss": 0.0132,
      "step": 14500
    },
    {
      "epoch": 4.2429526300494045,
      "grad_norm": 0.49993807077407837,
      "learning_rate": 1.4509983933899474e-05,
      "loss": 0.0137,
      "step": 14600
    },
    {
      "epoch": 4.272013949433305,
      "grad_norm": 0.045188985764980316,
      "learning_rate": 1.4356973452681511e-05,
      "loss": 0.0153,
      "step": 14700
    },
    {
      "epoch": 4.272013949433305,
      "eval_loss": 0.031595949083566666,
      "eval_runtime": 30.8789,
      "eval_samples_per_second": 396.16,
      "eval_steps_per_second": 12.403,
      "step": 14700
    },
    {
      "epoch": 4.301075268817204,
      "grad_norm": 0.030895119532942772,
      "learning_rate": 1.4205493076275728e-05,
      "loss": 0.0141,
      "step": 14800
    },
    {
      "epoch": 4.330136588201104,
      "grad_norm": 1.5610733032226562,
      "learning_rate": 1.4052482595057764e-05,
      "loss": 0.0117,
      "step": 14900
    },
    {
      "epoch": 4.3591979075850045,
      "grad_norm": 1.8453948497772217,
      "learning_rate": 1.3899472113839798e-05,
      "loss": 0.0169,
      "step": 15000
    },
    {
      "epoch": 4.3591979075850045,
      "eval_loss": 0.03333577141165733,
      "eval_runtime": 30.1339,
      "eval_samples_per_second": 405.955,
      "eval_steps_per_second": 12.71,
      "step": 15000
    },
    {
      "epoch": 4.388259226968905,
      "grad_norm": 1.0826033353805542,
      "learning_rate": 1.3746461632621835e-05,
      "loss": 0.0169,
      "step": 15100
    },
    {
      "epoch": 4.417320546352804,
      "grad_norm": 0.9390700459480286,
      "learning_rate": 1.3593451151403872e-05,
      "loss": 0.0134,
      "step": 15200
    },
    {
      "epoch": 4.446381865736704,
      "grad_norm": 0.7679401636123657,
      "learning_rate": 1.3440440670185907e-05,
      "loss": 0.0178,
      "step": 15300
    },
    {
      "epoch": 4.446381865736704,
      "eval_loss": 0.030997373163700104,
      "eval_runtime": 31.2406,
      "eval_samples_per_second": 391.574,
      "eval_steps_per_second": 12.26,
      "step": 15300
    },
    {
      "epoch": 4.4754431851206045,
      "grad_norm": 1.1060469150543213,
      "learning_rate": 1.3287430188967944e-05,
      "loss": 0.0155,
      "step": 15400
    },
    {
      "epoch": 4.504504504504505,
      "grad_norm": 2.564753293991089,
      "learning_rate": 1.313594981256216e-05,
      "loss": 0.0178,
      "step": 15500
    },
    {
      "epoch": 4.533565823888405,
      "grad_norm": 0.15116174519062042,
      "learning_rate": 1.2982939331344197e-05,
      "loss": 0.0098,
      "step": 15600
    },
    {
      "epoch": 4.533565823888405,
      "eval_loss": 0.030752593651413918,
      "eval_runtime": 29.6271,
      "eval_samples_per_second": 412.899,
      "eval_steps_per_second": 12.927,
      "step": 15600
    },
    {
      "epoch": 4.562627143272304,
      "grad_norm": 0.29445090889930725,
      "learning_rate": 1.2829928850126234e-05,
      "loss": 0.0118,
      "step": 15700
    },
    {
      "epoch": 4.5916884626562045,
      "grad_norm": 0.19161444902420044,
      "learning_rate": 1.2676918368908271e-05,
      "loss": 0.0122,
      "step": 15800
    },
    {
      "epoch": 4.620749782040105,
      "grad_norm": 0.15567593276500702,
      "learning_rate": 1.2523907887690308e-05,
      "loss": 0.0138,
      "step": 15900
    },
    {
      "epoch": 4.620749782040105,
      "eval_loss": 0.030682172626256943,
      "eval_runtime": 32.042,
      "eval_samples_per_second": 381.78,
      "eval_steps_per_second": 11.953,
      "step": 15900
    },
    {
      "epoch": 4.649811101424005,
      "grad_norm": 0.9265925884246826,
      "learning_rate": 1.2370897406472343e-05,
      "loss": 0.0125,
      "step": 16000
    },
    {
      "epoch": 4.678872420807904,
      "grad_norm": 0.4911397695541382,
      "learning_rate": 1.221788692525438e-05,
      "loss": 0.0135,
      "step": 16100
    },
    {
      "epoch": 4.707933740191804,
      "grad_norm": 4.223952293395996,
      "learning_rate": 1.2064876444036417e-05,
      "loss": 0.0155,
      "step": 16200
    },
    {
      "epoch": 4.707933740191804,
      "eval_loss": 0.029881037771701813,
      "eval_runtime": 29.3642,
      "eval_samples_per_second": 416.595,
      "eval_steps_per_second": 13.043,
      "step": 16200
    },
    {
      "epoch": 4.736995059575705,
      "grad_norm": 0.14090295135974884,
      "learning_rate": 1.1911865962818454e-05,
      "loss": 0.0129,
      "step": 16300
    },
    {
      "epoch": 4.766056378959605,
      "grad_norm": 4.120369911193848,
      "learning_rate": 1.1758855481600491e-05,
      "loss": 0.0131,
      "step": 16400
    },
    {
      "epoch": 4.795117698343505,
      "grad_norm": 0.8263002634048462,
      "learning_rate": 1.1605845000382527e-05,
      "loss": 0.0114,
      "step": 16500
    },
    {
      "epoch": 4.795117698343505,
      "eval_loss": 0.029694661498069763,
      "eval_runtime": 31.2583,
      "eval_samples_per_second": 391.352,
      "eval_steps_per_second": 12.253,
      "step": 16500
    },
    {
      "epoch": 4.824179017727404,
      "grad_norm": 0.40641260147094727,
      "learning_rate": 1.1452834519164563e-05,
      "loss": 0.0131,
      "step": 16600
    },
    {
      "epoch": 4.853240337111305,
      "grad_norm": 0.818534791469574,
      "learning_rate": 1.12998240379466e-05,
      "loss": 0.0131,
      "step": 16700
    },
    {
      "epoch": 4.882301656495205,
      "grad_norm": 0.12895511090755463,
      "learning_rate": 1.1146813556728637e-05,
      "loss": 0.0144,
      "step": 16800
    },
    {
      "epoch": 4.882301656495205,
      "eval_loss": 0.02863692305982113,
      "eval_runtime": 28.8059,
      "eval_samples_per_second": 424.671,
      "eval_steps_per_second": 13.296,
      "step": 16800
    },
    {
      "epoch": 4.911362975879105,
      "grad_norm": 0.09267265349626541,
      "learning_rate": 1.0993803075510674e-05,
      "loss": 0.0116,
      "step": 16900
    },
    {
      "epoch": 4.940424295263005,
      "grad_norm": 0.10310707986354828,
      "learning_rate": 1.0840792594292711e-05,
      "loss": 0.0141,
      "step": 17000
    },
    {
      "epoch": 4.969485614646905,
      "grad_norm": 0.27060624957084656,
      "learning_rate": 1.0687782113074747e-05,
      "loss": 0.019,
      "step": 17100
    },
    {
      "epoch": 4.969485614646905,
      "eval_loss": 0.02940661646425724,
      "eval_runtime": 31.7266,
      "eval_samples_per_second": 385.575,
      "eval_steps_per_second": 12.072,
      "step": 17100
    },
    {
      "epoch": 4.998546934030805,
      "grad_norm": 0.17917053401470184,
      "learning_rate": 1.0534771631856784e-05,
      "loss": 0.0141,
      "step": 17200
    },
    {
      "epoch": 5.027608253414705,
      "grad_norm": 1.1561379432678223,
      "learning_rate": 1.038176115063882e-05,
      "loss": 0.0136,
      "step": 17300
    },
    {
      "epoch": 5.056669572798605,
      "grad_norm": 0.08390885591506958,
      "learning_rate": 1.0228750669420858e-05,
      "loss": 0.0123,
      "step": 17400
    },
    {
      "epoch": 5.056669572798605,
      "eval_loss": 0.027470601722598076,
      "eval_runtime": 29.8277,
      "eval_samples_per_second": 410.122,
      "eval_steps_per_second": 12.84,
      "step": 17400
    },
    {
      "epoch": 5.0857308921825055,
      "grad_norm": 1.385046362876892,
      "learning_rate": 1.0075740188202894e-05,
      "loss": 0.0124,
      "step": 17500
    },
    {
      "epoch": 5.114792211566405,
      "grad_norm": 0.07293292135000229,
      "learning_rate": 9.92272970698493e-06,
      "loss": 0.0088,
      "step": 17600
    },
    {
      "epoch": 5.143853530950305,
      "grad_norm": 0.14433690905570984,
      "learning_rate": 9.769719225766967e-06,
      "loss": 0.0079,
      "step": 17700
    },
    {
      "epoch": 5.143853530950305,
      "eval_loss": 0.028321975842118263,
      "eval_runtime": 31.9063,
      "eval_samples_per_second": 383.404,
      "eval_steps_per_second": 12.004,
      "step": 17700
    },
    {
      "epoch": 5.172914850334205,
      "grad_norm": 0.06696674972772598,
      "learning_rate": 9.616708744549002e-06,
      "loss": 0.0097,
      "step": 17800
    },
    {
      "epoch": 5.2019761697181055,
      "grad_norm": 0.5811833739280701,
      "learning_rate": 9.463698263331039e-06,
      "loss": 0.0085,
      "step": 17900
    },
    {
      "epoch": 5.231037489102005,
      "grad_norm": 0.05103592202067375,
      "learning_rate": 9.310687782113076e-06,
      "loss": 0.0159,
      "step": 18000
    },
    {
      "epoch": 5.231037489102005,
      "eval_loss": 0.029669631272554398,
      "eval_runtime": 28.3217,
      "eval_samples_per_second": 431.931,
      "eval_steps_per_second": 13.523,
      "step": 18000
    },
    {
      "epoch": 5.260098808485905,
      "grad_norm": 0.31854337453842163,
      "learning_rate": 9.157677300895113e-06,
      "loss": 0.0092,
      "step": 18100
    },
    {
      "epoch": 5.289160127869805,
      "grad_norm": 0.05867089703679085,
      "learning_rate": 9.00466681967715e-06,
      "loss": 0.0168,
      "step": 18200
    },
    {
      "epoch": 5.3182214472537055,
      "grad_norm": 0.01062140055000782,
      "learning_rate": 8.851656338459185e-06,
      "loss": 0.01,
      "step": 18300
    },
    {
      "epoch": 5.3182214472537055,
      "eval_loss": 0.02898811176419258,
      "eval_runtime": 30.964,
      "eval_samples_per_second": 395.071,
      "eval_steps_per_second": 12.369,
      "step": 18300
    },
    {
      "epoch": 5.347282766637606,
      "grad_norm": 0.23528796434402466,
      "learning_rate": 8.698645857241222e-06,
      "loss": 0.0086,
      "step": 18400
    },
    {
      "epoch": 5.376344086021505,
      "grad_norm": 0.7098844647407532,
      "learning_rate": 8.545635376023257e-06,
      "loss": 0.0083,
      "step": 18500
    },
    {
      "epoch": 5.405405405405405,
      "grad_norm": 0.0675228089094162,
      "learning_rate": 8.392624894805294e-06,
      "loss": 0.0127,
      "step": 18600
    },
    {
      "epoch": 5.405405405405405,
      "eval_loss": 0.028332242742180824,
      "eval_runtime": 28.1702,
      "eval_samples_per_second": 434.253,
      "eval_steps_per_second": 13.596,
      "step": 18600
    },
    {
      "epoch": 5.4344667247893055,
      "grad_norm": 0.8665882349014282,
      "learning_rate": 8.239614413587331e-06,
      "loss": 0.0107,
      "step": 18700
    },
    {
      "epoch": 5.463528044173206,
      "grad_norm": 0.13911496102809906,
      "learning_rate": 8.086603932369368e-06,
      "loss": 0.0121,
      "step": 18800
    },
    {
      "epoch": 5.492589363557105,
      "grad_norm": 0.08510652929544449,
      "learning_rate": 7.935123555963584e-06,
      "loss": 0.0098,
      "step": 18900
    },
    {
      "epoch": 5.492589363557105,
      "eval_loss": 0.02792692370712757,
      "eval_runtime": 30.8767,
      "eval_samples_per_second": 396.189,
      "eval_steps_per_second": 12.404,
      "step": 18900
    },
    {
      "epoch": 5.521650682941005,
      "grad_norm": 0.048594117164611816,
      "learning_rate": 7.782113074745621e-06,
      "loss": 0.014,
      "step": 19000
    },
    {
      "epoch": 5.5507120023249055,
      "grad_norm": 0.09377363324165344,
      "learning_rate": 7.629102593527657e-06,
      "loss": 0.0114,
      "step": 19100
    },
    {
      "epoch": 5.579773321708806,
      "grad_norm": 0.033514928072690964,
      "learning_rate": 7.476092112309694e-06,
      "loss": 0.012,
      "step": 19200
    },
    {
      "epoch": 5.579773321708806,
      "eval_loss": 0.027103085070848465,
      "eval_runtime": 27.6915,
      "eval_samples_per_second": 441.76,
      "eval_steps_per_second": 13.831,
      "step": 19200
    },
    {
      "epoch": 5.608834641092706,
      "grad_norm": 0.8264727592468262,
      "learning_rate": 7.3230816310917305e-06,
      "loss": 0.0105,
      "step": 19300
    },
    {
      "epoch": 5.637895960476605,
      "grad_norm": 2.6640799045562744,
      "learning_rate": 7.1700711498737675e-06,
      "loss": 0.0142,
      "step": 19400
    },
    {
      "epoch": 5.6669572798605055,
      "grad_norm": 0.03503794968128204,
      "learning_rate": 7.017060668655804e-06,
      "loss": 0.0096,
      "step": 19500
    },
    {
      "epoch": 5.6669572798605055,
      "eval_loss": 0.02664261683821678,
      "eval_runtime": 29.6973,
      "eval_samples_per_second": 411.923,
      "eval_steps_per_second": 12.897,
      "step": 19500
    },
    {
      "epoch": 5.696018599244406,
      "grad_norm": 0.022000476717948914,
      "learning_rate": 6.86405018743784e-06,
      "loss": 0.0113,
      "step": 19600
    },
    {
      "epoch": 5.725079918628306,
      "grad_norm": 0.03974687680602074,
      "learning_rate": 6.711039706219876e-06,
      "loss": 0.0119,
      "step": 19700
    },
    {
      "epoch": 5.754141238012206,
      "grad_norm": 0.0372069887816906,
      "learning_rate": 6.558029225001913e-06,
      "loss": 0.0142,
      "step": 19800
    },
    {
      "epoch": 5.754141238012206,
      "eval_loss": 0.027487969025969505,
      "eval_runtime": 27.3827,
      "eval_samples_per_second": 446.742,
      "eval_steps_per_second": 13.987,
      "step": 19800
    },
    {
      "epoch": 5.7832025573961054,
      "grad_norm": 0.24150213599205017,
      "learning_rate": 6.40501874378395e-06,
      "loss": 0.0097,
      "step": 19900
    },
    {
      "epoch": 5.812263876780006,
      "grad_norm": 1.2151767015457153,
      "learning_rate": 6.252008262565986e-06,
      "loss": 0.008,
      "step": 20000
    },
    {
      "epoch": 5.841325196163906,
      "grad_norm": 0.12997518479824066,
      "learning_rate": 6.098997781348023e-06,
      "loss": 0.0103,
      "step": 20100
    },
    {
      "epoch": 5.841325196163906,
      "eval_loss": 0.027195250615477562,
      "eval_runtime": 30.0975,
      "eval_samples_per_second": 406.446,
      "eval_steps_per_second": 12.725,
      "step": 20100
    },
    {
      "epoch": 5.870386515547806,
      "grad_norm": 1.2384765148162842,
      "learning_rate": 5.94598730013006e-06,
      "loss": 0.0115,
      "step": 20200
    },
    {
      "epoch": 5.899447834931706,
      "grad_norm": 0.46222779154777527,
      "learning_rate": 5.792976818912096e-06,
      "loss": 0.0107,
      "step": 20300
    },
    {
      "epoch": 5.928509154315606,
      "grad_norm": 0.08042553812265396,
      "learning_rate": 5.639966337694133e-06,
      "loss": 0.0089,
      "step": 20400
    },
    {
      "epoch": 5.928509154315606,
      "eval_loss": 0.02774212509393692,
      "eval_runtime": 27.6569,
      "eval_samples_per_second": 442.313,
      "eval_steps_per_second": 13.848,
      "step": 20400
    },
    {
      "epoch": 5.957570473699506,
      "grad_norm": 0.01730126515030861,
      "learning_rate": 5.486955856476169e-06,
      "loss": 0.0114,
      "step": 20500
    },
    {
      "epoch": 5.986631793083406,
      "grad_norm": 0.11168185621500015,
      "learning_rate": 5.333945375258206e-06,
      "loss": 0.0097,
      "step": 20600
    },
    {
      "epoch": 6.015693112467306,
      "grad_norm": 0.294203519821167,
      "learning_rate": 5.180934894040243e-06,
      "loss": 0.0098,
      "step": 20700
    },
    {
      "epoch": 6.015693112467306,
      "eval_loss": 0.027102118358016014,
      "eval_runtime": 30.4406,
      "eval_samples_per_second": 401.864,
      "eval_steps_per_second": 12.582,
      "step": 20700
    },
    {
      "epoch": 6.044754431851206,
      "grad_norm": 0.7714498043060303,
      "learning_rate": 5.027924412822279e-06,
      "loss": 0.012,
      "step": 20800
    },
    {
      "epoch": 6.073815751235106,
      "grad_norm": 0.008903966285288334,
      "learning_rate": 4.874913931604315e-06,
      "loss": 0.0096,
      "step": 20900
    },
    {
      "epoch": 6.102877070619006,
      "grad_norm": 0.07710820436477661,
      "learning_rate": 4.721903450386352e-06,
      "loss": 0.0081,
      "step": 21000
    },
    {
      "epoch": 6.102877070619006,
      "eval_loss": 0.027398647740483284,
      "eval_runtime": 27.3332,
      "eval_samples_per_second": 447.551,
      "eval_steps_per_second": 14.012,
      "step": 21000
    },
    {
      "epoch": 6.131938390002906,
      "grad_norm": 0.07280821353197098,
      "learning_rate": 4.568892969168388e-06,
      "loss": 0.007,
      "step": 21100
    },
    {
      "epoch": 6.1609997093868065,
      "grad_norm": 0.15252958238124847,
      "learning_rate": 4.415882487950424e-06,
      "loss": 0.0086,
      "step": 21200
    },
    {
      "epoch": 6.190061028770706,
      "grad_norm": 0.058942727744579315,
      "learning_rate": 4.262872006732461e-06,
      "loss": 0.0077,
      "step": 21300
    },
    {
      "epoch": 6.190061028770706,
      "eval_loss": 0.027552621439099312,
      "eval_runtime": 31.1781,
      "eval_samples_per_second": 392.358,
      "eval_steps_per_second": 12.284,
      "step": 21300
    },
    {
      "epoch": 6.219122348154606,
      "grad_norm": 0.10249049216508865,
      "learning_rate": 4.109861525514498e-06,
      "loss": 0.0096,
      "step": 21400
    },
    {
      "epoch": 6.248183667538506,
      "grad_norm": 0.0850997343659401,
      "learning_rate": 3.956851044296534e-06,
      "loss": 0.0071,
      "step": 21500
    },
    {
      "epoch": 6.2772449869224065,
      "grad_norm": 0.39275994896888733,
      "learning_rate": 3.8038405630785714e-06,
      "loss": 0.0084,
      "step": 21600
    },
    {
      "epoch": 6.2772449869224065,
      "eval_loss": 0.027445398271083832,
      "eval_runtime": 27.3905,
      "eval_samples_per_second": 446.615,
      "eval_steps_per_second": 13.983,
      "step": 21600
    },
    {
      "epoch": 6.306306306306306,
      "grad_norm": 0.09213391691446304,
      "learning_rate": 3.650830081860608e-06,
      "loss": 0.0105,
      "step": 21700
    },
    {
      "epoch": 6.335367625690206,
      "grad_norm": 0.012103458866477013,
      "learning_rate": 3.4993497054548235e-06,
      "loss": 0.0089,
      "step": 21800
    },
    {
      "epoch": 6.364428945074106,
      "grad_norm": 1.7496978044509888,
      "learning_rate": 3.3463392242368605e-06,
      "loss": 0.0112,
      "step": 21900
    },
    {
      "epoch": 6.364428945074106,
      "eval_loss": 0.027640603482723236,
      "eval_runtime": 31.8396,
      "eval_samples_per_second": 384.207,
      "eval_steps_per_second": 12.029,
      "step": 21900
    },
    {
      "epoch": 6.3934902644580065,
      "grad_norm": 0.04138137400150299,
      "learning_rate": 3.193328743018897e-06,
      "loss": 0.0095,
      "step": 22000
    },
    {
      "epoch": 6.422551583841907,
      "grad_norm": 0.08476943522691727,
      "learning_rate": 3.0403182618009336e-06,
      "loss": 0.0081,
      "step": 22100
    },
    {
      "epoch": 6.451612903225806,
      "grad_norm": 2.349595785140991,
      "learning_rate": 2.88730778058297e-06,
      "loss": 0.0085,
      "step": 22200
    },
    {
      "epoch": 6.451612903225806,
      "eval_loss": 0.027077894657850266,
      "eval_runtime": 28.0475,
      "eval_samples_per_second": 436.153,
      "eval_steps_per_second": 13.655,
      "step": 22200
    },
    {
      "epoch": 6.480674222609706,
      "grad_norm": 1.728858232498169,
      "learning_rate": 2.7342972993650067e-06,
      "loss": 0.0113,
      "step": 22300
    },
    {
      "epoch": 6.5097355419936065,
      "grad_norm": 0.10394562035799026,
      "learning_rate": 2.5812868181470436e-06,
      "loss": 0.0088,
      "step": 22400
    },
    {
      "epoch": 6.538796861377507,
      "grad_norm": 0.07329772412776947,
      "learning_rate": 2.4282763369290798e-06,
      "loss": 0.0094,
      "step": 22500
    },
    {
      "epoch": 6.538796861377507,
      "eval_loss": 0.02670624852180481,
      "eval_runtime": 31.4268,
      "eval_samples_per_second": 389.254,
      "eval_steps_per_second": 12.187,
      "step": 22500
    },
    {
      "epoch": 6.567858180761407,
      "grad_norm": 0.09306680411100388,
      "learning_rate": 2.2752658557111167e-06,
      "loss": 0.0073,
      "step": 22600
    },
    {
      "epoch": 6.596919500145306,
      "grad_norm": 0.061711717396974564,
      "learning_rate": 2.122255374493153e-06,
      "loss": 0.0075,
      "step": 22700
    },
    {
      "epoch": 6.6259808195292065,
      "grad_norm": 1.4829280376434326,
      "learning_rate": 1.9692448932751894e-06,
      "loss": 0.0078,
      "step": 22800
    },
    {
      "epoch": 6.6259808195292065,
      "eval_loss": 0.02658041939139366,
      "eval_runtime": 27.5526,
      "eval_samples_per_second": 443.986,
      "eval_steps_per_second": 13.901,
      "step": 22800
    },
    {
      "epoch": 6.655042138913107,
      "grad_norm": 0.017336523160338402,
      "learning_rate": 1.8162344120572261e-06,
      "loss": 0.0108,
      "step": 22900
    },
    {
      "epoch": 6.684103458297007,
      "grad_norm": 0.16300371289253235,
      "learning_rate": 1.6632239308392625e-06,
      "loss": 0.0125,
      "step": 23000
    },
    {
      "epoch": 6.713164777680907,
      "grad_norm": 0.4278203845024109,
      "learning_rate": 1.5102134496212992e-06,
      "loss": 0.0099,
      "step": 23100
    },
    {
      "epoch": 6.713164777680907,
      "eval_loss": 0.02630489505827427,
      "eval_runtime": 31.2321,
      "eval_samples_per_second": 391.681,
      "eval_steps_per_second": 12.263,
      "step": 23100
    },
    {
      "epoch": 6.7422260970648065,
      "grad_norm": 0.5904819965362549,
      "learning_rate": 1.3572029684033358e-06,
      "loss": 0.0087,
      "step": 23200
    },
    {
      "epoch": 6.771287416448707,
      "grad_norm": 1.0529608726501465,
      "learning_rate": 1.2041924871853723e-06,
      "loss": 0.0078,
      "step": 23300
    },
    {
      "epoch": 6.800348735832607,
      "grad_norm": 0.13827526569366455,
      "learning_rate": 1.0511820059674089e-06,
      "loss": 0.0113,
      "step": 23400
    },
    {
      "epoch": 6.800348735832607,
      "eval_loss": 0.026266321539878845,
      "eval_runtime": 27.2864,
      "eval_samples_per_second": 448.319,
      "eval_steps_per_second": 14.036,
      "step": 23400
    },
    {
      "epoch": 6.829410055216507,
      "grad_norm": 0.8667532801628113,
      "learning_rate": 8.981715247494455e-07,
      "loss": 0.0097,
      "step": 23500
    },
    {
      "epoch": 6.858471374600407,
      "grad_norm": 0.045186400413513184,
      "learning_rate": 7.45161043531482e-07,
      "loss": 0.0066,
      "step": 23600
    },
    {
      "epoch": 6.887532693984307,
      "grad_norm": 0.05325442925095558,
      "learning_rate": 5.921505623135185e-07,
      "loss": 0.0053,
      "step": 23700
    },
    {
      "epoch": 6.887532693984307,
      "eval_loss": 0.02622653916478157,
      "eval_runtime": 30.4773,
      "eval_samples_per_second": 401.381,
      "eval_steps_per_second": 12.567,
      "step": 23700
    },
    {
      "epoch": 6.916594013368207,
      "grad_norm": 0.22071638703346252,
      "learning_rate": 4.3914008109555514e-07,
      "loss": 0.0095,
      "step": 23800
    },
    {
      "epoch": 6.945655332752107,
      "grad_norm": 0.026897892355918884,
      "learning_rate": 2.8612959987759163e-07,
      "loss": 0.0067,
      "step": 23900
    },
    {
      "epoch": 6.974716652136007,
      "grad_norm": 1.3487067222595215,
      "learning_rate": 1.331191186596282e-07,
      "loss": 0.01,
      "step": 24000
    },
    {
      "epoch": 6.974716652136007,
      "eval_loss": 0.026174582540988922,
      "eval_runtime": 27.261,
      "eval_samples_per_second": 448.736,
      "eval_steps_per_second": 14.049,
      "step": 24000
    }
  ],
  "logging_steps": 100,
  "max_steps": 24087,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 7,
  "save_steps": 300,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}