Baby-CoThought / trainer_state.json
yaanhaan's picture
Upload 9 files
0254c20
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 4.986592651832337,
"global_step": 106000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 4.999952956673096e-05,
"loss": 11.0435,
"step": 1
},
{
"epoch": 0.02,
"learning_rate": 4.9764783365479614e-05,
"loss": 6.9955,
"step": 500
},
{
"epoch": 0.05,
"learning_rate": 4.952956673095921e-05,
"loss": 6.0131,
"step": 1000
},
{
"epoch": 0.07,
"learning_rate": 4.9294350096438824e-05,
"loss": 5.5677,
"step": 1500
},
{
"epoch": 0.09,
"learning_rate": 4.905913346191843e-05,
"loss": 5.2936,
"step": 2000
},
{
"epoch": 0.12,
"learning_rate": 4.8823916827398035e-05,
"loss": 5.0561,
"step": 2500
},
{
"epoch": 0.14,
"learning_rate": 4.8588700192877647e-05,
"loss": 4.8787,
"step": 3000
},
{
"epoch": 0.16,
"learning_rate": 4.835348355835725e-05,
"loss": 4.7244,
"step": 3500
},
{
"epoch": 0.19,
"learning_rate": 4.811826692383686e-05,
"loss": 4.5982,
"step": 4000
},
{
"epoch": 0.21,
"learning_rate": 4.788305028931646e-05,
"loss": 4.4815,
"step": 4500
},
{
"epoch": 0.24,
"learning_rate": 4.764783365479607e-05,
"loss": 4.4,
"step": 5000
},
{
"epoch": 0.26,
"learning_rate": 4.741261702027567e-05,
"loss": 4.3032,
"step": 5500
},
{
"epoch": 0.28,
"learning_rate": 4.7177400385755284e-05,
"loss": 4.2385,
"step": 6000
},
{
"epoch": 0.31,
"learning_rate": 4.694218375123489e-05,
"loss": 4.1932,
"step": 6500
},
{
"epoch": 0.33,
"learning_rate": 4.6706967116714494e-05,
"loss": 4.1363,
"step": 7000
},
{
"epoch": 0.35,
"learning_rate": 4.6471750482194106e-05,
"loss": 4.0896,
"step": 7500
},
{
"epoch": 0.38,
"learning_rate": 4.6236533847673704e-05,
"loss": 4.0412,
"step": 8000
},
{
"epoch": 0.4,
"learning_rate": 4.6001317213153316e-05,
"loss": 3.9997,
"step": 8500
},
{
"epoch": 0.42,
"learning_rate": 4.576610057863293e-05,
"loss": 3.9699,
"step": 9000
},
{
"epoch": 0.45,
"learning_rate": 4.5530883944112527e-05,
"loss": 3.9251,
"step": 9500
},
{
"epoch": 0.47,
"learning_rate": 4.529566730959214e-05,
"loss": 3.9045,
"step": 10000
},
{
"epoch": 0.49,
"learning_rate": 4.5060450675071744e-05,
"loss": 3.8665,
"step": 10500
},
{
"epoch": 0.52,
"learning_rate": 4.482523404055135e-05,
"loss": 3.836,
"step": 11000
},
{
"epoch": 0.54,
"learning_rate": 4.4590017406030954e-05,
"loss": 3.8186,
"step": 11500
},
{
"epoch": 0.56,
"learning_rate": 4.4354800771510566e-05,
"loss": 3.8041,
"step": 12000
},
{
"epoch": 0.59,
"learning_rate": 4.411958413699017e-05,
"loss": 3.7728,
"step": 12500
},
{
"epoch": 0.61,
"learning_rate": 4.3884367502469776e-05,
"loss": 3.7502,
"step": 13000
},
{
"epoch": 0.64,
"learning_rate": 4.364915086794938e-05,
"loss": 3.7264,
"step": 13500
},
{
"epoch": 0.66,
"learning_rate": 4.3413934233428986e-05,
"loss": 3.7029,
"step": 14000
},
{
"epoch": 0.68,
"learning_rate": 4.31787175989086e-05,
"loss": 3.6765,
"step": 14500
},
{
"epoch": 0.71,
"learning_rate": 4.29435009643882e-05,
"loss": 3.6528,
"step": 15000
},
{
"epoch": 0.73,
"learning_rate": 4.270828432986781e-05,
"loss": 3.6563,
"step": 15500
},
{
"epoch": 0.75,
"learning_rate": 4.247306769534742e-05,
"loss": 3.6245,
"step": 16000
},
{
"epoch": 0.78,
"learning_rate": 4.223785106082702e-05,
"loss": 3.6056,
"step": 16500
},
{
"epoch": 0.8,
"learning_rate": 4.200263442630663e-05,
"loss": 3.5829,
"step": 17000
},
{
"epoch": 0.82,
"learning_rate": 4.176741779178624e-05,
"loss": 3.5855,
"step": 17500
},
{
"epoch": 0.85,
"learning_rate": 4.153220115726584e-05,
"loss": 3.5703,
"step": 18000
},
{
"epoch": 0.87,
"learning_rate": 4.129698452274545e-05,
"loss": 3.5545,
"step": 18500
},
{
"epoch": 0.89,
"learning_rate": 4.106176788822506e-05,
"loss": 3.5338,
"step": 19000
},
{
"epoch": 0.92,
"learning_rate": 4.082655125370466e-05,
"loss": 3.5202,
"step": 19500
},
{
"epoch": 0.94,
"learning_rate": 4.059133461918427e-05,
"loss": 3.5066,
"step": 20000
},
{
"epoch": 0.96,
"learning_rate": 4.035611798466388e-05,
"loss": 3.5027,
"step": 20500
},
{
"epoch": 0.99,
"learning_rate": 4.0120901350143485e-05,
"loss": 3.477,
"step": 21000
},
{
"epoch": 1.01,
"learning_rate": 3.988568471562309e-05,
"loss": 3.4658,
"step": 21500
},
{
"epoch": 1.03,
"learning_rate": 3.96504680811027e-05,
"loss": 3.4505,
"step": 22000
},
{
"epoch": 1.06,
"learning_rate": 3.94152514465823e-05,
"loss": 3.445,
"step": 22500
},
{
"epoch": 1.08,
"learning_rate": 3.918003481206191e-05,
"loss": 3.4314,
"step": 23000
},
{
"epoch": 1.11,
"learning_rate": 3.894481817754152e-05,
"loss": 3.4278,
"step": 23500
},
{
"epoch": 1.13,
"learning_rate": 3.870960154302112e-05,
"loss": 3.4179,
"step": 24000
},
{
"epoch": 1.15,
"learning_rate": 3.8474384908500734e-05,
"loss": 3.4051,
"step": 24500
},
{
"epoch": 1.18,
"learning_rate": 3.823916827398034e-05,
"loss": 3.3858,
"step": 25000
},
{
"epoch": 1.2,
"learning_rate": 3.8003951639459944e-05,
"loss": 3.3851,
"step": 25500
},
{
"epoch": 1.22,
"learning_rate": 3.7768735004939556e-05,
"loss": 3.3592,
"step": 26000
},
{
"epoch": 1.25,
"learning_rate": 3.7533518370419154e-05,
"loss": 3.3617,
"step": 26500
},
{
"epoch": 1.27,
"learning_rate": 3.7298301735898766e-05,
"loss": 3.351,
"step": 27000
},
{
"epoch": 1.29,
"learning_rate": 3.706308510137837e-05,
"loss": 3.3456,
"step": 27500
},
{
"epoch": 1.32,
"learning_rate": 3.6827868466857977e-05,
"loss": 3.3412,
"step": 28000
},
{
"epoch": 1.34,
"learning_rate": 3.659265183233758e-05,
"loss": 3.3275,
"step": 28500
},
{
"epoch": 1.36,
"learning_rate": 3.6357435197817194e-05,
"loss": 3.3048,
"step": 29000
},
{
"epoch": 1.39,
"learning_rate": 3.61222185632968e-05,
"loss": 3.3148,
"step": 29500
},
{
"epoch": 1.41,
"learning_rate": 3.5887001928776404e-05,
"loss": 3.3077,
"step": 30000
},
{
"epoch": 1.43,
"learning_rate": 3.5651785294256016e-05,
"loss": 3.2981,
"step": 30500
},
{
"epoch": 1.46,
"learning_rate": 3.5416568659735614e-05,
"loss": 3.2814,
"step": 31000
},
{
"epoch": 1.48,
"learning_rate": 3.5181352025215226e-05,
"loss": 3.2877,
"step": 31500
},
{
"epoch": 1.51,
"learning_rate": 3.494613539069483e-05,
"loss": 3.2722,
"step": 32000
},
{
"epoch": 1.53,
"learning_rate": 3.4710918756174436e-05,
"loss": 3.269,
"step": 32500
},
{
"epoch": 1.55,
"learning_rate": 3.447570212165405e-05,
"loss": 3.2491,
"step": 33000
},
{
"epoch": 1.58,
"learning_rate": 3.424048548713365e-05,
"loss": 3.2531,
"step": 33500
},
{
"epoch": 1.6,
"learning_rate": 3.400526885261326e-05,
"loss": 3.2399,
"step": 34000
},
{
"epoch": 1.62,
"learning_rate": 3.377005221809286e-05,
"loss": 3.24,
"step": 34500
},
{
"epoch": 1.65,
"learning_rate": 3.353483558357247e-05,
"loss": 3.2312,
"step": 35000
},
{
"epoch": 1.67,
"learning_rate": 3.329961894905208e-05,
"loss": 3.2293,
"step": 35500
},
{
"epoch": 1.69,
"learning_rate": 3.3064402314531685e-05,
"loss": 3.2214,
"step": 36000
},
{
"epoch": 1.72,
"learning_rate": 3.282918568001129e-05,
"loss": 3.2154,
"step": 36500
},
{
"epoch": 1.74,
"learning_rate": 3.2593969045490896e-05,
"loss": 3.2198,
"step": 37000
},
{
"epoch": 1.76,
"learning_rate": 3.235875241097051e-05,
"loss": 3.1968,
"step": 37500
},
{
"epoch": 1.79,
"learning_rate": 3.212353577645011e-05,
"loss": 3.1938,
"step": 38000
},
{
"epoch": 1.81,
"learning_rate": 3.188831914192972e-05,
"loss": 3.1861,
"step": 38500
},
{
"epoch": 1.83,
"learning_rate": 3.165310250740933e-05,
"loss": 3.1805,
"step": 39000
},
{
"epoch": 1.86,
"learning_rate": 3.141788587288893e-05,
"loss": 3.1778,
"step": 39500
},
{
"epoch": 1.88,
"learning_rate": 3.118266923836854e-05,
"loss": 3.1748,
"step": 40000
},
{
"epoch": 1.91,
"learning_rate": 3.0947452603848145e-05,
"loss": 3.1688,
"step": 40500
},
{
"epoch": 1.93,
"learning_rate": 3.071223596932775e-05,
"loss": 3.1617,
"step": 41000
},
{
"epoch": 1.95,
"learning_rate": 3.0477019334807362e-05,
"loss": 3.1537,
"step": 41500
},
{
"epoch": 1.98,
"learning_rate": 3.0241802700286964e-05,
"loss": 3.1581,
"step": 42000
},
{
"epoch": 2.0,
"learning_rate": 3.0006586065766572e-05,
"loss": 3.149,
"step": 42500
},
{
"epoch": 2.02,
"learning_rate": 2.9771369431246177e-05,
"loss": 3.1317,
"step": 43000
},
{
"epoch": 2.05,
"learning_rate": 2.9536152796725786e-05,
"loss": 3.1136,
"step": 43500
},
{
"epoch": 2.07,
"learning_rate": 2.9300936162205394e-05,
"loss": 3.1115,
"step": 44000
},
{
"epoch": 2.09,
"learning_rate": 2.9065719527685e-05,
"loss": 3.1165,
"step": 44500
},
{
"epoch": 2.12,
"learning_rate": 2.8830502893164608e-05,
"loss": 3.0959,
"step": 45000
},
{
"epoch": 2.14,
"learning_rate": 2.859528625864421e-05,
"loss": 3.1135,
"step": 45500
},
{
"epoch": 2.16,
"learning_rate": 2.8360069624123818e-05,
"loss": 3.0856,
"step": 46000
},
{
"epoch": 2.19,
"learning_rate": 2.812485298960343e-05,
"loss": 3.1023,
"step": 46500
},
{
"epoch": 2.21,
"learning_rate": 2.7889636355083032e-05,
"loss": 3.0848,
"step": 47000
},
{
"epoch": 2.23,
"learning_rate": 2.765441972056264e-05,
"loss": 3.0901,
"step": 47500
},
{
"epoch": 2.26,
"learning_rate": 2.7419203086042245e-05,
"loss": 3.0776,
"step": 48000
},
{
"epoch": 2.28,
"learning_rate": 2.7183986451521854e-05,
"loss": 3.0703,
"step": 48500
},
{
"epoch": 2.31,
"learning_rate": 2.6948769817001456e-05,
"loss": 3.0671,
"step": 49000
},
{
"epoch": 2.33,
"learning_rate": 2.6713553182481067e-05,
"loss": 3.0574,
"step": 49500
},
{
"epoch": 2.35,
"learning_rate": 2.6478336547960676e-05,
"loss": 3.0583,
"step": 50000
},
{
"epoch": 2.38,
"learning_rate": 2.6243119913440278e-05,
"loss": 3.0599,
"step": 50500
},
{
"epoch": 2.4,
"learning_rate": 2.6007903278919886e-05,
"loss": 3.0596,
"step": 51000
},
{
"epoch": 2.42,
"learning_rate": 2.577268664439949e-05,
"loss": 3.0522,
"step": 51500
},
{
"epoch": 2.45,
"learning_rate": 2.55374700098791e-05,
"loss": 3.0484,
"step": 52000
},
{
"epoch": 2.47,
"learning_rate": 2.5302253375358708e-05,
"loss": 3.0332,
"step": 52500
},
{
"epoch": 2.49,
"learning_rate": 2.5067036740838313e-05,
"loss": 3.0302,
"step": 53000
},
{
"epoch": 2.52,
"learning_rate": 2.4831820106317922e-05,
"loss": 3.0321,
"step": 53500
},
{
"epoch": 2.54,
"learning_rate": 2.4596603471797527e-05,
"loss": 3.0304,
"step": 54000
},
{
"epoch": 2.56,
"learning_rate": 2.4361386837277132e-05,
"loss": 3.0296,
"step": 54500
},
{
"epoch": 2.59,
"learning_rate": 2.412617020275674e-05,
"loss": 3.0164,
"step": 55000
},
{
"epoch": 2.61,
"learning_rate": 2.3890953568236346e-05,
"loss": 3.0118,
"step": 55500
},
{
"epoch": 2.63,
"learning_rate": 2.365573693371595e-05,
"loss": 3.0146,
"step": 56000
},
{
"epoch": 2.66,
"learning_rate": 2.3420520299195563e-05,
"loss": 3.0152,
"step": 56500
},
{
"epoch": 2.68,
"learning_rate": 2.3185303664675168e-05,
"loss": 3.0084,
"step": 57000
},
{
"epoch": 2.7,
"learning_rate": 2.2950087030154773e-05,
"loss": 2.9993,
"step": 57500
},
{
"epoch": 2.73,
"learning_rate": 2.271487039563438e-05,
"loss": 3.0059,
"step": 58000
},
{
"epoch": 2.75,
"learning_rate": 2.2479653761113986e-05,
"loss": 2.998,
"step": 58500
},
{
"epoch": 2.78,
"learning_rate": 2.224443712659359e-05,
"loss": 2.9874,
"step": 59000
},
{
"epoch": 2.8,
"learning_rate": 2.20092204920732e-05,
"loss": 2.9934,
"step": 59500
},
{
"epoch": 2.82,
"learning_rate": 2.177400385755281e-05,
"loss": 2.9885,
"step": 60000
},
{
"epoch": 2.85,
"learning_rate": 2.1538787223032414e-05,
"loss": 2.9806,
"step": 60500
},
{
"epoch": 2.87,
"learning_rate": 2.130357058851202e-05,
"loss": 2.9801,
"step": 61000
},
{
"epoch": 2.89,
"learning_rate": 2.1068353953991627e-05,
"loss": 2.9726,
"step": 61500
},
{
"epoch": 2.92,
"learning_rate": 2.0833137319471236e-05,
"loss": 2.975,
"step": 62000
},
{
"epoch": 2.94,
"learning_rate": 2.059792068495084e-05,
"loss": 2.9697,
"step": 62500
},
{
"epoch": 2.96,
"learning_rate": 2.036270405043045e-05,
"loss": 2.9604,
"step": 63000
},
{
"epoch": 2.99,
"learning_rate": 2.0127487415910055e-05,
"loss": 2.955,
"step": 63500
},
{
"epoch": 3.01,
"learning_rate": 1.989227078138966e-05,
"loss": 2.9571,
"step": 64000
},
{
"epoch": 3.03,
"learning_rate": 1.9657054146869268e-05,
"loss": 2.9388,
"step": 64500
},
{
"epoch": 3.06,
"learning_rate": 1.9421837512348877e-05,
"loss": 2.9447,
"step": 65000
},
{
"epoch": 3.08,
"learning_rate": 1.9186620877828482e-05,
"loss": 2.9359,
"step": 65500
},
{
"epoch": 3.1,
"learning_rate": 1.8951404243308087e-05,
"loss": 2.9429,
"step": 66000
},
{
"epoch": 3.13,
"learning_rate": 1.8716187608787695e-05,
"loss": 2.9315,
"step": 66500
},
{
"epoch": 3.15,
"learning_rate": 1.84809709742673e-05,
"loss": 2.9258,
"step": 67000
},
{
"epoch": 3.18,
"learning_rate": 1.8245754339746906e-05,
"loss": 2.9285,
"step": 67500
},
{
"epoch": 3.2,
"learning_rate": 1.8010537705226514e-05,
"loss": 2.9277,
"step": 68000
},
{
"epoch": 3.22,
"learning_rate": 1.7775321070706123e-05,
"loss": 2.9209,
"step": 68500
},
{
"epoch": 3.25,
"learning_rate": 1.7540104436185728e-05,
"loss": 2.925,
"step": 69000
},
{
"epoch": 3.27,
"learning_rate": 1.7304887801665333e-05,
"loss": 2.9119,
"step": 69500
},
{
"epoch": 3.29,
"learning_rate": 1.706967116714494e-05,
"loss": 2.9145,
"step": 70000
},
{
"epoch": 3.32,
"learning_rate": 1.6834454532624546e-05,
"loss": 2.9079,
"step": 70500
},
{
"epoch": 3.34,
"learning_rate": 1.6599237898104155e-05,
"loss": 2.9089,
"step": 71000
},
{
"epoch": 3.36,
"learning_rate": 1.6364021263583763e-05,
"loss": 2.9052,
"step": 71500
},
{
"epoch": 3.39,
"learning_rate": 1.612880462906337e-05,
"loss": 2.897,
"step": 72000
},
{
"epoch": 3.41,
"learning_rate": 1.5893587994542974e-05,
"loss": 2.9026,
"step": 72500
},
{
"epoch": 3.43,
"learning_rate": 1.5658371360022582e-05,
"loss": 2.9071,
"step": 73000
},
{
"epoch": 3.46,
"learning_rate": 1.5423154725502187e-05,
"loss": 2.9021,
"step": 73500
},
{
"epoch": 3.48,
"learning_rate": 1.5187938090981796e-05,
"loss": 2.8918,
"step": 74000
},
{
"epoch": 3.5,
"learning_rate": 1.4952721456461402e-05,
"loss": 2.8995,
"step": 74500
},
{
"epoch": 3.53,
"learning_rate": 1.4717504821941008e-05,
"loss": 2.8874,
"step": 75000
},
{
"epoch": 3.55,
"learning_rate": 1.4482288187420614e-05,
"loss": 2.8779,
"step": 75500
},
{
"epoch": 3.58,
"learning_rate": 1.4247071552900221e-05,
"loss": 2.8961,
"step": 76000
},
{
"epoch": 3.6,
"learning_rate": 1.401185491837983e-05,
"loss": 2.8852,
"step": 76500
},
{
"epoch": 3.62,
"learning_rate": 1.3776638283859436e-05,
"loss": 2.8786,
"step": 77000
},
{
"epoch": 3.65,
"learning_rate": 1.3541421649339042e-05,
"loss": 2.8713,
"step": 77500
},
{
"epoch": 3.67,
"learning_rate": 1.3306205014818648e-05,
"loss": 2.8909,
"step": 78000
},
{
"epoch": 3.69,
"learning_rate": 1.3070988380298255e-05,
"loss": 2.8746,
"step": 78500
},
{
"epoch": 3.72,
"learning_rate": 1.283577174577786e-05,
"loss": 2.8682,
"step": 79000
},
{
"epoch": 3.74,
"learning_rate": 1.260055511125747e-05,
"loss": 2.871,
"step": 79500
},
{
"epoch": 3.76,
"learning_rate": 1.2365338476737076e-05,
"loss": 2.8652,
"step": 80000
},
{
"epoch": 3.79,
"learning_rate": 1.2130121842216682e-05,
"loss": 2.8685,
"step": 80500
},
{
"epoch": 3.81,
"learning_rate": 1.189490520769629e-05,
"loss": 2.859,
"step": 81000
},
{
"epoch": 3.83,
"learning_rate": 1.1659688573175896e-05,
"loss": 2.8699,
"step": 81500
},
{
"epoch": 3.86,
"learning_rate": 1.1424471938655503e-05,
"loss": 2.8599,
"step": 82000
},
{
"epoch": 3.88,
"learning_rate": 1.1189255304135108e-05,
"loss": 2.8636,
"step": 82500
},
{
"epoch": 3.9,
"learning_rate": 1.0954038669614716e-05,
"loss": 2.8681,
"step": 83000
},
{
"epoch": 3.93,
"learning_rate": 1.0718822035094323e-05,
"loss": 2.8511,
"step": 83500
},
{
"epoch": 3.95,
"learning_rate": 1.0483605400573928e-05,
"loss": 2.8546,
"step": 84000
},
{
"epoch": 3.98,
"learning_rate": 1.0248388766053537e-05,
"loss": 2.8476,
"step": 84500
},
{
"epoch": 4.0,
"learning_rate": 1.0013172131533142e-05,
"loss": 2.8516,
"step": 85000
},
{
"epoch": 4.02,
"learning_rate": 9.777955497012749e-06,
"loss": 2.8439,
"step": 85500
},
{
"epoch": 4.05,
"learning_rate": 9.542738862492356e-06,
"loss": 2.8427,
"step": 86000
},
{
"epoch": 4.07,
"learning_rate": 9.307522227971962e-06,
"loss": 2.8392,
"step": 86500
},
{
"epoch": 4.09,
"learning_rate": 9.07230559345157e-06,
"loss": 2.8381,
"step": 87000
},
{
"epoch": 4.12,
"learning_rate": 8.837088958931176e-06,
"loss": 2.8347,
"step": 87500
},
{
"epoch": 4.14,
"learning_rate": 8.601872324410783e-06,
"loss": 2.8354,
"step": 88000
},
{
"epoch": 4.16,
"learning_rate": 8.36665568989039e-06,
"loss": 2.8351,
"step": 88500
},
{
"epoch": 4.19,
"learning_rate": 8.131439055369996e-06,
"loss": 2.8283,
"step": 89000
},
{
"epoch": 4.21,
"learning_rate": 7.896222420849603e-06,
"loss": 2.8277,
"step": 89500
},
{
"epoch": 4.23,
"learning_rate": 7.661005786329208e-06,
"loss": 2.83,
"step": 90000
},
{
"epoch": 4.26,
"learning_rate": 7.425789151808817e-06,
"loss": 2.8247,
"step": 90500
},
{
"epoch": 4.28,
"learning_rate": 7.190572517288423e-06,
"loss": 2.8238,
"step": 91000
},
{
"epoch": 4.3,
"learning_rate": 6.955355882768029e-06,
"loss": 2.8292,
"step": 91500
},
{
"epoch": 4.33,
"learning_rate": 6.720139248247636e-06,
"loss": 2.8368,
"step": 92000
},
{
"epoch": 4.35,
"learning_rate": 6.484922613727243e-06,
"loss": 2.8247,
"step": 92500
},
{
"epoch": 4.38,
"learning_rate": 6.24970597920685e-06,
"loss": 2.8258,
"step": 93000
},
{
"epoch": 4.4,
"learning_rate": 6.014489344686457e-06,
"loss": 2.8128,
"step": 93500
},
{
"epoch": 4.42,
"learning_rate": 5.779272710166063e-06,
"loss": 2.82,
"step": 94000
},
{
"epoch": 4.45,
"learning_rate": 5.5440560756456695e-06,
"loss": 2.8101,
"step": 94500
},
{
"epoch": 4.47,
"learning_rate": 5.308839441125277e-06,
"loss": 2.8218,
"step": 95000
},
{
"epoch": 4.49,
"learning_rate": 5.073622806604884e-06,
"loss": 2.8205,
"step": 95500
},
{
"epoch": 4.52,
"learning_rate": 4.83840617208449e-06,
"loss": 2.8134,
"step": 96000
},
{
"epoch": 4.54,
"learning_rate": 4.603189537564097e-06,
"loss": 2.8081,
"step": 96500
},
{
"epoch": 4.56,
"learning_rate": 4.3679729030437035e-06,
"loss": 2.8082,
"step": 97000
},
{
"epoch": 4.59,
"learning_rate": 4.13275626852331e-06,
"loss": 2.8099,
"step": 97500
},
{
"epoch": 4.61,
"learning_rate": 3.897539634002917e-06,
"loss": 2.8096,
"step": 98000
},
{
"epoch": 4.63,
"learning_rate": 3.662322999482524e-06,
"loss": 2.8078,
"step": 98500
},
{
"epoch": 4.66,
"learning_rate": 3.42710636496213e-06,
"loss": 2.8151,
"step": 99000
},
{
"epoch": 4.68,
"learning_rate": 3.191889730441737e-06,
"loss": 2.795,
"step": 99500
},
{
"epoch": 4.7,
"learning_rate": 2.956673095921344e-06,
"loss": 2.8095,
"step": 100000
},
{
"epoch": 4.73,
"learning_rate": 2.7214564614009503e-06,
"loss": 2.8146,
"step": 100500
},
{
"epoch": 4.75,
"learning_rate": 2.486239826880557e-06,
"loss": 2.8022,
"step": 101000
},
{
"epoch": 4.77,
"learning_rate": 2.251023192360164e-06,
"loss": 2.8,
"step": 101500
},
{
"epoch": 4.8,
"learning_rate": 2.0158065578397707e-06,
"loss": 2.7974,
"step": 102000
},
{
"epoch": 4.82,
"learning_rate": 1.780589923319377e-06,
"loss": 2.8006,
"step": 102500
},
{
"epoch": 4.85,
"learning_rate": 1.545373288798984e-06,
"loss": 2.795,
"step": 103000
},
{
"epoch": 4.87,
"learning_rate": 1.3101566542785905e-06,
"loss": 2.8012,
"step": 103500
},
{
"epoch": 4.89,
"learning_rate": 1.0749400197581973e-06,
"loss": 2.8025,
"step": 104000
},
{
"epoch": 4.92,
"learning_rate": 8.397233852378041e-07,
"loss": 2.8024,
"step": 104500
},
{
"epoch": 4.94,
"learning_rate": 6.045067507174108e-07,
"loss": 2.7982,
"step": 105000
},
{
"epoch": 4.96,
"learning_rate": 3.6929011619701747e-07,
"loss": 2.7953,
"step": 105500
},
{
"epoch": 4.99,
"learning_rate": 1.340734816766242e-07,
"loss": 2.8075,
"step": 106000
}
],
"max_steps": 106285,
"num_train_epochs": 5,
"total_flos": 3.567911344090743e+18,
"trial_name": null,
"trial_params": null
}