project-code-py-small / trainer_state.json
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 5.0,
"global_step": 60120,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.04,
"learning_rate": 4.958416500332668e-05,
"loss": 2.6853,
"step": 500
},
{
"epoch": 0.08,
"learning_rate": 4.916833000665336e-05,
"loss": 2.3888,
"step": 1000
},
{
"epoch": 0.12,
"learning_rate": 4.8752495009980045e-05,
"loss": 2.4518,
"step": 1500
},
{
"epoch": 0.17,
"learning_rate": 4.833666001330672e-05,
"loss": 2.4134,
"step": 2000
},
{
"epoch": 0.21,
"learning_rate": 4.79208250166334e-05,
"loss": 2.2574,
"step": 2500
},
{
"epoch": 0.25,
"learning_rate": 4.750499001996008e-05,
"loss": 2.1327,
"step": 3000
},
{
"epoch": 0.29,
"learning_rate": 4.7089155023286765e-05,
"loss": 2.2362,
"step": 3500
},
{
"epoch": 0.33,
"learning_rate": 4.6673320026613444e-05,
"loss": 2.1094,
"step": 4000
},
{
"epoch": 0.37,
"learning_rate": 4.625748502994012e-05,
"loss": 2.1622,
"step": 4500
},
{
"epoch": 0.42,
"learning_rate": 4.58416500332668e-05,
"loss": 2.0353,
"step": 5000
},
{
"epoch": 0.46,
"learning_rate": 4.5425815036593486e-05,
"loss": 2.0771,
"step": 5500
},
{
"epoch": 0.5,
"learning_rate": 4.5009980039920164e-05,
"loss": 2.1499,
"step": 6000
},
{
"epoch": 0.54,
"learning_rate": 4.459414504324684e-05,
"loss": 2.1235,
"step": 6500
},
{
"epoch": 0.58,
"learning_rate": 4.417831004657352e-05,
"loss": 2.061,
"step": 7000
},
{
"epoch": 0.62,
"learning_rate": 4.3762475049900207e-05,
"loss": 2.066,
"step": 7500
},
{
"epoch": 0.67,
"learning_rate": 4.3346640053226885e-05,
"loss": 2.055,
"step": 8000
},
{
"epoch": 0.71,
"learning_rate": 4.2930805056553564e-05,
"loss": 2.012,
"step": 8500
},
{
"epoch": 0.75,
"learning_rate": 4.251497005988024e-05,
"loss": 2.0656,
"step": 9000
},
{
"epoch": 0.79,
"learning_rate": 4.209913506320692e-05,
"loss": 2.1063,
"step": 9500
},
{
"epoch": 0.83,
"learning_rate": 4.1683300066533606e-05,
"loss": 2.007,
"step": 10000
},
{
"epoch": 0.87,
"learning_rate": 4.1267465069860284e-05,
"loss": 2.0566,
"step": 10500
},
{
"epoch": 0.91,
"learning_rate": 4.085163007318696e-05,
"loss": 2.0895,
"step": 11000
},
{
"epoch": 0.96,
"learning_rate": 4.043579507651364e-05,
"loss": 1.9675,
"step": 11500
},
{
"epoch": 1.0,
"learning_rate": 4.0019960079840326e-05,
"loss": 2.0837,
"step": 12000
},
{
"epoch": 1.04,
"learning_rate": 3.9604125083167e-05,
"loss": 1.964,
"step": 12500
},
{
"epoch": 1.08,
"learning_rate": 3.9188290086493677e-05,
"loss": 2.024,
"step": 13000
},
{
"epoch": 1.12,
"learning_rate": 3.877245508982036e-05,
"loss": 1.9672,
"step": 13500
},
{
"epoch": 1.16,
"learning_rate": 3.835662009314704e-05,
"loss": 1.9733,
"step": 14000
},
{
"epoch": 1.21,
"learning_rate": 3.794078509647372e-05,
"loss": 2.0202,
"step": 14500
},
{
"epoch": 1.25,
"learning_rate": 3.75249500998004e-05,
"loss": 1.9932,
"step": 15000
},
{
"epoch": 1.29,
"learning_rate": 3.7109115103127076e-05,
"loss": 1.956,
"step": 15500
},
{
"epoch": 1.33,
"learning_rate": 3.669328010645376e-05,
"loss": 1.9439,
"step": 16000
},
{
"epoch": 1.37,
"learning_rate": 3.627744510978044e-05,
"loss": 2.024,
"step": 16500
},
{
"epoch": 1.41,
"learning_rate": 3.586161011310712e-05,
"loss": 1.9913,
"step": 17000
},
{
"epoch": 1.46,
"learning_rate": 3.5445775116433796e-05,
"loss": 1.8802,
"step": 17500
},
{
"epoch": 1.5,
"learning_rate": 3.502994011976048e-05,
"loss": 1.9544,
"step": 18000
},
{
"epoch": 1.54,
"learning_rate": 3.461410512308716e-05,
"loss": 1.9787,
"step": 18500
},
{
"epoch": 1.58,
"learning_rate": 3.419827012641384e-05,
"loss": 1.81,
"step": 19000
},
{
"epoch": 1.62,
"learning_rate": 3.378243512974052e-05,
"loss": 1.8795,
"step": 19500
},
{
"epoch": 1.66,
"learning_rate": 3.33666001330672e-05,
"loss": 1.9943,
"step": 20000
},
{
"epoch": 1.7,
"learning_rate": 3.295076513639388e-05,
"loss": 1.9593,
"step": 20500
},
{
"epoch": 1.75,
"learning_rate": 3.253493013972056e-05,
"loss": 1.8693,
"step": 21000
},
{
"epoch": 1.79,
"learning_rate": 3.211909514304724e-05,
"loss": 1.8688,
"step": 21500
},
{
"epoch": 1.83,
"learning_rate": 3.170326014637392e-05,
"loss": 1.9299,
"step": 22000
},
{
"epoch": 1.87,
"learning_rate": 3.12874251497006e-05,
"loss": 1.9862,
"step": 22500
},
{
"epoch": 1.91,
"learning_rate": 3.087159015302728e-05,
"loss": 1.9493,
"step": 23000
},
{
"epoch": 1.95,
"learning_rate": 3.045575515635396e-05,
"loss": 1.8195,
"step": 23500
},
{
"epoch": 2.0,
"learning_rate": 3.003992015968064e-05,
"loss": 2.0447,
"step": 24000
},
{
"epoch": 2.04,
"learning_rate": 2.962408516300732e-05,
"loss": 1.9297,
"step": 24500
},
{
"epoch": 2.08,
"learning_rate": 2.9208250166334e-05,
"loss": 1.8736,
"step": 25000
},
{
"epoch": 2.12,
"learning_rate": 2.879241516966068e-05,
"loss": 1.7594,
"step": 25500
},
{
"epoch": 2.16,
"learning_rate": 2.837658017298736e-05,
"loss": 1.913,
"step": 26000
},
{
"epoch": 2.2,
"learning_rate": 2.796074517631404e-05,
"loss": 1.8812,
"step": 26500
},
{
"epoch": 2.25,
"learning_rate": 2.754491017964072e-05,
"loss": 1.7825,
"step": 27000
},
{
"epoch": 2.29,
"learning_rate": 2.71290751829674e-05,
"loss": 1.9509,
"step": 27500
},
{
"epoch": 2.33,
"learning_rate": 2.671324018629408e-05,
"loss": 1.8853,
"step": 28000
},
{
"epoch": 2.37,
"learning_rate": 2.629740518962076e-05,
"loss": 1.9285,
"step": 28500
},
{
"epoch": 2.41,
"learning_rate": 2.5881570192947442e-05,
"loss": 1.9212,
"step": 29000
},
{
"epoch": 2.45,
"learning_rate": 2.546573519627412e-05,
"loss": 1.8891,
"step": 29500
},
{
"epoch": 2.5,
"learning_rate": 2.5049900199600802e-05,
"loss": 1.9006,
"step": 30000
},
{
"epoch": 2.54,
"learning_rate": 2.4634065202927477e-05,
"loss": 1.9119,
"step": 30500
},
{
"epoch": 2.58,
"learning_rate": 2.421823020625416e-05,
"loss": 1.8479,
"step": 31000
},
{
"epoch": 2.62,
"learning_rate": 2.3802395209580838e-05,
"loss": 1.9032,
"step": 31500
},
{
"epoch": 2.66,
"learning_rate": 2.338656021290752e-05,
"loss": 1.9279,
"step": 32000
},
{
"epoch": 2.7,
"learning_rate": 2.2970725216234198e-05,
"loss": 1.8265,
"step": 32500
},
{
"epoch": 2.74,
"learning_rate": 2.255489021956088e-05,
"loss": 1.8167,
"step": 33000
},
{
"epoch": 2.79,
"learning_rate": 2.2139055222887558e-05,
"loss": 1.9173,
"step": 33500
},
{
"epoch": 2.83,
"learning_rate": 2.172322022621424e-05,
"loss": 1.7094,
"step": 34000
},
{
"epoch": 2.87,
"learning_rate": 2.130738522954092e-05,
"loss": 1.8966,
"step": 34500
},
{
"epoch": 2.91,
"learning_rate": 2.08915502328676e-05,
"loss": 1.9069,
"step": 35000
},
{
"epoch": 2.95,
"learning_rate": 2.047571523619428e-05,
"loss": 1.8751,
"step": 35500
},
{
"epoch": 2.99,
"learning_rate": 2.0059880239520957e-05,
"loss": 1.8776,
"step": 36000
},
{
"epoch": 3.04,
"learning_rate": 1.964404524284764e-05,
"loss": 1.907,
"step": 36500
},
{
"epoch": 3.08,
"learning_rate": 1.9228210246174318e-05,
"loss": 1.8639,
"step": 37000
},
{
"epoch": 3.12,
"learning_rate": 1.8812375249501e-05,
"loss": 1.8729,
"step": 37500
},
{
"epoch": 3.16,
"learning_rate": 1.8396540252827678e-05,
"loss": 1.8241,
"step": 38000
},
{
"epoch": 3.2,
"learning_rate": 1.798070525615436e-05,
"loss": 1.7862,
"step": 38500
},
{
"epoch": 3.24,
"learning_rate": 1.756487025948104e-05,
"loss": 1.8483,
"step": 39000
},
{
"epoch": 3.29,
"learning_rate": 1.714903526280772e-05,
"loss": 1.9772,
"step": 39500
},
{
"epoch": 3.33,
"learning_rate": 1.67332002661344e-05,
"loss": 1.825,
"step": 40000
},
{
"epoch": 3.37,
"learning_rate": 1.631736526946108e-05,
"loss": 1.9203,
"step": 40500
},
{
"epoch": 3.41,
"learning_rate": 1.590153027278776e-05,
"loss": 1.8294,
"step": 41000
},
{
"epoch": 3.45,
"learning_rate": 1.548569527611444e-05,
"loss": 1.7971,
"step": 41500
},
{
"epoch": 3.49,
"learning_rate": 1.506986027944112e-05,
"loss": 1.7856,
"step": 42000
},
{
"epoch": 3.53,
"learning_rate": 1.4654025282767798e-05,
"loss": 1.834,
"step": 42500
},
{
"epoch": 3.58,
"learning_rate": 1.4238190286094476e-05,
"loss": 1.7484,
"step": 43000
},
{
"epoch": 3.62,
"learning_rate": 1.3822355289421157e-05,
"loss": 1.7906,
"step": 43500
},
{
"epoch": 3.66,
"learning_rate": 1.3406520292747837e-05,
"loss": 1.8818,
"step": 44000
},
{
"epoch": 3.7,
"learning_rate": 1.2990685296074517e-05,
"loss": 1.9182,
"step": 44500
},
{
"epoch": 3.74,
"learning_rate": 1.2574850299401197e-05,
"loss": 1.8451,
"step": 45000
},
{
"epoch": 3.78,
"learning_rate": 1.2159015302727877e-05,
"loss": 1.8165,
"step": 45500
},
{
"epoch": 3.83,
"learning_rate": 1.1743180306054557e-05,
"loss": 1.8486,
"step": 46000
},
{
"epoch": 3.87,
"learning_rate": 1.1327345309381238e-05,
"loss": 1.6533,
"step": 46500
},
{
"epoch": 3.91,
"learning_rate": 1.0911510312707918e-05,
"loss": 1.7231,
"step": 47000
},
{
"epoch": 3.95,
"learning_rate": 1.0495675316034598e-05,
"loss": 1.884,
"step": 47500
},
{
"epoch": 3.99,
"learning_rate": 1.0079840319361278e-05,
"loss": 1.7961,
"step": 48000
},
{
"epoch": 4.03,
"learning_rate": 9.664005322687958e-06,
"loss": 1.8359,
"step": 48500
},
{
"epoch": 4.08,
"learning_rate": 9.248170326014638e-06,
"loss": 1.7858,
"step": 49000
},
{
"epoch": 4.12,
"learning_rate": 8.832335329341319e-06,
"loss": 1.8043,
"step": 49500
},
{
"epoch": 4.16,
"learning_rate": 8.416500332667999e-06,
"loss": 1.9039,
"step": 50000
},
{
"epoch": 4.2,
"learning_rate": 8.000665335994679e-06,
"loss": 1.7695,
"step": 50500
},
{
"epoch": 4.24,
"learning_rate": 7.584830339321358e-06,
"loss": 1.8137,
"step": 51000
},
{
"epoch": 4.28,
"learning_rate": 7.168995342648038e-06,
"loss": 1.7972,
"step": 51500
},
{
"epoch": 4.32,
"learning_rate": 6.753160345974717e-06,
"loss": 1.8118,
"step": 52000
},
{
"epoch": 4.37,
"learning_rate": 6.337325349301397e-06,
"loss": 1.8511,
"step": 52500
},
{
"epoch": 4.41,
"learning_rate": 5.921490352628078e-06,
"loss": 1.8975,
"step": 53000
},
{
"epoch": 4.45,
"learning_rate": 5.505655355954757e-06,
"loss": 1.7506,
"step": 53500
},
{
"epoch": 4.49,
"learning_rate": 5.0898203592814375e-06,
"loss": 1.7564,
"step": 54000
},
{
"epoch": 4.53,
"learning_rate": 4.673985362608118e-06,
"loss": 1.9306,
"step": 54500
},
{
"epoch": 4.57,
"learning_rate": 4.258150365934797e-06,
"loss": 1.7743,
"step": 55000
},
{
"epoch": 4.62,
"learning_rate": 3.842315369261477e-06,
"loss": 1.8154,
"step": 55500
},
{
"epoch": 4.66,
"learning_rate": 3.4264803725881573e-06,
"loss": 1.7763,
"step": 56000
},
{
"epoch": 4.7,
"learning_rate": 3.010645375914837e-06,
"loss": 1.7074,
"step": 56500
},
{
"epoch": 4.74,
"learning_rate": 2.594810379241517e-06,
"loss": 1.8081,
"step": 57000
},
{
"epoch": 4.78,
"learning_rate": 2.178975382568197e-06,
"loss": 1.7563,
"step": 57500
},
{
"epoch": 4.82,
"learning_rate": 1.7631403858948771e-06,
"loss": 1.818,
"step": 58000
},
{
"epoch": 4.87,
"learning_rate": 1.3473053892215569e-06,
"loss": 1.7706,
"step": 58500
},
{
"epoch": 4.91,
"learning_rate": 9.314703925482368e-07,
"loss": 1.8389,
"step": 59000
},
{
"epoch": 4.95,
"learning_rate": 5.156353958749169e-07,
"loss": 1.8149,
"step": 59500
},
{
"epoch": 4.99,
"learning_rate": 9.98003992015968e-08,
"loss": 1.7843,
"step": 60000
},
{
"epoch": 5.0,
"step": 60120,
"total_flos": 3.025664452067328e+16,
"train_runtime": 9052.4232,
"train_samples_per_second": 6.641
}
],
"max_steps": 60120,
"num_train_epochs": 5,
"total_flos": 3.025664452067328e+16,
"trial_name": null,
"trial_params": null
}