mrm8488's picture
Initial commit from mrm8488
a78317c
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.37524950099800397,
"global_step": 47000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 4.9966733200266135e-05,
"loss": 1.8086,
"step": 500
},
{
"epoch": 0.01,
"learning_rate": 4.993346640053227e-05,
"loss": 2.1136,
"step": 1000
},
{
"epoch": 0.01,
"learning_rate": 4.99001996007984e-05,
"loss": 1.0008,
"step": 1500
},
{
"epoch": 0.02,
"learning_rate": 4.986693280106454e-05,
"loss": 0.8661,
"step": 2000
},
{
"epoch": 0.02,
"learning_rate": 4.983366600133067e-05,
"loss": 0.8234,
"step": 2500
},
{
"epoch": 0.02,
"learning_rate": 4.980039920159681e-05,
"loss": 0.811,
"step": 3000
},
{
"epoch": 0.03,
"learning_rate": 4.9767132401862945e-05,
"loss": 0.7938,
"step": 3500
},
{
"epoch": 0.03,
"learning_rate": 4.973386560212908e-05,
"loss": 0.7741,
"step": 4000
},
{
"epoch": 0.04,
"learning_rate": 4.970059880239521e-05,
"loss": 0.795,
"step": 4500
},
{
"epoch": 0.04,
"learning_rate": 4.966733200266135e-05,
"loss": 0.745,
"step": 5000
},
{
"epoch": 0.04,
"learning_rate": 4.963406520292748e-05,
"loss": 0.7542,
"step": 5500
},
{
"epoch": 0.05,
"learning_rate": 4.9600798403193615e-05,
"loss": 0.749,
"step": 6000
},
{
"epoch": 0.05,
"learning_rate": 4.956753160345975e-05,
"loss": 0.7411,
"step": 6500
},
{
"epoch": 0.06,
"learning_rate": 4.953426480372588e-05,
"loss": 0.737,
"step": 7000
},
{
"epoch": 0.06,
"learning_rate": 4.950099800399202e-05,
"loss": 0.7304,
"step": 7500
},
{
"epoch": 0.06,
"learning_rate": 4.946773120425816e-05,
"loss": 0.7227,
"step": 8000
},
{
"epoch": 0.07,
"learning_rate": 4.943446440452429e-05,
"loss": 0.7153,
"step": 8500
},
{
"epoch": 0.07,
"learning_rate": 4.9401197604790424e-05,
"loss": 0.7292,
"step": 9000
},
{
"epoch": 0.08,
"learning_rate": 4.936793080505656e-05,
"loss": 0.698,
"step": 9500
},
{
"epoch": 0.08,
"learning_rate": 4.933466400532269e-05,
"loss": 0.7148,
"step": 10000
},
{
"epoch": 0.08,
"learning_rate": 4.930139720558882e-05,
"loss": 0.6963,
"step": 10500
},
{
"epoch": 0.09,
"learning_rate": 4.9268130405854955e-05,
"loss": 0.7101,
"step": 11000
},
{
"epoch": 0.09,
"learning_rate": 4.9234863606121094e-05,
"loss": 0.6888,
"step": 11500
},
{
"epoch": 0.1,
"learning_rate": 4.920159680638723e-05,
"loss": 0.702,
"step": 12000
},
{
"epoch": 0.1,
"learning_rate": 4.916833000665336e-05,
"loss": 0.6865,
"step": 12500
},
{
"epoch": 0.1,
"learning_rate": 4.91350632069195e-05,
"loss": 0.6816,
"step": 13000
},
{
"epoch": 0.11,
"learning_rate": 4.910179640718563e-05,
"loss": 0.6926,
"step": 13500
},
{
"epoch": 0.11,
"learning_rate": 4.9068529607451764e-05,
"loss": 0.6917,
"step": 14000
},
{
"epoch": 0.12,
"learning_rate": 4.9035262807717903e-05,
"loss": 0.6714,
"step": 14500
},
{
"epoch": 0.12,
"learning_rate": 4.9001996007984036e-05,
"loss": 0.6793,
"step": 15000
},
{
"epoch": 0.12,
"learning_rate": 4.896872920825017e-05,
"loss": 0.6826,
"step": 15500
},
{
"epoch": 0.13,
"learning_rate": 4.89354624085163e-05,
"loss": 0.6739,
"step": 16000
},
{
"epoch": 0.13,
"learning_rate": 4.8902195608782434e-05,
"loss": 0.697,
"step": 16500
},
{
"epoch": 0.14,
"learning_rate": 4.886892880904857e-05,
"loss": 0.6696,
"step": 17000
},
{
"epoch": 0.14,
"learning_rate": 4.8835662009314706e-05,
"loss": 0.6911,
"step": 17500
},
{
"epoch": 0.14,
"learning_rate": 4.8802395209580846e-05,
"loss": 0.6642,
"step": 18000
},
{
"epoch": 0.15,
"learning_rate": 4.876912840984698e-05,
"loss": 0.6697,
"step": 18500
},
{
"epoch": 0.15,
"learning_rate": 4.873586161011311e-05,
"loss": 0.6635,
"step": 19000
},
{
"epoch": 0.16,
"learning_rate": 4.8702594810379244e-05,
"loss": 0.6824,
"step": 19500
},
{
"epoch": 0.16,
"learning_rate": 4.8669328010645376e-05,
"loss": 0.6601,
"step": 20000
},
{
"epoch": 0.16,
"learning_rate": 4.863606121091151e-05,
"loss": 0.6871,
"step": 20500
},
{
"epoch": 0.17,
"learning_rate": 4.860279441117765e-05,
"loss": 0.661,
"step": 21000
},
{
"epoch": 0.17,
"learning_rate": 4.856952761144378e-05,
"loss": 0.6658,
"step": 21500
},
{
"epoch": 0.18,
"learning_rate": 4.8536260811709914e-05,
"loss": 0.6607,
"step": 22000
},
{
"epoch": 0.18,
"learning_rate": 4.8502994011976046e-05,
"loss": 0.6447,
"step": 22500
},
{
"epoch": 0.18,
"learning_rate": 4.8469727212242186e-05,
"loss": 0.6603,
"step": 23000
},
{
"epoch": 0.19,
"learning_rate": 4.843646041250832e-05,
"loss": 0.6559,
"step": 23500
},
{
"epoch": 0.19,
"learning_rate": 4.840319361277446e-05,
"loss": 0.6563,
"step": 24000
},
{
"epoch": 0.2,
"learning_rate": 4.836992681304059e-05,
"loss": 0.6681,
"step": 24500
},
{
"epoch": 0.2,
"learning_rate": 4.833666001330672e-05,
"loss": 0.6412,
"step": 25000
},
{
"epoch": 0.2,
"learning_rate": 4.8303393213572856e-05,
"loss": 0.6553,
"step": 25500
},
{
"epoch": 0.21,
"learning_rate": 4.827012641383899e-05,
"loss": 0.655,
"step": 26000
},
{
"epoch": 0.21,
"learning_rate": 4.823685961410512e-05,
"loss": 0.6623,
"step": 26500
},
{
"epoch": 0.22,
"learning_rate": 4.820359281437126e-05,
"loss": 0.6574,
"step": 27000
},
{
"epoch": 0.22,
"learning_rate": 4.817032601463739e-05,
"loss": 0.6492,
"step": 27500
},
{
"epoch": 0.22,
"learning_rate": 4.813705921490353e-05,
"loss": 0.6371,
"step": 28000
},
{
"epoch": 0.23,
"learning_rate": 4.8103792415169665e-05,
"loss": 0.6509,
"step": 28500
},
{
"epoch": 0.23,
"learning_rate": 4.80705256154358e-05,
"loss": 0.6443,
"step": 29000
},
{
"epoch": 0.24,
"learning_rate": 4.803725881570193e-05,
"loss": 0.634,
"step": 29500
},
{
"epoch": 0.24,
"learning_rate": 4.800399201596807e-05,
"loss": 0.6575,
"step": 30000
},
{
"epoch": 0.24,
"learning_rate": 4.79707252162342e-05,
"loss": 0.6518,
"step": 30500
},
{
"epoch": 0.25,
"learning_rate": 4.7937458416500335e-05,
"loss": 0.6584,
"step": 31000
},
{
"epoch": 0.25,
"learning_rate": 4.790419161676647e-05,
"loss": 0.6374,
"step": 31500
},
{
"epoch": 0.26,
"learning_rate": 4.78709248170326e-05,
"loss": 0.6481,
"step": 32000
},
{
"epoch": 0.26,
"learning_rate": 4.783765801729873e-05,
"loss": 0.6438,
"step": 32500
},
{
"epoch": 0.26,
"learning_rate": 4.780439121756487e-05,
"loss": 0.6349,
"step": 33000
},
{
"epoch": 0.27,
"learning_rate": 4.777112441783101e-05,
"loss": 0.63,
"step": 33500
},
{
"epoch": 0.27,
"learning_rate": 4.7737857618097145e-05,
"loss": 0.6447,
"step": 34000
},
{
"epoch": 0.28,
"learning_rate": 4.770459081836328e-05,
"loss": 0.6462,
"step": 34500
},
{
"epoch": 0.28,
"learning_rate": 4.767132401862941e-05,
"loss": 0.6317,
"step": 35000
},
{
"epoch": 0.28,
"learning_rate": 4.763805721889554e-05,
"loss": 0.641,
"step": 35500
},
{
"epoch": 0.29,
"learning_rate": 4.7604790419161675e-05,
"loss": 0.6407,
"step": 36000
},
{
"epoch": 0.29,
"learning_rate": 4.7571523619427815e-05,
"loss": 0.6353,
"step": 36500
},
{
"epoch": 0.3,
"learning_rate": 4.753825681969395e-05,
"loss": 0.6284,
"step": 37000
},
{
"epoch": 0.3,
"learning_rate": 4.750499001996008e-05,
"loss": 0.6333,
"step": 37500
},
{
"epoch": 0.3,
"learning_rate": 4.747172322022622e-05,
"loss": 0.6202,
"step": 38000
},
{
"epoch": 0.31,
"learning_rate": 4.743845642049235e-05,
"loss": 0.6277,
"step": 38500
},
{
"epoch": 0.31,
"learning_rate": 4.7405189620758485e-05,
"loss": 0.6416,
"step": 39000
},
{
"epoch": 0.32,
"learning_rate": 4.7371922821024624e-05,
"loss": 0.6256,
"step": 39500
},
{
"epoch": 0.32,
"learning_rate": 4.733865602129076e-05,
"loss": 0.628,
"step": 40000
},
{
"epoch": 0.32,
"learning_rate": 4.730538922155689e-05,
"loss": 0.6328,
"step": 40500
},
{
"epoch": 0.33,
"learning_rate": 4.727212242182302e-05,
"loss": 0.6379,
"step": 41000
},
{
"epoch": 0.33,
"learning_rate": 4.7238855622089155e-05,
"loss": 0.6279,
"step": 41500
},
{
"epoch": 0.34,
"learning_rate": 4.720558882235529e-05,
"loss": 0.6298,
"step": 42000
},
{
"epoch": 0.34,
"learning_rate": 4.717232202262143e-05,
"loss": 0.6369,
"step": 42500
},
{
"epoch": 0.34,
"learning_rate": 4.713905522288756e-05,
"loss": 0.6135,
"step": 43000
},
{
"epoch": 0.35,
"learning_rate": 4.71057884231537e-05,
"loss": 0.6188,
"step": 43500
},
{
"epoch": 0.35,
"learning_rate": 4.707252162341983e-05,
"loss": 0.638,
"step": 44000
},
{
"epoch": 0.36,
"learning_rate": 4.7039254823685964e-05,
"loss": 0.638,
"step": 44500
},
{
"epoch": 0.36,
"learning_rate": 4.70059880239521e-05,
"loss": 0.6346,
"step": 45000
},
{
"epoch": 0.36,
"learning_rate": 4.697272122421823e-05,
"loss": 0.6261,
"step": 45500
},
{
"epoch": 0.37,
"learning_rate": 4.693945442448437e-05,
"loss": 0.6294,
"step": 46000
},
{
"epoch": 0.37,
"learning_rate": 4.69061876247505e-05,
"loss": 0.6438,
"step": 46500
},
{
"epoch": 0.38,
"learning_rate": 4.6872920825016634e-05,
"loss": 0.6299,
"step": 47000
}
],
"max_steps": 751500,
"num_train_epochs": 6,
"total_flos": 57057686327623680,
"trial_name": null,
"trial_params": null
}