roberta-base-ca-cased-tc / trainer_state.json
ccasimiro's picture
upload model
488c07f
raw
history blame
11 kB
{
"best_metric": 0.7445234060287476,
"best_model_checkpoint": "/home/shared/dt01/temutauro/ccasimiro/berta/src/finetuning/tc/roberta-base-ca-cased-tecla/checkpoint-34440",
"epoch": 10.0,
"global_step": 34440,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.15,
"learning_rate": 4.9274099883855984e-05,
"loss": 1.3202,
"step": 500
},
{
"epoch": 0.29,
"learning_rate": 4.8548199767711965e-05,
"loss": 0.9902,
"step": 1000
},
{
"epoch": 0.44,
"learning_rate": 4.782229965156795e-05,
"loss": 0.9309,
"step": 1500
},
{
"epoch": 0.58,
"learning_rate": 4.709639953542393e-05,
"loss": 0.9006,
"step": 2000
},
{
"epoch": 0.73,
"learning_rate": 4.637049941927991e-05,
"loss": 0.8717,
"step": 2500
},
{
"epoch": 0.87,
"learning_rate": 4.564459930313589e-05,
"loss": 0.8612,
"step": 3000
},
{
"epoch": 1.0,
"eval_accuracy": 0.6943275928497314,
"eval_loss": 0.8295034766197205,
"eval_runtime": 288.6692,
"eval_samples_per_second": 47.757,
"step": 3444
},
{
"epoch": 1.02,
"learning_rate": 4.491869918699187e-05,
"loss": 0.8394,
"step": 3500
},
{
"epoch": 1.16,
"learning_rate": 4.4192799070847854e-05,
"loss": 0.7212,
"step": 4000
},
{
"epoch": 1.31,
"learning_rate": 4.3466898954703835e-05,
"loss": 0.7093,
"step": 4500
},
{
"epoch": 1.45,
"learning_rate": 4.2740998838559817e-05,
"loss": 0.6932,
"step": 5000
},
{
"epoch": 1.6,
"learning_rate": 4.20150987224158e-05,
"loss": 0.7032,
"step": 5500
},
{
"epoch": 1.74,
"learning_rate": 4.128919860627178e-05,
"loss": 0.6976,
"step": 6000
},
{
"epoch": 1.89,
"learning_rate": 4.056329849012776e-05,
"loss": 0.6866,
"step": 6500
},
{
"epoch": 2.0,
"eval_accuracy": 0.7147105932235718,
"eval_loss": 0.7905623912811279,
"eval_runtime": 288.7903,
"eval_samples_per_second": 47.737,
"step": 6888
},
{
"epoch": 2.03,
"learning_rate": 3.983739837398374e-05,
"loss": 0.648,
"step": 7000
},
{
"epoch": 2.18,
"learning_rate": 3.9111498257839724e-05,
"loss": 0.504,
"step": 7500
},
{
"epoch": 2.32,
"learning_rate": 3.8385598141695705e-05,
"loss": 0.5056,
"step": 8000
},
{
"epoch": 2.47,
"learning_rate": 3.7659698025551686e-05,
"loss": 0.5104,
"step": 8500
},
{
"epoch": 2.61,
"learning_rate": 3.693379790940767e-05,
"loss": 0.5021,
"step": 9000
},
{
"epoch": 2.76,
"learning_rate": 3.620789779326365e-05,
"loss": 0.502,
"step": 9500
},
{
"epoch": 2.9,
"learning_rate": 3.548199767711963e-05,
"loss": 0.501,
"step": 10000
},
{
"epoch": 3.0,
"eval_accuracy": 0.7299433946609497,
"eval_loss": 0.8311841487884521,
"eval_runtime": 289.0756,
"eval_samples_per_second": 47.69,
"step": 10332
},
{
"epoch": 3.05,
"learning_rate": 3.475609756097561e-05,
"loss": 0.44,
"step": 10500
},
{
"epoch": 3.19,
"learning_rate": 3.4030197444831594e-05,
"loss": 0.3205,
"step": 11000
},
{
"epoch": 3.34,
"learning_rate": 3.3304297328687575e-05,
"loss": 0.3217,
"step": 11500
},
{
"epoch": 3.48,
"learning_rate": 3.2578397212543556e-05,
"loss": 0.3295,
"step": 12000
},
{
"epoch": 3.63,
"learning_rate": 3.185249709639954e-05,
"loss": 0.3314,
"step": 12500
},
{
"epoch": 3.77,
"learning_rate": 3.112659698025552e-05,
"loss": 0.3384,
"step": 13000
},
{
"epoch": 3.92,
"learning_rate": 3.04006968641115e-05,
"loss": 0.3415,
"step": 13500
},
{
"epoch": 4.0,
"eval_accuracy": 0.722907304763794,
"eval_loss": 0.9545897245407104,
"eval_runtime": 288.9468,
"eval_samples_per_second": 47.711,
"step": 13776
},
{
"epoch": 4.07,
"learning_rate": 2.9674796747967482e-05,
"loss": 0.2816,
"step": 14000
},
{
"epoch": 4.21,
"learning_rate": 2.8948896631823464e-05,
"loss": 0.1995,
"step": 14500
},
{
"epoch": 4.36,
"learning_rate": 2.8222996515679445e-05,
"loss": 0.2125,
"step": 15000
},
{
"epoch": 4.5,
"learning_rate": 2.7497096399535426e-05,
"loss": 0.2144,
"step": 15500
},
{
"epoch": 4.65,
"learning_rate": 2.6771196283391408e-05,
"loss": 0.2109,
"step": 16000
},
{
"epoch": 4.79,
"learning_rate": 2.604529616724739e-05,
"loss": 0.2238,
"step": 16500
},
{
"epoch": 4.94,
"learning_rate": 2.5319396051103367e-05,
"loss": 0.223,
"step": 17000
},
{
"epoch": 5.0,
"eval_accuracy": 0.7305237054824829,
"eval_loss": 1.0980908870697021,
"eval_runtime": 288.7866,
"eval_samples_per_second": 47.738,
"step": 17220
},
{
"epoch": 5.08,
"learning_rate": 2.4593495934959352e-05,
"loss": 0.1694,
"step": 17500
},
{
"epoch": 5.23,
"learning_rate": 2.3867595818815333e-05,
"loss": 0.1347,
"step": 18000
},
{
"epoch": 5.37,
"learning_rate": 2.314169570267131e-05,
"loss": 0.1356,
"step": 18500
},
{
"epoch": 5.52,
"learning_rate": 2.2415795586527293e-05,
"loss": 0.1274,
"step": 19000
},
{
"epoch": 5.66,
"learning_rate": 2.1689895470383274e-05,
"loss": 0.1416,
"step": 19500
},
{
"epoch": 5.81,
"learning_rate": 2.096399535423926e-05,
"loss": 0.1375,
"step": 20000
},
{
"epoch": 5.95,
"learning_rate": 2.023809523809524e-05,
"loss": 0.15,
"step": 20500
},
{
"epoch": 6.0,
"eval_accuracy": 0.7300159335136414,
"eval_loss": 1.323264479637146,
"eval_runtime": 289.0229,
"eval_samples_per_second": 47.699,
"step": 20664
},
{
"epoch": 6.1,
"learning_rate": 1.9512195121951222e-05,
"loss": 0.1056,
"step": 21000
},
{
"epoch": 6.24,
"learning_rate": 1.8786295005807203e-05,
"loss": 0.0862,
"step": 21500
},
{
"epoch": 6.39,
"learning_rate": 1.806039488966318e-05,
"loss": 0.0916,
"step": 22000
},
{
"epoch": 6.53,
"learning_rate": 1.7334494773519163e-05,
"loss": 0.0954,
"step": 22500
},
{
"epoch": 6.68,
"learning_rate": 1.6608594657375144e-05,
"loss": 0.0956,
"step": 23000
},
{
"epoch": 6.82,
"learning_rate": 1.5882694541231126e-05,
"loss": 0.0891,
"step": 23500
},
{
"epoch": 6.97,
"learning_rate": 1.5156794425087109e-05,
"loss": 0.0935,
"step": 24000
},
{
"epoch": 7.0,
"eval_accuracy": 0.7382127046585083,
"eval_loss": 1.6066738367080688,
"eval_runtime": 288.9144,
"eval_samples_per_second": 47.717,
"step": 24108
},
{
"epoch": 7.11,
"learning_rate": 1.443089430894309e-05,
"loss": 0.0559,
"step": 24500
},
{
"epoch": 7.26,
"learning_rate": 1.3704994192799073e-05,
"loss": 0.0581,
"step": 25000
},
{
"epoch": 7.4,
"learning_rate": 1.2979094076655051e-05,
"loss": 0.06,
"step": 25500
},
{
"epoch": 7.55,
"learning_rate": 1.2253193960511034e-05,
"loss": 0.055,
"step": 26000
},
{
"epoch": 7.69,
"learning_rate": 1.1527293844367016e-05,
"loss": 0.0677,
"step": 26500
},
{
"epoch": 7.84,
"learning_rate": 1.0801393728222997e-05,
"loss": 0.0552,
"step": 27000
},
{
"epoch": 7.98,
"learning_rate": 1.0075493612078979e-05,
"loss": 0.0642,
"step": 27500
},
{
"epoch": 8.0,
"eval_accuracy": 0.7361816167831421,
"eval_loss": 1.801051378250122,
"eval_runtime": 288.9587,
"eval_samples_per_second": 47.709,
"step": 27552
},
{
"epoch": 8.13,
"learning_rate": 9.34959349593496e-06,
"loss": 0.0399,
"step": 28000
},
{
"epoch": 8.28,
"learning_rate": 8.623693379790942e-06,
"loss": 0.0324,
"step": 28500
},
{
"epoch": 8.42,
"learning_rate": 7.897793263646923e-06,
"loss": 0.0383,
"step": 29000
},
{
"epoch": 8.57,
"learning_rate": 7.1718931475029035e-06,
"loss": 0.0396,
"step": 29500
},
{
"epoch": 8.71,
"learning_rate": 6.445993031358885e-06,
"loss": 0.038,
"step": 30000
},
{
"epoch": 8.86,
"learning_rate": 5.720092915214867e-06,
"loss": 0.0355,
"step": 30500
},
{
"epoch": 9.0,
"eval_accuracy": 0.7436529994010925,
"eval_loss": 1.9809679985046387,
"eval_runtime": 288.9869,
"eval_samples_per_second": 47.705,
"step": 30996
},
{
"epoch": 9.0,
"learning_rate": 4.994192799070848e-06,
"loss": 0.0348,
"step": 31000
},
{
"epoch": 9.15,
"learning_rate": 4.26829268292683e-06,
"loss": 0.0194,
"step": 31500
},
{
"epoch": 9.29,
"learning_rate": 3.542392566782811e-06,
"loss": 0.0181,
"step": 32000
},
{
"epoch": 9.44,
"learning_rate": 2.8164924506387924e-06,
"loss": 0.0158,
"step": 32500
},
{
"epoch": 9.58,
"learning_rate": 2.090592334494774e-06,
"loss": 0.0201,
"step": 33000
},
{
"epoch": 9.73,
"learning_rate": 1.364692218350755e-06,
"loss": 0.0218,
"step": 33500
},
{
"epoch": 9.87,
"learning_rate": 6.387921022067365e-07,
"loss": 0.0203,
"step": 34000
},
{
"epoch": 10.0,
"eval_accuracy": 0.7445234060287476,
"eval_loss": 2.0729739665985107,
"eval_runtime": 288.9098,
"eval_samples_per_second": 47.717,
"step": 34440
},
{
"epoch": 10.0,
"step": 34440,
"total_flos": 0,
"train_runtime": 75009.1041,
"train_samples_per_second": 0.459
}
],
"max_steps": 34440,
"num_train_epochs": 10,
"total_flos": 0,
"trial_name": null,
"trial_params": null
}