bert-base-uncased-mrpc / trainer_state.json
Jeremiah Zhou
End of training
08705fe
{
"best_metric": 0.8578431372549019,
"best_model_checkpoint": "./fine-tune/bert-base-uncased/mrpc/checkpoint-690",
"epoch": 5.0,
"global_step": 1150,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 1.0,
"eval_accuracy": 0.8088235294117647,
"eval_combined_score": 0.8396277115497361,
"eval_f1": 0.8704318936877077,
"eval_loss": 0.4110628664493561,
"eval_runtime": 1.9445,
"eval_samples_per_second": 209.823,
"eval_steps_per_second": 26.228,
"step": 230
},
{
"epoch": 2.0,
"eval_accuracy": 0.8480392156862745,
"eval_combined_score": 0.8711185839523523,
"eval_f1": 0.89419795221843,
"eval_loss": 0.3761928379535675,
"eval_runtime": 2.0026,
"eval_samples_per_second": 203.738,
"eval_steps_per_second": 25.467,
"step": 460
},
{
"epoch": 2.17,
"learning_rate": 1.2025901942645698e-05,
"loss": 0.4287,
"step": 500
},
{
"epoch": 3.0,
"eval_accuracy": 0.8578431372549019,
"eval_combined_score": 0.8801000198059021,
"eval_f1": 0.9023569023569024,
"eval_loss": 0.557181179523468,
"eval_runtime": 1.4976,
"eval_samples_per_second": 272.437,
"eval_steps_per_second": 34.055,
"step": 690
},
{
"epoch": 4.0,
"eval_accuracy": 0.8553921568627451,
"eval_combined_score": 0.8765695619669012,
"eval_f1": 0.8977469670710572,
"eval_loss": 0.6086517572402954,
"eval_runtime": 1.1289,
"eval_samples_per_second": 361.413,
"eval_steps_per_second": 45.177,
"step": 920
},
{
"epoch": 4.35,
"learning_rate": 2.7752081406105456e-06,
"loss": 0.1172,
"step": 1000
},
{
"epoch": 5.0,
"eval_accuracy": 0.8455882352941176,
"eval_combined_score": 0.86782029565753,
"eval_f1": 0.8900523560209423,
"eval_loss": 0.6524226069450378,
"eval_runtime": 1.1704,
"eval_samples_per_second": 348.589,
"eval_steps_per_second": 43.574,
"step": 1150
},
{
"epoch": 5.0,
"step": 1150,
"total_flos": 1206364188825600.0,
"train_loss": 0.24623984295388926,
"train_runtime": 263.949,
"train_samples_per_second": 69.483,
"train_steps_per_second": 4.357
}
],
"max_steps": 1150,
"num_train_epochs": 5,
"total_flos": 1206364188825600.0,
"trial_name": null,
"trial_params": null
}