|
{ |
|
"best_metric": 77.48501965183796, |
|
"best_model_checkpoint": "/data/jcanete/all_results/tar/beto_cased/epochs_2_bs_32_lr_5e-5/checkpoint-5100", |
|
"epoch": 2.0, |
|
"global_step": 5486, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.11, |
|
"eval_exact_match": 50.879848628193, |
|
"eval_f1": 69.30387024901542, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 4.546117389719285e-05, |
|
"loss": 2.1254, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"eval_exact_match": 54.512771996215704, |
|
"eval_f1": 71.89755492209031, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"eval_exact_match": 55.96972563859981, |
|
"eval_f1": 73.46469051491528, |
|
"step": 900 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 4.090411957710536e-05, |
|
"loss": 1.7294, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"eval_exact_match": 56.60359508041627, |
|
"eval_f1": 73.90716618298387, |
|
"step": 1200 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 3.6347065257017864e-05, |
|
"loss": 1.6464, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"eval_exact_match": 56.773888363292336, |
|
"eval_f1": 74.46721951870393, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"eval_exact_match": 57.95648060548723, |
|
"eval_f1": 75.61075089328189, |
|
"step": 1800 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 3.179001093693037e-05, |
|
"loss": 1.5901, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"eval_exact_match": 58.240302743614, |
|
"eval_f1": 75.51663308469975, |
|
"step": 2100 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"eval_exact_match": 59.55534531693472, |
|
"eval_f1": 76.61330921600383, |
|
"step": 2400 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 2.7232956616842875e-05, |
|
"loss": 1.5281, |
|
"step": 2500 |
|
}, |
|
{ |
|
"epoch": 0.98, |
|
"eval_exact_match": 59.94323557237465, |
|
"eval_f1": 76.49598618587426, |
|
"step": 2700 |
|
}, |
|
{ |
|
"epoch": 1.09, |
|
"learning_rate": 2.2675902296755377e-05, |
|
"loss": 1.3384, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 1.09, |
|
"eval_exact_match": 58.99716177861873, |
|
"eval_f1": 76.52241925728033, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 1.2, |
|
"eval_exact_match": 59.38505203405866, |
|
"eval_f1": 76.35485242703709, |
|
"step": 3300 |
|
}, |
|
{ |
|
"epoch": 1.28, |
|
"learning_rate": 1.8118847976667883e-05, |
|
"loss": 1.1384, |
|
"step": 3500 |
|
}, |
|
{ |
|
"epoch": 1.31, |
|
"eval_exact_match": 59.66887417218543, |
|
"eval_f1": 76.86602160252082, |
|
"step": 3600 |
|
}, |
|
{ |
|
"epoch": 1.42, |
|
"eval_exact_match": 60.3689687795648, |
|
"eval_f1": 76.98844667865373, |
|
"step": 3900 |
|
}, |
|
{ |
|
"epoch": 1.46, |
|
"learning_rate": 1.3561793656580388e-05, |
|
"loss": 1.1451, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 1.53, |
|
"eval_exact_match": 60.3689687795648, |
|
"eval_f1": 77.01614995652554, |
|
"step": 4200 |
|
}, |
|
{ |
|
"epoch": 1.64, |
|
"learning_rate": 9.013853445133066e-06, |
|
"loss": 1.1248, |
|
"step": 4500 |
|
}, |
|
{ |
|
"epoch": 1.64, |
|
"eval_exact_match": 60.53926206244087, |
|
"eval_f1": 77.22917143472323, |
|
"step": 4500 |
|
}, |
|
{ |
|
"epoch": 1.75, |
|
"eval_exact_match": 60.68117313150426, |
|
"eval_f1": 77.3287559059371, |
|
"step": 4800 |
|
}, |
|
{ |
|
"epoch": 1.82, |
|
"learning_rate": 4.456799125045571e-06, |
|
"loss": 1.0917, |
|
"step": 5000 |
|
}, |
|
{ |
|
"epoch": 1.86, |
|
"eval_exact_match": 61.00283822138127, |
|
"eval_f1": 77.48501965183796, |
|
"step": 5100 |
|
}, |
|
{ |
|
"epoch": 1.97, |
|
"eval_exact_match": 61.031220435193944, |
|
"eval_f1": 77.37778335318785, |
|
"step": 5400 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"step": 5486, |
|
"total_flos": 3.4090493726506464e+16, |
|
"train_loss": 1.41421453160958, |
|
"train_runtime": 7497.7066, |
|
"train_samples_per_second": 23.407, |
|
"train_steps_per_second": 0.732 |
|
} |
|
], |
|
"max_steps": 5486, |
|
"num_train_epochs": 2, |
|
"total_flos": 3.4090493726506464e+16, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|