Upload trainer_state.json with huggingface_hub
Browse files- trainer_state.json +98 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,98 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 1784,
|
| 3 |
+
"best_metric": 0.8369551504363162,
|
| 4 |
+
"best_model_checkpoint": "outputs/bartpho-spam-binary/checkpoint-1784",
|
| 5 |
+
"epoch": 4.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 1784,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 1.0,
|
| 14 |
+
"eval_accuracy": 0.8298676748582231,
|
| 15 |
+
"eval_f1": 0.7260364185059205,
|
| 16 |
+
"eval_loss": 0.5030027627944946,
|
| 17 |
+
"eval_precision": 0.8798270552001894,
|
| 18 |
+
"eval_recall": 0.6933528974291974,
|
| 19 |
+
"eval_runtime": 7.8161,
|
| 20 |
+
"eval_samples_per_second": 203.043,
|
| 21 |
+
"eval_steps_per_second": 25.46,
|
| 22 |
+
"step": 446
|
| 23 |
+
},
|
| 24 |
+
{
|
| 25 |
+
"epoch": 2.0,
|
| 26 |
+
"eval_accuracy": 0.8412098298676749,
|
| 27 |
+
"eval_f1": 0.7812241521918941,
|
| 28 |
+
"eval_loss": 0.5157436728477478,
|
| 29 |
+
"eval_precision": 0.8159241618542106,
|
| 30 |
+
"eval_recall": 0.7610439991155957,
|
| 31 |
+
"eval_runtime": 7.8183,
|
| 32 |
+
"eval_samples_per_second": 202.986,
|
| 33 |
+
"eval_steps_per_second": 25.453,
|
| 34 |
+
"step": 892
|
| 35 |
+
},
|
| 36 |
+
{
|
| 37 |
+
"epoch": 3.0,
|
| 38 |
+
"eval_accuracy": 0.780718336483932,
|
| 39 |
+
"eval_f1": 0.7410714285714286,
|
| 40 |
+
"eval_loss": 0.5297622084617615,
|
| 41 |
+
"eval_precision": 0.730682044108667,
|
| 42 |
+
"eval_recall": 0.761203794898595,
|
| 43 |
+
"eval_runtime": 7.8172,
|
| 44 |
+
"eval_samples_per_second": 203.014,
|
| 45 |
+
"eval_steps_per_second": 25.457,
|
| 46 |
+
"step": 1338
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 3.9960740325294446,
|
| 50 |
+
"grad_norm": 3.245978593826294,
|
| 51 |
+
"learning_rate": 4.980352910793417e-05,
|
| 52 |
+
"loss": 0.3948,
|
| 53 |
+
"step": 1782
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 4.0,
|
| 57 |
+
"eval_accuracy": 0.8783868935097668,
|
| 58 |
+
"eval_f1": 0.8369551504363162,
|
| 59 |
+
"eval_loss": 0.31898775696754456,
|
| 60 |
+
"eval_precision": 0.8619013459042153,
|
| 61 |
+
"eval_recall": 0.8194197101565798,
|
| 62 |
+
"eval_runtime": 7.8168,
|
| 63 |
+
"eval_samples_per_second": 203.024,
|
| 64 |
+
"eval_steps_per_second": 25.458,
|
| 65 |
+
"step": 1784
|
| 66 |
+
}
|
| 67 |
+
],
|
| 68 |
+
"logging_steps": 1782,
|
| 69 |
+
"max_steps": 44600,
|
| 70 |
+
"num_input_tokens_seen": 0,
|
| 71 |
+
"num_train_epochs": 100,
|
| 72 |
+
"save_steps": 500,
|
| 73 |
+
"stateful_callbacks": {
|
| 74 |
+
"EarlyStoppingCallback": {
|
| 75 |
+
"args": {
|
| 76 |
+
"early_stopping_patience": 5,
|
| 77 |
+
"early_stopping_threshold": 0.0
|
| 78 |
+
},
|
| 79 |
+
"attributes": {
|
| 80 |
+
"early_stopping_patience_counter": 0
|
| 81 |
+
}
|
| 82 |
+
},
|
| 83 |
+
"TrainerControl": {
|
| 84 |
+
"args": {
|
| 85 |
+
"should_epoch_stop": false,
|
| 86 |
+
"should_evaluate": false,
|
| 87 |
+
"should_log": false,
|
| 88 |
+
"should_save": true,
|
| 89 |
+
"should_training_stop": false
|
| 90 |
+
},
|
| 91 |
+
"attributes": {}
|
| 92 |
+
}
|
| 93 |
+
},
|
| 94 |
+
"total_flos": 0.0,
|
| 95 |
+
"train_batch_size": 8,
|
| 96 |
+
"trial_name": null,
|
| 97 |
+
"trial_params": null
|
| 98 |
+
}
|