Upload trainer_state.json with huggingface_hub
Browse files- trainer_state.json +89 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,89 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9696969696969697,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 8,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.12121212121212122,
|
| 13 |
+
"grad_norm": 3.3049814701080322,
|
| 14 |
+
"learning_rate": 0.0002,
|
| 15 |
+
"loss": 4.8222,
|
| 16 |
+
"step": 1
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.24242424242424243,
|
| 20 |
+
"grad_norm": 3.172274112701416,
|
| 21 |
+
"learning_rate": 0.0001900968867902419,
|
| 22 |
+
"loss": 4.9618,
|
| 23 |
+
"step": 2
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.36363636363636365,
|
| 27 |
+
"grad_norm": 1.6254035234451294,
|
| 28 |
+
"learning_rate": 0.00016234898018587337,
|
| 29 |
+
"loss": 3.2341,
|
| 30 |
+
"step": 3
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.48484848484848486,
|
| 34 |
+
"grad_norm": 1.523720145225525,
|
| 35 |
+
"learning_rate": 0.00012225209339563145,
|
| 36 |
+
"loss": 2.9603,
|
| 37 |
+
"step": 4
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.6060606060606061,
|
| 41 |
+
"grad_norm": 1.194486379623413,
|
| 42 |
+
"learning_rate": 7.774790660436858e-05,
|
| 43 |
+
"loss": 2.4278,
|
| 44 |
+
"step": 5
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.7272727272727273,
|
| 48 |
+
"grad_norm": 1.3157984018325806,
|
| 49 |
+
"learning_rate": 3.7651019814126654e-05,
|
| 50 |
+
"loss": 1.9683,
|
| 51 |
+
"step": 6
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.8484848484848485,
|
| 55 |
+
"grad_norm": 1.3595160245895386,
|
| 56 |
+
"learning_rate": 9.903113209758096e-06,
|
| 57 |
+
"loss": 2.2561,
|
| 58 |
+
"step": 7
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.9696969696969697,
|
| 62 |
+
"grad_norm": 1.2659095525741577,
|
| 63 |
+
"learning_rate": 0.0,
|
| 64 |
+
"loss": 2.2029,
|
| 65 |
+
"step": 8
|
| 66 |
+
}
|
| 67 |
+
],
|
| 68 |
+
"logging_steps": 1,
|
| 69 |
+
"max_steps": 8,
|
| 70 |
+
"num_input_tokens_seen": 0,
|
| 71 |
+
"num_train_epochs": 1,
|
| 72 |
+
"save_steps": 500,
|
| 73 |
+
"stateful_callbacks": {
|
| 74 |
+
"TrainerControl": {
|
| 75 |
+
"args": {
|
| 76 |
+
"should_epoch_stop": false,
|
| 77 |
+
"should_evaluate": false,
|
| 78 |
+
"should_log": false,
|
| 79 |
+
"should_save": true,
|
| 80 |
+
"should_training_stop": true
|
| 81 |
+
},
|
| 82 |
+
"attributes": {}
|
| 83 |
+
}
|
| 84 |
+
},
|
| 85 |
+
"total_flos": 222371544858624.0,
|
| 86 |
+
"train_batch_size": 1,
|
| 87 |
+
"trial_name": null,
|
| 88 |
+
"trial_params": null
|
| 89 |
+
}
|