Upload /trainer_state.json with huggingface_hub
Browse files- trainer_state.json +87 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,87 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 5.319148936170213,
|
| 5 |
+
"global_step": 500,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 1.0,
|
| 12 |
+
"eval_gen_len": 132.7733,
|
| 13 |
+
"eval_loss": 1.3755453824996948,
|
| 14 |
+
"eval_rouge1": 53.8464,
|
| 15 |
+
"eval_rouge2": 25.6231,
|
| 16 |
+
"eval_rougeL": 36.8116,
|
| 17 |
+
"eval_rougeLsum": 49.9957,
|
| 18 |
+
"eval_runtime": 155.2181,
|
| 19 |
+
"eval_samples_per_second": 0.966,
|
| 20 |
+
"eval_steps_per_second": 0.245,
|
| 21 |
+
"step": 94
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 2.0,
|
| 25 |
+
"eval_gen_len": 137.1267,
|
| 26 |
+
"eval_loss": 1.2081164121627808,
|
| 27 |
+
"eval_rouge1": 55.5247,
|
| 28 |
+
"eval_rouge2": 27.5408,
|
| 29 |
+
"eval_rougeL": 37.7199,
|
| 30 |
+
"eval_rougeLsum": 51.7518,
|
| 31 |
+
"eval_runtime": 169.6036,
|
| 32 |
+
"eval_samples_per_second": 0.884,
|
| 33 |
+
"eval_steps_per_second": 0.224,
|
| 34 |
+
"step": 188
|
| 35 |
+
},
|
| 36 |
+
{
|
| 37 |
+
"epoch": 3.0,
|
| 38 |
+
"eval_gen_len": 137.62,
|
| 39 |
+
"eval_loss": 1.1149141788482666,
|
| 40 |
+
"eval_rouge1": 55.6412,
|
| 41 |
+
"eval_rouge2": 28.4076,
|
| 42 |
+
"eval_rougeL": 39.0516,
|
| 43 |
+
"eval_rougeLsum": 52.4525,
|
| 44 |
+
"eval_runtime": 161.1061,
|
| 45 |
+
"eval_samples_per_second": 0.931,
|
| 46 |
+
"eval_steps_per_second": 0.236,
|
| 47 |
+
"step": 282
|
| 48 |
+
},
|
| 49 |
+
{
|
| 50 |
+
"epoch": 4.0,
|
| 51 |
+
"eval_gen_len": 135.1733,
|
| 52 |
+
"eval_loss": 1.0563586950302124,
|
| 53 |
+
"eval_rouge1": 56.5376,
|
| 54 |
+
"eval_rouge2": 29.8226,
|
| 55 |
+
"eval_rougeL": 39.9324,
|
| 56 |
+
"eval_rougeLsum": 53.2425,
|
| 57 |
+
"eval_runtime": 157.7547,
|
| 58 |
+
"eval_samples_per_second": 0.951,
|
| 59 |
+
"eval_steps_per_second": 0.241,
|
| 60 |
+
"step": 376
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 5.0,
|
| 64 |
+
"eval_gen_len": 135.1733,
|
| 65 |
+
"eval_loss": 1.0107210874557495,
|
| 66 |
+
"eval_rouge1": 57.7194,
|
| 67 |
+
"eval_rouge2": 31.6091,
|
| 68 |
+
"eval_rougeL": 41.9389,
|
| 69 |
+
"eval_rougeLsum": 54.465,
|
| 70 |
+
"eval_runtime": 160.0847,
|
| 71 |
+
"eval_samples_per_second": 0.937,
|
| 72 |
+
"eval_steps_per_second": 0.237,
|
| 73 |
+
"step": 470
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 5.32,
|
| 77 |
+
"learning_rate": 9.361702127659576e-06,
|
| 78 |
+
"loss": 1.1609,
|
| 79 |
+
"step": 500
|
| 80 |
+
}
|
| 81 |
+
],
|
| 82 |
+
"max_steps": 940,
|
| 83 |
+
"num_train_epochs": 10,
|
| 84 |
+
"total_flos": 4323373681213440.0,
|
| 85 |
+
"trial_name": null,
|
| 86 |
+
"trial_params": null
|
| 87 |
+
}
|