Upload trainer_state.json
Browse files- trainer_state.json +105 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,105 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 7.0,
|
| 5 |
+
"global_step": 1477,
|
| 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_accuracy": 0.8411080332409973,
|
| 13 |
+
"eval_f1": 0.8052208835341366,
|
| 14 |
+
"eval_loss": 0.06445828080177307,
|
| 15 |
+
"eval_precision": 0.8038757099899766,
|
| 16 |
+
"eval_recall": 0.8065705665437479,
|
| 17 |
+
"eval_runtime": 4.9709,
|
| 18 |
+
"eval_samples_per_second": 678.346,
|
| 19 |
+
"eval_steps_per_second": 5.432,
|
| 20 |
+
"step": 211
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 2.0,
|
| 24 |
+
"eval_accuracy": 0.914016620498615,
|
| 25 |
+
"eval_f1": 0.9096102150537634,
|
| 26 |
+
"eval_loss": 0.03706464543938637,
|
| 27 |
+
"eval_precision": 0.9117547995958235,
|
| 28 |
+
"eval_recall": 0.9074756956084479,
|
| 29 |
+
"eval_runtime": 4.9541,
|
| 30 |
+
"eval_samples_per_second": 680.645,
|
| 31 |
+
"eval_steps_per_second": 5.45,
|
| 32 |
+
"step": 422
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 3.0,
|
| 36 |
+
"eval_accuracy": 0.9270914127423823,
|
| 37 |
+
"eval_f1": 0.9288702928870293,
|
| 38 |
+
"eval_loss": 0.03221126273274422,
|
| 39 |
+
"eval_precision": 0.9274732620320856,
|
| 40 |
+
"eval_recall": 0.93027153871941,
|
| 41 |
+
"eval_runtime": 5.0612,
|
| 42 |
+
"eval_samples_per_second": 666.244,
|
| 43 |
+
"eval_steps_per_second": 5.335,
|
| 44 |
+
"step": 633
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 4.0,
|
| 48 |
+
"eval_accuracy": 0.9380609418282548,
|
| 49 |
+
"eval_f1": 0.9411566346315967,
|
| 50 |
+
"eval_loss": 0.029083197936415672,
|
| 51 |
+
"eval_precision": 0.9467435549525102,
|
| 52 |
+
"eval_recall": 0.9356352665102247,
|
| 53 |
+
"eval_runtime": 4.9781,
|
| 54 |
+
"eval_samples_per_second": 677.372,
|
| 55 |
+
"eval_steps_per_second": 5.424,
|
| 56 |
+
"step": 844
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 4.74,
|
| 60 |
+
"loss": 0.1327,
|
| 61 |
+
"step": 1000
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"epoch": 5.0,
|
| 65 |
+
"eval_accuracy": 0.950803324099723,
|
| 66 |
+
"eval_f1": 0.953245879582913,
|
| 67 |
+
"eval_loss": 0.03200341761112213,
|
| 68 |
+
"eval_precision": 0.9564630442119474,
|
| 69 |
+
"eval_recall": 0.9500502849480389,
|
| 70 |
+
"eval_runtime": 4.9815,
|
| 71 |
+
"eval_samples_per_second": 676.904,
|
| 72 |
+
"eval_steps_per_second": 5.42,
|
| 73 |
+
"step": 1055
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 6.0,
|
| 77 |
+
"eval_accuracy": 0.9441551246537396,
|
| 78 |
+
"eval_f1": 0.9509078681909886,
|
| 79 |
+
"eval_loss": 0.03349672257900238,
|
| 80 |
+
"eval_precision": 0.9537942664418212,
|
| 81 |
+
"eval_recall": 0.9480388870264834,
|
| 82 |
+
"eval_runtime": 4.9755,
|
| 83 |
+
"eval_samples_per_second": 677.72,
|
| 84 |
+
"eval_steps_per_second": 5.427,
|
| 85 |
+
"step": 1266
|
| 86 |
+
},
|
| 87 |
+
{
|
| 88 |
+
"epoch": 7.0,
|
| 89 |
+
"eval_accuracy": 0.9472576177285319,
|
| 90 |
+
"eval_f1": 0.9518680578929655,
|
| 91 |
+
"eval_loss": 0.03363509103655815,
|
| 92 |
+
"eval_precision": 0.9557282865833052,
|
| 93 |
+
"eval_recall": 0.9480388870264834,
|
| 94 |
+
"eval_runtime": 5.0017,
|
| 95 |
+
"eval_samples_per_second": 674.172,
|
| 96 |
+
"eval_steps_per_second": 5.398,
|
| 97 |
+
"step": 1477
|
| 98 |
+
}
|
| 99 |
+
],
|
| 100 |
+
"max_steps": 1477,
|
| 101 |
+
"num_train_epochs": 7,
|
| 102 |
+
"total_flos": 5303134846185912.0,
|
| 103 |
+
"trial_name": null,
|
| 104 |
+
"trial_params": null
|
| 105 |
+
}
|