Upload trainer_state.json
Browse files- trainer_state.json +88 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,88 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 4.9689440993788825,
|
| 5 |
+
"eval_steps": 200,
|
| 6 |
+
"global_step": 600,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.83,
|
| 13 |
+
"learning_rate": 1e-05,
|
| 14 |
+
"loss": 6.109,
|
| 15 |
+
"step": 100
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 1.66,
|
| 19 |
+
"learning_rate": 2e-05,
|
| 20 |
+
"loss": 1.7606,
|
| 21 |
+
"step": 200
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 1.66,
|
| 25 |
+
"eval_loss": 0.9147409796714783,
|
| 26 |
+
"eval_runtime": 3.2133,
|
| 27 |
+
"eval_samples_per_second": 120.438,
|
| 28 |
+
"eval_steps_per_second": 15.249,
|
| 29 |
+
"step": 200
|
| 30 |
+
},
|
| 31 |
+
{
|
| 32 |
+
"epoch": 2.48,
|
| 33 |
+
"learning_rate": 3e-05,
|
| 34 |
+
"loss": 0.8584,
|
| 35 |
+
"step": 300
|
| 36 |
+
},
|
| 37 |
+
{
|
| 38 |
+
"epoch": 3.31,
|
| 39 |
+
"learning_rate": 4e-05,
|
| 40 |
+
"loss": 0.579,
|
| 41 |
+
"step": 400
|
| 42 |
+
},
|
| 43 |
+
{
|
| 44 |
+
"epoch": 3.31,
|
| 45 |
+
"eval_loss": 0.3369111120700836,
|
| 46 |
+
"eval_runtime": 3.2981,
|
| 47 |
+
"eval_samples_per_second": 117.339,
|
| 48 |
+
"eval_steps_per_second": 14.857,
|
| 49 |
+
"step": 400
|
| 50 |
+
},
|
| 51 |
+
{
|
| 52 |
+
"epoch": 4.14,
|
| 53 |
+
"learning_rate": 5e-05,
|
| 54 |
+
"loss": 0.4074,
|
| 55 |
+
"step": 500
|
| 56 |
+
},
|
| 57 |
+
{
|
| 58 |
+
"epoch": 4.97,
|
| 59 |
+
"learning_rate": 0.0,
|
| 60 |
+
"loss": 0.2482,
|
| 61 |
+
"step": 600
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"epoch": 4.97,
|
| 65 |
+
"eval_loss": 0.11852475255727768,
|
| 66 |
+
"eval_runtime": 3.2809,
|
| 67 |
+
"eval_samples_per_second": 117.954,
|
| 68 |
+
"eval_steps_per_second": 14.935,
|
| 69 |
+
"step": 600
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 4.97,
|
| 73 |
+
"step": 600,
|
| 74 |
+
"total_flos": 510083481600000.0,
|
| 75 |
+
"train_loss": 1.6604227765401205,
|
| 76 |
+
"train_runtime": 352.86,
|
| 77 |
+
"train_samples_per_second": 27.376,
|
| 78 |
+
"train_steps_per_second": 1.7
|
| 79 |
+
}
|
| 80 |
+
],
|
| 81 |
+
"logging_steps": 100,
|
| 82 |
+
"max_steps": 600,
|
| 83 |
+
"num_train_epochs": 5,
|
| 84 |
+
"save_steps": 1000,
|
| 85 |
+
"total_flos": 510083481600000.0,
|
| 86 |
+
"trial_name": null,
|
| 87 |
+
"trial_params": null
|
| 88 |
+
}
|