Commit
·
3d7b607
1
Parent(s):
683b5f3
Upload trainer_state.json
Browse files- trainer_state.json +70 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,70 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 4.991735537190083,
|
| 5 |
+
"global_step": 300,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 1.66,
|
| 12 |
+
"learning_rate": 6.666666666666667e-05,
|
| 13 |
+
"loss": 1.3311,
|
| 14 |
+
"step": 100
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 1.66,
|
| 18 |
+
"eval_accuracy": 0.6033057570457458,
|
| 19 |
+
"eval_loss": 1.0739655494689941,
|
| 20 |
+
"eval_runtime": 74.6016,
|
| 21 |
+
"eval_samples_per_second": 1.622,
|
| 22 |
+
"eval_steps_per_second": 0.416,
|
| 23 |
+
"step": 100
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 3.33,
|
| 27 |
+
"learning_rate": 3.3333333333333335e-05,
|
| 28 |
+
"loss": 0.6235,
|
| 29 |
+
"step": 200
|
| 30 |
+
},
|
| 31 |
+
{
|
| 32 |
+
"epoch": 3.33,
|
| 33 |
+
"eval_accuracy": 0.8512396812438965,
|
| 34 |
+
"eval_loss": 0.47899967432022095,
|
| 35 |
+
"eval_runtime": 71.6672,
|
| 36 |
+
"eval_samples_per_second": 1.688,
|
| 37 |
+
"eval_steps_per_second": 0.433,
|
| 38 |
+
"step": 200
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 4.99,
|
| 42 |
+
"learning_rate": 0.0,
|
| 43 |
+
"loss": 0.2154,
|
| 44 |
+
"step": 300
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 4.99,
|
| 48 |
+
"eval_accuracy": 0.9008264541625977,
|
| 49 |
+
"eval_loss": 0.3769935369491577,
|
| 50 |
+
"eval_runtime": 73.3118,
|
| 51 |
+
"eval_samples_per_second": 1.65,
|
| 52 |
+
"eval_steps_per_second": 0.423,
|
| 53 |
+
"step": 300
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 4.99,
|
| 57 |
+
"step": 300,
|
| 58 |
+
"total_flos": 4.279570856908188e+17,
|
| 59 |
+
"train_loss": 0.7233316548665365,
|
| 60 |
+
"train_runtime": 4793.1545,
|
| 61 |
+
"train_samples_per_second": 0.504,
|
| 62 |
+
"train_steps_per_second": 0.063
|
| 63 |
+
}
|
| 64 |
+
],
|
| 65 |
+
"max_steps": 300,
|
| 66 |
+
"num_train_epochs": 5,
|
| 67 |
+
"total_flos": 4.279570856908188e+17,
|
| 68 |
+
"trial_name": null,
|
| 69 |
+
"trial_params": null
|
| 70 |
+
}
|