Upload trainer_state.json with huggingface_hub
Browse files- trainer_state.json +83 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,83 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 263.1578947368421,
|
| 5 |
+
"global_step": 5000,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 26.32,
|
| 12 |
+
"learning_rate": 2.5e-05,
|
| 13 |
+
"loss": 0.2555,
|
| 14 |
+
"step": 500
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 52.63,
|
| 18 |
+
"learning_rate": 2.2222222222222223e-05,
|
| 19 |
+
"loss": 0.1603,
|
| 20 |
+
"step": 1000
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 78.95,
|
| 24 |
+
"learning_rate": 1.9444444444444445e-05,
|
| 25 |
+
"loss": 0.1074,
|
| 26 |
+
"step": 1500
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 105.26,
|
| 30 |
+
"learning_rate": 1.6666666666666667e-05,
|
| 31 |
+
"loss": 0.0881,
|
| 32 |
+
"step": 2000
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 131.58,
|
| 36 |
+
"learning_rate": 1.388888888888889e-05,
|
| 37 |
+
"loss": 0.0761,
|
| 38 |
+
"step": 2500
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 157.89,
|
| 42 |
+
"learning_rate": 1.1111111111111112e-05,
|
| 43 |
+
"loss": 0.0678,
|
| 44 |
+
"step": 3000
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 184.21,
|
| 48 |
+
"learning_rate": 8.333333333333334e-06,
|
| 49 |
+
"loss": 0.0619,
|
| 50 |
+
"step": 3500
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 210.53,
|
| 54 |
+
"learning_rate": 5.555555555555556e-06,
|
| 55 |
+
"loss": 0.0583,
|
| 56 |
+
"step": 4000
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 236.84,
|
| 60 |
+
"learning_rate": 2.777777777777778e-06,
|
| 61 |
+
"loss": 0.0549,
|
| 62 |
+
"step": 4500
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 263.16,
|
| 66 |
+
"learning_rate": 0.0,
|
| 67 |
+
"loss": 0.0543,
|
| 68 |
+
"step": 5000
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 263.16,
|
| 72 |
+
"step": 5000,
|
| 73 |
+
"total_flos": 3.53160818688e+16,
|
| 74 |
+
"train_runtime": 2716.3773,
|
| 75 |
+
"train_samples_per_second": 1.841
|
| 76 |
+
}
|
| 77 |
+
],
|
| 78 |
+
"max_steps": 5000,
|
| 79 |
+
"num_train_epochs": 264,
|
| 80 |
+
"total_flos": 3.53160818688e+16,
|
| 81 |
+
"trial_name": null,
|
| 82 |
+
"trial_params": null
|
| 83 |
+
}
|