Upload trainer_state.json
Browse files- trainer_state.json +72 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,72 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.9024694561958313,
|
| 3 |
+
"best_model_checkpoint": "models/checkpoints/checkpoint-2000",
|
| 4 |
+
"epoch": 2.0986358866736623,
|
| 5 |
+
"global_step": 2000,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.52,
|
| 12 |
+
"learning_rate": 1.9475341028331587e-05,
|
| 13 |
+
"loss": 1.2025,
|
| 14 |
+
"step": 500
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.52,
|
| 18 |
+
"eval_loss": 0.9926990270614624,
|
| 19 |
+
"eval_runtime": 325.2407,
|
| 20 |
+
"eval_samples_per_second": 43.454,
|
| 21 |
+
"eval_steps_per_second": 0.341,
|
| 22 |
+
"step": 500
|
| 23 |
+
},
|
| 24 |
+
{
|
| 25 |
+
"epoch": 1.05,
|
| 26 |
+
"learning_rate": 1.895068205666317e-05,
|
| 27 |
+
"loss": 1.0125,
|
| 28 |
+
"step": 1000
|
| 29 |
+
},
|
| 30 |
+
{
|
| 31 |
+
"epoch": 1.05,
|
| 32 |
+
"eval_loss": 0.948229968547821,
|
| 33 |
+
"eval_runtime": 325.6473,
|
| 34 |
+
"eval_samples_per_second": 43.4,
|
| 35 |
+
"eval_steps_per_second": 0.341,
|
| 36 |
+
"step": 1000
|
| 37 |
+
},
|
| 38 |
+
{
|
| 39 |
+
"epoch": 1.57,
|
| 40 |
+
"learning_rate": 1.8426023084994754e-05,
|
| 41 |
+
"loss": 0.9408,
|
| 42 |
+
"step": 1500
|
| 43 |
+
},
|
| 44 |
+
{
|
| 45 |
+
"epoch": 1.57,
|
| 46 |
+
"eval_loss": 0.9487974047660828,
|
| 47 |
+
"eval_runtime": 410.664,
|
| 48 |
+
"eval_samples_per_second": 34.415,
|
| 49 |
+
"eval_steps_per_second": 0.27,
|
| 50 |
+
"step": 1500
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 2.1,
|
| 54 |
+
"learning_rate": 1.790136411332634e-05,
|
| 55 |
+
"loss": 0.9073,
|
| 56 |
+
"step": 2000
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 2.1,
|
| 60 |
+
"eval_loss": 0.9024694561958313,
|
| 61 |
+
"eval_runtime": 409.1112,
|
| 62 |
+
"eval_samples_per_second": 34.546,
|
| 63 |
+
"eval_steps_per_second": 0.271,
|
| 64 |
+
"step": 2000
|
| 65 |
+
}
|
| 66 |
+
],
|
| 67 |
+
"max_steps": 19060,
|
| 68 |
+
"num_train_epochs": 20,
|
| 69 |
+
"total_flos": 1.3535851069855128e+16,
|
| 70 |
+
"trial_name": null,
|
| 71 |
+
"trial_params": null
|
| 72 |
+
}
|