Upload trainer_state.json with huggingface_hub
Browse files- trainer_state.json +3504 -4
trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -3507,6 +3507,3506 @@
|
|
| 3507 |
"learning_rate": 2.5e-05,
|
| 3508 |
"loss": 0.002,
|
| 3509 |
"step": 5000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3510 |
}
|
| 3511 |
],
|
| 3512 |
"logging_steps": 10,
|
|
@@ -3521,12 +7021,12 @@
|
|
| 3521 |
"should_evaluate": false,
|
| 3522 |
"should_log": false,
|
| 3523 |
"should_save": true,
|
| 3524 |
-
"should_training_stop":
|
| 3525 |
},
|
| 3526 |
"attributes": {}
|
| 3527 |
}
|
| 3528 |
},
|
| 3529 |
-
"total_flos":
|
| 3530 |
"train_batch_size": 1,
|
| 3531 |
"trial_name": null,
|
| 3532 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 10000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 3507 |
"learning_rate": 2.5e-05,
|
| 3508 |
"loss": 0.002,
|
| 3509 |
"step": 5000
|
| 3510 |
+
},
|
| 3511 |
+
{
|
| 3512 |
+
"epoch": 0.501,
|
| 3513 |
+
"grad_norm": 0.012683290056884289,
|
| 3514 |
+
"learning_rate": 2.495e-05,
|
| 3515 |
+
"loss": 0.0016,
|
| 3516 |
+
"step": 5010
|
| 3517 |
+
},
|
| 3518 |
+
{
|
| 3519 |
+
"epoch": 0.502,
|
| 3520 |
+
"grad_norm": 0.011495651677250862,
|
| 3521 |
+
"learning_rate": 2.4900000000000002e-05,
|
| 3522 |
+
"loss": 0.0016,
|
| 3523 |
+
"step": 5020
|
| 3524 |
+
},
|
| 3525 |
+
{
|
| 3526 |
+
"epoch": 0.503,
|
| 3527 |
+
"grad_norm": 0.014306634664535522,
|
| 3528 |
+
"learning_rate": 2.485e-05,
|
| 3529 |
+
"loss": 0.0018,
|
| 3530 |
+
"step": 5030
|
| 3531 |
+
},
|
| 3532 |
+
{
|
| 3533 |
+
"epoch": 0.504,
|
| 3534 |
+
"grad_norm": 0.02241896465420723,
|
| 3535 |
+
"learning_rate": 2.48e-05,
|
| 3536 |
+
"loss": 0.0021,
|
| 3537 |
+
"step": 5040
|
| 3538 |
+
},
|
| 3539 |
+
{
|
| 3540 |
+
"epoch": 0.505,
|
| 3541 |
+
"grad_norm": 0.017740361392498016,
|
| 3542 |
+
"learning_rate": 2.4750000000000002e-05,
|
| 3543 |
+
"loss": 0.0016,
|
| 3544 |
+
"step": 5050
|
| 3545 |
+
},
|
| 3546 |
+
{
|
| 3547 |
+
"epoch": 0.506,
|
| 3548 |
+
"grad_norm": 0.013199679553508759,
|
| 3549 |
+
"learning_rate": 2.47e-05,
|
| 3550 |
+
"loss": 0.0015,
|
| 3551 |
+
"step": 5060
|
| 3552 |
+
},
|
| 3553 |
+
{
|
| 3554 |
+
"epoch": 0.507,
|
| 3555 |
+
"grad_norm": 0.057298243045806885,
|
| 3556 |
+
"learning_rate": 2.465e-05,
|
| 3557 |
+
"loss": 0.0019,
|
| 3558 |
+
"step": 5070
|
| 3559 |
+
},
|
| 3560 |
+
{
|
| 3561 |
+
"epoch": 0.508,
|
| 3562 |
+
"grad_norm": 0.03238265961408615,
|
| 3563 |
+
"learning_rate": 2.46e-05,
|
| 3564 |
+
"loss": 0.0026,
|
| 3565 |
+
"step": 5080
|
| 3566 |
+
},
|
| 3567 |
+
{
|
| 3568 |
+
"epoch": 0.509,
|
| 3569 |
+
"grad_norm": 0.04820936918258667,
|
| 3570 |
+
"learning_rate": 2.455e-05,
|
| 3571 |
+
"loss": 0.0027,
|
| 3572 |
+
"step": 5090
|
| 3573 |
+
},
|
| 3574 |
+
{
|
| 3575 |
+
"epoch": 0.51,
|
| 3576 |
+
"grad_norm": 0.022526515647768974,
|
| 3577 |
+
"learning_rate": 2.45e-05,
|
| 3578 |
+
"loss": 0.0018,
|
| 3579 |
+
"step": 5100
|
| 3580 |
+
},
|
| 3581 |
+
{
|
| 3582 |
+
"epoch": 0.511,
|
| 3583 |
+
"grad_norm": 0.1899888962507248,
|
| 3584 |
+
"learning_rate": 2.445e-05,
|
| 3585 |
+
"loss": 0.0026,
|
| 3586 |
+
"step": 5110
|
| 3587 |
+
},
|
| 3588 |
+
{
|
| 3589 |
+
"epoch": 0.512,
|
| 3590 |
+
"grad_norm": 0.05366889387369156,
|
| 3591 |
+
"learning_rate": 2.44e-05,
|
| 3592 |
+
"loss": 0.003,
|
| 3593 |
+
"step": 5120
|
| 3594 |
+
},
|
| 3595 |
+
{
|
| 3596 |
+
"epoch": 0.513,
|
| 3597 |
+
"grad_norm": 0.028939131647348404,
|
| 3598 |
+
"learning_rate": 2.435e-05,
|
| 3599 |
+
"loss": 0.0021,
|
| 3600 |
+
"step": 5130
|
| 3601 |
+
},
|
| 3602 |
+
{
|
| 3603 |
+
"epoch": 0.514,
|
| 3604 |
+
"grad_norm": 0.023352844640612602,
|
| 3605 |
+
"learning_rate": 2.43e-05,
|
| 3606 |
+
"loss": 0.0019,
|
| 3607 |
+
"step": 5140
|
| 3608 |
+
},
|
| 3609 |
+
{
|
| 3610 |
+
"epoch": 0.515,
|
| 3611 |
+
"grad_norm": 0.015283104963600636,
|
| 3612 |
+
"learning_rate": 2.425e-05,
|
| 3613 |
+
"loss": 0.0017,
|
| 3614 |
+
"step": 5150
|
| 3615 |
+
},
|
| 3616 |
+
{
|
| 3617 |
+
"epoch": 0.516,
|
| 3618 |
+
"grad_norm": 0.0149134686216712,
|
| 3619 |
+
"learning_rate": 2.4200000000000002e-05,
|
| 3620 |
+
"loss": 0.0016,
|
| 3621 |
+
"step": 5160
|
| 3622 |
+
},
|
| 3623 |
+
{
|
| 3624 |
+
"epoch": 0.517,
|
| 3625 |
+
"grad_norm": 0.01739874854683876,
|
| 3626 |
+
"learning_rate": 2.415e-05,
|
| 3627 |
+
"loss": 0.0021,
|
| 3628 |
+
"step": 5170
|
| 3629 |
+
},
|
| 3630 |
+
{
|
| 3631 |
+
"epoch": 0.518,
|
| 3632 |
+
"grad_norm": 0.012562318705022335,
|
| 3633 |
+
"learning_rate": 2.41e-05,
|
| 3634 |
+
"loss": 0.0016,
|
| 3635 |
+
"step": 5180
|
| 3636 |
+
},
|
| 3637 |
+
{
|
| 3638 |
+
"epoch": 0.519,
|
| 3639 |
+
"grad_norm": 0.01181173324584961,
|
| 3640 |
+
"learning_rate": 2.4050000000000002e-05,
|
| 3641 |
+
"loss": 0.0017,
|
| 3642 |
+
"step": 5190
|
| 3643 |
+
},
|
| 3644 |
+
{
|
| 3645 |
+
"epoch": 0.52,
|
| 3646 |
+
"grad_norm": 0.0216183140873909,
|
| 3647 |
+
"learning_rate": 2.4e-05,
|
| 3648 |
+
"loss": 0.0017,
|
| 3649 |
+
"step": 5200
|
| 3650 |
+
},
|
| 3651 |
+
{
|
| 3652 |
+
"epoch": 0.521,
|
| 3653 |
+
"grad_norm": 0.014552557840943336,
|
| 3654 |
+
"learning_rate": 2.395e-05,
|
| 3655 |
+
"loss": 0.0017,
|
| 3656 |
+
"step": 5210
|
| 3657 |
+
},
|
| 3658 |
+
{
|
| 3659 |
+
"epoch": 0.522,
|
| 3660 |
+
"grad_norm": 0.013402258977293968,
|
| 3661 |
+
"learning_rate": 2.39e-05,
|
| 3662 |
+
"loss": 0.0015,
|
| 3663 |
+
"step": 5220
|
| 3664 |
+
},
|
| 3665 |
+
{
|
| 3666 |
+
"epoch": 0.523,
|
| 3667 |
+
"grad_norm": 0.017692307010293007,
|
| 3668 |
+
"learning_rate": 2.385e-05,
|
| 3669 |
+
"loss": 0.0017,
|
| 3670 |
+
"step": 5230
|
| 3671 |
+
},
|
| 3672 |
+
{
|
| 3673 |
+
"epoch": 0.524,
|
| 3674 |
+
"grad_norm": 0.007425515912473202,
|
| 3675 |
+
"learning_rate": 2.38e-05,
|
| 3676 |
+
"loss": 0.0015,
|
| 3677 |
+
"step": 5240
|
| 3678 |
+
},
|
| 3679 |
+
{
|
| 3680 |
+
"epoch": 0.525,
|
| 3681 |
+
"grad_norm": 0.010397032834589481,
|
| 3682 |
+
"learning_rate": 2.375e-05,
|
| 3683 |
+
"loss": 0.0014,
|
| 3684 |
+
"step": 5250
|
| 3685 |
+
},
|
| 3686 |
+
{
|
| 3687 |
+
"epoch": 0.526,
|
| 3688 |
+
"grad_norm": 0.013170558027923107,
|
| 3689 |
+
"learning_rate": 2.37e-05,
|
| 3690 |
+
"loss": 0.0017,
|
| 3691 |
+
"step": 5260
|
| 3692 |
+
},
|
| 3693 |
+
{
|
| 3694 |
+
"epoch": 0.527,
|
| 3695 |
+
"grad_norm": 0.47324055433273315,
|
| 3696 |
+
"learning_rate": 2.365e-05,
|
| 3697 |
+
"loss": 0.0037,
|
| 3698 |
+
"step": 5270
|
| 3699 |
+
},
|
| 3700 |
+
{
|
| 3701 |
+
"epoch": 0.528,
|
| 3702 |
+
"grad_norm": 0.06395496428012848,
|
| 3703 |
+
"learning_rate": 2.36e-05,
|
| 3704 |
+
"loss": 0.003,
|
| 3705 |
+
"step": 5280
|
| 3706 |
+
},
|
| 3707 |
+
{
|
| 3708 |
+
"epoch": 0.529,
|
| 3709 |
+
"grad_norm": 0.032293129712343216,
|
| 3710 |
+
"learning_rate": 2.355e-05,
|
| 3711 |
+
"loss": 0.0022,
|
| 3712 |
+
"step": 5290
|
| 3713 |
+
},
|
| 3714 |
+
{
|
| 3715 |
+
"epoch": 0.53,
|
| 3716 |
+
"grad_norm": 0.021514760330319405,
|
| 3717 |
+
"learning_rate": 2.35e-05,
|
| 3718 |
+
"loss": 0.002,
|
| 3719 |
+
"step": 5300
|
| 3720 |
+
},
|
| 3721 |
+
{
|
| 3722 |
+
"epoch": 0.531,
|
| 3723 |
+
"grad_norm": 0.016594447195529938,
|
| 3724 |
+
"learning_rate": 2.345e-05,
|
| 3725 |
+
"loss": 0.002,
|
| 3726 |
+
"step": 5310
|
| 3727 |
+
},
|
| 3728 |
+
{
|
| 3729 |
+
"epoch": 0.532,
|
| 3730 |
+
"grad_norm": 0.020661164075136185,
|
| 3731 |
+
"learning_rate": 2.3400000000000003e-05,
|
| 3732 |
+
"loss": 0.0018,
|
| 3733 |
+
"step": 5320
|
| 3734 |
+
},
|
| 3735 |
+
{
|
| 3736 |
+
"epoch": 0.533,
|
| 3737 |
+
"grad_norm": 0.01472094189375639,
|
| 3738 |
+
"learning_rate": 2.3350000000000002e-05,
|
| 3739 |
+
"loss": 0.0022,
|
| 3740 |
+
"step": 5330
|
| 3741 |
+
},
|
| 3742 |
+
{
|
| 3743 |
+
"epoch": 0.534,
|
| 3744 |
+
"grad_norm": 0.014501375146210194,
|
| 3745 |
+
"learning_rate": 2.3300000000000004e-05,
|
| 3746 |
+
"loss": 0.0017,
|
| 3747 |
+
"step": 5340
|
| 3748 |
+
},
|
| 3749 |
+
{
|
| 3750 |
+
"epoch": 0.535,
|
| 3751 |
+
"grad_norm": 0.01241264771670103,
|
| 3752 |
+
"learning_rate": 2.3250000000000003e-05,
|
| 3753 |
+
"loss": 0.0015,
|
| 3754 |
+
"step": 5350
|
| 3755 |
+
},
|
| 3756 |
+
{
|
| 3757 |
+
"epoch": 0.536,
|
| 3758 |
+
"grad_norm": 0.015589526854455471,
|
| 3759 |
+
"learning_rate": 2.32e-05,
|
| 3760 |
+
"loss": 0.0018,
|
| 3761 |
+
"step": 5360
|
| 3762 |
+
},
|
| 3763 |
+
{
|
| 3764 |
+
"epoch": 0.537,
|
| 3765 |
+
"grad_norm": 0.013468182645738125,
|
| 3766 |
+
"learning_rate": 2.3150000000000004e-05,
|
| 3767 |
+
"loss": 0.0018,
|
| 3768 |
+
"step": 5370
|
| 3769 |
+
},
|
| 3770 |
+
{
|
| 3771 |
+
"epoch": 0.538,
|
| 3772 |
+
"grad_norm": 0.015258733183145523,
|
| 3773 |
+
"learning_rate": 2.3100000000000002e-05,
|
| 3774 |
+
"loss": 0.0015,
|
| 3775 |
+
"step": 5380
|
| 3776 |
+
},
|
| 3777 |
+
{
|
| 3778 |
+
"epoch": 0.539,
|
| 3779 |
+
"grad_norm": 0.010932616889476776,
|
| 3780 |
+
"learning_rate": 2.305e-05,
|
| 3781 |
+
"loss": 0.0014,
|
| 3782 |
+
"step": 5390
|
| 3783 |
+
},
|
| 3784 |
+
{
|
| 3785 |
+
"epoch": 0.54,
|
| 3786 |
+
"grad_norm": 0.0102313794195652,
|
| 3787 |
+
"learning_rate": 2.3000000000000003e-05,
|
| 3788 |
+
"loss": 0.0014,
|
| 3789 |
+
"step": 5400
|
| 3790 |
+
},
|
| 3791 |
+
{
|
| 3792 |
+
"epoch": 0.541,
|
| 3793 |
+
"grad_norm": 0.00674120569601655,
|
| 3794 |
+
"learning_rate": 2.2950000000000002e-05,
|
| 3795 |
+
"loss": 0.0014,
|
| 3796 |
+
"step": 5410
|
| 3797 |
+
},
|
| 3798 |
+
{
|
| 3799 |
+
"epoch": 0.542,
|
| 3800 |
+
"grad_norm": 0.015179513022303581,
|
| 3801 |
+
"learning_rate": 2.29e-05,
|
| 3802 |
+
"loss": 0.0014,
|
| 3803 |
+
"step": 5420
|
| 3804 |
+
},
|
| 3805 |
+
{
|
| 3806 |
+
"epoch": 0.543,
|
| 3807 |
+
"grad_norm": 0.03448422998189926,
|
| 3808 |
+
"learning_rate": 2.2850000000000003e-05,
|
| 3809 |
+
"loss": 0.0019,
|
| 3810 |
+
"step": 5430
|
| 3811 |
+
},
|
| 3812 |
+
{
|
| 3813 |
+
"epoch": 0.544,
|
| 3814 |
+
"grad_norm": 0.028603358194231987,
|
| 3815 |
+
"learning_rate": 2.2800000000000002e-05,
|
| 3816 |
+
"loss": 0.0019,
|
| 3817 |
+
"step": 5440
|
| 3818 |
+
},
|
| 3819 |
+
{
|
| 3820 |
+
"epoch": 0.545,
|
| 3821 |
+
"grad_norm": 0.014372209087014198,
|
| 3822 |
+
"learning_rate": 2.275e-05,
|
| 3823 |
+
"loss": 0.0016,
|
| 3824 |
+
"step": 5450
|
| 3825 |
+
},
|
| 3826 |
+
{
|
| 3827 |
+
"epoch": 0.546,
|
| 3828 |
+
"grad_norm": 0.031532082706689835,
|
| 3829 |
+
"learning_rate": 2.2700000000000003e-05,
|
| 3830 |
+
"loss": 0.0017,
|
| 3831 |
+
"step": 5460
|
| 3832 |
+
},
|
| 3833 |
+
{
|
| 3834 |
+
"epoch": 0.547,
|
| 3835 |
+
"grad_norm": 0.018091056495904922,
|
| 3836 |
+
"learning_rate": 2.265e-05,
|
| 3837 |
+
"loss": 0.0016,
|
| 3838 |
+
"step": 5470
|
| 3839 |
+
},
|
| 3840 |
+
{
|
| 3841 |
+
"epoch": 0.548,
|
| 3842 |
+
"grad_norm": 0.014843069948256016,
|
| 3843 |
+
"learning_rate": 2.26e-05,
|
| 3844 |
+
"loss": 0.0015,
|
| 3845 |
+
"step": 5480
|
| 3846 |
+
},
|
| 3847 |
+
{
|
| 3848 |
+
"epoch": 0.549,
|
| 3849 |
+
"grad_norm": 0.011632148176431656,
|
| 3850 |
+
"learning_rate": 2.2550000000000003e-05,
|
| 3851 |
+
"loss": 0.0014,
|
| 3852 |
+
"step": 5490
|
| 3853 |
+
},
|
| 3854 |
+
{
|
| 3855 |
+
"epoch": 0.55,
|
| 3856 |
+
"grad_norm": 0.009511668235063553,
|
| 3857 |
+
"learning_rate": 2.25e-05,
|
| 3858 |
+
"loss": 0.0014,
|
| 3859 |
+
"step": 5500
|
| 3860 |
+
},
|
| 3861 |
+
{
|
| 3862 |
+
"epoch": 0.551,
|
| 3863 |
+
"grad_norm": 0.007981637492775917,
|
| 3864 |
+
"learning_rate": 2.245e-05,
|
| 3865 |
+
"loss": 0.0014,
|
| 3866 |
+
"step": 5510
|
| 3867 |
+
},
|
| 3868 |
+
{
|
| 3869 |
+
"epoch": 0.552,
|
| 3870 |
+
"grad_norm": 0.021288806572556496,
|
| 3871 |
+
"learning_rate": 2.2400000000000002e-05,
|
| 3872 |
+
"loss": 0.0015,
|
| 3873 |
+
"step": 5520
|
| 3874 |
+
},
|
| 3875 |
+
{
|
| 3876 |
+
"epoch": 0.553,
|
| 3877 |
+
"grad_norm": 0.01468642894178629,
|
| 3878 |
+
"learning_rate": 2.235e-05,
|
| 3879 |
+
"loss": 0.0018,
|
| 3880 |
+
"step": 5530
|
| 3881 |
+
},
|
| 3882 |
+
{
|
| 3883 |
+
"epoch": 0.554,
|
| 3884 |
+
"grad_norm": 0.011532713659107685,
|
| 3885 |
+
"learning_rate": 2.23e-05,
|
| 3886 |
+
"loss": 0.0012,
|
| 3887 |
+
"step": 5540
|
| 3888 |
+
},
|
| 3889 |
+
{
|
| 3890 |
+
"epoch": 0.555,
|
| 3891 |
+
"grad_norm": 0.00889046210795641,
|
| 3892 |
+
"learning_rate": 2.2250000000000002e-05,
|
| 3893 |
+
"loss": 0.0011,
|
| 3894 |
+
"step": 5550
|
| 3895 |
+
},
|
| 3896 |
+
{
|
| 3897 |
+
"epoch": 0.556,
|
| 3898 |
+
"grad_norm": 0.01401284895837307,
|
| 3899 |
+
"learning_rate": 2.22e-05,
|
| 3900 |
+
"loss": 0.0014,
|
| 3901 |
+
"step": 5560
|
| 3902 |
+
},
|
| 3903 |
+
{
|
| 3904 |
+
"epoch": 0.557,
|
| 3905 |
+
"grad_norm": 0.012369327247142792,
|
| 3906 |
+
"learning_rate": 2.215e-05,
|
| 3907 |
+
"loss": 0.0015,
|
| 3908 |
+
"step": 5570
|
| 3909 |
+
},
|
| 3910 |
+
{
|
| 3911 |
+
"epoch": 0.558,
|
| 3912 |
+
"grad_norm": 0.015258446335792542,
|
| 3913 |
+
"learning_rate": 2.2100000000000002e-05,
|
| 3914 |
+
"loss": 0.0015,
|
| 3915 |
+
"step": 5580
|
| 3916 |
+
},
|
| 3917 |
+
{
|
| 3918 |
+
"epoch": 0.559,
|
| 3919 |
+
"grad_norm": 0.009015046060085297,
|
| 3920 |
+
"learning_rate": 2.205e-05,
|
| 3921 |
+
"loss": 0.0012,
|
| 3922 |
+
"step": 5590
|
| 3923 |
+
},
|
| 3924 |
+
{
|
| 3925 |
+
"epoch": 0.56,
|
| 3926 |
+
"grad_norm": 0.011163819581270218,
|
| 3927 |
+
"learning_rate": 2.2000000000000003e-05,
|
| 3928 |
+
"loss": 0.0012,
|
| 3929 |
+
"step": 5600
|
| 3930 |
+
},
|
| 3931 |
+
{
|
| 3932 |
+
"epoch": 0.561,
|
| 3933 |
+
"grad_norm": 0.016389524564146996,
|
| 3934 |
+
"learning_rate": 2.195e-05,
|
| 3935 |
+
"loss": 0.0016,
|
| 3936 |
+
"step": 5610
|
| 3937 |
+
},
|
| 3938 |
+
{
|
| 3939 |
+
"epoch": 0.562,
|
| 3940 |
+
"grad_norm": 0.01325678639113903,
|
| 3941 |
+
"learning_rate": 2.19e-05,
|
| 3942 |
+
"loss": 0.0013,
|
| 3943 |
+
"step": 5620
|
| 3944 |
+
},
|
| 3945 |
+
{
|
| 3946 |
+
"epoch": 0.563,
|
| 3947 |
+
"grad_norm": 0.017966121435165405,
|
| 3948 |
+
"learning_rate": 2.1850000000000003e-05,
|
| 3949 |
+
"loss": 0.0014,
|
| 3950 |
+
"step": 5630
|
| 3951 |
+
},
|
| 3952 |
+
{
|
| 3953 |
+
"epoch": 0.564,
|
| 3954 |
+
"grad_norm": 0.012039076536893845,
|
| 3955 |
+
"learning_rate": 2.18e-05,
|
| 3956 |
+
"loss": 0.0013,
|
| 3957 |
+
"step": 5640
|
| 3958 |
+
},
|
| 3959 |
+
{
|
| 3960 |
+
"epoch": 0.565,
|
| 3961 |
+
"grad_norm": 0.006665175314992666,
|
| 3962 |
+
"learning_rate": 2.175e-05,
|
| 3963 |
+
"loss": 0.0012,
|
| 3964 |
+
"step": 5650
|
| 3965 |
+
},
|
| 3966 |
+
{
|
| 3967 |
+
"epoch": 0.566,
|
| 3968 |
+
"grad_norm": 0.0105441864579916,
|
| 3969 |
+
"learning_rate": 2.1700000000000002e-05,
|
| 3970 |
+
"loss": 0.0014,
|
| 3971 |
+
"step": 5660
|
| 3972 |
+
},
|
| 3973 |
+
{
|
| 3974 |
+
"epoch": 0.567,
|
| 3975 |
+
"grad_norm": 0.007554101757705212,
|
| 3976 |
+
"learning_rate": 2.165e-05,
|
| 3977 |
+
"loss": 0.0011,
|
| 3978 |
+
"step": 5670
|
| 3979 |
+
},
|
| 3980 |
+
{
|
| 3981 |
+
"epoch": 0.568,
|
| 3982 |
+
"grad_norm": 0.009823901578783989,
|
| 3983 |
+
"learning_rate": 2.16e-05,
|
| 3984 |
+
"loss": 0.0013,
|
| 3985 |
+
"step": 5680
|
| 3986 |
+
},
|
| 3987 |
+
{
|
| 3988 |
+
"epoch": 0.569,
|
| 3989 |
+
"grad_norm": 0.01720455475151539,
|
| 3990 |
+
"learning_rate": 2.1550000000000002e-05,
|
| 3991 |
+
"loss": 0.0015,
|
| 3992 |
+
"step": 5690
|
| 3993 |
+
},
|
| 3994 |
+
{
|
| 3995 |
+
"epoch": 0.57,
|
| 3996 |
+
"grad_norm": 0.01107338909059763,
|
| 3997 |
+
"learning_rate": 2.15e-05,
|
| 3998 |
+
"loss": 0.0012,
|
| 3999 |
+
"step": 5700
|
| 4000 |
+
},
|
| 4001 |
+
{
|
| 4002 |
+
"epoch": 0.571,
|
| 4003 |
+
"grad_norm": 0.01756761223077774,
|
| 4004 |
+
"learning_rate": 2.145e-05,
|
| 4005 |
+
"loss": 0.0014,
|
| 4006 |
+
"step": 5710
|
| 4007 |
+
},
|
| 4008 |
+
{
|
| 4009 |
+
"epoch": 0.572,
|
| 4010 |
+
"grad_norm": 0.022118983790278435,
|
| 4011 |
+
"learning_rate": 2.1400000000000002e-05,
|
| 4012 |
+
"loss": 0.0015,
|
| 4013 |
+
"step": 5720
|
| 4014 |
+
},
|
| 4015 |
+
{
|
| 4016 |
+
"epoch": 0.573,
|
| 4017 |
+
"grad_norm": 0.01616830937564373,
|
| 4018 |
+
"learning_rate": 2.135e-05,
|
| 4019 |
+
"loss": 0.0014,
|
| 4020 |
+
"step": 5730
|
| 4021 |
+
},
|
| 4022 |
+
{
|
| 4023 |
+
"epoch": 0.574,
|
| 4024 |
+
"grad_norm": 0.020481310784816742,
|
| 4025 |
+
"learning_rate": 2.13e-05,
|
| 4026 |
+
"loss": 0.0023,
|
| 4027 |
+
"step": 5740
|
| 4028 |
+
},
|
| 4029 |
+
{
|
| 4030 |
+
"epoch": 0.575,
|
| 4031 |
+
"grad_norm": 0.018176857382059097,
|
| 4032 |
+
"learning_rate": 2.125e-05,
|
| 4033 |
+
"loss": 0.0015,
|
| 4034 |
+
"step": 5750
|
| 4035 |
+
},
|
| 4036 |
+
{
|
| 4037 |
+
"epoch": 0.576,
|
| 4038 |
+
"grad_norm": 0.011317101307213306,
|
| 4039 |
+
"learning_rate": 2.12e-05,
|
| 4040 |
+
"loss": 0.0012,
|
| 4041 |
+
"step": 5760
|
| 4042 |
+
},
|
| 4043 |
+
{
|
| 4044 |
+
"epoch": 0.577,
|
| 4045 |
+
"grad_norm": 0.028791502118110657,
|
| 4046 |
+
"learning_rate": 2.115e-05,
|
| 4047 |
+
"loss": 0.0014,
|
| 4048 |
+
"step": 5770
|
| 4049 |
+
},
|
| 4050 |
+
{
|
| 4051 |
+
"epoch": 0.578,
|
| 4052 |
+
"grad_norm": 0.013037024065852165,
|
| 4053 |
+
"learning_rate": 2.11e-05,
|
| 4054 |
+
"loss": 0.0013,
|
| 4055 |
+
"step": 5780
|
| 4056 |
+
},
|
| 4057 |
+
{
|
| 4058 |
+
"epoch": 0.579,
|
| 4059 |
+
"grad_norm": 0.021426070481538773,
|
| 4060 |
+
"learning_rate": 2.105e-05,
|
| 4061 |
+
"loss": 0.0015,
|
| 4062 |
+
"step": 5790
|
| 4063 |
+
},
|
| 4064 |
+
{
|
| 4065 |
+
"epoch": 0.58,
|
| 4066 |
+
"grad_norm": 0.012033521197736263,
|
| 4067 |
+
"learning_rate": 2.1e-05,
|
| 4068 |
+
"loss": 0.0011,
|
| 4069 |
+
"step": 5800
|
| 4070 |
+
},
|
| 4071 |
+
{
|
| 4072 |
+
"epoch": 0.581,
|
| 4073 |
+
"grad_norm": 0.014337443746626377,
|
| 4074 |
+
"learning_rate": 2.095e-05,
|
| 4075 |
+
"loss": 0.0012,
|
| 4076 |
+
"step": 5810
|
| 4077 |
+
},
|
| 4078 |
+
{
|
| 4079 |
+
"epoch": 0.582,
|
| 4080 |
+
"grad_norm": 0.008603113703429699,
|
| 4081 |
+
"learning_rate": 2.09e-05,
|
| 4082 |
+
"loss": 0.0011,
|
| 4083 |
+
"step": 5820
|
| 4084 |
+
},
|
| 4085 |
+
{
|
| 4086 |
+
"epoch": 0.583,
|
| 4087 |
+
"grad_norm": 0.025418557226657867,
|
| 4088 |
+
"learning_rate": 2.085e-05,
|
| 4089 |
+
"loss": 0.0014,
|
| 4090 |
+
"step": 5830
|
| 4091 |
+
},
|
| 4092 |
+
{
|
| 4093 |
+
"epoch": 0.584,
|
| 4094 |
+
"grad_norm": 0.008621426299214363,
|
| 4095 |
+
"learning_rate": 2.08e-05,
|
| 4096 |
+
"loss": 0.0011,
|
| 4097 |
+
"step": 5840
|
| 4098 |
+
},
|
| 4099 |
+
{
|
| 4100 |
+
"epoch": 0.585,
|
| 4101 |
+
"grad_norm": 0.009969389997422695,
|
| 4102 |
+
"learning_rate": 2.075e-05,
|
| 4103 |
+
"loss": 0.0015,
|
| 4104 |
+
"step": 5850
|
| 4105 |
+
},
|
| 4106 |
+
{
|
| 4107 |
+
"epoch": 0.586,
|
| 4108 |
+
"grad_norm": 0.00997992418706417,
|
| 4109 |
+
"learning_rate": 2.07e-05,
|
| 4110 |
+
"loss": 0.0011,
|
| 4111 |
+
"step": 5860
|
| 4112 |
+
},
|
| 4113 |
+
{
|
| 4114 |
+
"epoch": 0.587,
|
| 4115 |
+
"grad_norm": 0.019949181005358696,
|
| 4116 |
+
"learning_rate": 2.065e-05,
|
| 4117 |
+
"loss": 0.001,
|
| 4118 |
+
"step": 5870
|
| 4119 |
+
},
|
| 4120 |
+
{
|
| 4121 |
+
"epoch": 0.588,
|
| 4122 |
+
"grad_norm": 0.009619793854653835,
|
| 4123 |
+
"learning_rate": 2.06e-05,
|
| 4124 |
+
"loss": 0.0011,
|
| 4125 |
+
"step": 5880
|
| 4126 |
+
},
|
| 4127 |
+
{
|
| 4128 |
+
"epoch": 0.589,
|
| 4129 |
+
"grad_norm": 0.007747489493340254,
|
| 4130 |
+
"learning_rate": 2.055e-05,
|
| 4131 |
+
"loss": 0.0012,
|
| 4132 |
+
"step": 5890
|
| 4133 |
+
},
|
| 4134 |
+
{
|
| 4135 |
+
"epoch": 0.59,
|
| 4136 |
+
"grad_norm": 0.01052554789930582,
|
| 4137 |
+
"learning_rate": 2.05e-05,
|
| 4138 |
+
"loss": 0.0014,
|
| 4139 |
+
"step": 5900
|
| 4140 |
+
},
|
| 4141 |
+
{
|
| 4142 |
+
"epoch": 0.591,
|
| 4143 |
+
"grad_norm": 0.014904200099408627,
|
| 4144 |
+
"learning_rate": 2.045e-05,
|
| 4145 |
+
"loss": 0.0012,
|
| 4146 |
+
"step": 5910
|
| 4147 |
+
},
|
| 4148 |
+
{
|
| 4149 |
+
"epoch": 0.592,
|
| 4150 |
+
"grad_norm": 0.00679561635479331,
|
| 4151 |
+
"learning_rate": 2.04e-05,
|
| 4152 |
+
"loss": 0.0011,
|
| 4153 |
+
"step": 5920
|
| 4154 |
+
},
|
| 4155 |
+
{
|
| 4156 |
+
"epoch": 0.593,
|
| 4157 |
+
"grad_norm": 0.006072670221328735,
|
| 4158 |
+
"learning_rate": 2.035e-05,
|
| 4159 |
+
"loss": 0.0011,
|
| 4160 |
+
"step": 5930
|
| 4161 |
+
},
|
| 4162 |
+
{
|
| 4163 |
+
"epoch": 0.594,
|
| 4164 |
+
"grad_norm": 0.014733157120645046,
|
| 4165 |
+
"learning_rate": 2.0300000000000002e-05,
|
| 4166 |
+
"loss": 0.0011,
|
| 4167 |
+
"step": 5940
|
| 4168 |
+
},
|
| 4169 |
+
{
|
| 4170 |
+
"epoch": 0.595,
|
| 4171 |
+
"grad_norm": 0.015511419624090195,
|
| 4172 |
+
"learning_rate": 2.025e-05,
|
| 4173 |
+
"loss": 0.0016,
|
| 4174 |
+
"step": 5950
|
| 4175 |
+
},
|
| 4176 |
+
{
|
| 4177 |
+
"epoch": 0.596,
|
| 4178 |
+
"grad_norm": 0.010620438493788242,
|
| 4179 |
+
"learning_rate": 2.0200000000000003e-05,
|
| 4180 |
+
"loss": 0.0012,
|
| 4181 |
+
"step": 5960
|
| 4182 |
+
},
|
| 4183 |
+
{
|
| 4184 |
+
"epoch": 0.597,
|
| 4185 |
+
"grad_norm": 0.0075794099830091,
|
| 4186 |
+
"learning_rate": 2.0150000000000002e-05,
|
| 4187 |
+
"loss": 0.0011,
|
| 4188 |
+
"step": 5970
|
| 4189 |
+
},
|
| 4190 |
+
{
|
| 4191 |
+
"epoch": 0.598,
|
| 4192 |
+
"grad_norm": 0.007882976904511452,
|
| 4193 |
+
"learning_rate": 2.01e-05,
|
| 4194 |
+
"loss": 0.0011,
|
| 4195 |
+
"step": 5980
|
| 4196 |
+
},
|
| 4197 |
+
{
|
| 4198 |
+
"epoch": 0.599,
|
| 4199 |
+
"grad_norm": 0.011548763141036034,
|
| 4200 |
+
"learning_rate": 2.0050000000000003e-05,
|
| 4201 |
+
"loss": 0.0013,
|
| 4202 |
+
"step": 5990
|
| 4203 |
+
},
|
| 4204 |
+
{
|
| 4205 |
+
"epoch": 0.6,
|
| 4206 |
+
"grad_norm": 0.0084703853353858,
|
| 4207 |
+
"learning_rate": 2e-05,
|
| 4208 |
+
"loss": 0.0011,
|
| 4209 |
+
"step": 6000
|
| 4210 |
+
},
|
| 4211 |
+
{
|
| 4212 |
+
"epoch": 0.601,
|
| 4213 |
+
"grad_norm": 0.007603704463690519,
|
| 4214 |
+
"learning_rate": 1.995e-05,
|
| 4215 |
+
"loss": 0.001,
|
| 4216 |
+
"step": 6010
|
| 4217 |
+
},
|
| 4218 |
+
{
|
| 4219 |
+
"epoch": 0.602,
|
| 4220 |
+
"grad_norm": 0.008562711998820305,
|
| 4221 |
+
"learning_rate": 1.9900000000000003e-05,
|
| 4222 |
+
"loss": 0.0012,
|
| 4223 |
+
"step": 6020
|
| 4224 |
+
},
|
| 4225 |
+
{
|
| 4226 |
+
"epoch": 0.603,
|
| 4227 |
+
"grad_norm": 0.007590813562273979,
|
| 4228 |
+
"learning_rate": 1.985e-05,
|
| 4229 |
+
"loss": 0.001,
|
| 4230 |
+
"step": 6030
|
| 4231 |
+
},
|
| 4232 |
+
{
|
| 4233 |
+
"epoch": 0.604,
|
| 4234 |
+
"grad_norm": 0.020342741161584854,
|
| 4235 |
+
"learning_rate": 1.9800000000000004e-05,
|
| 4236 |
+
"loss": 0.0017,
|
| 4237 |
+
"step": 6040
|
| 4238 |
+
},
|
| 4239 |
+
{
|
| 4240 |
+
"epoch": 0.605,
|
| 4241 |
+
"grad_norm": 0.16912633180618286,
|
| 4242 |
+
"learning_rate": 1.9750000000000002e-05,
|
| 4243 |
+
"loss": 0.0089,
|
| 4244 |
+
"step": 6050
|
| 4245 |
+
},
|
| 4246 |
+
{
|
| 4247 |
+
"epoch": 0.606,
|
| 4248 |
+
"grad_norm": 0.08793429285287857,
|
| 4249 |
+
"learning_rate": 1.97e-05,
|
| 4250 |
+
"loss": 0.0027,
|
| 4251 |
+
"step": 6060
|
| 4252 |
+
},
|
| 4253 |
+
{
|
| 4254 |
+
"epoch": 0.607,
|
| 4255 |
+
"grad_norm": 0.05196760594844818,
|
| 4256 |
+
"learning_rate": 1.9650000000000003e-05,
|
| 4257 |
+
"loss": 0.0022,
|
| 4258 |
+
"step": 6070
|
| 4259 |
+
},
|
| 4260 |
+
{
|
| 4261 |
+
"epoch": 0.608,
|
| 4262 |
+
"grad_norm": 0.02118327096104622,
|
| 4263 |
+
"learning_rate": 1.9600000000000002e-05,
|
| 4264 |
+
"loss": 0.0021,
|
| 4265 |
+
"step": 6080
|
| 4266 |
+
},
|
| 4267 |
+
{
|
| 4268 |
+
"epoch": 0.609,
|
| 4269 |
+
"grad_norm": 0.013289586640894413,
|
| 4270 |
+
"learning_rate": 1.955e-05,
|
| 4271 |
+
"loss": 0.0013,
|
| 4272 |
+
"step": 6090
|
| 4273 |
+
},
|
| 4274 |
+
{
|
| 4275 |
+
"epoch": 0.61,
|
| 4276 |
+
"grad_norm": 0.012911707162857056,
|
| 4277 |
+
"learning_rate": 1.9500000000000003e-05,
|
| 4278 |
+
"loss": 0.0013,
|
| 4279 |
+
"step": 6100
|
| 4280 |
+
},
|
| 4281 |
+
{
|
| 4282 |
+
"epoch": 0.611,
|
| 4283 |
+
"grad_norm": 0.018663186579942703,
|
| 4284 |
+
"learning_rate": 1.9450000000000002e-05,
|
| 4285 |
+
"loss": 0.0012,
|
| 4286 |
+
"step": 6110
|
| 4287 |
+
},
|
| 4288 |
+
{
|
| 4289 |
+
"epoch": 0.612,
|
| 4290 |
+
"grad_norm": 0.010551884770393372,
|
| 4291 |
+
"learning_rate": 1.94e-05,
|
| 4292 |
+
"loss": 0.0012,
|
| 4293 |
+
"step": 6120
|
| 4294 |
+
},
|
| 4295 |
+
{
|
| 4296 |
+
"epoch": 0.613,
|
| 4297 |
+
"grad_norm": 0.015853077173233032,
|
| 4298 |
+
"learning_rate": 1.9350000000000003e-05,
|
| 4299 |
+
"loss": 0.0013,
|
| 4300 |
+
"step": 6130
|
| 4301 |
+
},
|
| 4302 |
+
{
|
| 4303 |
+
"epoch": 0.614,
|
| 4304 |
+
"grad_norm": 0.020374910905957222,
|
| 4305 |
+
"learning_rate": 1.93e-05,
|
| 4306 |
+
"loss": 0.001,
|
| 4307 |
+
"step": 6140
|
| 4308 |
+
},
|
| 4309 |
+
{
|
| 4310 |
+
"epoch": 0.615,
|
| 4311 |
+
"grad_norm": 0.015159848146140575,
|
| 4312 |
+
"learning_rate": 1.925e-05,
|
| 4313 |
+
"loss": 0.0013,
|
| 4314 |
+
"step": 6150
|
| 4315 |
+
},
|
| 4316 |
+
{
|
| 4317 |
+
"epoch": 0.616,
|
| 4318 |
+
"grad_norm": 0.007991676218807697,
|
| 4319 |
+
"learning_rate": 1.9200000000000003e-05,
|
| 4320 |
+
"loss": 0.0013,
|
| 4321 |
+
"step": 6160
|
| 4322 |
+
},
|
| 4323 |
+
{
|
| 4324 |
+
"epoch": 0.617,
|
| 4325 |
+
"grad_norm": 0.007849587127566338,
|
| 4326 |
+
"learning_rate": 1.915e-05,
|
| 4327 |
+
"loss": 0.0011,
|
| 4328 |
+
"step": 6170
|
| 4329 |
+
},
|
| 4330 |
+
{
|
| 4331 |
+
"epoch": 0.618,
|
| 4332 |
+
"grad_norm": 0.022048622369766235,
|
| 4333 |
+
"learning_rate": 1.91e-05,
|
| 4334 |
+
"loss": 0.001,
|
| 4335 |
+
"step": 6180
|
| 4336 |
+
},
|
| 4337 |
+
{
|
| 4338 |
+
"epoch": 0.619,
|
| 4339 |
+
"grad_norm": 0.021215343847870827,
|
| 4340 |
+
"learning_rate": 1.9050000000000002e-05,
|
| 4341 |
+
"loss": 0.0011,
|
| 4342 |
+
"step": 6190
|
| 4343 |
+
},
|
| 4344 |
+
{
|
| 4345 |
+
"epoch": 0.62,
|
| 4346 |
+
"grad_norm": 0.012288344092667103,
|
| 4347 |
+
"learning_rate": 1.9e-05,
|
| 4348 |
+
"loss": 0.0012,
|
| 4349 |
+
"step": 6200
|
| 4350 |
+
},
|
| 4351 |
+
{
|
| 4352 |
+
"epoch": 0.621,
|
| 4353 |
+
"grad_norm": 0.020313331857323647,
|
| 4354 |
+
"learning_rate": 1.895e-05,
|
| 4355 |
+
"loss": 0.0011,
|
| 4356 |
+
"step": 6210
|
| 4357 |
+
},
|
| 4358 |
+
{
|
| 4359 |
+
"epoch": 0.622,
|
| 4360 |
+
"grad_norm": 0.008762447163462639,
|
| 4361 |
+
"learning_rate": 1.8900000000000002e-05,
|
| 4362 |
+
"loss": 0.001,
|
| 4363 |
+
"step": 6220
|
| 4364 |
+
},
|
| 4365 |
+
{
|
| 4366 |
+
"epoch": 0.623,
|
| 4367 |
+
"grad_norm": 0.0247616209089756,
|
| 4368 |
+
"learning_rate": 1.885e-05,
|
| 4369 |
+
"loss": 0.0011,
|
| 4370 |
+
"step": 6230
|
| 4371 |
+
},
|
| 4372 |
+
{
|
| 4373 |
+
"epoch": 0.624,
|
| 4374 |
+
"grad_norm": 0.09021363407373428,
|
| 4375 |
+
"learning_rate": 1.88e-05,
|
| 4376 |
+
"loss": 0.0016,
|
| 4377 |
+
"step": 6240
|
| 4378 |
+
},
|
| 4379 |
+
{
|
| 4380 |
+
"epoch": 0.625,
|
| 4381 |
+
"grad_norm": 0.017945896834135056,
|
| 4382 |
+
"learning_rate": 1.8750000000000002e-05,
|
| 4383 |
+
"loss": 0.0011,
|
| 4384 |
+
"step": 6250
|
| 4385 |
+
},
|
| 4386 |
+
{
|
| 4387 |
+
"epoch": 0.626,
|
| 4388 |
+
"grad_norm": 0.011303462088108063,
|
| 4389 |
+
"learning_rate": 1.87e-05,
|
| 4390 |
+
"loss": 0.0011,
|
| 4391 |
+
"step": 6260
|
| 4392 |
+
},
|
| 4393 |
+
{
|
| 4394 |
+
"epoch": 0.627,
|
| 4395 |
+
"grad_norm": 0.008381664752960205,
|
| 4396 |
+
"learning_rate": 1.865e-05,
|
| 4397 |
+
"loss": 0.0011,
|
| 4398 |
+
"step": 6270
|
| 4399 |
+
},
|
| 4400 |
+
{
|
| 4401 |
+
"epoch": 0.628,
|
| 4402 |
+
"grad_norm": 0.011003987863659859,
|
| 4403 |
+
"learning_rate": 1.86e-05,
|
| 4404 |
+
"loss": 0.0012,
|
| 4405 |
+
"step": 6280
|
| 4406 |
+
},
|
| 4407 |
+
{
|
| 4408 |
+
"epoch": 0.629,
|
| 4409 |
+
"grad_norm": 0.015965888276696205,
|
| 4410 |
+
"learning_rate": 1.855e-05,
|
| 4411 |
+
"loss": 0.001,
|
| 4412 |
+
"step": 6290
|
| 4413 |
+
},
|
| 4414 |
+
{
|
| 4415 |
+
"epoch": 0.63,
|
| 4416 |
+
"grad_norm": 0.006507181562483311,
|
| 4417 |
+
"learning_rate": 1.85e-05,
|
| 4418 |
+
"loss": 0.0009,
|
| 4419 |
+
"step": 6300
|
| 4420 |
+
},
|
| 4421 |
+
{
|
| 4422 |
+
"epoch": 0.631,
|
| 4423 |
+
"grad_norm": 0.015577591024339199,
|
| 4424 |
+
"learning_rate": 1.845e-05,
|
| 4425 |
+
"loss": 0.001,
|
| 4426 |
+
"step": 6310
|
| 4427 |
+
},
|
| 4428 |
+
{
|
| 4429 |
+
"epoch": 0.632,
|
| 4430 |
+
"grad_norm": 0.006741558667272329,
|
| 4431 |
+
"learning_rate": 1.84e-05,
|
| 4432 |
+
"loss": 0.0011,
|
| 4433 |
+
"step": 6320
|
| 4434 |
+
},
|
| 4435 |
+
{
|
| 4436 |
+
"epoch": 0.633,
|
| 4437 |
+
"grad_norm": 0.016030525788664818,
|
| 4438 |
+
"learning_rate": 1.8350000000000002e-05,
|
| 4439 |
+
"loss": 0.001,
|
| 4440 |
+
"step": 6330
|
| 4441 |
+
},
|
| 4442 |
+
{
|
| 4443 |
+
"epoch": 0.634,
|
| 4444 |
+
"grad_norm": 0.010763168334960938,
|
| 4445 |
+
"learning_rate": 1.83e-05,
|
| 4446 |
+
"loss": 0.0011,
|
| 4447 |
+
"step": 6340
|
| 4448 |
+
},
|
| 4449 |
+
{
|
| 4450 |
+
"epoch": 0.635,
|
| 4451 |
+
"grad_norm": 0.017273874953389168,
|
| 4452 |
+
"learning_rate": 1.825e-05,
|
| 4453 |
+
"loss": 0.001,
|
| 4454 |
+
"step": 6350
|
| 4455 |
+
},
|
| 4456 |
+
{
|
| 4457 |
+
"epoch": 0.636,
|
| 4458 |
+
"grad_norm": 0.010964670218527317,
|
| 4459 |
+
"learning_rate": 1.8200000000000002e-05,
|
| 4460 |
+
"loss": 0.0011,
|
| 4461 |
+
"step": 6360
|
| 4462 |
+
},
|
| 4463 |
+
{
|
| 4464 |
+
"epoch": 0.637,
|
| 4465 |
+
"grad_norm": 0.00803497713059187,
|
| 4466 |
+
"learning_rate": 1.815e-05,
|
| 4467 |
+
"loss": 0.0009,
|
| 4468 |
+
"step": 6370
|
| 4469 |
+
},
|
| 4470 |
+
{
|
| 4471 |
+
"epoch": 0.638,
|
| 4472 |
+
"grad_norm": 0.007479315157979727,
|
| 4473 |
+
"learning_rate": 1.81e-05,
|
| 4474 |
+
"loss": 0.0014,
|
| 4475 |
+
"step": 6380
|
| 4476 |
+
},
|
| 4477 |
+
{
|
| 4478 |
+
"epoch": 0.639,
|
| 4479 |
+
"grad_norm": 0.010598058812320232,
|
| 4480 |
+
"learning_rate": 1.805e-05,
|
| 4481 |
+
"loss": 0.001,
|
| 4482 |
+
"step": 6390
|
| 4483 |
+
},
|
| 4484 |
+
{
|
| 4485 |
+
"epoch": 0.64,
|
| 4486 |
+
"grad_norm": 0.009770036675035954,
|
| 4487 |
+
"learning_rate": 1.8e-05,
|
| 4488 |
+
"loss": 0.0009,
|
| 4489 |
+
"step": 6400
|
| 4490 |
+
},
|
| 4491 |
+
{
|
| 4492 |
+
"epoch": 0.641,
|
| 4493 |
+
"grad_norm": 0.011602561920881271,
|
| 4494 |
+
"learning_rate": 1.795e-05,
|
| 4495 |
+
"loss": 0.0008,
|
| 4496 |
+
"step": 6410
|
| 4497 |
+
},
|
| 4498 |
+
{
|
| 4499 |
+
"epoch": 0.642,
|
| 4500 |
+
"grad_norm": 0.0076597342267632484,
|
| 4501 |
+
"learning_rate": 1.79e-05,
|
| 4502 |
+
"loss": 0.0008,
|
| 4503 |
+
"step": 6420
|
| 4504 |
+
},
|
| 4505 |
+
{
|
| 4506 |
+
"epoch": 0.643,
|
| 4507 |
+
"grad_norm": 0.012248953804373741,
|
| 4508 |
+
"learning_rate": 1.785e-05,
|
| 4509 |
+
"loss": 0.0008,
|
| 4510 |
+
"step": 6430
|
| 4511 |
+
},
|
| 4512 |
+
{
|
| 4513 |
+
"epoch": 0.644,
|
| 4514 |
+
"grad_norm": 0.005626557394862175,
|
| 4515 |
+
"learning_rate": 1.78e-05,
|
| 4516 |
+
"loss": 0.0008,
|
| 4517 |
+
"step": 6440
|
| 4518 |
+
},
|
| 4519 |
+
{
|
| 4520 |
+
"epoch": 0.645,
|
| 4521 |
+
"grad_norm": 0.005482000298798084,
|
| 4522 |
+
"learning_rate": 1.775e-05,
|
| 4523 |
+
"loss": 0.0008,
|
| 4524 |
+
"step": 6450
|
| 4525 |
+
},
|
| 4526 |
+
{
|
| 4527 |
+
"epoch": 0.646,
|
| 4528 |
+
"grad_norm": 0.007456011138856411,
|
| 4529 |
+
"learning_rate": 1.77e-05,
|
| 4530 |
+
"loss": 0.0008,
|
| 4531 |
+
"step": 6460
|
| 4532 |
+
},
|
| 4533 |
+
{
|
| 4534 |
+
"epoch": 0.647,
|
| 4535 |
+
"grad_norm": 0.008909308351576328,
|
| 4536 |
+
"learning_rate": 1.765e-05,
|
| 4537 |
+
"loss": 0.0008,
|
| 4538 |
+
"step": 6470
|
| 4539 |
+
},
|
| 4540 |
+
{
|
| 4541 |
+
"epoch": 0.648,
|
| 4542 |
+
"grad_norm": 0.011135280132293701,
|
| 4543 |
+
"learning_rate": 1.76e-05,
|
| 4544 |
+
"loss": 0.0009,
|
| 4545 |
+
"step": 6480
|
| 4546 |
+
},
|
| 4547 |
+
{
|
| 4548 |
+
"epoch": 0.649,
|
| 4549 |
+
"grad_norm": 0.01595783233642578,
|
| 4550 |
+
"learning_rate": 1.755e-05,
|
| 4551 |
+
"loss": 0.001,
|
| 4552 |
+
"step": 6490
|
| 4553 |
+
},
|
| 4554 |
+
{
|
| 4555 |
+
"epoch": 0.65,
|
| 4556 |
+
"grad_norm": 0.013902807608246803,
|
| 4557 |
+
"learning_rate": 1.75e-05,
|
| 4558 |
+
"loss": 0.0011,
|
| 4559 |
+
"step": 6500
|
| 4560 |
+
},
|
| 4561 |
+
{
|
| 4562 |
+
"epoch": 0.651,
|
| 4563 |
+
"grad_norm": 0.010244622826576233,
|
| 4564 |
+
"learning_rate": 1.745e-05,
|
| 4565 |
+
"loss": 0.0009,
|
| 4566 |
+
"step": 6510
|
| 4567 |
+
},
|
| 4568 |
+
{
|
| 4569 |
+
"epoch": 0.652,
|
| 4570 |
+
"grad_norm": 0.007476091384887695,
|
| 4571 |
+
"learning_rate": 1.74e-05,
|
| 4572 |
+
"loss": 0.0009,
|
| 4573 |
+
"step": 6520
|
| 4574 |
+
},
|
| 4575 |
+
{
|
| 4576 |
+
"epoch": 0.653,
|
| 4577 |
+
"grad_norm": 0.013044660910964012,
|
| 4578 |
+
"learning_rate": 1.7349999999999998e-05,
|
| 4579 |
+
"loss": 0.0009,
|
| 4580 |
+
"step": 6530
|
| 4581 |
+
},
|
| 4582 |
+
{
|
| 4583 |
+
"epoch": 0.654,
|
| 4584 |
+
"grad_norm": 0.004804369527846575,
|
| 4585 |
+
"learning_rate": 1.73e-05,
|
| 4586 |
+
"loss": 0.0009,
|
| 4587 |
+
"step": 6540
|
| 4588 |
+
},
|
| 4589 |
+
{
|
| 4590 |
+
"epoch": 0.655,
|
| 4591 |
+
"grad_norm": 0.006042002234607935,
|
| 4592 |
+
"learning_rate": 1.725e-05,
|
| 4593 |
+
"loss": 0.0008,
|
| 4594 |
+
"step": 6550
|
| 4595 |
+
},
|
| 4596 |
+
{
|
| 4597 |
+
"epoch": 0.656,
|
| 4598 |
+
"grad_norm": 0.010785943828523159,
|
| 4599 |
+
"learning_rate": 1.7199999999999998e-05,
|
| 4600 |
+
"loss": 0.0009,
|
| 4601 |
+
"step": 6560
|
| 4602 |
+
},
|
| 4603 |
+
{
|
| 4604 |
+
"epoch": 0.657,
|
| 4605 |
+
"grad_norm": 0.011350172571837902,
|
| 4606 |
+
"learning_rate": 1.7150000000000004e-05,
|
| 4607 |
+
"loss": 0.0008,
|
| 4608 |
+
"step": 6570
|
| 4609 |
+
},
|
| 4610 |
+
{
|
| 4611 |
+
"epoch": 0.658,
|
| 4612 |
+
"grad_norm": 0.007638021372258663,
|
| 4613 |
+
"learning_rate": 1.7100000000000002e-05,
|
| 4614 |
+
"loss": 0.0009,
|
| 4615 |
+
"step": 6580
|
| 4616 |
+
},
|
| 4617 |
+
{
|
| 4618 |
+
"epoch": 0.659,
|
| 4619 |
+
"grad_norm": 0.005735939834266901,
|
| 4620 |
+
"learning_rate": 1.705e-05,
|
| 4621 |
+
"loss": 0.0009,
|
| 4622 |
+
"step": 6590
|
| 4623 |
+
},
|
| 4624 |
+
{
|
| 4625 |
+
"epoch": 0.66,
|
| 4626 |
+
"grad_norm": 0.02717960625886917,
|
| 4627 |
+
"learning_rate": 1.7000000000000003e-05,
|
| 4628 |
+
"loss": 0.0011,
|
| 4629 |
+
"step": 6600
|
| 4630 |
+
},
|
| 4631 |
+
{
|
| 4632 |
+
"epoch": 0.661,
|
| 4633 |
+
"grad_norm": 0.006012643221765757,
|
| 4634 |
+
"learning_rate": 1.6950000000000002e-05,
|
| 4635 |
+
"loss": 0.0008,
|
| 4636 |
+
"step": 6610
|
| 4637 |
+
},
|
| 4638 |
+
{
|
| 4639 |
+
"epoch": 0.662,
|
| 4640 |
+
"grad_norm": 0.00599683728069067,
|
| 4641 |
+
"learning_rate": 1.69e-05,
|
| 4642 |
+
"loss": 0.0008,
|
| 4643 |
+
"step": 6620
|
| 4644 |
+
},
|
| 4645 |
+
{
|
| 4646 |
+
"epoch": 0.663,
|
| 4647 |
+
"grad_norm": 0.026952974498271942,
|
| 4648 |
+
"learning_rate": 1.6850000000000003e-05,
|
| 4649 |
+
"loss": 0.0008,
|
| 4650 |
+
"step": 6630
|
| 4651 |
+
},
|
| 4652 |
+
{
|
| 4653 |
+
"epoch": 0.664,
|
| 4654 |
+
"grad_norm": 0.008171536959707737,
|
| 4655 |
+
"learning_rate": 1.6800000000000002e-05,
|
| 4656 |
+
"loss": 0.0008,
|
| 4657 |
+
"step": 6640
|
| 4658 |
+
},
|
| 4659 |
+
{
|
| 4660 |
+
"epoch": 0.665,
|
| 4661 |
+
"grad_norm": 0.007446442265063524,
|
| 4662 |
+
"learning_rate": 1.675e-05,
|
| 4663 |
+
"loss": 0.0009,
|
| 4664 |
+
"step": 6650
|
| 4665 |
+
},
|
| 4666 |
+
{
|
| 4667 |
+
"epoch": 0.666,
|
| 4668 |
+
"grad_norm": 0.006456063129007816,
|
| 4669 |
+
"learning_rate": 1.6700000000000003e-05,
|
| 4670 |
+
"loss": 0.0008,
|
| 4671 |
+
"step": 6660
|
| 4672 |
+
},
|
| 4673 |
+
{
|
| 4674 |
+
"epoch": 0.667,
|
| 4675 |
+
"grad_norm": 0.008162173442542553,
|
| 4676 |
+
"learning_rate": 1.665e-05,
|
| 4677 |
+
"loss": 0.0007,
|
| 4678 |
+
"step": 6670
|
| 4679 |
+
},
|
| 4680 |
+
{
|
| 4681 |
+
"epoch": 0.668,
|
| 4682 |
+
"grad_norm": 0.004432919900864363,
|
| 4683 |
+
"learning_rate": 1.66e-05,
|
| 4684 |
+
"loss": 0.0008,
|
| 4685 |
+
"step": 6680
|
| 4686 |
+
},
|
| 4687 |
+
{
|
| 4688 |
+
"epoch": 0.669,
|
| 4689 |
+
"grad_norm": 0.007158307824283838,
|
| 4690 |
+
"learning_rate": 1.6550000000000002e-05,
|
| 4691 |
+
"loss": 0.0008,
|
| 4692 |
+
"step": 6690
|
| 4693 |
+
},
|
| 4694 |
+
{
|
| 4695 |
+
"epoch": 0.67,
|
| 4696 |
+
"grad_norm": 0.003983801696449518,
|
| 4697 |
+
"learning_rate": 1.65e-05,
|
| 4698 |
+
"loss": 0.0007,
|
| 4699 |
+
"step": 6700
|
| 4700 |
+
},
|
| 4701 |
+
{
|
| 4702 |
+
"epoch": 0.671,
|
| 4703 |
+
"grad_norm": 0.005170087795704603,
|
| 4704 |
+
"learning_rate": 1.645e-05,
|
| 4705 |
+
"loss": 0.0008,
|
| 4706 |
+
"step": 6710
|
| 4707 |
+
},
|
| 4708 |
+
{
|
| 4709 |
+
"epoch": 0.672,
|
| 4710 |
+
"grad_norm": 0.004729804117232561,
|
| 4711 |
+
"learning_rate": 1.6400000000000002e-05,
|
| 4712 |
+
"loss": 0.0008,
|
| 4713 |
+
"step": 6720
|
| 4714 |
+
},
|
| 4715 |
+
{
|
| 4716 |
+
"epoch": 0.673,
|
| 4717 |
+
"grad_norm": 0.010037174448370934,
|
| 4718 |
+
"learning_rate": 1.635e-05,
|
| 4719 |
+
"loss": 0.001,
|
| 4720 |
+
"step": 6730
|
| 4721 |
+
},
|
| 4722 |
+
{
|
| 4723 |
+
"epoch": 0.674,
|
| 4724 |
+
"grad_norm": 0.050949569791555405,
|
| 4725 |
+
"learning_rate": 1.63e-05,
|
| 4726 |
+
"loss": 0.0023,
|
| 4727 |
+
"step": 6740
|
| 4728 |
+
},
|
| 4729 |
+
{
|
| 4730 |
+
"epoch": 0.675,
|
| 4731 |
+
"grad_norm": 0.0323474146425724,
|
| 4732 |
+
"learning_rate": 1.6250000000000002e-05,
|
| 4733 |
+
"loss": 0.0017,
|
| 4734 |
+
"step": 6750
|
| 4735 |
+
},
|
| 4736 |
+
{
|
| 4737 |
+
"epoch": 0.676,
|
| 4738 |
+
"grad_norm": 0.027231359854340553,
|
| 4739 |
+
"learning_rate": 1.62e-05,
|
| 4740 |
+
"loss": 0.0021,
|
| 4741 |
+
"step": 6760
|
| 4742 |
+
},
|
| 4743 |
+
{
|
| 4744 |
+
"epoch": 0.677,
|
| 4745 |
+
"grad_norm": 0.01555855292826891,
|
| 4746 |
+
"learning_rate": 1.6150000000000003e-05,
|
| 4747 |
+
"loss": 0.0013,
|
| 4748 |
+
"step": 6770
|
| 4749 |
+
},
|
| 4750 |
+
{
|
| 4751 |
+
"epoch": 0.678,
|
| 4752 |
+
"grad_norm": 0.01804298162460327,
|
| 4753 |
+
"learning_rate": 1.6100000000000002e-05,
|
| 4754 |
+
"loss": 0.0011,
|
| 4755 |
+
"step": 6780
|
| 4756 |
+
},
|
| 4757 |
+
{
|
| 4758 |
+
"epoch": 0.679,
|
| 4759 |
+
"grad_norm": 0.011248771101236343,
|
| 4760 |
+
"learning_rate": 1.605e-05,
|
| 4761 |
+
"loss": 0.0011,
|
| 4762 |
+
"step": 6790
|
| 4763 |
+
},
|
| 4764 |
+
{
|
| 4765 |
+
"epoch": 0.68,
|
| 4766 |
+
"grad_norm": 0.007389044389128685,
|
| 4767 |
+
"learning_rate": 1.6000000000000003e-05,
|
| 4768 |
+
"loss": 0.0009,
|
| 4769 |
+
"step": 6800
|
| 4770 |
+
},
|
| 4771 |
+
{
|
| 4772 |
+
"epoch": 0.681,
|
| 4773 |
+
"grad_norm": 0.014606145210564137,
|
| 4774 |
+
"learning_rate": 1.595e-05,
|
| 4775 |
+
"loss": 0.0012,
|
| 4776 |
+
"step": 6810
|
| 4777 |
+
},
|
| 4778 |
+
{
|
| 4779 |
+
"epoch": 0.682,
|
| 4780 |
+
"grad_norm": 0.012476052157580853,
|
| 4781 |
+
"learning_rate": 1.59e-05,
|
| 4782 |
+
"loss": 0.0009,
|
| 4783 |
+
"step": 6820
|
| 4784 |
+
},
|
| 4785 |
+
{
|
| 4786 |
+
"epoch": 0.683,
|
| 4787 |
+
"grad_norm": 0.009272475726902485,
|
| 4788 |
+
"learning_rate": 1.5850000000000002e-05,
|
| 4789 |
+
"loss": 0.0009,
|
| 4790 |
+
"step": 6830
|
| 4791 |
+
},
|
| 4792 |
+
{
|
| 4793 |
+
"epoch": 0.684,
|
| 4794 |
+
"grad_norm": 0.011705187149345875,
|
| 4795 |
+
"learning_rate": 1.58e-05,
|
| 4796 |
+
"loss": 0.0009,
|
| 4797 |
+
"step": 6840
|
| 4798 |
+
},
|
| 4799 |
+
{
|
| 4800 |
+
"epoch": 0.685,
|
| 4801 |
+
"grad_norm": 0.01874556578695774,
|
| 4802 |
+
"learning_rate": 1.575e-05,
|
| 4803 |
+
"loss": 0.0011,
|
| 4804 |
+
"step": 6850
|
| 4805 |
+
},
|
| 4806 |
+
{
|
| 4807 |
+
"epoch": 0.686,
|
| 4808 |
+
"grad_norm": 0.01463324110955,
|
| 4809 |
+
"learning_rate": 1.5700000000000002e-05,
|
| 4810 |
+
"loss": 0.0009,
|
| 4811 |
+
"step": 6860
|
| 4812 |
+
},
|
| 4813 |
+
{
|
| 4814 |
+
"epoch": 0.687,
|
| 4815 |
+
"grad_norm": 0.012001392431557178,
|
| 4816 |
+
"learning_rate": 1.565e-05,
|
| 4817 |
+
"loss": 0.001,
|
| 4818 |
+
"step": 6870
|
| 4819 |
+
},
|
| 4820 |
+
{
|
| 4821 |
+
"epoch": 0.688,
|
| 4822 |
+
"grad_norm": 0.009366356767714024,
|
| 4823 |
+
"learning_rate": 1.56e-05,
|
| 4824 |
+
"loss": 0.0008,
|
| 4825 |
+
"step": 6880
|
| 4826 |
+
},
|
| 4827 |
+
{
|
| 4828 |
+
"epoch": 0.689,
|
| 4829 |
+
"grad_norm": 0.010064000263810158,
|
| 4830 |
+
"learning_rate": 1.5550000000000002e-05,
|
| 4831 |
+
"loss": 0.0009,
|
| 4832 |
+
"step": 6890
|
| 4833 |
+
},
|
| 4834 |
+
{
|
| 4835 |
+
"epoch": 0.69,
|
| 4836 |
+
"grad_norm": 0.016703909263014793,
|
| 4837 |
+
"learning_rate": 1.55e-05,
|
| 4838 |
+
"loss": 0.0009,
|
| 4839 |
+
"step": 6900
|
| 4840 |
+
},
|
| 4841 |
+
{
|
| 4842 |
+
"epoch": 0.691,
|
| 4843 |
+
"grad_norm": 0.0146669652312994,
|
| 4844 |
+
"learning_rate": 1.545e-05,
|
| 4845 |
+
"loss": 0.001,
|
| 4846 |
+
"step": 6910
|
| 4847 |
+
},
|
| 4848 |
+
{
|
| 4849 |
+
"epoch": 0.692,
|
| 4850 |
+
"grad_norm": 0.006643705535680056,
|
| 4851 |
+
"learning_rate": 1.54e-05,
|
| 4852 |
+
"loss": 0.0009,
|
| 4853 |
+
"step": 6920
|
| 4854 |
+
},
|
| 4855 |
+
{
|
| 4856 |
+
"epoch": 0.693,
|
| 4857 |
+
"grad_norm": 0.011501871049404144,
|
| 4858 |
+
"learning_rate": 1.535e-05,
|
| 4859 |
+
"loss": 0.0008,
|
| 4860 |
+
"step": 6930
|
| 4861 |
+
},
|
| 4862 |
+
{
|
| 4863 |
+
"epoch": 0.694,
|
| 4864 |
+
"grad_norm": 0.008170065470039845,
|
| 4865 |
+
"learning_rate": 1.53e-05,
|
| 4866 |
+
"loss": 0.0008,
|
| 4867 |
+
"step": 6940
|
| 4868 |
+
},
|
| 4869 |
+
{
|
| 4870 |
+
"epoch": 0.695,
|
| 4871 |
+
"grad_norm": 0.00737554719671607,
|
| 4872 |
+
"learning_rate": 1.525e-05,
|
| 4873 |
+
"loss": 0.0007,
|
| 4874 |
+
"step": 6950
|
| 4875 |
+
},
|
| 4876 |
+
{
|
| 4877 |
+
"epoch": 0.696,
|
| 4878 |
+
"grad_norm": 0.006846282631158829,
|
| 4879 |
+
"learning_rate": 1.52e-05,
|
| 4880 |
+
"loss": 0.0009,
|
| 4881 |
+
"step": 6960
|
| 4882 |
+
},
|
| 4883 |
+
{
|
| 4884 |
+
"epoch": 0.697,
|
| 4885 |
+
"grad_norm": 0.007784941233694553,
|
| 4886 |
+
"learning_rate": 1.515e-05,
|
| 4887 |
+
"loss": 0.0008,
|
| 4888 |
+
"step": 6970
|
| 4889 |
+
},
|
| 4890 |
+
{
|
| 4891 |
+
"epoch": 0.698,
|
| 4892 |
+
"grad_norm": 0.009864069521427155,
|
| 4893 |
+
"learning_rate": 1.51e-05,
|
| 4894 |
+
"loss": 0.0008,
|
| 4895 |
+
"step": 6980
|
| 4896 |
+
},
|
| 4897 |
+
{
|
| 4898 |
+
"epoch": 0.699,
|
| 4899 |
+
"grad_norm": 0.007372863125056028,
|
| 4900 |
+
"learning_rate": 1.505e-05,
|
| 4901 |
+
"loss": 0.0009,
|
| 4902 |
+
"step": 6990
|
| 4903 |
+
},
|
| 4904 |
+
{
|
| 4905 |
+
"epoch": 0.7,
|
| 4906 |
+
"grad_norm": 0.006507135462015867,
|
| 4907 |
+
"learning_rate": 1.5e-05,
|
| 4908 |
+
"loss": 0.0008,
|
| 4909 |
+
"step": 7000
|
| 4910 |
+
},
|
| 4911 |
+
{
|
| 4912 |
+
"epoch": 0.701,
|
| 4913 |
+
"grad_norm": 0.03093353845179081,
|
| 4914 |
+
"learning_rate": 1.4950000000000001e-05,
|
| 4915 |
+
"loss": 0.0014,
|
| 4916 |
+
"step": 7010
|
| 4917 |
+
},
|
| 4918 |
+
{
|
| 4919 |
+
"epoch": 0.702,
|
| 4920 |
+
"grad_norm": 0.01417300570756197,
|
| 4921 |
+
"learning_rate": 1.49e-05,
|
| 4922 |
+
"loss": 0.001,
|
| 4923 |
+
"step": 7020
|
| 4924 |
+
},
|
| 4925 |
+
{
|
| 4926 |
+
"epoch": 0.703,
|
| 4927 |
+
"grad_norm": 0.010836401022970676,
|
| 4928 |
+
"learning_rate": 1.485e-05,
|
| 4929 |
+
"loss": 0.0012,
|
| 4930 |
+
"step": 7030
|
| 4931 |
+
},
|
| 4932 |
+
{
|
| 4933 |
+
"epoch": 0.704,
|
| 4934 |
+
"grad_norm": 0.01000068336725235,
|
| 4935 |
+
"learning_rate": 1.48e-05,
|
| 4936 |
+
"loss": 0.001,
|
| 4937 |
+
"step": 7040
|
| 4938 |
+
},
|
| 4939 |
+
{
|
| 4940 |
+
"epoch": 0.705,
|
| 4941 |
+
"grad_norm": 0.008654952049255371,
|
| 4942 |
+
"learning_rate": 1.475e-05,
|
| 4943 |
+
"loss": 0.0009,
|
| 4944 |
+
"step": 7050
|
| 4945 |
+
},
|
| 4946 |
+
{
|
| 4947 |
+
"epoch": 0.706,
|
| 4948 |
+
"grad_norm": 0.010761331766843796,
|
| 4949 |
+
"learning_rate": 1.47e-05,
|
| 4950 |
+
"loss": 0.001,
|
| 4951 |
+
"step": 7060
|
| 4952 |
+
},
|
| 4953 |
+
{
|
| 4954 |
+
"epoch": 0.707,
|
| 4955 |
+
"grad_norm": 0.006188638508319855,
|
| 4956 |
+
"learning_rate": 1.465e-05,
|
| 4957 |
+
"loss": 0.0008,
|
| 4958 |
+
"step": 7070
|
| 4959 |
+
},
|
| 4960 |
+
{
|
| 4961 |
+
"epoch": 0.708,
|
| 4962 |
+
"grad_norm": 0.007858789525926113,
|
| 4963 |
+
"learning_rate": 1.4599999999999999e-05,
|
| 4964 |
+
"loss": 0.0008,
|
| 4965 |
+
"step": 7080
|
| 4966 |
+
},
|
| 4967 |
+
{
|
| 4968 |
+
"epoch": 0.709,
|
| 4969 |
+
"grad_norm": 0.02773350477218628,
|
| 4970 |
+
"learning_rate": 1.455e-05,
|
| 4971 |
+
"loss": 0.0014,
|
| 4972 |
+
"step": 7090
|
| 4973 |
+
},
|
| 4974 |
+
{
|
| 4975 |
+
"epoch": 0.71,
|
| 4976 |
+
"grad_norm": 0.012381108477711678,
|
| 4977 |
+
"learning_rate": 1.45e-05,
|
| 4978 |
+
"loss": 0.0009,
|
| 4979 |
+
"step": 7100
|
| 4980 |
+
},
|
| 4981 |
+
{
|
| 4982 |
+
"epoch": 0.711,
|
| 4983 |
+
"grad_norm": 0.009256324730813503,
|
| 4984 |
+
"learning_rate": 1.4449999999999999e-05,
|
| 4985 |
+
"loss": 0.0008,
|
| 4986 |
+
"step": 7110
|
| 4987 |
+
},
|
| 4988 |
+
{
|
| 4989 |
+
"epoch": 0.712,
|
| 4990 |
+
"grad_norm": 0.007005748804658651,
|
| 4991 |
+
"learning_rate": 1.44e-05,
|
| 4992 |
+
"loss": 0.0009,
|
| 4993 |
+
"step": 7120
|
| 4994 |
+
},
|
| 4995 |
+
{
|
| 4996 |
+
"epoch": 0.713,
|
| 4997 |
+
"grad_norm": 0.0055755749344825745,
|
| 4998 |
+
"learning_rate": 1.435e-05,
|
| 4999 |
+
"loss": 0.0007,
|
| 5000 |
+
"step": 7130
|
| 5001 |
+
},
|
| 5002 |
+
{
|
| 5003 |
+
"epoch": 0.714,
|
| 5004 |
+
"grad_norm": 0.003967254888266325,
|
| 5005 |
+
"learning_rate": 1.43e-05,
|
| 5006 |
+
"loss": 0.0008,
|
| 5007 |
+
"step": 7140
|
| 5008 |
+
},
|
| 5009 |
+
{
|
| 5010 |
+
"epoch": 0.715,
|
| 5011 |
+
"grad_norm": 0.0079165268689394,
|
| 5012 |
+
"learning_rate": 1.4249999999999999e-05,
|
| 5013 |
+
"loss": 0.0011,
|
| 5014 |
+
"step": 7150
|
| 5015 |
+
},
|
| 5016 |
+
{
|
| 5017 |
+
"epoch": 0.716,
|
| 5018 |
+
"grad_norm": 0.004682580940425396,
|
| 5019 |
+
"learning_rate": 1.42e-05,
|
| 5020 |
+
"loss": 0.0007,
|
| 5021 |
+
"step": 7160
|
| 5022 |
+
},
|
| 5023 |
+
{
|
| 5024 |
+
"epoch": 0.717,
|
| 5025 |
+
"grad_norm": 0.008578700013458729,
|
| 5026 |
+
"learning_rate": 1.415e-05,
|
| 5027 |
+
"loss": 0.0011,
|
| 5028 |
+
"step": 7170
|
| 5029 |
+
},
|
| 5030 |
+
{
|
| 5031 |
+
"epoch": 0.718,
|
| 5032 |
+
"grad_norm": 0.006943961605429649,
|
| 5033 |
+
"learning_rate": 1.4099999999999999e-05,
|
| 5034 |
+
"loss": 0.0009,
|
| 5035 |
+
"step": 7180
|
| 5036 |
+
},
|
| 5037 |
+
{
|
| 5038 |
+
"epoch": 0.719,
|
| 5039 |
+
"grad_norm": 0.0072656250558793545,
|
| 5040 |
+
"learning_rate": 1.4050000000000003e-05,
|
| 5041 |
+
"loss": 0.0007,
|
| 5042 |
+
"step": 7190
|
| 5043 |
+
},
|
| 5044 |
+
{
|
| 5045 |
+
"epoch": 0.72,
|
| 5046 |
+
"grad_norm": 0.005639955401420593,
|
| 5047 |
+
"learning_rate": 1.4000000000000001e-05,
|
| 5048 |
+
"loss": 0.0007,
|
| 5049 |
+
"step": 7200
|
| 5050 |
+
},
|
| 5051 |
+
{
|
| 5052 |
+
"epoch": 0.721,
|
| 5053 |
+
"grad_norm": 0.005733838304877281,
|
| 5054 |
+
"learning_rate": 1.3950000000000002e-05,
|
| 5055 |
+
"loss": 0.0008,
|
| 5056 |
+
"step": 7210
|
| 5057 |
+
},
|
| 5058 |
+
{
|
| 5059 |
+
"epoch": 0.722,
|
| 5060 |
+
"grad_norm": 0.02654002234339714,
|
| 5061 |
+
"learning_rate": 1.3900000000000002e-05,
|
| 5062 |
+
"loss": 0.0008,
|
| 5063 |
+
"step": 7220
|
| 5064 |
+
},
|
| 5065 |
+
{
|
| 5066 |
+
"epoch": 0.723,
|
| 5067 |
+
"grad_norm": 0.007308628410100937,
|
| 5068 |
+
"learning_rate": 1.3850000000000001e-05,
|
| 5069 |
+
"loss": 0.0008,
|
| 5070 |
+
"step": 7230
|
| 5071 |
+
},
|
| 5072 |
+
{
|
| 5073 |
+
"epoch": 0.724,
|
| 5074 |
+
"grad_norm": 0.006939894054085016,
|
| 5075 |
+
"learning_rate": 1.3800000000000002e-05,
|
| 5076 |
+
"loss": 0.0007,
|
| 5077 |
+
"step": 7240
|
| 5078 |
+
},
|
| 5079 |
+
{
|
| 5080 |
+
"epoch": 0.725,
|
| 5081 |
+
"grad_norm": 0.03964811936020851,
|
| 5082 |
+
"learning_rate": 1.3750000000000002e-05,
|
| 5083 |
+
"loss": 0.0013,
|
| 5084 |
+
"step": 7250
|
| 5085 |
+
},
|
| 5086 |
+
{
|
| 5087 |
+
"epoch": 0.726,
|
| 5088 |
+
"grad_norm": 0.014138396829366684,
|
| 5089 |
+
"learning_rate": 1.3700000000000001e-05,
|
| 5090 |
+
"loss": 0.001,
|
| 5091 |
+
"step": 7260
|
| 5092 |
+
},
|
| 5093 |
+
{
|
| 5094 |
+
"epoch": 0.727,
|
| 5095 |
+
"grad_norm": 0.008445181883871555,
|
| 5096 |
+
"learning_rate": 1.3650000000000001e-05,
|
| 5097 |
+
"loss": 0.0008,
|
| 5098 |
+
"step": 7270
|
| 5099 |
+
},
|
| 5100 |
+
{
|
| 5101 |
+
"epoch": 0.728,
|
| 5102 |
+
"grad_norm": 0.01134855579584837,
|
| 5103 |
+
"learning_rate": 1.3600000000000002e-05,
|
| 5104 |
+
"loss": 0.0009,
|
| 5105 |
+
"step": 7280
|
| 5106 |
+
},
|
| 5107 |
+
{
|
| 5108 |
+
"epoch": 0.729,
|
| 5109 |
+
"grad_norm": 0.010982022620737553,
|
| 5110 |
+
"learning_rate": 1.3550000000000002e-05,
|
| 5111 |
+
"loss": 0.0015,
|
| 5112 |
+
"step": 7290
|
| 5113 |
+
},
|
| 5114 |
+
{
|
| 5115 |
+
"epoch": 0.73,
|
| 5116 |
+
"grad_norm": 0.011698734015226364,
|
| 5117 |
+
"learning_rate": 1.3500000000000001e-05,
|
| 5118 |
+
"loss": 0.0008,
|
| 5119 |
+
"step": 7300
|
| 5120 |
+
},
|
| 5121 |
+
{
|
| 5122 |
+
"epoch": 0.731,
|
| 5123 |
+
"grad_norm": 0.006420729216188192,
|
| 5124 |
+
"learning_rate": 1.3450000000000002e-05,
|
| 5125 |
+
"loss": 0.0008,
|
| 5126 |
+
"step": 7310
|
| 5127 |
+
},
|
| 5128 |
+
{
|
| 5129 |
+
"epoch": 0.732,
|
| 5130 |
+
"grad_norm": 0.006088167428970337,
|
| 5131 |
+
"learning_rate": 1.3400000000000002e-05,
|
| 5132 |
+
"loss": 0.0008,
|
| 5133 |
+
"step": 7320
|
| 5134 |
+
},
|
| 5135 |
+
{
|
| 5136 |
+
"epoch": 0.733,
|
| 5137 |
+
"grad_norm": 0.0071141645312309265,
|
| 5138 |
+
"learning_rate": 1.3350000000000001e-05,
|
| 5139 |
+
"loss": 0.0012,
|
| 5140 |
+
"step": 7330
|
| 5141 |
+
},
|
| 5142 |
+
{
|
| 5143 |
+
"epoch": 0.734,
|
| 5144 |
+
"grad_norm": 0.004975921008735895,
|
| 5145 |
+
"learning_rate": 1.3300000000000001e-05,
|
| 5146 |
+
"loss": 0.0006,
|
| 5147 |
+
"step": 7340
|
| 5148 |
+
},
|
| 5149 |
+
{
|
| 5150 |
+
"epoch": 0.735,
|
| 5151 |
+
"grad_norm": 0.004499469883739948,
|
| 5152 |
+
"learning_rate": 1.3250000000000002e-05,
|
| 5153 |
+
"loss": 0.0007,
|
| 5154 |
+
"step": 7350
|
| 5155 |
+
},
|
| 5156 |
+
{
|
| 5157 |
+
"epoch": 0.736,
|
| 5158 |
+
"grad_norm": 0.009738982655107975,
|
| 5159 |
+
"learning_rate": 1.32e-05,
|
| 5160 |
+
"loss": 0.001,
|
| 5161 |
+
"step": 7360
|
| 5162 |
+
},
|
| 5163 |
+
{
|
| 5164 |
+
"epoch": 0.737,
|
| 5165 |
+
"grad_norm": 0.006863337475806475,
|
| 5166 |
+
"learning_rate": 1.3150000000000001e-05,
|
| 5167 |
+
"loss": 0.001,
|
| 5168 |
+
"step": 7370
|
| 5169 |
+
},
|
| 5170 |
+
{
|
| 5171 |
+
"epoch": 0.738,
|
| 5172 |
+
"grad_norm": 0.008216536603868008,
|
| 5173 |
+
"learning_rate": 1.3100000000000002e-05,
|
| 5174 |
+
"loss": 0.0007,
|
| 5175 |
+
"step": 7380
|
| 5176 |
+
},
|
| 5177 |
+
{
|
| 5178 |
+
"epoch": 0.739,
|
| 5179 |
+
"grad_norm": 0.006803369149565697,
|
| 5180 |
+
"learning_rate": 1.305e-05,
|
| 5181 |
+
"loss": 0.0008,
|
| 5182 |
+
"step": 7390
|
| 5183 |
+
},
|
| 5184 |
+
{
|
| 5185 |
+
"epoch": 0.74,
|
| 5186 |
+
"grad_norm": 0.00551017839461565,
|
| 5187 |
+
"learning_rate": 1.3000000000000001e-05,
|
| 5188 |
+
"loss": 0.0008,
|
| 5189 |
+
"step": 7400
|
| 5190 |
+
},
|
| 5191 |
+
{
|
| 5192 |
+
"epoch": 0.741,
|
| 5193 |
+
"grad_norm": 0.009463651105761528,
|
| 5194 |
+
"learning_rate": 1.2950000000000001e-05,
|
| 5195 |
+
"loss": 0.0008,
|
| 5196 |
+
"step": 7410
|
| 5197 |
+
},
|
| 5198 |
+
{
|
| 5199 |
+
"epoch": 0.742,
|
| 5200 |
+
"grad_norm": 0.01233983039855957,
|
| 5201 |
+
"learning_rate": 1.29e-05,
|
| 5202 |
+
"loss": 0.0019,
|
| 5203 |
+
"step": 7420
|
| 5204 |
+
},
|
| 5205 |
+
{
|
| 5206 |
+
"epoch": 0.743,
|
| 5207 |
+
"grad_norm": 0.008470877073705196,
|
| 5208 |
+
"learning_rate": 1.285e-05,
|
| 5209 |
+
"loss": 0.0009,
|
| 5210 |
+
"step": 7430
|
| 5211 |
+
},
|
| 5212 |
+
{
|
| 5213 |
+
"epoch": 0.744,
|
| 5214 |
+
"grad_norm": 0.007592742796987295,
|
| 5215 |
+
"learning_rate": 1.2800000000000001e-05,
|
| 5216 |
+
"loss": 0.0008,
|
| 5217 |
+
"step": 7440
|
| 5218 |
+
},
|
| 5219 |
+
{
|
| 5220 |
+
"epoch": 0.745,
|
| 5221 |
+
"grad_norm": 0.03596987947821617,
|
| 5222 |
+
"learning_rate": 1.2750000000000002e-05,
|
| 5223 |
+
"loss": 0.001,
|
| 5224 |
+
"step": 7450
|
| 5225 |
+
},
|
| 5226 |
+
{
|
| 5227 |
+
"epoch": 0.746,
|
| 5228 |
+
"grad_norm": 0.005849502049386501,
|
| 5229 |
+
"learning_rate": 1.27e-05,
|
| 5230 |
+
"loss": 0.0008,
|
| 5231 |
+
"step": 7460
|
| 5232 |
+
},
|
| 5233 |
+
{
|
| 5234 |
+
"epoch": 0.747,
|
| 5235 |
+
"grad_norm": 0.009035659022629261,
|
| 5236 |
+
"learning_rate": 1.2650000000000001e-05,
|
| 5237 |
+
"loss": 0.0007,
|
| 5238 |
+
"step": 7470
|
| 5239 |
+
},
|
| 5240 |
+
{
|
| 5241 |
+
"epoch": 0.748,
|
| 5242 |
+
"grad_norm": 0.010397679172456264,
|
| 5243 |
+
"learning_rate": 1.2600000000000001e-05,
|
| 5244 |
+
"loss": 0.0014,
|
| 5245 |
+
"step": 7480
|
| 5246 |
+
},
|
| 5247 |
+
{
|
| 5248 |
+
"epoch": 0.749,
|
| 5249 |
+
"grad_norm": 0.014514378271996975,
|
| 5250 |
+
"learning_rate": 1.255e-05,
|
| 5251 |
+
"loss": 0.0008,
|
| 5252 |
+
"step": 7490
|
| 5253 |
+
},
|
| 5254 |
+
{
|
| 5255 |
+
"epoch": 0.75,
|
| 5256 |
+
"grad_norm": 0.004837281536310911,
|
| 5257 |
+
"learning_rate": 1.25e-05,
|
| 5258 |
+
"loss": 0.0006,
|
| 5259 |
+
"step": 7500
|
| 5260 |
+
},
|
| 5261 |
+
{
|
| 5262 |
+
"epoch": 0.751,
|
| 5263 |
+
"grad_norm": 0.007720770314335823,
|
| 5264 |
+
"learning_rate": 1.2450000000000001e-05,
|
| 5265 |
+
"loss": 0.0006,
|
| 5266 |
+
"step": 7510
|
| 5267 |
+
},
|
| 5268 |
+
{
|
| 5269 |
+
"epoch": 0.752,
|
| 5270 |
+
"grad_norm": 0.012046804651618004,
|
| 5271 |
+
"learning_rate": 1.24e-05,
|
| 5272 |
+
"loss": 0.0011,
|
| 5273 |
+
"step": 7520
|
| 5274 |
+
},
|
| 5275 |
+
{
|
| 5276 |
+
"epoch": 0.753,
|
| 5277 |
+
"grad_norm": 0.01343387458473444,
|
| 5278 |
+
"learning_rate": 1.235e-05,
|
| 5279 |
+
"loss": 0.0007,
|
| 5280 |
+
"step": 7530
|
| 5281 |
+
},
|
| 5282 |
+
{
|
| 5283 |
+
"epoch": 0.754,
|
| 5284 |
+
"grad_norm": 0.00810600072145462,
|
| 5285 |
+
"learning_rate": 1.23e-05,
|
| 5286 |
+
"loss": 0.0007,
|
| 5287 |
+
"step": 7540
|
| 5288 |
+
},
|
| 5289 |
+
{
|
| 5290 |
+
"epoch": 0.755,
|
| 5291 |
+
"grad_norm": 0.00925883837044239,
|
| 5292 |
+
"learning_rate": 1.225e-05,
|
| 5293 |
+
"loss": 0.0006,
|
| 5294 |
+
"step": 7550
|
| 5295 |
+
},
|
| 5296 |
+
{
|
| 5297 |
+
"epoch": 0.756,
|
| 5298 |
+
"grad_norm": 0.01927885413169861,
|
| 5299 |
+
"learning_rate": 1.22e-05,
|
| 5300 |
+
"loss": 0.0014,
|
| 5301 |
+
"step": 7560
|
| 5302 |
+
},
|
| 5303 |
+
{
|
| 5304 |
+
"epoch": 0.757,
|
| 5305 |
+
"grad_norm": 0.010129665955901146,
|
| 5306 |
+
"learning_rate": 1.215e-05,
|
| 5307 |
+
"loss": 0.0006,
|
| 5308 |
+
"step": 7570
|
| 5309 |
+
},
|
| 5310 |
+
{
|
| 5311 |
+
"epoch": 0.758,
|
| 5312 |
+
"grad_norm": 0.007863885723054409,
|
| 5313 |
+
"learning_rate": 1.2100000000000001e-05,
|
| 5314 |
+
"loss": 0.0006,
|
| 5315 |
+
"step": 7580
|
| 5316 |
+
},
|
| 5317 |
+
{
|
| 5318 |
+
"epoch": 0.759,
|
| 5319 |
+
"grad_norm": 0.005500464700162411,
|
| 5320 |
+
"learning_rate": 1.205e-05,
|
| 5321 |
+
"loss": 0.0007,
|
| 5322 |
+
"step": 7590
|
| 5323 |
+
},
|
| 5324 |
+
{
|
| 5325 |
+
"epoch": 0.76,
|
| 5326 |
+
"grad_norm": 0.0040563903748989105,
|
| 5327 |
+
"learning_rate": 1.2e-05,
|
| 5328 |
+
"loss": 0.0006,
|
| 5329 |
+
"step": 7600
|
| 5330 |
+
},
|
| 5331 |
+
{
|
| 5332 |
+
"epoch": 0.761,
|
| 5333 |
+
"grad_norm": 0.006361998151987791,
|
| 5334 |
+
"learning_rate": 1.195e-05,
|
| 5335 |
+
"loss": 0.0007,
|
| 5336 |
+
"step": 7610
|
| 5337 |
+
},
|
| 5338 |
+
{
|
| 5339 |
+
"epoch": 0.762,
|
| 5340 |
+
"grad_norm": 0.0136310625821352,
|
| 5341 |
+
"learning_rate": 1.19e-05,
|
| 5342 |
+
"loss": 0.0008,
|
| 5343 |
+
"step": 7620
|
| 5344 |
+
},
|
| 5345 |
+
{
|
| 5346 |
+
"epoch": 0.763,
|
| 5347 |
+
"grad_norm": 0.005384715739637613,
|
| 5348 |
+
"learning_rate": 1.185e-05,
|
| 5349 |
+
"loss": 0.0007,
|
| 5350 |
+
"step": 7630
|
| 5351 |
+
},
|
| 5352 |
+
{
|
| 5353 |
+
"epoch": 0.764,
|
| 5354 |
+
"grad_norm": 0.014707676135003567,
|
| 5355 |
+
"learning_rate": 1.18e-05,
|
| 5356 |
+
"loss": 0.0007,
|
| 5357 |
+
"step": 7640
|
| 5358 |
+
},
|
| 5359 |
+
{
|
| 5360 |
+
"epoch": 0.765,
|
| 5361 |
+
"grad_norm": 0.008092684671282768,
|
| 5362 |
+
"learning_rate": 1.175e-05,
|
| 5363 |
+
"loss": 0.0006,
|
| 5364 |
+
"step": 7650
|
| 5365 |
+
},
|
| 5366 |
+
{
|
| 5367 |
+
"epoch": 0.766,
|
| 5368 |
+
"grad_norm": 0.007185132242739201,
|
| 5369 |
+
"learning_rate": 1.1700000000000001e-05,
|
| 5370 |
+
"loss": 0.0006,
|
| 5371 |
+
"step": 7660
|
| 5372 |
+
},
|
| 5373 |
+
{
|
| 5374 |
+
"epoch": 0.767,
|
| 5375 |
+
"grad_norm": 0.005672789178788662,
|
| 5376 |
+
"learning_rate": 1.1650000000000002e-05,
|
| 5377 |
+
"loss": 0.0006,
|
| 5378 |
+
"step": 7670
|
| 5379 |
+
},
|
| 5380 |
+
{
|
| 5381 |
+
"epoch": 0.768,
|
| 5382 |
+
"grad_norm": 0.05434956029057503,
|
| 5383 |
+
"learning_rate": 1.16e-05,
|
| 5384 |
+
"loss": 0.001,
|
| 5385 |
+
"step": 7680
|
| 5386 |
+
},
|
| 5387 |
+
{
|
| 5388 |
+
"epoch": 0.769,
|
| 5389 |
+
"grad_norm": 0.00933472067117691,
|
| 5390 |
+
"learning_rate": 1.1550000000000001e-05,
|
| 5391 |
+
"loss": 0.0007,
|
| 5392 |
+
"step": 7690
|
| 5393 |
+
},
|
| 5394 |
+
{
|
| 5395 |
+
"epoch": 0.77,
|
| 5396 |
+
"grad_norm": 0.008684621192514896,
|
| 5397 |
+
"learning_rate": 1.1500000000000002e-05,
|
| 5398 |
+
"loss": 0.0006,
|
| 5399 |
+
"step": 7700
|
| 5400 |
+
},
|
| 5401 |
+
{
|
| 5402 |
+
"epoch": 0.771,
|
| 5403 |
+
"grad_norm": 0.03054739721119404,
|
| 5404 |
+
"learning_rate": 1.145e-05,
|
| 5405 |
+
"loss": 0.0006,
|
| 5406 |
+
"step": 7710
|
| 5407 |
+
},
|
| 5408 |
+
{
|
| 5409 |
+
"epoch": 0.772,
|
| 5410 |
+
"grad_norm": 0.005998207256197929,
|
| 5411 |
+
"learning_rate": 1.1400000000000001e-05,
|
| 5412 |
+
"loss": 0.0006,
|
| 5413 |
+
"step": 7720
|
| 5414 |
+
},
|
| 5415 |
+
{
|
| 5416 |
+
"epoch": 0.773,
|
| 5417 |
+
"grad_norm": 0.006153833121061325,
|
| 5418 |
+
"learning_rate": 1.1350000000000001e-05,
|
| 5419 |
+
"loss": 0.0006,
|
| 5420 |
+
"step": 7730
|
| 5421 |
+
},
|
| 5422 |
+
{
|
| 5423 |
+
"epoch": 0.774,
|
| 5424 |
+
"grad_norm": 0.007491481024771929,
|
| 5425 |
+
"learning_rate": 1.13e-05,
|
| 5426 |
+
"loss": 0.0007,
|
| 5427 |
+
"step": 7740
|
| 5428 |
+
},
|
| 5429 |
+
{
|
| 5430 |
+
"epoch": 0.775,
|
| 5431 |
+
"grad_norm": 0.01078925933688879,
|
| 5432 |
+
"learning_rate": 1.125e-05,
|
| 5433 |
+
"loss": 0.0006,
|
| 5434 |
+
"step": 7750
|
| 5435 |
+
},
|
| 5436 |
+
{
|
| 5437 |
+
"epoch": 0.776,
|
| 5438 |
+
"grad_norm": 0.005885554943233728,
|
| 5439 |
+
"learning_rate": 1.1200000000000001e-05,
|
| 5440 |
+
"loss": 0.0006,
|
| 5441 |
+
"step": 7760
|
| 5442 |
+
},
|
| 5443 |
+
{
|
| 5444 |
+
"epoch": 0.777,
|
| 5445 |
+
"grad_norm": 0.005423078313469887,
|
| 5446 |
+
"learning_rate": 1.115e-05,
|
| 5447 |
+
"loss": 0.0007,
|
| 5448 |
+
"step": 7770
|
| 5449 |
+
},
|
| 5450 |
+
{
|
| 5451 |
+
"epoch": 0.778,
|
| 5452 |
+
"grad_norm": 0.008044522255659103,
|
| 5453 |
+
"learning_rate": 1.11e-05,
|
| 5454 |
+
"loss": 0.0006,
|
| 5455 |
+
"step": 7780
|
| 5456 |
+
},
|
| 5457 |
+
{
|
| 5458 |
+
"epoch": 0.779,
|
| 5459 |
+
"grad_norm": 0.00733207818120718,
|
| 5460 |
+
"learning_rate": 1.1050000000000001e-05,
|
| 5461 |
+
"loss": 0.0007,
|
| 5462 |
+
"step": 7790
|
| 5463 |
+
},
|
| 5464 |
+
{
|
| 5465 |
+
"epoch": 0.78,
|
| 5466 |
+
"grad_norm": 0.0066906120628118515,
|
| 5467 |
+
"learning_rate": 1.1000000000000001e-05,
|
| 5468 |
+
"loss": 0.0009,
|
| 5469 |
+
"step": 7800
|
| 5470 |
+
},
|
| 5471 |
+
{
|
| 5472 |
+
"epoch": 0.781,
|
| 5473 |
+
"grad_norm": 0.004443836398422718,
|
| 5474 |
+
"learning_rate": 1.095e-05,
|
| 5475 |
+
"loss": 0.0006,
|
| 5476 |
+
"step": 7810
|
| 5477 |
+
},
|
| 5478 |
+
{
|
| 5479 |
+
"epoch": 0.782,
|
| 5480 |
+
"grad_norm": 0.0058379145339131355,
|
| 5481 |
+
"learning_rate": 1.09e-05,
|
| 5482 |
+
"loss": 0.0007,
|
| 5483 |
+
"step": 7820
|
| 5484 |
+
},
|
| 5485 |
+
{
|
| 5486 |
+
"epoch": 0.783,
|
| 5487 |
+
"grad_norm": 0.006808693055063486,
|
| 5488 |
+
"learning_rate": 1.0850000000000001e-05,
|
| 5489 |
+
"loss": 0.0006,
|
| 5490 |
+
"step": 7830
|
| 5491 |
+
},
|
| 5492 |
+
{
|
| 5493 |
+
"epoch": 0.784,
|
| 5494 |
+
"grad_norm": 0.008773542940616608,
|
| 5495 |
+
"learning_rate": 1.08e-05,
|
| 5496 |
+
"loss": 0.0006,
|
| 5497 |
+
"step": 7840
|
| 5498 |
+
},
|
| 5499 |
+
{
|
| 5500 |
+
"epoch": 0.785,
|
| 5501 |
+
"grad_norm": 0.006700740661472082,
|
| 5502 |
+
"learning_rate": 1.075e-05,
|
| 5503 |
+
"loss": 0.0006,
|
| 5504 |
+
"step": 7850
|
| 5505 |
+
},
|
| 5506 |
+
{
|
| 5507 |
+
"epoch": 0.786,
|
| 5508 |
+
"grad_norm": 0.00906393863260746,
|
| 5509 |
+
"learning_rate": 1.0700000000000001e-05,
|
| 5510 |
+
"loss": 0.0006,
|
| 5511 |
+
"step": 7860
|
| 5512 |
+
},
|
| 5513 |
+
{
|
| 5514 |
+
"epoch": 0.787,
|
| 5515 |
+
"grad_norm": 0.0030822190456092358,
|
| 5516 |
+
"learning_rate": 1.065e-05,
|
| 5517 |
+
"loss": 0.0005,
|
| 5518 |
+
"step": 7870
|
| 5519 |
+
},
|
| 5520 |
+
{
|
| 5521 |
+
"epoch": 0.788,
|
| 5522 |
+
"grad_norm": 0.0029632148798555136,
|
| 5523 |
+
"learning_rate": 1.06e-05,
|
| 5524 |
+
"loss": 0.0005,
|
| 5525 |
+
"step": 7880
|
| 5526 |
+
},
|
| 5527 |
+
{
|
| 5528 |
+
"epoch": 0.789,
|
| 5529 |
+
"grad_norm": 0.004798842128366232,
|
| 5530 |
+
"learning_rate": 1.055e-05,
|
| 5531 |
+
"loss": 0.0006,
|
| 5532 |
+
"step": 7890
|
| 5533 |
+
},
|
| 5534 |
+
{
|
| 5535 |
+
"epoch": 0.79,
|
| 5536 |
+
"grad_norm": 0.007376812864094973,
|
| 5537 |
+
"learning_rate": 1.05e-05,
|
| 5538 |
+
"loss": 0.0005,
|
| 5539 |
+
"step": 7900
|
| 5540 |
+
},
|
| 5541 |
+
{
|
| 5542 |
+
"epoch": 0.791,
|
| 5543 |
+
"grad_norm": 0.009337624534964561,
|
| 5544 |
+
"learning_rate": 1.045e-05,
|
| 5545 |
+
"loss": 0.0009,
|
| 5546 |
+
"step": 7910
|
| 5547 |
+
},
|
| 5548 |
+
{
|
| 5549 |
+
"epoch": 0.792,
|
| 5550 |
+
"grad_norm": 0.012847904115915298,
|
| 5551 |
+
"learning_rate": 1.04e-05,
|
| 5552 |
+
"loss": 0.0008,
|
| 5553 |
+
"step": 7920
|
| 5554 |
+
},
|
| 5555 |
+
{
|
| 5556 |
+
"epoch": 0.793,
|
| 5557 |
+
"grad_norm": 0.005587203428149223,
|
| 5558 |
+
"learning_rate": 1.035e-05,
|
| 5559 |
+
"loss": 0.0006,
|
| 5560 |
+
"step": 7930
|
| 5561 |
+
},
|
| 5562 |
+
{
|
| 5563 |
+
"epoch": 0.794,
|
| 5564 |
+
"grad_norm": 0.008464600890874863,
|
| 5565 |
+
"learning_rate": 1.03e-05,
|
| 5566 |
+
"loss": 0.0006,
|
| 5567 |
+
"step": 7940
|
| 5568 |
+
},
|
| 5569 |
+
{
|
| 5570 |
+
"epoch": 0.795,
|
| 5571 |
+
"grad_norm": 0.2516852617263794,
|
| 5572 |
+
"learning_rate": 1.025e-05,
|
| 5573 |
+
"loss": 0.002,
|
| 5574 |
+
"step": 7950
|
| 5575 |
+
},
|
| 5576 |
+
{
|
| 5577 |
+
"epoch": 0.796,
|
| 5578 |
+
"grad_norm": 0.04664693772792816,
|
| 5579 |
+
"learning_rate": 1.02e-05,
|
| 5580 |
+
"loss": 0.002,
|
| 5581 |
+
"step": 7960
|
| 5582 |
+
},
|
| 5583 |
+
{
|
| 5584 |
+
"epoch": 0.797,
|
| 5585 |
+
"grad_norm": 0.02456306852400303,
|
| 5586 |
+
"learning_rate": 1.0150000000000001e-05,
|
| 5587 |
+
"loss": 0.0013,
|
| 5588 |
+
"step": 7970
|
| 5589 |
+
},
|
| 5590 |
+
{
|
| 5591 |
+
"epoch": 0.798,
|
| 5592 |
+
"grad_norm": 0.011320951394736767,
|
| 5593 |
+
"learning_rate": 1.0100000000000002e-05,
|
| 5594 |
+
"loss": 0.0009,
|
| 5595 |
+
"step": 7980
|
| 5596 |
+
},
|
| 5597 |
+
{
|
| 5598 |
+
"epoch": 0.799,
|
| 5599 |
+
"grad_norm": 0.01860683411359787,
|
| 5600 |
+
"learning_rate": 1.005e-05,
|
| 5601 |
+
"loss": 0.0012,
|
| 5602 |
+
"step": 7990
|
| 5603 |
+
},
|
| 5604 |
+
{
|
| 5605 |
+
"epoch": 0.8,
|
| 5606 |
+
"grad_norm": 0.03227970749139786,
|
| 5607 |
+
"learning_rate": 1e-05,
|
| 5608 |
+
"loss": 0.0009,
|
| 5609 |
+
"step": 8000
|
| 5610 |
+
},
|
| 5611 |
+
{
|
| 5612 |
+
"epoch": 0.801,
|
| 5613 |
+
"grad_norm": 0.015873363241553307,
|
| 5614 |
+
"learning_rate": 9.950000000000001e-06,
|
| 5615 |
+
"loss": 0.0008,
|
| 5616 |
+
"step": 8010
|
| 5617 |
+
},
|
| 5618 |
+
{
|
| 5619 |
+
"epoch": 0.802,
|
| 5620 |
+
"grad_norm": 0.005454899277538061,
|
| 5621 |
+
"learning_rate": 9.900000000000002e-06,
|
| 5622 |
+
"loss": 0.0008,
|
| 5623 |
+
"step": 8020
|
| 5624 |
+
},
|
| 5625 |
+
{
|
| 5626 |
+
"epoch": 0.803,
|
| 5627 |
+
"grad_norm": 0.007948348298668861,
|
| 5628 |
+
"learning_rate": 9.85e-06,
|
| 5629 |
+
"loss": 0.0007,
|
| 5630 |
+
"step": 8030
|
| 5631 |
+
},
|
| 5632 |
+
{
|
| 5633 |
+
"epoch": 0.804,
|
| 5634 |
+
"grad_norm": 0.013328757137060165,
|
| 5635 |
+
"learning_rate": 9.800000000000001e-06,
|
| 5636 |
+
"loss": 0.0006,
|
| 5637 |
+
"step": 8040
|
| 5638 |
+
},
|
| 5639 |
+
{
|
| 5640 |
+
"epoch": 0.805,
|
| 5641 |
+
"grad_norm": 0.01018743496388197,
|
| 5642 |
+
"learning_rate": 9.750000000000002e-06,
|
| 5643 |
+
"loss": 0.0012,
|
| 5644 |
+
"step": 8050
|
| 5645 |
+
},
|
| 5646 |
+
{
|
| 5647 |
+
"epoch": 0.806,
|
| 5648 |
+
"grad_norm": 0.009421809576451778,
|
| 5649 |
+
"learning_rate": 9.7e-06,
|
| 5650 |
+
"loss": 0.0008,
|
| 5651 |
+
"step": 8060
|
| 5652 |
+
},
|
| 5653 |
+
{
|
| 5654 |
+
"epoch": 0.807,
|
| 5655 |
+
"grad_norm": 0.005202045664191246,
|
| 5656 |
+
"learning_rate": 9.65e-06,
|
| 5657 |
+
"loss": 0.0007,
|
| 5658 |
+
"step": 8070
|
| 5659 |
+
},
|
| 5660 |
+
{
|
| 5661 |
+
"epoch": 0.808,
|
| 5662 |
+
"grad_norm": 0.012956002727150917,
|
| 5663 |
+
"learning_rate": 9.600000000000001e-06,
|
| 5664 |
+
"loss": 0.0007,
|
| 5665 |
+
"step": 8080
|
| 5666 |
+
},
|
| 5667 |
+
{
|
| 5668 |
+
"epoch": 0.809,
|
| 5669 |
+
"grad_norm": 0.006403383333235979,
|
| 5670 |
+
"learning_rate": 9.55e-06,
|
| 5671 |
+
"loss": 0.0007,
|
| 5672 |
+
"step": 8090
|
| 5673 |
+
},
|
| 5674 |
+
{
|
| 5675 |
+
"epoch": 0.81,
|
| 5676 |
+
"grad_norm": 0.027560915797948837,
|
| 5677 |
+
"learning_rate": 9.5e-06,
|
| 5678 |
+
"loss": 0.0008,
|
| 5679 |
+
"step": 8100
|
| 5680 |
+
},
|
| 5681 |
+
{
|
| 5682 |
+
"epoch": 0.811,
|
| 5683 |
+
"grad_norm": 0.005196988116949797,
|
| 5684 |
+
"learning_rate": 9.450000000000001e-06,
|
| 5685 |
+
"loss": 0.0006,
|
| 5686 |
+
"step": 8110
|
| 5687 |
+
},
|
| 5688 |
+
{
|
| 5689 |
+
"epoch": 0.812,
|
| 5690 |
+
"grad_norm": 0.009510821662843227,
|
| 5691 |
+
"learning_rate": 9.4e-06,
|
| 5692 |
+
"loss": 0.0006,
|
| 5693 |
+
"step": 8120
|
| 5694 |
+
},
|
| 5695 |
+
{
|
| 5696 |
+
"epoch": 0.813,
|
| 5697 |
+
"grad_norm": 0.006430651992559433,
|
| 5698 |
+
"learning_rate": 9.35e-06,
|
| 5699 |
+
"loss": 0.0006,
|
| 5700 |
+
"step": 8130
|
| 5701 |
+
},
|
| 5702 |
+
{
|
| 5703 |
+
"epoch": 0.814,
|
| 5704 |
+
"grad_norm": 0.019426727667450905,
|
| 5705 |
+
"learning_rate": 9.3e-06,
|
| 5706 |
+
"loss": 0.0009,
|
| 5707 |
+
"step": 8140
|
| 5708 |
+
},
|
| 5709 |
+
{
|
| 5710 |
+
"epoch": 0.815,
|
| 5711 |
+
"grad_norm": 0.011564865708351135,
|
| 5712 |
+
"learning_rate": 9.25e-06,
|
| 5713 |
+
"loss": 0.0006,
|
| 5714 |
+
"step": 8150
|
| 5715 |
+
},
|
| 5716 |
+
{
|
| 5717 |
+
"epoch": 0.816,
|
| 5718 |
+
"grad_norm": 0.009036659263074398,
|
| 5719 |
+
"learning_rate": 9.2e-06,
|
| 5720 |
+
"loss": 0.0008,
|
| 5721 |
+
"step": 8160
|
| 5722 |
+
},
|
| 5723 |
+
{
|
| 5724 |
+
"epoch": 0.817,
|
| 5725 |
+
"grad_norm": 0.006685588974505663,
|
| 5726 |
+
"learning_rate": 9.15e-06,
|
| 5727 |
+
"loss": 0.0007,
|
| 5728 |
+
"step": 8170
|
| 5729 |
+
},
|
| 5730 |
+
{
|
| 5731 |
+
"epoch": 0.818,
|
| 5732 |
+
"grad_norm": 0.005980687215924263,
|
| 5733 |
+
"learning_rate": 9.100000000000001e-06,
|
| 5734 |
+
"loss": 0.0005,
|
| 5735 |
+
"step": 8180
|
| 5736 |
+
},
|
| 5737 |
+
{
|
| 5738 |
+
"epoch": 0.819,
|
| 5739 |
+
"grad_norm": 0.0029402158688753843,
|
| 5740 |
+
"learning_rate": 9.05e-06,
|
| 5741 |
+
"loss": 0.0005,
|
| 5742 |
+
"step": 8190
|
| 5743 |
+
},
|
| 5744 |
+
{
|
| 5745 |
+
"epoch": 0.82,
|
| 5746 |
+
"grad_norm": 0.0034720194526016712,
|
| 5747 |
+
"learning_rate": 9e-06,
|
| 5748 |
+
"loss": 0.0006,
|
| 5749 |
+
"step": 8200
|
| 5750 |
+
},
|
| 5751 |
+
{
|
| 5752 |
+
"epoch": 0.821,
|
| 5753 |
+
"grad_norm": 0.008967465721070766,
|
| 5754 |
+
"learning_rate": 8.95e-06,
|
| 5755 |
+
"loss": 0.0009,
|
| 5756 |
+
"step": 8210
|
| 5757 |
+
},
|
| 5758 |
+
{
|
| 5759 |
+
"epoch": 0.822,
|
| 5760 |
+
"grad_norm": 0.007418784312903881,
|
| 5761 |
+
"learning_rate": 8.9e-06,
|
| 5762 |
+
"loss": 0.0007,
|
| 5763 |
+
"step": 8220
|
| 5764 |
+
},
|
| 5765 |
+
{
|
| 5766 |
+
"epoch": 0.823,
|
| 5767 |
+
"grad_norm": 0.0077253603376448154,
|
| 5768 |
+
"learning_rate": 8.85e-06,
|
| 5769 |
+
"loss": 0.0006,
|
| 5770 |
+
"step": 8230
|
| 5771 |
+
},
|
| 5772 |
+
{
|
| 5773 |
+
"epoch": 0.824,
|
| 5774 |
+
"grad_norm": 0.011202674359083176,
|
| 5775 |
+
"learning_rate": 8.8e-06,
|
| 5776 |
+
"loss": 0.0013,
|
| 5777 |
+
"step": 8240
|
| 5778 |
+
},
|
| 5779 |
+
{
|
| 5780 |
+
"epoch": 0.825,
|
| 5781 |
+
"grad_norm": 0.022354573011398315,
|
| 5782 |
+
"learning_rate": 8.75e-06,
|
| 5783 |
+
"loss": 0.0008,
|
| 5784 |
+
"step": 8250
|
| 5785 |
+
},
|
| 5786 |
+
{
|
| 5787 |
+
"epoch": 0.826,
|
| 5788 |
+
"grad_norm": 0.01750505343079567,
|
| 5789 |
+
"learning_rate": 8.7e-06,
|
| 5790 |
+
"loss": 0.0013,
|
| 5791 |
+
"step": 8260
|
| 5792 |
+
},
|
| 5793 |
+
{
|
| 5794 |
+
"epoch": 0.827,
|
| 5795 |
+
"grad_norm": 0.01153852604329586,
|
| 5796 |
+
"learning_rate": 8.65e-06,
|
| 5797 |
+
"loss": 0.0009,
|
| 5798 |
+
"step": 8270
|
| 5799 |
+
},
|
| 5800 |
+
{
|
| 5801 |
+
"epoch": 0.828,
|
| 5802 |
+
"grad_norm": 0.008752427063882351,
|
| 5803 |
+
"learning_rate": 8.599999999999999e-06,
|
| 5804 |
+
"loss": 0.0006,
|
| 5805 |
+
"step": 8280
|
| 5806 |
+
},
|
| 5807 |
+
{
|
| 5808 |
+
"epoch": 0.829,
|
| 5809 |
+
"grad_norm": 0.007307702675461769,
|
| 5810 |
+
"learning_rate": 8.550000000000001e-06,
|
| 5811 |
+
"loss": 0.0007,
|
| 5812 |
+
"step": 8290
|
| 5813 |
+
},
|
| 5814 |
+
{
|
| 5815 |
+
"epoch": 0.83,
|
| 5816 |
+
"grad_norm": 0.0077101094648242,
|
| 5817 |
+
"learning_rate": 8.500000000000002e-06,
|
| 5818 |
+
"loss": 0.0006,
|
| 5819 |
+
"step": 8300
|
| 5820 |
+
},
|
| 5821 |
+
{
|
| 5822 |
+
"epoch": 0.831,
|
| 5823 |
+
"grad_norm": 0.006358897779136896,
|
| 5824 |
+
"learning_rate": 8.45e-06,
|
| 5825 |
+
"loss": 0.0005,
|
| 5826 |
+
"step": 8310
|
| 5827 |
+
},
|
| 5828 |
+
{
|
| 5829 |
+
"epoch": 0.832,
|
| 5830 |
+
"grad_norm": 0.003663134528324008,
|
| 5831 |
+
"learning_rate": 8.400000000000001e-06,
|
| 5832 |
+
"loss": 0.0006,
|
| 5833 |
+
"step": 8320
|
| 5834 |
+
},
|
| 5835 |
+
{
|
| 5836 |
+
"epoch": 0.833,
|
| 5837 |
+
"grad_norm": 0.005117372144013643,
|
| 5838 |
+
"learning_rate": 8.350000000000001e-06,
|
| 5839 |
+
"loss": 0.0006,
|
| 5840 |
+
"step": 8330
|
| 5841 |
+
},
|
| 5842 |
+
{
|
| 5843 |
+
"epoch": 0.834,
|
| 5844 |
+
"grad_norm": 0.004245636984705925,
|
| 5845 |
+
"learning_rate": 8.3e-06,
|
| 5846 |
+
"loss": 0.0005,
|
| 5847 |
+
"step": 8340
|
| 5848 |
+
},
|
| 5849 |
+
{
|
| 5850 |
+
"epoch": 0.835,
|
| 5851 |
+
"grad_norm": 0.005357146263122559,
|
| 5852 |
+
"learning_rate": 8.25e-06,
|
| 5853 |
+
"loss": 0.0006,
|
| 5854 |
+
"step": 8350
|
| 5855 |
+
},
|
| 5856 |
+
{
|
| 5857 |
+
"epoch": 0.836,
|
| 5858 |
+
"grad_norm": 0.01055213250219822,
|
| 5859 |
+
"learning_rate": 8.200000000000001e-06,
|
| 5860 |
+
"loss": 0.0008,
|
| 5861 |
+
"step": 8360
|
| 5862 |
+
},
|
| 5863 |
+
{
|
| 5864 |
+
"epoch": 0.837,
|
| 5865 |
+
"grad_norm": 0.01871907152235508,
|
| 5866 |
+
"learning_rate": 8.15e-06,
|
| 5867 |
+
"loss": 0.0007,
|
| 5868 |
+
"step": 8370
|
| 5869 |
+
},
|
| 5870 |
+
{
|
| 5871 |
+
"epoch": 0.838,
|
| 5872 |
+
"grad_norm": 0.013110162690281868,
|
| 5873 |
+
"learning_rate": 8.1e-06,
|
| 5874 |
+
"loss": 0.0005,
|
| 5875 |
+
"step": 8380
|
| 5876 |
+
},
|
| 5877 |
+
{
|
| 5878 |
+
"epoch": 0.839,
|
| 5879 |
+
"grad_norm": 0.005271353758871555,
|
| 5880 |
+
"learning_rate": 8.050000000000001e-06,
|
| 5881 |
+
"loss": 0.0007,
|
| 5882 |
+
"step": 8390
|
| 5883 |
+
},
|
| 5884 |
+
{
|
| 5885 |
+
"epoch": 0.84,
|
| 5886 |
+
"grad_norm": 0.004324494861066341,
|
| 5887 |
+
"learning_rate": 8.000000000000001e-06,
|
| 5888 |
+
"loss": 0.0005,
|
| 5889 |
+
"step": 8400
|
| 5890 |
+
},
|
| 5891 |
+
{
|
| 5892 |
+
"epoch": 0.841,
|
| 5893 |
+
"grad_norm": 0.0031851409003138542,
|
| 5894 |
+
"learning_rate": 7.95e-06,
|
| 5895 |
+
"loss": 0.0006,
|
| 5896 |
+
"step": 8410
|
| 5897 |
+
},
|
| 5898 |
+
{
|
| 5899 |
+
"epoch": 0.842,
|
| 5900 |
+
"grad_norm": 0.009736557491123676,
|
| 5901 |
+
"learning_rate": 7.9e-06,
|
| 5902 |
+
"loss": 0.0006,
|
| 5903 |
+
"step": 8420
|
| 5904 |
+
},
|
| 5905 |
+
{
|
| 5906 |
+
"epoch": 0.843,
|
| 5907 |
+
"grad_norm": 0.005168536212295294,
|
| 5908 |
+
"learning_rate": 7.850000000000001e-06,
|
| 5909 |
+
"loss": 0.0005,
|
| 5910 |
+
"step": 8430
|
| 5911 |
+
},
|
| 5912 |
+
{
|
| 5913 |
+
"epoch": 0.844,
|
| 5914 |
+
"grad_norm": 0.002579685300588608,
|
| 5915 |
+
"learning_rate": 7.8e-06,
|
| 5916 |
+
"loss": 0.0005,
|
| 5917 |
+
"step": 8440
|
| 5918 |
+
},
|
| 5919 |
+
{
|
| 5920 |
+
"epoch": 0.845,
|
| 5921 |
+
"grad_norm": 0.008710252121090889,
|
| 5922 |
+
"learning_rate": 7.75e-06,
|
| 5923 |
+
"loss": 0.0005,
|
| 5924 |
+
"step": 8450
|
| 5925 |
+
},
|
| 5926 |
+
{
|
| 5927 |
+
"epoch": 0.846,
|
| 5928 |
+
"grad_norm": 0.004952189512550831,
|
| 5929 |
+
"learning_rate": 7.7e-06,
|
| 5930 |
+
"loss": 0.0008,
|
| 5931 |
+
"step": 8460
|
| 5932 |
+
},
|
| 5933 |
+
{
|
| 5934 |
+
"epoch": 0.847,
|
| 5935 |
+
"grad_norm": 0.003375423140823841,
|
| 5936 |
+
"learning_rate": 7.65e-06,
|
| 5937 |
+
"loss": 0.0005,
|
| 5938 |
+
"step": 8470
|
| 5939 |
+
},
|
| 5940 |
+
{
|
| 5941 |
+
"epoch": 0.848,
|
| 5942 |
+
"grad_norm": 0.13184253871440887,
|
| 5943 |
+
"learning_rate": 7.6e-06,
|
| 5944 |
+
"loss": 0.0012,
|
| 5945 |
+
"step": 8480
|
| 5946 |
+
},
|
| 5947 |
+
{
|
| 5948 |
+
"epoch": 0.849,
|
| 5949 |
+
"grad_norm": 0.017549166455864906,
|
| 5950 |
+
"learning_rate": 7.55e-06,
|
| 5951 |
+
"loss": 0.0007,
|
| 5952 |
+
"step": 8490
|
| 5953 |
+
},
|
| 5954 |
+
{
|
| 5955 |
+
"epoch": 0.85,
|
| 5956 |
+
"grad_norm": 0.00852286908775568,
|
| 5957 |
+
"learning_rate": 7.5e-06,
|
| 5958 |
+
"loss": 0.0006,
|
| 5959 |
+
"step": 8500
|
| 5960 |
+
},
|
| 5961 |
+
{
|
| 5962 |
+
"epoch": 0.851,
|
| 5963 |
+
"grad_norm": 0.005547389388084412,
|
| 5964 |
+
"learning_rate": 7.45e-06,
|
| 5965 |
+
"loss": 0.0005,
|
| 5966 |
+
"step": 8510
|
| 5967 |
+
},
|
| 5968 |
+
{
|
| 5969 |
+
"epoch": 0.852,
|
| 5970 |
+
"grad_norm": 0.0061622606590390205,
|
| 5971 |
+
"learning_rate": 7.4e-06,
|
| 5972 |
+
"loss": 0.0005,
|
| 5973 |
+
"step": 8520
|
| 5974 |
+
},
|
| 5975 |
+
{
|
| 5976 |
+
"epoch": 0.853,
|
| 5977 |
+
"grad_norm": 0.005182339809834957,
|
| 5978 |
+
"learning_rate": 7.35e-06,
|
| 5979 |
+
"loss": 0.0008,
|
| 5980 |
+
"step": 8530
|
| 5981 |
+
},
|
| 5982 |
+
{
|
| 5983 |
+
"epoch": 0.854,
|
| 5984 |
+
"grad_norm": 0.005366960074752569,
|
| 5985 |
+
"learning_rate": 7.2999999999999996e-06,
|
| 5986 |
+
"loss": 0.0006,
|
| 5987 |
+
"step": 8540
|
| 5988 |
+
},
|
| 5989 |
+
{
|
| 5990 |
+
"epoch": 0.855,
|
| 5991 |
+
"grad_norm": 0.005542315077036619,
|
| 5992 |
+
"learning_rate": 7.25e-06,
|
| 5993 |
+
"loss": 0.0006,
|
| 5994 |
+
"step": 8550
|
| 5995 |
+
},
|
| 5996 |
+
{
|
| 5997 |
+
"epoch": 0.856,
|
| 5998 |
+
"grad_norm": 0.003940809518098831,
|
| 5999 |
+
"learning_rate": 7.2e-06,
|
| 6000 |
+
"loss": 0.0005,
|
| 6001 |
+
"step": 8560
|
| 6002 |
+
},
|
| 6003 |
+
{
|
| 6004 |
+
"epoch": 0.857,
|
| 6005 |
+
"grad_norm": 0.003730529686436057,
|
| 6006 |
+
"learning_rate": 7.15e-06,
|
| 6007 |
+
"loss": 0.0006,
|
| 6008 |
+
"step": 8570
|
| 6009 |
+
},
|
| 6010 |
+
{
|
| 6011 |
+
"epoch": 0.858,
|
| 6012 |
+
"grad_norm": 0.0033961348235607147,
|
| 6013 |
+
"learning_rate": 7.1e-06,
|
| 6014 |
+
"loss": 0.0005,
|
| 6015 |
+
"step": 8580
|
| 6016 |
+
},
|
| 6017 |
+
{
|
| 6018 |
+
"epoch": 0.859,
|
| 6019 |
+
"grad_norm": 0.004546662792563438,
|
| 6020 |
+
"learning_rate": 7.049999999999999e-06,
|
| 6021 |
+
"loss": 0.0006,
|
| 6022 |
+
"step": 8590
|
| 6023 |
+
},
|
| 6024 |
+
{
|
| 6025 |
+
"epoch": 0.86,
|
| 6026 |
+
"grad_norm": 0.009168008342385292,
|
| 6027 |
+
"learning_rate": 7.000000000000001e-06,
|
| 6028 |
+
"loss": 0.0005,
|
| 6029 |
+
"step": 8600
|
| 6030 |
+
},
|
| 6031 |
+
{
|
| 6032 |
+
"epoch": 0.861,
|
| 6033 |
+
"grad_norm": 0.008373426273465157,
|
| 6034 |
+
"learning_rate": 6.950000000000001e-06,
|
| 6035 |
+
"loss": 0.0008,
|
| 6036 |
+
"step": 8610
|
| 6037 |
+
},
|
| 6038 |
+
{
|
| 6039 |
+
"epoch": 0.862,
|
| 6040 |
+
"grad_norm": 0.004947313107550144,
|
| 6041 |
+
"learning_rate": 6.900000000000001e-06,
|
| 6042 |
+
"loss": 0.0006,
|
| 6043 |
+
"step": 8620
|
| 6044 |
+
},
|
| 6045 |
+
{
|
| 6046 |
+
"epoch": 0.863,
|
| 6047 |
+
"grad_norm": 0.015127859078347683,
|
| 6048 |
+
"learning_rate": 6.8500000000000005e-06,
|
| 6049 |
+
"loss": 0.0006,
|
| 6050 |
+
"step": 8630
|
| 6051 |
+
},
|
| 6052 |
+
{
|
| 6053 |
+
"epoch": 0.864,
|
| 6054 |
+
"grad_norm": 0.0056435600854456425,
|
| 6055 |
+
"learning_rate": 6.800000000000001e-06,
|
| 6056 |
+
"loss": 0.0006,
|
| 6057 |
+
"step": 8640
|
| 6058 |
+
},
|
| 6059 |
+
{
|
| 6060 |
+
"epoch": 0.865,
|
| 6061 |
+
"grad_norm": 0.004109732341021299,
|
| 6062 |
+
"learning_rate": 6.750000000000001e-06,
|
| 6063 |
+
"loss": 0.0005,
|
| 6064 |
+
"step": 8650
|
| 6065 |
+
},
|
| 6066 |
+
{
|
| 6067 |
+
"epoch": 0.866,
|
| 6068 |
+
"grad_norm": 0.006170314736664295,
|
| 6069 |
+
"learning_rate": 6.700000000000001e-06,
|
| 6070 |
+
"loss": 0.0005,
|
| 6071 |
+
"step": 8660
|
| 6072 |
+
},
|
| 6073 |
+
{
|
| 6074 |
+
"epoch": 0.867,
|
| 6075 |
+
"grad_norm": 0.002802550094202161,
|
| 6076 |
+
"learning_rate": 6.650000000000001e-06,
|
| 6077 |
+
"loss": 0.0005,
|
| 6078 |
+
"step": 8670
|
| 6079 |
+
},
|
| 6080 |
+
{
|
| 6081 |
+
"epoch": 0.868,
|
| 6082 |
+
"grad_norm": 0.0029788350220769644,
|
| 6083 |
+
"learning_rate": 6.6e-06,
|
| 6084 |
+
"loss": 0.0004,
|
| 6085 |
+
"step": 8680
|
| 6086 |
+
},
|
| 6087 |
+
{
|
| 6088 |
+
"epoch": 0.869,
|
| 6089 |
+
"grad_norm": 0.013022363185882568,
|
| 6090 |
+
"learning_rate": 6.550000000000001e-06,
|
| 6091 |
+
"loss": 0.0006,
|
| 6092 |
+
"step": 8690
|
| 6093 |
+
},
|
| 6094 |
+
{
|
| 6095 |
+
"epoch": 0.87,
|
| 6096 |
+
"grad_norm": 0.0036853367928415537,
|
| 6097 |
+
"learning_rate": 6.5000000000000004e-06,
|
| 6098 |
+
"loss": 0.0006,
|
| 6099 |
+
"step": 8700
|
| 6100 |
+
},
|
| 6101 |
+
{
|
| 6102 |
+
"epoch": 0.871,
|
| 6103 |
+
"grad_norm": 0.002578242914751172,
|
| 6104 |
+
"learning_rate": 6.45e-06,
|
| 6105 |
+
"loss": 0.0005,
|
| 6106 |
+
"step": 8710
|
| 6107 |
+
},
|
| 6108 |
+
{
|
| 6109 |
+
"epoch": 0.872,
|
| 6110 |
+
"grad_norm": 0.0036895396187901497,
|
| 6111 |
+
"learning_rate": 6.4000000000000006e-06,
|
| 6112 |
+
"loss": 0.0005,
|
| 6113 |
+
"step": 8720
|
| 6114 |
+
},
|
| 6115 |
+
{
|
| 6116 |
+
"epoch": 0.873,
|
| 6117 |
+
"grad_norm": 0.006020987406373024,
|
| 6118 |
+
"learning_rate": 6.35e-06,
|
| 6119 |
+
"loss": 0.0005,
|
| 6120 |
+
"step": 8730
|
| 6121 |
+
},
|
| 6122 |
+
{
|
| 6123 |
+
"epoch": 0.874,
|
| 6124 |
+
"grad_norm": 0.006671608425676823,
|
| 6125 |
+
"learning_rate": 6.300000000000001e-06,
|
| 6126 |
+
"loss": 0.0006,
|
| 6127 |
+
"step": 8740
|
| 6128 |
+
},
|
| 6129 |
+
{
|
| 6130 |
+
"epoch": 0.875,
|
| 6131 |
+
"grad_norm": 0.0038102639373391867,
|
| 6132 |
+
"learning_rate": 6.25e-06,
|
| 6133 |
+
"loss": 0.0006,
|
| 6134 |
+
"step": 8750
|
| 6135 |
+
},
|
| 6136 |
+
{
|
| 6137 |
+
"epoch": 0.876,
|
| 6138 |
+
"grad_norm": 0.006786294747143984,
|
| 6139 |
+
"learning_rate": 6.2e-06,
|
| 6140 |
+
"loss": 0.0004,
|
| 6141 |
+
"step": 8760
|
| 6142 |
+
},
|
| 6143 |
+
{
|
| 6144 |
+
"epoch": 0.877,
|
| 6145 |
+
"grad_norm": 0.00381205091252923,
|
| 6146 |
+
"learning_rate": 6.15e-06,
|
| 6147 |
+
"loss": 0.0004,
|
| 6148 |
+
"step": 8770
|
| 6149 |
+
},
|
| 6150 |
+
{
|
| 6151 |
+
"epoch": 0.878,
|
| 6152 |
+
"grad_norm": 0.007368630729615688,
|
| 6153 |
+
"learning_rate": 6.1e-06,
|
| 6154 |
+
"loss": 0.0005,
|
| 6155 |
+
"step": 8780
|
| 6156 |
+
},
|
| 6157 |
+
{
|
| 6158 |
+
"epoch": 0.879,
|
| 6159 |
+
"grad_norm": 0.0035172586794942617,
|
| 6160 |
+
"learning_rate": 6.0500000000000005e-06,
|
| 6161 |
+
"loss": 0.0006,
|
| 6162 |
+
"step": 8790
|
| 6163 |
+
},
|
| 6164 |
+
{
|
| 6165 |
+
"epoch": 0.88,
|
| 6166 |
+
"grad_norm": 0.005555720068514347,
|
| 6167 |
+
"learning_rate": 6e-06,
|
| 6168 |
+
"loss": 0.0007,
|
| 6169 |
+
"step": 8800
|
| 6170 |
+
},
|
| 6171 |
+
{
|
| 6172 |
+
"epoch": 0.881,
|
| 6173 |
+
"grad_norm": 0.0076825893484056,
|
| 6174 |
+
"learning_rate": 5.95e-06,
|
| 6175 |
+
"loss": 0.0005,
|
| 6176 |
+
"step": 8810
|
| 6177 |
+
},
|
| 6178 |
+
{
|
| 6179 |
+
"epoch": 0.882,
|
| 6180 |
+
"grad_norm": 0.0055446140468120575,
|
| 6181 |
+
"learning_rate": 5.9e-06,
|
| 6182 |
+
"loss": 0.0005,
|
| 6183 |
+
"step": 8820
|
| 6184 |
+
},
|
| 6185 |
+
{
|
| 6186 |
+
"epoch": 0.883,
|
| 6187 |
+
"grad_norm": 0.002265618182718754,
|
| 6188 |
+
"learning_rate": 5.850000000000001e-06,
|
| 6189 |
+
"loss": 0.0005,
|
| 6190 |
+
"step": 8830
|
| 6191 |
+
},
|
| 6192 |
+
{
|
| 6193 |
+
"epoch": 0.884,
|
| 6194 |
+
"grad_norm": 0.003428585361689329,
|
| 6195 |
+
"learning_rate": 5.8e-06,
|
| 6196 |
+
"loss": 0.0004,
|
| 6197 |
+
"step": 8840
|
| 6198 |
+
},
|
| 6199 |
+
{
|
| 6200 |
+
"epoch": 0.885,
|
| 6201 |
+
"grad_norm": 0.0044764927588403225,
|
| 6202 |
+
"learning_rate": 5.750000000000001e-06,
|
| 6203 |
+
"loss": 0.0005,
|
| 6204 |
+
"step": 8850
|
| 6205 |
+
},
|
| 6206 |
+
{
|
| 6207 |
+
"epoch": 0.886,
|
| 6208 |
+
"grad_norm": 0.003201392712071538,
|
| 6209 |
+
"learning_rate": 5.7000000000000005e-06,
|
| 6210 |
+
"loss": 0.0005,
|
| 6211 |
+
"step": 8860
|
| 6212 |
+
},
|
| 6213 |
+
{
|
| 6214 |
+
"epoch": 0.887,
|
| 6215 |
+
"grad_norm": 0.0029762780759483576,
|
| 6216 |
+
"learning_rate": 5.65e-06,
|
| 6217 |
+
"loss": 0.0006,
|
| 6218 |
+
"step": 8870
|
| 6219 |
+
},
|
| 6220 |
+
{
|
| 6221 |
+
"epoch": 0.888,
|
| 6222 |
+
"grad_norm": 0.07450267672538757,
|
| 6223 |
+
"learning_rate": 5.600000000000001e-06,
|
| 6224 |
+
"loss": 0.0009,
|
| 6225 |
+
"step": 8880
|
| 6226 |
+
},
|
| 6227 |
+
{
|
| 6228 |
+
"epoch": 0.889,
|
| 6229 |
+
"grad_norm": 0.006392148323357105,
|
| 6230 |
+
"learning_rate": 5.55e-06,
|
| 6231 |
+
"loss": 0.0006,
|
| 6232 |
+
"step": 8890
|
| 6233 |
+
},
|
| 6234 |
+
{
|
| 6235 |
+
"epoch": 0.89,
|
| 6236 |
+
"grad_norm": 0.0038995451759546995,
|
| 6237 |
+
"learning_rate": 5.500000000000001e-06,
|
| 6238 |
+
"loss": 0.0005,
|
| 6239 |
+
"step": 8900
|
| 6240 |
+
},
|
| 6241 |
+
{
|
| 6242 |
+
"epoch": 0.891,
|
| 6243 |
+
"grad_norm": 0.0028438065201044083,
|
| 6244 |
+
"learning_rate": 5.45e-06,
|
| 6245 |
+
"loss": 0.0004,
|
| 6246 |
+
"step": 8910
|
| 6247 |
+
},
|
| 6248 |
+
{
|
| 6249 |
+
"epoch": 0.892,
|
| 6250 |
+
"grad_norm": 0.003168331226333976,
|
| 6251 |
+
"learning_rate": 5.4e-06,
|
| 6252 |
+
"loss": 0.0004,
|
| 6253 |
+
"step": 8920
|
| 6254 |
+
},
|
| 6255 |
+
{
|
| 6256 |
+
"epoch": 0.893,
|
| 6257 |
+
"grad_norm": 0.0026163198053836823,
|
| 6258 |
+
"learning_rate": 5.3500000000000004e-06,
|
| 6259 |
+
"loss": 0.0004,
|
| 6260 |
+
"step": 8930
|
| 6261 |
+
},
|
| 6262 |
+
{
|
| 6263 |
+
"epoch": 0.894,
|
| 6264 |
+
"grad_norm": 0.0029086521826684475,
|
| 6265 |
+
"learning_rate": 5.3e-06,
|
| 6266 |
+
"loss": 0.0005,
|
| 6267 |
+
"step": 8940
|
| 6268 |
+
},
|
| 6269 |
+
{
|
| 6270 |
+
"epoch": 0.895,
|
| 6271 |
+
"grad_norm": 0.011433840729296207,
|
| 6272 |
+
"learning_rate": 5.25e-06,
|
| 6273 |
+
"loss": 0.0007,
|
| 6274 |
+
"step": 8950
|
| 6275 |
+
},
|
| 6276 |
+
{
|
| 6277 |
+
"epoch": 0.896,
|
| 6278 |
+
"grad_norm": 0.01782575435936451,
|
| 6279 |
+
"learning_rate": 5.2e-06,
|
| 6280 |
+
"loss": 0.0011,
|
| 6281 |
+
"step": 8960
|
| 6282 |
+
},
|
| 6283 |
+
{
|
| 6284 |
+
"epoch": 0.897,
|
| 6285 |
+
"grad_norm": 0.00613692682236433,
|
| 6286 |
+
"learning_rate": 5.15e-06,
|
| 6287 |
+
"loss": 0.0004,
|
| 6288 |
+
"step": 8970
|
| 6289 |
+
},
|
| 6290 |
+
{
|
| 6291 |
+
"epoch": 0.898,
|
| 6292 |
+
"grad_norm": 0.02408697083592415,
|
| 6293 |
+
"learning_rate": 5.1e-06,
|
| 6294 |
+
"loss": 0.0007,
|
| 6295 |
+
"step": 8980
|
| 6296 |
+
},
|
| 6297 |
+
{
|
| 6298 |
+
"epoch": 0.899,
|
| 6299 |
+
"grad_norm": 0.004028539173305035,
|
| 6300 |
+
"learning_rate": 5.050000000000001e-06,
|
| 6301 |
+
"loss": 0.0005,
|
| 6302 |
+
"step": 8990
|
| 6303 |
+
},
|
| 6304 |
+
{
|
| 6305 |
+
"epoch": 0.9,
|
| 6306 |
+
"grad_norm": 0.0032080088276416063,
|
| 6307 |
+
"learning_rate": 5e-06,
|
| 6308 |
+
"loss": 0.0005,
|
| 6309 |
+
"step": 9000
|
| 6310 |
+
},
|
| 6311 |
+
{
|
| 6312 |
+
"epoch": 0.901,
|
| 6313 |
+
"grad_norm": 0.0035681568551808596,
|
| 6314 |
+
"learning_rate": 4.950000000000001e-06,
|
| 6315 |
+
"loss": 0.0004,
|
| 6316 |
+
"step": 9010
|
| 6317 |
+
},
|
| 6318 |
+
{
|
| 6319 |
+
"epoch": 0.902,
|
| 6320 |
+
"grad_norm": 0.007591512985527515,
|
| 6321 |
+
"learning_rate": 4.9000000000000005e-06,
|
| 6322 |
+
"loss": 0.0005,
|
| 6323 |
+
"step": 9020
|
| 6324 |
+
},
|
| 6325 |
+
{
|
| 6326 |
+
"epoch": 0.903,
|
| 6327 |
+
"grad_norm": 0.004855870269238949,
|
| 6328 |
+
"learning_rate": 4.85e-06,
|
| 6329 |
+
"loss": 0.0004,
|
| 6330 |
+
"step": 9030
|
| 6331 |
+
},
|
| 6332 |
+
{
|
| 6333 |
+
"epoch": 0.904,
|
| 6334 |
+
"grad_norm": 0.004854188766330481,
|
| 6335 |
+
"learning_rate": 4.800000000000001e-06,
|
| 6336 |
+
"loss": 0.0004,
|
| 6337 |
+
"step": 9040
|
| 6338 |
+
},
|
| 6339 |
+
{
|
| 6340 |
+
"epoch": 0.905,
|
| 6341 |
+
"grad_norm": 0.004117886070162058,
|
| 6342 |
+
"learning_rate": 4.75e-06,
|
| 6343 |
+
"loss": 0.0005,
|
| 6344 |
+
"step": 9050
|
| 6345 |
+
},
|
| 6346 |
+
{
|
| 6347 |
+
"epoch": 0.906,
|
| 6348 |
+
"grad_norm": 0.0045243133790791035,
|
| 6349 |
+
"learning_rate": 4.7e-06,
|
| 6350 |
+
"loss": 0.0005,
|
| 6351 |
+
"step": 9060
|
| 6352 |
+
},
|
| 6353 |
+
{
|
| 6354 |
+
"epoch": 0.907,
|
| 6355 |
+
"grad_norm": 0.001863984507508576,
|
| 6356 |
+
"learning_rate": 4.65e-06,
|
| 6357 |
+
"loss": 0.0004,
|
| 6358 |
+
"step": 9070
|
| 6359 |
+
},
|
| 6360 |
+
{
|
| 6361 |
+
"epoch": 0.908,
|
| 6362 |
+
"grad_norm": 0.002472365740686655,
|
| 6363 |
+
"learning_rate": 4.6e-06,
|
| 6364 |
+
"loss": 0.0005,
|
| 6365 |
+
"step": 9080
|
| 6366 |
+
},
|
| 6367 |
+
{
|
| 6368 |
+
"epoch": 0.909,
|
| 6369 |
+
"grad_norm": 0.0020466954447329044,
|
| 6370 |
+
"learning_rate": 4.5500000000000005e-06,
|
| 6371 |
+
"loss": 0.0004,
|
| 6372 |
+
"step": 9090
|
| 6373 |
+
},
|
| 6374 |
+
{
|
| 6375 |
+
"epoch": 0.91,
|
| 6376 |
+
"grad_norm": 0.004180034622550011,
|
| 6377 |
+
"learning_rate": 4.5e-06,
|
| 6378 |
+
"loss": 0.0004,
|
| 6379 |
+
"step": 9100
|
| 6380 |
+
},
|
| 6381 |
+
{
|
| 6382 |
+
"epoch": 0.911,
|
| 6383 |
+
"grad_norm": 0.00341266137547791,
|
| 6384 |
+
"learning_rate": 4.45e-06,
|
| 6385 |
+
"loss": 0.0006,
|
| 6386 |
+
"step": 9110
|
| 6387 |
+
},
|
| 6388 |
+
{
|
| 6389 |
+
"epoch": 0.912,
|
| 6390 |
+
"grad_norm": 0.006567875389009714,
|
| 6391 |
+
"learning_rate": 4.4e-06,
|
| 6392 |
+
"loss": 0.0004,
|
| 6393 |
+
"step": 9120
|
| 6394 |
+
},
|
| 6395 |
+
{
|
| 6396 |
+
"epoch": 0.913,
|
| 6397 |
+
"grad_norm": 0.003975498490035534,
|
| 6398 |
+
"learning_rate": 4.35e-06,
|
| 6399 |
+
"loss": 0.0006,
|
| 6400 |
+
"step": 9130
|
| 6401 |
+
},
|
| 6402 |
+
{
|
| 6403 |
+
"epoch": 0.914,
|
| 6404 |
+
"grad_norm": 0.003391894046217203,
|
| 6405 |
+
"learning_rate": 4.2999999999999995e-06,
|
| 6406 |
+
"loss": 0.0006,
|
| 6407 |
+
"step": 9140
|
| 6408 |
+
},
|
| 6409 |
+
{
|
| 6410 |
+
"epoch": 0.915,
|
| 6411 |
+
"grad_norm": 0.005821021273732185,
|
| 6412 |
+
"learning_rate": 4.250000000000001e-06,
|
| 6413 |
+
"loss": 0.0004,
|
| 6414 |
+
"step": 9150
|
| 6415 |
+
},
|
| 6416 |
+
{
|
| 6417 |
+
"epoch": 0.916,
|
| 6418 |
+
"grad_norm": 0.0022448371164500713,
|
| 6419 |
+
"learning_rate": 4.2000000000000004e-06,
|
| 6420 |
+
"loss": 0.0004,
|
| 6421 |
+
"step": 9160
|
| 6422 |
+
},
|
| 6423 |
+
{
|
| 6424 |
+
"epoch": 0.917,
|
| 6425 |
+
"grad_norm": 0.003718709573149681,
|
| 6426 |
+
"learning_rate": 4.15e-06,
|
| 6427 |
+
"loss": 0.0004,
|
| 6428 |
+
"step": 9170
|
| 6429 |
+
},
|
| 6430 |
+
{
|
| 6431 |
+
"epoch": 0.918,
|
| 6432 |
+
"grad_norm": 0.008243223652243614,
|
| 6433 |
+
"learning_rate": 4.1000000000000006e-06,
|
| 6434 |
+
"loss": 0.0007,
|
| 6435 |
+
"step": 9180
|
| 6436 |
+
},
|
| 6437 |
+
{
|
| 6438 |
+
"epoch": 0.919,
|
| 6439 |
+
"grad_norm": 0.010773789137601852,
|
| 6440 |
+
"learning_rate": 4.05e-06,
|
| 6441 |
+
"loss": 0.0007,
|
| 6442 |
+
"step": 9190
|
| 6443 |
+
},
|
| 6444 |
+
{
|
| 6445 |
+
"epoch": 0.92,
|
| 6446 |
+
"grad_norm": 0.006589268799871206,
|
| 6447 |
+
"learning_rate": 4.000000000000001e-06,
|
| 6448 |
+
"loss": 0.0005,
|
| 6449 |
+
"step": 9200
|
| 6450 |
+
},
|
| 6451 |
+
{
|
| 6452 |
+
"epoch": 0.921,
|
| 6453 |
+
"grad_norm": 0.0026856744661927223,
|
| 6454 |
+
"learning_rate": 3.95e-06,
|
| 6455 |
+
"loss": 0.0004,
|
| 6456 |
+
"step": 9210
|
| 6457 |
+
},
|
| 6458 |
+
{
|
| 6459 |
+
"epoch": 0.922,
|
| 6460 |
+
"grad_norm": 0.012134186923503876,
|
| 6461 |
+
"learning_rate": 3.9e-06,
|
| 6462 |
+
"loss": 0.0005,
|
| 6463 |
+
"step": 9220
|
| 6464 |
+
},
|
| 6465 |
+
{
|
| 6466 |
+
"epoch": 0.923,
|
| 6467 |
+
"grad_norm": 0.004260225687175989,
|
| 6468 |
+
"learning_rate": 3.85e-06,
|
| 6469 |
+
"loss": 0.0005,
|
| 6470 |
+
"step": 9230
|
| 6471 |
+
},
|
| 6472 |
+
{
|
| 6473 |
+
"epoch": 0.924,
|
| 6474 |
+
"grad_norm": 0.0023803950753062963,
|
| 6475 |
+
"learning_rate": 3.8e-06,
|
| 6476 |
+
"loss": 0.0004,
|
| 6477 |
+
"step": 9240
|
| 6478 |
+
},
|
| 6479 |
+
{
|
| 6480 |
+
"epoch": 0.925,
|
| 6481 |
+
"grad_norm": 0.0037502460181713104,
|
| 6482 |
+
"learning_rate": 3.75e-06,
|
| 6483 |
+
"loss": 0.0005,
|
| 6484 |
+
"step": 9250
|
| 6485 |
+
},
|
| 6486 |
+
{
|
| 6487 |
+
"epoch": 0.926,
|
| 6488 |
+
"grad_norm": 0.0017525887815281749,
|
| 6489 |
+
"learning_rate": 3.7e-06,
|
| 6490 |
+
"loss": 0.0003,
|
| 6491 |
+
"step": 9260
|
| 6492 |
+
},
|
| 6493 |
+
{
|
| 6494 |
+
"epoch": 0.927,
|
| 6495 |
+
"grad_norm": 0.003996537532657385,
|
| 6496 |
+
"learning_rate": 3.6499999999999998e-06,
|
| 6497 |
+
"loss": 0.0005,
|
| 6498 |
+
"step": 9270
|
| 6499 |
+
},
|
| 6500 |
+
{
|
| 6501 |
+
"epoch": 0.928,
|
| 6502 |
+
"grad_norm": 0.009158821776509285,
|
| 6503 |
+
"learning_rate": 3.6e-06,
|
| 6504 |
+
"loss": 0.0007,
|
| 6505 |
+
"step": 9280
|
| 6506 |
+
},
|
| 6507 |
+
{
|
| 6508 |
+
"epoch": 0.929,
|
| 6509 |
+
"grad_norm": 0.003372638253495097,
|
| 6510 |
+
"learning_rate": 3.55e-06,
|
| 6511 |
+
"loss": 0.0004,
|
| 6512 |
+
"step": 9290
|
| 6513 |
+
},
|
| 6514 |
+
{
|
| 6515 |
+
"epoch": 0.93,
|
| 6516 |
+
"grad_norm": 0.0026602360885590315,
|
| 6517 |
+
"learning_rate": 3.5000000000000004e-06,
|
| 6518 |
+
"loss": 0.0004,
|
| 6519 |
+
"step": 9300
|
| 6520 |
+
},
|
| 6521 |
+
{
|
| 6522 |
+
"epoch": 0.931,
|
| 6523 |
+
"grad_norm": 0.014532738365232944,
|
| 6524 |
+
"learning_rate": 3.4500000000000004e-06,
|
| 6525 |
+
"loss": 0.0007,
|
| 6526 |
+
"step": 9310
|
| 6527 |
+
},
|
| 6528 |
+
{
|
| 6529 |
+
"epoch": 0.932,
|
| 6530 |
+
"grad_norm": 0.002912462456151843,
|
| 6531 |
+
"learning_rate": 3.4000000000000005e-06,
|
| 6532 |
+
"loss": 0.0004,
|
| 6533 |
+
"step": 9320
|
| 6534 |
+
},
|
| 6535 |
+
{
|
| 6536 |
+
"epoch": 0.933,
|
| 6537 |
+
"grad_norm": 0.0052029709331691265,
|
| 6538 |
+
"learning_rate": 3.3500000000000005e-06,
|
| 6539 |
+
"loss": 0.0006,
|
| 6540 |
+
"step": 9330
|
| 6541 |
+
},
|
| 6542 |
+
{
|
| 6543 |
+
"epoch": 0.934,
|
| 6544 |
+
"grad_norm": 0.016220854595303535,
|
| 6545 |
+
"learning_rate": 3.3e-06,
|
| 6546 |
+
"loss": 0.0004,
|
| 6547 |
+
"step": 9340
|
| 6548 |
+
},
|
| 6549 |
+
{
|
| 6550 |
+
"epoch": 0.935,
|
| 6551 |
+
"grad_norm": 0.0030162036418914795,
|
| 6552 |
+
"learning_rate": 3.2500000000000002e-06,
|
| 6553 |
+
"loss": 0.0004,
|
| 6554 |
+
"step": 9350
|
| 6555 |
+
},
|
| 6556 |
+
{
|
| 6557 |
+
"epoch": 0.936,
|
| 6558 |
+
"grad_norm": 0.002491691382601857,
|
| 6559 |
+
"learning_rate": 3.2000000000000003e-06,
|
| 6560 |
+
"loss": 0.0004,
|
| 6561 |
+
"step": 9360
|
| 6562 |
+
},
|
| 6563 |
+
{
|
| 6564 |
+
"epoch": 0.937,
|
| 6565 |
+
"grad_norm": 0.022630969062447548,
|
| 6566 |
+
"learning_rate": 3.1500000000000003e-06,
|
| 6567 |
+
"loss": 0.0005,
|
| 6568 |
+
"step": 9370
|
| 6569 |
+
},
|
| 6570 |
+
{
|
| 6571 |
+
"epoch": 0.938,
|
| 6572 |
+
"grad_norm": 0.005951160565018654,
|
| 6573 |
+
"learning_rate": 3.1e-06,
|
| 6574 |
+
"loss": 0.0005,
|
| 6575 |
+
"step": 9380
|
| 6576 |
+
},
|
| 6577 |
+
{
|
| 6578 |
+
"epoch": 0.939,
|
| 6579 |
+
"grad_norm": 0.0024763622786849737,
|
| 6580 |
+
"learning_rate": 3.05e-06,
|
| 6581 |
+
"loss": 0.0005,
|
| 6582 |
+
"step": 9390
|
| 6583 |
+
},
|
| 6584 |
+
{
|
| 6585 |
+
"epoch": 0.94,
|
| 6586 |
+
"grad_norm": 0.0049979290924966335,
|
| 6587 |
+
"learning_rate": 3e-06,
|
| 6588 |
+
"loss": 0.0005,
|
| 6589 |
+
"step": 9400
|
| 6590 |
+
},
|
| 6591 |
+
{
|
| 6592 |
+
"epoch": 0.941,
|
| 6593 |
+
"grad_norm": 0.0025999436620622873,
|
| 6594 |
+
"learning_rate": 2.95e-06,
|
| 6595 |
+
"loss": 0.0004,
|
| 6596 |
+
"step": 9410
|
| 6597 |
+
},
|
| 6598 |
+
{
|
| 6599 |
+
"epoch": 0.942,
|
| 6600 |
+
"grad_norm": 0.004584169946610928,
|
| 6601 |
+
"learning_rate": 2.9e-06,
|
| 6602 |
+
"loss": 0.0006,
|
| 6603 |
+
"step": 9420
|
| 6604 |
+
},
|
| 6605 |
+
{
|
| 6606 |
+
"epoch": 0.943,
|
| 6607 |
+
"grad_norm": 0.005211680196225643,
|
| 6608 |
+
"learning_rate": 2.8500000000000002e-06,
|
| 6609 |
+
"loss": 0.0005,
|
| 6610 |
+
"step": 9430
|
| 6611 |
+
},
|
| 6612 |
+
{
|
| 6613 |
+
"epoch": 0.944,
|
| 6614 |
+
"grad_norm": 0.0022507943212985992,
|
| 6615 |
+
"learning_rate": 2.8000000000000003e-06,
|
| 6616 |
+
"loss": 0.0004,
|
| 6617 |
+
"step": 9440
|
| 6618 |
+
},
|
| 6619 |
+
{
|
| 6620 |
+
"epoch": 0.945,
|
| 6621 |
+
"grad_norm": 0.0029024691320955753,
|
| 6622 |
+
"learning_rate": 2.7500000000000004e-06,
|
| 6623 |
+
"loss": 0.0004,
|
| 6624 |
+
"step": 9450
|
| 6625 |
+
},
|
| 6626 |
+
{
|
| 6627 |
+
"epoch": 0.946,
|
| 6628 |
+
"grad_norm": 0.003968573175370693,
|
| 6629 |
+
"learning_rate": 2.7e-06,
|
| 6630 |
+
"loss": 0.0004,
|
| 6631 |
+
"step": 9460
|
| 6632 |
+
},
|
| 6633 |
+
{
|
| 6634 |
+
"epoch": 0.947,
|
| 6635 |
+
"grad_norm": 0.003264777595177293,
|
| 6636 |
+
"learning_rate": 2.65e-06,
|
| 6637 |
+
"loss": 0.0005,
|
| 6638 |
+
"step": 9470
|
| 6639 |
+
},
|
| 6640 |
+
{
|
| 6641 |
+
"epoch": 0.948,
|
| 6642 |
+
"grad_norm": 0.0048127188347280025,
|
| 6643 |
+
"learning_rate": 2.6e-06,
|
| 6644 |
+
"loss": 0.0004,
|
| 6645 |
+
"step": 9480
|
| 6646 |
+
},
|
| 6647 |
+
{
|
| 6648 |
+
"epoch": 0.949,
|
| 6649 |
+
"grad_norm": 0.004405410494655371,
|
| 6650 |
+
"learning_rate": 2.55e-06,
|
| 6651 |
+
"loss": 0.0006,
|
| 6652 |
+
"step": 9490
|
| 6653 |
+
},
|
| 6654 |
+
{
|
| 6655 |
+
"epoch": 0.95,
|
| 6656 |
+
"grad_norm": 0.00462340796366334,
|
| 6657 |
+
"learning_rate": 2.5e-06,
|
| 6658 |
+
"loss": 0.0004,
|
| 6659 |
+
"step": 9500
|
| 6660 |
+
},
|
| 6661 |
+
{
|
| 6662 |
+
"epoch": 0.951,
|
| 6663 |
+
"grad_norm": 0.0021721452940255404,
|
| 6664 |
+
"learning_rate": 2.4500000000000003e-06,
|
| 6665 |
+
"loss": 0.0004,
|
| 6666 |
+
"step": 9510
|
| 6667 |
+
},
|
| 6668 |
+
{
|
| 6669 |
+
"epoch": 0.952,
|
| 6670 |
+
"grad_norm": 0.002355078933760524,
|
| 6671 |
+
"learning_rate": 2.4000000000000003e-06,
|
| 6672 |
+
"loss": 0.0006,
|
| 6673 |
+
"step": 9520
|
| 6674 |
+
},
|
| 6675 |
+
{
|
| 6676 |
+
"epoch": 0.953,
|
| 6677 |
+
"grad_norm": 0.0022414589766412973,
|
| 6678 |
+
"learning_rate": 2.35e-06,
|
| 6679 |
+
"loss": 0.0004,
|
| 6680 |
+
"step": 9530
|
| 6681 |
+
},
|
| 6682 |
+
{
|
| 6683 |
+
"epoch": 0.954,
|
| 6684 |
+
"grad_norm": 0.012005253694951534,
|
| 6685 |
+
"learning_rate": 2.3e-06,
|
| 6686 |
+
"loss": 0.0006,
|
| 6687 |
+
"step": 9540
|
| 6688 |
+
},
|
| 6689 |
+
{
|
| 6690 |
+
"epoch": 0.955,
|
| 6691 |
+
"grad_norm": 0.00513832364231348,
|
| 6692 |
+
"learning_rate": 2.25e-06,
|
| 6693 |
+
"loss": 0.0005,
|
| 6694 |
+
"step": 9550
|
| 6695 |
+
},
|
| 6696 |
+
{
|
| 6697 |
+
"epoch": 0.956,
|
| 6698 |
+
"grad_norm": 0.0027625642251223326,
|
| 6699 |
+
"learning_rate": 2.2e-06,
|
| 6700 |
+
"loss": 0.0005,
|
| 6701 |
+
"step": 9560
|
| 6702 |
+
},
|
| 6703 |
+
{
|
| 6704 |
+
"epoch": 0.957,
|
| 6705 |
+
"grad_norm": 0.008645957335829735,
|
| 6706 |
+
"learning_rate": 2.1499999999999997e-06,
|
| 6707 |
+
"loss": 0.0005,
|
| 6708 |
+
"step": 9570
|
| 6709 |
+
},
|
| 6710 |
+
{
|
| 6711 |
+
"epoch": 0.958,
|
| 6712 |
+
"grad_norm": 0.00188863230869174,
|
| 6713 |
+
"learning_rate": 2.1000000000000002e-06,
|
| 6714 |
+
"loss": 0.0004,
|
| 6715 |
+
"step": 9580
|
| 6716 |
+
},
|
| 6717 |
+
{
|
| 6718 |
+
"epoch": 0.959,
|
| 6719 |
+
"grad_norm": 0.0025561931543052197,
|
| 6720 |
+
"learning_rate": 2.0500000000000003e-06,
|
| 6721 |
+
"loss": 0.0004,
|
| 6722 |
+
"step": 9590
|
| 6723 |
+
},
|
| 6724 |
+
{
|
| 6725 |
+
"epoch": 0.96,
|
| 6726 |
+
"grad_norm": 0.0033618698362261057,
|
| 6727 |
+
"learning_rate": 2.0000000000000003e-06,
|
| 6728 |
+
"loss": 0.0004,
|
| 6729 |
+
"step": 9600
|
| 6730 |
+
},
|
| 6731 |
+
{
|
| 6732 |
+
"epoch": 0.961,
|
| 6733 |
+
"grad_norm": 0.0018735548947006464,
|
| 6734 |
+
"learning_rate": 1.95e-06,
|
| 6735 |
+
"loss": 0.0004,
|
| 6736 |
+
"step": 9610
|
| 6737 |
+
},
|
| 6738 |
+
{
|
| 6739 |
+
"epoch": 0.962,
|
| 6740 |
+
"grad_norm": 0.0019359014695510268,
|
| 6741 |
+
"learning_rate": 1.9e-06,
|
| 6742 |
+
"loss": 0.0005,
|
| 6743 |
+
"step": 9620
|
| 6744 |
+
},
|
| 6745 |
+
{
|
| 6746 |
+
"epoch": 0.963,
|
| 6747 |
+
"grad_norm": 0.005369434133172035,
|
| 6748 |
+
"learning_rate": 1.85e-06,
|
| 6749 |
+
"loss": 0.0004,
|
| 6750 |
+
"step": 9630
|
| 6751 |
+
},
|
| 6752 |
+
{
|
| 6753 |
+
"epoch": 0.964,
|
| 6754 |
+
"grad_norm": 0.0017576682148501277,
|
| 6755 |
+
"learning_rate": 1.8e-06,
|
| 6756 |
+
"loss": 0.0004,
|
| 6757 |
+
"step": 9640
|
| 6758 |
+
},
|
| 6759 |
+
{
|
| 6760 |
+
"epoch": 0.965,
|
| 6761 |
+
"grad_norm": 0.002633103635162115,
|
| 6762 |
+
"learning_rate": 1.7500000000000002e-06,
|
| 6763 |
+
"loss": 0.0004,
|
| 6764 |
+
"step": 9650
|
| 6765 |
+
},
|
| 6766 |
+
{
|
| 6767 |
+
"epoch": 0.966,
|
| 6768 |
+
"grad_norm": 0.007023205049335957,
|
| 6769 |
+
"learning_rate": 1.7000000000000002e-06,
|
| 6770 |
+
"loss": 0.0004,
|
| 6771 |
+
"step": 9660
|
| 6772 |
+
},
|
| 6773 |
+
{
|
| 6774 |
+
"epoch": 0.967,
|
| 6775 |
+
"grad_norm": 0.0026062438264489174,
|
| 6776 |
+
"learning_rate": 1.65e-06,
|
| 6777 |
+
"loss": 0.0005,
|
| 6778 |
+
"step": 9670
|
| 6779 |
+
},
|
| 6780 |
+
{
|
| 6781 |
+
"epoch": 0.968,
|
| 6782 |
+
"grad_norm": 0.0025111304130405188,
|
| 6783 |
+
"learning_rate": 1.6000000000000001e-06,
|
| 6784 |
+
"loss": 0.0005,
|
| 6785 |
+
"step": 9680
|
| 6786 |
+
},
|
| 6787 |
+
{
|
| 6788 |
+
"epoch": 0.969,
|
| 6789 |
+
"grad_norm": 0.0028218806255608797,
|
| 6790 |
+
"learning_rate": 1.55e-06,
|
| 6791 |
+
"loss": 0.0004,
|
| 6792 |
+
"step": 9690
|
| 6793 |
+
},
|
| 6794 |
+
{
|
| 6795 |
+
"epoch": 0.97,
|
| 6796 |
+
"grad_norm": 0.0024802633561193943,
|
| 6797 |
+
"learning_rate": 1.5e-06,
|
| 6798 |
+
"loss": 0.0005,
|
| 6799 |
+
"step": 9700
|
| 6800 |
+
},
|
| 6801 |
+
{
|
| 6802 |
+
"epoch": 0.971,
|
| 6803 |
+
"grad_norm": 0.002883223118260503,
|
| 6804 |
+
"learning_rate": 1.45e-06,
|
| 6805 |
+
"loss": 0.0005,
|
| 6806 |
+
"step": 9710
|
| 6807 |
+
},
|
| 6808 |
+
{
|
| 6809 |
+
"epoch": 0.972,
|
| 6810 |
+
"grad_norm": 0.002503247233107686,
|
| 6811 |
+
"learning_rate": 1.4000000000000001e-06,
|
| 6812 |
+
"loss": 0.0005,
|
| 6813 |
+
"step": 9720
|
| 6814 |
+
},
|
| 6815 |
+
{
|
| 6816 |
+
"epoch": 0.973,
|
| 6817 |
+
"grad_norm": 0.002495008986443281,
|
| 6818 |
+
"learning_rate": 1.35e-06,
|
| 6819 |
+
"loss": 0.0006,
|
| 6820 |
+
"step": 9730
|
| 6821 |
+
},
|
| 6822 |
+
{
|
| 6823 |
+
"epoch": 0.974,
|
| 6824 |
+
"grad_norm": 0.03429775312542915,
|
| 6825 |
+
"learning_rate": 1.3e-06,
|
| 6826 |
+
"loss": 0.0006,
|
| 6827 |
+
"step": 9740
|
| 6828 |
+
},
|
| 6829 |
+
{
|
| 6830 |
+
"epoch": 0.975,
|
| 6831 |
+
"grad_norm": 0.003482217201963067,
|
| 6832 |
+
"learning_rate": 1.25e-06,
|
| 6833 |
+
"loss": 0.0004,
|
| 6834 |
+
"step": 9750
|
| 6835 |
+
},
|
| 6836 |
+
{
|
| 6837 |
+
"epoch": 0.976,
|
| 6838 |
+
"grad_norm": 0.001837963704019785,
|
| 6839 |
+
"learning_rate": 1.2000000000000002e-06,
|
| 6840 |
+
"loss": 0.0004,
|
| 6841 |
+
"step": 9760
|
| 6842 |
+
},
|
| 6843 |
+
{
|
| 6844 |
+
"epoch": 0.977,
|
| 6845 |
+
"grad_norm": 0.0020507893059402704,
|
| 6846 |
+
"learning_rate": 1.15e-06,
|
| 6847 |
+
"loss": 0.0004,
|
| 6848 |
+
"step": 9770
|
| 6849 |
+
},
|
| 6850 |
+
{
|
| 6851 |
+
"epoch": 0.978,
|
| 6852 |
+
"grad_norm": 0.0022647210862487555,
|
| 6853 |
+
"learning_rate": 1.1e-06,
|
| 6854 |
+
"loss": 0.0005,
|
| 6855 |
+
"step": 9780
|
| 6856 |
+
},
|
| 6857 |
+
{
|
| 6858 |
+
"epoch": 0.979,
|
| 6859 |
+
"grad_norm": 0.0017425378318876028,
|
| 6860 |
+
"learning_rate": 1.0500000000000001e-06,
|
| 6861 |
+
"loss": 0.0004,
|
| 6862 |
+
"step": 9790
|
| 6863 |
+
},
|
| 6864 |
+
{
|
| 6865 |
+
"epoch": 0.98,
|
| 6866 |
+
"grad_norm": 0.2319187968969345,
|
| 6867 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 6868 |
+
"loss": 0.0021,
|
| 6869 |
+
"step": 9800
|
| 6870 |
+
},
|
| 6871 |
+
{
|
| 6872 |
+
"epoch": 0.981,
|
| 6873 |
+
"grad_norm": 0.01799739897251129,
|
| 6874 |
+
"learning_rate": 9.5e-07,
|
| 6875 |
+
"loss": 0.0006,
|
| 6876 |
+
"step": 9810
|
| 6877 |
+
},
|
| 6878 |
+
{
|
| 6879 |
+
"epoch": 0.982,
|
| 6880 |
+
"grad_norm": 0.007147952448576689,
|
| 6881 |
+
"learning_rate": 9e-07,
|
| 6882 |
+
"loss": 0.0005,
|
| 6883 |
+
"step": 9820
|
| 6884 |
+
},
|
| 6885 |
+
{
|
| 6886 |
+
"epoch": 0.983,
|
| 6887 |
+
"grad_norm": 0.004181794356554747,
|
| 6888 |
+
"learning_rate": 8.500000000000001e-07,
|
| 6889 |
+
"loss": 0.0005,
|
| 6890 |
+
"step": 9830
|
| 6891 |
+
},
|
| 6892 |
+
{
|
| 6893 |
+
"epoch": 0.984,
|
| 6894 |
+
"grad_norm": 0.00277232495136559,
|
| 6895 |
+
"learning_rate": 8.000000000000001e-07,
|
| 6896 |
+
"loss": 0.0004,
|
| 6897 |
+
"step": 9840
|
| 6898 |
+
},
|
| 6899 |
+
{
|
| 6900 |
+
"epoch": 0.985,
|
| 6901 |
+
"grad_norm": 0.0024797001387923956,
|
| 6902 |
+
"learning_rate": 7.5e-07,
|
| 6903 |
+
"loss": 0.0006,
|
| 6904 |
+
"step": 9850
|
| 6905 |
+
},
|
| 6906 |
+
{
|
| 6907 |
+
"epoch": 0.986,
|
| 6908 |
+
"grad_norm": 0.002748242113739252,
|
| 6909 |
+
"learning_rate": 7.000000000000001e-07,
|
| 6910 |
+
"loss": 0.0005,
|
| 6911 |
+
"step": 9860
|
| 6912 |
+
},
|
| 6913 |
+
{
|
| 6914 |
+
"epoch": 0.987,
|
| 6915 |
+
"grad_norm": 0.002988820429891348,
|
| 6916 |
+
"learning_rate": 6.5e-07,
|
| 6917 |
+
"loss": 0.0004,
|
| 6918 |
+
"step": 9870
|
| 6919 |
+
},
|
| 6920 |
+
{
|
| 6921 |
+
"epoch": 0.988,
|
| 6922 |
+
"grad_norm": 0.002272873418405652,
|
| 6923 |
+
"learning_rate": 6.000000000000001e-07,
|
| 6924 |
+
"loss": 0.0006,
|
| 6925 |
+
"step": 9880
|
| 6926 |
+
},
|
| 6927 |
+
{
|
| 6928 |
+
"epoch": 0.989,
|
| 6929 |
+
"grad_norm": 0.0028824047185480595,
|
| 6930 |
+
"learning_rate": 5.5e-07,
|
| 6931 |
+
"loss": 0.0005,
|
| 6932 |
+
"step": 9890
|
| 6933 |
+
},
|
| 6934 |
+
{
|
| 6935 |
+
"epoch": 0.99,
|
| 6936 |
+
"grad_norm": 0.013895529322326183,
|
| 6937 |
+
"learning_rate": 5.000000000000001e-07,
|
| 6938 |
+
"loss": 0.0005,
|
| 6939 |
+
"step": 9900
|
| 6940 |
+
},
|
| 6941 |
+
{
|
| 6942 |
+
"epoch": 0.991,
|
| 6943 |
+
"grad_norm": 0.004210934974253178,
|
| 6944 |
+
"learning_rate": 4.5e-07,
|
| 6945 |
+
"loss": 0.0004,
|
| 6946 |
+
"step": 9910
|
| 6947 |
+
},
|
| 6948 |
+
{
|
| 6949 |
+
"epoch": 0.992,
|
| 6950 |
+
"grad_norm": 0.0017349456902593374,
|
| 6951 |
+
"learning_rate": 4.0000000000000003e-07,
|
| 6952 |
+
"loss": 0.0005,
|
| 6953 |
+
"step": 9920
|
| 6954 |
+
},
|
| 6955 |
+
{
|
| 6956 |
+
"epoch": 0.993,
|
| 6957 |
+
"grad_norm": 0.0036622195038944483,
|
| 6958 |
+
"learning_rate": 3.5000000000000004e-07,
|
| 6959 |
+
"loss": 0.0003,
|
| 6960 |
+
"step": 9930
|
| 6961 |
+
},
|
| 6962 |
+
{
|
| 6963 |
+
"epoch": 0.994,
|
| 6964 |
+
"grad_norm": 0.02928483486175537,
|
| 6965 |
+
"learning_rate": 3.0000000000000004e-07,
|
| 6966 |
+
"loss": 0.0006,
|
| 6967 |
+
"step": 9940
|
| 6968 |
+
},
|
| 6969 |
+
{
|
| 6970 |
+
"epoch": 0.995,
|
| 6971 |
+
"grad_norm": 0.004271595273166895,
|
| 6972 |
+
"learning_rate": 2.5000000000000004e-07,
|
| 6973 |
+
"loss": 0.0004,
|
| 6974 |
+
"step": 9950
|
| 6975 |
+
},
|
| 6976 |
+
{
|
| 6977 |
+
"epoch": 0.996,
|
| 6978 |
+
"grad_norm": 0.004935207776725292,
|
| 6979 |
+
"learning_rate": 2.0000000000000002e-07,
|
| 6980 |
+
"loss": 0.0004,
|
| 6981 |
+
"step": 9960
|
| 6982 |
+
},
|
| 6983 |
+
{
|
| 6984 |
+
"epoch": 0.997,
|
| 6985 |
+
"grad_norm": 0.005258087068796158,
|
| 6986 |
+
"learning_rate": 1.5000000000000002e-07,
|
| 6987 |
+
"loss": 0.0004,
|
| 6988 |
+
"step": 9970
|
| 6989 |
+
},
|
| 6990 |
+
{
|
| 6991 |
+
"epoch": 0.998,
|
| 6992 |
+
"grad_norm": 0.0014150363858789206,
|
| 6993 |
+
"learning_rate": 1.0000000000000001e-07,
|
| 6994 |
+
"loss": 0.0004,
|
| 6995 |
+
"step": 9980
|
| 6996 |
+
},
|
| 6997 |
+
{
|
| 6998 |
+
"epoch": 0.999,
|
| 6999 |
+
"grad_norm": 0.003183445893228054,
|
| 7000 |
+
"learning_rate": 5.0000000000000004e-08,
|
| 7001 |
+
"loss": 0.0004,
|
| 7002 |
+
"step": 9990
|
| 7003 |
+
},
|
| 7004 |
+
{
|
| 7005 |
+
"epoch": 1.0,
|
| 7006 |
+
"grad_norm": 0.0063432566821575165,
|
| 7007 |
+
"learning_rate": 0.0,
|
| 7008 |
+
"loss": 0.0004,
|
| 7009 |
+
"step": 10000
|
| 7010 |
}
|
| 7011 |
],
|
| 7012 |
"logging_steps": 10,
|
|
|
|
| 7021 |
"should_evaluate": false,
|
| 7022 |
"should_log": false,
|
| 7023 |
"should_save": true,
|
| 7024 |
+
"should_training_stop": true
|
| 7025 |
},
|
| 7026 |
"attributes": {}
|
| 7027 |
}
|
| 7028 |
},
|
| 7029 |
+
"total_flos": 3.6962203336704e+17,
|
| 7030 |
"train_batch_size": 1,
|
| 7031 |
"trial_name": null,
|
| 7032 |
"trial_params": null
|