Upload folder using huggingface_hub
Browse files- scheduler.pt +1 -1
- trainer_state.json +3503 -3
- training_args.bin +1 -1
scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dd99d00c768cf310dacabe02dfac656daa677ad8744ec5c7b2354f852657c6f8
|
| 3 |
size 1064
|
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": 1.9e-05,
|
| 3508 |
"loss": 5.5315,
|
| 3509 |
"step": 5000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3510 |
}
|
| 3511 |
],
|
| 3512 |
"logging_steps": 10,
|
|
@@ -3514,7 +7014,7 @@
|
|
| 3514 |
"num_input_tokens_seen": 0,
|
| 3515 |
"num_train_epochs": 1,
|
| 3516 |
"save_steps": 1000,
|
| 3517 |
-
"total_flos": 1.
|
| 3518 |
"train_batch_size": 4,
|
| 3519 |
"trial_name": null,
|
| 3520 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.0031237114690190297,
|
| 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": 1.9e-05,
|
| 3508 |
"loss": 5.5315,
|
| 3509 |
"step": 5000
|
| 3510 |
+
},
|
| 3511 |
+
{
|
| 3512 |
+
"epoch": 0.0015649794459785338,
|
| 3513 |
+
"grad_norm": 17.125,
|
| 3514 |
+
"learning_rate": 1.8998e-05,
|
| 3515 |
+
"loss": 5.4787,
|
| 3516 |
+
"step": 5010
|
| 3517 |
+
},
|
| 3518 |
+
{
|
| 3519 |
+
"epoch": 0.001568103157447553,
|
| 3520 |
+
"grad_norm": 12.0625,
|
| 3521 |
+
"learning_rate": 1.8996000000000002e-05,
|
| 3522 |
+
"loss": 5.6937,
|
| 3523 |
+
"step": 5020
|
| 3524 |
+
},
|
| 3525 |
+
{
|
| 3526 |
+
"epoch": 0.001571226868916572,
|
| 3527 |
+
"grad_norm": 14.8125,
|
| 3528 |
+
"learning_rate": 1.8994e-05,
|
| 3529 |
+
"loss": 5.29,
|
| 3530 |
+
"step": 5030
|
| 3531 |
+
},
|
| 3532 |
+
{
|
| 3533 |
+
"epoch": 0.001574350580385591,
|
| 3534 |
+
"grad_norm": 22.375,
|
| 3535 |
+
"learning_rate": 1.8992e-05,
|
| 3536 |
+
"loss": 5.4227,
|
| 3537 |
+
"step": 5040
|
| 3538 |
+
},
|
| 3539 |
+
{
|
| 3540 |
+
"epoch": 0.00157747429185461,
|
| 3541 |
+
"grad_norm": 14.1875,
|
| 3542 |
+
"learning_rate": 1.8990000000000003e-05,
|
| 3543 |
+
"loss": 5.4948,
|
| 3544 |
+
"step": 5050
|
| 3545 |
+
},
|
| 3546 |
+
{
|
| 3547 |
+
"epoch": 0.001580598003323629,
|
| 3548 |
+
"grad_norm": 18.625,
|
| 3549 |
+
"learning_rate": 1.8988000000000002e-05,
|
| 3550 |
+
"loss": 5.6641,
|
| 3551 |
+
"step": 5060
|
| 3552 |
+
},
|
| 3553 |
+
{
|
| 3554 |
+
"epoch": 0.001583721714792648,
|
| 3555 |
+
"grad_norm": 21.375,
|
| 3556 |
+
"learning_rate": 1.8986e-05,
|
| 3557 |
+
"loss": 5.6905,
|
| 3558 |
+
"step": 5070
|
| 3559 |
+
},
|
| 3560 |
+
{
|
| 3561 |
+
"epoch": 0.0015868454262616672,
|
| 3562 |
+
"grad_norm": 21.75,
|
| 3563 |
+
"learning_rate": 1.8984000000000003e-05,
|
| 3564 |
+
"loss": 5.2701,
|
| 3565 |
+
"step": 5080
|
| 3566 |
+
},
|
| 3567 |
+
{
|
| 3568 |
+
"epoch": 0.0015899691377306861,
|
| 3569 |
+
"grad_norm": 18.625,
|
| 3570 |
+
"learning_rate": 1.8982000000000002e-05,
|
| 3571 |
+
"loss": 5.697,
|
| 3572 |
+
"step": 5090
|
| 3573 |
+
},
|
| 3574 |
+
{
|
| 3575 |
+
"epoch": 0.001593092849199705,
|
| 3576 |
+
"grad_norm": 20.0,
|
| 3577 |
+
"learning_rate": 1.898e-05,
|
| 3578 |
+
"loss": 5.455,
|
| 3579 |
+
"step": 5100
|
| 3580 |
+
},
|
| 3581 |
+
{
|
| 3582 |
+
"epoch": 0.0015962165606687242,
|
| 3583 |
+
"grad_norm": 16.125,
|
| 3584 |
+
"learning_rate": 1.8978e-05,
|
| 3585 |
+
"loss": 5.4842,
|
| 3586 |
+
"step": 5110
|
| 3587 |
+
},
|
| 3588 |
+
{
|
| 3589 |
+
"epoch": 0.0015993402721377432,
|
| 3590 |
+
"grad_norm": 15.5,
|
| 3591 |
+
"learning_rate": 1.8976000000000003e-05,
|
| 3592 |
+
"loss": 5.2105,
|
| 3593 |
+
"step": 5120
|
| 3594 |
+
},
|
| 3595 |
+
{
|
| 3596 |
+
"epoch": 0.0016024639836067622,
|
| 3597 |
+
"grad_norm": 19.125,
|
| 3598 |
+
"learning_rate": 1.8974e-05,
|
| 3599 |
+
"loss": 5.7378,
|
| 3600 |
+
"step": 5130
|
| 3601 |
+
},
|
| 3602 |
+
{
|
| 3603 |
+
"epoch": 0.0016055876950757813,
|
| 3604 |
+
"grad_norm": 15.625,
|
| 3605 |
+
"learning_rate": 1.8972e-05,
|
| 3606 |
+
"loss": 5.7174,
|
| 3607 |
+
"step": 5140
|
| 3608 |
+
},
|
| 3609 |
+
{
|
| 3610 |
+
"epoch": 0.0016087114065448003,
|
| 3611 |
+
"grad_norm": 14.4375,
|
| 3612 |
+
"learning_rate": 1.8970000000000003e-05,
|
| 3613 |
+
"loss": 5.6989,
|
| 3614 |
+
"step": 5150
|
| 3615 |
+
},
|
| 3616 |
+
{
|
| 3617 |
+
"epoch": 0.0016118351180138193,
|
| 3618 |
+
"grad_norm": 16.125,
|
| 3619 |
+
"learning_rate": 1.8968000000000002e-05,
|
| 3620 |
+
"loss": 5.7208,
|
| 3621 |
+
"step": 5160
|
| 3622 |
+
},
|
| 3623 |
+
{
|
| 3624 |
+
"epoch": 0.0016149588294828384,
|
| 3625 |
+
"grad_norm": 14.375,
|
| 3626 |
+
"learning_rate": 1.8966e-05,
|
| 3627 |
+
"loss": 5.5058,
|
| 3628 |
+
"step": 5170
|
| 3629 |
+
},
|
| 3630 |
+
{
|
| 3631 |
+
"epoch": 0.0016180825409518574,
|
| 3632 |
+
"grad_norm": 16.5,
|
| 3633 |
+
"learning_rate": 1.8964000000000003e-05,
|
| 3634 |
+
"loss": 5.2135,
|
| 3635 |
+
"step": 5180
|
| 3636 |
+
},
|
| 3637 |
+
{
|
| 3638 |
+
"epoch": 0.0016212062524208763,
|
| 3639 |
+
"grad_norm": 18.125,
|
| 3640 |
+
"learning_rate": 1.8962000000000002e-05,
|
| 3641 |
+
"loss": 5.6446,
|
| 3642 |
+
"step": 5190
|
| 3643 |
+
},
|
| 3644 |
+
{
|
| 3645 |
+
"epoch": 0.0016243299638898955,
|
| 3646 |
+
"grad_norm": 17.375,
|
| 3647 |
+
"learning_rate": 1.896e-05,
|
| 3648 |
+
"loss": 5.7546,
|
| 3649 |
+
"step": 5200
|
| 3650 |
+
},
|
| 3651 |
+
{
|
| 3652 |
+
"epoch": 0.0016274536753589145,
|
| 3653 |
+
"grad_norm": 13.125,
|
| 3654 |
+
"learning_rate": 1.8958e-05,
|
| 3655 |
+
"loss": 5.4918,
|
| 3656 |
+
"step": 5210
|
| 3657 |
+
},
|
| 3658 |
+
{
|
| 3659 |
+
"epoch": 0.0016305773868279334,
|
| 3660 |
+
"grad_norm": 21.375,
|
| 3661 |
+
"learning_rate": 1.8956e-05,
|
| 3662 |
+
"loss": 5.4425,
|
| 3663 |
+
"step": 5220
|
| 3664 |
+
},
|
| 3665 |
+
{
|
| 3666 |
+
"epoch": 0.0016337010982969526,
|
| 3667 |
+
"grad_norm": 13.5625,
|
| 3668 |
+
"learning_rate": 1.8954000000000002e-05,
|
| 3669 |
+
"loss": 5.5864,
|
| 3670 |
+
"step": 5230
|
| 3671 |
+
},
|
| 3672 |
+
{
|
| 3673 |
+
"epoch": 0.0016368248097659716,
|
| 3674 |
+
"grad_norm": 18.5,
|
| 3675 |
+
"learning_rate": 1.8952e-05,
|
| 3676 |
+
"loss": 5.7043,
|
| 3677 |
+
"step": 5240
|
| 3678 |
+
},
|
| 3679 |
+
{
|
| 3680 |
+
"epoch": 0.0016399485212349905,
|
| 3681 |
+
"grad_norm": 18.625,
|
| 3682 |
+
"learning_rate": 1.8950000000000003e-05,
|
| 3683 |
+
"loss": 5.5245,
|
| 3684 |
+
"step": 5250
|
| 3685 |
+
},
|
| 3686 |
+
{
|
| 3687 |
+
"epoch": 0.0016430722327040097,
|
| 3688 |
+
"grad_norm": 13.375,
|
| 3689 |
+
"learning_rate": 1.8948000000000002e-05,
|
| 3690 |
+
"loss": 5.5455,
|
| 3691 |
+
"step": 5260
|
| 3692 |
+
},
|
| 3693 |
+
{
|
| 3694 |
+
"epoch": 0.0016461959441730286,
|
| 3695 |
+
"grad_norm": 17.125,
|
| 3696 |
+
"learning_rate": 1.8946e-05,
|
| 3697 |
+
"loss": 5.5937,
|
| 3698 |
+
"step": 5270
|
| 3699 |
+
},
|
| 3700 |
+
{
|
| 3701 |
+
"epoch": 0.0016493196556420476,
|
| 3702 |
+
"grad_norm": 22.0,
|
| 3703 |
+
"learning_rate": 1.8944000000000004e-05,
|
| 3704 |
+
"loss": 5.4067,
|
| 3705 |
+
"step": 5280
|
| 3706 |
+
},
|
| 3707 |
+
{
|
| 3708 |
+
"epoch": 0.0016524433671110668,
|
| 3709 |
+
"grad_norm": 18.125,
|
| 3710 |
+
"learning_rate": 1.8942000000000003e-05,
|
| 3711 |
+
"loss": 5.7337,
|
| 3712 |
+
"step": 5290
|
| 3713 |
+
},
|
| 3714 |
+
{
|
| 3715 |
+
"epoch": 0.0016555670785800857,
|
| 3716 |
+
"grad_norm": 14.25,
|
| 3717 |
+
"learning_rate": 1.894e-05,
|
| 3718 |
+
"loss": 5.5917,
|
| 3719 |
+
"step": 5300
|
| 3720 |
+
},
|
| 3721 |
+
{
|
| 3722 |
+
"epoch": 0.0016586907900491047,
|
| 3723 |
+
"grad_norm": 17.25,
|
| 3724 |
+
"learning_rate": 1.8938e-05,
|
| 3725 |
+
"loss": 5.4728,
|
| 3726 |
+
"step": 5310
|
| 3727 |
+
},
|
| 3728 |
+
{
|
| 3729 |
+
"epoch": 0.0016618145015181239,
|
| 3730 |
+
"grad_norm": 15.4375,
|
| 3731 |
+
"learning_rate": 1.8936e-05,
|
| 3732 |
+
"loss": 5.5482,
|
| 3733 |
+
"step": 5320
|
| 3734 |
+
},
|
| 3735 |
+
{
|
| 3736 |
+
"epoch": 0.0016649382129871428,
|
| 3737 |
+
"grad_norm": 20.25,
|
| 3738 |
+
"learning_rate": 1.8934000000000002e-05,
|
| 3739 |
+
"loss": 5.6185,
|
| 3740 |
+
"step": 5330
|
| 3741 |
+
},
|
| 3742 |
+
{
|
| 3743 |
+
"epoch": 0.0016680619244561618,
|
| 3744 |
+
"grad_norm": 14.6875,
|
| 3745 |
+
"learning_rate": 1.8932e-05,
|
| 3746 |
+
"loss": 5.7582,
|
| 3747 |
+
"step": 5340
|
| 3748 |
+
},
|
| 3749 |
+
{
|
| 3750 |
+
"epoch": 0.001671185635925181,
|
| 3751 |
+
"grad_norm": 20.875,
|
| 3752 |
+
"learning_rate": 1.893e-05,
|
| 3753 |
+
"loss": 5.733,
|
| 3754 |
+
"step": 5350
|
| 3755 |
+
},
|
| 3756 |
+
{
|
| 3757 |
+
"epoch": 0.0016743093473942,
|
| 3758 |
+
"grad_norm": 11.625,
|
| 3759 |
+
"learning_rate": 1.8928000000000002e-05,
|
| 3760 |
+
"loss": 5.754,
|
| 3761 |
+
"step": 5360
|
| 3762 |
+
},
|
| 3763 |
+
{
|
| 3764 |
+
"epoch": 0.0016774330588632189,
|
| 3765 |
+
"grad_norm": 14.6875,
|
| 3766 |
+
"learning_rate": 1.8926e-05,
|
| 3767 |
+
"loss": 5.5967,
|
| 3768 |
+
"step": 5370
|
| 3769 |
+
},
|
| 3770 |
+
{
|
| 3771 |
+
"epoch": 0.001680556770332238,
|
| 3772 |
+
"grad_norm": 15.375,
|
| 3773 |
+
"learning_rate": 1.8924000000000004e-05,
|
| 3774 |
+
"loss": 5.5708,
|
| 3775 |
+
"step": 5380
|
| 3776 |
+
},
|
| 3777 |
+
{
|
| 3778 |
+
"epoch": 0.001683680481801257,
|
| 3779 |
+
"grad_norm": 19.625,
|
| 3780 |
+
"learning_rate": 1.8922000000000003e-05,
|
| 3781 |
+
"loss": 5.4091,
|
| 3782 |
+
"step": 5390
|
| 3783 |
+
},
|
| 3784 |
+
{
|
| 3785 |
+
"epoch": 0.001686804193270276,
|
| 3786 |
+
"grad_norm": 14.125,
|
| 3787 |
+
"learning_rate": 1.8920000000000002e-05,
|
| 3788 |
+
"loss": 5.7056,
|
| 3789 |
+
"step": 5400
|
| 3790 |
+
},
|
| 3791 |
+
{
|
| 3792 |
+
"epoch": 0.0016899279047392951,
|
| 3793 |
+
"grad_norm": 14.1875,
|
| 3794 |
+
"learning_rate": 1.8918e-05,
|
| 3795 |
+
"loss": 5.7095,
|
| 3796 |
+
"step": 5410
|
| 3797 |
+
},
|
| 3798 |
+
{
|
| 3799 |
+
"epoch": 0.001693051616208314,
|
| 3800 |
+
"grad_norm": 17.625,
|
| 3801 |
+
"learning_rate": 1.8916e-05,
|
| 3802 |
+
"loss": 5.632,
|
| 3803 |
+
"step": 5420
|
| 3804 |
+
},
|
| 3805 |
+
{
|
| 3806 |
+
"epoch": 0.001696175327677333,
|
| 3807 |
+
"grad_norm": 17.0,
|
| 3808 |
+
"learning_rate": 1.8914000000000002e-05,
|
| 3809 |
+
"loss": 5.5599,
|
| 3810 |
+
"step": 5430
|
| 3811 |
+
},
|
| 3812 |
+
{
|
| 3813 |
+
"epoch": 0.0016992990391463522,
|
| 3814 |
+
"grad_norm": 14.5625,
|
| 3815 |
+
"learning_rate": 1.8912e-05,
|
| 3816 |
+
"loss": 5.6591,
|
| 3817 |
+
"step": 5440
|
| 3818 |
+
},
|
| 3819 |
+
{
|
| 3820 |
+
"epoch": 0.0017024227506153712,
|
| 3821 |
+
"grad_norm": 14.9375,
|
| 3822 |
+
"learning_rate": 1.891e-05,
|
| 3823 |
+
"loss": 5.5385,
|
| 3824 |
+
"step": 5450
|
| 3825 |
+
},
|
| 3826 |
+
{
|
| 3827 |
+
"epoch": 0.0017055464620843901,
|
| 3828 |
+
"grad_norm": 19.625,
|
| 3829 |
+
"learning_rate": 1.8908000000000003e-05,
|
| 3830 |
+
"loss": 5.4329,
|
| 3831 |
+
"step": 5460
|
| 3832 |
+
},
|
| 3833 |
+
{
|
| 3834 |
+
"epoch": 0.0017086701735534093,
|
| 3835 |
+
"grad_norm": 22.375,
|
| 3836 |
+
"learning_rate": 1.8906e-05,
|
| 3837 |
+
"loss": 5.2307,
|
| 3838 |
+
"step": 5470
|
| 3839 |
+
},
|
| 3840 |
+
{
|
| 3841 |
+
"epoch": 0.0017117938850224283,
|
| 3842 |
+
"grad_norm": 19.125,
|
| 3843 |
+
"learning_rate": 1.8904000000000004e-05,
|
| 3844 |
+
"loss": 5.2522,
|
| 3845 |
+
"step": 5480
|
| 3846 |
+
},
|
| 3847 |
+
{
|
| 3848 |
+
"epoch": 0.0017149175964914472,
|
| 3849 |
+
"grad_norm": 11.625,
|
| 3850 |
+
"learning_rate": 1.8902000000000003e-05,
|
| 3851 |
+
"loss": 4.9953,
|
| 3852 |
+
"step": 5490
|
| 3853 |
+
},
|
| 3854 |
+
{
|
| 3855 |
+
"epoch": 0.0017180413079604664,
|
| 3856 |
+
"grad_norm": 19.125,
|
| 3857 |
+
"learning_rate": 1.8900000000000002e-05,
|
| 3858 |
+
"loss": 5.6609,
|
| 3859 |
+
"step": 5500
|
| 3860 |
+
},
|
| 3861 |
+
{
|
| 3862 |
+
"epoch": 0.0017211650194294853,
|
| 3863 |
+
"grad_norm": 18.75,
|
| 3864 |
+
"learning_rate": 1.8898e-05,
|
| 3865 |
+
"loss": 5.4622,
|
| 3866 |
+
"step": 5510
|
| 3867 |
+
},
|
| 3868 |
+
{
|
| 3869 |
+
"epoch": 0.0017242887308985043,
|
| 3870 |
+
"grad_norm": 12.8125,
|
| 3871 |
+
"learning_rate": 1.8896e-05,
|
| 3872 |
+
"loss": 5.2735,
|
| 3873 |
+
"step": 5520
|
| 3874 |
+
},
|
| 3875 |
+
{
|
| 3876 |
+
"epoch": 0.0017274124423675235,
|
| 3877 |
+
"grad_norm": 13.5625,
|
| 3878 |
+
"learning_rate": 1.8894000000000002e-05,
|
| 3879 |
+
"loss": 5.6178,
|
| 3880 |
+
"step": 5530
|
| 3881 |
+
},
|
| 3882 |
+
{
|
| 3883 |
+
"epoch": 0.0017305361538365424,
|
| 3884 |
+
"grad_norm": 19.125,
|
| 3885 |
+
"learning_rate": 1.8892e-05,
|
| 3886 |
+
"loss": 5.0805,
|
| 3887 |
+
"step": 5540
|
| 3888 |
+
},
|
| 3889 |
+
{
|
| 3890 |
+
"epoch": 0.0017336598653055614,
|
| 3891 |
+
"grad_norm": 20.125,
|
| 3892 |
+
"learning_rate": 1.889e-05,
|
| 3893 |
+
"loss": 5.3146,
|
| 3894 |
+
"step": 5550
|
| 3895 |
+
},
|
| 3896 |
+
{
|
| 3897 |
+
"epoch": 0.0017367835767745806,
|
| 3898 |
+
"grad_norm": 11.375,
|
| 3899 |
+
"learning_rate": 1.8888000000000003e-05,
|
| 3900 |
+
"loss": 5.741,
|
| 3901 |
+
"step": 5560
|
| 3902 |
+
},
|
| 3903 |
+
{
|
| 3904 |
+
"epoch": 0.0017399072882435995,
|
| 3905 |
+
"grad_norm": 15.5625,
|
| 3906 |
+
"learning_rate": 1.8886000000000002e-05,
|
| 3907 |
+
"loss": 5.7295,
|
| 3908 |
+
"step": 5570
|
| 3909 |
+
},
|
| 3910 |
+
{
|
| 3911 |
+
"epoch": 0.0017430309997126185,
|
| 3912 |
+
"grad_norm": 12.1875,
|
| 3913 |
+
"learning_rate": 1.8884e-05,
|
| 3914 |
+
"loss": 5.4365,
|
| 3915 |
+
"step": 5580
|
| 3916 |
+
},
|
| 3917 |
+
{
|
| 3918 |
+
"epoch": 0.0017461547111816376,
|
| 3919 |
+
"grad_norm": 18.125,
|
| 3920 |
+
"learning_rate": 1.8882000000000003e-05,
|
| 3921 |
+
"loss": 5.6188,
|
| 3922 |
+
"step": 5590
|
| 3923 |
+
},
|
| 3924 |
+
{
|
| 3925 |
+
"epoch": 0.0017492784226506566,
|
| 3926 |
+
"grad_norm": 18.75,
|
| 3927 |
+
"learning_rate": 1.8880000000000002e-05,
|
| 3928 |
+
"loss": 5.4696,
|
| 3929 |
+
"step": 5600
|
| 3930 |
+
},
|
| 3931 |
+
{
|
| 3932 |
+
"epoch": 0.0017524021341196756,
|
| 3933 |
+
"grad_norm": 12.25,
|
| 3934 |
+
"learning_rate": 1.8878e-05,
|
| 3935 |
+
"loss": 5.5221,
|
| 3936 |
+
"step": 5610
|
| 3937 |
+
},
|
| 3938 |
+
{
|
| 3939 |
+
"epoch": 0.0017555258455886947,
|
| 3940 |
+
"grad_norm": 21.75,
|
| 3941 |
+
"learning_rate": 1.8876e-05,
|
| 3942 |
+
"loss": 5.1922,
|
| 3943 |
+
"step": 5620
|
| 3944 |
+
},
|
| 3945 |
+
{
|
| 3946 |
+
"epoch": 0.0017586495570577137,
|
| 3947 |
+
"grad_norm": 17.125,
|
| 3948 |
+
"learning_rate": 1.8874000000000003e-05,
|
| 3949 |
+
"loss": 5.4216,
|
| 3950 |
+
"step": 5630
|
| 3951 |
+
},
|
| 3952 |
+
{
|
| 3953 |
+
"epoch": 0.0017617732685267326,
|
| 3954 |
+
"grad_norm": 17.75,
|
| 3955 |
+
"learning_rate": 1.8872e-05,
|
| 3956 |
+
"loss": 5.4876,
|
| 3957 |
+
"step": 5640
|
| 3958 |
+
},
|
| 3959 |
+
{
|
| 3960 |
+
"epoch": 0.0017648969799957518,
|
| 3961 |
+
"grad_norm": 17.0,
|
| 3962 |
+
"learning_rate": 1.887e-05,
|
| 3963 |
+
"loss": 5.595,
|
| 3964 |
+
"step": 5650
|
| 3965 |
+
},
|
| 3966 |
+
{
|
| 3967 |
+
"epoch": 0.0017680206914647708,
|
| 3968 |
+
"grad_norm": 20.25,
|
| 3969 |
+
"learning_rate": 1.8868000000000003e-05,
|
| 3970 |
+
"loss": 5.6682,
|
| 3971 |
+
"step": 5660
|
| 3972 |
+
},
|
| 3973 |
+
{
|
| 3974 |
+
"epoch": 0.0017711444029337897,
|
| 3975 |
+
"grad_norm": 17.375,
|
| 3976 |
+
"learning_rate": 1.8866000000000002e-05,
|
| 3977 |
+
"loss": 5.2636,
|
| 3978 |
+
"step": 5670
|
| 3979 |
+
},
|
| 3980 |
+
{
|
| 3981 |
+
"epoch": 0.001774268114402809,
|
| 3982 |
+
"grad_norm": 14.875,
|
| 3983 |
+
"learning_rate": 1.8864e-05,
|
| 3984 |
+
"loss": 5.4282,
|
| 3985 |
+
"step": 5680
|
| 3986 |
+
},
|
| 3987 |
+
{
|
| 3988 |
+
"epoch": 0.0017773918258718279,
|
| 3989 |
+
"grad_norm": 18.375,
|
| 3990 |
+
"learning_rate": 1.8862000000000003e-05,
|
| 3991 |
+
"loss": 5.0176,
|
| 3992 |
+
"step": 5690
|
| 3993 |
+
},
|
| 3994 |
+
{
|
| 3995 |
+
"epoch": 0.0017805155373408468,
|
| 3996 |
+
"grad_norm": 16.375,
|
| 3997 |
+
"learning_rate": 1.886e-05,
|
| 3998 |
+
"loss": 5.6029,
|
| 3999 |
+
"step": 5700
|
| 4000 |
+
},
|
| 4001 |
+
{
|
| 4002 |
+
"epoch": 0.001783639248809866,
|
| 4003 |
+
"grad_norm": 9.875,
|
| 4004 |
+
"learning_rate": 1.8858e-05,
|
| 4005 |
+
"loss": 5.7118,
|
| 4006 |
+
"step": 5710
|
| 4007 |
+
},
|
| 4008 |
+
{
|
| 4009 |
+
"epoch": 0.001786762960278885,
|
| 4010 |
+
"grad_norm": 21.25,
|
| 4011 |
+
"learning_rate": 1.8856e-05,
|
| 4012 |
+
"loss": 5.0426,
|
| 4013 |
+
"step": 5720
|
| 4014 |
+
},
|
| 4015 |
+
{
|
| 4016 |
+
"epoch": 0.001789886671747904,
|
| 4017 |
+
"grad_norm": 20.5,
|
| 4018 |
+
"learning_rate": 1.8854000000000003e-05,
|
| 4019 |
+
"loss": 5.5434,
|
| 4020 |
+
"step": 5730
|
| 4021 |
+
},
|
| 4022 |
+
{
|
| 4023 |
+
"epoch": 0.001793010383216923,
|
| 4024 |
+
"grad_norm": 12.75,
|
| 4025 |
+
"learning_rate": 1.8852000000000002e-05,
|
| 4026 |
+
"loss": 5.665,
|
| 4027 |
+
"step": 5740
|
| 4028 |
+
},
|
| 4029 |
+
{
|
| 4030 |
+
"epoch": 0.001796134094685942,
|
| 4031 |
+
"grad_norm": 16.375,
|
| 4032 |
+
"learning_rate": 1.885e-05,
|
| 4033 |
+
"loss": 5.6768,
|
| 4034 |
+
"step": 5750
|
| 4035 |
+
},
|
| 4036 |
+
{
|
| 4037 |
+
"epoch": 0.001799257806154961,
|
| 4038 |
+
"grad_norm": 19.25,
|
| 4039 |
+
"learning_rate": 1.8848000000000003e-05,
|
| 4040 |
+
"loss": 5.3965,
|
| 4041 |
+
"step": 5760
|
| 4042 |
+
},
|
| 4043 |
+
{
|
| 4044 |
+
"epoch": 0.0018023815176239802,
|
| 4045 |
+
"grad_norm": 13.0625,
|
| 4046 |
+
"learning_rate": 1.8846000000000002e-05,
|
| 4047 |
+
"loss": 5.6074,
|
| 4048 |
+
"step": 5770
|
| 4049 |
+
},
|
| 4050 |
+
{
|
| 4051 |
+
"epoch": 0.0018055052290929991,
|
| 4052 |
+
"grad_norm": 20.875,
|
| 4053 |
+
"learning_rate": 1.8844e-05,
|
| 4054 |
+
"loss": 5.6579,
|
| 4055 |
+
"step": 5780
|
| 4056 |
+
},
|
| 4057 |
+
{
|
| 4058 |
+
"epoch": 0.001808628940562018,
|
| 4059 |
+
"grad_norm": 15.75,
|
| 4060 |
+
"learning_rate": 1.8842000000000004e-05,
|
| 4061 |
+
"loss": 5.5465,
|
| 4062 |
+
"step": 5790
|
| 4063 |
+
},
|
| 4064 |
+
{
|
| 4065 |
+
"epoch": 0.0018117526520310373,
|
| 4066 |
+
"grad_norm": 14.8125,
|
| 4067 |
+
"learning_rate": 1.884e-05,
|
| 4068 |
+
"loss": 5.542,
|
| 4069 |
+
"step": 5800
|
| 4070 |
+
},
|
| 4071 |
+
{
|
| 4072 |
+
"epoch": 0.0018148763635000562,
|
| 4073 |
+
"grad_norm": 14.3125,
|
| 4074 |
+
"learning_rate": 1.8838e-05,
|
| 4075 |
+
"loss": 5.7126,
|
| 4076 |
+
"step": 5810
|
| 4077 |
+
},
|
| 4078 |
+
{
|
| 4079 |
+
"epoch": 0.0018180000749690752,
|
| 4080 |
+
"grad_norm": 14.125,
|
| 4081 |
+
"learning_rate": 1.8836e-05,
|
| 4082 |
+
"loss": 5.6531,
|
| 4083 |
+
"step": 5820
|
| 4084 |
+
},
|
| 4085 |
+
{
|
| 4086 |
+
"epoch": 0.0018211237864380943,
|
| 4087 |
+
"grad_norm": 11.9375,
|
| 4088 |
+
"learning_rate": 1.8834e-05,
|
| 4089 |
+
"loss": 5.1662,
|
| 4090 |
+
"step": 5830
|
| 4091 |
+
},
|
| 4092 |
+
{
|
| 4093 |
+
"epoch": 0.0018242474979071133,
|
| 4094 |
+
"grad_norm": 18.625,
|
| 4095 |
+
"learning_rate": 1.8832000000000002e-05,
|
| 4096 |
+
"loss": 5.7536,
|
| 4097 |
+
"step": 5840
|
| 4098 |
+
},
|
| 4099 |
+
{
|
| 4100 |
+
"epoch": 0.0018273712093761323,
|
| 4101 |
+
"grad_norm": 13.625,
|
| 4102 |
+
"learning_rate": 1.883e-05,
|
| 4103 |
+
"loss": 5.7136,
|
| 4104 |
+
"step": 5850
|
| 4105 |
+
},
|
| 4106 |
+
{
|
| 4107 |
+
"epoch": 0.0018304949208451514,
|
| 4108 |
+
"grad_norm": 13.75,
|
| 4109 |
+
"learning_rate": 1.8828000000000003e-05,
|
| 4110 |
+
"loss": 5.4302,
|
| 4111 |
+
"step": 5860
|
| 4112 |
+
},
|
| 4113 |
+
{
|
| 4114 |
+
"epoch": 0.0018336186323141704,
|
| 4115 |
+
"grad_norm": 15.75,
|
| 4116 |
+
"learning_rate": 1.8826000000000002e-05,
|
| 4117 |
+
"loss": 5.3569,
|
| 4118 |
+
"step": 5870
|
| 4119 |
+
},
|
| 4120 |
+
{
|
| 4121 |
+
"epoch": 0.0018367423437831893,
|
| 4122 |
+
"grad_norm": 15.25,
|
| 4123 |
+
"learning_rate": 1.8824e-05,
|
| 4124 |
+
"loss": 4.9267,
|
| 4125 |
+
"step": 5880
|
| 4126 |
+
},
|
| 4127 |
+
{
|
| 4128 |
+
"epoch": 0.0018398660552522085,
|
| 4129 |
+
"grad_norm": 13.0,
|
| 4130 |
+
"learning_rate": 1.8822000000000004e-05,
|
| 4131 |
+
"loss": 5.2913,
|
| 4132 |
+
"step": 5890
|
| 4133 |
+
},
|
| 4134 |
+
{
|
| 4135 |
+
"epoch": 0.0018429897667212275,
|
| 4136 |
+
"grad_norm": 22.875,
|
| 4137 |
+
"learning_rate": 1.882e-05,
|
| 4138 |
+
"loss": 5.4008,
|
| 4139 |
+
"step": 5900
|
| 4140 |
+
},
|
| 4141 |
+
{
|
| 4142 |
+
"epoch": 0.0018461134781902464,
|
| 4143 |
+
"grad_norm": 15.1875,
|
| 4144 |
+
"learning_rate": 1.8818000000000002e-05,
|
| 4145 |
+
"loss": 5.378,
|
| 4146 |
+
"step": 5910
|
| 4147 |
+
},
|
| 4148 |
+
{
|
| 4149 |
+
"epoch": 0.0018492371896592656,
|
| 4150 |
+
"grad_norm": 11.375,
|
| 4151 |
+
"learning_rate": 1.8816e-05,
|
| 4152 |
+
"loss": 5.8182,
|
| 4153 |
+
"step": 5920
|
| 4154 |
+
},
|
| 4155 |
+
{
|
| 4156 |
+
"epoch": 0.0018523609011282846,
|
| 4157 |
+
"grad_norm": 18.0,
|
| 4158 |
+
"learning_rate": 1.8814e-05,
|
| 4159 |
+
"loss": 4.9918,
|
| 4160 |
+
"step": 5930
|
| 4161 |
+
},
|
| 4162 |
+
{
|
| 4163 |
+
"epoch": 0.0018554846125973035,
|
| 4164 |
+
"grad_norm": 12.4375,
|
| 4165 |
+
"learning_rate": 1.8812000000000002e-05,
|
| 4166 |
+
"loss": 5.6478,
|
| 4167 |
+
"step": 5940
|
| 4168 |
+
},
|
| 4169 |
+
{
|
| 4170 |
+
"epoch": 0.0018586083240663227,
|
| 4171 |
+
"grad_norm": 11.875,
|
| 4172 |
+
"learning_rate": 1.881e-05,
|
| 4173 |
+
"loss": 5.4287,
|
| 4174 |
+
"step": 5950
|
| 4175 |
+
},
|
| 4176 |
+
{
|
| 4177 |
+
"epoch": 0.0018617320355353417,
|
| 4178 |
+
"grad_norm": 15.3125,
|
| 4179 |
+
"learning_rate": 1.8808e-05,
|
| 4180 |
+
"loss": 5.6799,
|
| 4181 |
+
"step": 5960
|
| 4182 |
+
},
|
| 4183 |
+
{
|
| 4184 |
+
"epoch": 0.0018648557470043606,
|
| 4185 |
+
"grad_norm": 14.625,
|
| 4186 |
+
"learning_rate": 1.8806000000000003e-05,
|
| 4187 |
+
"loss": 5.4295,
|
| 4188 |
+
"step": 5970
|
| 4189 |
+
},
|
| 4190 |
+
{
|
| 4191 |
+
"epoch": 0.0018679794584733798,
|
| 4192 |
+
"grad_norm": 13.5625,
|
| 4193 |
+
"learning_rate": 1.8804e-05,
|
| 4194 |
+
"loss": 5.7169,
|
| 4195 |
+
"step": 5980
|
| 4196 |
+
},
|
| 4197 |
+
{
|
| 4198 |
+
"epoch": 0.0018711031699423987,
|
| 4199 |
+
"grad_norm": 14.625,
|
| 4200 |
+
"learning_rate": 1.8802000000000004e-05,
|
| 4201 |
+
"loss": 5.6801,
|
| 4202 |
+
"step": 5990
|
| 4203 |
+
},
|
| 4204 |
+
{
|
| 4205 |
+
"epoch": 0.0018742268814114177,
|
| 4206 |
+
"grad_norm": 17.875,
|
| 4207 |
+
"learning_rate": 1.88e-05,
|
| 4208 |
+
"loss": 5.5941,
|
| 4209 |
+
"step": 6000
|
| 4210 |
+
},
|
| 4211 |
+
{
|
| 4212 |
+
"epoch": 0.0018773505928804369,
|
| 4213 |
+
"grad_norm": 13.1875,
|
| 4214 |
+
"learning_rate": 1.8798000000000002e-05,
|
| 4215 |
+
"loss": 5.4782,
|
| 4216 |
+
"step": 6010
|
| 4217 |
+
},
|
| 4218 |
+
{
|
| 4219 |
+
"epoch": 0.0018804743043494558,
|
| 4220 |
+
"grad_norm": 19.5,
|
| 4221 |
+
"learning_rate": 1.8796e-05,
|
| 4222 |
+
"loss": 5.371,
|
| 4223 |
+
"step": 6020
|
| 4224 |
+
},
|
| 4225 |
+
{
|
| 4226 |
+
"epoch": 0.0018835980158184748,
|
| 4227 |
+
"grad_norm": 14.0,
|
| 4228 |
+
"learning_rate": 1.8794e-05,
|
| 4229 |
+
"loss": 5.0534,
|
| 4230 |
+
"step": 6030
|
| 4231 |
+
},
|
| 4232 |
+
{
|
| 4233 |
+
"epoch": 0.001886721727287494,
|
| 4234 |
+
"grad_norm": 18.125,
|
| 4235 |
+
"learning_rate": 1.8792000000000002e-05,
|
| 4236 |
+
"loss": 5.5754,
|
| 4237 |
+
"step": 6040
|
| 4238 |
+
},
|
| 4239 |
+
{
|
| 4240 |
+
"epoch": 0.001889845438756513,
|
| 4241 |
+
"grad_norm": 14.875,
|
| 4242 |
+
"learning_rate": 1.879e-05,
|
| 4243 |
+
"loss": 5.4178,
|
| 4244 |
+
"step": 6050
|
| 4245 |
+
},
|
| 4246 |
+
{
|
| 4247 |
+
"epoch": 0.0018929691502255319,
|
| 4248 |
+
"grad_norm": 10.875,
|
| 4249 |
+
"learning_rate": 1.8788e-05,
|
| 4250 |
+
"loss": 5.6989,
|
| 4251 |
+
"step": 6060
|
| 4252 |
+
},
|
| 4253 |
+
{
|
| 4254 |
+
"epoch": 0.001896092861694551,
|
| 4255 |
+
"grad_norm": 14.125,
|
| 4256 |
+
"learning_rate": 1.8786000000000003e-05,
|
| 4257 |
+
"loss": 5.1613,
|
| 4258 |
+
"step": 6070
|
| 4259 |
+
},
|
| 4260 |
+
{
|
| 4261 |
+
"epoch": 0.00189921657316357,
|
| 4262 |
+
"grad_norm": 13.3125,
|
| 4263 |
+
"learning_rate": 1.8784000000000002e-05,
|
| 4264 |
+
"loss": 5.7284,
|
| 4265 |
+
"step": 6080
|
| 4266 |
+
},
|
| 4267 |
+
{
|
| 4268 |
+
"epoch": 0.001902340284632589,
|
| 4269 |
+
"grad_norm": 17.375,
|
| 4270 |
+
"learning_rate": 1.8782e-05,
|
| 4271 |
+
"loss": 5.4167,
|
| 4272 |
+
"step": 6090
|
| 4273 |
+
},
|
| 4274 |
+
{
|
| 4275 |
+
"epoch": 0.0019054639961016081,
|
| 4276 |
+
"grad_norm": 21.375,
|
| 4277 |
+
"learning_rate": 1.878e-05,
|
| 4278 |
+
"loss": 5.3408,
|
| 4279 |
+
"step": 6100
|
| 4280 |
+
},
|
| 4281 |
+
{
|
| 4282 |
+
"epoch": 0.001908587707570627,
|
| 4283 |
+
"grad_norm": 16.5,
|
| 4284 |
+
"learning_rate": 1.8778000000000002e-05,
|
| 4285 |
+
"loss": 5.6557,
|
| 4286 |
+
"step": 6110
|
| 4287 |
+
},
|
| 4288 |
+
{
|
| 4289 |
+
"epoch": 0.001911711419039646,
|
| 4290 |
+
"grad_norm": 13.0,
|
| 4291 |
+
"learning_rate": 1.8776e-05,
|
| 4292 |
+
"loss": 5.4712,
|
| 4293 |
+
"step": 6120
|
| 4294 |
+
},
|
| 4295 |
+
{
|
| 4296 |
+
"epoch": 0.0019148351305086652,
|
| 4297 |
+
"grad_norm": 16.0,
|
| 4298 |
+
"learning_rate": 1.8774e-05,
|
| 4299 |
+
"loss": 5.7737,
|
| 4300 |
+
"step": 6130
|
| 4301 |
+
},
|
| 4302 |
+
{
|
| 4303 |
+
"epoch": 0.0019179588419776842,
|
| 4304 |
+
"grad_norm": 13.0,
|
| 4305 |
+
"learning_rate": 1.8772000000000003e-05,
|
| 4306 |
+
"loss": 5.6721,
|
| 4307 |
+
"step": 6140
|
| 4308 |
+
},
|
| 4309 |
+
{
|
| 4310 |
+
"epoch": 0.0019210825534467031,
|
| 4311 |
+
"grad_norm": 13.0625,
|
| 4312 |
+
"learning_rate": 1.877e-05,
|
| 4313 |
+
"loss": 5.729,
|
| 4314 |
+
"step": 6150
|
| 4315 |
+
},
|
| 4316 |
+
{
|
| 4317 |
+
"epoch": 0.0019242062649157223,
|
| 4318 |
+
"grad_norm": 13.8125,
|
| 4319 |
+
"learning_rate": 1.8768e-05,
|
| 4320 |
+
"loss": 4.9197,
|
| 4321 |
+
"step": 6160
|
| 4322 |
+
},
|
| 4323 |
+
{
|
| 4324 |
+
"epoch": 0.0019273299763847413,
|
| 4325 |
+
"grad_norm": 18.75,
|
| 4326 |
+
"learning_rate": 1.8766000000000003e-05,
|
| 4327 |
+
"loss": 5.704,
|
| 4328 |
+
"step": 6170
|
| 4329 |
+
},
|
| 4330 |
+
{
|
| 4331 |
+
"epoch": 0.0019304536878537602,
|
| 4332 |
+
"grad_norm": 13.3125,
|
| 4333 |
+
"learning_rate": 1.8764000000000002e-05,
|
| 4334 |
+
"loss": 5.7139,
|
| 4335 |
+
"step": 6180
|
| 4336 |
+
},
|
| 4337 |
+
{
|
| 4338 |
+
"epoch": 0.0019335773993227794,
|
| 4339 |
+
"grad_norm": 15.875,
|
| 4340 |
+
"learning_rate": 1.8762e-05,
|
| 4341 |
+
"loss": 5.1654,
|
| 4342 |
+
"step": 6190
|
| 4343 |
+
},
|
| 4344 |
+
{
|
| 4345 |
+
"epoch": 0.0019367011107917984,
|
| 4346 |
+
"grad_norm": 16.625,
|
| 4347 |
+
"learning_rate": 1.876e-05,
|
| 4348 |
+
"loss": 5.3066,
|
| 4349 |
+
"step": 6200
|
| 4350 |
+
},
|
| 4351 |
+
{
|
| 4352 |
+
"epoch": 0.0019398248222608173,
|
| 4353 |
+
"grad_norm": 11.125,
|
| 4354 |
+
"learning_rate": 1.8758000000000002e-05,
|
| 4355 |
+
"loss": 5.585,
|
| 4356 |
+
"step": 6210
|
| 4357 |
+
},
|
| 4358 |
+
{
|
| 4359 |
+
"epoch": 0.0019429485337298365,
|
| 4360 |
+
"grad_norm": 12.75,
|
| 4361 |
+
"learning_rate": 1.8756e-05,
|
| 4362 |
+
"loss": 5.4716,
|
| 4363 |
+
"step": 6220
|
| 4364 |
+
},
|
| 4365 |
+
{
|
| 4366 |
+
"epoch": 0.0019460722451988554,
|
| 4367 |
+
"grad_norm": 11.375,
|
| 4368 |
+
"learning_rate": 1.8754e-05,
|
| 4369 |
+
"loss": 5.4859,
|
| 4370 |
+
"step": 6230
|
| 4371 |
+
},
|
| 4372 |
+
{
|
| 4373 |
+
"epoch": 0.0019491959566678744,
|
| 4374 |
+
"grad_norm": 12.5,
|
| 4375 |
+
"learning_rate": 1.8752000000000003e-05,
|
| 4376 |
+
"loss": 5.6636,
|
| 4377 |
+
"step": 6240
|
| 4378 |
+
},
|
| 4379 |
+
{
|
| 4380 |
+
"epoch": 0.0019523196681368936,
|
| 4381 |
+
"grad_norm": 11.1875,
|
| 4382 |
+
"learning_rate": 1.8750000000000002e-05,
|
| 4383 |
+
"loss": 5.4264,
|
| 4384 |
+
"step": 6250
|
| 4385 |
+
},
|
| 4386 |
+
{
|
| 4387 |
+
"epoch": 0.0019554433796059125,
|
| 4388 |
+
"grad_norm": 19.375,
|
| 4389 |
+
"learning_rate": 1.8748e-05,
|
| 4390 |
+
"loss": 5.3095,
|
| 4391 |
+
"step": 6260
|
| 4392 |
+
},
|
| 4393 |
+
{
|
| 4394 |
+
"epoch": 0.0019585670910749315,
|
| 4395 |
+
"grad_norm": 12.75,
|
| 4396 |
+
"learning_rate": 1.8746000000000003e-05,
|
| 4397 |
+
"loss": 5.4934,
|
| 4398 |
+
"step": 6270
|
| 4399 |
+
},
|
| 4400 |
+
{
|
| 4401 |
+
"epoch": 0.0019616908025439504,
|
| 4402 |
+
"grad_norm": 15.9375,
|
| 4403 |
+
"learning_rate": 1.8744000000000002e-05,
|
| 4404 |
+
"loss": 5.6084,
|
| 4405 |
+
"step": 6280
|
| 4406 |
+
},
|
| 4407 |
+
{
|
| 4408 |
+
"epoch": 0.00196481451401297,
|
| 4409 |
+
"grad_norm": 11.5,
|
| 4410 |
+
"learning_rate": 1.8742e-05,
|
| 4411 |
+
"loss": 5.695,
|
| 4412 |
+
"step": 6290
|
| 4413 |
+
},
|
| 4414 |
+
{
|
| 4415 |
+
"epoch": 0.001967938225481989,
|
| 4416 |
+
"grad_norm": 17.5,
|
| 4417 |
+
"learning_rate": 1.8740000000000004e-05,
|
| 4418 |
+
"loss": 5.456,
|
| 4419 |
+
"step": 6300
|
| 4420 |
+
},
|
| 4421 |
+
{
|
| 4422 |
+
"epoch": 0.0019710619369510077,
|
| 4423 |
+
"grad_norm": 21.375,
|
| 4424 |
+
"learning_rate": 1.8738e-05,
|
| 4425 |
+
"loss": 5.2945,
|
| 4426 |
+
"step": 6310
|
| 4427 |
+
},
|
| 4428 |
+
{
|
| 4429 |
+
"epoch": 0.0019741856484200267,
|
| 4430 |
+
"grad_norm": 15.875,
|
| 4431 |
+
"learning_rate": 1.8736e-05,
|
| 4432 |
+
"loss": 5.7415,
|
| 4433 |
+
"step": 6320
|
| 4434 |
+
},
|
| 4435 |
+
{
|
| 4436 |
+
"epoch": 0.0019773093598890457,
|
| 4437 |
+
"grad_norm": 15.6875,
|
| 4438 |
+
"learning_rate": 1.8734e-05,
|
| 4439 |
+
"loss": 5.7126,
|
| 4440 |
+
"step": 6330
|
| 4441 |
+
},
|
| 4442 |
+
{
|
| 4443 |
+
"epoch": 0.0019804330713580646,
|
| 4444 |
+
"grad_norm": 19.5,
|
| 4445 |
+
"learning_rate": 1.8732000000000003e-05,
|
| 4446 |
+
"loss": 5.5993,
|
| 4447 |
+
"step": 6340
|
| 4448 |
+
},
|
| 4449 |
+
{
|
| 4450 |
+
"epoch": 0.001983556782827084,
|
| 4451 |
+
"grad_norm": 15.9375,
|
| 4452 |
+
"learning_rate": 1.8730000000000002e-05,
|
| 4453 |
+
"loss": 5.6778,
|
| 4454 |
+
"step": 6350
|
| 4455 |
+
},
|
| 4456 |
+
{
|
| 4457 |
+
"epoch": 0.001986680494296103,
|
| 4458 |
+
"grad_norm": 15.8125,
|
| 4459 |
+
"learning_rate": 1.8728e-05,
|
| 4460 |
+
"loss": 5.5652,
|
| 4461 |
+
"step": 6360
|
| 4462 |
+
},
|
| 4463 |
+
{
|
| 4464 |
+
"epoch": 0.001989804205765122,
|
| 4465 |
+
"grad_norm": 11.25,
|
| 4466 |
+
"learning_rate": 1.8726000000000003e-05,
|
| 4467 |
+
"loss": 5.5042,
|
| 4468 |
+
"step": 6370
|
| 4469 |
+
},
|
| 4470 |
+
{
|
| 4471 |
+
"epoch": 0.001992927917234141,
|
| 4472 |
+
"grad_norm": 19.125,
|
| 4473 |
+
"learning_rate": 1.8724000000000002e-05,
|
| 4474 |
+
"loss": 5.7456,
|
| 4475 |
+
"step": 6380
|
| 4476 |
+
},
|
| 4477 |
+
{
|
| 4478 |
+
"epoch": 0.00199605162870316,
|
| 4479 |
+
"grad_norm": 18.75,
|
| 4480 |
+
"learning_rate": 1.8722e-05,
|
| 4481 |
+
"loss": 5.7884,
|
| 4482 |
+
"step": 6390
|
| 4483 |
+
},
|
| 4484 |
+
{
|
| 4485 |
+
"epoch": 0.001999175340172179,
|
| 4486 |
+
"grad_norm": 16.875,
|
| 4487 |
+
"learning_rate": 1.8720000000000004e-05,
|
| 4488 |
+
"loss": 5.3617,
|
| 4489 |
+
"step": 6400
|
| 4490 |
+
},
|
| 4491 |
+
{
|
| 4492 |
+
"epoch": 0.002002299051641198,
|
| 4493 |
+
"grad_norm": 17.125,
|
| 4494 |
+
"learning_rate": 1.8718e-05,
|
| 4495 |
+
"loss": 5.0735,
|
| 4496 |
+
"step": 6410
|
| 4497 |
+
},
|
| 4498 |
+
{
|
| 4499 |
+
"epoch": 0.002005422763110217,
|
| 4500 |
+
"grad_norm": 12.5625,
|
| 4501 |
+
"learning_rate": 1.8716000000000002e-05,
|
| 4502 |
+
"loss": 5.5877,
|
| 4503 |
+
"step": 6420
|
| 4504 |
+
},
|
| 4505 |
+
{
|
| 4506 |
+
"epoch": 0.002008546474579236,
|
| 4507 |
+
"grad_norm": 15.0625,
|
| 4508 |
+
"learning_rate": 1.8714e-05,
|
| 4509 |
+
"loss": 5.5967,
|
| 4510 |
+
"step": 6430
|
| 4511 |
+
},
|
| 4512 |
+
{
|
| 4513 |
+
"epoch": 0.002011670186048255,
|
| 4514 |
+
"grad_norm": 18.875,
|
| 4515 |
+
"learning_rate": 1.8712e-05,
|
| 4516 |
+
"loss": 5.4915,
|
| 4517 |
+
"step": 6440
|
| 4518 |
+
},
|
| 4519 |
+
{
|
| 4520 |
+
"epoch": 0.002014793897517274,
|
| 4521 |
+
"grad_norm": 11.0625,
|
| 4522 |
+
"learning_rate": 1.8710000000000002e-05,
|
| 4523 |
+
"loss": 5.4465,
|
| 4524 |
+
"step": 6450
|
| 4525 |
+
},
|
| 4526 |
+
{
|
| 4527 |
+
"epoch": 0.002017917608986293,
|
| 4528 |
+
"grad_norm": 18.125,
|
| 4529 |
+
"learning_rate": 1.8708e-05,
|
| 4530 |
+
"loss": 5.7355,
|
| 4531 |
+
"step": 6460
|
| 4532 |
+
},
|
| 4533 |
+
{
|
| 4534 |
+
"epoch": 0.0020210413204553124,
|
| 4535 |
+
"grad_norm": 11.8125,
|
| 4536 |
+
"learning_rate": 1.8706000000000004e-05,
|
| 4537 |
+
"loss": 5.7497,
|
| 4538 |
+
"step": 6470
|
| 4539 |
+
},
|
| 4540 |
+
{
|
| 4541 |
+
"epoch": 0.0020241650319243313,
|
| 4542 |
+
"grad_norm": 21.375,
|
| 4543 |
+
"learning_rate": 1.8704000000000003e-05,
|
| 4544 |
+
"loss": 5.501,
|
| 4545 |
+
"step": 6480
|
| 4546 |
+
},
|
| 4547 |
+
{
|
| 4548 |
+
"epoch": 0.0020272887433933503,
|
| 4549 |
+
"grad_norm": 14.25,
|
| 4550 |
+
"learning_rate": 1.8702e-05,
|
| 4551 |
+
"loss": 5.3654,
|
| 4552 |
+
"step": 6490
|
| 4553 |
+
},
|
| 4554 |
+
{
|
| 4555 |
+
"epoch": 0.0020304124548623692,
|
| 4556 |
+
"grad_norm": 14.25,
|
| 4557 |
+
"learning_rate": 1.8700000000000004e-05,
|
| 4558 |
+
"loss": 5.4165,
|
| 4559 |
+
"step": 6500
|
| 4560 |
+
},
|
| 4561 |
+
{
|
| 4562 |
+
"epoch": 0.002033536166331388,
|
| 4563 |
+
"grad_norm": 14.75,
|
| 4564 |
+
"learning_rate": 1.8698e-05,
|
| 4565 |
+
"loss": 5.0379,
|
| 4566 |
+
"step": 6510
|
| 4567 |
+
},
|
| 4568 |
+
{
|
| 4569 |
+
"epoch": 0.002036659877800407,
|
| 4570 |
+
"grad_norm": 18.25,
|
| 4571 |
+
"learning_rate": 1.8696000000000002e-05,
|
| 4572 |
+
"loss": 5.718,
|
| 4573 |
+
"step": 6520
|
| 4574 |
+
},
|
| 4575 |
+
{
|
| 4576 |
+
"epoch": 0.0020397835892694265,
|
| 4577 |
+
"grad_norm": 21.0,
|
| 4578 |
+
"learning_rate": 1.8694e-05,
|
| 4579 |
+
"loss": 5.3976,
|
| 4580 |
+
"step": 6530
|
| 4581 |
+
},
|
| 4582 |
+
{
|
| 4583 |
+
"epoch": 0.0020429073007384455,
|
| 4584 |
+
"grad_norm": 12.5,
|
| 4585 |
+
"learning_rate": 1.8692e-05,
|
| 4586 |
+
"loss": 5.745,
|
| 4587 |
+
"step": 6540
|
| 4588 |
+
},
|
| 4589 |
+
{
|
| 4590 |
+
"epoch": 0.0020460310122074644,
|
| 4591 |
+
"grad_norm": 27.5,
|
| 4592 |
+
"learning_rate": 1.8690000000000002e-05,
|
| 4593 |
+
"loss": 5.4708,
|
| 4594 |
+
"step": 6550
|
| 4595 |
+
},
|
| 4596 |
+
{
|
| 4597 |
+
"epoch": 0.0020491547236764834,
|
| 4598 |
+
"grad_norm": 15.75,
|
| 4599 |
+
"learning_rate": 1.8688e-05,
|
| 4600 |
+
"loss": 5.5559,
|
| 4601 |
+
"step": 6560
|
| 4602 |
+
},
|
| 4603 |
+
{
|
| 4604 |
+
"epoch": 0.0020522784351455024,
|
| 4605 |
+
"grad_norm": 14.25,
|
| 4606 |
+
"learning_rate": 1.8686e-05,
|
| 4607 |
+
"loss": 5.4009,
|
| 4608 |
+
"step": 6570
|
| 4609 |
+
},
|
| 4610 |
+
{
|
| 4611 |
+
"epoch": 0.0020554021466145213,
|
| 4612 |
+
"grad_norm": 14.375,
|
| 4613 |
+
"learning_rate": 1.8684000000000003e-05,
|
| 4614 |
+
"loss": 5.4685,
|
| 4615 |
+
"step": 6580
|
| 4616 |
+
},
|
| 4617 |
+
{
|
| 4618 |
+
"epoch": 0.0020585258580835407,
|
| 4619 |
+
"grad_norm": 22.25,
|
| 4620 |
+
"learning_rate": 1.8682000000000002e-05,
|
| 4621 |
+
"loss": 5.4608,
|
| 4622 |
+
"step": 6590
|
| 4623 |
+
},
|
| 4624 |
+
{
|
| 4625 |
+
"epoch": 0.0020616495695525597,
|
| 4626 |
+
"grad_norm": 12.8125,
|
| 4627 |
+
"learning_rate": 1.8680000000000004e-05,
|
| 4628 |
+
"loss": 5.6396,
|
| 4629 |
+
"step": 6600
|
| 4630 |
+
},
|
| 4631 |
+
{
|
| 4632 |
+
"epoch": 0.0020647732810215786,
|
| 4633 |
+
"grad_norm": 17.375,
|
| 4634 |
+
"learning_rate": 1.8678e-05,
|
| 4635 |
+
"loss": 5.4356,
|
| 4636 |
+
"step": 6610
|
| 4637 |
+
},
|
| 4638 |
+
{
|
| 4639 |
+
"epoch": 0.0020678969924905976,
|
| 4640 |
+
"grad_norm": 19.375,
|
| 4641 |
+
"learning_rate": 1.8676000000000002e-05,
|
| 4642 |
+
"loss": 5.4307,
|
| 4643 |
+
"step": 6620
|
| 4644 |
+
},
|
| 4645 |
+
{
|
| 4646 |
+
"epoch": 0.0020710207039596165,
|
| 4647 |
+
"grad_norm": 15.3125,
|
| 4648 |
+
"learning_rate": 1.8674e-05,
|
| 4649 |
+
"loss": 5.7456,
|
| 4650 |
+
"step": 6630
|
| 4651 |
+
},
|
| 4652 |
+
{
|
| 4653 |
+
"epoch": 0.0020741444154286355,
|
| 4654 |
+
"grad_norm": 18.625,
|
| 4655 |
+
"learning_rate": 1.8672e-05,
|
| 4656 |
+
"loss": 5.5859,
|
| 4657 |
+
"step": 6640
|
| 4658 |
+
},
|
| 4659 |
+
{
|
| 4660 |
+
"epoch": 0.002077268126897655,
|
| 4661 |
+
"grad_norm": 16.5,
|
| 4662 |
+
"learning_rate": 1.8670000000000003e-05,
|
| 4663 |
+
"loss": 5.5676,
|
| 4664 |
+
"step": 6650
|
| 4665 |
+
},
|
| 4666 |
+
{
|
| 4667 |
+
"epoch": 0.002080391838366674,
|
| 4668 |
+
"grad_norm": 17.875,
|
| 4669 |
+
"learning_rate": 1.8668e-05,
|
| 4670 |
+
"loss": 5.6596,
|
| 4671 |
+
"step": 6660
|
| 4672 |
+
},
|
| 4673 |
+
{
|
| 4674 |
+
"epoch": 0.002083515549835693,
|
| 4675 |
+
"grad_norm": 16.375,
|
| 4676 |
+
"learning_rate": 1.8666e-05,
|
| 4677 |
+
"loss": 5.1641,
|
| 4678 |
+
"step": 6670
|
| 4679 |
+
},
|
| 4680 |
+
{
|
| 4681 |
+
"epoch": 0.0020866392613047118,
|
| 4682 |
+
"grad_norm": 17.125,
|
| 4683 |
+
"learning_rate": 1.8664000000000003e-05,
|
| 4684 |
+
"loss": 5.5482,
|
| 4685 |
+
"step": 6680
|
| 4686 |
+
},
|
| 4687 |
+
{
|
| 4688 |
+
"epoch": 0.0020897629727737307,
|
| 4689 |
+
"grad_norm": 17.375,
|
| 4690 |
+
"learning_rate": 1.8662000000000002e-05,
|
| 4691 |
+
"loss": 5.5011,
|
| 4692 |
+
"step": 6690
|
| 4693 |
+
},
|
| 4694 |
+
{
|
| 4695 |
+
"epoch": 0.0020928866842427497,
|
| 4696 |
+
"grad_norm": 15.4375,
|
| 4697 |
+
"learning_rate": 1.866e-05,
|
| 4698 |
+
"loss": 5.7944,
|
| 4699 |
+
"step": 6700
|
| 4700 |
+
},
|
| 4701 |
+
{
|
| 4702 |
+
"epoch": 0.002096010395711769,
|
| 4703 |
+
"grad_norm": 18.625,
|
| 4704 |
+
"learning_rate": 1.8658e-05,
|
| 4705 |
+
"loss": 5.5027,
|
| 4706 |
+
"step": 6710
|
| 4707 |
+
},
|
| 4708 |
+
{
|
| 4709 |
+
"epoch": 0.002099134107180788,
|
| 4710 |
+
"grad_norm": 14.0625,
|
| 4711 |
+
"learning_rate": 1.8656000000000002e-05,
|
| 4712 |
+
"loss": 5.1993,
|
| 4713 |
+
"step": 6720
|
| 4714 |
+
},
|
| 4715 |
+
{
|
| 4716 |
+
"epoch": 0.002102257818649807,
|
| 4717 |
+
"grad_norm": 38.25,
|
| 4718 |
+
"learning_rate": 1.8654e-05,
|
| 4719 |
+
"loss": 5.6462,
|
| 4720 |
+
"step": 6730
|
| 4721 |
+
},
|
| 4722 |
+
{
|
| 4723 |
+
"epoch": 0.002105381530118826,
|
| 4724 |
+
"grad_norm": 12.8125,
|
| 4725 |
+
"learning_rate": 1.8652e-05,
|
| 4726 |
+
"loss": 5.1891,
|
| 4727 |
+
"step": 6740
|
| 4728 |
+
},
|
| 4729 |
+
{
|
| 4730 |
+
"epoch": 0.002108505241587845,
|
| 4731 |
+
"grad_norm": 11.875,
|
| 4732 |
+
"learning_rate": 1.8650000000000003e-05,
|
| 4733 |
+
"loss": 5.4365,
|
| 4734 |
+
"step": 6750
|
| 4735 |
+
},
|
| 4736 |
+
{
|
| 4737 |
+
"epoch": 0.002111628953056864,
|
| 4738 |
+
"grad_norm": 16.125,
|
| 4739 |
+
"learning_rate": 1.8648000000000002e-05,
|
| 4740 |
+
"loss": 5.3265,
|
| 4741 |
+
"step": 6760
|
| 4742 |
+
},
|
| 4743 |
+
{
|
| 4744 |
+
"epoch": 0.0021147526645258832,
|
| 4745 |
+
"grad_norm": 18.25,
|
| 4746 |
+
"learning_rate": 1.8646e-05,
|
| 4747 |
+
"loss": 5.4414,
|
| 4748 |
+
"step": 6770
|
| 4749 |
+
},
|
| 4750 |
+
{
|
| 4751 |
+
"epoch": 0.002117876375994902,
|
| 4752 |
+
"grad_norm": 13.8125,
|
| 4753 |
+
"learning_rate": 1.8644000000000003e-05,
|
| 4754 |
+
"loss": 5.447,
|
| 4755 |
+
"step": 6780
|
| 4756 |
+
},
|
| 4757 |
+
{
|
| 4758 |
+
"epoch": 0.002121000087463921,
|
| 4759 |
+
"grad_norm": 13.5625,
|
| 4760 |
+
"learning_rate": 1.8642000000000002e-05,
|
| 4761 |
+
"loss": 5.6893,
|
| 4762 |
+
"step": 6790
|
| 4763 |
+
},
|
| 4764 |
+
{
|
| 4765 |
+
"epoch": 0.00212412379893294,
|
| 4766 |
+
"grad_norm": 16.75,
|
| 4767 |
+
"learning_rate": 1.864e-05,
|
| 4768 |
+
"loss": 5.7147,
|
| 4769 |
+
"step": 6800
|
| 4770 |
+
},
|
| 4771 |
+
{
|
| 4772 |
+
"epoch": 0.002127247510401959,
|
| 4773 |
+
"grad_norm": 13.125,
|
| 4774 |
+
"learning_rate": 1.8638e-05,
|
| 4775 |
+
"loss": 4.9141,
|
| 4776 |
+
"step": 6810
|
| 4777 |
+
},
|
| 4778 |
+
{
|
| 4779 |
+
"epoch": 0.002130371221870978,
|
| 4780 |
+
"grad_norm": 17.625,
|
| 4781 |
+
"learning_rate": 1.8636e-05,
|
| 4782 |
+
"loss": 5.6681,
|
| 4783 |
+
"step": 6820
|
| 4784 |
+
},
|
| 4785 |
+
{
|
| 4786 |
+
"epoch": 0.0021334949333399974,
|
| 4787 |
+
"grad_norm": 13.4375,
|
| 4788 |
+
"learning_rate": 1.8634e-05,
|
| 4789 |
+
"loss": 5.753,
|
| 4790 |
+
"step": 6830
|
| 4791 |
+
},
|
| 4792 |
+
{
|
| 4793 |
+
"epoch": 0.0021366186448090164,
|
| 4794 |
+
"grad_norm": 14.5,
|
| 4795 |
+
"learning_rate": 1.8632e-05,
|
| 4796 |
+
"loss": 5.7403,
|
| 4797 |
+
"step": 6840
|
| 4798 |
+
},
|
| 4799 |
+
{
|
| 4800 |
+
"epoch": 0.0021397423562780353,
|
| 4801 |
+
"grad_norm": 15.5,
|
| 4802 |
+
"learning_rate": 1.8630000000000003e-05,
|
| 4803 |
+
"loss": 5.6975,
|
| 4804 |
+
"step": 6850
|
| 4805 |
+
},
|
| 4806 |
+
{
|
| 4807 |
+
"epoch": 0.0021428660677470543,
|
| 4808 |
+
"grad_norm": 19.125,
|
| 4809 |
+
"learning_rate": 1.8628000000000002e-05,
|
| 4810 |
+
"loss": 5.2514,
|
| 4811 |
+
"step": 6860
|
| 4812 |
+
},
|
| 4813 |
+
{
|
| 4814 |
+
"epoch": 0.0021459897792160732,
|
| 4815 |
+
"grad_norm": 12.4375,
|
| 4816 |
+
"learning_rate": 1.8626e-05,
|
| 4817 |
+
"loss": 5.7343,
|
| 4818 |
+
"step": 6870
|
| 4819 |
+
},
|
| 4820 |
+
{
|
| 4821 |
+
"epoch": 0.002149113490685092,
|
| 4822 |
+
"grad_norm": 12.75,
|
| 4823 |
+
"learning_rate": 1.8624000000000003e-05,
|
| 4824 |
+
"loss": 5.7297,
|
| 4825 |
+
"step": 6880
|
| 4826 |
+
},
|
| 4827 |
+
{
|
| 4828 |
+
"epoch": 0.0021522372021541116,
|
| 4829 |
+
"grad_norm": 19.0,
|
| 4830 |
+
"learning_rate": 1.8622000000000002e-05,
|
| 4831 |
+
"loss": 5.4692,
|
| 4832 |
+
"step": 6890
|
| 4833 |
+
},
|
| 4834 |
+
{
|
| 4835 |
+
"epoch": 0.0021553609136231305,
|
| 4836 |
+
"grad_norm": 18.75,
|
| 4837 |
+
"learning_rate": 1.862e-05,
|
| 4838 |
+
"loss": 5.602,
|
| 4839 |
+
"step": 6900
|
| 4840 |
+
},
|
| 4841 |
+
{
|
| 4842 |
+
"epoch": 0.0021584846250921495,
|
| 4843 |
+
"grad_norm": 11.8125,
|
| 4844 |
+
"learning_rate": 1.8618e-05,
|
| 4845 |
+
"loss": 5.4007,
|
| 4846 |
+
"step": 6910
|
| 4847 |
+
},
|
| 4848 |
+
{
|
| 4849 |
+
"epoch": 0.0021616083365611685,
|
| 4850 |
+
"grad_norm": 16.125,
|
| 4851 |
+
"learning_rate": 1.8616e-05,
|
| 4852 |
+
"loss": 5.4039,
|
| 4853 |
+
"step": 6920
|
| 4854 |
+
},
|
| 4855 |
+
{
|
| 4856 |
+
"epoch": 0.0021647320480301874,
|
| 4857 |
+
"grad_norm": 18.625,
|
| 4858 |
+
"learning_rate": 1.8614e-05,
|
| 4859 |
+
"loss": 4.9848,
|
| 4860 |
+
"step": 6930
|
| 4861 |
+
},
|
| 4862 |
+
{
|
| 4863 |
+
"epoch": 0.0021678557594992064,
|
| 4864 |
+
"grad_norm": 19.875,
|
| 4865 |
+
"learning_rate": 1.8612e-05,
|
| 4866 |
+
"loss": 5.3236,
|
| 4867 |
+
"step": 6940
|
| 4868 |
+
},
|
| 4869 |
+
{
|
| 4870 |
+
"epoch": 0.0021709794709682258,
|
| 4871 |
+
"grad_norm": 15.3125,
|
| 4872 |
+
"learning_rate": 1.8610000000000003e-05,
|
| 4873 |
+
"loss": 5.5269,
|
| 4874 |
+
"step": 6950
|
| 4875 |
+
},
|
| 4876 |
+
{
|
| 4877 |
+
"epoch": 0.0021741031824372447,
|
| 4878 |
+
"grad_norm": 18.125,
|
| 4879 |
+
"learning_rate": 1.8608000000000002e-05,
|
| 4880 |
+
"loss": 5.515,
|
| 4881 |
+
"step": 6960
|
| 4882 |
+
},
|
| 4883 |
+
{
|
| 4884 |
+
"epoch": 0.0021772268939062637,
|
| 4885 |
+
"grad_norm": 19.125,
|
| 4886 |
+
"learning_rate": 1.8606e-05,
|
| 4887 |
+
"loss": 5.392,
|
| 4888 |
+
"step": 6970
|
| 4889 |
+
},
|
| 4890 |
+
{
|
| 4891 |
+
"epoch": 0.0021803506053752826,
|
| 4892 |
+
"grad_norm": 15.6875,
|
| 4893 |
+
"learning_rate": 1.8604000000000003e-05,
|
| 4894 |
+
"loss": 5.2582,
|
| 4895 |
+
"step": 6980
|
| 4896 |
+
},
|
| 4897 |
+
{
|
| 4898 |
+
"epoch": 0.0021834743168443016,
|
| 4899 |
+
"grad_norm": 14.25,
|
| 4900 |
+
"learning_rate": 1.8602000000000002e-05,
|
| 4901 |
+
"loss": 5.7263,
|
| 4902 |
+
"step": 6990
|
| 4903 |
+
},
|
| 4904 |
+
{
|
| 4905 |
+
"epoch": 0.0021865980283133205,
|
| 4906 |
+
"grad_norm": 15.0625,
|
| 4907 |
+
"learning_rate": 1.86e-05,
|
| 4908 |
+
"loss": 5.7057,
|
| 4909 |
+
"step": 7000
|
| 4910 |
+
},
|
| 4911 |
+
{
|
| 4912 |
+
"epoch": 0.00218972173978234,
|
| 4913 |
+
"grad_norm": 14.9375,
|
| 4914 |
+
"learning_rate": 1.8598e-05,
|
| 4915 |
+
"loss": 5.0865,
|
| 4916 |
+
"step": 7010
|
| 4917 |
+
},
|
| 4918 |
+
{
|
| 4919 |
+
"epoch": 0.002192845451251359,
|
| 4920 |
+
"grad_norm": 13.125,
|
| 4921 |
+
"learning_rate": 1.8596e-05,
|
| 4922 |
+
"loss": 5.5477,
|
| 4923 |
+
"step": 7020
|
| 4924 |
+
},
|
| 4925 |
+
{
|
| 4926 |
+
"epoch": 0.002195969162720378,
|
| 4927 |
+
"grad_norm": 20.25,
|
| 4928 |
+
"learning_rate": 1.8594000000000002e-05,
|
| 4929 |
+
"loss": 5.3479,
|
| 4930 |
+
"step": 7030
|
| 4931 |
+
},
|
| 4932 |
+
{
|
| 4933 |
+
"epoch": 0.002199092874189397,
|
| 4934 |
+
"grad_norm": 16.25,
|
| 4935 |
+
"learning_rate": 1.8592e-05,
|
| 4936 |
+
"loss": 5.7431,
|
| 4937 |
+
"step": 7040
|
| 4938 |
+
},
|
| 4939 |
+
{
|
| 4940 |
+
"epoch": 0.0022022165856584158,
|
| 4941 |
+
"grad_norm": 14.375,
|
| 4942 |
+
"learning_rate": 1.859e-05,
|
| 4943 |
+
"loss": 5.6489,
|
| 4944 |
+
"step": 7050
|
| 4945 |
+
},
|
| 4946 |
+
{
|
| 4947 |
+
"epoch": 0.002205340297127435,
|
| 4948 |
+
"grad_norm": 13.0,
|
| 4949 |
+
"learning_rate": 1.8588000000000002e-05,
|
| 4950 |
+
"loss": 5.7156,
|
| 4951 |
+
"step": 7060
|
| 4952 |
+
},
|
| 4953 |
+
{
|
| 4954 |
+
"epoch": 0.002208464008596454,
|
| 4955 |
+
"grad_norm": 11.5,
|
| 4956 |
+
"learning_rate": 1.8586e-05,
|
| 4957 |
+
"loss": 5.2813,
|
| 4958 |
+
"step": 7070
|
| 4959 |
+
},
|
| 4960 |
+
{
|
| 4961 |
+
"epoch": 0.002211587720065473,
|
| 4962 |
+
"grad_norm": 13.5625,
|
| 4963 |
+
"learning_rate": 1.8584000000000004e-05,
|
| 4964 |
+
"loss": 5.5629,
|
| 4965 |
+
"step": 7080
|
| 4966 |
+
},
|
| 4967 |
+
{
|
| 4968 |
+
"epoch": 0.002214711431534492,
|
| 4969 |
+
"grad_norm": 16.125,
|
| 4970 |
+
"learning_rate": 1.8582000000000003e-05,
|
| 4971 |
+
"loss": 5.2464,
|
| 4972 |
+
"step": 7090
|
| 4973 |
+
},
|
| 4974 |
+
{
|
| 4975 |
+
"epoch": 0.002217835143003511,
|
| 4976 |
+
"grad_norm": 12.6875,
|
| 4977 |
+
"learning_rate": 1.858e-05,
|
| 4978 |
+
"loss": 5.327,
|
| 4979 |
+
"step": 7100
|
| 4980 |
+
},
|
| 4981 |
+
{
|
| 4982 |
+
"epoch": 0.00222095885447253,
|
| 4983 |
+
"grad_norm": 15.3125,
|
| 4984 |
+
"learning_rate": 1.8578e-05,
|
| 4985 |
+
"loss": 5.5996,
|
| 4986 |
+
"step": 7110
|
| 4987 |
+
},
|
| 4988 |
+
{
|
| 4989 |
+
"epoch": 0.0022240825659415493,
|
| 4990 |
+
"grad_norm": 21.25,
|
| 4991 |
+
"learning_rate": 1.8576e-05,
|
| 4992 |
+
"loss": 5.7259,
|
| 4993 |
+
"step": 7120
|
| 4994 |
+
},
|
| 4995 |
+
{
|
| 4996 |
+
"epoch": 0.0022272062774105683,
|
| 4997 |
+
"grad_norm": 10.75,
|
| 4998 |
+
"learning_rate": 1.8574000000000002e-05,
|
| 4999 |
+
"loss": 5.6215,
|
| 5000 |
+
"step": 7130
|
| 5001 |
+
},
|
| 5002 |
+
{
|
| 5003 |
+
"epoch": 0.0022303299888795872,
|
| 5004 |
+
"grad_norm": 11.625,
|
| 5005 |
+
"learning_rate": 1.8572e-05,
|
| 5006 |
+
"loss": 5.7345,
|
| 5007 |
+
"step": 7140
|
| 5008 |
+
},
|
| 5009 |
+
{
|
| 5010 |
+
"epoch": 0.002233453700348606,
|
| 5011 |
+
"grad_norm": 11.4375,
|
| 5012 |
+
"learning_rate": 1.857e-05,
|
| 5013 |
+
"loss": 5.5052,
|
| 5014 |
+
"step": 7150
|
| 5015 |
+
},
|
| 5016 |
+
{
|
| 5017 |
+
"epoch": 0.002236577411817625,
|
| 5018 |
+
"grad_norm": 15.375,
|
| 5019 |
+
"learning_rate": 1.8568000000000002e-05,
|
| 5020 |
+
"loss": 5.2982,
|
| 5021 |
+
"step": 7160
|
| 5022 |
+
},
|
| 5023 |
+
{
|
| 5024 |
+
"epoch": 0.002239701123286644,
|
| 5025 |
+
"grad_norm": 16.125,
|
| 5026 |
+
"learning_rate": 1.8566e-05,
|
| 5027 |
+
"loss": 5.6919,
|
| 5028 |
+
"step": 7170
|
| 5029 |
+
},
|
| 5030 |
+
{
|
| 5031 |
+
"epoch": 0.0022428248347556635,
|
| 5032 |
+
"grad_norm": 12.375,
|
| 5033 |
+
"learning_rate": 1.8564e-05,
|
| 5034 |
+
"loss": 5.7517,
|
| 5035 |
+
"step": 7180
|
| 5036 |
+
},
|
| 5037 |
+
{
|
| 5038 |
+
"epoch": 0.0022459485462246825,
|
| 5039 |
+
"grad_norm": 20.125,
|
| 5040 |
+
"learning_rate": 1.8562000000000003e-05,
|
| 5041 |
+
"loss": 5.4861,
|
| 5042 |
+
"step": 7190
|
| 5043 |
+
},
|
| 5044 |
+
{
|
| 5045 |
+
"epoch": 0.0022490722576937014,
|
| 5046 |
+
"grad_norm": 20.875,
|
| 5047 |
+
"learning_rate": 1.8560000000000002e-05,
|
| 5048 |
+
"loss": 5.6446,
|
| 5049 |
+
"step": 7200
|
| 5050 |
+
},
|
| 5051 |
+
{
|
| 5052 |
+
"epoch": 0.0022521959691627204,
|
| 5053 |
+
"grad_norm": 12.0,
|
| 5054 |
+
"learning_rate": 1.8558e-05,
|
| 5055 |
+
"loss": 5.4446,
|
| 5056 |
+
"step": 7210
|
| 5057 |
+
},
|
| 5058 |
+
{
|
| 5059 |
+
"epoch": 0.0022553196806317393,
|
| 5060 |
+
"grad_norm": 15.0625,
|
| 5061 |
+
"learning_rate": 1.8556e-05,
|
| 5062 |
+
"loss": 5.4171,
|
| 5063 |
+
"step": 7220
|
| 5064 |
+
},
|
| 5065 |
+
{
|
| 5066 |
+
"epoch": 0.0022584433921007583,
|
| 5067 |
+
"grad_norm": 14.4375,
|
| 5068 |
+
"learning_rate": 1.8554000000000002e-05,
|
| 5069 |
+
"loss": 5.2308,
|
| 5070 |
+
"step": 7230
|
| 5071 |
+
},
|
| 5072 |
+
{
|
| 5073 |
+
"epoch": 0.0022615671035697777,
|
| 5074 |
+
"grad_norm": 18.75,
|
| 5075 |
+
"learning_rate": 1.8552e-05,
|
| 5076 |
+
"loss": 5.7176,
|
| 5077 |
+
"step": 7240
|
| 5078 |
+
},
|
| 5079 |
+
{
|
| 5080 |
+
"epoch": 0.0022646908150387966,
|
| 5081 |
+
"grad_norm": 18.375,
|
| 5082 |
+
"learning_rate": 1.855e-05,
|
| 5083 |
+
"loss": 5.3464,
|
| 5084 |
+
"step": 7250
|
| 5085 |
+
},
|
| 5086 |
+
{
|
| 5087 |
+
"epoch": 0.0022678145265078156,
|
| 5088 |
+
"grad_norm": 15.375,
|
| 5089 |
+
"learning_rate": 1.8548000000000003e-05,
|
| 5090 |
+
"loss": 5.7412,
|
| 5091 |
+
"step": 7260
|
| 5092 |
+
},
|
| 5093 |
+
{
|
| 5094 |
+
"epoch": 0.0022709382379768345,
|
| 5095 |
+
"grad_norm": 21.0,
|
| 5096 |
+
"learning_rate": 1.8546e-05,
|
| 5097 |
+
"loss": 5.434,
|
| 5098 |
+
"step": 7270
|
| 5099 |
+
},
|
| 5100 |
+
{
|
| 5101 |
+
"epoch": 0.0022740619494458535,
|
| 5102 |
+
"grad_norm": 12.375,
|
| 5103 |
+
"learning_rate": 1.8544e-05,
|
| 5104 |
+
"loss": 5.572,
|
| 5105 |
+
"step": 7280
|
| 5106 |
+
},
|
| 5107 |
+
{
|
| 5108 |
+
"epoch": 0.0022771856609148725,
|
| 5109 |
+
"grad_norm": 13.125,
|
| 5110 |
+
"learning_rate": 1.8542000000000003e-05,
|
| 5111 |
+
"loss": 5.4166,
|
| 5112 |
+
"step": 7290
|
| 5113 |
+
},
|
| 5114 |
+
{
|
| 5115 |
+
"epoch": 0.002280309372383892,
|
| 5116 |
+
"grad_norm": 13.25,
|
| 5117 |
+
"learning_rate": 1.8540000000000002e-05,
|
| 5118 |
+
"loss": 5.7315,
|
| 5119 |
+
"step": 7300
|
| 5120 |
+
},
|
| 5121 |
+
{
|
| 5122 |
+
"epoch": 0.002283433083852911,
|
| 5123 |
+
"grad_norm": 12.25,
|
| 5124 |
+
"learning_rate": 1.8538e-05,
|
| 5125 |
+
"loss": 5.0891,
|
| 5126 |
+
"step": 7310
|
| 5127 |
+
},
|
| 5128 |
+
{
|
| 5129 |
+
"epoch": 0.0022865567953219298,
|
| 5130 |
+
"grad_norm": 18.125,
|
| 5131 |
+
"learning_rate": 1.8536e-05,
|
| 5132 |
+
"loss": 5.3659,
|
| 5133 |
+
"step": 7320
|
| 5134 |
+
},
|
| 5135 |
+
{
|
| 5136 |
+
"epoch": 0.0022896805067909487,
|
| 5137 |
+
"grad_norm": 19.125,
|
| 5138 |
+
"learning_rate": 1.8534000000000002e-05,
|
| 5139 |
+
"loss": 5.2582,
|
| 5140 |
+
"step": 7330
|
| 5141 |
+
},
|
| 5142 |
+
{
|
| 5143 |
+
"epoch": 0.0022928042182599677,
|
| 5144 |
+
"grad_norm": 15.8125,
|
| 5145 |
+
"learning_rate": 1.8532e-05,
|
| 5146 |
+
"loss": 5.6617,
|
| 5147 |
+
"step": 7340
|
| 5148 |
+
},
|
| 5149 |
+
{
|
| 5150 |
+
"epoch": 0.0022959279297289866,
|
| 5151 |
+
"grad_norm": 15.1875,
|
| 5152 |
+
"learning_rate": 1.853e-05,
|
| 5153 |
+
"loss": 5.2852,
|
| 5154 |
+
"step": 7350
|
| 5155 |
+
},
|
| 5156 |
+
{
|
| 5157 |
+
"epoch": 0.002299051641198006,
|
| 5158 |
+
"grad_norm": 17.625,
|
| 5159 |
+
"learning_rate": 1.8528000000000003e-05,
|
| 5160 |
+
"loss": 5.1184,
|
| 5161 |
+
"step": 7360
|
| 5162 |
+
},
|
| 5163 |
+
{
|
| 5164 |
+
"epoch": 0.002302175352667025,
|
| 5165 |
+
"grad_norm": 14.6875,
|
| 5166 |
+
"learning_rate": 1.8526000000000002e-05,
|
| 5167 |
+
"loss": 5.4026,
|
| 5168 |
+
"step": 7370
|
| 5169 |
+
},
|
| 5170 |
+
{
|
| 5171 |
+
"epoch": 0.002305299064136044,
|
| 5172 |
+
"grad_norm": 14.4375,
|
| 5173 |
+
"learning_rate": 1.8524e-05,
|
| 5174 |
+
"loss": 5.5041,
|
| 5175 |
+
"step": 7380
|
| 5176 |
+
},
|
| 5177 |
+
{
|
| 5178 |
+
"epoch": 0.002308422775605063,
|
| 5179 |
+
"grad_norm": 19.5,
|
| 5180 |
+
"learning_rate": 1.8522000000000003e-05,
|
| 5181 |
+
"loss": 5.3296,
|
| 5182 |
+
"step": 7390
|
| 5183 |
+
},
|
| 5184 |
+
{
|
| 5185 |
+
"epoch": 0.002311546487074082,
|
| 5186 |
+
"grad_norm": 16.75,
|
| 5187 |
+
"learning_rate": 1.8520000000000002e-05,
|
| 5188 |
+
"loss": 5.39,
|
| 5189 |
+
"step": 7400
|
| 5190 |
+
},
|
| 5191 |
+
{
|
| 5192 |
+
"epoch": 0.002314670198543101,
|
| 5193 |
+
"grad_norm": 18.25,
|
| 5194 |
+
"learning_rate": 1.8518e-05,
|
| 5195 |
+
"loss": 5.6362,
|
| 5196 |
+
"step": 7410
|
| 5197 |
+
},
|
| 5198 |
+
{
|
| 5199 |
+
"epoch": 0.00231779391001212,
|
| 5200 |
+
"grad_norm": 13.625,
|
| 5201 |
+
"learning_rate": 1.8516e-05,
|
| 5202 |
+
"loss": 5.3206,
|
| 5203 |
+
"step": 7420
|
| 5204 |
+
},
|
| 5205 |
+
{
|
| 5206 |
+
"epoch": 0.002320917621481139,
|
| 5207 |
+
"grad_norm": 18.875,
|
| 5208 |
+
"learning_rate": 1.8514e-05,
|
| 5209 |
+
"loss": 5.7378,
|
| 5210 |
+
"step": 7430
|
| 5211 |
+
},
|
| 5212 |
+
{
|
| 5213 |
+
"epoch": 0.002324041332950158,
|
| 5214 |
+
"grad_norm": 18.0,
|
| 5215 |
+
"learning_rate": 1.8512e-05,
|
| 5216 |
+
"loss": 5.7009,
|
| 5217 |
+
"step": 7440
|
| 5218 |
+
},
|
| 5219 |
+
{
|
| 5220 |
+
"epoch": 0.002327165044419177,
|
| 5221 |
+
"grad_norm": 14.75,
|
| 5222 |
+
"learning_rate": 1.851e-05,
|
| 5223 |
+
"loss": 5.6616,
|
| 5224 |
+
"step": 7450
|
| 5225 |
+
},
|
| 5226 |
+
{
|
| 5227 |
+
"epoch": 0.002330288755888196,
|
| 5228 |
+
"grad_norm": 17.125,
|
| 5229 |
+
"learning_rate": 1.8508000000000003e-05,
|
| 5230 |
+
"loss": 5.4864,
|
| 5231 |
+
"step": 7460
|
| 5232 |
+
},
|
| 5233 |
+
{
|
| 5234 |
+
"epoch": 0.002333412467357215,
|
| 5235 |
+
"grad_norm": 15.0,
|
| 5236 |
+
"learning_rate": 1.8506000000000002e-05,
|
| 5237 |
+
"loss": 5.725,
|
| 5238 |
+
"step": 7470
|
| 5239 |
+
},
|
| 5240 |
+
{
|
| 5241 |
+
"epoch": 0.0023365361788262344,
|
| 5242 |
+
"grad_norm": 17.5,
|
| 5243 |
+
"learning_rate": 1.8504e-05,
|
| 5244 |
+
"loss": 5.4064,
|
| 5245 |
+
"step": 7480
|
| 5246 |
+
},
|
| 5247 |
+
{
|
| 5248 |
+
"epoch": 0.0023396598902952533,
|
| 5249 |
+
"grad_norm": 13.6875,
|
| 5250 |
+
"learning_rate": 1.8502000000000003e-05,
|
| 5251 |
+
"loss": 5.4624,
|
| 5252 |
+
"step": 7490
|
| 5253 |
+
},
|
| 5254 |
+
{
|
| 5255 |
+
"epoch": 0.0023427836017642723,
|
| 5256 |
+
"grad_norm": 18.625,
|
| 5257 |
+
"learning_rate": 1.8500000000000002e-05,
|
| 5258 |
+
"loss": 5.4293,
|
| 5259 |
+
"step": 7500
|
| 5260 |
+
},
|
| 5261 |
+
{
|
| 5262 |
+
"epoch": 0.0023459073132332912,
|
| 5263 |
+
"grad_norm": 11.1875,
|
| 5264 |
+
"learning_rate": 1.8498e-05,
|
| 5265 |
+
"loss": 5.7245,
|
| 5266 |
+
"step": 7510
|
| 5267 |
+
},
|
| 5268 |
+
{
|
| 5269 |
+
"epoch": 0.00234903102470231,
|
| 5270 |
+
"grad_norm": 15.75,
|
| 5271 |
+
"learning_rate": 1.8496e-05,
|
| 5272 |
+
"loss": 5.6408,
|
| 5273 |
+
"step": 7520
|
| 5274 |
+
},
|
| 5275 |
+
{
|
| 5276 |
+
"epoch": 0.002352154736171329,
|
| 5277 |
+
"grad_norm": 18.875,
|
| 5278 |
+
"learning_rate": 1.8494e-05,
|
| 5279 |
+
"loss": 5.6747,
|
| 5280 |
+
"step": 7530
|
| 5281 |
+
},
|
| 5282 |
+
{
|
| 5283 |
+
"epoch": 0.0023552784476403485,
|
| 5284 |
+
"grad_norm": 13.875,
|
| 5285 |
+
"learning_rate": 1.8492000000000002e-05,
|
| 5286 |
+
"loss": 5.0168,
|
| 5287 |
+
"step": 7540
|
| 5288 |
+
},
|
| 5289 |
+
{
|
| 5290 |
+
"epoch": 0.0023584021591093675,
|
| 5291 |
+
"grad_norm": 12.125,
|
| 5292 |
+
"learning_rate": 1.849e-05,
|
| 5293 |
+
"loss": 5.0578,
|
| 5294 |
+
"step": 7550
|
| 5295 |
+
},
|
| 5296 |
+
{
|
| 5297 |
+
"epoch": 0.0023615258705783865,
|
| 5298 |
+
"grad_norm": 22.875,
|
| 5299 |
+
"learning_rate": 1.8488e-05,
|
| 5300 |
+
"loss": 5.3784,
|
| 5301 |
+
"step": 7560
|
| 5302 |
+
},
|
| 5303 |
+
{
|
| 5304 |
+
"epoch": 0.0023646495820474054,
|
| 5305 |
+
"grad_norm": 15.5,
|
| 5306 |
+
"learning_rate": 1.8486000000000002e-05,
|
| 5307 |
+
"loss": 5.3212,
|
| 5308 |
+
"step": 7570
|
| 5309 |
+
},
|
| 5310 |
+
{
|
| 5311 |
+
"epoch": 0.0023677732935164244,
|
| 5312 |
+
"grad_norm": 11.9375,
|
| 5313 |
+
"learning_rate": 1.8484e-05,
|
| 5314 |
+
"loss": 5.4473,
|
| 5315 |
+
"step": 7580
|
| 5316 |
+
},
|
| 5317 |
+
{
|
| 5318 |
+
"epoch": 0.0023708970049854433,
|
| 5319 |
+
"grad_norm": 13.9375,
|
| 5320 |
+
"learning_rate": 1.8482000000000004e-05,
|
| 5321 |
+
"loss": 5.2285,
|
| 5322 |
+
"step": 7590
|
| 5323 |
+
},
|
| 5324 |
+
{
|
| 5325 |
+
"epoch": 0.0023740207164544627,
|
| 5326 |
+
"grad_norm": 21.5,
|
| 5327 |
+
"learning_rate": 1.8480000000000003e-05,
|
| 5328 |
+
"loss": 5.6203,
|
| 5329 |
+
"step": 7600
|
| 5330 |
+
},
|
| 5331 |
+
{
|
| 5332 |
+
"epoch": 0.0023771444279234817,
|
| 5333 |
+
"grad_norm": 21.875,
|
| 5334 |
+
"learning_rate": 1.8478e-05,
|
| 5335 |
+
"loss": 5.5058,
|
| 5336 |
+
"step": 7610
|
| 5337 |
+
},
|
| 5338 |
+
{
|
| 5339 |
+
"epoch": 0.0023802681393925006,
|
| 5340 |
+
"grad_norm": 15.8125,
|
| 5341 |
+
"learning_rate": 1.8476e-05,
|
| 5342 |
+
"loss": 5.4249,
|
| 5343 |
+
"step": 7620
|
| 5344 |
+
},
|
| 5345 |
+
{
|
| 5346 |
+
"epoch": 0.0023833918508615196,
|
| 5347 |
+
"grad_norm": 13.0625,
|
| 5348 |
+
"learning_rate": 1.8474e-05,
|
| 5349 |
+
"loss": 5.3753,
|
| 5350 |
+
"step": 7630
|
| 5351 |
+
},
|
| 5352 |
+
{
|
| 5353 |
+
"epoch": 0.0023865155623305385,
|
| 5354 |
+
"grad_norm": 18.0,
|
| 5355 |
+
"learning_rate": 1.8472000000000002e-05,
|
| 5356 |
+
"loss": 5.4768,
|
| 5357 |
+
"step": 7640
|
| 5358 |
+
},
|
| 5359 |
+
{
|
| 5360 |
+
"epoch": 0.0023896392737995575,
|
| 5361 |
+
"grad_norm": 16.0,
|
| 5362 |
+
"learning_rate": 1.847e-05,
|
| 5363 |
+
"loss": 5.4579,
|
| 5364 |
+
"step": 7650
|
| 5365 |
+
},
|
| 5366 |
+
{
|
| 5367 |
+
"epoch": 0.002392762985268577,
|
| 5368 |
+
"grad_norm": 14.8125,
|
| 5369 |
+
"learning_rate": 1.8468e-05,
|
| 5370 |
+
"loss": 5.6076,
|
| 5371 |
+
"step": 7660
|
| 5372 |
+
},
|
| 5373 |
+
{
|
| 5374 |
+
"epoch": 0.002395886696737596,
|
| 5375 |
+
"grad_norm": 15.4375,
|
| 5376 |
+
"learning_rate": 1.8466000000000002e-05,
|
| 5377 |
+
"loss": 5.3755,
|
| 5378 |
+
"step": 7670
|
| 5379 |
+
},
|
| 5380 |
+
{
|
| 5381 |
+
"epoch": 0.002399010408206615,
|
| 5382 |
+
"grad_norm": 15.9375,
|
| 5383 |
+
"learning_rate": 1.8464e-05,
|
| 5384 |
+
"loss": 5.3464,
|
| 5385 |
+
"step": 7680
|
| 5386 |
+
},
|
| 5387 |
+
{
|
| 5388 |
+
"epoch": 0.0024021341196756338,
|
| 5389 |
+
"grad_norm": 16.25,
|
| 5390 |
+
"learning_rate": 1.8462000000000004e-05,
|
| 5391 |
+
"loss": 5.4139,
|
| 5392 |
+
"step": 7690
|
| 5393 |
+
},
|
| 5394 |
+
{
|
| 5395 |
+
"epoch": 0.0024052578311446527,
|
| 5396 |
+
"grad_norm": 54.25,
|
| 5397 |
+
"learning_rate": 1.8460000000000003e-05,
|
| 5398 |
+
"loss": 5.1928,
|
| 5399 |
+
"step": 7700
|
| 5400 |
+
},
|
| 5401 |
+
{
|
| 5402 |
+
"epoch": 0.0024083815426136717,
|
| 5403 |
+
"grad_norm": 11.75,
|
| 5404 |
+
"learning_rate": 1.8458000000000002e-05,
|
| 5405 |
+
"loss": 5.8011,
|
| 5406 |
+
"step": 7710
|
| 5407 |
+
},
|
| 5408 |
+
{
|
| 5409 |
+
"epoch": 0.002411505254082691,
|
| 5410 |
+
"grad_norm": 11.9375,
|
| 5411 |
+
"learning_rate": 1.8456e-05,
|
| 5412 |
+
"loss": 5.7197,
|
| 5413 |
+
"step": 7720
|
| 5414 |
+
},
|
| 5415 |
+
{
|
| 5416 |
+
"epoch": 0.00241462896555171,
|
| 5417 |
+
"grad_norm": 15.8125,
|
| 5418 |
+
"learning_rate": 1.8454e-05,
|
| 5419 |
+
"loss": 5.7809,
|
| 5420 |
+
"step": 7730
|
| 5421 |
+
},
|
| 5422 |
+
{
|
| 5423 |
+
"epoch": 0.002417752677020729,
|
| 5424 |
+
"grad_norm": 18.75,
|
| 5425 |
+
"learning_rate": 1.8452000000000002e-05,
|
| 5426 |
+
"loss": 5.5839,
|
| 5427 |
+
"step": 7740
|
| 5428 |
+
},
|
| 5429 |
+
{
|
| 5430 |
+
"epoch": 0.002420876388489748,
|
| 5431 |
+
"grad_norm": 14.5625,
|
| 5432 |
+
"learning_rate": 1.845e-05,
|
| 5433 |
+
"loss": 5.7785,
|
| 5434 |
+
"step": 7750
|
| 5435 |
+
},
|
| 5436 |
+
{
|
| 5437 |
+
"epoch": 0.002424000099958767,
|
| 5438 |
+
"grad_norm": 20.5,
|
| 5439 |
+
"learning_rate": 1.8448e-05,
|
| 5440 |
+
"loss": 5.124,
|
| 5441 |
+
"step": 7760
|
| 5442 |
+
},
|
| 5443 |
+
{
|
| 5444 |
+
"epoch": 0.002427123811427786,
|
| 5445 |
+
"grad_norm": 21.75,
|
| 5446 |
+
"learning_rate": 1.8446000000000003e-05,
|
| 5447 |
+
"loss": 5.7176,
|
| 5448 |
+
"step": 7770
|
| 5449 |
+
},
|
| 5450 |
+
{
|
| 5451 |
+
"epoch": 0.0024302475228968052,
|
| 5452 |
+
"grad_norm": 16.25,
|
| 5453 |
+
"learning_rate": 1.8444e-05,
|
| 5454 |
+
"loss": 5.5727,
|
| 5455 |
+
"step": 7780
|
| 5456 |
+
},
|
| 5457 |
+
{
|
| 5458 |
+
"epoch": 0.002433371234365824,
|
| 5459 |
+
"grad_norm": 19.0,
|
| 5460 |
+
"learning_rate": 1.8442e-05,
|
| 5461 |
+
"loss": 5.5922,
|
| 5462 |
+
"step": 7790
|
| 5463 |
+
},
|
| 5464 |
+
{
|
| 5465 |
+
"epoch": 0.002436494945834843,
|
| 5466 |
+
"grad_norm": 15.9375,
|
| 5467 |
+
"learning_rate": 1.8440000000000003e-05,
|
| 5468 |
+
"loss": 5.4033,
|
| 5469 |
+
"step": 7800
|
| 5470 |
+
},
|
| 5471 |
+
{
|
| 5472 |
+
"epoch": 0.002439618657303862,
|
| 5473 |
+
"grad_norm": 11.875,
|
| 5474 |
+
"learning_rate": 1.8438000000000002e-05,
|
| 5475 |
+
"loss": 5.3833,
|
| 5476 |
+
"step": 7810
|
| 5477 |
+
},
|
| 5478 |
+
{
|
| 5479 |
+
"epoch": 0.002442742368772881,
|
| 5480 |
+
"grad_norm": 14.875,
|
| 5481 |
+
"learning_rate": 1.8436e-05,
|
| 5482 |
+
"loss": 5.5731,
|
| 5483 |
+
"step": 7820
|
| 5484 |
+
},
|
| 5485 |
+
{
|
| 5486 |
+
"epoch": 0.0024458660802419,
|
| 5487 |
+
"grad_norm": 13.875,
|
| 5488 |
+
"learning_rate": 1.8434e-05,
|
| 5489 |
+
"loss": 5.2196,
|
| 5490 |
+
"step": 7830
|
| 5491 |
+
},
|
| 5492 |
+
{
|
| 5493 |
+
"epoch": 0.0024489897917109194,
|
| 5494 |
+
"grad_norm": 17.0,
|
| 5495 |
+
"learning_rate": 1.8432000000000002e-05,
|
| 5496 |
+
"loss": 5.5524,
|
| 5497 |
+
"step": 7840
|
| 5498 |
+
},
|
| 5499 |
+
{
|
| 5500 |
+
"epoch": 0.0024521135031799384,
|
| 5501 |
+
"grad_norm": 19.75,
|
| 5502 |
+
"learning_rate": 1.843e-05,
|
| 5503 |
+
"loss": 5.3533,
|
| 5504 |
+
"step": 7850
|
| 5505 |
+
},
|
| 5506 |
+
{
|
| 5507 |
+
"epoch": 0.0024552372146489573,
|
| 5508 |
+
"grad_norm": 18.375,
|
| 5509 |
+
"learning_rate": 1.8428e-05,
|
| 5510 |
+
"loss": 5.2578,
|
| 5511 |
+
"step": 7860
|
| 5512 |
+
},
|
| 5513 |
+
{
|
| 5514 |
+
"epoch": 0.0024583609261179763,
|
| 5515 |
+
"grad_norm": 13.0,
|
| 5516 |
+
"learning_rate": 1.8426000000000003e-05,
|
| 5517 |
+
"loss": 5.6817,
|
| 5518 |
+
"step": 7870
|
| 5519 |
+
},
|
| 5520 |
+
{
|
| 5521 |
+
"epoch": 0.0024614846375869952,
|
| 5522 |
+
"grad_norm": 20.375,
|
| 5523 |
+
"learning_rate": 1.8424000000000002e-05,
|
| 5524 |
+
"loss": 5.6943,
|
| 5525 |
+
"step": 7880
|
| 5526 |
+
},
|
| 5527 |
+
{
|
| 5528 |
+
"epoch": 0.002464608349056014,
|
| 5529 |
+
"grad_norm": 17.875,
|
| 5530 |
+
"learning_rate": 1.8422e-05,
|
| 5531 |
+
"loss": 5.7388,
|
| 5532 |
+
"step": 7890
|
| 5533 |
+
},
|
| 5534 |
+
{
|
| 5535 |
+
"epoch": 0.0024677320605250336,
|
| 5536 |
+
"grad_norm": 14.9375,
|
| 5537 |
+
"learning_rate": 1.8420000000000003e-05,
|
| 5538 |
+
"loss": 5.6932,
|
| 5539 |
+
"step": 7900
|
| 5540 |
+
},
|
| 5541 |
+
{
|
| 5542 |
+
"epoch": 0.0024708557719940526,
|
| 5543 |
+
"grad_norm": 15.75,
|
| 5544 |
+
"learning_rate": 1.8418000000000002e-05,
|
| 5545 |
+
"loss": 5.6659,
|
| 5546 |
+
"step": 7910
|
| 5547 |
+
},
|
| 5548 |
+
{
|
| 5549 |
+
"epoch": 0.0024739794834630715,
|
| 5550 |
+
"grad_norm": 15.8125,
|
| 5551 |
+
"learning_rate": 1.8416e-05,
|
| 5552 |
+
"loss": 5.7062,
|
| 5553 |
+
"step": 7920
|
| 5554 |
+
},
|
| 5555 |
+
{
|
| 5556 |
+
"epoch": 0.0024771031949320905,
|
| 5557 |
+
"grad_norm": 17.5,
|
| 5558 |
+
"learning_rate": 1.8414e-05,
|
| 5559 |
+
"loss": 5.4439,
|
| 5560 |
+
"step": 7930
|
| 5561 |
+
},
|
| 5562 |
+
{
|
| 5563 |
+
"epoch": 0.0024802269064011094,
|
| 5564 |
+
"grad_norm": 17.875,
|
| 5565 |
+
"learning_rate": 1.8412000000000003e-05,
|
| 5566 |
+
"loss": 5.5437,
|
| 5567 |
+
"step": 7940
|
| 5568 |
+
},
|
| 5569 |
+
{
|
| 5570 |
+
"epoch": 0.0024833506178701284,
|
| 5571 |
+
"grad_norm": 19.0,
|
| 5572 |
+
"learning_rate": 1.841e-05,
|
| 5573 |
+
"loss": 5.6529,
|
| 5574 |
+
"step": 7950
|
| 5575 |
+
},
|
| 5576 |
+
{
|
| 5577 |
+
"epoch": 0.0024864743293391478,
|
| 5578 |
+
"grad_norm": 11.75,
|
| 5579 |
+
"learning_rate": 1.8408e-05,
|
| 5580 |
+
"loss": 5.5425,
|
| 5581 |
+
"step": 7960
|
| 5582 |
+
},
|
| 5583 |
+
{
|
| 5584 |
+
"epoch": 0.0024895980408081667,
|
| 5585 |
+
"grad_norm": 19.375,
|
| 5586 |
+
"learning_rate": 1.8406000000000003e-05,
|
| 5587 |
+
"loss": 5.6127,
|
| 5588 |
+
"step": 7970
|
| 5589 |
+
},
|
| 5590 |
+
{
|
| 5591 |
+
"epoch": 0.0024927217522771857,
|
| 5592 |
+
"grad_norm": 14.25,
|
| 5593 |
+
"learning_rate": 1.8404000000000002e-05,
|
| 5594 |
+
"loss": 5.5463,
|
| 5595 |
+
"step": 7980
|
| 5596 |
+
},
|
| 5597 |
+
{
|
| 5598 |
+
"epoch": 0.0024958454637462046,
|
| 5599 |
+
"grad_norm": 11.1875,
|
| 5600 |
+
"learning_rate": 1.8402e-05,
|
| 5601 |
+
"loss": 5.6528,
|
| 5602 |
+
"step": 7990
|
| 5603 |
+
},
|
| 5604 |
+
{
|
| 5605 |
+
"epoch": 0.0024989691752152236,
|
| 5606 |
+
"grad_norm": 16.125,
|
| 5607 |
+
"learning_rate": 1.8400000000000003e-05,
|
| 5608 |
+
"loss": 5.6768,
|
| 5609 |
+
"step": 8000
|
| 5610 |
+
},
|
| 5611 |
+
{
|
| 5612 |
+
"epoch": 0.0025020928866842426,
|
| 5613 |
+
"grad_norm": 12.9375,
|
| 5614 |
+
"learning_rate": 1.8398000000000002e-05,
|
| 5615 |
+
"loss": 5.529,
|
| 5616 |
+
"step": 8010
|
| 5617 |
+
},
|
| 5618 |
+
{
|
| 5619 |
+
"epoch": 0.002505216598153262,
|
| 5620 |
+
"grad_norm": 19.25,
|
| 5621 |
+
"learning_rate": 1.8396e-05,
|
| 5622 |
+
"loss": 5.6349,
|
| 5623 |
+
"step": 8020
|
| 5624 |
+
},
|
| 5625 |
+
{
|
| 5626 |
+
"epoch": 0.002508340309622281,
|
| 5627 |
+
"grad_norm": 15.8125,
|
| 5628 |
+
"learning_rate": 1.8394e-05,
|
| 5629 |
+
"loss": 5.7431,
|
| 5630 |
+
"step": 8030
|
| 5631 |
+
},
|
| 5632 |
+
{
|
| 5633 |
+
"epoch": 0.0025114640210913,
|
| 5634 |
+
"grad_norm": 18.125,
|
| 5635 |
+
"learning_rate": 1.8392e-05,
|
| 5636 |
+
"loss": 5.6143,
|
| 5637 |
+
"step": 8040
|
| 5638 |
+
},
|
| 5639 |
+
{
|
| 5640 |
+
"epoch": 0.002514587732560319,
|
| 5641 |
+
"grad_norm": 25.75,
|
| 5642 |
+
"learning_rate": 1.8390000000000002e-05,
|
| 5643 |
+
"loss": 5.5138,
|
| 5644 |
+
"step": 8050
|
| 5645 |
+
},
|
| 5646 |
+
{
|
| 5647 |
+
"epoch": 0.0025177114440293378,
|
| 5648 |
+
"grad_norm": 12.375,
|
| 5649 |
+
"learning_rate": 1.8388e-05,
|
| 5650 |
+
"loss": 5.5673,
|
| 5651 |
+
"step": 8060
|
| 5652 |
+
},
|
| 5653 |
+
{
|
| 5654 |
+
"epoch": 0.0025208351554983567,
|
| 5655 |
+
"grad_norm": 12.875,
|
| 5656 |
+
"learning_rate": 1.8386000000000003e-05,
|
| 5657 |
+
"loss": 5.7416,
|
| 5658 |
+
"step": 8070
|
| 5659 |
+
},
|
| 5660 |
+
{
|
| 5661 |
+
"epoch": 0.002523958866967376,
|
| 5662 |
+
"grad_norm": 13.9375,
|
| 5663 |
+
"learning_rate": 1.8384000000000002e-05,
|
| 5664 |
+
"loss": 5.6055,
|
| 5665 |
+
"step": 8080
|
| 5666 |
+
},
|
| 5667 |
+
{
|
| 5668 |
+
"epoch": 0.002527082578436395,
|
| 5669 |
+
"grad_norm": 15.4375,
|
| 5670 |
+
"learning_rate": 1.8382e-05,
|
| 5671 |
+
"loss": 5.5338,
|
| 5672 |
+
"step": 8090
|
| 5673 |
+
},
|
| 5674 |
+
{
|
| 5675 |
+
"epoch": 0.002530206289905414,
|
| 5676 |
+
"grad_norm": 24.75,
|
| 5677 |
+
"learning_rate": 1.8380000000000004e-05,
|
| 5678 |
+
"loss": 5.5553,
|
| 5679 |
+
"step": 8100
|
| 5680 |
+
},
|
| 5681 |
+
{
|
| 5682 |
+
"epoch": 0.002533330001374433,
|
| 5683 |
+
"grad_norm": 16.625,
|
| 5684 |
+
"learning_rate": 1.8378000000000003e-05,
|
| 5685 |
+
"loss": 5.4739,
|
| 5686 |
+
"step": 8110
|
| 5687 |
+
},
|
| 5688 |
+
{
|
| 5689 |
+
"epoch": 0.002536453712843452,
|
| 5690 |
+
"grad_norm": 20.25,
|
| 5691 |
+
"learning_rate": 1.8376e-05,
|
| 5692 |
+
"loss": 5.4292,
|
| 5693 |
+
"step": 8120
|
| 5694 |
+
},
|
| 5695 |
+
{
|
| 5696 |
+
"epoch": 0.002539577424312471,
|
| 5697 |
+
"grad_norm": 22.5,
|
| 5698 |
+
"learning_rate": 1.8374e-05,
|
| 5699 |
+
"loss": 5.6956,
|
| 5700 |
+
"step": 8130
|
| 5701 |
+
},
|
| 5702 |
+
{
|
| 5703 |
+
"epoch": 0.0025427011357814903,
|
| 5704 |
+
"grad_norm": 12.8125,
|
| 5705 |
+
"learning_rate": 1.8372e-05,
|
| 5706 |
+
"loss": 5.4388,
|
| 5707 |
+
"step": 8140
|
| 5708 |
+
},
|
| 5709 |
+
{
|
| 5710 |
+
"epoch": 0.0025458248472505093,
|
| 5711 |
+
"grad_norm": 13.6875,
|
| 5712 |
+
"learning_rate": 1.8370000000000002e-05,
|
| 5713 |
+
"loss": 5.3834,
|
| 5714 |
+
"step": 8150
|
| 5715 |
+
},
|
| 5716 |
+
{
|
| 5717 |
+
"epoch": 0.002548948558719528,
|
| 5718 |
+
"grad_norm": 17.625,
|
| 5719 |
+
"learning_rate": 1.8368e-05,
|
| 5720 |
+
"loss": 5.6013,
|
| 5721 |
+
"step": 8160
|
| 5722 |
+
},
|
| 5723 |
+
{
|
| 5724 |
+
"epoch": 0.002552072270188547,
|
| 5725 |
+
"grad_norm": 13.1875,
|
| 5726 |
+
"learning_rate": 1.8366e-05,
|
| 5727 |
+
"loss": 5.6469,
|
| 5728 |
+
"step": 8170
|
| 5729 |
+
},
|
| 5730 |
+
{
|
| 5731 |
+
"epoch": 0.002555195981657566,
|
| 5732 |
+
"grad_norm": 13.4375,
|
| 5733 |
+
"learning_rate": 1.8364000000000002e-05,
|
| 5734 |
+
"loss": 5.7029,
|
| 5735 |
+
"step": 8180
|
| 5736 |
+
},
|
| 5737 |
+
{
|
| 5738 |
+
"epoch": 0.002558319693126585,
|
| 5739 |
+
"grad_norm": 16.75,
|
| 5740 |
+
"learning_rate": 1.8362e-05,
|
| 5741 |
+
"loss": 5.6941,
|
| 5742 |
+
"step": 8190
|
| 5743 |
+
},
|
| 5744 |
+
{
|
| 5745 |
+
"epoch": 0.0025614434045956045,
|
| 5746 |
+
"grad_norm": 14.375,
|
| 5747 |
+
"learning_rate": 1.8360000000000004e-05,
|
| 5748 |
+
"loss": 5.6255,
|
| 5749 |
+
"step": 8200
|
| 5750 |
+
},
|
| 5751 |
+
{
|
| 5752 |
+
"epoch": 0.0025645671160646234,
|
| 5753 |
+
"grad_norm": 12.8125,
|
| 5754 |
+
"learning_rate": 1.8358000000000003e-05,
|
| 5755 |
+
"loss": 5.5778,
|
| 5756 |
+
"step": 8210
|
| 5757 |
+
},
|
| 5758 |
+
{
|
| 5759 |
+
"epoch": 0.0025676908275336424,
|
| 5760 |
+
"grad_norm": 13.0625,
|
| 5761 |
+
"learning_rate": 1.8356000000000002e-05,
|
| 5762 |
+
"loss": 5.7121,
|
| 5763 |
+
"step": 8220
|
| 5764 |
+
},
|
| 5765 |
+
{
|
| 5766 |
+
"epoch": 0.0025708145390026613,
|
| 5767 |
+
"grad_norm": 18.5,
|
| 5768 |
+
"learning_rate": 1.8354e-05,
|
| 5769 |
+
"loss": 5.6337,
|
| 5770 |
+
"step": 8230
|
| 5771 |
+
},
|
| 5772 |
+
{
|
| 5773 |
+
"epoch": 0.0025739382504716803,
|
| 5774 |
+
"grad_norm": 14.75,
|
| 5775 |
+
"learning_rate": 1.8352e-05,
|
| 5776 |
+
"loss": 5.781,
|
| 5777 |
+
"step": 8240
|
| 5778 |
+
},
|
| 5779 |
+
{
|
| 5780 |
+
"epoch": 0.0025770619619406993,
|
| 5781 |
+
"grad_norm": 16.375,
|
| 5782 |
+
"learning_rate": 1.8350000000000002e-05,
|
| 5783 |
+
"loss": 5.7368,
|
| 5784 |
+
"step": 8250
|
| 5785 |
+
},
|
| 5786 |
+
{
|
| 5787 |
+
"epoch": 0.0025801856734097186,
|
| 5788 |
+
"grad_norm": 14.4375,
|
| 5789 |
+
"learning_rate": 1.8348e-05,
|
| 5790 |
+
"loss": 5.614,
|
| 5791 |
+
"step": 8260
|
| 5792 |
+
},
|
| 5793 |
+
{
|
| 5794 |
+
"epoch": 0.0025833093848787376,
|
| 5795 |
+
"grad_norm": 12.9375,
|
| 5796 |
+
"learning_rate": 1.8346e-05,
|
| 5797 |
+
"loss": 5.2018,
|
| 5798 |
+
"step": 8270
|
| 5799 |
+
},
|
| 5800 |
+
{
|
| 5801 |
+
"epoch": 0.0025864330963477566,
|
| 5802 |
+
"grad_norm": 16.25,
|
| 5803 |
+
"learning_rate": 1.8344000000000003e-05,
|
| 5804 |
+
"loss": 5.5791,
|
| 5805 |
+
"step": 8280
|
| 5806 |
+
},
|
| 5807 |
+
{
|
| 5808 |
+
"epoch": 0.0025895568078167755,
|
| 5809 |
+
"grad_norm": 13.3125,
|
| 5810 |
+
"learning_rate": 1.8342e-05,
|
| 5811 |
+
"loss": 5.7152,
|
| 5812 |
+
"step": 8290
|
| 5813 |
+
},
|
| 5814 |
+
{
|
| 5815 |
+
"epoch": 0.0025926805192857945,
|
| 5816 |
+
"grad_norm": 20.5,
|
| 5817 |
+
"learning_rate": 1.834e-05,
|
| 5818 |
+
"loss": 5.6225,
|
| 5819 |
+
"step": 8300
|
| 5820 |
+
},
|
| 5821 |
+
{
|
| 5822 |
+
"epoch": 0.0025958042307548134,
|
| 5823 |
+
"grad_norm": 18.625,
|
| 5824 |
+
"learning_rate": 1.8338000000000003e-05,
|
| 5825 |
+
"loss": 5.5868,
|
| 5826 |
+
"step": 8310
|
| 5827 |
+
},
|
| 5828 |
+
{
|
| 5829 |
+
"epoch": 0.002598927942223833,
|
| 5830 |
+
"grad_norm": 15.625,
|
| 5831 |
+
"learning_rate": 1.8336000000000002e-05,
|
| 5832 |
+
"loss": 5.1618,
|
| 5833 |
+
"step": 8320
|
| 5834 |
+
},
|
| 5835 |
+
{
|
| 5836 |
+
"epoch": 0.0026020516536928518,
|
| 5837 |
+
"grad_norm": 15.25,
|
| 5838 |
+
"learning_rate": 1.8334e-05,
|
| 5839 |
+
"loss": 5.6222,
|
| 5840 |
+
"step": 8330
|
| 5841 |
+
},
|
| 5842 |
+
{
|
| 5843 |
+
"epoch": 0.0026051753651618707,
|
| 5844 |
+
"grad_norm": 14.5625,
|
| 5845 |
+
"learning_rate": 1.8332e-05,
|
| 5846 |
+
"loss": 5.441,
|
| 5847 |
+
"step": 8340
|
| 5848 |
+
},
|
| 5849 |
+
{
|
| 5850 |
+
"epoch": 0.0026082990766308897,
|
| 5851 |
+
"grad_norm": 20.625,
|
| 5852 |
+
"learning_rate": 1.8330000000000002e-05,
|
| 5853 |
+
"loss": 5.8026,
|
| 5854 |
+
"step": 8350
|
| 5855 |
+
},
|
| 5856 |
+
{
|
| 5857 |
+
"epoch": 0.0026114227880999086,
|
| 5858 |
+
"grad_norm": 16.0,
|
| 5859 |
+
"learning_rate": 1.8328e-05,
|
| 5860 |
+
"loss": 5.643,
|
| 5861 |
+
"step": 8360
|
| 5862 |
+
},
|
| 5863 |
+
{
|
| 5864 |
+
"epoch": 0.0026145464995689276,
|
| 5865 |
+
"grad_norm": 18.5,
|
| 5866 |
+
"learning_rate": 1.8326e-05,
|
| 5867 |
+
"loss": 5.744,
|
| 5868 |
+
"step": 8370
|
| 5869 |
+
},
|
| 5870 |
+
{
|
| 5871 |
+
"epoch": 0.002617670211037947,
|
| 5872 |
+
"grad_norm": 12.3125,
|
| 5873 |
+
"learning_rate": 1.8324000000000003e-05,
|
| 5874 |
+
"loss": 5.8405,
|
| 5875 |
+
"step": 8380
|
| 5876 |
+
},
|
| 5877 |
+
{
|
| 5878 |
+
"epoch": 0.002620793922506966,
|
| 5879 |
+
"grad_norm": 14.125,
|
| 5880 |
+
"learning_rate": 1.8322000000000002e-05,
|
| 5881 |
+
"loss": 5.6224,
|
| 5882 |
+
"step": 8390
|
| 5883 |
+
},
|
| 5884 |
+
{
|
| 5885 |
+
"epoch": 0.002623917633975985,
|
| 5886 |
+
"grad_norm": 14.3125,
|
| 5887 |
+
"learning_rate": 1.832e-05,
|
| 5888 |
+
"loss": 5.5866,
|
| 5889 |
+
"step": 8400
|
| 5890 |
+
},
|
| 5891 |
+
{
|
| 5892 |
+
"epoch": 0.002627041345445004,
|
| 5893 |
+
"grad_norm": 14.875,
|
| 5894 |
+
"learning_rate": 1.8318000000000003e-05,
|
| 5895 |
+
"loss": 5.6274,
|
| 5896 |
+
"step": 8410
|
| 5897 |
+
},
|
| 5898 |
+
{
|
| 5899 |
+
"epoch": 0.002630165056914023,
|
| 5900 |
+
"grad_norm": 18.375,
|
| 5901 |
+
"learning_rate": 1.8316e-05,
|
| 5902 |
+
"loss": 5.71,
|
| 5903 |
+
"step": 8420
|
| 5904 |
+
},
|
| 5905 |
+
{
|
| 5906 |
+
"epoch": 0.002633288768383042,
|
| 5907 |
+
"grad_norm": 14.625,
|
| 5908 |
+
"learning_rate": 1.8314e-05,
|
| 5909 |
+
"loss": 5.8016,
|
| 5910 |
+
"step": 8430
|
| 5911 |
+
},
|
| 5912 |
+
{
|
| 5913 |
+
"epoch": 0.002636412479852061,
|
| 5914 |
+
"grad_norm": 13.6875,
|
| 5915 |
+
"learning_rate": 1.8312e-05,
|
| 5916 |
+
"loss": 5.7552,
|
| 5917 |
+
"step": 8440
|
| 5918 |
+
},
|
| 5919 |
+
{
|
| 5920 |
+
"epoch": 0.00263953619132108,
|
| 5921 |
+
"grad_norm": 15.8125,
|
| 5922 |
+
"learning_rate": 1.8310000000000003e-05,
|
| 5923 |
+
"loss": 5.7443,
|
| 5924 |
+
"step": 8450
|
| 5925 |
+
},
|
| 5926 |
+
{
|
| 5927 |
+
"epoch": 0.002642659902790099,
|
| 5928 |
+
"grad_norm": 17.875,
|
| 5929 |
+
"learning_rate": 1.8308e-05,
|
| 5930 |
+
"loss": 5.7018,
|
| 5931 |
+
"step": 8460
|
| 5932 |
+
},
|
| 5933 |
+
{
|
| 5934 |
+
"epoch": 0.002645783614259118,
|
| 5935 |
+
"grad_norm": 14.0,
|
| 5936 |
+
"learning_rate": 1.8306e-05,
|
| 5937 |
+
"loss": 5.7637,
|
| 5938 |
+
"step": 8470
|
| 5939 |
+
},
|
| 5940 |
+
{
|
| 5941 |
+
"epoch": 0.002648907325728137,
|
| 5942 |
+
"grad_norm": 19.375,
|
| 5943 |
+
"learning_rate": 1.8304000000000003e-05,
|
| 5944 |
+
"loss": 5.6508,
|
| 5945 |
+
"step": 8480
|
| 5946 |
+
},
|
| 5947 |
+
{
|
| 5948 |
+
"epoch": 0.0026520310371971564,
|
| 5949 |
+
"grad_norm": 11.5625,
|
| 5950 |
+
"learning_rate": 1.8302000000000002e-05,
|
| 5951 |
+
"loss": 5.3358,
|
| 5952 |
+
"step": 8490
|
| 5953 |
+
},
|
| 5954 |
+
{
|
| 5955 |
+
"epoch": 0.0026551547486661753,
|
| 5956 |
+
"grad_norm": 13.3125,
|
| 5957 |
+
"learning_rate": 1.83e-05,
|
| 5958 |
+
"loss": 5.5337,
|
| 5959 |
+
"step": 8500
|
| 5960 |
+
},
|
| 5961 |
+
{
|
| 5962 |
+
"epoch": 0.0026582784601351943,
|
| 5963 |
+
"grad_norm": 14.875,
|
| 5964 |
+
"learning_rate": 1.8298000000000003e-05,
|
| 5965 |
+
"loss": 5.5015,
|
| 5966 |
+
"step": 8510
|
| 5967 |
+
},
|
| 5968 |
+
{
|
| 5969 |
+
"epoch": 0.0026614021716042133,
|
| 5970 |
+
"grad_norm": 15.3125,
|
| 5971 |
+
"learning_rate": 1.8296e-05,
|
| 5972 |
+
"loss": 5.3137,
|
| 5973 |
+
"step": 8520
|
| 5974 |
+
},
|
| 5975 |
+
{
|
| 5976 |
+
"epoch": 0.002664525883073232,
|
| 5977 |
+
"grad_norm": 14.25,
|
| 5978 |
+
"learning_rate": 1.8294e-05,
|
| 5979 |
+
"loss": 5.6978,
|
| 5980 |
+
"step": 8530
|
| 5981 |
+
},
|
| 5982 |
+
{
|
| 5983 |
+
"epoch": 0.002667649594542251,
|
| 5984 |
+
"grad_norm": 12.5625,
|
| 5985 |
+
"learning_rate": 1.8292e-05,
|
| 5986 |
+
"loss": 5.6982,
|
| 5987 |
+
"step": 8540
|
| 5988 |
+
},
|
| 5989 |
+
{
|
| 5990 |
+
"epoch": 0.0026707733060112706,
|
| 5991 |
+
"grad_norm": 15.5,
|
| 5992 |
+
"learning_rate": 1.8290000000000003e-05,
|
| 5993 |
+
"loss": 5.6831,
|
| 5994 |
+
"step": 8550
|
| 5995 |
+
},
|
| 5996 |
+
{
|
| 5997 |
+
"epoch": 0.0026738970174802895,
|
| 5998 |
+
"grad_norm": 16.875,
|
| 5999 |
+
"learning_rate": 1.8288000000000002e-05,
|
| 6000 |
+
"loss": 5.607,
|
| 6001 |
+
"step": 8560
|
| 6002 |
+
},
|
| 6003 |
+
{
|
| 6004 |
+
"epoch": 0.0026770207289493085,
|
| 6005 |
+
"grad_norm": 15.3125,
|
| 6006 |
+
"learning_rate": 1.8286e-05,
|
| 6007 |
+
"loss": 5.4306,
|
| 6008 |
+
"step": 8570
|
| 6009 |
+
},
|
| 6010 |
+
{
|
| 6011 |
+
"epoch": 0.0026801444404183274,
|
| 6012 |
+
"grad_norm": 15.8125,
|
| 6013 |
+
"learning_rate": 1.8284000000000003e-05,
|
| 6014 |
+
"loss": 5.5547,
|
| 6015 |
+
"step": 8580
|
| 6016 |
+
},
|
| 6017 |
+
{
|
| 6018 |
+
"epoch": 0.0026832681518873464,
|
| 6019 |
+
"grad_norm": 15.25,
|
| 6020 |
+
"learning_rate": 1.8282000000000002e-05,
|
| 6021 |
+
"loss": 5.6788,
|
| 6022 |
+
"step": 8590
|
| 6023 |
+
},
|
| 6024 |
+
{
|
| 6025 |
+
"epoch": 0.0026863918633563653,
|
| 6026 |
+
"grad_norm": 14.4375,
|
| 6027 |
+
"learning_rate": 1.828e-05,
|
| 6028 |
+
"loss": 5.7335,
|
| 6029 |
+
"step": 8600
|
| 6030 |
+
},
|
| 6031 |
+
{
|
| 6032 |
+
"epoch": 0.0026895155748253847,
|
| 6033 |
+
"grad_norm": 17.375,
|
| 6034 |
+
"learning_rate": 1.8278000000000004e-05,
|
| 6035 |
+
"loss": 5.6141,
|
| 6036 |
+
"step": 8610
|
| 6037 |
+
},
|
| 6038 |
+
{
|
| 6039 |
+
"epoch": 0.0026926392862944037,
|
| 6040 |
+
"grad_norm": 15.25,
|
| 6041 |
+
"learning_rate": 1.8276e-05,
|
| 6042 |
+
"loss": 5.1071,
|
| 6043 |
+
"step": 8620
|
| 6044 |
+
},
|
| 6045 |
+
{
|
| 6046 |
+
"epoch": 0.0026957629977634226,
|
| 6047 |
+
"grad_norm": 13.875,
|
| 6048 |
+
"learning_rate": 1.8274e-05,
|
| 6049 |
+
"loss": 5.3496,
|
| 6050 |
+
"step": 8630
|
| 6051 |
+
},
|
| 6052 |
+
{
|
| 6053 |
+
"epoch": 0.0026988867092324416,
|
| 6054 |
+
"grad_norm": 13.25,
|
| 6055 |
+
"learning_rate": 1.8272e-05,
|
| 6056 |
+
"loss": 5.6896,
|
| 6057 |
+
"step": 8640
|
| 6058 |
+
},
|
| 6059 |
+
{
|
| 6060 |
+
"epoch": 0.0027020104207014606,
|
| 6061 |
+
"grad_norm": 14.8125,
|
| 6062 |
+
"learning_rate": 1.827e-05,
|
| 6063 |
+
"loss": 5.6356,
|
| 6064 |
+
"step": 8650
|
| 6065 |
+
},
|
| 6066 |
+
{
|
| 6067 |
+
"epoch": 0.0027051341321704795,
|
| 6068 |
+
"grad_norm": 14.0,
|
| 6069 |
+
"learning_rate": 1.8268000000000002e-05,
|
| 6070 |
+
"loss": 5.3134,
|
| 6071 |
+
"step": 8660
|
| 6072 |
+
},
|
| 6073 |
+
{
|
| 6074 |
+
"epoch": 0.002708257843639499,
|
| 6075 |
+
"grad_norm": 13.8125,
|
| 6076 |
+
"learning_rate": 1.8266e-05,
|
| 6077 |
+
"loss": 5.7228,
|
| 6078 |
+
"step": 8670
|
| 6079 |
+
},
|
| 6080 |
+
{
|
| 6081 |
+
"epoch": 0.002711381555108518,
|
| 6082 |
+
"grad_norm": 12.0,
|
| 6083 |
+
"learning_rate": 1.8264000000000003e-05,
|
| 6084 |
+
"loss": 5.7139,
|
| 6085 |
+
"step": 8680
|
| 6086 |
+
},
|
| 6087 |
+
{
|
| 6088 |
+
"epoch": 0.002714505266577537,
|
| 6089 |
+
"grad_norm": 16.625,
|
| 6090 |
+
"learning_rate": 1.8262000000000002e-05,
|
| 6091 |
+
"loss": 5.5275,
|
| 6092 |
+
"step": 8690
|
| 6093 |
+
},
|
| 6094 |
+
{
|
| 6095 |
+
"epoch": 0.0027176289780465558,
|
| 6096 |
+
"grad_norm": 17.0,
|
| 6097 |
+
"learning_rate": 1.826e-05,
|
| 6098 |
+
"loss": 5.4343,
|
| 6099 |
+
"step": 8700
|
| 6100 |
+
},
|
| 6101 |
+
{
|
| 6102 |
+
"epoch": 0.0027207526895155747,
|
| 6103 |
+
"grad_norm": 17.5,
|
| 6104 |
+
"learning_rate": 1.8258000000000004e-05,
|
| 6105 |
+
"loss": 5.7115,
|
| 6106 |
+
"step": 8710
|
| 6107 |
+
},
|
| 6108 |
+
{
|
| 6109 |
+
"epoch": 0.0027238764009845937,
|
| 6110 |
+
"grad_norm": 18.875,
|
| 6111 |
+
"learning_rate": 1.8256e-05,
|
| 6112 |
+
"loss": 5.7312,
|
| 6113 |
+
"step": 8720
|
| 6114 |
+
},
|
| 6115 |
+
{
|
| 6116 |
+
"epoch": 0.002727000112453613,
|
| 6117 |
+
"grad_norm": 13.4375,
|
| 6118 |
+
"learning_rate": 1.8254000000000002e-05,
|
| 6119 |
+
"loss": 5.783,
|
| 6120 |
+
"step": 8730
|
| 6121 |
+
},
|
| 6122 |
+
{
|
| 6123 |
+
"epoch": 0.002730123823922632,
|
| 6124 |
+
"grad_norm": 19.875,
|
| 6125 |
+
"learning_rate": 1.8252e-05,
|
| 6126 |
+
"loss": 5.6312,
|
| 6127 |
+
"step": 8740
|
| 6128 |
+
},
|
| 6129 |
+
{
|
| 6130 |
+
"epoch": 0.002733247535391651,
|
| 6131 |
+
"grad_norm": 18.625,
|
| 6132 |
+
"learning_rate": 1.825e-05,
|
| 6133 |
+
"loss": 5.285,
|
| 6134 |
+
"step": 8750
|
| 6135 |
+
},
|
| 6136 |
+
{
|
| 6137 |
+
"epoch": 0.00273637124686067,
|
| 6138 |
+
"grad_norm": 17.125,
|
| 6139 |
+
"learning_rate": 1.8248000000000002e-05,
|
| 6140 |
+
"loss": 5.5014,
|
| 6141 |
+
"step": 8760
|
| 6142 |
+
},
|
| 6143 |
+
{
|
| 6144 |
+
"epoch": 0.002739494958329689,
|
| 6145 |
+
"grad_norm": 14.9375,
|
| 6146 |
+
"learning_rate": 1.8246e-05,
|
| 6147 |
+
"loss": 5.6598,
|
| 6148 |
+
"step": 8770
|
| 6149 |
+
},
|
| 6150 |
+
{
|
| 6151 |
+
"epoch": 0.002742618669798708,
|
| 6152 |
+
"grad_norm": 15.75,
|
| 6153 |
+
"learning_rate": 1.8244e-05,
|
| 6154 |
+
"loss": 5.323,
|
| 6155 |
+
"step": 8780
|
| 6156 |
+
},
|
| 6157 |
+
{
|
| 6158 |
+
"epoch": 0.0027457423812677273,
|
| 6159 |
+
"grad_norm": 17.25,
|
| 6160 |
+
"learning_rate": 1.8242000000000003e-05,
|
| 6161 |
+
"loss": 5.7593,
|
| 6162 |
+
"step": 8790
|
| 6163 |
+
},
|
| 6164 |
+
{
|
| 6165 |
+
"epoch": 0.002748866092736746,
|
| 6166 |
+
"grad_norm": 16.875,
|
| 6167 |
+
"learning_rate": 1.824e-05,
|
| 6168 |
+
"loss": 5.2561,
|
| 6169 |
+
"step": 8800
|
| 6170 |
+
},
|
| 6171 |
+
{
|
| 6172 |
+
"epoch": 0.002751989804205765,
|
| 6173 |
+
"grad_norm": 16.625,
|
| 6174 |
+
"learning_rate": 1.8238000000000004e-05,
|
| 6175 |
+
"loss": 5.7641,
|
| 6176 |
+
"step": 8810
|
| 6177 |
+
},
|
| 6178 |
+
{
|
| 6179 |
+
"epoch": 0.002755113515674784,
|
| 6180 |
+
"grad_norm": 12.5,
|
| 6181 |
+
"learning_rate": 1.8236000000000003e-05,
|
| 6182 |
+
"loss": 5.2643,
|
| 6183 |
+
"step": 8820
|
| 6184 |
+
},
|
| 6185 |
+
{
|
| 6186 |
+
"epoch": 0.002758237227143803,
|
| 6187 |
+
"grad_norm": 18.75,
|
| 6188 |
+
"learning_rate": 1.8234000000000002e-05,
|
| 6189 |
+
"loss": 5.7293,
|
| 6190 |
+
"step": 8830
|
| 6191 |
+
},
|
| 6192 |
+
{
|
| 6193 |
+
"epoch": 0.002761360938612822,
|
| 6194 |
+
"grad_norm": 16.5,
|
| 6195 |
+
"learning_rate": 1.8232e-05,
|
| 6196 |
+
"loss": 5.5185,
|
| 6197 |
+
"step": 8840
|
| 6198 |
+
},
|
| 6199 |
+
{
|
| 6200 |
+
"epoch": 0.0027644846500818414,
|
| 6201 |
+
"grad_norm": 14.1875,
|
| 6202 |
+
"learning_rate": 1.823e-05,
|
| 6203 |
+
"loss": 5.2863,
|
| 6204 |
+
"step": 8850
|
| 6205 |
+
},
|
| 6206 |
+
{
|
| 6207 |
+
"epoch": 0.0027676083615508604,
|
| 6208 |
+
"grad_norm": 13.5625,
|
| 6209 |
+
"learning_rate": 1.8228000000000002e-05,
|
| 6210 |
+
"loss": 5.6614,
|
| 6211 |
+
"step": 8860
|
| 6212 |
+
},
|
| 6213 |
+
{
|
| 6214 |
+
"epoch": 0.0027707320730198793,
|
| 6215 |
+
"grad_norm": 13.3125,
|
| 6216 |
+
"learning_rate": 1.8226e-05,
|
| 6217 |
+
"loss": 5.6269,
|
| 6218 |
+
"step": 8870
|
| 6219 |
+
},
|
| 6220 |
+
{
|
| 6221 |
+
"epoch": 0.0027738557844888983,
|
| 6222 |
+
"grad_norm": 17.375,
|
| 6223 |
+
"learning_rate": 1.8224e-05,
|
| 6224 |
+
"loss": 5.6308,
|
| 6225 |
+
"step": 8880
|
| 6226 |
+
},
|
| 6227 |
+
{
|
| 6228 |
+
"epoch": 0.0027769794959579173,
|
| 6229 |
+
"grad_norm": 12.5,
|
| 6230 |
+
"learning_rate": 1.8222000000000003e-05,
|
| 6231 |
+
"loss": 5.5156,
|
| 6232 |
+
"step": 8890
|
| 6233 |
+
},
|
| 6234 |
+
{
|
| 6235 |
+
"epoch": 0.0027801032074269362,
|
| 6236 |
+
"grad_norm": 11.5,
|
| 6237 |
+
"learning_rate": 1.8220000000000002e-05,
|
| 6238 |
+
"loss": 5.7255,
|
| 6239 |
+
"step": 8900
|
| 6240 |
+
},
|
| 6241 |
+
{
|
| 6242 |
+
"epoch": 0.0027832269188959556,
|
| 6243 |
+
"grad_norm": 14.625,
|
| 6244 |
+
"learning_rate": 1.8218e-05,
|
| 6245 |
+
"loss": 5.7292,
|
| 6246 |
+
"step": 8910
|
| 6247 |
+
},
|
| 6248 |
+
{
|
| 6249 |
+
"epoch": 0.0027863506303649746,
|
| 6250 |
+
"grad_norm": 17.625,
|
| 6251 |
+
"learning_rate": 1.8216000000000003e-05,
|
| 6252 |
+
"loss": 5.709,
|
| 6253 |
+
"step": 8920
|
| 6254 |
+
},
|
| 6255 |
+
{
|
| 6256 |
+
"epoch": 0.0027894743418339935,
|
| 6257 |
+
"grad_norm": 15.1875,
|
| 6258 |
+
"learning_rate": 1.8214000000000002e-05,
|
| 6259 |
+
"loss": 5.5746,
|
| 6260 |
+
"step": 8930
|
| 6261 |
+
},
|
| 6262 |
+
{
|
| 6263 |
+
"epoch": 0.0027925980533030125,
|
| 6264 |
+
"grad_norm": 14.5,
|
| 6265 |
+
"learning_rate": 1.8212e-05,
|
| 6266 |
+
"loss": 5.7584,
|
| 6267 |
+
"step": 8940
|
| 6268 |
+
},
|
| 6269 |
+
{
|
| 6270 |
+
"epoch": 0.0027957217647720314,
|
| 6271 |
+
"grad_norm": 12.3125,
|
| 6272 |
+
"learning_rate": 1.821e-05,
|
| 6273 |
+
"loss": 5.6617,
|
| 6274 |
+
"step": 8950
|
| 6275 |
+
},
|
| 6276 |
+
{
|
| 6277 |
+
"epoch": 0.0027988454762410504,
|
| 6278 |
+
"grad_norm": 18.5,
|
| 6279 |
+
"learning_rate": 1.8208000000000003e-05,
|
| 6280 |
+
"loss": 5.4972,
|
| 6281 |
+
"step": 8960
|
| 6282 |
+
},
|
| 6283 |
+
{
|
| 6284 |
+
"epoch": 0.00280196918771007,
|
| 6285 |
+
"grad_norm": 13.625,
|
| 6286 |
+
"learning_rate": 1.8206e-05,
|
| 6287 |
+
"loss": 5.7059,
|
| 6288 |
+
"step": 8970
|
| 6289 |
+
},
|
| 6290 |
+
{
|
| 6291 |
+
"epoch": 0.0028050928991790887,
|
| 6292 |
+
"grad_norm": 13.125,
|
| 6293 |
+
"learning_rate": 1.8204e-05,
|
| 6294 |
+
"loss": 5.6355,
|
| 6295 |
+
"step": 8980
|
| 6296 |
+
},
|
| 6297 |
+
{
|
| 6298 |
+
"epoch": 0.0028082166106481077,
|
| 6299 |
+
"grad_norm": 18.875,
|
| 6300 |
+
"learning_rate": 1.8202000000000003e-05,
|
| 6301 |
+
"loss": 5.3638,
|
| 6302 |
+
"step": 8990
|
| 6303 |
+
},
|
| 6304 |
+
{
|
| 6305 |
+
"epoch": 0.0028113403221171267,
|
| 6306 |
+
"grad_norm": 24.375,
|
| 6307 |
+
"learning_rate": 1.8200000000000002e-05,
|
| 6308 |
+
"loss": 5.4894,
|
| 6309 |
+
"step": 9000
|
| 6310 |
+
},
|
| 6311 |
+
{
|
| 6312 |
+
"epoch": 0.0028144640335861456,
|
| 6313 |
+
"grad_norm": 30.75,
|
| 6314 |
+
"learning_rate": 1.8198e-05,
|
| 6315 |
+
"loss": 5.4093,
|
| 6316 |
+
"step": 9010
|
| 6317 |
+
},
|
| 6318 |
+
{
|
| 6319 |
+
"epoch": 0.0028175877450551646,
|
| 6320 |
+
"grad_norm": 14.6875,
|
| 6321 |
+
"learning_rate": 1.8196000000000003e-05,
|
| 6322 |
+
"loss": 5.5249,
|
| 6323 |
+
"step": 9020
|
| 6324 |
+
},
|
| 6325 |
+
{
|
| 6326 |
+
"epoch": 0.002820711456524184,
|
| 6327 |
+
"grad_norm": 14.5625,
|
| 6328 |
+
"learning_rate": 1.8194e-05,
|
| 6329 |
+
"loss": 5.5573,
|
| 6330 |
+
"step": 9030
|
| 6331 |
+
},
|
| 6332 |
+
{
|
| 6333 |
+
"epoch": 0.002823835167993203,
|
| 6334 |
+
"grad_norm": 14.8125,
|
| 6335 |
+
"learning_rate": 1.8192e-05,
|
| 6336 |
+
"loss": 5.7256,
|
| 6337 |
+
"step": 9040
|
| 6338 |
+
},
|
| 6339 |
+
{
|
| 6340 |
+
"epoch": 0.002826958879462222,
|
| 6341 |
+
"grad_norm": 13.1875,
|
| 6342 |
+
"learning_rate": 1.819e-05,
|
| 6343 |
+
"loss": 5.5939,
|
| 6344 |
+
"step": 9050
|
| 6345 |
+
},
|
| 6346 |
+
{
|
| 6347 |
+
"epoch": 0.002830082590931241,
|
| 6348 |
+
"grad_norm": 16.75,
|
| 6349 |
+
"learning_rate": 1.8188000000000003e-05,
|
| 6350 |
+
"loss": 5.6293,
|
| 6351 |
+
"step": 9060
|
| 6352 |
+
},
|
| 6353 |
+
{
|
| 6354 |
+
"epoch": 0.00283320630240026,
|
| 6355 |
+
"grad_norm": 16.75,
|
| 6356 |
+
"learning_rate": 1.8186000000000002e-05,
|
| 6357 |
+
"loss": 5.7227,
|
| 6358 |
+
"step": 9070
|
| 6359 |
+
},
|
| 6360 |
+
{
|
| 6361 |
+
"epoch": 0.0028363300138692787,
|
| 6362 |
+
"grad_norm": 14.6875,
|
| 6363 |
+
"learning_rate": 1.8184e-05,
|
| 6364 |
+
"loss": 5.6747,
|
| 6365 |
+
"step": 9080
|
| 6366 |
+
},
|
| 6367 |
+
{
|
| 6368 |
+
"epoch": 0.002839453725338298,
|
| 6369 |
+
"grad_norm": 17.875,
|
| 6370 |
+
"learning_rate": 1.8182000000000003e-05,
|
| 6371 |
+
"loss": 5.7045,
|
| 6372 |
+
"step": 9090
|
| 6373 |
+
},
|
| 6374 |
+
{
|
| 6375 |
+
"epoch": 0.002842577436807317,
|
| 6376 |
+
"grad_norm": 15.125,
|
| 6377 |
+
"learning_rate": 1.8180000000000002e-05,
|
| 6378 |
+
"loss": 5.7102,
|
| 6379 |
+
"step": 9100
|
| 6380 |
+
},
|
| 6381 |
+
{
|
| 6382 |
+
"epoch": 0.002845701148276336,
|
| 6383 |
+
"grad_norm": 11.8125,
|
| 6384 |
+
"learning_rate": 1.8178e-05,
|
| 6385 |
+
"loss": 5.5916,
|
| 6386 |
+
"step": 9110
|
| 6387 |
+
},
|
| 6388 |
+
{
|
| 6389 |
+
"epoch": 0.002848824859745355,
|
| 6390 |
+
"grad_norm": 16.625,
|
| 6391 |
+
"learning_rate": 1.8176000000000004e-05,
|
| 6392 |
+
"loss": 5.6392,
|
| 6393 |
+
"step": 9120
|
| 6394 |
+
},
|
| 6395 |
+
{
|
| 6396 |
+
"epoch": 0.002851948571214374,
|
| 6397 |
+
"grad_norm": 17.25,
|
| 6398 |
+
"learning_rate": 1.8174e-05,
|
| 6399 |
+
"loss": 5.4425,
|
| 6400 |
+
"step": 9130
|
| 6401 |
+
},
|
| 6402 |
+
{
|
| 6403 |
+
"epoch": 0.002855072282683393,
|
| 6404 |
+
"grad_norm": 13.8125,
|
| 6405 |
+
"learning_rate": 1.8172e-05,
|
| 6406 |
+
"loss": 5.6837,
|
| 6407 |
+
"step": 9140
|
| 6408 |
+
},
|
| 6409 |
+
{
|
| 6410 |
+
"epoch": 0.0028581959941524123,
|
| 6411 |
+
"grad_norm": 14.375,
|
| 6412 |
+
"learning_rate": 1.817e-05,
|
| 6413 |
+
"loss": 5.6339,
|
| 6414 |
+
"step": 9150
|
| 6415 |
+
},
|
| 6416 |
+
{
|
| 6417 |
+
"epoch": 0.0028613197056214313,
|
| 6418 |
+
"grad_norm": 22.75,
|
| 6419 |
+
"learning_rate": 1.8168e-05,
|
| 6420 |
+
"loss": 5.3867,
|
| 6421 |
+
"step": 9160
|
| 6422 |
+
},
|
| 6423 |
+
{
|
| 6424 |
+
"epoch": 0.0028644434170904502,
|
| 6425 |
+
"grad_norm": 13.75,
|
| 6426 |
+
"learning_rate": 1.8166000000000002e-05,
|
| 6427 |
+
"loss": 5.5268,
|
| 6428 |
+
"step": 9170
|
| 6429 |
+
},
|
| 6430 |
+
{
|
| 6431 |
+
"epoch": 0.002867567128559469,
|
| 6432 |
+
"grad_norm": 29.375,
|
| 6433 |
+
"learning_rate": 1.8164e-05,
|
| 6434 |
+
"loss": 5.5336,
|
| 6435 |
+
"step": 9180
|
| 6436 |
+
},
|
| 6437 |
+
{
|
| 6438 |
+
"epoch": 0.002870690840028488,
|
| 6439 |
+
"grad_norm": 18.125,
|
| 6440 |
+
"learning_rate": 1.8162000000000003e-05,
|
| 6441 |
+
"loss": 5.3273,
|
| 6442 |
+
"step": 9190
|
| 6443 |
+
},
|
| 6444 |
+
{
|
| 6445 |
+
"epoch": 0.002873814551497507,
|
| 6446 |
+
"grad_norm": 13.75,
|
| 6447 |
+
"learning_rate": 1.8160000000000002e-05,
|
| 6448 |
+
"loss": 5.7044,
|
| 6449 |
+
"step": 9200
|
| 6450 |
+
},
|
| 6451 |
+
{
|
| 6452 |
+
"epoch": 0.0028769382629665265,
|
| 6453 |
+
"grad_norm": 16.5,
|
| 6454 |
+
"learning_rate": 1.8158e-05,
|
| 6455 |
+
"loss": 5.5071,
|
| 6456 |
+
"step": 9210
|
| 6457 |
+
},
|
| 6458 |
+
{
|
| 6459 |
+
"epoch": 0.0028800619744355454,
|
| 6460 |
+
"grad_norm": 21.625,
|
| 6461 |
+
"learning_rate": 1.8156000000000004e-05,
|
| 6462 |
+
"loss": 5.5489,
|
| 6463 |
+
"step": 9220
|
| 6464 |
+
},
|
| 6465 |
+
{
|
| 6466 |
+
"epoch": 0.0028831856859045644,
|
| 6467 |
+
"grad_norm": 18.125,
|
| 6468 |
+
"learning_rate": 1.8154e-05,
|
| 6469 |
+
"loss": 5.7425,
|
| 6470 |
+
"step": 9230
|
| 6471 |
+
},
|
| 6472 |
+
{
|
| 6473 |
+
"epoch": 0.0028863093973735834,
|
| 6474 |
+
"grad_norm": 19.625,
|
| 6475 |
+
"learning_rate": 1.8152000000000002e-05,
|
| 6476 |
+
"loss": 5.4189,
|
| 6477 |
+
"step": 9240
|
| 6478 |
+
},
|
| 6479 |
+
{
|
| 6480 |
+
"epoch": 0.0028894331088426023,
|
| 6481 |
+
"grad_norm": 15.6875,
|
| 6482 |
+
"learning_rate": 1.815e-05,
|
| 6483 |
+
"loss": 5.607,
|
| 6484 |
+
"step": 9250
|
| 6485 |
+
},
|
| 6486 |
+
{
|
| 6487 |
+
"epoch": 0.0028925568203116213,
|
| 6488 |
+
"grad_norm": 14.0625,
|
| 6489 |
+
"learning_rate": 1.8148e-05,
|
| 6490 |
+
"loss": 5.441,
|
| 6491 |
+
"step": 9260
|
| 6492 |
+
},
|
| 6493 |
+
{
|
| 6494 |
+
"epoch": 0.0028956805317806407,
|
| 6495 |
+
"grad_norm": 16.375,
|
| 6496 |
+
"learning_rate": 1.8146000000000002e-05,
|
| 6497 |
+
"loss": 5.7174,
|
| 6498 |
+
"step": 9270
|
| 6499 |
+
},
|
| 6500 |
+
{
|
| 6501 |
+
"epoch": 0.0028988042432496596,
|
| 6502 |
+
"grad_norm": 14.0,
|
| 6503 |
+
"learning_rate": 1.8144e-05,
|
| 6504 |
+
"loss": 5.736,
|
| 6505 |
+
"step": 9280
|
| 6506 |
+
},
|
| 6507 |
+
{
|
| 6508 |
+
"epoch": 0.0029019279547186786,
|
| 6509 |
+
"grad_norm": 16.0,
|
| 6510 |
+
"learning_rate": 1.8142000000000004e-05,
|
| 6511 |
+
"loss": 5.5643,
|
| 6512 |
+
"step": 9290
|
| 6513 |
+
},
|
| 6514 |
+
{
|
| 6515 |
+
"epoch": 0.0029050516661876975,
|
| 6516 |
+
"grad_norm": 14.875,
|
| 6517 |
+
"learning_rate": 1.8140000000000003e-05,
|
| 6518 |
+
"loss": 5.6589,
|
| 6519 |
+
"step": 9300
|
| 6520 |
+
},
|
| 6521 |
+
{
|
| 6522 |
+
"epoch": 0.0029081753776567165,
|
| 6523 |
+
"grad_norm": 16.25,
|
| 6524 |
+
"learning_rate": 1.8138e-05,
|
| 6525 |
+
"loss": 5.5534,
|
| 6526 |
+
"step": 9310
|
| 6527 |
+
},
|
| 6528 |
+
{
|
| 6529 |
+
"epoch": 0.0029112990891257354,
|
| 6530 |
+
"grad_norm": 15.375,
|
| 6531 |
+
"learning_rate": 1.8136000000000004e-05,
|
| 6532 |
+
"loss": 5.5841,
|
| 6533 |
+
"step": 9320
|
| 6534 |
+
},
|
| 6535 |
+
{
|
| 6536 |
+
"epoch": 0.002914422800594755,
|
| 6537 |
+
"grad_norm": 14.875,
|
| 6538 |
+
"learning_rate": 1.8134e-05,
|
| 6539 |
+
"loss": 5.6218,
|
| 6540 |
+
"step": 9330
|
| 6541 |
+
},
|
| 6542 |
+
{
|
| 6543 |
+
"epoch": 0.002917546512063774,
|
| 6544 |
+
"grad_norm": 12.8125,
|
| 6545 |
+
"learning_rate": 1.8132000000000002e-05,
|
| 6546 |
+
"loss": 5.5351,
|
| 6547 |
+
"step": 9340
|
| 6548 |
+
},
|
| 6549 |
+
{
|
| 6550 |
+
"epoch": 0.0029206702235327927,
|
| 6551 |
+
"grad_norm": 14.3125,
|
| 6552 |
+
"learning_rate": 1.813e-05,
|
| 6553 |
+
"loss": 5.5595,
|
| 6554 |
+
"step": 9350
|
| 6555 |
+
},
|
| 6556 |
+
{
|
| 6557 |
+
"epoch": 0.0029237939350018117,
|
| 6558 |
+
"grad_norm": 12.375,
|
| 6559 |
+
"learning_rate": 1.8128e-05,
|
| 6560 |
+
"loss": 5.6869,
|
| 6561 |
+
"step": 9360
|
| 6562 |
+
},
|
| 6563 |
+
{
|
| 6564 |
+
"epoch": 0.0029269176464708307,
|
| 6565 |
+
"grad_norm": 18.125,
|
| 6566 |
+
"learning_rate": 1.8126000000000002e-05,
|
| 6567 |
+
"loss": 5.4235,
|
| 6568 |
+
"step": 9370
|
| 6569 |
+
},
|
| 6570 |
+
{
|
| 6571 |
+
"epoch": 0.0029300413579398496,
|
| 6572 |
+
"grad_norm": 18.125,
|
| 6573 |
+
"learning_rate": 1.8124e-05,
|
| 6574 |
+
"loss": 5.583,
|
| 6575 |
+
"step": 9380
|
| 6576 |
+
},
|
| 6577 |
+
{
|
| 6578 |
+
"epoch": 0.002933165069408869,
|
| 6579 |
+
"grad_norm": 13.375,
|
| 6580 |
+
"learning_rate": 1.8122e-05,
|
| 6581 |
+
"loss": 5.6377,
|
| 6582 |
+
"step": 9390
|
| 6583 |
+
},
|
| 6584 |
+
{
|
| 6585 |
+
"epoch": 0.002936288780877888,
|
| 6586 |
+
"grad_norm": 14.9375,
|
| 6587 |
+
"learning_rate": 1.8120000000000003e-05,
|
| 6588 |
+
"loss": 5.5843,
|
| 6589 |
+
"step": 9400
|
| 6590 |
+
},
|
| 6591 |
+
{
|
| 6592 |
+
"epoch": 0.002939412492346907,
|
| 6593 |
+
"grad_norm": 18.875,
|
| 6594 |
+
"learning_rate": 1.8118000000000002e-05,
|
| 6595 |
+
"loss": 5.4062,
|
| 6596 |
+
"step": 9410
|
| 6597 |
+
},
|
| 6598 |
+
{
|
| 6599 |
+
"epoch": 0.002942536203815926,
|
| 6600 |
+
"grad_norm": 13.375,
|
| 6601 |
+
"learning_rate": 1.8116000000000004e-05,
|
| 6602 |
+
"loss": 5.6681,
|
| 6603 |
+
"step": 9420
|
| 6604 |
+
},
|
| 6605 |
+
{
|
| 6606 |
+
"epoch": 0.002945659915284945,
|
| 6607 |
+
"grad_norm": 11.6875,
|
| 6608 |
+
"learning_rate": 1.8114e-05,
|
| 6609 |
+
"loss": 5.5529,
|
| 6610 |
+
"step": 9430
|
| 6611 |
+
},
|
| 6612 |
+
{
|
| 6613 |
+
"epoch": 0.002948783626753964,
|
| 6614 |
+
"grad_norm": 20.875,
|
| 6615 |
+
"learning_rate": 1.8112000000000002e-05,
|
| 6616 |
+
"loss": 5.6737,
|
| 6617 |
+
"step": 9440
|
| 6618 |
+
},
|
| 6619 |
+
{
|
| 6620 |
+
"epoch": 0.002951907338222983,
|
| 6621 |
+
"grad_norm": 13.5625,
|
| 6622 |
+
"learning_rate": 1.811e-05,
|
| 6623 |
+
"loss": 5.6382,
|
| 6624 |
+
"step": 9450
|
| 6625 |
+
},
|
| 6626 |
+
{
|
| 6627 |
+
"epoch": 0.002955031049692002,
|
| 6628 |
+
"grad_norm": 17.75,
|
| 6629 |
+
"learning_rate": 1.8108e-05,
|
| 6630 |
+
"loss": 5.5691,
|
| 6631 |
+
"step": 9460
|
| 6632 |
+
},
|
| 6633 |
+
{
|
| 6634 |
+
"epoch": 0.002958154761161021,
|
| 6635 |
+
"grad_norm": 13.8125,
|
| 6636 |
+
"learning_rate": 1.8106000000000002e-05,
|
| 6637 |
+
"loss": 5.5743,
|
| 6638 |
+
"step": 9470
|
| 6639 |
+
},
|
| 6640 |
+
{
|
| 6641 |
+
"epoch": 0.00296127847263004,
|
| 6642 |
+
"grad_norm": 14.3125,
|
| 6643 |
+
"learning_rate": 1.8104e-05,
|
| 6644 |
+
"loss": 5.3204,
|
| 6645 |
+
"step": 9480
|
| 6646 |
+
},
|
| 6647 |
+
{
|
| 6648 |
+
"epoch": 0.002964402184099059,
|
| 6649 |
+
"grad_norm": 14.625,
|
| 6650 |
+
"learning_rate": 1.8102e-05,
|
| 6651 |
+
"loss": 5.6013,
|
| 6652 |
+
"step": 9490
|
| 6653 |
+
},
|
| 6654 |
+
{
|
| 6655 |
+
"epoch": 0.002967525895568078,
|
| 6656 |
+
"grad_norm": 14.5625,
|
| 6657 |
+
"learning_rate": 1.8100000000000003e-05,
|
| 6658 |
+
"loss": 5.619,
|
| 6659 |
+
"step": 9500
|
| 6660 |
+
},
|
| 6661 |
+
{
|
| 6662 |
+
"epoch": 0.0029706496070370974,
|
| 6663 |
+
"grad_norm": 20.75,
|
| 6664 |
+
"learning_rate": 1.8098000000000002e-05,
|
| 6665 |
+
"loss": 5.2921,
|
| 6666 |
+
"step": 9510
|
| 6667 |
+
},
|
| 6668 |
+
{
|
| 6669 |
+
"epoch": 0.0029737733185061163,
|
| 6670 |
+
"grad_norm": 15.6875,
|
| 6671 |
+
"learning_rate": 1.8096e-05,
|
| 6672 |
+
"loss": 5.5717,
|
| 6673 |
+
"step": 9520
|
| 6674 |
+
},
|
| 6675 |
+
{
|
| 6676 |
+
"epoch": 0.0029768970299751353,
|
| 6677 |
+
"grad_norm": 12.0,
|
| 6678 |
+
"learning_rate": 1.8094e-05,
|
| 6679 |
+
"loss": 5.2209,
|
| 6680 |
+
"step": 9530
|
| 6681 |
+
},
|
| 6682 |
+
{
|
| 6683 |
+
"epoch": 0.0029800207414441542,
|
| 6684 |
+
"grad_norm": 16.5,
|
| 6685 |
+
"learning_rate": 1.8092000000000002e-05,
|
| 6686 |
+
"loss": 5.6648,
|
| 6687 |
+
"step": 9540
|
| 6688 |
+
},
|
| 6689 |
+
{
|
| 6690 |
+
"epoch": 0.002983144452913173,
|
| 6691 |
+
"grad_norm": 14.125,
|
| 6692 |
+
"learning_rate": 1.809e-05,
|
| 6693 |
+
"loss": 5.5017,
|
| 6694 |
+
"step": 9550
|
| 6695 |
+
},
|
| 6696 |
+
{
|
| 6697 |
+
"epoch": 0.002986268164382192,
|
| 6698 |
+
"grad_norm": 14.625,
|
| 6699 |
+
"learning_rate": 1.8088e-05,
|
| 6700 |
+
"loss": 5.6437,
|
| 6701 |
+
"step": 9560
|
| 6702 |
+
},
|
| 6703 |
+
{
|
| 6704 |
+
"epoch": 0.0029893918758512115,
|
| 6705 |
+
"grad_norm": 12.875,
|
| 6706 |
+
"learning_rate": 1.8086000000000003e-05,
|
| 6707 |
+
"loss": 5.4502,
|
| 6708 |
+
"step": 9570
|
| 6709 |
+
},
|
| 6710 |
+
{
|
| 6711 |
+
"epoch": 0.0029925155873202305,
|
| 6712 |
+
"grad_norm": 17.75,
|
| 6713 |
+
"learning_rate": 1.8084e-05,
|
| 6714 |
+
"loss": 5.4951,
|
| 6715 |
+
"step": 9580
|
| 6716 |
+
},
|
| 6717 |
+
{
|
| 6718 |
+
"epoch": 0.0029956392987892494,
|
| 6719 |
+
"grad_norm": 14.5625,
|
| 6720 |
+
"learning_rate": 1.8082e-05,
|
| 6721 |
+
"loss": 5.5379,
|
| 6722 |
+
"step": 9590
|
| 6723 |
+
},
|
| 6724 |
+
{
|
| 6725 |
+
"epoch": 0.0029987630102582684,
|
| 6726 |
+
"grad_norm": 12.0625,
|
| 6727 |
+
"learning_rate": 1.8080000000000003e-05,
|
| 6728 |
+
"loss": 5.3454,
|
| 6729 |
+
"step": 9600
|
| 6730 |
+
},
|
| 6731 |
+
{
|
| 6732 |
+
"epoch": 0.0030018867217272874,
|
| 6733 |
+
"grad_norm": 12.25,
|
| 6734 |
+
"learning_rate": 1.8078000000000002e-05,
|
| 6735 |
+
"loss": 5.4042,
|
| 6736 |
+
"step": 9610
|
| 6737 |
+
},
|
| 6738 |
+
{
|
| 6739 |
+
"epoch": 0.0030050104331963063,
|
| 6740 |
+
"grad_norm": 21.25,
|
| 6741 |
+
"learning_rate": 1.8076e-05,
|
| 6742 |
+
"loss": 5.1163,
|
| 6743 |
+
"step": 9620
|
| 6744 |
+
},
|
| 6745 |
+
{
|
| 6746 |
+
"epoch": 0.0030081341446653257,
|
| 6747 |
+
"grad_norm": 13.4375,
|
| 6748 |
+
"learning_rate": 1.8074e-05,
|
| 6749 |
+
"loss": 5.622,
|
| 6750 |
+
"step": 9630
|
| 6751 |
+
},
|
| 6752 |
+
{
|
| 6753 |
+
"epoch": 0.0030112578561343447,
|
| 6754 |
+
"grad_norm": 17.5,
|
| 6755 |
+
"learning_rate": 1.8072e-05,
|
| 6756 |
+
"loss": 5.6973,
|
| 6757 |
+
"step": 9640
|
| 6758 |
+
},
|
| 6759 |
+
{
|
| 6760 |
+
"epoch": 0.0030143815676033636,
|
| 6761 |
+
"grad_norm": 18.5,
|
| 6762 |
+
"learning_rate": 1.807e-05,
|
| 6763 |
+
"loss": 5.6359,
|
| 6764 |
+
"step": 9650
|
| 6765 |
+
},
|
| 6766 |
+
{
|
| 6767 |
+
"epoch": 0.0030175052790723826,
|
| 6768 |
+
"grad_norm": 17.625,
|
| 6769 |
+
"learning_rate": 1.8068e-05,
|
| 6770 |
+
"loss": 5.3276,
|
| 6771 |
+
"step": 9660
|
| 6772 |
+
},
|
| 6773 |
+
{
|
| 6774 |
+
"epoch": 0.0030206289905414015,
|
| 6775 |
+
"grad_norm": 14.5625,
|
| 6776 |
+
"learning_rate": 1.8066000000000003e-05,
|
| 6777 |
+
"loss": 5.3333,
|
| 6778 |
+
"step": 9670
|
| 6779 |
+
},
|
| 6780 |
+
{
|
| 6781 |
+
"epoch": 0.0030237527020104205,
|
| 6782 |
+
"grad_norm": 13.375,
|
| 6783 |
+
"learning_rate": 1.8064000000000002e-05,
|
| 6784 |
+
"loss": 5.6219,
|
| 6785 |
+
"step": 9680
|
| 6786 |
+
},
|
| 6787 |
+
{
|
| 6788 |
+
"epoch": 0.00302687641347944,
|
| 6789 |
+
"grad_norm": 12.8125,
|
| 6790 |
+
"learning_rate": 1.8062e-05,
|
| 6791 |
+
"loss": 5.6096,
|
| 6792 |
+
"step": 9690
|
| 6793 |
+
},
|
| 6794 |
+
{
|
| 6795 |
+
"epoch": 0.003030000124948459,
|
| 6796 |
+
"grad_norm": 17.5,
|
| 6797 |
+
"learning_rate": 1.8060000000000003e-05,
|
| 6798 |
+
"loss": 5.4421,
|
| 6799 |
+
"step": 9700
|
| 6800 |
+
},
|
| 6801 |
+
{
|
| 6802 |
+
"epoch": 0.003033123836417478,
|
| 6803 |
+
"grad_norm": 18.25,
|
| 6804 |
+
"learning_rate": 1.8058000000000002e-05,
|
| 6805 |
+
"loss": 5.6122,
|
| 6806 |
+
"step": 9710
|
| 6807 |
+
},
|
| 6808 |
+
{
|
| 6809 |
+
"epoch": 0.0030362475478864968,
|
| 6810 |
+
"grad_norm": 14.1875,
|
| 6811 |
+
"learning_rate": 1.8056e-05,
|
| 6812 |
+
"loss": 5.5429,
|
| 6813 |
+
"step": 9720
|
| 6814 |
+
},
|
| 6815 |
+
{
|
| 6816 |
+
"epoch": 0.0030393712593555157,
|
| 6817 |
+
"grad_norm": 16.875,
|
| 6818 |
+
"learning_rate": 1.8054e-05,
|
| 6819 |
+
"loss": 5.7268,
|
| 6820 |
+
"step": 9730
|
| 6821 |
+
},
|
| 6822 |
+
{
|
| 6823 |
+
"epoch": 0.0030424949708245347,
|
| 6824 |
+
"grad_norm": 13.8125,
|
| 6825 |
+
"learning_rate": 1.8052e-05,
|
| 6826 |
+
"loss": 5.5956,
|
| 6827 |
+
"step": 9740
|
| 6828 |
+
},
|
| 6829 |
+
{
|
| 6830 |
+
"epoch": 0.003045618682293554,
|
| 6831 |
+
"grad_norm": 11.1875,
|
| 6832 |
+
"learning_rate": 1.805e-05,
|
| 6833 |
+
"loss": 5.8065,
|
| 6834 |
+
"step": 9750
|
| 6835 |
+
},
|
| 6836 |
+
{
|
| 6837 |
+
"epoch": 0.003048742393762573,
|
| 6838 |
+
"grad_norm": 12.875,
|
| 6839 |
+
"learning_rate": 1.8048e-05,
|
| 6840 |
+
"loss": 5.445,
|
| 6841 |
+
"step": 9760
|
| 6842 |
+
},
|
| 6843 |
+
{
|
| 6844 |
+
"epoch": 0.003051866105231592,
|
| 6845 |
+
"grad_norm": 18.0,
|
| 6846 |
+
"learning_rate": 1.8046e-05,
|
| 6847 |
+
"loss": 5.5073,
|
| 6848 |
+
"step": 9770
|
| 6849 |
+
},
|
| 6850 |
+
{
|
| 6851 |
+
"epoch": 0.003054989816700611,
|
| 6852 |
+
"grad_norm": 12.25,
|
| 6853 |
+
"learning_rate": 1.8044000000000002e-05,
|
| 6854 |
+
"loss": 5.6303,
|
| 6855 |
+
"step": 9780
|
| 6856 |
+
},
|
| 6857 |
+
{
|
| 6858 |
+
"epoch": 0.00305811352816963,
|
| 6859 |
+
"grad_norm": 12.375,
|
| 6860 |
+
"learning_rate": 1.8042e-05,
|
| 6861 |
+
"loss": 5.4904,
|
| 6862 |
+
"step": 9790
|
| 6863 |
+
},
|
| 6864 |
+
{
|
| 6865 |
+
"epoch": 0.003061237239638649,
|
| 6866 |
+
"grad_norm": 14.375,
|
| 6867 |
+
"learning_rate": 1.8040000000000003e-05,
|
| 6868 |
+
"loss": 5.455,
|
| 6869 |
+
"step": 9800
|
| 6870 |
+
},
|
| 6871 |
+
{
|
| 6872 |
+
"epoch": 0.0030643609511076682,
|
| 6873 |
+
"grad_norm": 14.1875,
|
| 6874 |
+
"learning_rate": 1.8038000000000002e-05,
|
| 6875 |
+
"loss": 5.5094,
|
| 6876 |
+
"step": 9810
|
| 6877 |
+
},
|
| 6878 |
+
{
|
| 6879 |
+
"epoch": 0.003067484662576687,
|
| 6880 |
+
"grad_norm": 18.25,
|
| 6881 |
+
"learning_rate": 1.8036e-05,
|
| 6882 |
+
"loss": 5.7042,
|
| 6883 |
+
"step": 9820
|
| 6884 |
+
},
|
| 6885 |
+
{
|
| 6886 |
+
"epoch": 0.003070608374045706,
|
| 6887 |
+
"grad_norm": 15.25,
|
| 6888 |
+
"learning_rate": 1.8034e-05,
|
| 6889 |
+
"loss": 5.6813,
|
| 6890 |
+
"step": 9830
|
| 6891 |
+
},
|
| 6892 |
+
{
|
| 6893 |
+
"epoch": 0.003073732085514725,
|
| 6894 |
+
"grad_norm": 11.75,
|
| 6895 |
+
"learning_rate": 1.8032e-05,
|
| 6896 |
+
"loss": 5.5455,
|
| 6897 |
+
"step": 9840
|
| 6898 |
+
},
|
| 6899 |
+
{
|
| 6900 |
+
"epoch": 0.003076855796983744,
|
| 6901 |
+
"grad_norm": 12.6875,
|
| 6902 |
+
"learning_rate": 1.8030000000000002e-05,
|
| 6903 |
+
"loss": 5.7319,
|
| 6904 |
+
"step": 9850
|
| 6905 |
+
},
|
| 6906 |
+
{
|
| 6907 |
+
"epoch": 0.0030799795084527634,
|
| 6908 |
+
"grad_norm": 11.4375,
|
| 6909 |
+
"learning_rate": 1.8028e-05,
|
| 6910 |
+
"loss": 5.7984,
|
| 6911 |
+
"step": 9860
|
| 6912 |
+
},
|
| 6913 |
+
{
|
| 6914 |
+
"epoch": 0.0030831032199217824,
|
| 6915 |
+
"grad_norm": 12.0625,
|
| 6916 |
+
"learning_rate": 1.8026e-05,
|
| 6917 |
+
"loss": 5.6482,
|
| 6918 |
+
"step": 9870
|
| 6919 |
+
},
|
| 6920 |
+
{
|
| 6921 |
+
"epoch": 0.0030862269313908014,
|
| 6922 |
+
"grad_norm": 12.75,
|
| 6923 |
+
"learning_rate": 1.8024000000000002e-05,
|
| 6924 |
+
"loss": 5.8138,
|
| 6925 |
+
"step": 9880
|
| 6926 |
+
},
|
| 6927 |
+
{
|
| 6928 |
+
"epoch": 0.0030893506428598203,
|
| 6929 |
+
"grad_norm": 15.1875,
|
| 6930 |
+
"learning_rate": 1.8022e-05,
|
| 6931 |
+
"loss": 5.542,
|
| 6932 |
+
"step": 9890
|
| 6933 |
+
},
|
| 6934 |
+
{
|
| 6935 |
+
"epoch": 0.0030924743543288393,
|
| 6936 |
+
"grad_norm": 15.4375,
|
| 6937 |
+
"learning_rate": 1.802e-05,
|
| 6938 |
+
"loss": 5.3519,
|
| 6939 |
+
"step": 9900
|
| 6940 |
+
},
|
| 6941 |
+
{
|
| 6942 |
+
"epoch": 0.0030955980657978582,
|
| 6943 |
+
"grad_norm": 12.875,
|
| 6944 |
+
"learning_rate": 1.8018000000000003e-05,
|
| 6945 |
+
"loss": 5.6849,
|
| 6946 |
+
"step": 9910
|
| 6947 |
+
},
|
| 6948 |
+
{
|
| 6949 |
+
"epoch": 0.0030987217772668776,
|
| 6950 |
+
"grad_norm": 16.25,
|
| 6951 |
+
"learning_rate": 1.8016e-05,
|
| 6952 |
+
"loss": 5.5522,
|
| 6953 |
+
"step": 9920
|
| 6954 |
+
},
|
| 6955 |
+
{
|
| 6956 |
+
"epoch": 0.0031018454887358966,
|
| 6957 |
+
"grad_norm": 17.875,
|
| 6958 |
+
"learning_rate": 1.8014e-05,
|
| 6959 |
+
"loss": 5.3861,
|
| 6960 |
+
"step": 9930
|
| 6961 |
+
},
|
| 6962 |
+
{
|
| 6963 |
+
"epoch": 0.0031049692002049155,
|
| 6964 |
+
"grad_norm": 18.125,
|
| 6965 |
+
"learning_rate": 1.8012e-05,
|
| 6966 |
+
"loss": 5.7198,
|
| 6967 |
+
"step": 9940
|
| 6968 |
+
},
|
| 6969 |
+
{
|
| 6970 |
+
"epoch": 0.0031080929116739345,
|
| 6971 |
+
"grad_norm": 18.375,
|
| 6972 |
+
"learning_rate": 1.8010000000000002e-05,
|
| 6973 |
+
"loss": 5.722,
|
| 6974 |
+
"step": 9950
|
| 6975 |
+
},
|
| 6976 |
+
{
|
| 6977 |
+
"epoch": 0.0031112166231429535,
|
| 6978 |
+
"grad_norm": 9.5625,
|
| 6979 |
+
"learning_rate": 1.8008e-05,
|
| 6980 |
+
"loss": 5.4564,
|
| 6981 |
+
"step": 9960
|
| 6982 |
+
},
|
| 6983 |
+
{
|
| 6984 |
+
"epoch": 0.0031143403346119724,
|
| 6985 |
+
"grad_norm": 12.75,
|
| 6986 |
+
"learning_rate": 1.8006e-05,
|
| 6987 |
+
"loss": 5.6358,
|
| 6988 |
+
"step": 9970
|
| 6989 |
+
},
|
| 6990 |
+
{
|
| 6991 |
+
"epoch": 0.003117464046080992,
|
| 6992 |
+
"grad_norm": 13.0,
|
| 6993 |
+
"learning_rate": 1.8004000000000002e-05,
|
| 6994 |
+
"loss": 5.6605,
|
| 6995 |
+
"step": 9980
|
| 6996 |
+
},
|
| 6997 |
+
{
|
| 6998 |
+
"epoch": 0.0031205877575500108,
|
| 6999 |
+
"grad_norm": 14.375,
|
| 7000 |
+
"learning_rate": 1.8002e-05,
|
| 7001 |
+
"loss": 5.6693,
|
| 7002 |
+
"step": 9990
|
| 7003 |
+
},
|
| 7004 |
+
{
|
| 7005 |
+
"epoch": 0.0031237114690190297,
|
| 7006 |
+
"grad_norm": 17.875,
|
| 7007 |
+
"learning_rate": 1.8e-05,
|
| 7008 |
+
"loss": 5.1117,
|
| 7009 |
+
"step": 10000
|
| 7010 |
}
|
| 7011 |
],
|
| 7012 |
"logging_steps": 10,
|
|
|
|
| 7014 |
"num_input_tokens_seen": 0,
|
| 7015 |
"num_train_epochs": 1,
|
| 7016 |
"save_steps": 1000,
|
| 7017 |
+
"total_flos": 1.567748744282112e+18,
|
| 7018 |
"train_batch_size": 4,
|
| 7019 |
"trial_name": null,
|
| 7020 |
"trial_params": null
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4984
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f1c12f37866e3e4956ef1c53aa3d780282f4658da9f00127876e8020ddc68ee1
|
| 3 |
size 4984
|