Upload folder using huggingface_hub
Browse files- adapter_model.safetensors +1 -1
- trainer_state.json +1402 -2
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 167829552
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:45a400a6169db337b1e03de36a0fe40f8682afd90b03b3162e53bd9bfd97a571
|
| 3 |
size 167829552
|
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,
|
|
@@ -4207,6 +4207,1406 @@
|
|
| 4207 |
"learning_rate": 1.9487364483851983e-05,
|
| 4208 |
"loss": 0.0079,
|
| 4209 |
"step": 30000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 4210 |
}
|
| 4211 |
],
|
| 4212 |
"logging_steps": 50,
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.03451197479245361,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 40000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 4207 |
"learning_rate": 1.9487364483851983e-05,
|
| 4208 |
"loss": 0.0079,
|
| 4209 |
"step": 30000
|
| 4210 |
+
},
|
| 4211 |
+
{
|
| 4212 |
+
"epoch": 0.025927121062830776,
|
| 4213 |
+
"grad_norm": 0.026366397738456726,
|
| 4214 |
+
"learning_rate": 1.9486501461097523e-05,
|
| 4215 |
+
"loss": 0.0381,
|
| 4216 |
+
"step": 30050
|
| 4217 |
+
},
|
| 4218 |
+
{
|
| 4219 |
+
"epoch": 0.025970261031321343,
|
| 4220 |
+
"grad_norm": 5.6284894943237305,
|
| 4221 |
+
"learning_rate": 1.9485638438343066e-05,
|
| 4222 |
+
"loss": 0.0618,
|
| 4223 |
+
"step": 30100
|
| 4224 |
+
},
|
| 4225 |
+
{
|
| 4226 |
+
"epoch": 0.02601340099981191,
|
| 4227 |
+
"grad_norm": 0.00824633240699768,
|
| 4228 |
+
"learning_rate": 1.948477541558861e-05,
|
| 4229 |
+
"loss": 0.0173,
|
| 4230 |
+
"step": 30150
|
| 4231 |
+
},
|
| 4232 |
+
{
|
| 4233 |
+
"epoch": 0.026056540968302476,
|
| 4234 |
+
"grad_norm": 0.0007174229249358177,
|
| 4235 |
+
"learning_rate": 1.948391239283415e-05,
|
| 4236 |
+
"loss": 0.0199,
|
| 4237 |
+
"step": 30200
|
| 4238 |
+
},
|
| 4239 |
+
{
|
| 4240 |
+
"epoch": 0.026099680936793043,
|
| 4241 |
+
"grad_norm": 0.02488381415605545,
|
| 4242 |
+
"learning_rate": 1.9483049370079693e-05,
|
| 4243 |
+
"loss": 0.0312,
|
| 4244 |
+
"step": 30250
|
| 4245 |
+
},
|
| 4246 |
+
{
|
| 4247 |
+
"epoch": 0.02614282090528361,
|
| 4248 |
+
"grad_norm": 1.9344063997268677,
|
| 4249 |
+
"learning_rate": 1.9482186347325236e-05,
|
| 4250 |
+
"loss": 0.0357,
|
| 4251 |
+
"step": 30300
|
| 4252 |
+
},
|
| 4253 |
+
{
|
| 4254 |
+
"epoch": 0.02618596087377418,
|
| 4255 |
+
"grad_norm": 3.485973834991455,
|
| 4256 |
+
"learning_rate": 1.9481323324570776e-05,
|
| 4257 |
+
"loss": 0.0266,
|
| 4258 |
+
"step": 30350
|
| 4259 |
+
},
|
| 4260 |
+
{
|
| 4261 |
+
"epoch": 0.026229100842264746,
|
| 4262 |
+
"grad_norm": 6.07471008606808e-07,
|
| 4263 |
+
"learning_rate": 1.948046030181632e-05,
|
| 4264 |
+
"loss": 0.0308,
|
| 4265 |
+
"step": 30400
|
| 4266 |
+
},
|
| 4267 |
+
{
|
| 4268 |
+
"epoch": 0.026272240810755312,
|
| 4269 |
+
"grad_norm": 7.532801760135044e-08,
|
| 4270 |
+
"learning_rate": 1.9479597279061863e-05,
|
| 4271 |
+
"loss": 0.047,
|
| 4272 |
+
"step": 30450
|
| 4273 |
+
},
|
| 4274 |
+
{
|
| 4275 |
+
"epoch": 0.02631538077924588,
|
| 4276 |
+
"grad_norm": 0.0005202541360631585,
|
| 4277 |
+
"learning_rate": 1.9478734256307403e-05,
|
| 4278 |
+
"loss": 0.0239,
|
| 4279 |
+
"step": 30500
|
| 4280 |
+
},
|
| 4281 |
+
{
|
| 4282 |
+
"epoch": 0.026358520747736446,
|
| 4283 |
+
"grad_norm": 26.940954208374023,
|
| 4284 |
+
"learning_rate": 1.9477871233552946e-05,
|
| 4285 |
+
"loss": 0.0709,
|
| 4286 |
+
"step": 30550
|
| 4287 |
+
},
|
| 4288 |
+
{
|
| 4289 |
+
"epoch": 0.026401660716227012,
|
| 4290 |
+
"grad_norm": 7.362630470575393e-12,
|
| 4291 |
+
"learning_rate": 1.9477008210798486e-05,
|
| 4292 |
+
"loss": 0.0117,
|
| 4293 |
+
"step": 30600
|
| 4294 |
+
},
|
| 4295 |
+
{
|
| 4296 |
+
"epoch": 0.02644480068471758,
|
| 4297 |
+
"grad_norm": 1.2002854418824427e-05,
|
| 4298 |
+
"learning_rate": 1.947614518804403e-05,
|
| 4299 |
+
"loss": 0.0392,
|
| 4300 |
+
"step": 30650
|
| 4301 |
+
},
|
| 4302 |
+
{
|
| 4303 |
+
"epoch": 0.026487940653208145,
|
| 4304 |
+
"grad_norm": 0.4743211269378662,
|
| 4305 |
+
"learning_rate": 1.947528216528957e-05,
|
| 4306 |
+
"loss": 0.0461,
|
| 4307 |
+
"step": 30700
|
| 4308 |
+
},
|
| 4309 |
+
{
|
| 4310 |
+
"epoch": 0.026531080621698715,
|
| 4311 |
+
"grad_norm": 5.520277568393794e-10,
|
| 4312 |
+
"learning_rate": 1.9474419142535113e-05,
|
| 4313 |
+
"loss": 0.002,
|
| 4314 |
+
"step": 30750
|
| 4315 |
+
},
|
| 4316 |
+
{
|
| 4317 |
+
"epoch": 0.026574220590189282,
|
| 4318 |
+
"grad_norm": 5.655643420254819e-08,
|
| 4319 |
+
"learning_rate": 1.9473556119780656e-05,
|
| 4320 |
+
"loss": 0.002,
|
| 4321 |
+
"step": 30800
|
| 4322 |
+
},
|
| 4323 |
+
{
|
| 4324 |
+
"epoch": 0.02661736055867985,
|
| 4325 |
+
"grad_norm": 3.0585747481381986e-06,
|
| 4326 |
+
"learning_rate": 1.94726930970262e-05,
|
| 4327 |
+
"loss": 0.0208,
|
| 4328 |
+
"step": 30850
|
| 4329 |
+
},
|
| 4330 |
+
{
|
| 4331 |
+
"epoch": 0.026660500527170415,
|
| 4332 |
+
"grad_norm": 0.00038789489190094173,
|
| 4333 |
+
"learning_rate": 1.947183007427174e-05,
|
| 4334 |
+
"loss": 0.0221,
|
| 4335 |
+
"step": 30900
|
| 4336 |
+
},
|
| 4337 |
+
{
|
| 4338 |
+
"epoch": 0.02670364049566098,
|
| 4339 |
+
"grad_norm": 0.006069442722946405,
|
| 4340 |
+
"learning_rate": 1.9470967051517283e-05,
|
| 4341 |
+
"loss": 0.0332,
|
| 4342 |
+
"step": 30950
|
| 4343 |
+
},
|
| 4344 |
+
{
|
| 4345 |
+
"epoch": 0.02674678046415155,
|
| 4346 |
+
"grad_norm": 2.1992854204455625e-09,
|
| 4347 |
+
"learning_rate": 1.9470104028762823e-05,
|
| 4348 |
+
"loss": 0.0133,
|
| 4349 |
+
"step": 31000
|
| 4350 |
+
},
|
| 4351 |
+
{
|
| 4352 |
+
"epoch": 0.026789920432642115,
|
| 4353 |
+
"grad_norm": 0.0005120674031786621,
|
| 4354 |
+
"learning_rate": 1.9469241006008366e-05,
|
| 4355 |
+
"loss": 0.0549,
|
| 4356 |
+
"step": 31050
|
| 4357 |
+
},
|
| 4358 |
+
{
|
| 4359 |
+
"epoch": 0.02683306040113268,
|
| 4360 |
+
"grad_norm": 3.589123298297636e-05,
|
| 4361 |
+
"learning_rate": 1.946837798325391e-05,
|
| 4362 |
+
"loss": 0.0172,
|
| 4363 |
+
"step": 31100
|
| 4364 |
+
},
|
| 4365 |
+
{
|
| 4366 |
+
"epoch": 0.02687620036962325,
|
| 4367 |
+
"grad_norm": 4.615823812059716e-08,
|
| 4368 |
+
"learning_rate": 1.946751496049945e-05,
|
| 4369 |
+
"loss": 0.013,
|
| 4370 |
+
"step": 31150
|
| 4371 |
+
},
|
| 4372 |
+
{
|
| 4373 |
+
"epoch": 0.026919340338113818,
|
| 4374 |
+
"grad_norm": 7.231820475794848e-09,
|
| 4375 |
+
"learning_rate": 1.9466651937744993e-05,
|
| 4376 |
+
"loss": 0.0037,
|
| 4377 |
+
"step": 31200
|
| 4378 |
+
},
|
| 4379 |
+
{
|
| 4380 |
+
"epoch": 0.026962480306604385,
|
| 4381 |
+
"grad_norm": 5.052131157867734e-09,
|
| 4382 |
+
"learning_rate": 1.9465788914990532e-05,
|
| 4383 |
+
"loss": 0.0491,
|
| 4384 |
+
"step": 31250
|
| 4385 |
+
},
|
| 4386 |
+
{
|
| 4387 |
+
"epoch": 0.02700562027509495,
|
| 4388 |
+
"grad_norm": 0.00010309406206943095,
|
| 4389 |
+
"learning_rate": 1.9464925892236076e-05,
|
| 4390 |
+
"loss": 0.0028,
|
| 4391 |
+
"step": 31300
|
| 4392 |
+
},
|
| 4393 |
+
{
|
| 4394 |
+
"epoch": 0.027048760243585518,
|
| 4395 |
+
"grad_norm": 5.6031745771178976e-05,
|
| 4396 |
+
"learning_rate": 1.946406286948162e-05,
|
| 4397 |
+
"loss": 0.015,
|
| 4398 |
+
"step": 31350
|
| 4399 |
+
},
|
| 4400 |
+
{
|
| 4401 |
+
"epoch": 0.027091900212076084,
|
| 4402 |
+
"grad_norm": 0.00024476449470967054,
|
| 4403 |
+
"learning_rate": 1.9463199846727162e-05,
|
| 4404 |
+
"loss": 0.0154,
|
| 4405 |
+
"step": 31400
|
| 4406 |
+
},
|
| 4407 |
+
{
|
| 4408 |
+
"epoch": 0.02713504018056665,
|
| 4409 |
+
"grad_norm": 2.0063467331965512e-07,
|
| 4410 |
+
"learning_rate": 1.9462336823972702e-05,
|
| 4411 |
+
"loss": 0.0212,
|
| 4412 |
+
"step": 31450
|
| 4413 |
+
},
|
| 4414 |
+
{
|
| 4415 |
+
"epoch": 0.027178180149057218,
|
| 4416 |
+
"grad_norm": 6.659844075329602e-05,
|
| 4417 |
+
"learning_rate": 1.9461473801218246e-05,
|
| 4418 |
+
"loss": 0.0216,
|
| 4419 |
+
"step": 31500
|
| 4420 |
+
},
|
| 4421 |
+
{
|
| 4422 |
+
"epoch": 0.027221320117547788,
|
| 4423 |
+
"grad_norm": 4.053091470268555e-05,
|
| 4424 |
+
"learning_rate": 1.9460610778463786e-05,
|
| 4425 |
+
"loss": 0.026,
|
| 4426 |
+
"step": 31550
|
| 4427 |
+
},
|
| 4428 |
+
{
|
| 4429 |
+
"epoch": 0.027264460086038354,
|
| 4430 |
+
"grad_norm": 2.6744512382492758e-08,
|
| 4431 |
+
"learning_rate": 1.945974775570933e-05,
|
| 4432 |
+
"loss": 0.0284,
|
| 4433 |
+
"step": 31600
|
| 4434 |
+
},
|
| 4435 |
+
{
|
| 4436 |
+
"epoch": 0.02730760005452892,
|
| 4437 |
+
"grad_norm": 0.1950395703315735,
|
| 4438 |
+
"learning_rate": 1.945888473295487e-05,
|
| 4439 |
+
"loss": 0.0064,
|
| 4440 |
+
"step": 31650
|
| 4441 |
+
},
|
| 4442 |
+
{
|
| 4443 |
+
"epoch": 0.027350740023019487,
|
| 4444 |
+
"grad_norm": 41.71430587768555,
|
| 4445 |
+
"learning_rate": 1.9458021710200412e-05,
|
| 4446 |
+
"loss": 0.0379,
|
| 4447 |
+
"step": 31700
|
| 4448 |
+
},
|
| 4449 |
+
{
|
| 4450 |
+
"epoch": 0.027393879991510054,
|
| 4451 |
+
"grad_norm": 2.8257717943347416e-08,
|
| 4452 |
+
"learning_rate": 1.9457158687445952e-05,
|
| 4453 |
+
"loss": 0.0263,
|
| 4454 |
+
"step": 31750
|
| 4455 |
+
},
|
| 4456 |
+
{
|
| 4457 |
+
"epoch": 0.02743701996000062,
|
| 4458 |
+
"grad_norm": 0.002763712080195546,
|
| 4459 |
+
"learning_rate": 1.9456295664691496e-05,
|
| 4460 |
+
"loss": 0.0189,
|
| 4461 |
+
"step": 31800
|
| 4462 |
+
},
|
| 4463 |
+
{
|
| 4464 |
+
"epoch": 0.027480159928491187,
|
| 4465 |
+
"grad_norm": 1.0972726061098115e-09,
|
| 4466 |
+
"learning_rate": 1.945543264193704e-05,
|
| 4467 |
+
"loss": 0.0156,
|
| 4468 |
+
"step": 31850
|
| 4469 |
+
},
|
| 4470 |
+
{
|
| 4471 |
+
"epoch": 0.027523299896981754,
|
| 4472 |
+
"grad_norm": 0.0012834984809160233,
|
| 4473 |
+
"learning_rate": 1.9454569619182582e-05,
|
| 4474 |
+
"loss": 0.0076,
|
| 4475 |
+
"step": 31900
|
| 4476 |
+
},
|
| 4477 |
+
{
|
| 4478 |
+
"epoch": 0.027566439865472324,
|
| 4479 |
+
"grad_norm": 2.497445628080186e-08,
|
| 4480 |
+
"learning_rate": 1.9453706596428122e-05,
|
| 4481 |
+
"loss": 0.0209,
|
| 4482 |
+
"step": 31950
|
| 4483 |
+
},
|
| 4484 |
+
{
|
| 4485 |
+
"epoch": 0.02760957983396289,
|
| 4486 |
+
"grad_norm": 23.704517364501953,
|
| 4487 |
+
"learning_rate": 1.9452843573673666e-05,
|
| 4488 |
+
"loss": 0.0603,
|
| 4489 |
+
"step": 32000
|
| 4490 |
+
},
|
| 4491 |
+
{
|
| 4492 |
+
"epoch": 0.027652719802453457,
|
| 4493 |
+
"grad_norm": 0.0009068456711247563,
|
| 4494 |
+
"learning_rate": 1.945198055091921e-05,
|
| 4495 |
+
"loss": 0.035,
|
| 4496 |
+
"step": 32050
|
| 4497 |
+
},
|
| 4498 |
+
{
|
| 4499 |
+
"epoch": 0.027695859770944024,
|
| 4500 |
+
"grad_norm": 5.298162460327148,
|
| 4501 |
+
"learning_rate": 1.945111752816475e-05,
|
| 4502 |
+
"loss": 0.0053,
|
| 4503 |
+
"step": 32100
|
| 4504 |
+
},
|
| 4505 |
+
{
|
| 4506 |
+
"epoch": 0.02773899973943459,
|
| 4507 |
+
"grad_norm": 0.017380917444825172,
|
| 4508 |
+
"learning_rate": 1.9450254505410292e-05,
|
| 4509 |
+
"loss": 0.0471,
|
| 4510 |
+
"step": 32150
|
| 4511 |
+
},
|
| 4512 |
+
{
|
| 4513 |
+
"epoch": 0.027782139707925157,
|
| 4514 |
+
"grad_norm": 0.02581915073096752,
|
| 4515 |
+
"learning_rate": 1.9449391482655832e-05,
|
| 4516 |
+
"loss": 0.0396,
|
| 4517 |
+
"step": 32200
|
| 4518 |
+
},
|
| 4519 |
+
{
|
| 4520 |
+
"epoch": 0.027825279676415723,
|
| 4521 |
+
"grad_norm": 1.437704066908907e-09,
|
| 4522 |
+
"learning_rate": 1.9448528459901375e-05,
|
| 4523 |
+
"loss": 0.0283,
|
| 4524 |
+
"step": 32250
|
| 4525 |
+
},
|
| 4526 |
+
{
|
| 4527 |
+
"epoch": 0.02786841964490629,
|
| 4528 |
+
"grad_norm": 1.0882466483508324e-08,
|
| 4529 |
+
"learning_rate": 1.9447665437146915e-05,
|
| 4530 |
+
"loss": 0.0066,
|
| 4531 |
+
"step": 32300
|
| 4532 |
+
},
|
| 4533 |
+
{
|
| 4534 |
+
"epoch": 0.02791155961339686,
|
| 4535 |
+
"grad_norm": 5.027173122229556e-11,
|
| 4536 |
+
"learning_rate": 1.9446802414392462e-05,
|
| 4537 |
+
"loss": 0.0082,
|
| 4538 |
+
"step": 32350
|
| 4539 |
+
},
|
| 4540 |
+
{
|
| 4541 |
+
"epoch": 0.027954699581887427,
|
| 4542 |
+
"grad_norm": 4.071168899536133,
|
| 4543 |
+
"learning_rate": 1.9445939391638002e-05,
|
| 4544 |
+
"loss": 0.0217,
|
| 4545 |
+
"step": 32400
|
| 4546 |
+
},
|
| 4547 |
+
{
|
| 4548 |
+
"epoch": 0.027997839550377993,
|
| 4549 |
+
"grad_norm": 0.0017136979149654508,
|
| 4550 |
+
"learning_rate": 1.9445076368883545e-05,
|
| 4551 |
+
"loss": 0.0665,
|
| 4552 |
+
"step": 32450
|
| 4553 |
+
},
|
| 4554 |
+
{
|
| 4555 |
+
"epoch": 0.02804097951886856,
|
| 4556 |
+
"grad_norm": 1.7071112301536573e-09,
|
| 4557 |
+
"learning_rate": 1.9444213346129085e-05,
|
| 4558 |
+
"loss": 0.0283,
|
| 4559 |
+
"step": 32500
|
| 4560 |
+
},
|
| 4561 |
+
{
|
| 4562 |
+
"epoch": 0.028084119487359126,
|
| 4563 |
+
"grad_norm": 2.8745741897928667e-10,
|
| 4564 |
+
"learning_rate": 1.944335032337463e-05,
|
| 4565 |
+
"loss": 0.0165,
|
| 4566 |
+
"step": 32550
|
| 4567 |
+
},
|
| 4568 |
+
{
|
| 4569 |
+
"epoch": 0.028127259455849693,
|
| 4570 |
+
"grad_norm": 0.06553611904382706,
|
| 4571 |
+
"learning_rate": 1.944248730062017e-05,
|
| 4572 |
+
"loss": 0.0039,
|
| 4573 |
+
"step": 32600
|
| 4574 |
+
},
|
| 4575 |
+
{
|
| 4576 |
+
"epoch": 0.02817039942434026,
|
| 4577 |
+
"grad_norm": 1.2114237506466452e-05,
|
| 4578 |
+
"learning_rate": 1.9441624277865712e-05,
|
| 4579 |
+
"loss": 0.0053,
|
| 4580 |
+
"step": 32650
|
| 4581 |
+
},
|
| 4582 |
+
{
|
| 4583 |
+
"epoch": 0.028213539392830826,
|
| 4584 |
+
"grad_norm": 5.977819910185644e-06,
|
| 4585 |
+
"learning_rate": 1.9440761255111252e-05,
|
| 4586 |
+
"loss": 0.016,
|
| 4587 |
+
"step": 32700
|
| 4588 |
+
},
|
| 4589 |
+
{
|
| 4590 |
+
"epoch": 0.028256679361321393,
|
| 4591 |
+
"grad_norm": 0.00414885301142931,
|
| 4592 |
+
"learning_rate": 1.9439898232356795e-05,
|
| 4593 |
+
"loss": 0.0064,
|
| 4594 |
+
"step": 32750
|
| 4595 |
+
},
|
| 4596 |
+
{
|
| 4597 |
+
"epoch": 0.028299819329811963,
|
| 4598 |
+
"grad_norm": 0.001667422242462635,
|
| 4599 |
+
"learning_rate": 1.943903520960234e-05,
|
| 4600 |
+
"loss": 0.0013,
|
| 4601 |
+
"step": 32800
|
| 4602 |
+
},
|
| 4603 |
+
{
|
| 4604 |
+
"epoch": 0.02834295929830253,
|
| 4605 |
+
"grad_norm": 1.7196412045450415e-06,
|
| 4606 |
+
"learning_rate": 1.943817218684788e-05,
|
| 4607 |
+
"loss": 0.0022,
|
| 4608 |
+
"step": 32850
|
| 4609 |
+
},
|
| 4610 |
+
{
|
| 4611 |
+
"epoch": 0.028386099266793096,
|
| 4612 |
+
"grad_norm": 4.220390792397666e-07,
|
| 4613 |
+
"learning_rate": 1.9437309164093422e-05,
|
| 4614 |
+
"loss": 0.0278,
|
| 4615 |
+
"step": 32900
|
| 4616 |
+
},
|
| 4617 |
+
{
|
| 4618 |
+
"epoch": 0.028429239235283663,
|
| 4619 |
+
"grad_norm": 8.6249691833018e-09,
|
| 4620 |
+
"learning_rate": 1.9436446141338965e-05,
|
| 4621 |
+
"loss": 0.0155,
|
| 4622 |
+
"step": 32950
|
| 4623 |
+
},
|
| 4624 |
+
{
|
| 4625 |
+
"epoch": 0.02847237920377423,
|
| 4626 |
+
"grad_norm": 21.435453414916992,
|
| 4627 |
+
"learning_rate": 1.9435583118584505e-05,
|
| 4628 |
+
"loss": 0.0234,
|
| 4629 |
+
"step": 33000
|
| 4630 |
+
},
|
| 4631 |
+
{
|
| 4632 |
+
"epoch": 0.028515519172264796,
|
| 4633 |
+
"grad_norm": 9.135671461990569e-06,
|
| 4634 |
+
"learning_rate": 1.943472009583005e-05,
|
| 4635 |
+
"loss": 0.0028,
|
| 4636 |
+
"step": 33050
|
| 4637 |
+
},
|
| 4638 |
+
{
|
| 4639 |
+
"epoch": 0.028558659140755362,
|
| 4640 |
+
"grad_norm": 1.085790088950489e-07,
|
| 4641 |
+
"learning_rate": 1.9433857073075592e-05,
|
| 4642 |
+
"loss": 0.0189,
|
| 4643 |
+
"step": 33100
|
| 4644 |
+
},
|
| 4645 |
+
{
|
| 4646 |
+
"epoch": 0.02860179910924593,
|
| 4647 |
+
"grad_norm": 1.0733113288879395,
|
| 4648 |
+
"learning_rate": 1.9432994050321132e-05,
|
| 4649 |
+
"loss": 0.0188,
|
| 4650 |
+
"step": 33150
|
| 4651 |
+
},
|
| 4652 |
+
{
|
| 4653 |
+
"epoch": 0.0286449390777365,
|
| 4654 |
+
"grad_norm": 5.325038046066766e-07,
|
| 4655 |
+
"learning_rate": 1.9432131027566675e-05,
|
| 4656 |
+
"loss": 0.0025,
|
| 4657 |
+
"step": 33200
|
| 4658 |
+
},
|
| 4659 |
+
{
|
| 4660 |
+
"epoch": 0.028688079046227066,
|
| 4661 |
+
"grad_norm": 0.001730454503558576,
|
| 4662 |
+
"learning_rate": 1.9431268004812215e-05,
|
| 4663 |
+
"loss": 0.0429,
|
| 4664 |
+
"step": 33250
|
| 4665 |
+
},
|
| 4666 |
+
{
|
| 4667 |
+
"epoch": 0.028731219014717632,
|
| 4668 |
+
"grad_norm": 0.03524341806769371,
|
| 4669 |
+
"learning_rate": 1.943040498205776e-05,
|
| 4670 |
+
"loss": 0.0147,
|
| 4671 |
+
"step": 33300
|
| 4672 |
+
},
|
| 4673 |
+
{
|
| 4674 |
+
"epoch": 0.0287743589832082,
|
| 4675 |
+
"grad_norm": 8.027368769703003e-10,
|
| 4676 |
+
"learning_rate": 1.9429541959303298e-05,
|
| 4677 |
+
"loss": 0.0074,
|
| 4678 |
+
"step": 33350
|
| 4679 |
+
},
|
| 4680 |
+
{
|
| 4681 |
+
"epoch": 0.028817498951698765,
|
| 4682 |
+
"grad_norm": 2.603889299734874e-07,
|
| 4683 |
+
"learning_rate": 1.9428678936548845e-05,
|
| 4684 |
+
"loss": 0.0015,
|
| 4685 |
+
"step": 33400
|
| 4686 |
+
},
|
| 4687 |
+
{
|
| 4688 |
+
"epoch": 0.028860638920189332,
|
| 4689 |
+
"grad_norm": 12.171298027038574,
|
| 4690 |
+
"learning_rate": 1.9427815913794385e-05,
|
| 4691 |
+
"loss": 0.0188,
|
| 4692 |
+
"step": 33450
|
| 4693 |
+
},
|
| 4694 |
+
{
|
| 4695 |
+
"epoch": 0.0289037788886799,
|
| 4696 |
+
"grad_norm": 3.4058632536471123e-06,
|
| 4697 |
+
"learning_rate": 1.9426952891039928e-05,
|
| 4698 |
+
"loss": 0.0529,
|
| 4699 |
+
"step": 33500
|
| 4700 |
+
},
|
| 4701 |
+
{
|
| 4702 |
+
"epoch": 0.028946918857170465,
|
| 4703 |
+
"grad_norm": 17.399200439453125,
|
| 4704 |
+
"learning_rate": 1.9426089868285468e-05,
|
| 4705 |
+
"loss": 0.0294,
|
| 4706 |
+
"step": 33550
|
| 4707 |
+
},
|
| 4708 |
+
{
|
| 4709 |
+
"epoch": 0.028990058825661035,
|
| 4710 |
+
"grad_norm": 0.011678768321871758,
|
| 4711 |
+
"learning_rate": 1.942522684553101e-05,
|
| 4712 |
+
"loss": 0.0211,
|
| 4713 |
+
"step": 33600
|
| 4714 |
+
},
|
| 4715 |
+
{
|
| 4716 |
+
"epoch": 0.029033198794151602,
|
| 4717 |
+
"grad_norm": 2.466938212819514e-06,
|
| 4718 |
+
"learning_rate": 1.942436382277655e-05,
|
| 4719 |
+
"loss": 0.03,
|
| 4720 |
+
"step": 33650
|
| 4721 |
+
},
|
| 4722 |
+
{
|
| 4723 |
+
"epoch": 0.02907633876264217,
|
| 4724 |
+
"grad_norm": 3.6094334986136456e-12,
|
| 4725 |
+
"learning_rate": 1.9423500800022095e-05,
|
| 4726 |
+
"loss": 0.0381,
|
| 4727 |
+
"step": 33700
|
| 4728 |
+
},
|
| 4729 |
+
{
|
| 4730 |
+
"epoch": 0.029119478731132735,
|
| 4731 |
+
"grad_norm": 0.08116328716278076,
|
| 4732 |
+
"learning_rate": 1.9422637777267638e-05,
|
| 4733 |
+
"loss": 0.0016,
|
| 4734 |
+
"step": 33750
|
| 4735 |
+
},
|
| 4736 |
+
{
|
| 4737 |
+
"epoch": 0.0291626186996233,
|
| 4738 |
+
"grad_norm": 0.2594936788082123,
|
| 4739 |
+
"learning_rate": 1.9421774754513178e-05,
|
| 4740 |
+
"loss": 0.0145,
|
| 4741 |
+
"step": 33800
|
| 4742 |
+
},
|
| 4743 |
+
{
|
| 4744 |
+
"epoch": 0.029205758668113868,
|
| 4745 |
+
"grad_norm": 1.6326714103342965e-05,
|
| 4746 |
+
"learning_rate": 1.942091173175872e-05,
|
| 4747 |
+
"loss": 0.014,
|
| 4748 |
+
"step": 33850
|
| 4749 |
+
},
|
| 4750 |
+
{
|
| 4751 |
+
"epoch": 0.029248898636604435,
|
| 4752 |
+
"grad_norm": 6.704578368044167e-07,
|
| 4753 |
+
"learning_rate": 1.942004870900426e-05,
|
| 4754 |
+
"loss": 0.0138,
|
| 4755 |
+
"step": 33900
|
| 4756 |
+
},
|
| 4757 |
+
{
|
| 4758 |
+
"epoch": 0.029292038605095,
|
| 4759 |
+
"grad_norm": 1.600632737464025e-09,
|
| 4760 |
+
"learning_rate": 1.9419185686249805e-05,
|
| 4761 |
+
"loss": 0.0044,
|
| 4762 |
+
"step": 33950
|
| 4763 |
+
},
|
| 4764 |
+
{
|
| 4765 |
+
"epoch": 0.02933517857358557,
|
| 4766 |
+
"grad_norm": 2.9473580070771277e-05,
|
| 4767 |
+
"learning_rate": 1.9418322663495348e-05,
|
| 4768 |
+
"loss": 0.0209,
|
| 4769 |
+
"step": 34000
|
| 4770 |
+
},
|
| 4771 |
+
{
|
| 4772 |
+
"epoch": 0.029378318542076138,
|
| 4773 |
+
"grad_norm": 0.013792168349027634,
|
| 4774 |
+
"learning_rate": 1.941745964074089e-05,
|
| 4775 |
+
"loss": 0.009,
|
| 4776 |
+
"step": 34050
|
| 4777 |
+
},
|
| 4778 |
+
{
|
| 4779 |
+
"epoch": 0.029421458510566705,
|
| 4780 |
+
"grad_norm": 1.5911604123175493e-07,
|
| 4781 |
+
"learning_rate": 1.941659661798643e-05,
|
| 4782 |
+
"loss": 0.0272,
|
| 4783 |
+
"step": 34100
|
| 4784 |
+
},
|
| 4785 |
+
{
|
| 4786 |
+
"epoch": 0.02946459847905727,
|
| 4787 |
+
"grad_norm": 0.29515737295150757,
|
| 4788 |
+
"learning_rate": 1.9415733595231975e-05,
|
| 4789 |
+
"loss": 0.0595,
|
| 4790 |
+
"step": 34150
|
| 4791 |
+
},
|
| 4792 |
+
{
|
| 4793 |
+
"epoch": 0.029507738447547838,
|
| 4794 |
+
"grad_norm": 2.744394862475019e-07,
|
| 4795 |
+
"learning_rate": 1.9414870572477515e-05,
|
| 4796 |
+
"loss": 0.046,
|
| 4797 |
+
"step": 34200
|
| 4798 |
+
},
|
| 4799 |
+
{
|
| 4800 |
+
"epoch": 0.029550878416038404,
|
| 4801 |
+
"grad_norm": 0.028887495398521423,
|
| 4802 |
+
"learning_rate": 1.9414007549723058e-05,
|
| 4803 |
+
"loss": 0.0014,
|
| 4804 |
+
"step": 34250
|
| 4805 |
+
},
|
| 4806 |
+
{
|
| 4807 |
+
"epoch": 0.02959401838452897,
|
| 4808 |
+
"grad_norm": 1.5995985449990258e-05,
|
| 4809 |
+
"learning_rate": 1.9413144526968598e-05,
|
| 4810 |
+
"loss": 0.0072,
|
| 4811 |
+
"step": 34300
|
| 4812 |
+
},
|
| 4813 |
+
{
|
| 4814 |
+
"epoch": 0.029637158353019537,
|
| 4815 |
+
"grad_norm": 1.774524207576178e-05,
|
| 4816 |
+
"learning_rate": 1.941228150421414e-05,
|
| 4817 |
+
"loss": 0.0072,
|
| 4818 |
+
"step": 34350
|
| 4819 |
+
},
|
| 4820 |
+
{
|
| 4821 |
+
"epoch": 0.029680298321510108,
|
| 4822 |
+
"grad_norm": 3.840292084333896e-09,
|
| 4823 |
+
"learning_rate": 1.9411418481459685e-05,
|
| 4824 |
+
"loss": 0.015,
|
| 4825 |
+
"step": 34400
|
| 4826 |
+
},
|
| 4827 |
+
{
|
| 4828 |
+
"epoch": 0.029723438290000674,
|
| 4829 |
+
"grad_norm": 4.855828592553735e-06,
|
| 4830 |
+
"learning_rate": 1.9410555458705225e-05,
|
| 4831 |
+
"loss": 0.0101,
|
| 4832 |
+
"step": 34450
|
| 4833 |
+
},
|
| 4834 |
+
{
|
| 4835 |
+
"epoch": 0.02976657825849124,
|
| 4836 |
+
"grad_norm": 5.043638229370117,
|
| 4837 |
+
"learning_rate": 1.9409692435950768e-05,
|
| 4838 |
+
"loss": 0.0598,
|
| 4839 |
+
"step": 34500
|
| 4840 |
+
},
|
| 4841 |
+
{
|
| 4842 |
+
"epoch": 0.029809718226981807,
|
| 4843 |
+
"grad_norm": 3.365451473058556e-09,
|
| 4844 |
+
"learning_rate": 1.940882941319631e-05,
|
| 4845 |
+
"loss": 0.012,
|
| 4846 |
+
"step": 34550
|
| 4847 |
+
},
|
| 4848 |
+
{
|
| 4849 |
+
"epoch": 0.029852858195472374,
|
| 4850 |
+
"grad_norm": 2.5963392999983625e-06,
|
| 4851 |
+
"learning_rate": 1.940796639044185e-05,
|
| 4852 |
+
"loss": 0.0195,
|
| 4853 |
+
"step": 34600
|
| 4854 |
+
},
|
| 4855 |
+
{
|
| 4856 |
+
"epoch": 0.02989599816396294,
|
| 4857 |
+
"grad_norm": 0.0003348338359501213,
|
| 4858 |
+
"learning_rate": 1.9407103367687394e-05,
|
| 4859 |
+
"loss": 0.0289,
|
| 4860 |
+
"step": 34650
|
| 4861 |
+
},
|
| 4862 |
+
{
|
| 4863 |
+
"epoch": 0.029939138132453507,
|
| 4864 |
+
"grad_norm": 6.386066436767578,
|
| 4865 |
+
"learning_rate": 1.9406240344932938e-05,
|
| 4866 |
+
"loss": 0.0308,
|
| 4867 |
+
"step": 34700
|
| 4868 |
+
},
|
| 4869 |
+
{
|
| 4870 |
+
"epoch": 0.029982278100944074,
|
| 4871 |
+
"grad_norm": 0.00012195282033644617,
|
| 4872 |
+
"learning_rate": 1.9405377322178478e-05,
|
| 4873 |
+
"loss": 0.0522,
|
| 4874 |
+
"step": 34750
|
| 4875 |
+
},
|
| 4876 |
+
{
|
| 4877 |
+
"epoch": 0.030025418069434644,
|
| 4878 |
+
"grad_norm": 0.0025203858967870474,
|
| 4879 |
+
"learning_rate": 1.940451429942402e-05,
|
| 4880 |
+
"loss": 0.0275,
|
| 4881 |
+
"step": 34800
|
| 4882 |
+
},
|
| 4883 |
+
{
|
| 4884 |
+
"epoch": 0.03006855803792521,
|
| 4885 |
+
"grad_norm": 4.238718820381848e-10,
|
| 4886 |
+
"learning_rate": 1.940365127666956e-05,
|
| 4887 |
+
"loss": 0.0164,
|
| 4888 |
+
"step": 34850
|
| 4889 |
+
},
|
| 4890 |
+
{
|
| 4891 |
+
"epoch": 0.030111698006415777,
|
| 4892 |
+
"grad_norm": 1.477847000330712e-08,
|
| 4893 |
+
"learning_rate": 1.9402788253915104e-05,
|
| 4894 |
+
"loss": 0.0227,
|
| 4895 |
+
"step": 34900
|
| 4896 |
+
},
|
| 4897 |
+
{
|
| 4898 |
+
"epoch": 0.030154837974906343,
|
| 4899 |
+
"grad_norm": 8.416482621953492e-09,
|
| 4900 |
+
"learning_rate": 1.9401925231160644e-05,
|
| 4901 |
+
"loss": 0.0379,
|
| 4902 |
+
"step": 34950
|
| 4903 |
+
},
|
| 4904 |
+
{
|
| 4905 |
+
"epoch": 0.03019797794339691,
|
| 4906 |
+
"grad_norm": 2.9379866646195296e-06,
|
| 4907 |
+
"learning_rate": 1.940106220840619e-05,
|
| 4908 |
+
"loss": 0.0449,
|
| 4909 |
+
"step": 35000
|
| 4910 |
+
},
|
| 4911 |
+
{
|
| 4912 |
+
"epoch": 0.030241117911887477,
|
| 4913 |
+
"grad_norm": 13.662910461425781,
|
| 4914 |
+
"learning_rate": 1.940019918565173e-05,
|
| 4915 |
+
"loss": 0.0245,
|
| 4916 |
+
"step": 35050
|
| 4917 |
+
},
|
| 4918 |
+
{
|
| 4919 |
+
"epoch": 0.030284257880378043,
|
| 4920 |
+
"grad_norm": 2.694193881325191e-06,
|
| 4921 |
+
"learning_rate": 1.9399336162897274e-05,
|
| 4922 |
+
"loss": 0.0231,
|
| 4923 |
+
"step": 35100
|
| 4924 |
+
},
|
| 4925 |
+
{
|
| 4926 |
+
"epoch": 0.03032739784886861,
|
| 4927 |
+
"grad_norm": 19.55348014831543,
|
| 4928 |
+
"learning_rate": 1.9398473140142814e-05,
|
| 4929 |
+
"loss": 0.0253,
|
| 4930 |
+
"step": 35150
|
| 4931 |
+
},
|
| 4932 |
+
{
|
| 4933 |
+
"epoch": 0.03037053781735918,
|
| 4934 |
+
"grad_norm": 7.588599970631549e-09,
|
| 4935 |
+
"learning_rate": 1.9397610117388358e-05,
|
| 4936 |
+
"loss": 0.026,
|
| 4937 |
+
"step": 35200
|
| 4938 |
+
},
|
| 4939 |
+
{
|
| 4940 |
+
"epoch": 0.030413677785849746,
|
| 4941 |
+
"grad_norm": 6.923779882761494e-10,
|
| 4942 |
+
"learning_rate": 1.9396747094633898e-05,
|
| 4943 |
+
"loss": 0.008,
|
| 4944 |
+
"step": 35250
|
| 4945 |
+
},
|
| 4946 |
+
{
|
| 4947 |
+
"epoch": 0.030456817754340313,
|
| 4948 |
+
"grad_norm": 5.178381456971692e-07,
|
| 4949 |
+
"learning_rate": 1.939588407187944e-05,
|
| 4950 |
+
"loss": 0.0512,
|
| 4951 |
+
"step": 35300
|
| 4952 |
+
},
|
| 4953 |
+
{
|
| 4954 |
+
"epoch": 0.03049995772283088,
|
| 4955 |
+
"grad_norm": 3.179905760930524e-08,
|
| 4956 |
+
"learning_rate": 1.939502104912498e-05,
|
| 4957 |
+
"loss": 0.0314,
|
| 4958 |
+
"step": 35350
|
| 4959 |
+
},
|
| 4960 |
+
{
|
| 4961 |
+
"epoch": 0.030543097691321446,
|
| 4962 |
+
"grad_norm": 0.00010464258957654238,
|
| 4963 |
+
"learning_rate": 1.9394158026370524e-05,
|
| 4964 |
+
"loss": 0.0015,
|
| 4965 |
+
"step": 35400
|
| 4966 |
+
},
|
| 4967 |
+
{
|
| 4968 |
+
"epoch": 0.030586237659812013,
|
| 4969 |
+
"grad_norm": 11.300006866455078,
|
| 4970 |
+
"learning_rate": 1.9393295003616067e-05,
|
| 4971 |
+
"loss": 0.0298,
|
| 4972 |
+
"step": 35450
|
| 4973 |
+
},
|
| 4974 |
+
{
|
| 4975 |
+
"epoch": 0.03062937762830258,
|
| 4976 |
+
"grad_norm": 1.0112120918392975e-07,
|
| 4977 |
+
"learning_rate": 1.9392431980861607e-05,
|
| 4978 |
+
"loss": 0.0235,
|
| 4979 |
+
"step": 35500
|
| 4980 |
+
},
|
| 4981 |
+
{
|
| 4982 |
+
"epoch": 0.030672517596793146,
|
| 4983 |
+
"grad_norm": 0.0002930278715211898,
|
| 4984 |
+
"learning_rate": 1.939156895810715e-05,
|
| 4985 |
+
"loss": 0.0422,
|
| 4986 |
+
"step": 35550
|
| 4987 |
+
},
|
| 4988 |
+
{
|
| 4989 |
+
"epoch": 0.030715657565283716,
|
| 4990 |
+
"grad_norm": 3.265949146680214e-07,
|
| 4991 |
+
"learning_rate": 1.9390705935352694e-05,
|
| 4992 |
+
"loss": 0.0453,
|
| 4993 |
+
"step": 35600
|
| 4994 |
+
},
|
| 4995 |
+
{
|
| 4996 |
+
"epoch": 0.030758797533774283,
|
| 4997 |
+
"grad_norm": 0.01071107853204012,
|
| 4998 |
+
"learning_rate": 1.9389842912598234e-05,
|
| 4999 |
+
"loss": 0.0088,
|
| 5000 |
+
"step": 35650
|
| 5001 |
+
},
|
| 5002 |
+
{
|
| 5003 |
+
"epoch": 0.03080193750226485,
|
| 5004 |
+
"grad_norm": 2.198061288538611e-09,
|
| 5005 |
+
"learning_rate": 1.9388979889843777e-05,
|
| 5006 |
+
"loss": 0.0344,
|
| 5007 |
+
"step": 35700
|
| 5008 |
+
},
|
| 5009 |
+
{
|
| 5010 |
+
"epoch": 0.030845077470755416,
|
| 5011 |
+
"grad_norm": 2.0178050874619657e-07,
|
| 5012 |
+
"learning_rate": 1.938811686708932e-05,
|
| 5013 |
+
"loss": 0.0112,
|
| 5014 |
+
"step": 35750
|
| 5015 |
+
},
|
| 5016 |
+
{
|
| 5017 |
+
"epoch": 0.030888217439245982,
|
| 5018 |
+
"grad_norm": 0.03751551732420921,
|
| 5019 |
+
"learning_rate": 1.938725384433486e-05,
|
| 5020 |
+
"loss": 0.0112,
|
| 5021 |
+
"step": 35800
|
| 5022 |
+
},
|
| 5023 |
+
{
|
| 5024 |
+
"epoch": 0.03093135740773655,
|
| 5025 |
+
"grad_norm": 0.00011108023318229243,
|
| 5026 |
+
"learning_rate": 1.9386390821580404e-05,
|
| 5027 |
+
"loss": 0.0275,
|
| 5028 |
+
"step": 35850
|
| 5029 |
+
},
|
| 5030 |
+
{
|
| 5031 |
+
"epoch": 0.030974497376227116,
|
| 5032 |
+
"grad_norm": 1.5553026644354873e-09,
|
| 5033 |
+
"learning_rate": 1.9385527798825944e-05,
|
| 5034 |
+
"loss": 0.0118,
|
| 5035 |
+
"step": 35900
|
| 5036 |
+
},
|
| 5037 |
+
{
|
| 5038 |
+
"epoch": 0.031017637344717682,
|
| 5039 |
+
"grad_norm": 2.6839693418878596e-06,
|
| 5040 |
+
"learning_rate": 1.9384664776071487e-05,
|
| 5041 |
+
"loss": 0.0054,
|
| 5042 |
+
"step": 35950
|
| 5043 |
+
},
|
| 5044 |
+
{
|
| 5045 |
+
"epoch": 0.03106077731320825,
|
| 5046 |
+
"grad_norm": 2.178272318076324e-08,
|
| 5047 |
+
"learning_rate": 1.9383801753317027e-05,
|
| 5048 |
+
"loss": 0.0331,
|
| 5049 |
+
"step": 36000
|
| 5050 |
+
},
|
| 5051 |
+
{
|
| 5052 |
+
"epoch": 0.03110391728169882,
|
| 5053 |
+
"grad_norm": 2.3207785204704123e-07,
|
| 5054 |
+
"learning_rate": 1.9382938730562574e-05,
|
| 5055 |
+
"loss": 0.0102,
|
| 5056 |
+
"step": 36050
|
| 5057 |
+
},
|
| 5058 |
+
{
|
| 5059 |
+
"epoch": 0.031147057250189385,
|
| 5060 |
+
"grad_norm": 1.738131345518923e-07,
|
| 5061 |
+
"learning_rate": 1.9382075707808114e-05,
|
| 5062 |
+
"loss": 0.0588,
|
| 5063 |
+
"step": 36100
|
| 5064 |
+
},
|
| 5065 |
+
{
|
| 5066 |
+
"epoch": 0.031190197218679952,
|
| 5067 |
+
"grad_norm": 0.019147371873259544,
|
| 5068 |
+
"learning_rate": 1.9381212685053657e-05,
|
| 5069 |
+
"loss": 0.043,
|
| 5070 |
+
"step": 36150
|
| 5071 |
+
},
|
| 5072 |
+
{
|
| 5073 |
+
"epoch": 0.03123333718717052,
|
| 5074 |
+
"grad_norm": 0.0022545859683305025,
|
| 5075 |
+
"learning_rate": 1.9380349662299197e-05,
|
| 5076 |
+
"loss": 0.0191,
|
| 5077 |
+
"step": 36200
|
| 5078 |
+
},
|
| 5079 |
+
{
|
| 5080 |
+
"epoch": 0.031276477155661085,
|
| 5081 |
+
"grad_norm": 0.00014786762767471373,
|
| 5082 |
+
"learning_rate": 1.937948663954474e-05,
|
| 5083 |
+
"loss": 0.0016,
|
| 5084 |
+
"step": 36250
|
| 5085 |
+
},
|
| 5086 |
+
{
|
| 5087 |
+
"epoch": 0.031319617124151655,
|
| 5088 |
+
"grad_norm": 1.8323513018003723e-07,
|
| 5089 |
+
"learning_rate": 1.937862361679028e-05,
|
| 5090 |
+
"loss": 0.0007,
|
| 5091 |
+
"step": 36300
|
| 5092 |
+
},
|
| 5093 |
+
{
|
| 5094 |
+
"epoch": 0.03136275709264222,
|
| 5095 |
+
"grad_norm": 15.16702651977539,
|
| 5096 |
+
"learning_rate": 1.9377760594035824e-05,
|
| 5097 |
+
"loss": 0.0363,
|
| 5098 |
+
"step": 36350
|
| 5099 |
+
},
|
| 5100 |
+
{
|
| 5101 |
+
"epoch": 0.03140589706113279,
|
| 5102 |
+
"grad_norm": 0.061391185969114304,
|
| 5103 |
+
"learning_rate": 1.9376897571281367e-05,
|
| 5104 |
+
"loss": 0.0393,
|
| 5105 |
+
"step": 36400
|
| 5106 |
+
},
|
| 5107 |
+
{
|
| 5108 |
+
"epoch": 0.03144903702962335,
|
| 5109 |
+
"grad_norm": 0.0035098083317279816,
|
| 5110 |
+
"learning_rate": 1.9376034548526907e-05,
|
| 5111 |
+
"loss": 0.0147,
|
| 5112 |
+
"step": 36450
|
| 5113 |
+
},
|
| 5114 |
+
{
|
| 5115 |
+
"epoch": 0.03149217699811392,
|
| 5116 |
+
"grad_norm": 0.06623140722513199,
|
| 5117 |
+
"learning_rate": 1.937517152577245e-05,
|
| 5118 |
+
"loss": 0.0543,
|
| 5119 |
+
"step": 36500
|
| 5120 |
+
},
|
| 5121 |
+
{
|
| 5122 |
+
"epoch": 0.031535316966604485,
|
| 5123 |
+
"grad_norm": 8.011748832359444e-06,
|
| 5124 |
+
"learning_rate": 1.937430850301799e-05,
|
| 5125 |
+
"loss": 0.0447,
|
| 5126 |
+
"step": 36550
|
| 5127 |
+
},
|
| 5128 |
+
{
|
| 5129 |
+
"epoch": 0.031578456935095055,
|
| 5130 |
+
"grad_norm": 2.976227278850274e-06,
|
| 5131 |
+
"learning_rate": 1.9373445480263534e-05,
|
| 5132 |
+
"loss": 0.0238,
|
| 5133 |
+
"step": 36600
|
| 5134 |
+
},
|
| 5135 |
+
{
|
| 5136 |
+
"epoch": 0.031621596903585625,
|
| 5137 |
+
"grad_norm": 4.54370677971383e-07,
|
| 5138 |
+
"learning_rate": 1.9372582457509077e-05,
|
| 5139 |
+
"loss": 0.0282,
|
| 5140 |
+
"step": 36650
|
| 5141 |
+
},
|
| 5142 |
+
{
|
| 5143 |
+
"epoch": 0.03166473687207619,
|
| 5144 |
+
"grad_norm": 1.2593355247503268e-09,
|
| 5145 |
+
"learning_rate": 1.937171943475462e-05,
|
| 5146 |
+
"loss": 0.0475,
|
| 5147 |
+
"step": 36700
|
| 5148 |
+
},
|
| 5149 |
+
{
|
| 5150 |
+
"epoch": 0.03170787684056676,
|
| 5151 |
+
"grad_norm": 0.0001775699929567054,
|
| 5152 |
+
"learning_rate": 1.937085641200016e-05,
|
| 5153 |
+
"loss": 0.0005,
|
| 5154 |
+
"step": 36750
|
| 5155 |
+
},
|
| 5156 |
+
{
|
| 5157 |
+
"epoch": 0.03175101680905732,
|
| 5158 |
+
"grad_norm": 1.9041050336454646e-07,
|
| 5159 |
+
"learning_rate": 1.9369993389245704e-05,
|
| 5160 |
+
"loss": 0.0008,
|
| 5161 |
+
"step": 36800
|
| 5162 |
+
},
|
| 5163 |
+
{
|
| 5164 |
+
"epoch": 0.03179415677754789,
|
| 5165 |
+
"grad_norm": 0.0002166083868360147,
|
| 5166 |
+
"learning_rate": 1.9369130366491244e-05,
|
| 5167 |
+
"loss": 0.0064,
|
| 5168 |
+
"step": 36850
|
| 5169 |
+
},
|
| 5170 |
+
{
|
| 5171 |
+
"epoch": 0.031837296746038454,
|
| 5172 |
+
"grad_norm": 2.4730157921482032e-09,
|
| 5173 |
+
"learning_rate": 1.9368267343736787e-05,
|
| 5174 |
+
"loss": 0.0747,
|
| 5175 |
+
"step": 36900
|
| 5176 |
+
},
|
| 5177 |
+
{
|
| 5178 |
+
"epoch": 0.031880436714529024,
|
| 5179 |
+
"grad_norm": 6.864386705274228e-06,
|
| 5180 |
+
"learning_rate": 1.9367404320982327e-05,
|
| 5181 |
+
"loss": 0.0022,
|
| 5182 |
+
"step": 36950
|
| 5183 |
+
},
|
| 5184 |
+
{
|
| 5185 |
+
"epoch": 0.03192357668301959,
|
| 5186 |
+
"grad_norm": 2.638907517393818e-06,
|
| 5187 |
+
"learning_rate": 1.936654129822787e-05,
|
| 5188 |
+
"loss": 0.0239,
|
| 5189 |
+
"step": 37000
|
| 5190 |
+
},
|
| 5191 |
+
{
|
| 5192 |
+
"epoch": 0.03196671665151016,
|
| 5193 |
+
"grad_norm": 8.631070522824302e-05,
|
| 5194 |
+
"learning_rate": 1.9365678275473413e-05,
|
| 5195 |
+
"loss": 0.0191,
|
| 5196 |
+
"step": 37050
|
| 5197 |
+
},
|
| 5198 |
+
{
|
| 5199 |
+
"epoch": 0.03200985662000073,
|
| 5200 |
+
"grad_norm": 14.52698802947998,
|
| 5201 |
+
"learning_rate": 1.9364815252718953e-05,
|
| 5202 |
+
"loss": 0.0188,
|
| 5203 |
+
"step": 37100
|
| 5204 |
+
},
|
| 5205 |
+
{
|
| 5206 |
+
"epoch": 0.03205299658849129,
|
| 5207 |
+
"grad_norm": 0.07407932728528976,
|
| 5208 |
+
"learning_rate": 1.9363952229964497e-05,
|
| 5209 |
+
"loss": 0.0136,
|
| 5210 |
+
"step": 37150
|
| 5211 |
+
},
|
| 5212 |
+
{
|
| 5213 |
+
"epoch": 0.03209613655698186,
|
| 5214 |
+
"grad_norm": 0.002848062664270401,
|
| 5215 |
+
"learning_rate": 1.936308920721004e-05,
|
| 5216 |
+
"loss": 0.0451,
|
| 5217 |
+
"step": 37200
|
| 5218 |
+
},
|
| 5219 |
+
{
|
| 5220 |
+
"epoch": 0.032139276525472424,
|
| 5221 |
+
"grad_norm": 2.2414766931433405e-07,
|
| 5222 |
+
"learning_rate": 1.936222618445558e-05,
|
| 5223 |
+
"loss": 0.0395,
|
| 5224 |
+
"step": 37250
|
| 5225 |
+
},
|
| 5226 |
+
{
|
| 5227 |
+
"epoch": 0.032182416493962994,
|
| 5228 |
+
"grad_norm": 5.524349830920983e-07,
|
| 5229 |
+
"learning_rate": 1.9361363161701123e-05,
|
| 5230 |
+
"loss": 0.0468,
|
| 5231 |
+
"step": 37300
|
| 5232 |
+
},
|
| 5233 |
+
{
|
| 5234 |
+
"epoch": 0.03222555646245356,
|
| 5235 |
+
"grad_norm": 2.2004120182828046e-05,
|
| 5236 |
+
"learning_rate": 1.9360500138946667e-05,
|
| 5237 |
+
"loss": 0.0599,
|
| 5238 |
+
"step": 37350
|
| 5239 |
+
},
|
| 5240 |
+
{
|
| 5241 |
+
"epoch": 0.03226869643094413,
|
| 5242 |
+
"grad_norm": 5.064206831661977e-08,
|
| 5243 |
+
"learning_rate": 1.9359637116192207e-05,
|
| 5244 |
+
"loss": 0.0191,
|
| 5245 |
+
"step": 37400
|
| 5246 |
+
},
|
| 5247 |
+
{
|
| 5248 |
+
"epoch": 0.0323118363994347,
|
| 5249 |
+
"grad_norm": 5.038096060161479e-05,
|
| 5250 |
+
"learning_rate": 1.935877409343775e-05,
|
| 5251 |
+
"loss": 0.0094,
|
| 5252 |
+
"step": 37450
|
| 5253 |
+
},
|
| 5254 |
+
{
|
| 5255 |
+
"epoch": 0.03235497636792526,
|
| 5256 |
+
"grad_norm": 0.002139901742339134,
|
| 5257 |
+
"learning_rate": 1.935791107068329e-05,
|
| 5258 |
+
"loss": 0.0026,
|
| 5259 |
+
"step": 37500
|
| 5260 |
+
},
|
| 5261 |
+
{
|
| 5262 |
+
"epoch": 0.03239811633641583,
|
| 5263 |
+
"grad_norm": 0.025793571025133133,
|
| 5264 |
+
"learning_rate": 1.9357048047928833e-05,
|
| 5265 |
+
"loss": 0.0503,
|
| 5266 |
+
"step": 37550
|
| 5267 |
+
},
|
| 5268 |
+
{
|
| 5269 |
+
"epoch": 0.03244125630490639,
|
| 5270 |
+
"grad_norm": 1.497374176979065,
|
| 5271 |
+
"learning_rate": 1.9356185025174373e-05,
|
| 5272 |
+
"loss": 0.0239,
|
| 5273 |
+
"step": 37600
|
| 5274 |
+
},
|
| 5275 |
+
{
|
| 5276 |
+
"epoch": 0.032484396273396964,
|
| 5277 |
+
"grad_norm": 9.68094241216022e-07,
|
| 5278 |
+
"learning_rate": 1.935532200241992e-05,
|
| 5279 |
+
"loss": 0.0362,
|
| 5280 |
+
"step": 37650
|
| 5281 |
+
},
|
| 5282 |
+
{
|
| 5283 |
+
"epoch": 0.03252753624188753,
|
| 5284 |
+
"grad_norm": 7.437192266479542e-07,
|
| 5285 |
+
"learning_rate": 1.935445897966546e-05,
|
| 5286 |
+
"loss": 0.0174,
|
| 5287 |
+
"step": 37700
|
| 5288 |
+
},
|
| 5289 |
+
{
|
| 5290 |
+
"epoch": 0.0325706762103781,
|
| 5291 |
+
"grad_norm": 1.591896947594762e-09,
|
| 5292 |
+
"learning_rate": 1.9353595956911003e-05,
|
| 5293 |
+
"loss": 0.0253,
|
| 5294 |
+
"step": 37750
|
| 5295 |
+
},
|
| 5296 |
+
{
|
| 5297 |
+
"epoch": 0.03261381617886866,
|
| 5298 |
+
"grad_norm": 14.039113998413086,
|
| 5299 |
+
"learning_rate": 1.9352732934156543e-05,
|
| 5300 |
+
"loss": 0.0201,
|
| 5301 |
+
"step": 37800
|
| 5302 |
+
},
|
| 5303 |
+
{
|
| 5304 |
+
"epoch": 0.03265695614735923,
|
| 5305 |
+
"grad_norm": 2.0073053747182712e-05,
|
| 5306 |
+
"learning_rate": 1.9351869911402087e-05,
|
| 5307 |
+
"loss": 0.043,
|
| 5308 |
+
"step": 37850
|
| 5309 |
+
},
|
| 5310 |
+
{
|
| 5311 |
+
"epoch": 0.0327000961158498,
|
| 5312 |
+
"grad_norm": 1.3844499768822516e-08,
|
| 5313 |
+
"learning_rate": 1.9351006888647626e-05,
|
| 5314 |
+
"loss": 0.007,
|
| 5315 |
+
"step": 37900
|
| 5316 |
+
},
|
| 5317 |
+
{
|
| 5318 |
+
"epoch": 0.03274323608434036,
|
| 5319 |
+
"grad_norm": 0.02289557084441185,
|
| 5320 |
+
"learning_rate": 1.935014386589317e-05,
|
| 5321 |
+
"loss": 0.0268,
|
| 5322 |
+
"step": 37950
|
| 5323 |
+
},
|
| 5324 |
+
{
|
| 5325 |
+
"epoch": 0.03278637605283093,
|
| 5326 |
+
"grad_norm": 2.7390053766729316e-11,
|
| 5327 |
+
"learning_rate": 1.934928084313871e-05,
|
| 5328 |
+
"loss": 0.0141,
|
| 5329 |
+
"step": 38000
|
| 5330 |
+
},
|
| 5331 |
+
{
|
| 5332 |
+
"epoch": 0.032829516021321496,
|
| 5333 |
+
"grad_norm": 2.0595265937117802e-07,
|
| 5334 |
+
"learning_rate": 1.9348417820384253e-05,
|
| 5335 |
+
"loss": 0.1202,
|
| 5336 |
+
"step": 38050
|
| 5337 |
+
},
|
| 5338 |
+
{
|
| 5339 |
+
"epoch": 0.032872655989812066,
|
| 5340 |
+
"grad_norm": 0.00014018621004652232,
|
| 5341 |
+
"learning_rate": 1.9347554797629796e-05,
|
| 5342 |
+
"loss": 0.0277,
|
| 5343 |
+
"step": 38100
|
| 5344 |
+
},
|
| 5345 |
+
{
|
| 5346 |
+
"epoch": 0.03291579595830263,
|
| 5347 |
+
"grad_norm": 9.558748570270836e-05,
|
| 5348 |
+
"learning_rate": 1.9346691774875336e-05,
|
| 5349 |
+
"loss": 0.0498,
|
| 5350 |
+
"step": 38150
|
| 5351 |
+
},
|
| 5352 |
+
{
|
| 5353 |
+
"epoch": 0.0329589359267932,
|
| 5354 |
+
"grad_norm": 2.20267253325801e-07,
|
| 5355 |
+
"learning_rate": 1.934582875212088e-05,
|
| 5356 |
+
"loss": 0.0357,
|
| 5357 |
+
"step": 38200
|
| 5358 |
+
},
|
| 5359 |
+
{
|
| 5360 |
+
"epoch": 0.03300207589528377,
|
| 5361 |
+
"grad_norm": 0.002117832424119115,
|
| 5362 |
+
"learning_rate": 1.9344965729366423e-05,
|
| 5363 |
+
"loss": 0.0478,
|
| 5364 |
+
"step": 38250
|
| 5365 |
+
},
|
| 5366 |
+
{
|
| 5367 |
+
"epoch": 0.03304521586377433,
|
| 5368 |
+
"grad_norm": 0.0015125697245821357,
|
| 5369 |
+
"learning_rate": 1.9344102706611963e-05,
|
| 5370 |
+
"loss": 0.0049,
|
| 5371 |
+
"step": 38300
|
| 5372 |
+
},
|
| 5373 |
+
{
|
| 5374 |
+
"epoch": 0.0330883558322649,
|
| 5375 |
+
"grad_norm": 0.001929111429490149,
|
| 5376 |
+
"learning_rate": 1.9343239683857506e-05,
|
| 5377 |
+
"loss": 0.0321,
|
| 5378 |
+
"step": 38350
|
| 5379 |
+
},
|
| 5380 |
+
{
|
| 5381 |
+
"epoch": 0.033131495800755466,
|
| 5382 |
+
"grad_norm": 14.052818298339844,
|
| 5383 |
+
"learning_rate": 1.934237666110305e-05,
|
| 5384 |
+
"loss": 0.0126,
|
| 5385 |
+
"step": 38400
|
| 5386 |
+
},
|
| 5387 |
+
{
|
| 5388 |
+
"epoch": 0.033174635769246036,
|
| 5389 |
+
"grad_norm": 0.04780351743102074,
|
| 5390 |
+
"learning_rate": 1.934151363834859e-05,
|
| 5391 |
+
"loss": 0.0145,
|
| 5392 |
+
"step": 38450
|
| 5393 |
+
},
|
| 5394 |
+
{
|
| 5395 |
+
"epoch": 0.0332177757377366,
|
| 5396 |
+
"grad_norm": 1.625859908926941e-07,
|
| 5397 |
+
"learning_rate": 1.9340650615594133e-05,
|
| 5398 |
+
"loss": 0.0006,
|
| 5399 |
+
"step": 38500
|
| 5400 |
+
},
|
| 5401 |
+
{
|
| 5402 |
+
"epoch": 0.03326091570622717,
|
| 5403 |
+
"grad_norm": 4.171390173723921e-06,
|
| 5404 |
+
"learning_rate": 1.9339787592839673e-05,
|
| 5405 |
+
"loss": 0.0052,
|
| 5406 |
+
"step": 38550
|
| 5407 |
+
},
|
| 5408 |
+
{
|
| 5409 |
+
"epoch": 0.03330405567471773,
|
| 5410 |
+
"grad_norm": 9.933991532307118e-05,
|
| 5411 |
+
"learning_rate": 1.9338924570085216e-05,
|
| 5412 |
+
"loss": 0.0149,
|
| 5413 |
+
"step": 38600
|
| 5414 |
+
},
|
| 5415 |
+
{
|
| 5416 |
+
"epoch": 0.0333471956432083,
|
| 5417 |
+
"grad_norm": 5.527433510899016e-10,
|
| 5418 |
+
"learning_rate": 1.9338061547330756e-05,
|
| 5419 |
+
"loss": 0.0569,
|
| 5420 |
+
"step": 38650
|
| 5421 |
+
},
|
| 5422 |
+
{
|
| 5423 |
+
"epoch": 0.03339033561169887,
|
| 5424 |
+
"grad_norm": 1.7711924149566016e-09,
|
| 5425 |
+
"learning_rate": 1.9337198524576303e-05,
|
| 5426 |
+
"loss": 0.0089,
|
| 5427 |
+
"step": 38700
|
| 5428 |
+
},
|
| 5429 |
+
{
|
| 5430 |
+
"epoch": 0.033433475580189435,
|
| 5431 |
+
"grad_norm": 5.876652497960322e-09,
|
| 5432 |
+
"learning_rate": 1.9336335501821843e-05,
|
| 5433 |
+
"loss": 0.0412,
|
| 5434 |
+
"step": 38750
|
| 5435 |
+
},
|
| 5436 |
+
{
|
| 5437 |
+
"epoch": 0.033476615548680005,
|
| 5438 |
+
"grad_norm": 1.2611899375915527,
|
| 5439 |
+
"learning_rate": 1.9335472479067386e-05,
|
| 5440 |
+
"loss": 0.0057,
|
| 5441 |
+
"step": 38800
|
| 5442 |
+
},
|
| 5443 |
+
{
|
| 5444 |
+
"epoch": 0.03351975551717057,
|
| 5445 |
+
"grad_norm": 0.00011541438288986683,
|
| 5446 |
+
"learning_rate": 1.9334609456312926e-05,
|
| 5447 |
+
"loss": 0.0264,
|
| 5448 |
+
"step": 38850
|
| 5449 |
+
},
|
| 5450 |
+
{
|
| 5451 |
+
"epoch": 0.03356289548566114,
|
| 5452 |
+
"grad_norm": 0.7902683019638062,
|
| 5453 |
+
"learning_rate": 1.933374643355847e-05,
|
| 5454 |
+
"loss": 0.0269,
|
| 5455 |
+
"step": 38900
|
| 5456 |
+
},
|
| 5457 |
+
{
|
| 5458 |
+
"epoch": 0.0336060354541517,
|
| 5459 |
+
"grad_norm": 1.6534098904230632e-05,
|
| 5460 |
+
"learning_rate": 1.933288341080401e-05,
|
| 5461 |
+
"loss": 0.0041,
|
| 5462 |
+
"step": 38950
|
| 5463 |
+
},
|
| 5464 |
+
{
|
| 5465 |
+
"epoch": 0.03364917542264227,
|
| 5466 |
+
"grad_norm": 0.029098449274897575,
|
| 5467 |
+
"learning_rate": 1.9332020388049553e-05,
|
| 5468 |
+
"loss": 0.0208,
|
| 5469 |
+
"step": 39000
|
| 5470 |
+
},
|
| 5471 |
+
{
|
| 5472 |
+
"epoch": 0.033692315391132835,
|
| 5473 |
+
"grad_norm": 0.0004794780688825995,
|
| 5474 |
+
"learning_rate": 1.9331157365295096e-05,
|
| 5475 |
+
"loss": 0.0595,
|
| 5476 |
+
"step": 39050
|
| 5477 |
+
},
|
| 5478 |
+
{
|
| 5479 |
+
"epoch": 0.033735455359623405,
|
| 5480 |
+
"grad_norm": 16.320070266723633,
|
| 5481 |
+
"learning_rate": 1.9330294342540636e-05,
|
| 5482 |
+
"loss": 0.0735,
|
| 5483 |
+
"step": 39100
|
| 5484 |
+
},
|
| 5485 |
+
{
|
| 5486 |
+
"epoch": 0.033778595328113975,
|
| 5487 |
+
"grad_norm": 3.635158840609165e-09,
|
| 5488 |
+
"learning_rate": 1.932943131978618e-05,
|
| 5489 |
+
"loss": 0.0164,
|
| 5490 |
+
"step": 39150
|
| 5491 |
+
},
|
| 5492 |
+
{
|
| 5493 |
+
"epoch": 0.03382173529660454,
|
| 5494 |
+
"grad_norm": 1.406357796440716e-06,
|
| 5495 |
+
"learning_rate": 1.932856829703172e-05,
|
| 5496 |
+
"loss": 0.0237,
|
| 5497 |
+
"step": 39200
|
| 5498 |
+
},
|
| 5499 |
+
{
|
| 5500 |
+
"epoch": 0.03386487526509511,
|
| 5501 |
+
"grad_norm": 0.05031180754303932,
|
| 5502 |
+
"learning_rate": 1.9327705274277263e-05,
|
| 5503 |
+
"loss": 0.0264,
|
| 5504 |
+
"step": 39250
|
| 5505 |
+
},
|
| 5506 |
+
{
|
| 5507 |
+
"epoch": 0.03390801523358567,
|
| 5508 |
+
"grad_norm": 0.022205352783203125,
|
| 5509 |
+
"learning_rate": 1.9326842251522806e-05,
|
| 5510 |
+
"loss": 0.0076,
|
| 5511 |
+
"step": 39300
|
| 5512 |
+
},
|
| 5513 |
+
{
|
| 5514 |
+
"epoch": 0.03395115520207624,
|
| 5515 |
+
"grad_norm": 3.1427214707946405e-05,
|
| 5516 |
+
"learning_rate": 1.932597922876835e-05,
|
| 5517 |
+
"loss": 0.0093,
|
| 5518 |
+
"step": 39350
|
| 5519 |
+
},
|
| 5520 |
+
{
|
| 5521 |
+
"epoch": 0.033994295170566804,
|
| 5522 |
+
"grad_norm": 0.0015017461264505982,
|
| 5523 |
+
"learning_rate": 1.932511620601389e-05,
|
| 5524 |
+
"loss": 0.0016,
|
| 5525 |
+
"step": 39400
|
| 5526 |
+
},
|
| 5527 |
+
{
|
| 5528 |
+
"epoch": 0.034037435139057375,
|
| 5529 |
+
"grad_norm": 3.3295341483885466e-10,
|
| 5530 |
+
"learning_rate": 1.9324253183259433e-05,
|
| 5531 |
+
"loss": 0.0297,
|
| 5532 |
+
"step": 39450
|
| 5533 |
+
},
|
| 5534 |
+
{
|
| 5535 |
+
"epoch": 0.034080575107547945,
|
| 5536 |
+
"grad_norm": 1.431539747853705e-10,
|
| 5537 |
+
"learning_rate": 1.9323390160504972e-05,
|
| 5538 |
+
"loss": 0.008,
|
| 5539 |
+
"step": 39500
|
| 5540 |
+
},
|
| 5541 |
+
{
|
| 5542 |
+
"epoch": 0.03412371507603851,
|
| 5543 |
+
"grad_norm": 9.472168188695562e-11,
|
| 5544 |
+
"learning_rate": 1.9322527137750516e-05,
|
| 5545 |
+
"loss": 0.0526,
|
| 5546 |
+
"step": 39550
|
| 5547 |
+
},
|
| 5548 |
+
{
|
| 5549 |
+
"epoch": 0.03416685504452908,
|
| 5550 |
+
"grad_norm": 1.1010347078510563e-09,
|
| 5551 |
+
"learning_rate": 1.9321664114996056e-05,
|
| 5552 |
+
"loss": 0.0438,
|
| 5553 |
+
"step": 39600
|
| 5554 |
+
},
|
| 5555 |
+
{
|
| 5556 |
+
"epoch": 0.03420999501301964,
|
| 5557 |
+
"grad_norm": 0.0038324242923408747,
|
| 5558 |
+
"learning_rate": 1.93208010922416e-05,
|
| 5559 |
+
"loss": 0.0068,
|
| 5560 |
+
"step": 39650
|
| 5561 |
+
},
|
| 5562 |
+
{
|
| 5563 |
+
"epoch": 0.03425313498151021,
|
| 5564 |
+
"grad_norm": 1.2454121067762003e-10,
|
| 5565 |
+
"learning_rate": 1.9319938069487142e-05,
|
| 5566 |
+
"loss": 0.0105,
|
| 5567 |
+
"step": 39700
|
| 5568 |
+
},
|
| 5569 |
+
{
|
| 5570 |
+
"epoch": 0.034296274950000774,
|
| 5571 |
+
"grad_norm": 3.0910987103283105e-09,
|
| 5572 |
+
"learning_rate": 1.9319075046732682e-05,
|
| 5573 |
+
"loss": 0.003,
|
| 5574 |
+
"step": 39750
|
| 5575 |
+
},
|
| 5576 |
+
{
|
| 5577 |
+
"epoch": 0.034339414918491344,
|
| 5578 |
+
"grad_norm": 0.20711366832256317,
|
| 5579 |
+
"learning_rate": 1.9318212023978226e-05,
|
| 5580 |
+
"loss": 0.0072,
|
| 5581 |
+
"step": 39800
|
| 5582 |
+
},
|
| 5583 |
+
{
|
| 5584 |
+
"epoch": 0.03438255488698191,
|
| 5585 |
+
"grad_norm": 0.0013983896933495998,
|
| 5586 |
+
"learning_rate": 1.931734900122377e-05,
|
| 5587 |
+
"loss": 0.0357,
|
| 5588 |
+
"step": 39850
|
| 5589 |
+
},
|
| 5590 |
+
{
|
| 5591 |
+
"epoch": 0.03442569485547248,
|
| 5592 |
+
"grad_norm": 1.195646859741828e-06,
|
| 5593 |
+
"learning_rate": 1.931648597846931e-05,
|
| 5594 |
+
"loss": 0.0252,
|
| 5595 |
+
"step": 39900
|
| 5596 |
+
},
|
| 5597 |
+
{
|
| 5598 |
+
"epoch": 0.03446883482396305,
|
| 5599 |
+
"grad_norm": 0.0007419702014885843,
|
| 5600 |
+
"learning_rate": 1.9315622955714852e-05,
|
| 5601 |
+
"loss": 0.0203,
|
| 5602 |
+
"step": 39950
|
| 5603 |
+
},
|
| 5604 |
+
{
|
| 5605 |
+
"epoch": 0.03451197479245361,
|
| 5606 |
+
"grad_norm": 1.9399341908865608e-05,
|
| 5607 |
+
"learning_rate": 1.9314759932960396e-05,
|
| 5608 |
+
"loss": 0.0273,
|
| 5609 |
+
"step": 40000
|
| 5610 |
}
|
| 5611 |
],
|
| 5612 |
"logging_steps": 50,
|