Upload folder using huggingface_hub
#4
by
CreatorPhan
- opened
- adapter_model.bin +1 -1
- optimizer.pt +1 -1
- rng_state.pth +1 -1
- scheduler.pt +1 -1
- trainer_state.json +1803 -3
adapter_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 39409357
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0bfdd727772683806731fba15bf8f4caf691af6184bc5b21e2ea5a4f390f41dc
|
| 3 |
size 39409357
|
optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 78844421
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c80d296d52ab4d4c11a0695940c0dadb792928e126cde3111312316dd3d33438
|
| 3 |
size 78844421
|
rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14575
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3e46ce4eb16240da9f3a8b3066acb6f59a234249ee2a3052f3323786da479838
|
| 3 |
size 14575
|
scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 627
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:64035578adb33297a89b3be176cbd0b7d7adb0f34d904267be99bb01d2a849d0
|
| 3 |
size 627
|
trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch":
|
| 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,
|
|
@@ -1207,13 +1207,1813 @@
|
|
| 1207 |
"learning_rate": 0.002642857142857143,
|
| 1208 |
"loss": 1.678,
|
| 1209 |
"step": 200
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1210 |
}
|
| 1211 |
],
|
| 1212 |
"logging_steps": 1,
|
| 1213 |
"max_steps": 1680,
|
| 1214 |
"num_train_epochs": 48,
|
| 1215 |
"save_steps": 100,
|
| 1216 |
-
"total_flos":
|
| 1217 |
"trial_name": null,
|
| 1218 |
"trial_params": null
|
| 1219 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 14.285714285714286,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 500,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 1207 |
"learning_rate": 0.002642857142857143,
|
| 1208 |
"loss": 1.678,
|
| 1209 |
"step": 200
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 5.74,
|
| 1213 |
+
"learning_rate": 0.0026410714285714284,
|
| 1214 |
+
"loss": 1.7613,
|
| 1215 |
+
"step": 201
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 5.77,
|
| 1219 |
+
"learning_rate": 0.0026392857142857142,
|
| 1220 |
+
"loss": 1.7541,
|
| 1221 |
+
"step": 202
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 5.8,
|
| 1225 |
+
"learning_rate": 0.0026375,
|
| 1226 |
+
"loss": 1.798,
|
| 1227 |
+
"step": 203
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 5.83,
|
| 1231 |
+
"learning_rate": 0.002635714285714286,
|
| 1232 |
+
"loss": 1.821,
|
| 1233 |
+
"step": 204
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 5.86,
|
| 1237 |
+
"learning_rate": 0.0026339285714285714,
|
| 1238 |
+
"loss": 1.8385,
|
| 1239 |
+
"step": 205
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 5.89,
|
| 1243 |
+
"learning_rate": 0.002632142857142857,
|
| 1244 |
+
"loss": 1.8613,
|
| 1245 |
+
"step": 206
|
| 1246 |
+
},
|
| 1247 |
+
{
|
| 1248 |
+
"epoch": 5.91,
|
| 1249 |
+
"learning_rate": 0.002630357142857143,
|
| 1250 |
+
"loss": 1.902,
|
| 1251 |
+
"step": 207
|
| 1252 |
+
},
|
| 1253 |
+
{
|
| 1254 |
+
"epoch": 5.94,
|
| 1255 |
+
"learning_rate": 0.0026285714285714285,
|
| 1256 |
+
"loss": 2.0848,
|
| 1257 |
+
"step": 208
|
| 1258 |
+
},
|
| 1259 |
+
{
|
| 1260 |
+
"epoch": 5.97,
|
| 1261 |
+
"learning_rate": 0.0026267857142857143,
|
| 1262 |
+
"loss": 2.3277,
|
| 1263 |
+
"step": 209
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 6.0,
|
| 1267 |
+
"learning_rate": 0.002625,
|
| 1268 |
+
"loss": 2.8535,
|
| 1269 |
+
"step": 210
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 6.03,
|
| 1273 |
+
"learning_rate": 0.0026232142857142856,
|
| 1274 |
+
"loss": 6.2197,
|
| 1275 |
+
"step": 211
|
| 1276 |
+
},
|
| 1277 |
+
{
|
| 1278 |
+
"epoch": 6.06,
|
| 1279 |
+
"learning_rate": 0.0026214285714285714,
|
| 1280 |
+
"loss": 10.2288,
|
| 1281 |
+
"step": 212
|
| 1282 |
+
},
|
| 1283 |
+
{
|
| 1284 |
+
"epoch": 6.09,
|
| 1285 |
+
"learning_rate": 0.0026196428571428573,
|
| 1286 |
+
"loss": 12.5006,
|
| 1287 |
+
"step": 213
|
| 1288 |
+
},
|
| 1289 |
+
{
|
| 1290 |
+
"epoch": 6.11,
|
| 1291 |
+
"learning_rate": 0.002617857142857143,
|
| 1292 |
+
"loss": 10.5184,
|
| 1293 |
+
"step": 214
|
| 1294 |
+
},
|
| 1295 |
+
{
|
| 1296 |
+
"epoch": 6.14,
|
| 1297 |
+
"learning_rate": 0.0026160714285714285,
|
| 1298 |
+
"loss": 9.4834,
|
| 1299 |
+
"step": 215
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 6.17,
|
| 1303 |
+
"learning_rate": 0.0026142857142857144,
|
| 1304 |
+
"loss": 16.0513,
|
| 1305 |
+
"step": 216
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 6.2,
|
| 1309 |
+
"learning_rate": 0.0026125000000000002,
|
| 1310 |
+
"loss": 11.0576,
|
| 1311 |
+
"step": 217
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 6.23,
|
| 1315 |
+
"learning_rate": 0.0026107142857142857,
|
| 1316 |
+
"loss": 15.3574,
|
| 1317 |
+
"step": 218
|
| 1318 |
+
},
|
| 1319 |
+
{
|
| 1320 |
+
"epoch": 6.26,
|
| 1321 |
+
"learning_rate": 0.0026089285714285715,
|
| 1322 |
+
"loss": 15.5239,
|
| 1323 |
+
"step": 219
|
| 1324 |
+
},
|
| 1325 |
+
{
|
| 1326 |
+
"epoch": 6.29,
|
| 1327 |
+
"learning_rate": 0.0026071428571428574,
|
| 1328 |
+
"loss": 15.3973,
|
| 1329 |
+
"step": 220
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 6.31,
|
| 1333 |
+
"learning_rate": 0.0026053571428571428,
|
| 1334 |
+
"loss": 12.059,
|
| 1335 |
+
"step": 221
|
| 1336 |
+
},
|
| 1337 |
+
{
|
| 1338 |
+
"epoch": 6.34,
|
| 1339 |
+
"learning_rate": 0.0026035714285714286,
|
| 1340 |
+
"loss": 10.8352,
|
| 1341 |
+
"step": 222
|
| 1342 |
+
},
|
| 1343 |
+
{
|
| 1344 |
+
"epoch": 6.37,
|
| 1345 |
+
"learning_rate": 0.0026017857142857145,
|
| 1346 |
+
"loss": 10.1507,
|
| 1347 |
+
"step": 223
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 6.4,
|
| 1351 |
+
"learning_rate": 0.0026000000000000003,
|
| 1352 |
+
"loss": 10.651,
|
| 1353 |
+
"step": 224
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 6.43,
|
| 1357 |
+
"learning_rate": 0.0025982142857142857,
|
| 1358 |
+
"loss": 9.8363,
|
| 1359 |
+
"step": 225
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 6.46,
|
| 1363 |
+
"learning_rate": 0.0025964285714285716,
|
| 1364 |
+
"loss": 9.3673,
|
| 1365 |
+
"step": 226
|
| 1366 |
+
},
|
| 1367 |
+
{
|
| 1368 |
+
"epoch": 6.49,
|
| 1369 |
+
"learning_rate": 0.0025946428571428574,
|
| 1370 |
+
"loss": 9.5433,
|
| 1371 |
+
"step": 227
|
| 1372 |
+
},
|
| 1373 |
+
{
|
| 1374 |
+
"epoch": 6.51,
|
| 1375 |
+
"learning_rate": 0.002592857142857143,
|
| 1376 |
+
"loss": 9.9206,
|
| 1377 |
+
"step": 228
|
| 1378 |
+
},
|
| 1379 |
+
{
|
| 1380 |
+
"epoch": 6.54,
|
| 1381 |
+
"learning_rate": 0.0025910714285714287,
|
| 1382 |
+
"loss": 9.5516,
|
| 1383 |
+
"step": 229
|
| 1384 |
+
},
|
| 1385 |
+
{
|
| 1386 |
+
"epoch": 6.57,
|
| 1387 |
+
"learning_rate": 0.0025892857142857145,
|
| 1388 |
+
"loss": 9.2165,
|
| 1389 |
+
"step": 230
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 6.6,
|
| 1393 |
+
"learning_rate": 0.0025875000000000004,
|
| 1394 |
+
"loss": 9.0825,
|
| 1395 |
+
"step": 231
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 6.63,
|
| 1399 |
+
"learning_rate": 0.002585714285714286,
|
| 1400 |
+
"loss": 8.7437,
|
| 1401 |
+
"step": 232
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 6.66,
|
| 1405 |
+
"learning_rate": 0.0025839285714285717,
|
| 1406 |
+
"loss": 8.6366,
|
| 1407 |
+
"step": 233
|
| 1408 |
+
},
|
| 1409 |
+
{
|
| 1410 |
+
"epoch": 6.69,
|
| 1411 |
+
"learning_rate": 0.0025821428571428575,
|
| 1412 |
+
"loss": 9.7431,
|
| 1413 |
+
"step": 234
|
| 1414 |
+
},
|
| 1415 |
+
{
|
| 1416 |
+
"epoch": 6.71,
|
| 1417 |
+
"learning_rate": 0.002580357142857143,
|
| 1418 |
+
"loss": 8.1876,
|
| 1419 |
+
"step": 235
|
| 1420 |
+
},
|
| 1421 |
+
{
|
| 1422 |
+
"epoch": 6.74,
|
| 1423 |
+
"learning_rate": 0.0025785714285714288,
|
| 1424 |
+
"loss": 8.4559,
|
| 1425 |
+
"step": 236
|
| 1426 |
+
},
|
| 1427 |
+
{
|
| 1428 |
+
"epoch": 6.77,
|
| 1429 |
+
"learning_rate": 0.002576785714285714,
|
| 1430 |
+
"loss": 8.0092,
|
| 1431 |
+
"step": 237
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 6.8,
|
| 1435 |
+
"learning_rate": 0.002575,
|
| 1436 |
+
"loss": 8.028,
|
| 1437 |
+
"step": 238
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 6.83,
|
| 1441 |
+
"learning_rate": 0.0025732142857142854,
|
| 1442 |
+
"loss": 7.8379,
|
| 1443 |
+
"step": 239
|
| 1444 |
+
},
|
| 1445 |
+
{
|
| 1446 |
+
"epoch": 6.86,
|
| 1447 |
+
"learning_rate": 0.0025714285714285713,
|
| 1448 |
+
"loss": 7.8127,
|
| 1449 |
+
"step": 240
|
| 1450 |
+
},
|
| 1451 |
+
{
|
| 1452 |
+
"epoch": 6.89,
|
| 1453 |
+
"learning_rate": 0.002569642857142857,
|
| 1454 |
+
"loss": 7.8252,
|
| 1455 |
+
"step": 241
|
| 1456 |
+
},
|
| 1457 |
+
{
|
| 1458 |
+
"epoch": 6.91,
|
| 1459 |
+
"learning_rate": 0.002567857142857143,
|
| 1460 |
+
"loss": 7.7094,
|
| 1461 |
+
"step": 242
|
| 1462 |
+
},
|
| 1463 |
+
{
|
| 1464 |
+
"epoch": 6.94,
|
| 1465 |
+
"learning_rate": 0.0025660714285714284,
|
| 1466 |
+
"loss": 7.7962,
|
| 1467 |
+
"step": 243
|
| 1468 |
+
},
|
| 1469 |
+
{
|
| 1470 |
+
"epoch": 6.97,
|
| 1471 |
+
"learning_rate": 0.0025642857142857143,
|
| 1472 |
+
"loss": 7.4966,
|
| 1473 |
+
"step": 244
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 7.0,
|
| 1477 |
+
"learning_rate": 0.0025625,
|
| 1478 |
+
"loss": 7.4851,
|
| 1479 |
+
"step": 245
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 7.03,
|
| 1483 |
+
"learning_rate": 0.0025607142857142855,
|
| 1484 |
+
"loss": 7.5188,
|
| 1485 |
+
"step": 246
|
| 1486 |
+
},
|
| 1487 |
+
{
|
| 1488 |
+
"epoch": 7.06,
|
| 1489 |
+
"learning_rate": 0.0025589285714285714,
|
| 1490 |
+
"loss": 7.7866,
|
| 1491 |
+
"step": 247
|
| 1492 |
+
},
|
| 1493 |
+
{
|
| 1494 |
+
"epoch": 7.09,
|
| 1495 |
+
"learning_rate": 0.0025571428571428572,
|
| 1496 |
+
"loss": 7.5743,
|
| 1497 |
+
"step": 248
|
| 1498 |
+
},
|
| 1499 |
+
{
|
| 1500 |
+
"epoch": 7.11,
|
| 1501 |
+
"learning_rate": 0.0025553571428571426,
|
| 1502 |
+
"loss": 7.4608,
|
| 1503 |
+
"step": 249
|
| 1504 |
+
},
|
| 1505 |
+
{
|
| 1506 |
+
"epoch": 7.14,
|
| 1507 |
+
"learning_rate": 0.0025535714285714285,
|
| 1508 |
+
"loss": 7.4655,
|
| 1509 |
+
"step": 250
|
| 1510 |
+
},
|
| 1511 |
+
{
|
| 1512 |
+
"epoch": 7.17,
|
| 1513 |
+
"learning_rate": 0.0025517857142857143,
|
| 1514 |
+
"loss": 7.5474,
|
| 1515 |
+
"step": 251
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 7.2,
|
| 1519 |
+
"learning_rate": 0.00255,
|
| 1520 |
+
"loss": 7.6983,
|
| 1521 |
+
"step": 252
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 7.23,
|
| 1525 |
+
"learning_rate": 0.0025482142857142856,
|
| 1526 |
+
"loss": 7.4936,
|
| 1527 |
+
"step": 253
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 7.26,
|
| 1531 |
+
"learning_rate": 0.0025464285714285714,
|
| 1532 |
+
"loss": 7.6966,
|
| 1533 |
+
"step": 254
|
| 1534 |
+
},
|
| 1535 |
+
{
|
| 1536 |
+
"epoch": 7.29,
|
| 1537 |
+
"learning_rate": 0.0025446428571428573,
|
| 1538 |
+
"loss": 7.4701,
|
| 1539 |
+
"step": 255
|
| 1540 |
+
},
|
| 1541 |
+
{
|
| 1542 |
+
"epoch": 7.31,
|
| 1543 |
+
"learning_rate": 0.0025428571428571427,
|
| 1544 |
+
"loss": 7.511,
|
| 1545 |
+
"step": 256
|
| 1546 |
+
},
|
| 1547 |
+
{
|
| 1548 |
+
"epoch": 7.34,
|
| 1549 |
+
"learning_rate": 0.0025410714285714286,
|
| 1550 |
+
"loss": 7.3709,
|
| 1551 |
+
"step": 257
|
| 1552 |
+
},
|
| 1553 |
+
{
|
| 1554 |
+
"epoch": 7.37,
|
| 1555 |
+
"learning_rate": 0.0025392857142857144,
|
| 1556 |
+
"loss": 7.4582,
|
| 1557 |
+
"step": 258
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 7.4,
|
| 1561 |
+
"learning_rate": 0.0025375,
|
| 1562 |
+
"loss": 7.4263,
|
| 1563 |
+
"step": 259
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 7.43,
|
| 1567 |
+
"learning_rate": 0.0025357142857142857,
|
| 1568 |
+
"loss": 7.3134,
|
| 1569 |
+
"step": 260
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 7.46,
|
| 1573 |
+
"learning_rate": 0.0025339285714285715,
|
| 1574 |
+
"loss": 7.3849,
|
| 1575 |
+
"step": 261
|
| 1576 |
+
},
|
| 1577 |
+
{
|
| 1578 |
+
"epoch": 7.49,
|
| 1579 |
+
"learning_rate": 0.0025321428571428574,
|
| 1580 |
+
"loss": 7.292,
|
| 1581 |
+
"step": 262
|
| 1582 |
+
},
|
| 1583 |
+
{
|
| 1584 |
+
"epoch": 7.51,
|
| 1585 |
+
"learning_rate": 0.002530357142857143,
|
| 1586 |
+
"loss": 7.343,
|
| 1587 |
+
"step": 263
|
| 1588 |
+
},
|
| 1589 |
+
{
|
| 1590 |
+
"epoch": 7.54,
|
| 1591 |
+
"learning_rate": 0.0025285714285714286,
|
| 1592 |
+
"loss": 7.3166,
|
| 1593 |
+
"step": 264
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 7.57,
|
| 1597 |
+
"learning_rate": 0.0025267857142857145,
|
| 1598 |
+
"loss": 7.2676,
|
| 1599 |
+
"step": 265
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 7.6,
|
| 1603 |
+
"learning_rate": 0.002525,
|
| 1604 |
+
"loss": 7.2955,
|
| 1605 |
+
"step": 266
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 7.63,
|
| 1609 |
+
"learning_rate": 0.0025232142857142857,
|
| 1610 |
+
"loss": 7.3386,
|
| 1611 |
+
"step": 267
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 7.66,
|
| 1615 |
+
"learning_rate": 0.0025214285714285716,
|
| 1616 |
+
"loss": 7.2682,
|
| 1617 |
+
"step": 268
|
| 1618 |
+
},
|
| 1619 |
+
{
|
| 1620 |
+
"epoch": 7.69,
|
| 1621 |
+
"learning_rate": 0.0025196428571428574,
|
| 1622 |
+
"loss": 7.2359,
|
| 1623 |
+
"step": 269
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 7.71,
|
| 1627 |
+
"learning_rate": 0.002517857142857143,
|
| 1628 |
+
"loss": 7.1849,
|
| 1629 |
+
"step": 270
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 7.74,
|
| 1633 |
+
"learning_rate": 0.0025160714285714287,
|
| 1634 |
+
"loss": 7.2421,
|
| 1635 |
+
"step": 271
|
| 1636 |
+
},
|
| 1637 |
+
{
|
| 1638 |
+
"epoch": 7.77,
|
| 1639 |
+
"learning_rate": 0.0025142857142857146,
|
| 1640 |
+
"loss": 7.2341,
|
| 1641 |
+
"step": 272
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 7.8,
|
| 1645 |
+
"learning_rate": 0.0025125,
|
| 1646 |
+
"loss": 7.2901,
|
| 1647 |
+
"step": 273
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 7.83,
|
| 1651 |
+
"learning_rate": 0.002510714285714286,
|
| 1652 |
+
"loss": 7.1931,
|
| 1653 |
+
"step": 274
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 7.86,
|
| 1657 |
+
"learning_rate": 0.0025089285714285717,
|
| 1658 |
+
"loss": 7.1907,
|
| 1659 |
+
"step": 275
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 7.89,
|
| 1663 |
+
"learning_rate": 0.002507142857142857,
|
| 1664 |
+
"loss": 7.2369,
|
| 1665 |
+
"step": 276
|
| 1666 |
+
},
|
| 1667 |
+
{
|
| 1668 |
+
"epoch": 7.91,
|
| 1669 |
+
"learning_rate": 0.002505357142857143,
|
| 1670 |
+
"loss": 7.1764,
|
| 1671 |
+
"step": 277
|
| 1672 |
+
},
|
| 1673 |
+
{
|
| 1674 |
+
"epoch": 7.94,
|
| 1675 |
+
"learning_rate": 0.002503571428571429,
|
| 1676 |
+
"loss": 7.1928,
|
| 1677 |
+
"step": 278
|
| 1678 |
+
},
|
| 1679 |
+
{
|
| 1680 |
+
"epoch": 7.97,
|
| 1681 |
+
"learning_rate": 0.0025017857142857146,
|
| 1682 |
+
"loss": 7.2114,
|
| 1683 |
+
"step": 279
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 8.0,
|
| 1687 |
+
"learning_rate": 0.0025,
|
| 1688 |
+
"loss": 7.2307,
|
| 1689 |
+
"step": 280
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 8.03,
|
| 1693 |
+
"learning_rate": 0.002498214285714286,
|
| 1694 |
+
"loss": 7.2477,
|
| 1695 |
+
"step": 281
|
| 1696 |
+
},
|
| 1697 |
+
{
|
| 1698 |
+
"epoch": 8.06,
|
| 1699 |
+
"learning_rate": 0.0024964285714285718,
|
| 1700 |
+
"loss": 7.2069,
|
| 1701 |
+
"step": 282
|
| 1702 |
+
},
|
| 1703 |
+
{
|
| 1704 |
+
"epoch": 8.09,
|
| 1705 |
+
"learning_rate": 0.002494642857142857,
|
| 1706 |
+
"loss": 7.1484,
|
| 1707 |
+
"step": 283
|
| 1708 |
+
},
|
| 1709 |
+
{
|
| 1710 |
+
"epoch": 8.11,
|
| 1711 |
+
"learning_rate": 0.002492857142857143,
|
| 1712 |
+
"loss": 7.1076,
|
| 1713 |
+
"step": 284
|
| 1714 |
+
},
|
| 1715 |
+
{
|
| 1716 |
+
"epoch": 8.14,
|
| 1717 |
+
"learning_rate": 0.002491071428571429,
|
| 1718 |
+
"loss": 7.0819,
|
| 1719 |
+
"step": 285
|
| 1720 |
+
},
|
| 1721 |
+
{
|
| 1722 |
+
"epoch": 8.17,
|
| 1723 |
+
"learning_rate": 0.0024892857142857143,
|
| 1724 |
+
"loss": 7.0708,
|
| 1725 |
+
"step": 286
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 8.2,
|
| 1729 |
+
"learning_rate": 0.0024875,
|
| 1730 |
+
"loss": 7.0763,
|
| 1731 |
+
"step": 287
|
| 1732 |
+
},
|
| 1733 |
+
{
|
| 1734 |
+
"epoch": 8.23,
|
| 1735 |
+
"learning_rate": 0.002485714285714286,
|
| 1736 |
+
"loss": 7.0792,
|
| 1737 |
+
"step": 288
|
| 1738 |
+
},
|
| 1739 |
+
{
|
| 1740 |
+
"epoch": 8.26,
|
| 1741 |
+
"learning_rate": 0.0024839285714285714,
|
| 1742 |
+
"loss": 7.1397,
|
| 1743 |
+
"step": 289
|
| 1744 |
+
},
|
| 1745 |
+
{
|
| 1746 |
+
"epoch": 8.29,
|
| 1747 |
+
"learning_rate": 0.0024821428571428572,
|
| 1748 |
+
"loss": 7.0893,
|
| 1749 |
+
"step": 290
|
| 1750 |
+
},
|
| 1751 |
+
{
|
| 1752 |
+
"epoch": 8.31,
|
| 1753 |
+
"learning_rate": 0.0024803571428571427,
|
| 1754 |
+
"loss": 7.1263,
|
| 1755 |
+
"step": 291
|
| 1756 |
+
},
|
| 1757 |
+
{
|
| 1758 |
+
"epoch": 8.34,
|
| 1759 |
+
"learning_rate": 0.0024785714285714285,
|
| 1760 |
+
"loss": 7.0226,
|
| 1761 |
+
"step": 292
|
| 1762 |
+
},
|
| 1763 |
+
{
|
| 1764 |
+
"epoch": 8.37,
|
| 1765 |
+
"learning_rate": 0.0024767857142857144,
|
| 1766 |
+
"loss": 7.1017,
|
| 1767 |
+
"step": 293
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 8.4,
|
| 1771 |
+
"learning_rate": 0.0024749999999999998,
|
| 1772 |
+
"loss": 7.0161,
|
| 1773 |
+
"step": 294
|
| 1774 |
+
},
|
| 1775 |
+
{
|
| 1776 |
+
"epoch": 8.43,
|
| 1777 |
+
"learning_rate": 0.0024732142857142856,
|
| 1778 |
+
"loss": 7.117,
|
| 1779 |
+
"step": 295
|
| 1780 |
+
},
|
| 1781 |
+
{
|
| 1782 |
+
"epoch": 8.46,
|
| 1783 |
+
"learning_rate": 0.0024714285714285715,
|
| 1784 |
+
"loss": 7.0234,
|
| 1785 |
+
"step": 296
|
| 1786 |
+
},
|
| 1787 |
+
{
|
| 1788 |
+
"epoch": 8.49,
|
| 1789 |
+
"learning_rate": 0.002469642857142857,
|
| 1790 |
+
"loss": 7.0663,
|
| 1791 |
+
"step": 297
|
| 1792 |
+
},
|
| 1793 |
+
{
|
| 1794 |
+
"epoch": 8.51,
|
| 1795 |
+
"learning_rate": 0.0024678571428571427,
|
| 1796 |
+
"loss": 7.1604,
|
| 1797 |
+
"step": 298
|
| 1798 |
+
},
|
| 1799 |
+
{
|
| 1800 |
+
"epoch": 8.54,
|
| 1801 |
+
"learning_rate": 0.0024660714285714286,
|
| 1802 |
+
"loss": 7.0543,
|
| 1803 |
+
"step": 299
|
| 1804 |
+
},
|
| 1805 |
+
{
|
| 1806 |
+
"epoch": 8.57,
|
| 1807 |
+
"learning_rate": 0.0024642857142857144,
|
| 1808 |
+
"loss": 7.0131,
|
| 1809 |
+
"step": 300
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 8.6,
|
| 1813 |
+
"learning_rate": 0.0024625,
|
| 1814 |
+
"loss": 7.0294,
|
| 1815 |
+
"step": 301
|
| 1816 |
+
},
|
| 1817 |
+
{
|
| 1818 |
+
"epoch": 8.63,
|
| 1819 |
+
"learning_rate": 0.0024607142857142857,
|
| 1820 |
+
"loss": 7.0273,
|
| 1821 |
+
"step": 302
|
| 1822 |
+
},
|
| 1823 |
+
{
|
| 1824 |
+
"epoch": 8.66,
|
| 1825 |
+
"learning_rate": 0.0024589285714285715,
|
| 1826 |
+
"loss": 7.0074,
|
| 1827 |
+
"step": 303
|
| 1828 |
+
},
|
| 1829 |
+
{
|
| 1830 |
+
"epoch": 8.69,
|
| 1831 |
+
"learning_rate": 0.002457142857142857,
|
| 1832 |
+
"loss": 6.9747,
|
| 1833 |
+
"step": 304
|
| 1834 |
+
},
|
| 1835 |
+
{
|
| 1836 |
+
"epoch": 8.71,
|
| 1837 |
+
"learning_rate": 0.002455357142857143,
|
| 1838 |
+
"loss": 7.0617,
|
| 1839 |
+
"step": 305
|
| 1840 |
+
},
|
| 1841 |
+
{
|
| 1842 |
+
"epoch": 8.74,
|
| 1843 |
+
"learning_rate": 0.0024535714285714287,
|
| 1844 |
+
"loss": 7.0907,
|
| 1845 |
+
"step": 306
|
| 1846 |
+
},
|
| 1847 |
+
{
|
| 1848 |
+
"epoch": 8.77,
|
| 1849 |
+
"learning_rate": 0.002451785714285714,
|
| 1850 |
+
"loss": 7.0037,
|
| 1851 |
+
"step": 307
|
| 1852 |
+
},
|
| 1853 |
+
{
|
| 1854 |
+
"epoch": 8.8,
|
| 1855 |
+
"learning_rate": 0.00245,
|
| 1856 |
+
"loss": 6.969,
|
| 1857 |
+
"step": 308
|
| 1858 |
+
},
|
| 1859 |
+
{
|
| 1860 |
+
"epoch": 8.83,
|
| 1861 |
+
"learning_rate": 0.0024482142857142858,
|
| 1862 |
+
"loss": 7.0575,
|
| 1863 |
+
"step": 309
|
| 1864 |
+
},
|
| 1865 |
+
{
|
| 1866 |
+
"epoch": 8.86,
|
| 1867 |
+
"learning_rate": 0.0024464285714285716,
|
| 1868 |
+
"loss": 6.9494,
|
| 1869 |
+
"step": 310
|
| 1870 |
+
},
|
| 1871 |
+
{
|
| 1872 |
+
"epoch": 8.89,
|
| 1873 |
+
"learning_rate": 0.002444642857142857,
|
| 1874 |
+
"loss": 6.969,
|
| 1875 |
+
"step": 311
|
| 1876 |
+
},
|
| 1877 |
+
{
|
| 1878 |
+
"epoch": 8.91,
|
| 1879 |
+
"learning_rate": 0.002442857142857143,
|
| 1880 |
+
"loss": 6.8827,
|
| 1881 |
+
"step": 312
|
| 1882 |
+
},
|
| 1883 |
+
{
|
| 1884 |
+
"epoch": 8.94,
|
| 1885 |
+
"learning_rate": 0.0024410714285714287,
|
| 1886 |
+
"loss": 6.9058,
|
| 1887 |
+
"step": 313
|
| 1888 |
+
},
|
| 1889 |
+
{
|
| 1890 |
+
"epoch": 8.97,
|
| 1891 |
+
"learning_rate": 0.002439285714285714,
|
| 1892 |
+
"loss": 6.8808,
|
| 1893 |
+
"step": 314
|
| 1894 |
+
},
|
| 1895 |
+
{
|
| 1896 |
+
"epoch": 9.0,
|
| 1897 |
+
"learning_rate": 0.0024375,
|
| 1898 |
+
"loss": 6.9516,
|
| 1899 |
+
"step": 315
|
| 1900 |
+
},
|
| 1901 |
+
{
|
| 1902 |
+
"epoch": 9.03,
|
| 1903 |
+
"learning_rate": 0.002435714285714286,
|
| 1904 |
+
"loss": 6.9132,
|
| 1905 |
+
"step": 316
|
| 1906 |
+
},
|
| 1907 |
+
{
|
| 1908 |
+
"epoch": 9.06,
|
| 1909 |
+
"learning_rate": 0.0024339285714285717,
|
| 1910 |
+
"loss": 6.9058,
|
| 1911 |
+
"step": 317
|
| 1912 |
+
},
|
| 1913 |
+
{
|
| 1914 |
+
"epoch": 9.09,
|
| 1915 |
+
"learning_rate": 0.002432142857142857,
|
| 1916 |
+
"loss": 6.9332,
|
| 1917 |
+
"step": 318
|
| 1918 |
+
},
|
| 1919 |
+
{
|
| 1920 |
+
"epoch": 9.11,
|
| 1921 |
+
"learning_rate": 0.002430357142857143,
|
| 1922 |
+
"loss": 6.9757,
|
| 1923 |
+
"step": 319
|
| 1924 |
+
},
|
| 1925 |
+
{
|
| 1926 |
+
"epoch": 9.14,
|
| 1927 |
+
"learning_rate": 0.002428571428571429,
|
| 1928 |
+
"loss": 6.8261,
|
| 1929 |
+
"step": 320
|
| 1930 |
+
},
|
| 1931 |
+
{
|
| 1932 |
+
"epoch": 9.17,
|
| 1933 |
+
"learning_rate": 0.0024267857142857142,
|
| 1934 |
+
"loss": 6.8571,
|
| 1935 |
+
"step": 321
|
| 1936 |
+
},
|
| 1937 |
+
{
|
| 1938 |
+
"epoch": 9.2,
|
| 1939 |
+
"learning_rate": 0.002425,
|
| 1940 |
+
"loss": 6.8435,
|
| 1941 |
+
"step": 322
|
| 1942 |
+
},
|
| 1943 |
+
{
|
| 1944 |
+
"epoch": 9.23,
|
| 1945 |
+
"learning_rate": 0.002423214285714286,
|
| 1946 |
+
"loss": 6.9033,
|
| 1947 |
+
"step": 323
|
| 1948 |
+
},
|
| 1949 |
+
{
|
| 1950 |
+
"epoch": 9.26,
|
| 1951 |
+
"learning_rate": 0.0024214285714285713,
|
| 1952 |
+
"loss": 6.8042,
|
| 1953 |
+
"step": 324
|
| 1954 |
+
},
|
| 1955 |
+
{
|
| 1956 |
+
"epoch": 9.29,
|
| 1957 |
+
"learning_rate": 0.002419642857142857,
|
| 1958 |
+
"loss": 6.8732,
|
| 1959 |
+
"step": 325
|
| 1960 |
+
},
|
| 1961 |
+
{
|
| 1962 |
+
"epoch": 9.31,
|
| 1963 |
+
"learning_rate": 0.002417857142857143,
|
| 1964 |
+
"loss": 6.752,
|
| 1965 |
+
"step": 326
|
| 1966 |
+
},
|
| 1967 |
+
{
|
| 1968 |
+
"epoch": 9.34,
|
| 1969 |
+
"learning_rate": 0.002416071428571429,
|
| 1970 |
+
"loss": 6.8016,
|
| 1971 |
+
"step": 327
|
| 1972 |
+
},
|
| 1973 |
+
{
|
| 1974 |
+
"epoch": 9.37,
|
| 1975 |
+
"learning_rate": 0.0024142857142857143,
|
| 1976 |
+
"loss": 6.8879,
|
| 1977 |
+
"step": 328
|
| 1978 |
+
},
|
| 1979 |
+
{
|
| 1980 |
+
"epoch": 9.4,
|
| 1981 |
+
"learning_rate": 0.0024125,
|
| 1982 |
+
"loss": 6.7643,
|
| 1983 |
+
"step": 329
|
| 1984 |
+
},
|
| 1985 |
+
{
|
| 1986 |
+
"epoch": 9.43,
|
| 1987 |
+
"learning_rate": 0.002410714285714286,
|
| 1988 |
+
"loss": 6.7084,
|
| 1989 |
+
"step": 330
|
| 1990 |
+
},
|
| 1991 |
+
{
|
| 1992 |
+
"epoch": 9.46,
|
| 1993 |
+
"learning_rate": 0.0024089285714285714,
|
| 1994 |
+
"loss": 6.8049,
|
| 1995 |
+
"step": 331
|
| 1996 |
+
},
|
| 1997 |
+
{
|
| 1998 |
+
"epoch": 9.49,
|
| 1999 |
+
"learning_rate": 0.0024071428571428573,
|
| 2000 |
+
"loss": 6.7925,
|
| 2001 |
+
"step": 332
|
| 2002 |
+
},
|
| 2003 |
+
{
|
| 2004 |
+
"epoch": 9.51,
|
| 2005 |
+
"learning_rate": 0.002405357142857143,
|
| 2006 |
+
"loss": 6.7289,
|
| 2007 |
+
"step": 333
|
| 2008 |
+
},
|
| 2009 |
+
{
|
| 2010 |
+
"epoch": 9.54,
|
| 2011 |
+
"learning_rate": 0.0024035714285714285,
|
| 2012 |
+
"loss": 6.7439,
|
| 2013 |
+
"step": 334
|
| 2014 |
+
},
|
| 2015 |
+
{
|
| 2016 |
+
"epoch": 9.57,
|
| 2017 |
+
"learning_rate": 0.0024017857142857144,
|
| 2018 |
+
"loss": 6.7119,
|
| 2019 |
+
"step": 335
|
| 2020 |
+
},
|
| 2021 |
+
{
|
| 2022 |
+
"epoch": 9.6,
|
| 2023 |
+
"learning_rate": 0.0024000000000000002,
|
| 2024 |
+
"loss": 6.7251,
|
| 2025 |
+
"step": 336
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 9.63,
|
| 2029 |
+
"learning_rate": 0.002398214285714286,
|
| 2030 |
+
"loss": 6.6659,
|
| 2031 |
+
"step": 337
|
| 2032 |
+
},
|
| 2033 |
+
{
|
| 2034 |
+
"epoch": 9.66,
|
| 2035 |
+
"learning_rate": 0.0023964285714285715,
|
| 2036 |
+
"loss": 6.7422,
|
| 2037 |
+
"step": 338
|
| 2038 |
+
},
|
| 2039 |
+
{
|
| 2040 |
+
"epoch": 9.69,
|
| 2041 |
+
"learning_rate": 0.0023946428571428573,
|
| 2042 |
+
"loss": 6.7852,
|
| 2043 |
+
"step": 339
|
| 2044 |
+
},
|
| 2045 |
+
{
|
| 2046 |
+
"epoch": 9.71,
|
| 2047 |
+
"learning_rate": 0.002392857142857143,
|
| 2048 |
+
"loss": 6.6828,
|
| 2049 |
+
"step": 340
|
| 2050 |
+
},
|
| 2051 |
+
{
|
| 2052 |
+
"epoch": 9.74,
|
| 2053 |
+
"learning_rate": 0.0023910714285714286,
|
| 2054 |
+
"loss": 6.686,
|
| 2055 |
+
"step": 341
|
| 2056 |
+
},
|
| 2057 |
+
{
|
| 2058 |
+
"epoch": 9.77,
|
| 2059 |
+
"learning_rate": 0.002389285714285714,
|
| 2060 |
+
"loss": 6.7326,
|
| 2061 |
+
"step": 342
|
| 2062 |
+
},
|
| 2063 |
+
{
|
| 2064 |
+
"epoch": 9.8,
|
| 2065 |
+
"learning_rate": 0.0023875,
|
| 2066 |
+
"loss": 6.5601,
|
| 2067 |
+
"step": 343
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 9.83,
|
| 2071 |
+
"learning_rate": 0.0023857142857142857,
|
| 2072 |
+
"loss": 6.6646,
|
| 2073 |
+
"step": 344
|
| 2074 |
+
},
|
| 2075 |
+
{
|
| 2076 |
+
"epoch": 9.86,
|
| 2077 |
+
"learning_rate": 0.002383928571428571,
|
| 2078 |
+
"loss": 6.5673,
|
| 2079 |
+
"step": 345
|
| 2080 |
+
},
|
| 2081 |
+
{
|
| 2082 |
+
"epoch": 9.89,
|
| 2083 |
+
"learning_rate": 0.002382142857142857,
|
| 2084 |
+
"loss": 6.6227,
|
| 2085 |
+
"step": 346
|
| 2086 |
+
},
|
| 2087 |
+
{
|
| 2088 |
+
"epoch": 9.91,
|
| 2089 |
+
"learning_rate": 0.002380357142857143,
|
| 2090 |
+
"loss": 6.5526,
|
| 2091 |
+
"step": 347
|
| 2092 |
+
},
|
| 2093 |
+
{
|
| 2094 |
+
"epoch": 9.94,
|
| 2095 |
+
"learning_rate": 0.0023785714285714287,
|
| 2096 |
+
"loss": 6.6842,
|
| 2097 |
+
"step": 348
|
| 2098 |
+
},
|
| 2099 |
+
{
|
| 2100 |
+
"epoch": 9.97,
|
| 2101 |
+
"learning_rate": 0.002376785714285714,
|
| 2102 |
+
"loss": 6.6211,
|
| 2103 |
+
"step": 349
|
| 2104 |
+
},
|
| 2105 |
+
{
|
| 2106 |
+
"epoch": 10.0,
|
| 2107 |
+
"learning_rate": 0.002375,
|
| 2108 |
+
"loss": 6.6952,
|
| 2109 |
+
"step": 350
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 10.03,
|
| 2113 |
+
"learning_rate": 0.002373214285714286,
|
| 2114 |
+
"loss": 6.5324,
|
| 2115 |
+
"step": 351
|
| 2116 |
+
},
|
| 2117 |
+
{
|
| 2118 |
+
"epoch": 10.06,
|
| 2119 |
+
"learning_rate": 0.002371428571428571,
|
| 2120 |
+
"loss": 6.5792,
|
| 2121 |
+
"step": 352
|
| 2122 |
+
},
|
| 2123 |
+
{
|
| 2124 |
+
"epoch": 10.09,
|
| 2125 |
+
"learning_rate": 0.002369642857142857,
|
| 2126 |
+
"loss": 6.5276,
|
| 2127 |
+
"step": 353
|
| 2128 |
+
},
|
| 2129 |
+
{
|
| 2130 |
+
"epoch": 10.11,
|
| 2131 |
+
"learning_rate": 0.002367857142857143,
|
| 2132 |
+
"loss": 6.5634,
|
| 2133 |
+
"step": 354
|
| 2134 |
+
},
|
| 2135 |
+
{
|
| 2136 |
+
"epoch": 10.14,
|
| 2137 |
+
"learning_rate": 0.0023660714285714288,
|
| 2138 |
+
"loss": 6.5385,
|
| 2139 |
+
"step": 355
|
| 2140 |
+
},
|
| 2141 |
+
{
|
| 2142 |
+
"epoch": 10.17,
|
| 2143 |
+
"learning_rate": 0.002364285714285714,
|
| 2144 |
+
"loss": 6.4516,
|
| 2145 |
+
"step": 356
|
| 2146 |
+
},
|
| 2147 |
+
{
|
| 2148 |
+
"epoch": 10.2,
|
| 2149 |
+
"learning_rate": 0.0023625,
|
| 2150 |
+
"loss": 6.5641,
|
| 2151 |
+
"step": 357
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 10.23,
|
| 2155 |
+
"learning_rate": 0.002360714285714286,
|
| 2156 |
+
"loss": 6.5001,
|
| 2157 |
+
"step": 358
|
| 2158 |
+
},
|
| 2159 |
+
{
|
| 2160 |
+
"epoch": 10.26,
|
| 2161 |
+
"learning_rate": 0.0023589285714285713,
|
| 2162 |
+
"loss": 6.4846,
|
| 2163 |
+
"step": 359
|
| 2164 |
+
},
|
| 2165 |
+
{
|
| 2166 |
+
"epoch": 10.29,
|
| 2167 |
+
"learning_rate": 0.002357142857142857,
|
| 2168 |
+
"loss": 6.4638,
|
| 2169 |
+
"step": 360
|
| 2170 |
+
},
|
| 2171 |
+
{
|
| 2172 |
+
"epoch": 10.31,
|
| 2173 |
+
"learning_rate": 0.002355357142857143,
|
| 2174 |
+
"loss": 6.5217,
|
| 2175 |
+
"step": 361
|
| 2176 |
+
},
|
| 2177 |
+
{
|
| 2178 |
+
"epoch": 10.34,
|
| 2179 |
+
"learning_rate": 0.0023535714285714284,
|
| 2180 |
+
"loss": 6.5444,
|
| 2181 |
+
"step": 362
|
| 2182 |
+
},
|
| 2183 |
+
{
|
| 2184 |
+
"epoch": 10.37,
|
| 2185 |
+
"learning_rate": 0.0023517857142857142,
|
| 2186 |
+
"loss": 6.496,
|
| 2187 |
+
"step": 363
|
| 2188 |
+
},
|
| 2189 |
+
{
|
| 2190 |
+
"epoch": 10.4,
|
| 2191 |
+
"learning_rate": 0.00235,
|
| 2192 |
+
"loss": 6.5345,
|
| 2193 |
+
"step": 364
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"epoch": 10.43,
|
| 2197 |
+
"learning_rate": 0.002348214285714286,
|
| 2198 |
+
"loss": 6.4732,
|
| 2199 |
+
"step": 365
|
| 2200 |
+
},
|
| 2201 |
+
{
|
| 2202 |
+
"epoch": 10.46,
|
| 2203 |
+
"learning_rate": 0.0023464285714285714,
|
| 2204 |
+
"loss": 6.4765,
|
| 2205 |
+
"step": 366
|
| 2206 |
+
},
|
| 2207 |
+
{
|
| 2208 |
+
"epoch": 10.49,
|
| 2209 |
+
"learning_rate": 0.002344642857142857,
|
| 2210 |
+
"loss": 6.3881,
|
| 2211 |
+
"step": 367
|
| 2212 |
+
},
|
| 2213 |
+
{
|
| 2214 |
+
"epoch": 10.51,
|
| 2215 |
+
"learning_rate": 0.002342857142857143,
|
| 2216 |
+
"loss": 6.4908,
|
| 2217 |
+
"step": 368
|
| 2218 |
+
},
|
| 2219 |
+
{
|
| 2220 |
+
"epoch": 10.54,
|
| 2221 |
+
"learning_rate": 0.0023410714285714285,
|
| 2222 |
+
"loss": 6.4593,
|
| 2223 |
+
"step": 369
|
| 2224 |
+
},
|
| 2225 |
+
{
|
| 2226 |
+
"epoch": 10.57,
|
| 2227 |
+
"learning_rate": 0.0023392857142857143,
|
| 2228 |
+
"loss": 6.5006,
|
| 2229 |
+
"step": 370
|
| 2230 |
+
},
|
| 2231 |
+
{
|
| 2232 |
+
"epoch": 10.6,
|
| 2233 |
+
"learning_rate": 0.0023375,
|
| 2234 |
+
"loss": 6.4495,
|
| 2235 |
+
"step": 371
|
| 2236 |
+
},
|
| 2237 |
+
{
|
| 2238 |
+
"epoch": 10.63,
|
| 2239 |
+
"learning_rate": 0.0023357142857142856,
|
| 2240 |
+
"loss": 6.3569,
|
| 2241 |
+
"step": 372
|
| 2242 |
+
},
|
| 2243 |
+
{
|
| 2244 |
+
"epoch": 10.66,
|
| 2245 |
+
"learning_rate": 0.0023339285714285714,
|
| 2246 |
+
"loss": 6.3592,
|
| 2247 |
+
"step": 373
|
| 2248 |
+
},
|
| 2249 |
+
{
|
| 2250 |
+
"epoch": 10.69,
|
| 2251 |
+
"learning_rate": 0.0023321428571428573,
|
| 2252 |
+
"loss": 6.3258,
|
| 2253 |
+
"step": 374
|
| 2254 |
+
},
|
| 2255 |
+
{
|
| 2256 |
+
"epoch": 10.71,
|
| 2257 |
+
"learning_rate": 0.002330357142857143,
|
| 2258 |
+
"loss": 6.3216,
|
| 2259 |
+
"step": 375
|
| 2260 |
+
},
|
| 2261 |
+
{
|
| 2262 |
+
"epoch": 10.74,
|
| 2263 |
+
"learning_rate": 0.0023285714285714285,
|
| 2264 |
+
"loss": 6.4878,
|
| 2265 |
+
"step": 376
|
| 2266 |
+
},
|
| 2267 |
+
{
|
| 2268 |
+
"epoch": 10.77,
|
| 2269 |
+
"learning_rate": 0.0023267857142857144,
|
| 2270 |
+
"loss": 6.3412,
|
| 2271 |
+
"step": 377
|
| 2272 |
+
},
|
| 2273 |
+
{
|
| 2274 |
+
"epoch": 10.8,
|
| 2275 |
+
"learning_rate": 0.0023250000000000002,
|
| 2276 |
+
"loss": 6.3925,
|
| 2277 |
+
"step": 378
|
| 2278 |
+
},
|
| 2279 |
+
{
|
| 2280 |
+
"epoch": 10.83,
|
| 2281 |
+
"learning_rate": 0.0023232142857142857,
|
| 2282 |
+
"loss": 6.275,
|
| 2283 |
+
"step": 379
|
| 2284 |
+
},
|
| 2285 |
+
{
|
| 2286 |
+
"epoch": 10.86,
|
| 2287 |
+
"learning_rate": 0.0023214285714285715,
|
| 2288 |
+
"loss": 6.3575,
|
| 2289 |
+
"step": 380
|
| 2290 |
+
},
|
| 2291 |
+
{
|
| 2292 |
+
"epoch": 10.89,
|
| 2293 |
+
"learning_rate": 0.0023196428571428574,
|
| 2294 |
+
"loss": 6.3259,
|
| 2295 |
+
"step": 381
|
| 2296 |
+
},
|
| 2297 |
+
{
|
| 2298 |
+
"epoch": 10.91,
|
| 2299 |
+
"learning_rate": 0.002317857142857143,
|
| 2300 |
+
"loss": 6.315,
|
| 2301 |
+
"step": 382
|
| 2302 |
+
},
|
| 2303 |
+
{
|
| 2304 |
+
"epoch": 10.94,
|
| 2305 |
+
"learning_rate": 0.0023160714285714286,
|
| 2306 |
+
"loss": 6.277,
|
| 2307 |
+
"step": 383
|
| 2308 |
+
},
|
| 2309 |
+
{
|
| 2310 |
+
"epoch": 10.97,
|
| 2311 |
+
"learning_rate": 0.0023142857142857145,
|
| 2312 |
+
"loss": 6.3259,
|
| 2313 |
+
"step": 384
|
| 2314 |
+
},
|
| 2315 |
+
{
|
| 2316 |
+
"epoch": 11.0,
|
| 2317 |
+
"learning_rate": 0.0023125000000000003,
|
| 2318 |
+
"loss": 6.3747,
|
| 2319 |
+
"step": 385
|
| 2320 |
+
},
|
| 2321 |
+
{
|
| 2322 |
+
"epoch": 11.03,
|
| 2323 |
+
"learning_rate": 0.0023107142857142857,
|
| 2324 |
+
"loss": 6.3646,
|
| 2325 |
+
"step": 386
|
| 2326 |
+
},
|
| 2327 |
+
{
|
| 2328 |
+
"epoch": 11.06,
|
| 2329 |
+
"learning_rate": 0.0023089285714285716,
|
| 2330 |
+
"loss": 6.3687,
|
| 2331 |
+
"step": 387
|
| 2332 |
+
},
|
| 2333 |
+
{
|
| 2334 |
+
"epoch": 11.09,
|
| 2335 |
+
"learning_rate": 0.0023071428571428574,
|
| 2336 |
+
"loss": 6.3374,
|
| 2337 |
+
"step": 388
|
| 2338 |
+
},
|
| 2339 |
+
{
|
| 2340 |
+
"epoch": 11.11,
|
| 2341 |
+
"learning_rate": 0.002305357142857143,
|
| 2342 |
+
"loss": 6.3129,
|
| 2343 |
+
"step": 389
|
| 2344 |
+
},
|
| 2345 |
+
{
|
| 2346 |
+
"epoch": 11.14,
|
| 2347 |
+
"learning_rate": 0.0023035714285714287,
|
| 2348 |
+
"loss": 6.3425,
|
| 2349 |
+
"step": 390
|
| 2350 |
+
},
|
| 2351 |
+
{
|
| 2352 |
+
"epoch": 11.17,
|
| 2353 |
+
"learning_rate": 0.0023017857142857145,
|
| 2354 |
+
"loss": 6.2122,
|
| 2355 |
+
"step": 391
|
| 2356 |
+
},
|
| 2357 |
+
{
|
| 2358 |
+
"epoch": 11.2,
|
| 2359 |
+
"learning_rate": 0.0023000000000000004,
|
| 2360 |
+
"loss": 6.2768,
|
| 2361 |
+
"step": 392
|
| 2362 |
+
},
|
| 2363 |
+
{
|
| 2364 |
+
"epoch": 11.23,
|
| 2365 |
+
"learning_rate": 0.002298214285714286,
|
| 2366 |
+
"loss": 6.2853,
|
| 2367 |
+
"step": 393
|
| 2368 |
+
},
|
| 2369 |
+
{
|
| 2370 |
+
"epoch": 11.26,
|
| 2371 |
+
"learning_rate": 0.0022964285714285712,
|
| 2372 |
+
"loss": 6.3215,
|
| 2373 |
+
"step": 394
|
| 2374 |
+
},
|
| 2375 |
+
{
|
| 2376 |
+
"epoch": 11.29,
|
| 2377 |
+
"learning_rate": 0.002294642857142857,
|
| 2378 |
+
"loss": 6.3244,
|
| 2379 |
+
"step": 395
|
| 2380 |
+
},
|
| 2381 |
+
{
|
| 2382 |
+
"epoch": 11.31,
|
| 2383 |
+
"learning_rate": 0.002292857142857143,
|
| 2384 |
+
"loss": 6.2399,
|
| 2385 |
+
"step": 396
|
| 2386 |
+
},
|
| 2387 |
+
{
|
| 2388 |
+
"epoch": 11.34,
|
| 2389 |
+
"learning_rate": 0.0022910714285714283,
|
| 2390 |
+
"loss": 6.2457,
|
| 2391 |
+
"step": 397
|
| 2392 |
+
},
|
| 2393 |
+
{
|
| 2394 |
+
"epoch": 11.37,
|
| 2395 |
+
"learning_rate": 0.002289285714285714,
|
| 2396 |
+
"loss": 6.2018,
|
| 2397 |
+
"step": 398
|
| 2398 |
+
},
|
| 2399 |
+
{
|
| 2400 |
+
"epoch": 11.4,
|
| 2401 |
+
"learning_rate": 0.0022875,
|
| 2402 |
+
"loss": 6.2101,
|
| 2403 |
+
"step": 399
|
| 2404 |
+
},
|
| 2405 |
+
{
|
| 2406 |
+
"epoch": 11.43,
|
| 2407 |
+
"learning_rate": 0.0022857142857142855,
|
| 2408 |
+
"loss": 6.2257,
|
| 2409 |
+
"step": 400
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 11.46,
|
| 2413 |
+
"learning_rate": 0.0022839285714285713,
|
| 2414 |
+
"loss": 6.3029,
|
| 2415 |
+
"step": 401
|
| 2416 |
+
},
|
| 2417 |
+
{
|
| 2418 |
+
"epoch": 11.49,
|
| 2419 |
+
"learning_rate": 0.002282142857142857,
|
| 2420 |
+
"loss": 6.2312,
|
| 2421 |
+
"step": 402
|
| 2422 |
+
},
|
| 2423 |
+
{
|
| 2424 |
+
"epoch": 11.51,
|
| 2425 |
+
"learning_rate": 0.002280357142857143,
|
| 2426 |
+
"loss": 6.203,
|
| 2427 |
+
"step": 403
|
| 2428 |
+
},
|
| 2429 |
+
{
|
| 2430 |
+
"epoch": 11.54,
|
| 2431 |
+
"learning_rate": 0.0022785714285714284,
|
| 2432 |
+
"loss": 6.2881,
|
| 2433 |
+
"step": 404
|
| 2434 |
+
},
|
| 2435 |
+
{
|
| 2436 |
+
"epoch": 11.57,
|
| 2437 |
+
"learning_rate": 0.0022767857142857143,
|
| 2438 |
+
"loss": 6.3466,
|
| 2439 |
+
"step": 405
|
| 2440 |
+
},
|
| 2441 |
+
{
|
| 2442 |
+
"epoch": 11.6,
|
| 2443 |
+
"learning_rate": 0.002275,
|
| 2444 |
+
"loss": 6.1908,
|
| 2445 |
+
"step": 406
|
| 2446 |
+
},
|
| 2447 |
+
{
|
| 2448 |
+
"epoch": 11.63,
|
| 2449 |
+
"learning_rate": 0.0022732142857142855,
|
| 2450 |
+
"loss": 6.196,
|
| 2451 |
+
"step": 407
|
| 2452 |
+
},
|
| 2453 |
+
{
|
| 2454 |
+
"epoch": 11.66,
|
| 2455 |
+
"learning_rate": 0.0022714285714285714,
|
| 2456 |
+
"loss": 6.1726,
|
| 2457 |
+
"step": 408
|
| 2458 |
+
},
|
| 2459 |
+
{
|
| 2460 |
+
"epoch": 11.69,
|
| 2461 |
+
"learning_rate": 0.0022696428571428572,
|
| 2462 |
+
"loss": 6.1207,
|
| 2463 |
+
"step": 409
|
| 2464 |
+
},
|
| 2465 |
+
{
|
| 2466 |
+
"epoch": 11.71,
|
| 2467 |
+
"learning_rate": 0.0022678571428571426,
|
| 2468 |
+
"loss": 6.2382,
|
| 2469 |
+
"step": 410
|
| 2470 |
+
},
|
| 2471 |
+
{
|
| 2472 |
+
"epoch": 11.74,
|
| 2473 |
+
"learning_rate": 0.0022660714285714285,
|
| 2474 |
+
"loss": 6.1757,
|
| 2475 |
+
"step": 411
|
| 2476 |
+
},
|
| 2477 |
+
{
|
| 2478 |
+
"epoch": 11.77,
|
| 2479 |
+
"learning_rate": 0.0022642857142857143,
|
| 2480 |
+
"loss": 6.1153,
|
| 2481 |
+
"step": 412
|
| 2482 |
+
},
|
| 2483 |
+
{
|
| 2484 |
+
"epoch": 11.8,
|
| 2485 |
+
"learning_rate": 0.0022625,
|
| 2486 |
+
"loss": 6.1261,
|
| 2487 |
+
"step": 413
|
| 2488 |
+
},
|
| 2489 |
+
{
|
| 2490 |
+
"epoch": 11.83,
|
| 2491 |
+
"learning_rate": 0.0022607142857142856,
|
| 2492 |
+
"loss": 6.0762,
|
| 2493 |
+
"step": 414
|
| 2494 |
+
},
|
| 2495 |
+
{
|
| 2496 |
+
"epoch": 11.86,
|
| 2497 |
+
"learning_rate": 0.0022589285714285715,
|
| 2498 |
+
"loss": 6.1386,
|
| 2499 |
+
"step": 415
|
| 2500 |
+
},
|
| 2501 |
+
{
|
| 2502 |
+
"epoch": 11.89,
|
| 2503 |
+
"learning_rate": 0.0022571428571428573,
|
| 2504 |
+
"loss": 6.1204,
|
| 2505 |
+
"step": 416
|
| 2506 |
+
},
|
| 2507 |
+
{
|
| 2508 |
+
"epoch": 11.91,
|
| 2509 |
+
"learning_rate": 0.0022553571428571427,
|
| 2510 |
+
"loss": 6.1059,
|
| 2511 |
+
"step": 417
|
| 2512 |
+
},
|
| 2513 |
+
{
|
| 2514 |
+
"epoch": 11.94,
|
| 2515 |
+
"learning_rate": 0.0022535714285714286,
|
| 2516 |
+
"loss": 6.0591,
|
| 2517 |
+
"step": 418
|
| 2518 |
+
},
|
| 2519 |
+
{
|
| 2520 |
+
"epoch": 11.97,
|
| 2521 |
+
"learning_rate": 0.0022517857142857144,
|
| 2522 |
+
"loss": 6.1713,
|
| 2523 |
+
"step": 419
|
| 2524 |
+
},
|
| 2525 |
+
{
|
| 2526 |
+
"epoch": 12.0,
|
| 2527 |
+
"learning_rate": 0.0022500000000000003,
|
| 2528 |
+
"loss": 6.2039,
|
| 2529 |
+
"step": 420
|
| 2530 |
+
},
|
| 2531 |
+
{
|
| 2532 |
+
"epoch": 12.03,
|
| 2533 |
+
"learning_rate": 0.0022482142857142857,
|
| 2534 |
+
"loss": 6.0168,
|
| 2535 |
+
"step": 421
|
| 2536 |
+
},
|
| 2537 |
+
{
|
| 2538 |
+
"epoch": 12.06,
|
| 2539 |
+
"learning_rate": 0.0022464285714285715,
|
| 2540 |
+
"loss": 6.0206,
|
| 2541 |
+
"step": 422
|
| 2542 |
+
},
|
| 2543 |
+
{
|
| 2544 |
+
"epoch": 12.09,
|
| 2545 |
+
"learning_rate": 0.0022446428571428574,
|
| 2546 |
+
"loss": 6.0642,
|
| 2547 |
+
"step": 423
|
| 2548 |
+
},
|
| 2549 |
+
{
|
| 2550 |
+
"epoch": 12.11,
|
| 2551 |
+
"learning_rate": 0.002242857142857143,
|
| 2552 |
+
"loss": 6.0665,
|
| 2553 |
+
"step": 424
|
| 2554 |
+
},
|
| 2555 |
+
{
|
| 2556 |
+
"epoch": 12.14,
|
| 2557 |
+
"learning_rate": 0.0022410714285714286,
|
| 2558 |
+
"loss": 5.9766,
|
| 2559 |
+
"step": 425
|
| 2560 |
+
},
|
| 2561 |
+
{
|
| 2562 |
+
"epoch": 12.17,
|
| 2563 |
+
"learning_rate": 0.0022392857142857145,
|
| 2564 |
+
"loss": 6.2167,
|
| 2565 |
+
"step": 426
|
| 2566 |
+
},
|
| 2567 |
+
{
|
| 2568 |
+
"epoch": 12.2,
|
| 2569 |
+
"learning_rate": 0.0022375,
|
| 2570 |
+
"loss": 6.002,
|
| 2571 |
+
"step": 427
|
| 2572 |
+
},
|
| 2573 |
+
{
|
| 2574 |
+
"epoch": 12.23,
|
| 2575 |
+
"learning_rate": 0.0022357142857142858,
|
| 2576 |
+
"loss": 6.0266,
|
| 2577 |
+
"step": 428
|
| 2578 |
+
},
|
| 2579 |
+
{
|
| 2580 |
+
"epoch": 12.26,
|
| 2581 |
+
"learning_rate": 0.0022339285714285716,
|
| 2582 |
+
"loss": 5.9339,
|
| 2583 |
+
"step": 429
|
| 2584 |
+
},
|
| 2585 |
+
{
|
| 2586 |
+
"epoch": 12.29,
|
| 2587 |
+
"learning_rate": 0.0022321428571428575,
|
| 2588 |
+
"loss": 6.1066,
|
| 2589 |
+
"step": 430
|
| 2590 |
+
},
|
| 2591 |
+
{
|
| 2592 |
+
"epoch": 12.31,
|
| 2593 |
+
"learning_rate": 0.002230357142857143,
|
| 2594 |
+
"loss": 5.9262,
|
| 2595 |
+
"step": 431
|
| 2596 |
+
},
|
| 2597 |
+
{
|
| 2598 |
+
"epoch": 12.34,
|
| 2599 |
+
"learning_rate": 0.0022285714285714287,
|
| 2600 |
+
"loss": 6.0696,
|
| 2601 |
+
"step": 432
|
| 2602 |
+
},
|
| 2603 |
+
{
|
| 2604 |
+
"epoch": 12.37,
|
| 2605 |
+
"learning_rate": 0.0022267857142857146,
|
| 2606 |
+
"loss": 5.9181,
|
| 2607 |
+
"step": 433
|
| 2608 |
+
},
|
| 2609 |
+
{
|
| 2610 |
+
"epoch": 12.4,
|
| 2611 |
+
"learning_rate": 0.002225,
|
| 2612 |
+
"loss": 6.0291,
|
| 2613 |
+
"step": 434
|
| 2614 |
+
},
|
| 2615 |
+
{
|
| 2616 |
+
"epoch": 12.43,
|
| 2617 |
+
"learning_rate": 0.002223214285714286,
|
| 2618 |
+
"loss": 5.9493,
|
| 2619 |
+
"step": 435
|
| 2620 |
+
},
|
| 2621 |
+
{
|
| 2622 |
+
"epoch": 12.46,
|
| 2623 |
+
"learning_rate": 0.0022214285714285717,
|
| 2624 |
+
"loss": 5.9639,
|
| 2625 |
+
"step": 436
|
| 2626 |
+
},
|
| 2627 |
+
{
|
| 2628 |
+
"epoch": 12.49,
|
| 2629 |
+
"learning_rate": 0.002219642857142857,
|
| 2630 |
+
"loss": 6.0303,
|
| 2631 |
+
"step": 437
|
| 2632 |
+
},
|
| 2633 |
+
{
|
| 2634 |
+
"epoch": 12.51,
|
| 2635 |
+
"learning_rate": 0.002217857142857143,
|
| 2636 |
+
"loss": 6.0157,
|
| 2637 |
+
"step": 438
|
| 2638 |
+
},
|
| 2639 |
+
{
|
| 2640 |
+
"epoch": 12.54,
|
| 2641 |
+
"learning_rate": 0.002216071428571429,
|
| 2642 |
+
"loss": 5.9309,
|
| 2643 |
+
"step": 439
|
| 2644 |
+
},
|
| 2645 |
+
{
|
| 2646 |
+
"epoch": 12.57,
|
| 2647 |
+
"learning_rate": 0.0022142857142857146,
|
| 2648 |
+
"loss": 5.9554,
|
| 2649 |
+
"step": 440
|
| 2650 |
+
},
|
| 2651 |
+
{
|
| 2652 |
+
"epoch": 12.6,
|
| 2653 |
+
"learning_rate": 0.0022125,
|
| 2654 |
+
"loss": 5.9761,
|
| 2655 |
+
"step": 441
|
| 2656 |
+
},
|
| 2657 |
+
{
|
| 2658 |
+
"epoch": 12.63,
|
| 2659 |
+
"learning_rate": 0.002210714285714286,
|
| 2660 |
+
"loss": 5.9042,
|
| 2661 |
+
"step": 442
|
| 2662 |
+
},
|
| 2663 |
+
{
|
| 2664 |
+
"epoch": 12.66,
|
| 2665 |
+
"learning_rate": 0.0022089285714285718,
|
| 2666 |
+
"loss": 6.0009,
|
| 2667 |
+
"step": 443
|
| 2668 |
+
},
|
| 2669 |
+
{
|
| 2670 |
+
"epoch": 12.69,
|
| 2671 |
+
"learning_rate": 0.002207142857142857,
|
| 2672 |
+
"loss": 5.9199,
|
| 2673 |
+
"step": 444
|
| 2674 |
+
},
|
| 2675 |
+
{
|
| 2676 |
+
"epoch": 12.71,
|
| 2677 |
+
"learning_rate": 0.002205357142857143,
|
| 2678 |
+
"loss": 5.9472,
|
| 2679 |
+
"step": 445
|
| 2680 |
+
},
|
| 2681 |
+
{
|
| 2682 |
+
"epoch": 12.74,
|
| 2683 |
+
"learning_rate": 0.002203571428571429,
|
| 2684 |
+
"loss": 6.0478,
|
| 2685 |
+
"step": 446
|
| 2686 |
+
},
|
| 2687 |
+
{
|
| 2688 |
+
"epoch": 12.77,
|
| 2689 |
+
"learning_rate": 0.0022017857142857143,
|
| 2690 |
+
"loss": 6.0131,
|
| 2691 |
+
"step": 447
|
| 2692 |
+
},
|
| 2693 |
+
{
|
| 2694 |
+
"epoch": 12.8,
|
| 2695 |
+
"learning_rate": 0.0021999999999999997,
|
| 2696 |
+
"loss": 5.9161,
|
| 2697 |
+
"step": 448
|
| 2698 |
+
},
|
| 2699 |
+
{
|
| 2700 |
+
"epoch": 12.83,
|
| 2701 |
+
"learning_rate": 0.0021982142857142855,
|
| 2702 |
+
"loss": 5.935,
|
| 2703 |
+
"step": 449
|
| 2704 |
+
},
|
| 2705 |
+
{
|
| 2706 |
+
"epoch": 12.86,
|
| 2707 |
+
"learning_rate": 0.0021964285714285714,
|
| 2708 |
+
"loss": 5.9035,
|
| 2709 |
+
"step": 450
|
| 2710 |
+
},
|
| 2711 |
+
{
|
| 2712 |
+
"epoch": 12.89,
|
| 2713 |
+
"learning_rate": 0.0021946428571428572,
|
| 2714 |
+
"loss": 5.9422,
|
| 2715 |
+
"step": 451
|
| 2716 |
+
},
|
| 2717 |
+
{
|
| 2718 |
+
"epoch": 12.91,
|
| 2719 |
+
"learning_rate": 0.0021928571428571427,
|
| 2720 |
+
"loss": 6.0135,
|
| 2721 |
+
"step": 452
|
| 2722 |
+
},
|
| 2723 |
+
{
|
| 2724 |
+
"epoch": 12.94,
|
| 2725 |
+
"learning_rate": 0.0021910714285714285,
|
| 2726 |
+
"loss": 5.9757,
|
| 2727 |
+
"step": 453
|
| 2728 |
+
},
|
| 2729 |
+
{
|
| 2730 |
+
"epoch": 12.97,
|
| 2731 |
+
"learning_rate": 0.0021892857142857144,
|
| 2732 |
+
"loss": 5.942,
|
| 2733 |
+
"step": 454
|
| 2734 |
+
},
|
| 2735 |
+
{
|
| 2736 |
+
"epoch": 13.0,
|
| 2737 |
+
"learning_rate": 0.0021874999999999998,
|
| 2738 |
+
"loss": 5.943,
|
| 2739 |
+
"step": 455
|
| 2740 |
+
},
|
| 2741 |
+
{
|
| 2742 |
+
"epoch": 13.03,
|
| 2743 |
+
"learning_rate": 0.0021857142857142856,
|
| 2744 |
+
"loss": 5.8982,
|
| 2745 |
+
"step": 456
|
| 2746 |
+
},
|
| 2747 |
+
{
|
| 2748 |
+
"epoch": 13.06,
|
| 2749 |
+
"learning_rate": 0.0021839285714285715,
|
| 2750 |
+
"loss": 5.9874,
|
| 2751 |
+
"step": 457
|
| 2752 |
+
},
|
| 2753 |
+
{
|
| 2754 |
+
"epoch": 13.09,
|
| 2755 |
+
"learning_rate": 0.002182142857142857,
|
| 2756 |
+
"loss": 5.8677,
|
| 2757 |
+
"step": 458
|
| 2758 |
+
},
|
| 2759 |
+
{
|
| 2760 |
+
"epoch": 13.11,
|
| 2761 |
+
"learning_rate": 0.0021803571428571427,
|
| 2762 |
+
"loss": 5.8782,
|
| 2763 |
+
"step": 459
|
| 2764 |
+
},
|
| 2765 |
+
{
|
| 2766 |
+
"epoch": 13.14,
|
| 2767 |
+
"learning_rate": 0.0021785714285714286,
|
| 2768 |
+
"loss": 5.787,
|
| 2769 |
+
"step": 460
|
| 2770 |
+
},
|
| 2771 |
+
{
|
| 2772 |
+
"epoch": 13.17,
|
| 2773 |
+
"learning_rate": 0.0021767857142857144,
|
| 2774 |
+
"loss": 5.8339,
|
| 2775 |
+
"step": 461
|
| 2776 |
+
},
|
| 2777 |
+
{
|
| 2778 |
+
"epoch": 13.2,
|
| 2779 |
+
"learning_rate": 0.002175,
|
| 2780 |
+
"loss": 5.8303,
|
| 2781 |
+
"step": 462
|
| 2782 |
+
},
|
| 2783 |
+
{
|
| 2784 |
+
"epoch": 13.23,
|
| 2785 |
+
"learning_rate": 0.0021732142857142857,
|
| 2786 |
+
"loss": 5.8187,
|
| 2787 |
+
"step": 463
|
| 2788 |
+
},
|
| 2789 |
+
{
|
| 2790 |
+
"epoch": 13.26,
|
| 2791 |
+
"learning_rate": 0.0021714285714285715,
|
| 2792 |
+
"loss": 5.7448,
|
| 2793 |
+
"step": 464
|
| 2794 |
+
},
|
| 2795 |
+
{
|
| 2796 |
+
"epoch": 13.29,
|
| 2797 |
+
"learning_rate": 0.002169642857142857,
|
| 2798 |
+
"loss": 5.8681,
|
| 2799 |
+
"step": 465
|
| 2800 |
+
},
|
| 2801 |
+
{
|
| 2802 |
+
"epoch": 13.31,
|
| 2803 |
+
"learning_rate": 0.002167857142857143,
|
| 2804 |
+
"loss": 5.8039,
|
| 2805 |
+
"step": 466
|
| 2806 |
+
},
|
| 2807 |
+
{
|
| 2808 |
+
"epoch": 13.34,
|
| 2809 |
+
"learning_rate": 0.0021660714285714287,
|
| 2810 |
+
"loss": 5.8511,
|
| 2811 |
+
"step": 467
|
| 2812 |
+
},
|
| 2813 |
+
{
|
| 2814 |
+
"epoch": 13.37,
|
| 2815 |
+
"learning_rate": 0.0021642857142857145,
|
| 2816 |
+
"loss": 5.8184,
|
| 2817 |
+
"step": 468
|
| 2818 |
+
},
|
| 2819 |
+
{
|
| 2820 |
+
"epoch": 13.4,
|
| 2821 |
+
"learning_rate": 0.0021625,
|
| 2822 |
+
"loss": 5.7656,
|
| 2823 |
+
"step": 469
|
| 2824 |
+
},
|
| 2825 |
+
{
|
| 2826 |
+
"epoch": 13.43,
|
| 2827 |
+
"learning_rate": 0.0021607142857142858,
|
| 2828 |
+
"loss": 5.8613,
|
| 2829 |
+
"step": 470
|
| 2830 |
+
},
|
| 2831 |
+
{
|
| 2832 |
+
"epoch": 13.46,
|
| 2833 |
+
"learning_rate": 0.0021589285714285716,
|
| 2834 |
+
"loss": 5.849,
|
| 2835 |
+
"step": 471
|
| 2836 |
+
},
|
| 2837 |
+
{
|
| 2838 |
+
"epoch": 13.49,
|
| 2839 |
+
"learning_rate": 0.002157142857142857,
|
| 2840 |
+
"loss": 5.8011,
|
| 2841 |
+
"step": 472
|
| 2842 |
+
},
|
| 2843 |
+
{
|
| 2844 |
+
"epoch": 13.51,
|
| 2845 |
+
"learning_rate": 0.002155357142857143,
|
| 2846 |
+
"loss": 5.7813,
|
| 2847 |
+
"step": 473
|
| 2848 |
+
},
|
| 2849 |
+
{
|
| 2850 |
+
"epoch": 13.54,
|
| 2851 |
+
"learning_rate": 0.0021535714285714287,
|
| 2852 |
+
"loss": 5.8186,
|
| 2853 |
+
"step": 474
|
| 2854 |
+
},
|
| 2855 |
+
{
|
| 2856 |
+
"epoch": 13.57,
|
| 2857 |
+
"learning_rate": 0.002151785714285714,
|
| 2858 |
+
"loss": 5.8303,
|
| 2859 |
+
"step": 475
|
| 2860 |
+
},
|
| 2861 |
+
{
|
| 2862 |
+
"epoch": 13.6,
|
| 2863 |
+
"learning_rate": 0.00215,
|
| 2864 |
+
"loss": 5.7879,
|
| 2865 |
+
"step": 476
|
| 2866 |
+
},
|
| 2867 |
+
{
|
| 2868 |
+
"epoch": 13.63,
|
| 2869 |
+
"learning_rate": 0.002148214285714286,
|
| 2870 |
+
"loss": 5.6829,
|
| 2871 |
+
"step": 477
|
| 2872 |
+
},
|
| 2873 |
+
{
|
| 2874 |
+
"epoch": 13.66,
|
| 2875 |
+
"learning_rate": 0.0021464285714285717,
|
| 2876 |
+
"loss": 5.7869,
|
| 2877 |
+
"step": 478
|
| 2878 |
+
},
|
| 2879 |
+
{
|
| 2880 |
+
"epoch": 13.69,
|
| 2881 |
+
"learning_rate": 0.002144642857142857,
|
| 2882 |
+
"loss": 5.6489,
|
| 2883 |
+
"step": 479
|
| 2884 |
+
},
|
| 2885 |
+
{
|
| 2886 |
+
"epoch": 13.71,
|
| 2887 |
+
"learning_rate": 0.002142857142857143,
|
| 2888 |
+
"loss": 5.8708,
|
| 2889 |
+
"step": 480
|
| 2890 |
+
},
|
| 2891 |
+
{
|
| 2892 |
+
"epoch": 13.74,
|
| 2893 |
+
"learning_rate": 0.002141071428571429,
|
| 2894 |
+
"loss": 5.7791,
|
| 2895 |
+
"step": 481
|
| 2896 |
+
},
|
| 2897 |
+
{
|
| 2898 |
+
"epoch": 13.77,
|
| 2899 |
+
"learning_rate": 0.0021392857142857142,
|
| 2900 |
+
"loss": 5.7497,
|
| 2901 |
+
"step": 482
|
| 2902 |
+
},
|
| 2903 |
+
{
|
| 2904 |
+
"epoch": 13.8,
|
| 2905 |
+
"learning_rate": 0.0021375,
|
| 2906 |
+
"loss": 5.827,
|
| 2907 |
+
"step": 483
|
| 2908 |
+
},
|
| 2909 |
+
{
|
| 2910 |
+
"epoch": 13.83,
|
| 2911 |
+
"learning_rate": 0.002135714285714286,
|
| 2912 |
+
"loss": 5.7286,
|
| 2913 |
+
"step": 484
|
| 2914 |
+
},
|
| 2915 |
+
{
|
| 2916 |
+
"epoch": 13.86,
|
| 2917 |
+
"learning_rate": 0.0021339285714285713,
|
| 2918 |
+
"loss": 5.8183,
|
| 2919 |
+
"step": 485
|
| 2920 |
+
},
|
| 2921 |
+
{
|
| 2922 |
+
"epoch": 13.89,
|
| 2923 |
+
"learning_rate": 0.002132142857142857,
|
| 2924 |
+
"loss": 5.7191,
|
| 2925 |
+
"step": 486
|
| 2926 |
+
},
|
| 2927 |
+
{
|
| 2928 |
+
"epoch": 13.91,
|
| 2929 |
+
"learning_rate": 0.002130357142857143,
|
| 2930 |
+
"loss": 5.7647,
|
| 2931 |
+
"step": 487
|
| 2932 |
+
},
|
| 2933 |
+
{
|
| 2934 |
+
"epoch": 13.94,
|
| 2935 |
+
"learning_rate": 0.002128571428571429,
|
| 2936 |
+
"loss": 5.799,
|
| 2937 |
+
"step": 488
|
| 2938 |
+
},
|
| 2939 |
+
{
|
| 2940 |
+
"epoch": 13.97,
|
| 2941 |
+
"learning_rate": 0.0021267857142857143,
|
| 2942 |
+
"loss": 5.7583,
|
| 2943 |
+
"step": 489
|
| 2944 |
+
},
|
| 2945 |
+
{
|
| 2946 |
+
"epoch": 14.0,
|
| 2947 |
+
"learning_rate": 0.002125,
|
| 2948 |
+
"loss": 5.6326,
|
| 2949 |
+
"step": 490
|
| 2950 |
+
},
|
| 2951 |
+
{
|
| 2952 |
+
"epoch": 14.03,
|
| 2953 |
+
"learning_rate": 0.002123214285714286,
|
| 2954 |
+
"loss": 5.614,
|
| 2955 |
+
"step": 491
|
| 2956 |
+
},
|
| 2957 |
+
{
|
| 2958 |
+
"epoch": 14.06,
|
| 2959 |
+
"learning_rate": 0.0021214285714285714,
|
| 2960 |
+
"loss": 5.7278,
|
| 2961 |
+
"step": 492
|
| 2962 |
+
},
|
| 2963 |
+
{
|
| 2964 |
+
"epoch": 14.09,
|
| 2965 |
+
"learning_rate": 0.0021196428571428573,
|
| 2966 |
+
"loss": 5.6661,
|
| 2967 |
+
"step": 493
|
| 2968 |
+
},
|
| 2969 |
+
{
|
| 2970 |
+
"epoch": 14.11,
|
| 2971 |
+
"learning_rate": 0.002117857142857143,
|
| 2972 |
+
"loss": 5.6822,
|
| 2973 |
+
"step": 494
|
| 2974 |
+
},
|
| 2975 |
+
{
|
| 2976 |
+
"epoch": 14.14,
|
| 2977 |
+
"learning_rate": 0.002116071428571429,
|
| 2978 |
+
"loss": 5.7356,
|
| 2979 |
+
"step": 495
|
| 2980 |
+
},
|
| 2981 |
+
{
|
| 2982 |
+
"epoch": 14.17,
|
| 2983 |
+
"learning_rate": 0.0021142857142857144,
|
| 2984 |
+
"loss": 5.6169,
|
| 2985 |
+
"step": 496
|
| 2986 |
+
},
|
| 2987 |
+
{
|
| 2988 |
+
"epoch": 14.2,
|
| 2989 |
+
"learning_rate": 0.0021125000000000002,
|
| 2990 |
+
"loss": 5.7203,
|
| 2991 |
+
"step": 497
|
| 2992 |
+
},
|
| 2993 |
+
{
|
| 2994 |
+
"epoch": 14.23,
|
| 2995 |
+
"learning_rate": 0.002110714285714286,
|
| 2996 |
+
"loss": 5.6377,
|
| 2997 |
+
"step": 498
|
| 2998 |
+
},
|
| 2999 |
+
{
|
| 3000 |
+
"epoch": 14.26,
|
| 3001 |
+
"learning_rate": 0.0021089285714285715,
|
| 3002 |
+
"loss": 5.6836,
|
| 3003 |
+
"step": 499
|
| 3004 |
+
},
|
| 3005 |
+
{
|
| 3006 |
+
"epoch": 14.29,
|
| 3007 |
+
"learning_rate": 0.002107142857142857,
|
| 3008 |
+
"loss": 5.6531,
|
| 3009 |
+
"step": 500
|
| 3010 |
}
|
| 3011 |
],
|
| 3012 |
"logging_steps": 1,
|
| 3013 |
"max_steps": 1680,
|
| 3014 |
"num_train_epochs": 48,
|
| 3015 |
"save_steps": 100,
|
| 3016 |
+
"total_flos": 2.884638740186112e+17,
|
| 3017 |
"trial_name": null,
|
| 3018 |
"trial_params": null
|
| 3019 |
}
|