Upload folder using huggingface_hub
#6
by
CreatorPhan
- opened
- adapter_model.bin +1 -1
- optimizer.pt +1 -1
- rng_state.pth +1 -1
- scheduler.pt +1 -1
- trainer_state.json +1203 -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:ed6e90196f5a274b5d5ff0f18e648b6396fc99189dc82111fcfd2e83656a72f7
|
| 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 sha256:2fe640e1cc583a2fcc1f7be360da696c6de30f28f00d5f64bbb3f586eab33160
|
| 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:ab62043be50b93d4eb28964be2d945176db3d64fe73ddd052a7656ba9141c683
|
| 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 sha256:7b34852b759fa1b56e14c21698a89fa4737f1160151f73515f2ad9b2b1334acd
|
| 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,
|
|
@@ -4207,13 +4207,1213 @@
|
|
| 4207 |
"learning_rate": 0.00010056818181818181,
|
| 4208 |
"loss": 0.6347,
|
| 4209 |
"step": 700
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 4210 |
}
|
| 4211 |
],
|
| 4212 |
"logging_steps": 1,
|
| 4213 |
"max_steps": 1408,
|
| 4214 |
"num_train_epochs": 16,
|
| 4215 |
"save_steps": 100,
|
| 4216 |
-
"total_flos":
|
| 4217 |
"trial_name": null,
|
| 4218 |
"trial_params": null
|
| 4219 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 10.167696381288614,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 900,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 4207 |
"learning_rate": 0.00010056818181818181,
|
| 4208 |
"loss": 0.6347,
|
| 4209 |
"step": 700
|
| 4210 |
+
},
|
| 4211 |
+
{
|
| 4212 |
+
"epoch": 7.92,
|
| 4213 |
+
"learning_rate": 0.00010042613636363636,
|
| 4214 |
+
"loss": 0.6857,
|
| 4215 |
+
"step": 701
|
| 4216 |
+
},
|
| 4217 |
+
{
|
| 4218 |
+
"epoch": 7.93,
|
| 4219 |
+
"learning_rate": 0.00010028409090909093,
|
| 4220 |
+
"loss": 0.6677,
|
| 4221 |
+
"step": 702
|
| 4222 |
+
},
|
| 4223 |
+
{
|
| 4224 |
+
"epoch": 7.94,
|
| 4225 |
+
"learning_rate": 0.00010014204545454546,
|
| 4226 |
+
"loss": 0.6697,
|
| 4227 |
+
"step": 703
|
| 4228 |
+
},
|
| 4229 |
+
{
|
| 4230 |
+
"epoch": 7.95,
|
| 4231 |
+
"learning_rate": 0.0001,
|
| 4232 |
+
"loss": 0.6375,
|
| 4233 |
+
"step": 704
|
| 4234 |
+
},
|
| 4235 |
+
{
|
| 4236 |
+
"epoch": 7.96,
|
| 4237 |
+
"learning_rate": 9.985795454545455e-05,
|
| 4238 |
+
"loss": 0.6572,
|
| 4239 |
+
"step": 705
|
| 4240 |
+
},
|
| 4241 |
+
{
|
| 4242 |
+
"epoch": 7.98,
|
| 4243 |
+
"learning_rate": 9.97159090909091e-05,
|
| 4244 |
+
"loss": 0.668,
|
| 4245 |
+
"step": 706
|
| 4246 |
+
},
|
| 4247 |
+
{
|
| 4248 |
+
"epoch": 7.99,
|
| 4249 |
+
"learning_rate": 9.957386363636364e-05,
|
| 4250 |
+
"loss": 0.6797,
|
| 4251 |
+
"step": 707
|
| 4252 |
+
},
|
| 4253 |
+
{
|
| 4254 |
+
"epoch": 8.0,
|
| 4255 |
+
"learning_rate": 9.943181818181819e-05,
|
| 4256 |
+
"loss": 0.6784,
|
| 4257 |
+
"step": 708
|
| 4258 |
+
},
|
| 4259 |
+
{
|
| 4260 |
+
"epoch": 8.01,
|
| 4261 |
+
"learning_rate": 9.928977272727273e-05,
|
| 4262 |
+
"loss": 0.6192,
|
| 4263 |
+
"step": 709
|
| 4264 |
+
},
|
| 4265 |
+
{
|
| 4266 |
+
"epoch": 8.02,
|
| 4267 |
+
"learning_rate": 9.914772727272728e-05,
|
| 4268 |
+
"loss": 0.6287,
|
| 4269 |
+
"step": 710
|
| 4270 |
+
},
|
| 4271 |
+
{
|
| 4272 |
+
"epoch": 8.03,
|
| 4273 |
+
"learning_rate": 9.900568181818183e-05,
|
| 4274 |
+
"loss": 0.6034,
|
| 4275 |
+
"step": 711
|
| 4276 |
+
},
|
| 4277 |
+
{
|
| 4278 |
+
"epoch": 8.04,
|
| 4279 |
+
"learning_rate": 9.886363636363637e-05,
|
| 4280 |
+
"loss": 0.6167,
|
| 4281 |
+
"step": 712
|
| 4282 |
+
},
|
| 4283 |
+
{
|
| 4284 |
+
"epoch": 8.06,
|
| 4285 |
+
"learning_rate": 9.872159090909091e-05,
|
| 4286 |
+
"loss": 0.6353,
|
| 4287 |
+
"step": 713
|
| 4288 |
+
},
|
| 4289 |
+
{
|
| 4290 |
+
"epoch": 8.07,
|
| 4291 |
+
"learning_rate": 9.857954545454547e-05,
|
| 4292 |
+
"loss": 0.6222,
|
| 4293 |
+
"step": 714
|
| 4294 |
+
},
|
| 4295 |
+
{
|
| 4296 |
+
"epoch": 8.08,
|
| 4297 |
+
"learning_rate": 9.84375e-05,
|
| 4298 |
+
"loss": 0.5963,
|
| 4299 |
+
"step": 715
|
| 4300 |
+
},
|
| 4301 |
+
{
|
| 4302 |
+
"epoch": 8.09,
|
| 4303 |
+
"learning_rate": 9.829545454545455e-05,
|
| 4304 |
+
"loss": 0.6042,
|
| 4305 |
+
"step": 716
|
| 4306 |
+
},
|
| 4307 |
+
{
|
| 4308 |
+
"epoch": 8.1,
|
| 4309 |
+
"learning_rate": 9.81534090909091e-05,
|
| 4310 |
+
"loss": 0.612,
|
| 4311 |
+
"step": 717
|
| 4312 |
+
},
|
| 4313 |
+
{
|
| 4314 |
+
"epoch": 8.11,
|
| 4315 |
+
"learning_rate": 9.801136363636364e-05,
|
| 4316 |
+
"loss": 0.6069,
|
| 4317 |
+
"step": 718
|
| 4318 |
+
},
|
| 4319 |
+
{
|
| 4320 |
+
"epoch": 8.12,
|
| 4321 |
+
"learning_rate": 9.786931818181818e-05,
|
| 4322 |
+
"loss": 0.6001,
|
| 4323 |
+
"step": 719
|
| 4324 |
+
},
|
| 4325 |
+
{
|
| 4326 |
+
"epoch": 8.13,
|
| 4327 |
+
"learning_rate": 9.772727272727274e-05,
|
| 4328 |
+
"loss": 0.6007,
|
| 4329 |
+
"step": 720
|
| 4330 |
+
},
|
| 4331 |
+
{
|
| 4332 |
+
"epoch": 8.15,
|
| 4333 |
+
"learning_rate": 9.758522727272727e-05,
|
| 4334 |
+
"loss": 0.6079,
|
| 4335 |
+
"step": 721
|
| 4336 |
+
},
|
| 4337 |
+
{
|
| 4338 |
+
"epoch": 8.16,
|
| 4339 |
+
"learning_rate": 9.744318181818183e-05,
|
| 4340 |
+
"loss": 0.6216,
|
| 4341 |
+
"step": 722
|
| 4342 |
+
},
|
| 4343 |
+
{
|
| 4344 |
+
"epoch": 8.17,
|
| 4345 |
+
"learning_rate": 9.730113636363637e-05,
|
| 4346 |
+
"loss": 0.6321,
|
| 4347 |
+
"step": 723
|
| 4348 |
+
},
|
| 4349 |
+
{
|
| 4350 |
+
"epoch": 8.18,
|
| 4351 |
+
"learning_rate": 9.71590909090909e-05,
|
| 4352 |
+
"loss": 0.6044,
|
| 4353 |
+
"step": 724
|
| 4354 |
+
},
|
| 4355 |
+
{
|
| 4356 |
+
"epoch": 8.19,
|
| 4357 |
+
"learning_rate": 9.701704545454547e-05,
|
| 4358 |
+
"loss": 0.6028,
|
| 4359 |
+
"step": 725
|
| 4360 |
+
},
|
| 4361 |
+
{
|
| 4362 |
+
"epoch": 8.2,
|
| 4363 |
+
"learning_rate": 9.687500000000001e-05,
|
| 4364 |
+
"loss": 0.6098,
|
| 4365 |
+
"step": 726
|
| 4366 |
+
},
|
| 4367 |
+
{
|
| 4368 |
+
"epoch": 8.21,
|
| 4369 |
+
"learning_rate": 9.673295454545454e-05,
|
| 4370 |
+
"loss": 0.6032,
|
| 4371 |
+
"step": 727
|
| 4372 |
+
},
|
| 4373 |
+
{
|
| 4374 |
+
"epoch": 8.22,
|
| 4375 |
+
"learning_rate": 9.65909090909091e-05,
|
| 4376 |
+
"loss": 0.6298,
|
| 4377 |
+
"step": 728
|
| 4378 |
+
},
|
| 4379 |
+
{
|
| 4380 |
+
"epoch": 8.24,
|
| 4381 |
+
"learning_rate": 9.644886363636365e-05,
|
| 4382 |
+
"loss": 0.6115,
|
| 4383 |
+
"step": 729
|
| 4384 |
+
},
|
| 4385 |
+
{
|
| 4386 |
+
"epoch": 8.25,
|
| 4387 |
+
"learning_rate": 9.630681818181818e-05,
|
| 4388 |
+
"loss": 0.6052,
|
| 4389 |
+
"step": 730
|
| 4390 |
+
},
|
| 4391 |
+
{
|
| 4392 |
+
"epoch": 8.26,
|
| 4393 |
+
"learning_rate": 9.616477272727274e-05,
|
| 4394 |
+
"loss": 0.6097,
|
| 4395 |
+
"step": 731
|
| 4396 |
+
},
|
| 4397 |
+
{
|
| 4398 |
+
"epoch": 8.27,
|
| 4399 |
+
"learning_rate": 9.602272727272728e-05,
|
| 4400 |
+
"loss": 0.6062,
|
| 4401 |
+
"step": 732
|
| 4402 |
+
},
|
| 4403 |
+
{
|
| 4404 |
+
"epoch": 8.28,
|
| 4405 |
+
"learning_rate": 9.588068181818183e-05,
|
| 4406 |
+
"loss": 0.5984,
|
| 4407 |
+
"step": 733
|
| 4408 |
+
},
|
| 4409 |
+
{
|
| 4410 |
+
"epoch": 8.29,
|
| 4411 |
+
"learning_rate": 9.573863636363637e-05,
|
| 4412 |
+
"loss": 0.6432,
|
| 4413 |
+
"step": 734
|
| 4414 |
+
},
|
| 4415 |
+
{
|
| 4416 |
+
"epoch": 8.3,
|
| 4417 |
+
"learning_rate": 9.559659090909092e-05,
|
| 4418 |
+
"loss": 0.5814,
|
| 4419 |
+
"step": 735
|
| 4420 |
+
},
|
| 4421 |
+
{
|
| 4422 |
+
"epoch": 8.31,
|
| 4423 |
+
"learning_rate": 9.545454545454546e-05,
|
| 4424 |
+
"loss": 0.5965,
|
| 4425 |
+
"step": 736
|
| 4426 |
+
},
|
| 4427 |
+
{
|
| 4428 |
+
"epoch": 8.33,
|
| 4429 |
+
"learning_rate": 9.53125e-05,
|
| 4430 |
+
"loss": 0.6102,
|
| 4431 |
+
"step": 737
|
| 4432 |
+
},
|
| 4433 |
+
{
|
| 4434 |
+
"epoch": 8.34,
|
| 4435 |
+
"learning_rate": 9.517045454545455e-05,
|
| 4436 |
+
"loss": 0.5849,
|
| 4437 |
+
"step": 738
|
| 4438 |
+
},
|
| 4439 |
+
{
|
| 4440 |
+
"epoch": 8.35,
|
| 4441 |
+
"learning_rate": 9.50284090909091e-05,
|
| 4442 |
+
"loss": 0.6062,
|
| 4443 |
+
"step": 739
|
| 4444 |
+
},
|
| 4445 |
+
{
|
| 4446 |
+
"epoch": 8.36,
|
| 4447 |
+
"learning_rate": 9.488636363636364e-05,
|
| 4448 |
+
"loss": 0.6031,
|
| 4449 |
+
"step": 740
|
| 4450 |
+
},
|
| 4451 |
+
{
|
| 4452 |
+
"epoch": 8.37,
|
| 4453 |
+
"learning_rate": 9.474431818181819e-05,
|
| 4454 |
+
"loss": 0.5932,
|
| 4455 |
+
"step": 741
|
| 4456 |
+
},
|
| 4457 |
+
{
|
| 4458 |
+
"epoch": 8.38,
|
| 4459 |
+
"learning_rate": 9.460227272727273e-05,
|
| 4460 |
+
"loss": 0.589,
|
| 4461 |
+
"step": 742
|
| 4462 |
+
},
|
| 4463 |
+
{
|
| 4464 |
+
"epoch": 8.39,
|
| 4465 |
+
"learning_rate": 9.446022727272728e-05,
|
| 4466 |
+
"loss": 0.6096,
|
| 4467 |
+
"step": 743
|
| 4468 |
+
},
|
| 4469 |
+
{
|
| 4470 |
+
"epoch": 8.41,
|
| 4471 |
+
"learning_rate": 9.431818181818182e-05,
|
| 4472 |
+
"loss": 0.601,
|
| 4473 |
+
"step": 744
|
| 4474 |
+
},
|
| 4475 |
+
{
|
| 4476 |
+
"epoch": 8.42,
|
| 4477 |
+
"learning_rate": 9.417613636363637e-05,
|
| 4478 |
+
"loss": 0.5798,
|
| 4479 |
+
"step": 745
|
| 4480 |
+
},
|
| 4481 |
+
{
|
| 4482 |
+
"epoch": 8.43,
|
| 4483 |
+
"learning_rate": 9.403409090909091e-05,
|
| 4484 |
+
"loss": 0.59,
|
| 4485 |
+
"step": 746
|
| 4486 |
+
},
|
| 4487 |
+
{
|
| 4488 |
+
"epoch": 8.44,
|
| 4489 |
+
"learning_rate": 9.389204545454546e-05,
|
| 4490 |
+
"loss": 0.5988,
|
| 4491 |
+
"step": 747
|
| 4492 |
+
},
|
| 4493 |
+
{
|
| 4494 |
+
"epoch": 8.45,
|
| 4495 |
+
"learning_rate": 9.375e-05,
|
| 4496 |
+
"loss": 0.5591,
|
| 4497 |
+
"step": 748
|
| 4498 |
+
},
|
| 4499 |
+
{
|
| 4500 |
+
"epoch": 8.46,
|
| 4501 |
+
"learning_rate": 9.360795454545455e-05,
|
| 4502 |
+
"loss": 0.5939,
|
| 4503 |
+
"step": 749
|
| 4504 |
+
},
|
| 4505 |
+
{
|
| 4506 |
+
"epoch": 8.47,
|
| 4507 |
+
"learning_rate": 9.346590909090909e-05,
|
| 4508 |
+
"loss": 0.5886,
|
| 4509 |
+
"step": 750
|
| 4510 |
+
},
|
| 4511 |
+
{
|
| 4512 |
+
"epoch": 8.48,
|
| 4513 |
+
"learning_rate": 9.332386363636364e-05,
|
| 4514 |
+
"loss": 0.5994,
|
| 4515 |
+
"step": 751
|
| 4516 |
+
},
|
| 4517 |
+
{
|
| 4518 |
+
"epoch": 8.5,
|
| 4519 |
+
"learning_rate": 9.318181818181818e-05,
|
| 4520 |
+
"loss": 0.5821,
|
| 4521 |
+
"step": 752
|
| 4522 |
+
},
|
| 4523 |
+
{
|
| 4524 |
+
"epoch": 8.51,
|
| 4525 |
+
"learning_rate": 9.303977272727273e-05,
|
| 4526 |
+
"loss": 0.602,
|
| 4527 |
+
"step": 753
|
| 4528 |
+
},
|
| 4529 |
+
{
|
| 4530 |
+
"epoch": 8.52,
|
| 4531 |
+
"learning_rate": 9.289772727272727e-05,
|
| 4532 |
+
"loss": 0.5708,
|
| 4533 |
+
"step": 754
|
| 4534 |
+
},
|
| 4535 |
+
{
|
| 4536 |
+
"epoch": 8.53,
|
| 4537 |
+
"learning_rate": 9.275568181818183e-05,
|
| 4538 |
+
"loss": 0.5902,
|
| 4539 |
+
"step": 755
|
| 4540 |
+
},
|
| 4541 |
+
{
|
| 4542 |
+
"epoch": 8.54,
|
| 4543 |
+
"learning_rate": 9.261363636363636e-05,
|
| 4544 |
+
"loss": 0.6053,
|
| 4545 |
+
"step": 756
|
| 4546 |
+
},
|
| 4547 |
+
{
|
| 4548 |
+
"epoch": 8.55,
|
| 4549 |
+
"learning_rate": 9.247159090909091e-05,
|
| 4550 |
+
"loss": 0.5797,
|
| 4551 |
+
"step": 757
|
| 4552 |
+
},
|
| 4553 |
+
{
|
| 4554 |
+
"epoch": 8.56,
|
| 4555 |
+
"learning_rate": 9.232954545454547e-05,
|
| 4556 |
+
"loss": 0.5965,
|
| 4557 |
+
"step": 758
|
| 4558 |
+
},
|
| 4559 |
+
{
|
| 4560 |
+
"epoch": 8.57,
|
| 4561 |
+
"learning_rate": 9.21875e-05,
|
| 4562 |
+
"loss": 0.5738,
|
| 4563 |
+
"step": 759
|
| 4564 |
+
},
|
| 4565 |
+
{
|
| 4566 |
+
"epoch": 8.59,
|
| 4567 |
+
"learning_rate": 9.204545454545454e-05,
|
| 4568 |
+
"loss": 0.5819,
|
| 4569 |
+
"step": 760
|
| 4570 |
+
},
|
| 4571 |
+
{
|
| 4572 |
+
"epoch": 8.6,
|
| 4573 |
+
"learning_rate": 9.19034090909091e-05,
|
| 4574 |
+
"loss": 0.5994,
|
| 4575 |
+
"step": 761
|
| 4576 |
+
},
|
| 4577 |
+
{
|
| 4578 |
+
"epoch": 8.61,
|
| 4579 |
+
"learning_rate": 9.176136363636363e-05,
|
| 4580 |
+
"loss": 0.5738,
|
| 4581 |
+
"step": 762
|
| 4582 |
+
},
|
| 4583 |
+
{
|
| 4584 |
+
"epoch": 8.62,
|
| 4585 |
+
"learning_rate": 9.161931818181818e-05,
|
| 4586 |
+
"loss": 0.5663,
|
| 4587 |
+
"step": 763
|
| 4588 |
+
},
|
| 4589 |
+
{
|
| 4590 |
+
"epoch": 8.63,
|
| 4591 |
+
"learning_rate": 9.147727272727274e-05,
|
| 4592 |
+
"loss": 0.5798,
|
| 4593 |
+
"step": 764
|
| 4594 |
+
},
|
| 4595 |
+
{
|
| 4596 |
+
"epoch": 8.64,
|
| 4597 |
+
"learning_rate": 9.133522727272727e-05,
|
| 4598 |
+
"loss": 0.5705,
|
| 4599 |
+
"step": 765
|
| 4600 |
+
},
|
| 4601 |
+
{
|
| 4602 |
+
"epoch": 8.65,
|
| 4603 |
+
"learning_rate": 9.119318181818183e-05,
|
| 4604 |
+
"loss": 0.5943,
|
| 4605 |
+
"step": 766
|
| 4606 |
+
},
|
| 4607 |
+
{
|
| 4608 |
+
"epoch": 8.67,
|
| 4609 |
+
"learning_rate": 9.105113636363637e-05,
|
| 4610 |
+
"loss": 0.6019,
|
| 4611 |
+
"step": 767
|
| 4612 |
+
},
|
| 4613 |
+
{
|
| 4614 |
+
"epoch": 8.68,
|
| 4615 |
+
"learning_rate": 9.090909090909092e-05,
|
| 4616 |
+
"loss": 0.5733,
|
| 4617 |
+
"step": 768
|
| 4618 |
+
},
|
| 4619 |
+
{
|
| 4620 |
+
"epoch": 8.69,
|
| 4621 |
+
"learning_rate": 9.076704545454546e-05,
|
| 4622 |
+
"loss": 0.575,
|
| 4623 |
+
"step": 769
|
| 4624 |
+
},
|
| 4625 |
+
{
|
| 4626 |
+
"epoch": 8.7,
|
| 4627 |
+
"learning_rate": 9.062500000000001e-05,
|
| 4628 |
+
"loss": 0.5675,
|
| 4629 |
+
"step": 770
|
| 4630 |
+
},
|
| 4631 |
+
{
|
| 4632 |
+
"epoch": 8.71,
|
| 4633 |
+
"learning_rate": 9.048295454545455e-05,
|
| 4634 |
+
"loss": 0.566,
|
| 4635 |
+
"step": 771
|
| 4636 |
+
},
|
| 4637 |
+
{
|
| 4638 |
+
"epoch": 8.72,
|
| 4639 |
+
"learning_rate": 9.03409090909091e-05,
|
| 4640 |
+
"loss": 0.5513,
|
| 4641 |
+
"step": 772
|
| 4642 |
+
},
|
| 4643 |
+
{
|
| 4644 |
+
"epoch": 8.73,
|
| 4645 |
+
"learning_rate": 9.019886363636364e-05,
|
| 4646 |
+
"loss": 0.5682,
|
| 4647 |
+
"step": 773
|
| 4648 |
+
},
|
| 4649 |
+
{
|
| 4650 |
+
"epoch": 8.74,
|
| 4651 |
+
"learning_rate": 9.005681818181819e-05,
|
| 4652 |
+
"loss": 0.5508,
|
| 4653 |
+
"step": 774
|
| 4654 |
+
},
|
| 4655 |
+
{
|
| 4656 |
+
"epoch": 8.76,
|
| 4657 |
+
"learning_rate": 8.991477272727273e-05,
|
| 4658 |
+
"loss": 0.5668,
|
| 4659 |
+
"step": 775
|
| 4660 |
+
},
|
| 4661 |
+
{
|
| 4662 |
+
"epoch": 8.77,
|
| 4663 |
+
"learning_rate": 8.977272727272728e-05,
|
| 4664 |
+
"loss": 0.569,
|
| 4665 |
+
"step": 776
|
| 4666 |
+
},
|
| 4667 |
+
{
|
| 4668 |
+
"epoch": 8.78,
|
| 4669 |
+
"learning_rate": 8.963068181818182e-05,
|
| 4670 |
+
"loss": 0.5897,
|
| 4671 |
+
"step": 777
|
| 4672 |
+
},
|
| 4673 |
+
{
|
| 4674 |
+
"epoch": 8.79,
|
| 4675 |
+
"learning_rate": 8.948863636363637e-05,
|
| 4676 |
+
"loss": 0.5738,
|
| 4677 |
+
"step": 778
|
| 4678 |
+
},
|
| 4679 |
+
{
|
| 4680 |
+
"epoch": 8.8,
|
| 4681 |
+
"learning_rate": 8.934659090909091e-05,
|
| 4682 |
+
"loss": 0.5511,
|
| 4683 |
+
"step": 779
|
| 4684 |
+
},
|
| 4685 |
+
{
|
| 4686 |
+
"epoch": 8.81,
|
| 4687 |
+
"learning_rate": 8.920454545454546e-05,
|
| 4688 |
+
"loss": 0.5659,
|
| 4689 |
+
"step": 780
|
| 4690 |
+
},
|
| 4691 |
+
{
|
| 4692 |
+
"epoch": 8.82,
|
| 4693 |
+
"learning_rate": 8.90625e-05,
|
| 4694 |
+
"loss": 0.5649,
|
| 4695 |
+
"step": 781
|
| 4696 |
+
},
|
| 4697 |
+
{
|
| 4698 |
+
"epoch": 8.83,
|
| 4699 |
+
"learning_rate": 8.892045454545455e-05,
|
| 4700 |
+
"loss": 0.5618,
|
| 4701 |
+
"step": 782
|
| 4702 |
+
},
|
| 4703 |
+
{
|
| 4704 |
+
"epoch": 8.85,
|
| 4705 |
+
"learning_rate": 8.87784090909091e-05,
|
| 4706 |
+
"loss": 0.5602,
|
| 4707 |
+
"step": 783
|
| 4708 |
+
},
|
| 4709 |
+
{
|
| 4710 |
+
"epoch": 8.86,
|
| 4711 |
+
"learning_rate": 8.863636363636364e-05,
|
| 4712 |
+
"loss": 0.5723,
|
| 4713 |
+
"step": 784
|
| 4714 |
+
},
|
| 4715 |
+
{
|
| 4716 |
+
"epoch": 8.87,
|
| 4717 |
+
"learning_rate": 8.849431818181818e-05,
|
| 4718 |
+
"loss": 0.5816,
|
| 4719 |
+
"step": 785
|
| 4720 |
+
},
|
| 4721 |
+
{
|
| 4722 |
+
"epoch": 8.88,
|
| 4723 |
+
"learning_rate": 8.835227272727273e-05,
|
| 4724 |
+
"loss": 0.555,
|
| 4725 |
+
"step": 786
|
| 4726 |
+
},
|
| 4727 |
+
{
|
| 4728 |
+
"epoch": 8.89,
|
| 4729 |
+
"learning_rate": 8.821022727272727e-05,
|
| 4730 |
+
"loss": 0.5563,
|
| 4731 |
+
"step": 787
|
| 4732 |
+
},
|
| 4733 |
+
{
|
| 4734 |
+
"epoch": 8.9,
|
| 4735 |
+
"learning_rate": 8.806818181818183e-05,
|
| 4736 |
+
"loss": 0.554,
|
| 4737 |
+
"step": 788
|
| 4738 |
+
},
|
| 4739 |
+
{
|
| 4740 |
+
"epoch": 8.91,
|
| 4741 |
+
"learning_rate": 8.792613636363636e-05,
|
| 4742 |
+
"loss": 0.5671,
|
| 4743 |
+
"step": 789
|
| 4744 |
+
},
|
| 4745 |
+
{
|
| 4746 |
+
"epoch": 8.92,
|
| 4747 |
+
"learning_rate": 8.778409090909091e-05,
|
| 4748 |
+
"loss": 0.5485,
|
| 4749 |
+
"step": 790
|
| 4750 |
+
},
|
| 4751 |
+
{
|
| 4752 |
+
"epoch": 8.94,
|
| 4753 |
+
"learning_rate": 8.764204545454547e-05,
|
| 4754 |
+
"loss": 0.5712,
|
| 4755 |
+
"step": 791
|
| 4756 |
+
},
|
| 4757 |
+
{
|
| 4758 |
+
"epoch": 8.95,
|
| 4759 |
+
"learning_rate": 8.75e-05,
|
| 4760 |
+
"loss": 0.5507,
|
| 4761 |
+
"step": 792
|
| 4762 |
+
},
|
| 4763 |
+
{
|
| 4764 |
+
"epoch": 8.96,
|
| 4765 |
+
"learning_rate": 8.735795454545454e-05,
|
| 4766 |
+
"loss": 0.5718,
|
| 4767 |
+
"step": 793
|
| 4768 |
+
},
|
| 4769 |
+
{
|
| 4770 |
+
"epoch": 8.97,
|
| 4771 |
+
"learning_rate": 8.72159090909091e-05,
|
| 4772 |
+
"loss": 0.5585,
|
| 4773 |
+
"step": 794
|
| 4774 |
+
},
|
| 4775 |
+
{
|
| 4776 |
+
"epoch": 8.98,
|
| 4777 |
+
"learning_rate": 8.707386363636363e-05,
|
| 4778 |
+
"loss": 0.5563,
|
| 4779 |
+
"step": 795
|
| 4780 |
+
},
|
| 4781 |
+
{
|
| 4782 |
+
"epoch": 8.99,
|
| 4783 |
+
"learning_rate": 8.693181818181818e-05,
|
| 4784 |
+
"loss": 0.581,
|
| 4785 |
+
"step": 796
|
| 4786 |
+
},
|
| 4787 |
+
{
|
| 4788 |
+
"epoch": 9.0,
|
| 4789 |
+
"learning_rate": 8.678977272727274e-05,
|
| 4790 |
+
"loss": 0.5511,
|
| 4791 |
+
"step": 797
|
| 4792 |
+
},
|
| 4793 |
+
{
|
| 4794 |
+
"epoch": 9.02,
|
| 4795 |
+
"learning_rate": 8.664772727272727e-05,
|
| 4796 |
+
"loss": 0.5103,
|
| 4797 |
+
"step": 798
|
| 4798 |
+
},
|
| 4799 |
+
{
|
| 4800 |
+
"epoch": 9.03,
|
| 4801 |
+
"learning_rate": 8.650568181818183e-05,
|
| 4802 |
+
"loss": 0.5323,
|
| 4803 |
+
"step": 799
|
| 4804 |
+
},
|
| 4805 |
+
{
|
| 4806 |
+
"epoch": 9.04,
|
| 4807 |
+
"learning_rate": 8.636363636363637e-05,
|
| 4808 |
+
"loss": 0.5092,
|
| 4809 |
+
"step": 800
|
| 4810 |
+
},
|
| 4811 |
+
{
|
| 4812 |
+
"epoch": 9.05,
|
| 4813 |
+
"learning_rate": 8.62215909090909e-05,
|
| 4814 |
+
"loss": 0.5247,
|
| 4815 |
+
"step": 801
|
| 4816 |
+
},
|
| 4817 |
+
{
|
| 4818 |
+
"epoch": 9.06,
|
| 4819 |
+
"learning_rate": 8.607954545454546e-05,
|
| 4820 |
+
"loss": 0.5403,
|
| 4821 |
+
"step": 802
|
| 4822 |
+
},
|
| 4823 |
+
{
|
| 4824 |
+
"epoch": 9.07,
|
| 4825 |
+
"learning_rate": 8.593750000000001e-05,
|
| 4826 |
+
"loss": 0.5252,
|
| 4827 |
+
"step": 803
|
| 4828 |
+
},
|
| 4829 |
+
{
|
| 4830 |
+
"epoch": 9.08,
|
| 4831 |
+
"learning_rate": 8.579545454545454e-05,
|
| 4832 |
+
"loss": 0.5296,
|
| 4833 |
+
"step": 804
|
| 4834 |
+
},
|
| 4835 |
+
{
|
| 4836 |
+
"epoch": 9.09,
|
| 4837 |
+
"learning_rate": 8.56534090909091e-05,
|
| 4838 |
+
"loss": 0.5223,
|
| 4839 |
+
"step": 805
|
| 4840 |
+
},
|
| 4841 |
+
{
|
| 4842 |
+
"epoch": 9.11,
|
| 4843 |
+
"learning_rate": 8.551136363636364e-05,
|
| 4844 |
+
"loss": 0.4972,
|
| 4845 |
+
"step": 806
|
| 4846 |
+
},
|
| 4847 |
+
{
|
| 4848 |
+
"epoch": 9.12,
|
| 4849 |
+
"learning_rate": 8.536931818181818e-05,
|
| 4850 |
+
"loss": 0.5005,
|
| 4851 |
+
"step": 807
|
| 4852 |
+
},
|
| 4853 |
+
{
|
| 4854 |
+
"epoch": 9.13,
|
| 4855 |
+
"learning_rate": 8.522727272727273e-05,
|
| 4856 |
+
"loss": 0.5249,
|
| 4857 |
+
"step": 808
|
| 4858 |
+
},
|
| 4859 |
+
{
|
| 4860 |
+
"epoch": 9.14,
|
| 4861 |
+
"learning_rate": 8.508522727272728e-05,
|
| 4862 |
+
"loss": 0.5135,
|
| 4863 |
+
"step": 809
|
| 4864 |
+
},
|
| 4865 |
+
{
|
| 4866 |
+
"epoch": 9.15,
|
| 4867 |
+
"learning_rate": 8.494318181818182e-05,
|
| 4868 |
+
"loss": 0.5053,
|
| 4869 |
+
"step": 810
|
| 4870 |
+
},
|
| 4871 |
+
{
|
| 4872 |
+
"epoch": 9.16,
|
| 4873 |
+
"learning_rate": 8.480113636363637e-05,
|
| 4874 |
+
"loss": 0.5158,
|
| 4875 |
+
"step": 811
|
| 4876 |
+
},
|
| 4877 |
+
{
|
| 4878 |
+
"epoch": 9.17,
|
| 4879 |
+
"learning_rate": 8.465909090909091e-05,
|
| 4880 |
+
"loss": 0.5061,
|
| 4881 |
+
"step": 812
|
| 4882 |
+
},
|
| 4883 |
+
{
|
| 4884 |
+
"epoch": 9.18,
|
| 4885 |
+
"learning_rate": 8.451704545454546e-05,
|
| 4886 |
+
"loss": 0.4988,
|
| 4887 |
+
"step": 813
|
| 4888 |
+
},
|
| 4889 |
+
{
|
| 4890 |
+
"epoch": 9.2,
|
| 4891 |
+
"learning_rate": 8.4375e-05,
|
| 4892 |
+
"loss": 0.5273,
|
| 4893 |
+
"step": 814
|
| 4894 |
+
},
|
| 4895 |
+
{
|
| 4896 |
+
"epoch": 9.21,
|
| 4897 |
+
"learning_rate": 8.423295454545455e-05,
|
| 4898 |
+
"loss": 0.5332,
|
| 4899 |
+
"step": 815
|
| 4900 |
+
},
|
| 4901 |
+
{
|
| 4902 |
+
"epoch": 9.22,
|
| 4903 |
+
"learning_rate": 8.40909090909091e-05,
|
| 4904 |
+
"loss": 0.5181,
|
| 4905 |
+
"step": 816
|
| 4906 |
+
},
|
| 4907 |
+
{
|
| 4908 |
+
"epoch": 9.23,
|
| 4909 |
+
"learning_rate": 8.394886363636364e-05,
|
| 4910 |
+
"loss": 0.5085,
|
| 4911 |
+
"step": 817
|
| 4912 |
+
},
|
| 4913 |
+
{
|
| 4914 |
+
"epoch": 9.24,
|
| 4915 |
+
"learning_rate": 8.380681818181818e-05,
|
| 4916 |
+
"loss": 0.5137,
|
| 4917 |
+
"step": 818
|
| 4918 |
+
},
|
| 4919 |
+
{
|
| 4920 |
+
"epoch": 9.25,
|
| 4921 |
+
"learning_rate": 8.366477272727273e-05,
|
| 4922 |
+
"loss": 0.5195,
|
| 4923 |
+
"step": 819
|
| 4924 |
+
},
|
| 4925 |
+
{
|
| 4926 |
+
"epoch": 9.26,
|
| 4927 |
+
"learning_rate": 8.352272727272727e-05,
|
| 4928 |
+
"loss": 0.5077,
|
| 4929 |
+
"step": 820
|
| 4930 |
+
},
|
| 4931 |
+
{
|
| 4932 |
+
"epoch": 9.28,
|
| 4933 |
+
"learning_rate": 8.338068181818183e-05,
|
| 4934 |
+
"loss": 0.5074,
|
| 4935 |
+
"step": 821
|
| 4936 |
+
},
|
| 4937 |
+
{
|
| 4938 |
+
"epoch": 9.29,
|
| 4939 |
+
"learning_rate": 8.323863636363637e-05,
|
| 4940 |
+
"loss": 0.5142,
|
| 4941 |
+
"step": 822
|
| 4942 |
+
},
|
| 4943 |
+
{
|
| 4944 |
+
"epoch": 9.3,
|
| 4945 |
+
"learning_rate": 8.309659090909091e-05,
|
| 4946 |
+
"loss": 0.5116,
|
| 4947 |
+
"step": 823
|
| 4948 |
+
},
|
| 4949 |
+
{
|
| 4950 |
+
"epoch": 9.31,
|
| 4951 |
+
"learning_rate": 8.295454545454547e-05,
|
| 4952 |
+
"loss": 0.4974,
|
| 4953 |
+
"step": 824
|
| 4954 |
+
},
|
| 4955 |
+
{
|
| 4956 |
+
"epoch": 9.32,
|
| 4957 |
+
"learning_rate": 8.28125e-05,
|
| 4958 |
+
"loss": 0.5117,
|
| 4959 |
+
"step": 825
|
| 4960 |
+
},
|
| 4961 |
+
{
|
| 4962 |
+
"epoch": 9.33,
|
| 4963 |
+
"learning_rate": 8.267045454545455e-05,
|
| 4964 |
+
"loss": 0.5114,
|
| 4965 |
+
"step": 826
|
| 4966 |
+
},
|
| 4967 |
+
{
|
| 4968 |
+
"epoch": 9.34,
|
| 4969 |
+
"learning_rate": 8.25284090909091e-05,
|
| 4970 |
+
"loss": 0.5039,
|
| 4971 |
+
"step": 827
|
| 4972 |
+
},
|
| 4973 |
+
{
|
| 4974 |
+
"epoch": 9.35,
|
| 4975 |
+
"learning_rate": 8.238636363636364e-05,
|
| 4976 |
+
"loss": 0.498,
|
| 4977 |
+
"step": 828
|
| 4978 |
+
},
|
| 4979 |
+
{
|
| 4980 |
+
"epoch": 9.37,
|
| 4981 |
+
"learning_rate": 8.224431818181818e-05,
|
| 4982 |
+
"loss": 0.5042,
|
| 4983 |
+
"step": 829
|
| 4984 |
+
},
|
| 4985 |
+
{
|
| 4986 |
+
"epoch": 9.38,
|
| 4987 |
+
"learning_rate": 8.210227272727274e-05,
|
| 4988 |
+
"loss": 0.5049,
|
| 4989 |
+
"step": 830
|
| 4990 |
+
},
|
| 4991 |
+
{
|
| 4992 |
+
"epoch": 9.39,
|
| 4993 |
+
"learning_rate": 8.196022727272727e-05,
|
| 4994 |
+
"loss": 0.5123,
|
| 4995 |
+
"step": 831
|
| 4996 |
+
},
|
| 4997 |
+
{
|
| 4998 |
+
"epoch": 9.4,
|
| 4999 |
+
"learning_rate": 8.181818181818183e-05,
|
| 5000 |
+
"loss": 0.4907,
|
| 5001 |
+
"step": 832
|
| 5002 |
+
},
|
| 5003 |
+
{
|
| 5004 |
+
"epoch": 9.41,
|
| 5005 |
+
"learning_rate": 8.167613636363637e-05,
|
| 5006 |
+
"loss": 0.5267,
|
| 5007 |
+
"step": 833
|
| 5008 |
+
},
|
| 5009 |
+
{
|
| 5010 |
+
"epoch": 9.42,
|
| 5011 |
+
"learning_rate": 8.15340909090909e-05,
|
| 5012 |
+
"loss": 0.5314,
|
| 5013 |
+
"step": 834
|
| 5014 |
+
},
|
| 5015 |
+
{
|
| 5016 |
+
"epoch": 9.43,
|
| 5017 |
+
"learning_rate": 8.139204545454546e-05,
|
| 5018 |
+
"loss": 0.4952,
|
| 5019 |
+
"step": 835
|
| 5020 |
+
},
|
| 5021 |
+
{
|
| 5022 |
+
"epoch": 9.44,
|
| 5023 |
+
"learning_rate": 8.125000000000001e-05,
|
| 5024 |
+
"loss": 0.5014,
|
| 5025 |
+
"step": 836
|
| 5026 |
+
},
|
| 5027 |
+
{
|
| 5028 |
+
"epoch": 9.46,
|
| 5029 |
+
"learning_rate": 8.110795454545454e-05,
|
| 5030 |
+
"loss": 0.4967,
|
| 5031 |
+
"step": 837
|
| 5032 |
+
},
|
| 5033 |
+
{
|
| 5034 |
+
"epoch": 9.47,
|
| 5035 |
+
"learning_rate": 8.09659090909091e-05,
|
| 5036 |
+
"loss": 0.5116,
|
| 5037 |
+
"step": 838
|
| 5038 |
+
},
|
| 5039 |
+
{
|
| 5040 |
+
"epoch": 9.48,
|
| 5041 |
+
"learning_rate": 8.082386363636365e-05,
|
| 5042 |
+
"loss": 0.5119,
|
| 5043 |
+
"step": 839
|
| 5044 |
+
},
|
| 5045 |
+
{
|
| 5046 |
+
"epoch": 9.49,
|
| 5047 |
+
"learning_rate": 8.068181818181818e-05,
|
| 5048 |
+
"loss": 0.4987,
|
| 5049 |
+
"step": 840
|
| 5050 |
+
},
|
| 5051 |
+
{
|
| 5052 |
+
"epoch": 9.5,
|
| 5053 |
+
"learning_rate": 8.053977272727274e-05,
|
| 5054 |
+
"loss": 0.5063,
|
| 5055 |
+
"step": 841
|
| 5056 |
+
},
|
| 5057 |
+
{
|
| 5058 |
+
"epoch": 9.51,
|
| 5059 |
+
"learning_rate": 8.039772727272728e-05,
|
| 5060 |
+
"loss": 0.5019,
|
| 5061 |
+
"step": 842
|
| 5062 |
+
},
|
| 5063 |
+
{
|
| 5064 |
+
"epoch": 9.52,
|
| 5065 |
+
"learning_rate": 8.025568181818183e-05,
|
| 5066 |
+
"loss": 0.5272,
|
| 5067 |
+
"step": 843
|
| 5068 |
+
},
|
| 5069 |
+
{
|
| 5070 |
+
"epoch": 9.54,
|
| 5071 |
+
"learning_rate": 8.011363636363637e-05,
|
| 5072 |
+
"loss": 0.4969,
|
| 5073 |
+
"step": 844
|
| 5074 |
+
},
|
| 5075 |
+
{
|
| 5076 |
+
"epoch": 9.55,
|
| 5077 |
+
"learning_rate": 7.997159090909092e-05,
|
| 5078 |
+
"loss": 0.5222,
|
| 5079 |
+
"step": 845
|
| 5080 |
+
},
|
| 5081 |
+
{
|
| 5082 |
+
"epoch": 9.56,
|
| 5083 |
+
"learning_rate": 7.982954545454546e-05,
|
| 5084 |
+
"loss": 0.4729,
|
| 5085 |
+
"step": 846
|
| 5086 |
+
},
|
| 5087 |
+
{
|
| 5088 |
+
"epoch": 9.57,
|
| 5089 |
+
"learning_rate": 7.96875e-05,
|
| 5090 |
+
"loss": 0.4976,
|
| 5091 |
+
"step": 847
|
| 5092 |
+
},
|
| 5093 |
+
{
|
| 5094 |
+
"epoch": 9.58,
|
| 5095 |
+
"learning_rate": 7.954545454545455e-05,
|
| 5096 |
+
"loss": 0.4974,
|
| 5097 |
+
"step": 848
|
| 5098 |
+
},
|
| 5099 |
+
{
|
| 5100 |
+
"epoch": 9.59,
|
| 5101 |
+
"learning_rate": 7.94034090909091e-05,
|
| 5102 |
+
"loss": 0.4849,
|
| 5103 |
+
"step": 849
|
| 5104 |
+
},
|
| 5105 |
+
{
|
| 5106 |
+
"epoch": 9.6,
|
| 5107 |
+
"learning_rate": 7.926136363636364e-05,
|
| 5108 |
+
"loss": 0.4897,
|
| 5109 |
+
"step": 850
|
| 5110 |
+
},
|
| 5111 |
+
{
|
| 5112 |
+
"epoch": 9.61,
|
| 5113 |
+
"learning_rate": 7.911931818181819e-05,
|
| 5114 |
+
"loss": 0.4962,
|
| 5115 |
+
"step": 851
|
| 5116 |
+
},
|
| 5117 |
+
{
|
| 5118 |
+
"epoch": 9.63,
|
| 5119 |
+
"learning_rate": 7.897727272727273e-05,
|
| 5120 |
+
"loss": 0.4877,
|
| 5121 |
+
"step": 852
|
| 5122 |
+
},
|
| 5123 |
+
{
|
| 5124 |
+
"epoch": 9.64,
|
| 5125 |
+
"learning_rate": 7.883522727272728e-05,
|
| 5126 |
+
"loss": 0.4921,
|
| 5127 |
+
"step": 853
|
| 5128 |
+
},
|
| 5129 |
+
{
|
| 5130 |
+
"epoch": 9.65,
|
| 5131 |
+
"learning_rate": 7.869318181818182e-05,
|
| 5132 |
+
"loss": 0.4969,
|
| 5133 |
+
"step": 854
|
| 5134 |
+
},
|
| 5135 |
+
{
|
| 5136 |
+
"epoch": 9.66,
|
| 5137 |
+
"learning_rate": 7.855113636363637e-05,
|
| 5138 |
+
"loss": 0.5045,
|
| 5139 |
+
"step": 855
|
| 5140 |
+
},
|
| 5141 |
+
{
|
| 5142 |
+
"epoch": 9.67,
|
| 5143 |
+
"learning_rate": 7.840909090909091e-05,
|
| 5144 |
+
"loss": 0.5207,
|
| 5145 |
+
"step": 856
|
| 5146 |
+
},
|
| 5147 |
+
{
|
| 5148 |
+
"epoch": 9.68,
|
| 5149 |
+
"learning_rate": 7.826704545454546e-05,
|
| 5150 |
+
"loss": 0.5098,
|
| 5151 |
+
"step": 857
|
| 5152 |
+
},
|
| 5153 |
+
{
|
| 5154 |
+
"epoch": 9.69,
|
| 5155 |
+
"learning_rate": 7.8125e-05,
|
| 5156 |
+
"loss": 0.5005,
|
| 5157 |
+
"step": 858
|
| 5158 |
+
},
|
| 5159 |
+
{
|
| 5160 |
+
"epoch": 9.7,
|
| 5161 |
+
"learning_rate": 7.798295454545455e-05,
|
| 5162 |
+
"loss": 0.5028,
|
| 5163 |
+
"step": 859
|
| 5164 |
+
},
|
| 5165 |
+
{
|
| 5166 |
+
"epoch": 9.72,
|
| 5167 |
+
"learning_rate": 7.784090909090909e-05,
|
| 5168 |
+
"loss": 0.5067,
|
| 5169 |
+
"step": 860
|
| 5170 |
+
},
|
| 5171 |
+
{
|
| 5172 |
+
"epoch": 9.73,
|
| 5173 |
+
"learning_rate": 7.769886363636364e-05,
|
| 5174 |
+
"loss": 0.484,
|
| 5175 |
+
"step": 861
|
| 5176 |
+
},
|
| 5177 |
+
{
|
| 5178 |
+
"epoch": 9.74,
|
| 5179 |
+
"learning_rate": 7.755681818181818e-05,
|
| 5180 |
+
"loss": 0.5029,
|
| 5181 |
+
"step": 862
|
| 5182 |
+
},
|
| 5183 |
+
{
|
| 5184 |
+
"epoch": 9.75,
|
| 5185 |
+
"learning_rate": 7.741477272727273e-05,
|
| 5186 |
+
"loss": 0.5077,
|
| 5187 |
+
"step": 863
|
| 5188 |
+
},
|
| 5189 |
+
{
|
| 5190 |
+
"epoch": 9.76,
|
| 5191 |
+
"learning_rate": 7.727272727272727e-05,
|
| 5192 |
+
"loss": 0.5091,
|
| 5193 |
+
"step": 864
|
| 5194 |
+
},
|
| 5195 |
+
{
|
| 5196 |
+
"epoch": 9.77,
|
| 5197 |
+
"learning_rate": 7.713068181818183e-05,
|
| 5198 |
+
"loss": 0.4781,
|
| 5199 |
+
"step": 865
|
| 5200 |
+
},
|
| 5201 |
+
{
|
| 5202 |
+
"epoch": 9.78,
|
| 5203 |
+
"learning_rate": 7.698863636363636e-05,
|
| 5204 |
+
"loss": 0.5124,
|
| 5205 |
+
"step": 866
|
| 5206 |
+
},
|
| 5207 |
+
{
|
| 5208 |
+
"epoch": 9.79,
|
| 5209 |
+
"learning_rate": 7.684659090909091e-05,
|
| 5210 |
+
"loss": 0.4859,
|
| 5211 |
+
"step": 867
|
| 5212 |
+
},
|
| 5213 |
+
{
|
| 5214 |
+
"epoch": 9.81,
|
| 5215 |
+
"learning_rate": 7.670454545454547e-05,
|
| 5216 |
+
"loss": 0.4872,
|
| 5217 |
+
"step": 868
|
| 5218 |
+
},
|
| 5219 |
+
{
|
| 5220 |
+
"epoch": 9.82,
|
| 5221 |
+
"learning_rate": 7.65625e-05,
|
| 5222 |
+
"loss": 0.4675,
|
| 5223 |
+
"step": 869
|
| 5224 |
+
},
|
| 5225 |
+
{
|
| 5226 |
+
"epoch": 9.83,
|
| 5227 |
+
"learning_rate": 7.642045454545454e-05,
|
| 5228 |
+
"loss": 0.5056,
|
| 5229 |
+
"step": 870
|
| 5230 |
+
},
|
| 5231 |
+
{
|
| 5232 |
+
"epoch": 9.84,
|
| 5233 |
+
"learning_rate": 7.62784090909091e-05,
|
| 5234 |
+
"loss": 0.4868,
|
| 5235 |
+
"step": 871
|
| 5236 |
+
},
|
| 5237 |
+
{
|
| 5238 |
+
"epoch": 9.85,
|
| 5239 |
+
"learning_rate": 7.613636363636363e-05,
|
| 5240 |
+
"loss": 0.4907,
|
| 5241 |
+
"step": 872
|
| 5242 |
+
},
|
| 5243 |
+
{
|
| 5244 |
+
"epoch": 9.86,
|
| 5245 |
+
"learning_rate": 7.599431818181818e-05,
|
| 5246 |
+
"loss": 0.474,
|
| 5247 |
+
"step": 873
|
| 5248 |
+
},
|
| 5249 |
+
{
|
| 5250 |
+
"epoch": 9.87,
|
| 5251 |
+
"learning_rate": 7.585227272727274e-05,
|
| 5252 |
+
"loss": 0.4813,
|
| 5253 |
+
"step": 874
|
| 5254 |
+
},
|
| 5255 |
+
{
|
| 5256 |
+
"epoch": 9.89,
|
| 5257 |
+
"learning_rate": 7.571022727272727e-05,
|
| 5258 |
+
"loss": 0.4838,
|
| 5259 |
+
"step": 875
|
| 5260 |
+
},
|
| 5261 |
+
{
|
| 5262 |
+
"epoch": 9.9,
|
| 5263 |
+
"learning_rate": 7.556818181818183e-05,
|
| 5264 |
+
"loss": 0.4935,
|
| 5265 |
+
"step": 876
|
| 5266 |
+
},
|
| 5267 |
+
{
|
| 5268 |
+
"epoch": 9.91,
|
| 5269 |
+
"learning_rate": 7.542613636363637e-05,
|
| 5270 |
+
"loss": 0.4884,
|
| 5271 |
+
"step": 877
|
| 5272 |
+
},
|
| 5273 |
+
{
|
| 5274 |
+
"epoch": 9.92,
|
| 5275 |
+
"learning_rate": 7.52840909090909e-05,
|
| 5276 |
+
"loss": 0.4797,
|
| 5277 |
+
"step": 878
|
| 5278 |
+
},
|
| 5279 |
+
{
|
| 5280 |
+
"epoch": 9.93,
|
| 5281 |
+
"learning_rate": 7.514204545454546e-05,
|
| 5282 |
+
"loss": 0.479,
|
| 5283 |
+
"step": 879
|
| 5284 |
+
},
|
| 5285 |
+
{
|
| 5286 |
+
"epoch": 9.94,
|
| 5287 |
+
"learning_rate": 7.500000000000001e-05,
|
| 5288 |
+
"loss": 0.4727,
|
| 5289 |
+
"step": 880
|
| 5290 |
+
},
|
| 5291 |
+
{
|
| 5292 |
+
"epoch": 9.95,
|
| 5293 |
+
"learning_rate": 7.485795454545454e-05,
|
| 5294 |
+
"loss": 0.4758,
|
| 5295 |
+
"step": 881
|
| 5296 |
+
},
|
| 5297 |
+
{
|
| 5298 |
+
"epoch": 9.96,
|
| 5299 |
+
"learning_rate": 7.47159090909091e-05,
|
| 5300 |
+
"loss": 0.482,
|
| 5301 |
+
"step": 882
|
| 5302 |
+
},
|
| 5303 |
+
{
|
| 5304 |
+
"epoch": 9.98,
|
| 5305 |
+
"learning_rate": 7.457386363636364e-05,
|
| 5306 |
+
"loss": 0.4951,
|
| 5307 |
+
"step": 883
|
| 5308 |
+
},
|
| 5309 |
+
{
|
| 5310 |
+
"epoch": 9.99,
|
| 5311 |
+
"learning_rate": 7.443181818181817e-05,
|
| 5312 |
+
"loss": 0.4823,
|
| 5313 |
+
"step": 884
|
| 5314 |
+
},
|
| 5315 |
+
{
|
| 5316 |
+
"epoch": 10.0,
|
| 5317 |
+
"learning_rate": 7.428977272727273e-05,
|
| 5318 |
+
"loss": 0.4638,
|
| 5319 |
+
"step": 885
|
| 5320 |
+
},
|
| 5321 |
+
{
|
| 5322 |
+
"epoch": 10.01,
|
| 5323 |
+
"learning_rate": 7.414772727272728e-05,
|
| 5324 |
+
"loss": 0.4715,
|
| 5325 |
+
"step": 886
|
| 5326 |
+
},
|
| 5327 |
+
{
|
| 5328 |
+
"epoch": 10.02,
|
| 5329 |
+
"learning_rate": 7.400568181818182e-05,
|
| 5330 |
+
"loss": 0.461,
|
| 5331 |
+
"step": 887
|
| 5332 |
+
},
|
| 5333 |
+
{
|
| 5334 |
+
"epoch": 10.03,
|
| 5335 |
+
"learning_rate": 7.386363636363637e-05,
|
| 5336 |
+
"loss": 0.4429,
|
| 5337 |
+
"step": 888
|
| 5338 |
+
},
|
| 5339 |
+
{
|
| 5340 |
+
"epoch": 10.04,
|
| 5341 |
+
"learning_rate": 7.372159090909091e-05,
|
| 5342 |
+
"loss": 0.4403,
|
| 5343 |
+
"step": 889
|
| 5344 |
+
},
|
| 5345 |
+
{
|
| 5346 |
+
"epoch": 10.05,
|
| 5347 |
+
"learning_rate": 7.357954545454546e-05,
|
| 5348 |
+
"loss": 0.4519,
|
| 5349 |
+
"step": 890
|
| 5350 |
+
},
|
| 5351 |
+
{
|
| 5352 |
+
"epoch": 10.07,
|
| 5353 |
+
"learning_rate": 7.34375e-05,
|
| 5354 |
+
"loss": 0.4611,
|
| 5355 |
+
"step": 891
|
| 5356 |
+
},
|
| 5357 |
+
{
|
| 5358 |
+
"epoch": 10.08,
|
| 5359 |
+
"learning_rate": 7.329545454545455e-05,
|
| 5360 |
+
"loss": 0.4543,
|
| 5361 |
+
"step": 892
|
| 5362 |
+
},
|
| 5363 |
+
{
|
| 5364 |
+
"epoch": 10.09,
|
| 5365 |
+
"learning_rate": 7.315340909090909e-05,
|
| 5366 |
+
"loss": 0.4528,
|
| 5367 |
+
"step": 893
|
| 5368 |
+
},
|
| 5369 |
+
{
|
| 5370 |
+
"epoch": 10.1,
|
| 5371 |
+
"learning_rate": 7.301136363636364e-05,
|
| 5372 |
+
"loss": 0.4586,
|
| 5373 |
+
"step": 894
|
| 5374 |
+
},
|
| 5375 |
+
{
|
| 5376 |
+
"epoch": 10.11,
|
| 5377 |
+
"learning_rate": 7.286931818181818e-05,
|
| 5378 |
+
"loss": 0.4418,
|
| 5379 |
+
"step": 895
|
| 5380 |
+
},
|
| 5381 |
+
{
|
| 5382 |
+
"epoch": 10.12,
|
| 5383 |
+
"learning_rate": 7.272727272727273e-05,
|
| 5384 |
+
"loss": 0.4435,
|
| 5385 |
+
"step": 896
|
| 5386 |
+
},
|
| 5387 |
+
{
|
| 5388 |
+
"epoch": 10.13,
|
| 5389 |
+
"learning_rate": 7.258522727272727e-05,
|
| 5390 |
+
"loss": 0.44,
|
| 5391 |
+
"step": 897
|
| 5392 |
+
},
|
| 5393 |
+
{
|
| 5394 |
+
"epoch": 10.15,
|
| 5395 |
+
"learning_rate": 7.244318181818183e-05,
|
| 5396 |
+
"loss": 0.4589,
|
| 5397 |
+
"step": 898
|
| 5398 |
+
},
|
| 5399 |
+
{
|
| 5400 |
+
"epoch": 10.16,
|
| 5401 |
+
"learning_rate": 7.230113636363636e-05,
|
| 5402 |
+
"loss": 0.4597,
|
| 5403 |
+
"step": 899
|
| 5404 |
+
},
|
| 5405 |
+
{
|
| 5406 |
+
"epoch": 10.17,
|
| 5407 |
+
"learning_rate": 7.215909090909091e-05,
|
| 5408 |
+
"loss": 0.4479,
|
| 5409 |
+
"step": 900
|
| 5410 |
}
|
| 5411 |
],
|
| 5412 |
"logging_steps": 1,
|
| 5413 |
"max_steps": 1408,
|
| 5414 |
"num_train_epochs": 16,
|
| 5415 |
"save_steps": 100,
|
| 5416 |
+
"total_flos": 1.227986631604562e+18,
|
| 5417 |
"trial_name": null,
|
| 5418 |
"trial_params": null
|
| 5419 |
}
|