Commit
•
28812b3
1
Parent(s):
dc202b3
Upload folder using huggingface_hub (#5)
Browse files- Upload folder using huggingface_hub (03d09198311838452509ba48c594dcdb0f40d044)
- 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 sha256:83a2bd54ab68a3dcad8fe0e5a630e45f5342750c381fd29ab4100c5db3d4e0e9
|
| 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 sha256:81ef498b8b8b31a2a5c36ffe3529e1378029eb4c3e4ba4770de0c248e4c62950
|
| 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 sha256:36fc71bd44bd7f04f2599c5dface64c517de1a7ab7bac3600f3f6470c6c72673
|
| 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 sha256:7c6badaa6737fa9306a1f25b19c54f91672e90aa94d505b95ab467e7d08dd541
|
| 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,
|
|
@@ -3007,13 +3007,1213 @@
|
|
| 3007 |
"learning_rate": 0.00012897727272727274,
|
| 3008 |
"loss": 1.0534,
|
| 3009 |
"step": 500
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3010 |
}
|
| 3011 |
],
|
| 3012 |
"logging_steps": 1,
|
| 3013 |
"max_steps": 1408,
|
| 3014 |
"num_train_epochs": 16,
|
| 3015 |
"save_steps": 100,
|
| 3016 |
-
"total_flos":
|
| 3017 |
"trial_name": null,
|
| 3018 |
"trial_params": null
|
| 3019 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 7.908208296557811,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 700,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 3007 |
"learning_rate": 0.00012897727272727274,
|
| 3008 |
"loss": 1.0534,
|
| 3009 |
"step": 500
|
| 3010 |
+
},
|
| 3011 |
+
{
|
| 3012 |
+
"epoch": 5.66,
|
| 3013 |
+
"learning_rate": 0.00012883522727272727,
|
| 3014 |
+
"loss": 1.0755,
|
| 3015 |
+
"step": 501
|
| 3016 |
+
},
|
| 3017 |
+
{
|
| 3018 |
+
"epoch": 5.67,
|
| 3019 |
+
"learning_rate": 0.00012869318181818183,
|
| 3020 |
+
"loss": 1.0755,
|
| 3021 |
+
"step": 502
|
| 3022 |
+
},
|
| 3023 |
+
{
|
| 3024 |
+
"epoch": 5.68,
|
| 3025 |
+
"learning_rate": 0.00012855113636363636,
|
| 3026 |
+
"loss": 1.0869,
|
| 3027 |
+
"step": 503
|
| 3028 |
+
},
|
| 3029 |
+
{
|
| 3030 |
+
"epoch": 5.69,
|
| 3031 |
+
"learning_rate": 0.00012840909090909092,
|
| 3032 |
+
"loss": 1.0673,
|
| 3033 |
+
"step": 504
|
| 3034 |
+
},
|
| 3035 |
+
{
|
| 3036 |
+
"epoch": 5.71,
|
| 3037 |
+
"learning_rate": 0.00012826704545454545,
|
| 3038 |
+
"loss": 1.0692,
|
| 3039 |
+
"step": 505
|
| 3040 |
+
},
|
| 3041 |
+
{
|
| 3042 |
+
"epoch": 5.72,
|
| 3043 |
+
"learning_rate": 0.000128125,
|
| 3044 |
+
"loss": 1.0474,
|
| 3045 |
+
"step": 506
|
| 3046 |
+
},
|
| 3047 |
+
{
|
| 3048 |
+
"epoch": 5.73,
|
| 3049 |
+
"learning_rate": 0.00012798295454545454,
|
| 3050 |
+
"loss": 1.0749,
|
| 3051 |
+
"step": 507
|
| 3052 |
+
},
|
| 3053 |
+
{
|
| 3054 |
+
"epoch": 5.74,
|
| 3055 |
+
"learning_rate": 0.0001278409090909091,
|
| 3056 |
+
"loss": 1.0519,
|
| 3057 |
+
"step": 508
|
| 3058 |
+
},
|
| 3059 |
+
{
|
| 3060 |
+
"epoch": 5.75,
|
| 3061 |
+
"learning_rate": 0.00012769886363636366,
|
| 3062 |
+
"loss": 1.0566,
|
| 3063 |
+
"step": 509
|
| 3064 |
+
},
|
| 3065 |
+
{
|
| 3066 |
+
"epoch": 5.76,
|
| 3067 |
+
"learning_rate": 0.0001275568181818182,
|
| 3068 |
+
"loss": 1.06,
|
| 3069 |
+
"step": 510
|
| 3070 |
+
},
|
| 3071 |
+
{
|
| 3072 |
+
"epoch": 5.77,
|
| 3073 |
+
"learning_rate": 0.00012741477272727272,
|
| 3074 |
+
"loss": 1.0618,
|
| 3075 |
+
"step": 511
|
| 3076 |
+
},
|
| 3077 |
+
{
|
| 3078 |
+
"epoch": 5.78,
|
| 3079 |
+
"learning_rate": 0.00012727272727272728,
|
| 3080 |
+
"loss": 1.0643,
|
| 3081 |
+
"step": 512
|
| 3082 |
+
},
|
| 3083 |
+
{
|
| 3084 |
+
"epoch": 5.8,
|
| 3085 |
+
"learning_rate": 0.0001271306818181818,
|
| 3086 |
+
"loss": 1.026,
|
| 3087 |
+
"step": 513
|
| 3088 |
+
},
|
| 3089 |
+
{
|
| 3090 |
+
"epoch": 5.81,
|
| 3091 |
+
"learning_rate": 0.00012698863636363637,
|
| 3092 |
+
"loss": 1.0335,
|
| 3093 |
+
"step": 514
|
| 3094 |
+
},
|
| 3095 |
+
{
|
| 3096 |
+
"epoch": 5.82,
|
| 3097 |
+
"learning_rate": 0.00012684659090909093,
|
| 3098 |
+
"loss": 1.0205,
|
| 3099 |
+
"step": 515
|
| 3100 |
+
},
|
| 3101 |
+
{
|
| 3102 |
+
"epoch": 5.83,
|
| 3103 |
+
"learning_rate": 0.00012670454545454546,
|
| 3104 |
+
"loss": 1.0594,
|
| 3105 |
+
"step": 516
|
| 3106 |
+
},
|
| 3107 |
+
{
|
| 3108 |
+
"epoch": 5.84,
|
| 3109 |
+
"learning_rate": 0.0001265625,
|
| 3110 |
+
"loss": 1.0136,
|
| 3111 |
+
"step": 517
|
| 3112 |
+
},
|
| 3113 |
+
{
|
| 3114 |
+
"epoch": 5.85,
|
| 3115 |
+
"learning_rate": 0.00012642045454545455,
|
| 3116 |
+
"loss": 1.0244,
|
| 3117 |
+
"step": 518
|
| 3118 |
+
},
|
| 3119 |
+
{
|
| 3120 |
+
"epoch": 5.86,
|
| 3121 |
+
"learning_rate": 0.00012627840909090908,
|
| 3122 |
+
"loss": 1.0569,
|
| 3123 |
+
"step": 519
|
| 3124 |
+
},
|
| 3125 |
+
{
|
| 3126 |
+
"epoch": 5.87,
|
| 3127 |
+
"learning_rate": 0.00012613636363636364,
|
| 3128 |
+
"loss": 1.0416,
|
| 3129 |
+
"step": 520
|
| 3130 |
+
},
|
| 3131 |
+
{
|
| 3132 |
+
"epoch": 5.89,
|
| 3133 |
+
"learning_rate": 0.0001259943181818182,
|
| 3134 |
+
"loss": 0.9884,
|
| 3135 |
+
"step": 521
|
| 3136 |
+
},
|
| 3137 |
+
{
|
| 3138 |
+
"epoch": 5.9,
|
| 3139 |
+
"learning_rate": 0.00012585227272727273,
|
| 3140 |
+
"loss": 1.0351,
|
| 3141 |
+
"step": 522
|
| 3142 |
+
},
|
| 3143 |
+
{
|
| 3144 |
+
"epoch": 5.91,
|
| 3145 |
+
"learning_rate": 0.00012571022727272726,
|
| 3146 |
+
"loss": 1.0037,
|
| 3147 |
+
"step": 523
|
| 3148 |
+
},
|
| 3149 |
+
{
|
| 3150 |
+
"epoch": 5.92,
|
| 3151 |
+
"learning_rate": 0.00012556818181818182,
|
| 3152 |
+
"loss": 1.0219,
|
| 3153 |
+
"step": 524
|
| 3154 |
+
},
|
| 3155 |
+
{
|
| 3156 |
+
"epoch": 5.93,
|
| 3157 |
+
"learning_rate": 0.00012542613636363635,
|
| 3158 |
+
"loss": 1.0533,
|
| 3159 |
+
"step": 525
|
| 3160 |
+
},
|
| 3161 |
+
{
|
| 3162 |
+
"epoch": 5.94,
|
| 3163 |
+
"learning_rate": 0.0001252840909090909,
|
| 3164 |
+
"loss": 1.0031,
|
| 3165 |
+
"step": 526
|
| 3166 |
+
},
|
| 3167 |
+
{
|
| 3168 |
+
"epoch": 5.95,
|
| 3169 |
+
"learning_rate": 0.00012514204545454547,
|
| 3170 |
+
"loss": 1.0454,
|
| 3171 |
+
"step": 527
|
| 3172 |
+
},
|
| 3173 |
+
{
|
| 3174 |
+
"epoch": 5.97,
|
| 3175 |
+
"learning_rate": 0.000125,
|
| 3176 |
+
"loss": 1.0195,
|
| 3177 |
+
"step": 528
|
| 3178 |
+
},
|
| 3179 |
+
{
|
| 3180 |
+
"epoch": 5.98,
|
| 3181 |
+
"learning_rate": 0.00012485795454545453,
|
| 3182 |
+
"loss": 1.0076,
|
| 3183 |
+
"step": 529
|
| 3184 |
+
},
|
| 3185 |
+
{
|
| 3186 |
+
"epoch": 5.99,
|
| 3187 |
+
"learning_rate": 0.0001247159090909091,
|
| 3188 |
+
"loss": 1.0378,
|
| 3189 |
+
"step": 530
|
| 3190 |
+
},
|
| 3191 |
+
{
|
| 3192 |
+
"epoch": 6.0,
|
| 3193 |
+
"learning_rate": 0.00012457386363636365,
|
| 3194 |
+
"loss": 0.9795,
|
| 3195 |
+
"step": 531
|
| 3196 |
+
},
|
| 3197 |
+
{
|
| 3198 |
+
"epoch": 6.01,
|
| 3199 |
+
"learning_rate": 0.00012443181818181818,
|
| 3200 |
+
"loss": 0.9405,
|
| 3201 |
+
"step": 532
|
| 3202 |
+
},
|
| 3203 |
+
{
|
| 3204 |
+
"epoch": 6.02,
|
| 3205 |
+
"learning_rate": 0.00012428977272727274,
|
| 3206 |
+
"loss": 0.9503,
|
| 3207 |
+
"step": 533
|
| 3208 |
+
},
|
| 3209 |
+
{
|
| 3210 |
+
"epoch": 6.03,
|
| 3211 |
+
"learning_rate": 0.00012414772727272727,
|
| 3212 |
+
"loss": 0.9456,
|
| 3213 |
+
"step": 534
|
| 3214 |
+
},
|
| 3215 |
+
{
|
| 3216 |
+
"epoch": 6.04,
|
| 3217 |
+
"learning_rate": 0.0001240056818181818,
|
| 3218 |
+
"loss": 0.9536,
|
| 3219 |
+
"step": 535
|
| 3220 |
+
},
|
| 3221 |
+
{
|
| 3222 |
+
"epoch": 6.06,
|
| 3223 |
+
"learning_rate": 0.00012386363636363636,
|
| 3224 |
+
"loss": 0.9412,
|
| 3225 |
+
"step": 536
|
| 3226 |
+
},
|
| 3227 |
+
{
|
| 3228 |
+
"epoch": 6.07,
|
| 3229 |
+
"learning_rate": 0.00012372159090909092,
|
| 3230 |
+
"loss": 0.9315,
|
| 3231 |
+
"step": 537
|
| 3232 |
+
},
|
| 3233 |
+
{
|
| 3234 |
+
"epoch": 6.08,
|
| 3235 |
+
"learning_rate": 0.00012357954545454545,
|
| 3236 |
+
"loss": 0.9486,
|
| 3237 |
+
"step": 538
|
| 3238 |
+
},
|
| 3239 |
+
{
|
| 3240 |
+
"epoch": 6.09,
|
| 3241 |
+
"learning_rate": 0.0001234375,
|
| 3242 |
+
"loss": 0.9405,
|
| 3243 |
+
"step": 539
|
| 3244 |
+
},
|
| 3245 |
+
{
|
| 3246 |
+
"epoch": 6.1,
|
| 3247 |
+
"learning_rate": 0.00012329545454545454,
|
| 3248 |
+
"loss": 0.9269,
|
| 3249 |
+
"step": 540
|
| 3250 |
+
},
|
| 3251 |
+
{
|
| 3252 |
+
"epoch": 6.11,
|
| 3253 |
+
"learning_rate": 0.0001231534090909091,
|
| 3254 |
+
"loss": 0.9378,
|
| 3255 |
+
"step": 541
|
| 3256 |
+
},
|
| 3257 |
+
{
|
| 3258 |
+
"epoch": 6.12,
|
| 3259 |
+
"learning_rate": 0.00012301136363636366,
|
| 3260 |
+
"loss": 0.9431,
|
| 3261 |
+
"step": 542
|
| 3262 |
+
},
|
| 3263 |
+
{
|
| 3264 |
+
"epoch": 6.13,
|
| 3265 |
+
"learning_rate": 0.0001228693181818182,
|
| 3266 |
+
"loss": 0.9256,
|
| 3267 |
+
"step": 543
|
| 3268 |
+
},
|
| 3269 |
+
{
|
| 3270 |
+
"epoch": 6.15,
|
| 3271 |
+
"learning_rate": 0.00012272727272727272,
|
| 3272 |
+
"loss": 0.919,
|
| 3273 |
+
"step": 544
|
| 3274 |
+
},
|
| 3275 |
+
{
|
| 3276 |
+
"epoch": 6.16,
|
| 3277 |
+
"learning_rate": 0.00012258522727272728,
|
| 3278 |
+
"loss": 0.9188,
|
| 3279 |
+
"step": 545
|
| 3280 |
+
},
|
| 3281 |
+
{
|
| 3282 |
+
"epoch": 6.17,
|
| 3283 |
+
"learning_rate": 0.00012244318181818181,
|
| 3284 |
+
"loss": 0.9447,
|
| 3285 |
+
"step": 546
|
| 3286 |
+
},
|
| 3287 |
+
{
|
| 3288 |
+
"epoch": 6.18,
|
| 3289 |
+
"learning_rate": 0.00012230113636363637,
|
| 3290 |
+
"loss": 0.9261,
|
| 3291 |
+
"step": 547
|
| 3292 |
+
},
|
| 3293 |
+
{
|
| 3294 |
+
"epoch": 6.19,
|
| 3295 |
+
"learning_rate": 0.00012215909090909093,
|
| 3296 |
+
"loss": 0.9302,
|
| 3297 |
+
"step": 548
|
| 3298 |
+
},
|
| 3299 |
+
{
|
| 3300 |
+
"epoch": 6.2,
|
| 3301 |
+
"learning_rate": 0.00012201704545454546,
|
| 3302 |
+
"loss": 0.9161,
|
| 3303 |
+
"step": 549
|
| 3304 |
+
},
|
| 3305 |
+
{
|
| 3306 |
+
"epoch": 6.21,
|
| 3307 |
+
"learning_rate": 0.00012187500000000001,
|
| 3308 |
+
"loss": 0.9521,
|
| 3309 |
+
"step": 550
|
| 3310 |
+
},
|
| 3311 |
+
{
|
| 3312 |
+
"epoch": 6.22,
|
| 3313 |
+
"learning_rate": 0.00012173295454545455,
|
| 3314 |
+
"loss": 0.9026,
|
| 3315 |
+
"step": 551
|
| 3316 |
+
},
|
| 3317 |
+
{
|
| 3318 |
+
"epoch": 6.24,
|
| 3319 |
+
"learning_rate": 0.00012159090909090908,
|
| 3320 |
+
"loss": 0.9361,
|
| 3321 |
+
"step": 552
|
| 3322 |
+
},
|
| 3323 |
+
{
|
| 3324 |
+
"epoch": 6.25,
|
| 3325 |
+
"learning_rate": 0.00012144886363636366,
|
| 3326 |
+
"loss": 0.8944,
|
| 3327 |
+
"step": 553
|
| 3328 |
+
},
|
| 3329 |
+
{
|
| 3330 |
+
"epoch": 6.26,
|
| 3331 |
+
"learning_rate": 0.00012130681818181819,
|
| 3332 |
+
"loss": 0.895,
|
| 3333 |
+
"step": 554
|
| 3334 |
+
},
|
| 3335 |
+
{
|
| 3336 |
+
"epoch": 6.27,
|
| 3337 |
+
"learning_rate": 0.00012116477272727273,
|
| 3338 |
+
"loss": 0.8956,
|
| 3339 |
+
"step": 555
|
| 3340 |
+
},
|
| 3341 |
+
{
|
| 3342 |
+
"epoch": 6.28,
|
| 3343 |
+
"learning_rate": 0.00012102272727272728,
|
| 3344 |
+
"loss": 0.8998,
|
| 3345 |
+
"step": 556
|
| 3346 |
+
},
|
| 3347 |
+
{
|
| 3348 |
+
"epoch": 6.29,
|
| 3349 |
+
"learning_rate": 0.00012088068181818182,
|
| 3350 |
+
"loss": 0.915,
|
| 3351 |
+
"step": 557
|
| 3352 |
+
},
|
| 3353 |
+
{
|
| 3354 |
+
"epoch": 6.3,
|
| 3355 |
+
"learning_rate": 0.00012073863636363636,
|
| 3356 |
+
"loss": 0.9282,
|
| 3357 |
+
"step": 558
|
| 3358 |
+
},
|
| 3359 |
+
{
|
| 3360 |
+
"epoch": 6.32,
|
| 3361 |
+
"learning_rate": 0.00012059659090909093,
|
| 3362 |
+
"loss": 0.8938,
|
| 3363 |
+
"step": 559
|
| 3364 |
+
},
|
| 3365 |
+
{
|
| 3366 |
+
"epoch": 6.33,
|
| 3367 |
+
"learning_rate": 0.00012045454545454546,
|
| 3368 |
+
"loss": 0.8886,
|
| 3369 |
+
"step": 560
|
| 3370 |
+
},
|
| 3371 |
+
{
|
| 3372 |
+
"epoch": 6.34,
|
| 3373 |
+
"learning_rate": 0.0001203125,
|
| 3374 |
+
"loss": 0.8988,
|
| 3375 |
+
"step": 561
|
| 3376 |
+
},
|
| 3377 |
+
{
|
| 3378 |
+
"epoch": 6.35,
|
| 3379 |
+
"learning_rate": 0.00012017045454545455,
|
| 3380 |
+
"loss": 0.8852,
|
| 3381 |
+
"step": 562
|
| 3382 |
+
},
|
| 3383 |
+
{
|
| 3384 |
+
"epoch": 6.36,
|
| 3385 |
+
"learning_rate": 0.0001200284090909091,
|
| 3386 |
+
"loss": 0.8818,
|
| 3387 |
+
"step": 563
|
| 3388 |
+
},
|
| 3389 |
+
{
|
| 3390 |
+
"epoch": 6.37,
|
| 3391 |
+
"learning_rate": 0.00011988636363636365,
|
| 3392 |
+
"loss": 0.8881,
|
| 3393 |
+
"step": 564
|
| 3394 |
+
},
|
| 3395 |
+
{
|
| 3396 |
+
"epoch": 6.38,
|
| 3397 |
+
"learning_rate": 0.0001197443181818182,
|
| 3398 |
+
"loss": 0.9226,
|
| 3399 |
+
"step": 565
|
| 3400 |
+
},
|
| 3401 |
+
{
|
| 3402 |
+
"epoch": 6.39,
|
| 3403 |
+
"learning_rate": 0.00011960227272727273,
|
| 3404 |
+
"loss": 0.8849,
|
| 3405 |
+
"step": 566
|
| 3406 |
+
},
|
| 3407 |
+
{
|
| 3408 |
+
"epoch": 6.41,
|
| 3409 |
+
"learning_rate": 0.00011946022727272727,
|
| 3410 |
+
"loss": 0.8894,
|
| 3411 |
+
"step": 567
|
| 3412 |
+
},
|
| 3413 |
+
{
|
| 3414 |
+
"epoch": 6.42,
|
| 3415 |
+
"learning_rate": 0.00011931818181818182,
|
| 3416 |
+
"loss": 0.9207,
|
| 3417 |
+
"step": 568
|
| 3418 |
+
},
|
| 3419 |
+
{
|
| 3420 |
+
"epoch": 6.43,
|
| 3421 |
+
"learning_rate": 0.00011917613636363636,
|
| 3422 |
+
"loss": 0.9105,
|
| 3423 |
+
"step": 569
|
| 3424 |
+
},
|
| 3425 |
+
{
|
| 3426 |
+
"epoch": 6.44,
|
| 3427 |
+
"learning_rate": 0.00011903409090909092,
|
| 3428 |
+
"loss": 0.8762,
|
| 3429 |
+
"step": 570
|
| 3430 |
+
},
|
| 3431 |
+
{
|
| 3432 |
+
"epoch": 6.45,
|
| 3433 |
+
"learning_rate": 0.00011889204545454547,
|
| 3434 |
+
"loss": 0.8926,
|
| 3435 |
+
"step": 571
|
| 3436 |
+
},
|
| 3437 |
+
{
|
| 3438 |
+
"epoch": 6.46,
|
| 3439 |
+
"learning_rate": 0.00011875,
|
| 3440 |
+
"loss": 0.8719,
|
| 3441 |
+
"step": 572
|
| 3442 |
+
},
|
| 3443 |
+
{
|
| 3444 |
+
"epoch": 6.47,
|
| 3445 |
+
"learning_rate": 0.00011860795454545454,
|
| 3446 |
+
"loss": 0.9198,
|
| 3447 |
+
"step": 573
|
| 3448 |
+
},
|
| 3449 |
+
{
|
| 3450 |
+
"epoch": 6.48,
|
| 3451 |
+
"learning_rate": 0.00011846590909090909,
|
| 3452 |
+
"loss": 0.8846,
|
| 3453 |
+
"step": 574
|
| 3454 |
+
},
|
| 3455 |
+
{
|
| 3456 |
+
"epoch": 6.5,
|
| 3457 |
+
"learning_rate": 0.00011832386363636365,
|
| 3458 |
+
"loss": 0.8495,
|
| 3459 |
+
"step": 575
|
| 3460 |
+
},
|
| 3461 |
+
{
|
| 3462 |
+
"epoch": 6.51,
|
| 3463 |
+
"learning_rate": 0.0001181818181818182,
|
| 3464 |
+
"loss": 0.8953,
|
| 3465 |
+
"step": 576
|
| 3466 |
+
},
|
| 3467 |
+
{
|
| 3468 |
+
"epoch": 6.52,
|
| 3469 |
+
"learning_rate": 0.00011803977272727274,
|
| 3470 |
+
"loss": 0.8686,
|
| 3471 |
+
"step": 577
|
| 3472 |
+
},
|
| 3473 |
+
{
|
| 3474 |
+
"epoch": 6.53,
|
| 3475 |
+
"learning_rate": 0.00011789772727272727,
|
| 3476 |
+
"loss": 0.8841,
|
| 3477 |
+
"step": 578
|
| 3478 |
+
},
|
| 3479 |
+
{
|
| 3480 |
+
"epoch": 6.54,
|
| 3481 |
+
"learning_rate": 0.00011775568181818182,
|
| 3482 |
+
"loss": 0.8681,
|
| 3483 |
+
"step": 579
|
| 3484 |
+
},
|
| 3485 |
+
{
|
| 3486 |
+
"epoch": 6.55,
|
| 3487 |
+
"learning_rate": 0.00011761363636363636,
|
| 3488 |
+
"loss": 0.8732,
|
| 3489 |
+
"step": 580
|
| 3490 |
+
},
|
| 3491 |
+
{
|
| 3492 |
+
"epoch": 6.56,
|
| 3493 |
+
"learning_rate": 0.00011747159090909092,
|
| 3494 |
+
"loss": 0.8582,
|
| 3495 |
+
"step": 581
|
| 3496 |
+
},
|
| 3497 |
+
{
|
| 3498 |
+
"epoch": 6.58,
|
| 3499 |
+
"learning_rate": 0.00011732954545454546,
|
| 3500 |
+
"loss": 0.8744,
|
| 3501 |
+
"step": 582
|
| 3502 |
+
},
|
| 3503 |
+
{
|
| 3504 |
+
"epoch": 6.59,
|
| 3505 |
+
"learning_rate": 0.00011718750000000001,
|
| 3506 |
+
"loss": 0.8694,
|
| 3507 |
+
"step": 583
|
| 3508 |
+
},
|
| 3509 |
+
{
|
| 3510 |
+
"epoch": 6.6,
|
| 3511 |
+
"learning_rate": 0.00011704545454545454,
|
| 3512 |
+
"loss": 0.8565,
|
| 3513 |
+
"step": 584
|
| 3514 |
+
},
|
| 3515 |
+
{
|
| 3516 |
+
"epoch": 6.61,
|
| 3517 |
+
"learning_rate": 0.00011690340909090909,
|
| 3518 |
+
"loss": 0.8584,
|
| 3519 |
+
"step": 585
|
| 3520 |
+
},
|
| 3521 |
+
{
|
| 3522 |
+
"epoch": 6.62,
|
| 3523 |
+
"learning_rate": 0.00011676136363636366,
|
| 3524 |
+
"loss": 0.8859,
|
| 3525 |
+
"step": 586
|
| 3526 |
+
},
|
| 3527 |
+
{
|
| 3528 |
+
"epoch": 6.63,
|
| 3529 |
+
"learning_rate": 0.00011661931818181819,
|
| 3530 |
+
"loss": 0.8452,
|
| 3531 |
+
"step": 587
|
| 3532 |
+
},
|
| 3533 |
+
{
|
| 3534 |
+
"epoch": 6.64,
|
| 3535 |
+
"learning_rate": 0.00011647727272727273,
|
| 3536 |
+
"loss": 0.8323,
|
| 3537 |
+
"step": 588
|
| 3538 |
+
},
|
| 3539 |
+
{
|
| 3540 |
+
"epoch": 6.65,
|
| 3541 |
+
"learning_rate": 0.00011633522727272728,
|
| 3542 |
+
"loss": 0.8548,
|
| 3543 |
+
"step": 589
|
| 3544 |
+
},
|
| 3545 |
+
{
|
| 3546 |
+
"epoch": 6.67,
|
| 3547 |
+
"learning_rate": 0.00011619318181818181,
|
| 3548 |
+
"loss": 0.8506,
|
| 3549 |
+
"step": 590
|
| 3550 |
+
},
|
| 3551 |
+
{
|
| 3552 |
+
"epoch": 6.68,
|
| 3553 |
+
"learning_rate": 0.00011605113636363636,
|
| 3554 |
+
"loss": 0.8556,
|
| 3555 |
+
"step": 591
|
| 3556 |
+
},
|
| 3557 |
+
{
|
| 3558 |
+
"epoch": 6.69,
|
| 3559 |
+
"learning_rate": 0.00011590909090909093,
|
| 3560 |
+
"loss": 0.8459,
|
| 3561 |
+
"step": 592
|
| 3562 |
+
},
|
| 3563 |
+
{
|
| 3564 |
+
"epoch": 6.7,
|
| 3565 |
+
"learning_rate": 0.00011576704545454546,
|
| 3566 |
+
"loss": 0.8432,
|
| 3567 |
+
"step": 593
|
| 3568 |
+
},
|
| 3569 |
+
{
|
| 3570 |
+
"epoch": 6.71,
|
| 3571 |
+
"learning_rate": 0.000115625,
|
| 3572 |
+
"loss": 0.8645,
|
| 3573 |
+
"step": 594
|
| 3574 |
+
},
|
| 3575 |
+
{
|
| 3576 |
+
"epoch": 6.72,
|
| 3577 |
+
"learning_rate": 0.00011548295454545455,
|
| 3578 |
+
"loss": 0.86,
|
| 3579 |
+
"step": 595
|
| 3580 |
+
},
|
| 3581 |
+
{
|
| 3582 |
+
"epoch": 6.73,
|
| 3583 |
+
"learning_rate": 0.00011534090909090908,
|
| 3584 |
+
"loss": 0.8161,
|
| 3585 |
+
"step": 596
|
| 3586 |
+
},
|
| 3587 |
+
{
|
| 3588 |
+
"epoch": 6.74,
|
| 3589 |
+
"learning_rate": 0.00011519886363636365,
|
| 3590 |
+
"loss": 0.8133,
|
| 3591 |
+
"step": 597
|
| 3592 |
+
},
|
| 3593 |
+
{
|
| 3594 |
+
"epoch": 6.76,
|
| 3595 |
+
"learning_rate": 0.0001150568181818182,
|
| 3596 |
+
"loss": 0.8372,
|
| 3597 |
+
"step": 598
|
| 3598 |
+
},
|
| 3599 |
+
{
|
| 3600 |
+
"epoch": 6.77,
|
| 3601 |
+
"learning_rate": 0.00011491477272727273,
|
| 3602 |
+
"loss": 0.8222,
|
| 3603 |
+
"step": 599
|
| 3604 |
+
},
|
| 3605 |
+
{
|
| 3606 |
+
"epoch": 6.78,
|
| 3607 |
+
"learning_rate": 0.00011477272727272728,
|
| 3608 |
+
"loss": 0.8372,
|
| 3609 |
+
"step": 600
|
| 3610 |
+
},
|
| 3611 |
+
{
|
| 3612 |
+
"epoch": 6.79,
|
| 3613 |
+
"learning_rate": 0.00011463068181818182,
|
| 3614 |
+
"loss": 0.837,
|
| 3615 |
+
"step": 601
|
| 3616 |
+
},
|
| 3617 |
+
{
|
| 3618 |
+
"epoch": 6.8,
|
| 3619 |
+
"learning_rate": 0.00011448863636363637,
|
| 3620 |
+
"loss": 0.8406,
|
| 3621 |
+
"step": 602
|
| 3622 |
+
},
|
| 3623 |
+
{
|
| 3624 |
+
"epoch": 6.81,
|
| 3625 |
+
"learning_rate": 0.00011434659090909092,
|
| 3626 |
+
"loss": 0.836,
|
| 3627 |
+
"step": 603
|
| 3628 |
+
},
|
| 3629 |
+
{
|
| 3630 |
+
"epoch": 6.82,
|
| 3631 |
+
"learning_rate": 0.00011420454545454547,
|
| 3632 |
+
"loss": 0.8476,
|
| 3633 |
+
"step": 604
|
| 3634 |
+
},
|
| 3635 |
+
{
|
| 3636 |
+
"epoch": 6.83,
|
| 3637 |
+
"learning_rate": 0.0001140625,
|
| 3638 |
+
"loss": 0.8368,
|
| 3639 |
+
"step": 605
|
| 3640 |
+
},
|
| 3641 |
+
{
|
| 3642 |
+
"epoch": 6.85,
|
| 3643 |
+
"learning_rate": 0.00011392045454545455,
|
| 3644 |
+
"loss": 0.822,
|
| 3645 |
+
"step": 606
|
| 3646 |
+
},
|
| 3647 |
+
{
|
| 3648 |
+
"epoch": 6.86,
|
| 3649 |
+
"learning_rate": 0.00011377840909090909,
|
| 3650 |
+
"loss": 0.8107,
|
| 3651 |
+
"step": 607
|
| 3652 |
+
},
|
| 3653 |
+
{
|
| 3654 |
+
"epoch": 6.87,
|
| 3655 |
+
"learning_rate": 0.00011363636363636365,
|
| 3656 |
+
"loss": 0.8395,
|
| 3657 |
+
"step": 608
|
| 3658 |
+
},
|
| 3659 |
+
{
|
| 3660 |
+
"epoch": 6.88,
|
| 3661 |
+
"learning_rate": 0.0001134943181818182,
|
| 3662 |
+
"loss": 0.8083,
|
| 3663 |
+
"step": 609
|
| 3664 |
+
},
|
| 3665 |
+
{
|
| 3666 |
+
"epoch": 6.89,
|
| 3667 |
+
"learning_rate": 0.00011335227272727274,
|
| 3668 |
+
"loss": 0.828,
|
| 3669 |
+
"step": 610
|
| 3670 |
+
},
|
| 3671 |
+
{
|
| 3672 |
+
"epoch": 6.9,
|
| 3673 |
+
"learning_rate": 0.00011321022727272727,
|
| 3674 |
+
"loss": 0.8494,
|
| 3675 |
+
"step": 611
|
| 3676 |
+
},
|
| 3677 |
+
{
|
| 3678 |
+
"epoch": 6.91,
|
| 3679 |
+
"learning_rate": 0.00011306818181818182,
|
| 3680 |
+
"loss": 0.8169,
|
| 3681 |
+
"step": 612
|
| 3682 |
+
},
|
| 3683 |
+
{
|
| 3684 |
+
"epoch": 6.93,
|
| 3685 |
+
"learning_rate": 0.00011292613636363636,
|
| 3686 |
+
"loss": 0.8224,
|
| 3687 |
+
"step": 613
|
| 3688 |
+
},
|
| 3689 |
+
{
|
| 3690 |
+
"epoch": 6.94,
|
| 3691 |
+
"learning_rate": 0.00011278409090909092,
|
| 3692 |
+
"loss": 0.8173,
|
| 3693 |
+
"step": 614
|
| 3694 |
+
},
|
| 3695 |
+
{
|
| 3696 |
+
"epoch": 6.95,
|
| 3697 |
+
"learning_rate": 0.00011264204545454547,
|
| 3698 |
+
"loss": 0.7961,
|
| 3699 |
+
"step": 615
|
| 3700 |
+
},
|
| 3701 |
+
{
|
| 3702 |
+
"epoch": 6.96,
|
| 3703 |
+
"learning_rate": 0.00011250000000000001,
|
| 3704 |
+
"loss": 0.7948,
|
| 3705 |
+
"step": 616
|
| 3706 |
+
},
|
| 3707 |
+
{
|
| 3708 |
+
"epoch": 6.97,
|
| 3709 |
+
"learning_rate": 0.00011235795454545454,
|
| 3710 |
+
"loss": 0.7746,
|
| 3711 |
+
"step": 617
|
| 3712 |
+
},
|
| 3713 |
+
{
|
| 3714 |
+
"epoch": 6.98,
|
| 3715 |
+
"learning_rate": 0.00011221590909090909,
|
| 3716 |
+
"loss": 0.8325,
|
| 3717 |
+
"step": 618
|
| 3718 |
+
},
|
| 3719 |
+
{
|
| 3720 |
+
"epoch": 6.99,
|
| 3721 |
+
"learning_rate": 0.00011207386363636365,
|
| 3722 |
+
"loss": 0.8149,
|
| 3723 |
+
"step": 619
|
| 3724 |
+
},
|
| 3725 |
+
{
|
| 3726 |
+
"epoch": 7.0,
|
| 3727 |
+
"learning_rate": 0.00011193181818181819,
|
| 3728 |
+
"loss": 0.7516,
|
| 3729 |
+
"step": 620
|
| 3730 |
+
},
|
| 3731 |
+
{
|
| 3732 |
+
"epoch": 7.02,
|
| 3733 |
+
"learning_rate": 0.00011178977272727274,
|
| 3734 |
+
"loss": 0.7571,
|
| 3735 |
+
"step": 621
|
| 3736 |
+
},
|
| 3737 |
+
{
|
| 3738 |
+
"epoch": 7.03,
|
| 3739 |
+
"learning_rate": 0.00011164772727272728,
|
| 3740 |
+
"loss": 0.7397,
|
| 3741 |
+
"step": 622
|
| 3742 |
+
},
|
| 3743 |
+
{
|
| 3744 |
+
"epoch": 7.04,
|
| 3745 |
+
"learning_rate": 0.00011150568181818181,
|
| 3746 |
+
"loss": 0.761,
|
| 3747 |
+
"step": 623
|
| 3748 |
+
},
|
| 3749 |
+
{
|
| 3750 |
+
"epoch": 7.05,
|
| 3751 |
+
"learning_rate": 0.00011136363636363636,
|
| 3752 |
+
"loss": 0.7783,
|
| 3753 |
+
"step": 624
|
| 3754 |
+
},
|
| 3755 |
+
{
|
| 3756 |
+
"epoch": 7.06,
|
| 3757 |
+
"learning_rate": 0.00011122159090909092,
|
| 3758 |
+
"loss": 0.7571,
|
| 3759 |
+
"step": 625
|
| 3760 |
+
},
|
| 3761 |
+
{
|
| 3762 |
+
"epoch": 7.07,
|
| 3763 |
+
"learning_rate": 0.00011107954545454546,
|
| 3764 |
+
"loss": 0.7628,
|
| 3765 |
+
"step": 626
|
| 3766 |
+
},
|
| 3767 |
+
{
|
| 3768 |
+
"epoch": 7.08,
|
| 3769 |
+
"learning_rate": 0.0001109375,
|
| 3770 |
+
"loss": 0.7561,
|
| 3771 |
+
"step": 627
|
| 3772 |
+
},
|
| 3773 |
+
{
|
| 3774 |
+
"epoch": 7.09,
|
| 3775 |
+
"learning_rate": 0.00011079545454545455,
|
| 3776 |
+
"loss": 0.7432,
|
| 3777 |
+
"step": 628
|
| 3778 |
+
},
|
| 3779 |
+
{
|
| 3780 |
+
"epoch": 7.11,
|
| 3781 |
+
"learning_rate": 0.00011065340909090908,
|
| 3782 |
+
"loss": 0.7245,
|
| 3783 |
+
"step": 629
|
| 3784 |
+
},
|
| 3785 |
+
{
|
| 3786 |
+
"epoch": 7.12,
|
| 3787 |
+
"learning_rate": 0.00011051136363636366,
|
| 3788 |
+
"loss": 0.7279,
|
| 3789 |
+
"step": 630
|
| 3790 |
+
},
|
| 3791 |
+
{
|
| 3792 |
+
"epoch": 7.13,
|
| 3793 |
+
"learning_rate": 0.00011036931818181819,
|
| 3794 |
+
"loss": 0.7347,
|
| 3795 |
+
"step": 631
|
| 3796 |
+
},
|
| 3797 |
+
{
|
| 3798 |
+
"epoch": 7.14,
|
| 3799 |
+
"learning_rate": 0.00011022727272727273,
|
| 3800 |
+
"loss": 0.7427,
|
| 3801 |
+
"step": 632
|
| 3802 |
+
},
|
| 3803 |
+
{
|
| 3804 |
+
"epoch": 7.15,
|
| 3805 |
+
"learning_rate": 0.00011008522727272728,
|
| 3806 |
+
"loss": 0.7339,
|
| 3807 |
+
"step": 633
|
| 3808 |
+
},
|
| 3809 |
+
{
|
| 3810 |
+
"epoch": 7.16,
|
| 3811 |
+
"learning_rate": 0.00010994318181818182,
|
| 3812 |
+
"loss": 0.7375,
|
| 3813 |
+
"step": 634
|
| 3814 |
+
},
|
| 3815 |
+
{
|
| 3816 |
+
"epoch": 7.17,
|
| 3817 |
+
"learning_rate": 0.00010980113636363635,
|
| 3818 |
+
"loss": 0.7182,
|
| 3819 |
+
"step": 635
|
| 3820 |
+
},
|
| 3821 |
+
{
|
| 3822 |
+
"epoch": 7.19,
|
| 3823 |
+
"learning_rate": 0.00010965909090909093,
|
| 3824 |
+
"loss": 0.7452,
|
| 3825 |
+
"step": 636
|
| 3826 |
+
},
|
| 3827 |
+
{
|
| 3828 |
+
"epoch": 7.2,
|
| 3829 |
+
"learning_rate": 0.00010951704545454546,
|
| 3830 |
+
"loss": 0.7565,
|
| 3831 |
+
"step": 637
|
| 3832 |
+
},
|
| 3833 |
+
{
|
| 3834 |
+
"epoch": 7.21,
|
| 3835 |
+
"learning_rate": 0.000109375,
|
| 3836 |
+
"loss": 0.7296,
|
| 3837 |
+
"step": 638
|
| 3838 |
+
},
|
| 3839 |
+
{
|
| 3840 |
+
"epoch": 7.22,
|
| 3841 |
+
"learning_rate": 0.00010923295454545455,
|
| 3842 |
+
"loss": 0.7484,
|
| 3843 |
+
"step": 639
|
| 3844 |
+
},
|
| 3845 |
+
{
|
| 3846 |
+
"epoch": 7.23,
|
| 3847 |
+
"learning_rate": 0.00010909090909090909,
|
| 3848 |
+
"loss": 0.732,
|
| 3849 |
+
"step": 640
|
| 3850 |
+
},
|
| 3851 |
+
{
|
| 3852 |
+
"epoch": 7.24,
|
| 3853 |
+
"learning_rate": 0.00010894886363636365,
|
| 3854 |
+
"loss": 0.7415,
|
| 3855 |
+
"step": 641
|
| 3856 |
+
},
|
| 3857 |
+
{
|
| 3858 |
+
"epoch": 7.25,
|
| 3859 |
+
"learning_rate": 0.0001088068181818182,
|
| 3860 |
+
"loss": 0.7344,
|
| 3861 |
+
"step": 642
|
| 3862 |
+
},
|
| 3863 |
+
{
|
| 3864 |
+
"epoch": 7.26,
|
| 3865 |
+
"learning_rate": 0.00010866477272727274,
|
| 3866 |
+
"loss": 0.7267,
|
| 3867 |
+
"step": 643
|
| 3868 |
+
},
|
| 3869 |
+
{
|
| 3870 |
+
"epoch": 7.28,
|
| 3871 |
+
"learning_rate": 0.00010852272727272727,
|
| 3872 |
+
"loss": 0.7543,
|
| 3873 |
+
"step": 644
|
| 3874 |
+
},
|
| 3875 |
+
{
|
| 3876 |
+
"epoch": 7.29,
|
| 3877 |
+
"learning_rate": 0.00010838068181818182,
|
| 3878 |
+
"loss": 0.7266,
|
| 3879 |
+
"step": 645
|
| 3880 |
+
},
|
| 3881 |
+
{
|
| 3882 |
+
"epoch": 7.3,
|
| 3883 |
+
"learning_rate": 0.00010823863636363636,
|
| 3884 |
+
"loss": 0.7449,
|
| 3885 |
+
"step": 646
|
| 3886 |
+
},
|
| 3887 |
+
{
|
| 3888 |
+
"epoch": 7.31,
|
| 3889 |
+
"learning_rate": 0.00010809659090909092,
|
| 3890 |
+
"loss": 0.7324,
|
| 3891 |
+
"step": 647
|
| 3892 |
+
},
|
| 3893 |
+
{
|
| 3894 |
+
"epoch": 7.32,
|
| 3895 |
+
"learning_rate": 0.00010795454545454547,
|
| 3896 |
+
"loss": 0.7268,
|
| 3897 |
+
"step": 648
|
| 3898 |
+
},
|
| 3899 |
+
{
|
| 3900 |
+
"epoch": 7.33,
|
| 3901 |
+
"learning_rate": 0.00010781250000000001,
|
| 3902 |
+
"loss": 0.7172,
|
| 3903 |
+
"step": 649
|
| 3904 |
+
},
|
| 3905 |
+
{
|
| 3906 |
+
"epoch": 7.34,
|
| 3907 |
+
"learning_rate": 0.00010767045454545454,
|
| 3908 |
+
"loss": 0.7169,
|
| 3909 |
+
"step": 650
|
| 3910 |
+
},
|
| 3911 |
+
{
|
| 3912 |
+
"epoch": 7.35,
|
| 3913 |
+
"learning_rate": 0.00010752840909090909,
|
| 3914 |
+
"loss": 0.7194,
|
| 3915 |
+
"step": 651
|
| 3916 |
+
},
|
| 3917 |
+
{
|
| 3918 |
+
"epoch": 7.37,
|
| 3919 |
+
"learning_rate": 0.00010738636363636365,
|
| 3920 |
+
"loss": 0.7223,
|
| 3921 |
+
"step": 652
|
| 3922 |
+
},
|
| 3923 |
+
{
|
| 3924 |
+
"epoch": 7.38,
|
| 3925 |
+
"learning_rate": 0.00010724431818181819,
|
| 3926 |
+
"loss": 0.7158,
|
| 3927 |
+
"step": 653
|
| 3928 |
+
},
|
| 3929 |
+
{
|
| 3930 |
+
"epoch": 7.39,
|
| 3931 |
+
"learning_rate": 0.00010710227272727274,
|
| 3932 |
+
"loss": 0.7122,
|
| 3933 |
+
"step": 654
|
| 3934 |
+
},
|
| 3935 |
+
{
|
| 3936 |
+
"epoch": 7.4,
|
| 3937 |
+
"learning_rate": 0.00010696022727272728,
|
| 3938 |
+
"loss": 0.7225,
|
| 3939 |
+
"step": 655
|
| 3940 |
+
},
|
| 3941 |
+
{
|
| 3942 |
+
"epoch": 7.41,
|
| 3943 |
+
"learning_rate": 0.00010681818181818181,
|
| 3944 |
+
"loss": 0.7102,
|
| 3945 |
+
"step": 656
|
| 3946 |
+
},
|
| 3947 |
+
{
|
| 3948 |
+
"epoch": 7.42,
|
| 3949 |
+
"learning_rate": 0.00010667613636363636,
|
| 3950 |
+
"loss": 0.7251,
|
| 3951 |
+
"step": 657
|
| 3952 |
+
},
|
| 3953 |
+
{
|
| 3954 |
+
"epoch": 7.43,
|
| 3955 |
+
"learning_rate": 0.00010653409090909092,
|
| 3956 |
+
"loss": 0.7191,
|
| 3957 |
+
"step": 658
|
| 3958 |
+
},
|
| 3959 |
+
{
|
| 3960 |
+
"epoch": 7.45,
|
| 3961 |
+
"learning_rate": 0.00010639204545454546,
|
| 3962 |
+
"loss": 0.7015,
|
| 3963 |
+
"step": 659
|
| 3964 |
+
},
|
| 3965 |
+
{
|
| 3966 |
+
"epoch": 7.46,
|
| 3967 |
+
"learning_rate": 0.00010625000000000001,
|
| 3968 |
+
"loss": 0.693,
|
| 3969 |
+
"step": 660
|
| 3970 |
+
},
|
| 3971 |
+
{
|
| 3972 |
+
"epoch": 7.47,
|
| 3973 |
+
"learning_rate": 0.00010610795454545455,
|
| 3974 |
+
"loss": 0.7039,
|
| 3975 |
+
"step": 661
|
| 3976 |
+
},
|
| 3977 |
+
{
|
| 3978 |
+
"epoch": 7.48,
|
| 3979 |
+
"learning_rate": 0.00010596590909090908,
|
| 3980 |
+
"loss": 0.7305,
|
| 3981 |
+
"step": 662
|
| 3982 |
+
},
|
| 3983 |
+
{
|
| 3984 |
+
"epoch": 7.49,
|
| 3985 |
+
"learning_rate": 0.00010582386363636366,
|
| 3986 |
+
"loss": 0.6978,
|
| 3987 |
+
"step": 663
|
| 3988 |
+
},
|
| 3989 |
+
{
|
| 3990 |
+
"epoch": 7.5,
|
| 3991 |
+
"learning_rate": 0.00010568181818181819,
|
| 3992 |
+
"loss": 0.7219,
|
| 3993 |
+
"step": 664
|
| 3994 |
+
},
|
| 3995 |
+
{
|
| 3996 |
+
"epoch": 7.51,
|
| 3997 |
+
"learning_rate": 0.00010553977272727273,
|
| 3998 |
+
"loss": 0.7199,
|
| 3999 |
+
"step": 665
|
| 4000 |
+
},
|
| 4001 |
+
{
|
| 4002 |
+
"epoch": 7.52,
|
| 4003 |
+
"learning_rate": 0.00010539772727272728,
|
| 4004 |
+
"loss": 0.6979,
|
| 4005 |
+
"step": 666
|
| 4006 |
+
},
|
| 4007 |
+
{
|
| 4008 |
+
"epoch": 7.54,
|
| 4009 |
+
"learning_rate": 0.00010525568181818182,
|
| 4010 |
+
"loss": 0.7058,
|
| 4011 |
+
"step": 667
|
| 4012 |
+
},
|
| 4013 |
+
{
|
| 4014 |
+
"epoch": 7.55,
|
| 4015 |
+
"learning_rate": 0.00010511363636363635,
|
| 4016 |
+
"loss": 0.6994,
|
| 4017 |
+
"step": 668
|
| 4018 |
+
},
|
| 4019 |
+
{
|
| 4020 |
+
"epoch": 7.56,
|
| 4021 |
+
"learning_rate": 0.00010497159090909093,
|
| 4022 |
+
"loss": 0.7141,
|
| 4023 |
+
"step": 669
|
| 4024 |
+
},
|
| 4025 |
+
{
|
| 4026 |
+
"epoch": 7.57,
|
| 4027 |
+
"learning_rate": 0.00010482954545454546,
|
| 4028 |
+
"loss": 0.7092,
|
| 4029 |
+
"step": 670
|
| 4030 |
+
},
|
| 4031 |
+
{
|
| 4032 |
+
"epoch": 7.58,
|
| 4033 |
+
"learning_rate": 0.0001046875,
|
| 4034 |
+
"loss": 0.7059,
|
| 4035 |
+
"step": 671
|
| 4036 |
+
},
|
| 4037 |
+
{
|
| 4038 |
+
"epoch": 7.59,
|
| 4039 |
+
"learning_rate": 0.00010454545454545455,
|
| 4040 |
+
"loss": 0.6904,
|
| 4041 |
+
"step": 672
|
| 4042 |
+
},
|
| 4043 |
+
{
|
| 4044 |
+
"epoch": 7.6,
|
| 4045 |
+
"learning_rate": 0.0001044034090909091,
|
| 4046 |
+
"loss": 0.7115,
|
| 4047 |
+
"step": 673
|
| 4048 |
+
},
|
| 4049 |
+
{
|
| 4050 |
+
"epoch": 7.61,
|
| 4051 |
+
"learning_rate": 0.00010426136363636365,
|
| 4052 |
+
"loss": 0.7254,
|
| 4053 |
+
"step": 674
|
| 4054 |
+
},
|
| 4055 |
+
{
|
| 4056 |
+
"epoch": 7.63,
|
| 4057 |
+
"learning_rate": 0.0001041193181818182,
|
| 4058 |
+
"loss": 0.7181,
|
| 4059 |
+
"step": 675
|
| 4060 |
+
},
|
| 4061 |
+
{
|
| 4062 |
+
"epoch": 7.64,
|
| 4063 |
+
"learning_rate": 0.00010397727272727273,
|
| 4064 |
+
"loss": 0.6867,
|
| 4065 |
+
"step": 676
|
| 4066 |
+
},
|
| 4067 |
+
{
|
| 4068 |
+
"epoch": 7.65,
|
| 4069 |
+
"learning_rate": 0.00010383522727272727,
|
| 4070 |
+
"loss": 0.6917,
|
| 4071 |
+
"step": 677
|
| 4072 |
+
},
|
| 4073 |
+
{
|
| 4074 |
+
"epoch": 7.66,
|
| 4075 |
+
"learning_rate": 0.00010369318181818182,
|
| 4076 |
+
"loss": 0.6908,
|
| 4077 |
+
"step": 678
|
| 4078 |
+
},
|
| 4079 |
+
{
|
| 4080 |
+
"epoch": 7.67,
|
| 4081 |
+
"learning_rate": 0.00010355113636363636,
|
| 4082 |
+
"loss": 0.6871,
|
| 4083 |
+
"step": 679
|
| 4084 |
+
},
|
| 4085 |
+
{
|
| 4086 |
+
"epoch": 7.68,
|
| 4087 |
+
"learning_rate": 0.00010340909090909092,
|
| 4088 |
+
"loss": 0.682,
|
| 4089 |
+
"step": 680
|
| 4090 |
+
},
|
| 4091 |
+
{
|
| 4092 |
+
"epoch": 7.69,
|
| 4093 |
+
"learning_rate": 0.00010326704545454547,
|
| 4094 |
+
"loss": 0.6737,
|
| 4095 |
+
"step": 681
|
| 4096 |
+
},
|
| 4097 |
+
{
|
| 4098 |
+
"epoch": 7.7,
|
| 4099 |
+
"learning_rate": 0.000103125,
|
| 4100 |
+
"loss": 0.7023,
|
| 4101 |
+
"step": 682
|
| 4102 |
+
},
|
| 4103 |
+
{
|
| 4104 |
+
"epoch": 7.72,
|
| 4105 |
+
"learning_rate": 0.00010298295454545454,
|
| 4106 |
+
"loss": 0.7079,
|
| 4107 |
+
"step": 683
|
| 4108 |
+
},
|
| 4109 |
+
{
|
| 4110 |
+
"epoch": 7.73,
|
| 4111 |
+
"learning_rate": 0.00010284090909090909,
|
| 4112 |
+
"loss": 0.6954,
|
| 4113 |
+
"step": 684
|
| 4114 |
+
},
|
| 4115 |
+
{
|
| 4116 |
+
"epoch": 7.74,
|
| 4117 |
+
"learning_rate": 0.00010269886363636365,
|
| 4118 |
+
"loss": 0.6834,
|
| 4119 |
+
"step": 685
|
| 4120 |
+
},
|
| 4121 |
+
{
|
| 4122 |
+
"epoch": 7.75,
|
| 4123 |
+
"learning_rate": 0.0001025568181818182,
|
| 4124 |
+
"loss": 0.6706,
|
| 4125 |
+
"step": 686
|
| 4126 |
+
},
|
| 4127 |
+
{
|
| 4128 |
+
"epoch": 7.76,
|
| 4129 |
+
"learning_rate": 0.00010241477272727274,
|
| 4130 |
+
"loss": 0.6706,
|
| 4131 |
+
"step": 687
|
| 4132 |
+
},
|
| 4133 |
+
{
|
| 4134 |
+
"epoch": 7.77,
|
| 4135 |
+
"learning_rate": 0.00010227272727272727,
|
| 4136 |
+
"loss": 0.681,
|
| 4137 |
+
"step": 688
|
| 4138 |
+
},
|
| 4139 |
+
{
|
| 4140 |
+
"epoch": 7.78,
|
| 4141 |
+
"learning_rate": 0.00010213068181818182,
|
| 4142 |
+
"loss": 0.6853,
|
| 4143 |
+
"step": 689
|
| 4144 |
+
},
|
| 4145 |
+
{
|
| 4146 |
+
"epoch": 7.8,
|
| 4147 |
+
"learning_rate": 0.00010198863636363636,
|
| 4148 |
+
"loss": 0.6772,
|
| 4149 |
+
"step": 690
|
| 4150 |
+
},
|
| 4151 |
+
{
|
| 4152 |
+
"epoch": 7.81,
|
| 4153 |
+
"learning_rate": 0.00010184659090909092,
|
| 4154 |
+
"loss": 0.6635,
|
| 4155 |
+
"step": 691
|
| 4156 |
+
},
|
| 4157 |
+
{
|
| 4158 |
+
"epoch": 7.82,
|
| 4159 |
+
"learning_rate": 0.00010170454545454546,
|
| 4160 |
+
"loss": 0.6712,
|
| 4161 |
+
"step": 692
|
| 4162 |
+
},
|
| 4163 |
+
{
|
| 4164 |
+
"epoch": 7.83,
|
| 4165 |
+
"learning_rate": 0.00010156250000000001,
|
| 4166 |
+
"loss": 0.6884,
|
| 4167 |
+
"step": 693
|
| 4168 |
+
},
|
| 4169 |
+
{
|
| 4170 |
+
"epoch": 7.84,
|
| 4171 |
+
"learning_rate": 0.00010142045454545454,
|
| 4172 |
+
"loss": 0.6641,
|
| 4173 |
+
"step": 694
|
| 4174 |
+
},
|
| 4175 |
+
{
|
| 4176 |
+
"epoch": 7.85,
|
| 4177 |
+
"learning_rate": 0.00010127840909090909,
|
| 4178 |
+
"loss": 0.6838,
|
| 4179 |
+
"step": 695
|
| 4180 |
+
},
|
| 4181 |
+
{
|
| 4182 |
+
"epoch": 7.86,
|
| 4183 |
+
"learning_rate": 0.00010113636363636366,
|
| 4184 |
+
"loss": 0.675,
|
| 4185 |
+
"step": 696
|
| 4186 |
+
},
|
| 4187 |
+
{
|
| 4188 |
+
"epoch": 7.87,
|
| 4189 |
+
"learning_rate": 0.00010099431818181819,
|
| 4190 |
+
"loss": 0.6626,
|
| 4191 |
+
"step": 697
|
| 4192 |
+
},
|
| 4193 |
+
{
|
| 4194 |
+
"epoch": 7.89,
|
| 4195 |
+
"learning_rate": 0.00010085227272727273,
|
| 4196 |
+
"loss": 0.6605,
|
| 4197 |
+
"step": 698
|
| 4198 |
+
},
|
| 4199 |
+
{
|
| 4200 |
+
"epoch": 7.9,
|
| 4201 |
+
"learning_rate": 0.00010071022727272728,
|
| 4202 |
+
"loss": 0.6777,
|
| 4203 |
+
"step": 699
|
| 4204 |
+
},
|
| 4205 |
+
{
|
| 4206 |
+
"epoch": 7.91,
|
| 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": 9.550264840009421e+17,
|
| 4217 |
"trial_name": null,
|
| 4218 |
"trial_params": null
|
| 4219 |
}
|