Training in progress, step 1600
Browse files- last-checkpoint/optimizer.pt +1 -1
- last-checkpoint/pytorch_model.bin +1 -1
- last-checkpoint/rng_state.pth +1 -1
- last-checkpoint/scheduler.pt +1 -1
- last-checkpoint/trainer_state.json +1218 -5
- pytorch_model.bin +1 -1
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4736616809
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d827ca908186ff4d1eeac2cd097d10d72152f52a20d876b710c6ab92928f405b
|
| 3 |
size 4736616809
|
last-checkpoint/pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 2368281769
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:728b6dfaa1c989be32093514baf4c7d88f3ff72d43aa5c614b342b108779fd92
|
| 3 |
size 2368281769
|
last-checkpoint/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 sha256:204741013cff7ad3367b61fa8ef614471a5423dec181a95b8549965efd787d9c
|
| 3 |
size 14575
|
last-checkpoint/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 sha256:504d7c79830240f80701a38d5e784c442b3685f4ab42e22f30cf40ef402b0086
|
| 3 |
size 627
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 2.
|
| 3 |
-
"best_model_checkpoint": "output/checkpoint-
|
| 4 |
-
"epoch": 1.
|
| 5 |
-
"global_step":
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
@@ -8497,11 +8497,1224 @@
|
|
| 8497 |
"eval_samples_per_second": 0.054,
|
| 8498 |
"eval_steps_per_second": 0.054,
|
| 8499 |
"step": 1400
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8500 |
}
|
| 8501 |
],
|
| 8502 |
"max_steps": 2000,
|
| 8503 |
"num_train_epochs": 9223372036854775807,
|
| 8504 |
-
"total_flos": 1.
|
| 8505 |
"trial_name": null,
|
| 8506 |
"trial_params": null
|
| 8507 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 2.1070616245269775,
|
| 3 |
+
"best_model_checkpoint": "output/checkpoint-1600",
|
| 4 |
+
"epoch": 1.18650390625,
|
| 5 |
+
"global_step": 1600,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
|
|
| 8497 |
"eval_samples_per_second": 0.054,
|
| 8498 |
"eval_steps_per_second": 0.054,
|
| 8499 |
"step": 1400
|
| 8500 |
+
},
|
| 8501 |
+
{
|
| 8502 |
+
"epoch": 1.09,
|
| 8503 |
+
"learning_rate": 0.0002662222222222222,
|
| 8504 |
+
"loss": 2.0898,
|
| 8505 |
+
"step": 1401
|
| 8506 |
+
},
|
| 8507 |
+
{
|
| 8508 |
+
"epoch": 1.09,
|
| 8509 |
+
"learning_rate": 0.00026577777777777776,
|
| 8510 |
+
"loss": 2.1313,
|
| 8511 |
+
"step": 1402
|
| 8512 |
+
},
|
| 8513 |
+
{
|
| 8514 |
+
"epoch": 1.09,
|
| 8515 |
+
"learning_rate": 0.00026533333333333335,
|
| 8516 |
+
"loss": 2.1198,
|
| 8517 |
+
"step": 1403
|
| 8518 |
+
},
|
| 8519 |
+
{
|
| 8520 |
+
"epoch": 1.09,
|
| 8521 |
+
"learning_rate": 0.00026488888888888894,
|
| 8522 |
+
"loss": 2.1396,
|
| 8523 |
+
"step": 1404
|
| 8524 |
+
},
|
| 8525 |
+
{
|
| 8526 |
+
"epoch": 1.09,
|
| 8527 |
+
"learning_rate": 0.00026444444444444443,
|
| 8528 |
+
"loss": 2.1515,
|
| 8529 |
+
"step": 1405
|
| 8530 |
+
},
|
| 8531 |
+
{
|
| 8532 |
+
"epoch": 1.09,
|
| 8533 |
+
"learning_rate": 0.000264,
|
| 8534 |
+
"loss": 2.1336,
|
| 8535 |
+
"step": 1406
|
| 8536 |
+
},
|
| 8537 |
+
{
|
| 8538 |
+
"epoch": 1.09,
|
| 8539 |
+
"learning_rate": 0.00026355555555555556,
|
| 8540 |
+
"loss": 2.1756,
|
| 8541 |
+
"step": 1407
|
| 8542 |
+
},
|
| 8543 |
+
{
|
| 8544 |
+
"epoch": 1.09,
|
| 8545 |
+
"learning_rate": 0.00026311111111111115,
|
| 8546 |
+
"loss": 2.1563,
|
| 8547 |
+
"step": 1408
|
| 8548 |
+
},
|
| 8549 |
+
{
|
| 8550 |
+
"epoch": 1.09,
|
| 8551 |
+
"learning_rate": 0.00026266666666666664,
|
| 8552 |
+
"loss": 2.1252,
|
| 8553 |
+
"step": 1409
|
| 8554 |
+
},
|
| 8555 |
+
{
|
| 8556 |
+
"epoch": 1.09,
|
| 8557 |
+
"learning_rate": 0.00026222222222222223,
|
| 8558 |
+
"loss": 2.118,
|
| 8559 |
+
"step": 1410
|
| 8560 |
+
},
|
| 8561 |
+
{
|
| 8562 |
+
"epoch": 1.09,
|
| 8563 |
+
"learning_rate": 0.0002617777777777778,
|
| 8564 |
+
"loss": 2.142,
|
| 8565 |
+
"step": 1411
|
| 8566 |
+
},
|
| 8567 |
+
{
|
| 8568 |
+
"epoch": 1.09,
|
| 8569 |
+
"learning_rate": 0.00026133333333333336,
|
| 8570 |
+
"loss": 2.2093,
|
| 8571 |
+
"step": 1412
|
| 8572 |
+
},
|
| 8573 |
+
{
|
| 8574 |
+
"epoch": 1.09,
|
| 8575 |
+
"learning_rate": 0.0002608888888888889,
|
| 8576 |
+
"loss": 2.1199,
|
| 8577 |
+
"step": 1413
|
| 8578 |
+
},
|
| 8579 |
+
{
|
| 8580 |
+
"epoch": 1.09,
|
| 8581 |
+
"learning_rate": 0.00026044444444444444,
|
| 8582 |
+
"loss": 2.0734,
|
| 8583 |
+
"step": 1414
|
| 8584 |
+
},
|
| 8585 |
+
{
|
| 8586 |
+
"epoch": 1.09,
|
| 8587 |
+
"learning_rate": 0.00026000000000000003,
|
| 8588 |
+
"loss": 2.0976,
|
| 8589 |
+
"step": 1415
|
| 8590 |
+
},
|
| 8591 |
+
{
|
| 8592 |
+
"epoch": 1.09,
|
| 8593 |
+
"learning_rate": 0.00025955555555555557,
|
| 8594 |
+
"loss": 2.0571,
|
| 8595 |
+
"step": 1416
|
| 8596 |
+
},
|
| 8597 |
+
{
|
| 8598 |
+
"epoch": 1.1,
|
| 8599 |
+
"learning_rate": 0.0002591111111111111,
|
| 8600 |
+
"loss": 2.1415,
|
| 8601 |
+
"step": 1417
|
| 8602 |
+
},
|
| 8603 |
+
{
|
| 8604 |
+
"epoch": 1.1,
|
| 8605 |
+
"learning_rate": 0.00025866666666666665,
|
| 8606 |
+
"loss": 2.0729,
|
| 8607 |
+
"step": 1418
|
| 8608 |
+
},
|
| 8609 |
+
{
|
| 8610 |
+
"epoch": 1.1,
|
| 8611 |
+
"learning_rate": 0.00025822222222222224,
|
| 8612 |
+
"loss": 2.1535,
|
| 8613 |
+
"step": 1419
|
| 8614 |
+
},
|
| 8615 |
+
{
|
| 8616 |
+
"epoch": 1.1,
|
| 8617 |
+
"learning_rate": 0.00025777777777777783,
|
| 8618 |
+
"loss": 2.1454,
|
| 8619 |
+
"step": 1420
|
| 8620 |
+
},
|
| 8621 |
+
{
|
| 8622 |
+
"epoch": 1.1,
|
| 8623 |
+
"learning_rate": 0.0002573333333333333,
|
| 8624 |
+
"loss": 2.1955,
|
| 8625 |
+
"step": 1421
|
| 8626 |
+
},
|
| 8627 |
+
{
|
| 8628 |
+
"epoch": 1.1,
|
| 8629 |
+
"learning_rate": 0.0002568888888888889,
|
| 8630 |
+
"loss": 2.1006,
|
| 8631 |
+
"step": 1422
|
| 8632 |
+
},
|
| 8633 |
+
{
|
| 8634 |
+
"epoch": 1.1,
|
| 8635 |
+
"learning_rate": 0.00025644444444444445,
|
| 8636 |
+
"loss": 2.0917,
|
| 8637 |
+
"step": 1423
|
| 8638 |
+
},
|
| 8639 |
+
{
|
| 8640 |
+
"epoch": 1.1,
|
| 8641 |
+
"learning_rate": 0.00025600000000000004,
|
| 8642 |
+
"loss": 2.0921,
|
| 8643 |
+
"step": 1424
|
| 8644 |
+
},
|
| 8645 |
+
{
|
| 8646 |
+
"epoch": 1.1,
|
| 8647 |
+
"learning_rate": 0.00025555555555555553,
|
| 8648 |
+
"loss": 2.1228,
|
| 8649 |
+
"step": 1425
|
| 8650 |
+
},
|
| 8651 |
+
{
|
| 8652 |
+
"epoch": 1.1,
|
| 8653 |
+
"learning_rate": 0.0002551111111111111,
|
| 8654 |
+
"loss": 2.1433,
|
| 8655 |
+
"step": 1426
|
| 8656 |
+
},
|
| 8657 |
+
{
|
| 8658 |
+
"epoch": 1.1,
|
| 8659 |
+
"learning_rate": 0.0002546666666666667,
|
| 8660 |
+
"loss": 2.1351,
|
| 8661 |
+
"step": 1427
|
| 8662 |
+
},
|
| 8663 |
+
{
|
| 8664 |
+
"epoch": 1.1,
|
| 8665 |
+
"learning_rate": 0.00025422222222222225,
|
| 8666 |
+
"loss": 2.1113,
|
| 8667 |
+
"step": 1428
|
| 8668 |
+
},
|
| 8669 |
+
{
|
| 8670 |
+
"epoch": 1.1,
|
| 8671 |
+
"learning_rate": 0.0002537777777777778,
|
| 8672 |
+
"loss": 2.0421,
|
| 8673 |
+
"step": 1429
|
| 8674 |
+
},
|
| 8675 |
+
{
|
| 8676 |
+
"epoch": 1.1,
|
| 8677 |
+
"learning_rate": 0.00025333333333333333,
|
| 8678 |
+
"loss": 2.0756,
|
| 8679 |
+
"step": 1430
|
| 8680 |
+
},
|
| 8681 |
+
{
|
| 8682 |
+
"epoch": 1.1,
|
| 8683 |
+
"learning_rate": 0.0002528888888888889,
|
| 8684 |
+
"loss": 2.1356,
|
| 8685 |
+
"step": 1431
|
| 8686 |
+
},
|
| 8687 |
+
{
|
| 8688 |
+
"epoch": 1.1,
|
| 8689 |
+
"learning_rate": 0.00025244444444444446,
|
| 8690 |
+
"loss": 2.1675,
|
| 8691 |
+
"step": 1432
|
| 8692 |
+
},
|
| 8693 |
+
{
|
| 8694 |
+
"epoch": 1.1,
|
| 8695 |
+
"learning_rate": 0.000252,
|
| 8696 |
+
"loss": 2.1583,
|
| 8697 |
+
"step": 1433
|
| 8698 |
+
},
|
| 8699 |
+
{
|
| 8700 |
+
"epoch": 1.1,
|
| 8701 |
+
"learning_rate": 0.0002515555555555556,
|
| 8702 |
+
"loss": 2.1534,
|
| 8703 |
+
"step": 1434
|
| 8704 |
+
},
|
| 8705 |
+
{
|
| 8706 |
+
"epoch": 1.1,
|
| 8707 |
+
"learning_rate": 0.00025111111111111113,
|
| 8708 |
+
"loss": 2.1404,
|
| 8709 |
+
"step": 1435
|
| 8710 |
+
},
|
| 8711 |
+
{
|
| 8712 |
+
"epoch": 1.1,
|
| 8713 |
+
"learning_rate": 0.00025066666666666667,
|
| 8714 |
+
"loss": 2.1038,
|
| 8715 |
+
"step": 1436
|
| 8716 |
+
},
|
| 8717 |
+
{
|
| 8718 |
+
"epoch": 1.11,
|
| 8719 |
+
"learning_rate": 0.0002502222222222222,
|
| 8720 |
+
"loss": 2.1705,
|
| 8721 |
+
"step": 1437
|
| 8722 |
+
},
|
| 8723 |
+
{
|
| 8724 |
+
"epoch": 1.11,
|
| 8725 |
+
"learning_rate": 0.0002497777777777778,
|
| 8726 |
+
"loss": 2.1683,
|
| 8727 |
+
"step": 1438
|
| 8728 |
+
},
|
| 8729 |
+
{
|
| 8730 |
+
"epoch": 1.11,
|
| 8731 |
+
"learning_rate": 0.00024933333333333334,
|
| 8732 |
+
"loss": 2.1288,
|
| 8733 |
+
"step": 1439
|
| 8734 |
+
},
|
| 8735 |
+
{
|
| 8736 |
+
"epoch": 1.11,
|
| 8737 |
+
"learning_rate": 0.0002488888888888889,
|
| 8738 |
+
"loss": 2.1502,
|
| 8739 |
+
"step": 1440
|
| 8740 |
+
},
|
| 8741 |
+
{
|
| 8742 |
+
"epoch": 1.11,
|
| 8743 |
+
"learning_rate": 0.0002484444444444444,
|
| 8744 |
+
"loss": 2.1077,
|
| 8745 |
+
"step": 1441
|
| 8746 |
+
},
|
| 8747 |
+
{
|
| 8748 |
+
"epoch": 1.11,
|
| 8749 |
+
"learning_rate": 0.000248,
|
| 8750 |
+
"loss": 2.1261,
|
| 8751 |
+
"step": 1442
|
| 8752 |
+
},
|
| 8753 |
+
{
|
| 8754 |
+
"epoch": 1.11,
|
| 8755 |
+
"learning_rate": 0.0002475555555555556,
|
| 8756 |
+
"loss": 2.1351,
|
| 8757 |
+
"step": 1443
|
| 8758 |
+
},
|
| 8759 |
+
{
|
| 8760 |
+
"epoch": 1.11,
|
| 8761 |
+
"learning_rate": 0.00024711111111111114,
|
| 8762 |
+
"loss": 2.1047,
|
| 8763 |
+
"step": 1444
|
| 8764 |
+
},
|
| 8765 |
+
{
|
| 8766 |
+
"epoch": 1.11,
|
| 8767 |
+
"learning_rate": 0.0002466666666666667,
|
| 8768 |
+
"loss": 2.1223,
|
| 8769 |
+
"step": 1445
|
| 8770 |
+
},
|
| 8771 |
+
{
|
| 8772 |
+
"epoch": 1.11,
|
| 8773 |
+
"learning_rate": 0.0002462222222222222,
|
| 8774 |
+
"loss": 2.0726,
|
| 8775 |
+
"step": 1446
|
| 8776 |
+
},
|
| 8777 |
+
{
|
| 8778 |
+
"epoch": 1.11,
|
| 8779 |
+
"learning_rate": 0.0002457777777777778,
|
| 8780 |
+
"loss": 2.1112,
|
| 8781 |
+
"step": 1447
|
| 8782 |
+
},
|
| 8783 |
+
{
|
| 8784 |
+
"epoch": 1.11,
|
| 8785 |
+
"learning_rate": 0.00024533333333333335,
|
| 8786 |
+
"loss": 2.1283,
|
| 8787 |
+
"step": 1448
|
| 8788 |
+
},
|
| 8789 |
+
{
|
| 8790 |
+
"epoch": 1.11,
|
| 8791 |
+
"learning_rate": 0.0002448888888888889,
|
| 8792 |
+
"loss": 2.1205,
|
| 8793 |
+
"step": 1449
|
| 8794 |
+
},
|
| 8795 |
+
{
|
| 8796 |
+
"epoch": 1.11,
|
| 8797 |
+
"learning_rate": 0.0002444444444444445,
|
| 8798 |
+
"loss": 2.1715,
|
| 8799 |
+
"step": 1450
|
| 8800 |
+
},
|
| 8801 |
+
{
|
| 8802 |
+
"epoch": 1.11,
|
| 8803 |
+
"learning_rate": 0.000244,
|
| 8804 |
+
"loss": 2.1517,
|
| 8805 |
+
"step": 1451
|
| 8806 |
+
},
|
| 8807 |
+
{
|
| 8808 |
+
"epoch": 1.11,
|
| 8809 |
+
"learning_rate": 0.0002435555555555556,
|
| 8810 |
+
"loss": 2.1007,
|
| 8811 |
+
"step": 1452
|
| 8812 |
+
},
|
| 8813 |
+
{
|
| 8814 |
+
"epoch": 1.11,
|
| 8815 |
+
"learning_rate": 0.0002431111111111111,
|
| 8816 |
+
"loss": 2.1804,
|
| 8817 |
+
"step": 1453
|
| 8818 |
+
},
|
| 8819 |
+
{
|
| 8820 |
+
"epoch": 1.11,
|
| 8821 |
+
"learning_rate": 0.0002426666666666667,
|
| 8822 |
+
"loss": 2.1121,
|
| 8823 |
+
"step": 1454
|
| 8824 |
+
},
|
| 8825 |
+
{
|
| 8826 |
+
"epoch": 1.11,
|
| 8827 |
+
"learning_rate": 0.0002422222222222222,
|
| 8828 |
+
"loss": 2.1229,
|
| 8829 |
+
"step": 1455
|
| 8830 |
+
},
|
| 8831 |
+
{
|
| 8832 |
+
"epoch": 1.11,
|
| 8833 |
+
"learning_rate": 0.0002417777777777778,
|
| 8834 |
+
"loss": 2.0234,
|
| 8835 |
+
"step": 1456
|
| 8836 |
+
},
|
| 8837 |
+
{
|
| 8838 |
+
"epoch": 1.12,
|
| 8839 |
+
"learning_rate": 0.00024133333333333336,
|
| 8840 |
+
"loss": 2.0615,
|
| 8841 |
+
"step": 1457
|
| 8842 |
+
},
|
| 8843 |
+
{
|
| 8844 |
+
"epoch": 1.12,
|
| 8845 |
+
"learning_rate": 0.0002408888888888889,
|
| 8846 |
+
"loss": 2.1373,
|
| 8847 |
+
"step": 1458
|
| 8848 |
+
},
|
| 8849 |
+
{
|
| 8850 |
+
"epoch": 1.12,
|
| 8851 |
+
"learning_rate": 0.00024044444444444447,
|
| 8852 |
+
"loss": 2.1106,
|
| 8853 |
+
"step": 1459
|
| 8854 |
+
},
|
| 8855 |
+
{
|
| 8856 |
+
"epoch": 1.12,
|
| 8857 |
+
"learning_rate": 0.00024,
|
| 8858 |
+
"loss": 2.1396,
|
| 8859 |
+
"step": 1460
|
| 8860 |
+
},
|
| 8861 |
+
{
|
| 8862 |
+
"epoch": 1.12,
|
| 8863 |
+
"learning_rate": 0.00023955555555555557,
|
| 8864 |
+
"loss": 2.1163,
|
| 8865 |
+
"step": 1461
|
| 8866 |
+
},
|
| 8867 |
+
{
|
| 8868 |
+
"epoch": 1.12,
|
| 8869 |
+
"learning_rate": 0.0002391111111111111,
|
| 8870 |
+
"loss": 2.0869,
|
| 8871 |
+
"step": 1462
|
| 8872 |
+
},
|
| 8873 |
+
{
|
| 8874 |
+
"epoch": 1.12,
|
| 8875 |
+
"learning_rate": 0.00023866666666666668,
|
| 8876 |
+
"loss": 2.1535,
|
| 8877 |
+
"step": 1463
|
| 8878 |
+
},
|
| 8879 |
+
{
|
| 8880 |
+
"epoch": 1.12,
|
| 8881 |
+
"learning_rate": 0.00023822222222222222,
|
| 8882 |
+
"loss": 2.0669,
|
| 8883 |
+
"step": 1464
|
| 8884 |
+
},
|
| 8885 |
+
{
|
| 8886 |
+
"epoch": 1.12,
|
| 8887 |
+
"learning_rate": 0.00023777777777777778,
|
| 8888 |
+
"loss": 2.1315,
|
| 8889 |
+
"step": 1465
|
| 8890 |
+
},
|
| 8891 |
+
{
|
| 8892 |
+
"epoch": 1.12,
|
| 8893 |
+
"learning_rate": 0.00023733333333333337,
|
| 8894 |
+
"loss": 2.1385,
|
| 8895 |
+
"step": 1466
|
| 8896 |
+
},
|
| 8897 |
+
{
|
| 8898 |
+
"epoch": 1.12,
|
| 8899 |
+
"learning_rate": 0.00023688888888888889,
|
| 8900 |
+
"loss": 2.1548,
|
| 8901 |
+
"step": 1467
|
| 8902 |
+
},
|
| 8903 |
+
{
|
| 8904 |
+
"epoch": 1.12,
|
| 8905 |
+
"learning_rate": 0.00023644444444444448,
|
| 8906 |
+
"loss": 2.1021,
|
| 8907 |
+
"step": 1468
|
| 8908 |
+
},
|
| 8909 |
+
{
|
| 8910 |
+
"epoch": 1.12,
|
| 8911 |
+
"learning_rate": 0.000236,
|
| 8912 |
+
"loss": 2.1829,
|
| 8913 |
+
"step": 1469
|
| 8914 |
+
},
|
| 8915 |
+
{
|
| 8916 |
+
"epoch": 1.12,
|
| 8917 |
+
"learning_rate": 0.00023555555555555558,
|
| 8918 |
+
"loss": 2.1002,
|
| 8919 |
+
"step": 1470
|
| 8920 |
+
},
|
| 8921 |
+
{
|
| 8922 |
+
"epoch": 1.12,
|
| 8923 |
+
"learning_rate": 0.0002351111111111111,
|
| 8924 |
+
"loss": 2.162,
|
| 8925 |
+
"step": 1471
|
| 8926 |
+
},
|
| 8927 |
+
{
|
| 8928 |
+
"epoch": 1.12,
|
| 8929 |
+
"learning_rate": 0.0002346666666666667,
|
| 8930 |
+
"loss": 2.099,
|
| 8931 |
+
"step": 1472
|
| 8932 |
+
},
|
| 8933 |
+
{
|
| 8934 |
+
"epoch": 1.12,
|
| 8935 |
+
"learning_rate": 0.00023422222222222225,
|
| 8936 |
+
"loss": 2.1477,
|
| 8937 |
+
"step": 1473
|
| 8938 |
+
},
|
| 8939 |
+
{
|
| 8940 |
+
"epoch": 1.12,
|
| 8941 |
+
"learning_rate": 0.0002337777777777778,
|
| 8942 |
+
"loss": 2.1612,
|
| 8943 |
+
"step": 1474
|
| 8944 |
+
},
|
| 8945 |
+
{
|
| 8946 |
+
"epoch": 1.12,
|
| 8947 |
+
"learning_rate": 0.00023333333333333336,
|
| 8948 |
+
"loss": 2.0684,
|
| 8949 |
+
"step": 1475
|
| 8950 |
+
},
|
| 8951 |
+
{
|
| 8952 |
+
"epoch": 1.12,
|
| 8953 |
+
"learning_rate": 0.0002328888888888889,
|
| 8954 |
+
"loss": 2.1182,
|
| 8955 |
+
"step": 1476
|
| 8956 |
+
},
|
| 8957 |
+
{
|
| 8958 |
+
"epoch": 1.13,
|
| 8959 |
+
"learning_rate": 0.00023244444444444446,
|
| 8960 |
+
"loss": 2.1298,
|
| 8961 |
+
"step": 1477
|
| 8962 |
+
},
|
| 8963 |
+
{
|
| 8964 |
+
"epoch": 1.13,
|
| 8965 |
+
"learning_rate": 0.000232,
|
| 8966 |
+
"loss": 2.0761,
|
| 8967 |
+
"step": 1478
|
| 8968 |
+
},
|
| 8969 |
+
{
|
| 8970 |
+
"epoch": 1.13,
|
| 8971 |
+
"learning_rate": 0.00023155555555555557,
|
| 8972 |
+
"loss": 2.1133,
|
| 8973 |
+
"step": 1479
|
| 8974 |
+
},
|
| 8975 |
+
{
|
| 8976 |
+
"epoch": 1.13,
|
| 8977 |
+
"learning_rate": 0.0002311111111111111,
|
| 8978 |
+
"loss": 2.1749,
|
| 8979 |
+
"step": 1480
|
| 8980 |
+
},
|
| 8981 |
+
{
|
| 8982 |
+
"epoch": 1.13,
|
| 8983 |
+
"learning_rate": 0.00023066666666666667,
|
| 8984 |
+
"loss": 2.1653,
|
| 8985 |
+
"step": 1481
|
| 8986 |
+
},
|
| 8987 |
+
{
|
| 8988 |
+
"epoch": 1.13,
|
| 8989 |
+
"learning_rate": 0.00023022222222222224,
|
| 8990 |
+
"loss": 2.1304,
|
| 8991 |
+
"step": 1482
|
| 8992 |
+
},
|
| 8993 |
+
{
|
| 8994 |
+
"epoch": 1.13,
|
| 8995 |
+
"learning_rate": 0.00022977777777777778,
|
| 8996 |
+
"loss": 2.0739,
|
| 8997 |
+
"step": 1483
|
| 8998 |
+
},
|
| 8999 |
+
{
|
| 9000 |
+
"epoch": 1.13,
|
| 9001 |
+
"learning_rate": 0.00022933333333333334,
|
| 9002 |
+
"loss": 2.0529,
|
| 9003 |
+
"step": 1484
|
| 9004 |
+
},
|
| 9005 |
+
{
|
| 9006 |
+
"epoch": 1.13,
|
| 9007 |
+
"learning_rate": 0.00022888888888888888,
|
| 9008 |
+
"loss": 2.1136,
|
| 9009 |
+
"step": 1485
|
| 9010 |
+
},
|
| 9011 |
+
{
|
| 9012 |
+
"epoch": 1.13,
|
| 9013 |
+
"learning_rate": 0.00022844444444444447,
|
| 9014 |
+
"loss": 2.11,
|
| 9015 |
+
"step": 1486
|
| 9016 |
+
},
|
| 9017 |
+
{
|
| 9018 |
+
"epoch": 1.13,
|
| 9019 |
+
"learning_rate": 0.00022799999999999999,
|
| 9020 |
+
"loss": 2.1093,
|
| 9021 |
+
"step": 1487
|
| 9022 |
+
},
|
| 9023 |
+
{
|
| 9024 |
+
"epoch": 1.13,
|
| 9025 |
+
"learning_rate": 0.00022755555555555558,
|
| 9026 |
+
"loss": 2.1102,
|
| 9027 |
+
"step": 1488
|
| 9028 |
+
},
|
| 9029 |
+
{
|
| 9030 |
+
"epoch": 1.13,
|
| 9031 |
+
"learning_rate": 0.00022711111111111114,
|
| 9032 |
+
"loss": 2.1033,
|
| 9033 |
+
"step": 1489
|
| 9034 |
+
},
|
| 9035 |
+
{
|
| 9036 |
+
"epoch": 1.13,
|
| 9037 |
+
"learning_rate": 0.00022666666666666668,
|
| 9038 |
+
"loss": 2.1908,
|
| 9039 |
+
"step": 1490
|
| 9040 |
+
},
|
| 9041 |
+
{
|
| 9042 |
+
"epoch": 1.13,
|
| 9043 |
+
"learning_rate": 0.00022622222222222225,
|
| 9044 |
+
"loss": 2.0719,
|
| 9045 |
+
"step": 1491
|
| 9046 |
+
},
|
| 9047 |
+
{
|
| 9048 |
+
"epoch": 1.13,
|
| 9049 |
+
"learning_rate": 0.0002257777777777778,
|
| 9050 |
+
"loss": 2.0688,
|
| 9051 |
+
"step": 1492
|
| 9052 |
+
},
|
| 9053 |
+
{
|
| 9054 |
+
"epoch": 1.13,
|
| 9055 |
+
"learning_rate": 0.00022533333333333335,
|
| 9056 |
+
"loss": 2.1859,
|
| 9057 |
+
"step": 1493
|
| 9058 |
+
},
|
| 9059 |
+
{
|
| 9060 |
+
"epoch": 1.13,
|
| 9061 |
+
"learning_rate": 0.0002248888888888889,
|
| 9062 |
+
"loss": 2.1219,
|
| 9063 |
+
"step": 1494
|
| 9064 |
+
},
|
| 9065 |
+
{
|
| 9066 |
+
"epoch": 1.13,
|
| 9067 |
+
"learning_rate": 0.00022444444444444446,
|
| 9068 |
+
"loss": 2.0813,
|
| 9069 |
+
"step": 1495
|
| 9070 |
+
},
|
| 9071 |
+
{
|
| 9072 |
+
"epoch": 1.13,
|
| 9073 |
+
"learning_rate": 0.00022400000000000002,
|
| 9074 |
+
"loss": 2.133,
|
| 9075 |
+
"step": 1496
|
| 9076 |
+
},
|
| 9077 |
+
{
|
| 9078 |
+
"epoch": 1.14,
|
| 9079 |
+
"learning_rate": 0.00022355555555555556,
|
| 9080 |
+
"loss": 2.2231,
|
| 9081 |
+
"step": 1497
|
| 9082 |
+
},
|
| 9083 |
+
{
|
| 9084 |
+
"epoch": 1.14,
|
| 9085 |
+
"learning_rate": 0.00022311111111111113,
|
| 9086 |
+
"loss": 2.1299,
|
| 9087 |
+
"step": 1498
|
| 9088 |
+
},
|
| 9089 |
+
{
|
| 9090 |
+
"epoch": 1.14,
|
| 9091 |
+
"learning_rate": 0.00022266666666666667,
|
| 9092 |
+
"loss": 2.1343,
|
| 9093 |
+
"step": 1499
|
| 9094 |
+
},
|
| 9095 |
+
{
|
| 9096 |
+
"epoch": 1.14,
|
| 9097 |
+
"learning_rate": 0.00022222222222222223,
|
| 9098 |
+
"loss": 2.1702,
|
| 9099 |
+
"step": 1500
|
| 9100 |
+
},
|
| 9101 |
+
{
|
| 9102 |
+
"epoch": 1.14,
|
| 9103 |
+
"learning_rate": 0.00022177777777777777,
|
| 9104 |
+
"loss": 2.1268,
|
| 9105 |
+
"step": 1501
|
| 9106 |
+
},
|
| 9107 |
+
{
|
| 9108 |
+
"epoch": 1.14,
|
| 9109 |
+
"learning_rate": 0.00022133333333333334,
|
| 9110 |
+
"loss": 2.1472,
|
| 9111 |
+
"step": 1502
|
| 9112 |
+
},
|
| 9113 |
+
{
|
| 9114 |
+
"epoch": 1.14,
|
| 9115 |
+
"learning_rate": 0.00022088888888888888,
|
| 9116 |
+
"loss": 2.1203,
|
| 9117 |
+
"step": 1503
|
| 9118 |
+
},
|
| 9119 |
+
{
|
| 9120 |
+
"epoch": 1.14,
|
| 9121 |
+
"learning_rate": 0.00022044444444444444,
|
| 9122 |
+
"loss": 2.1968,
|
| 9123 |
+
"step": 1504
|
| 9124 |
+
},
|
| 9125 |
+
{
|
| 9126 |
+
"epoch": 1.14,
|
| 9127 |
+
"learning_rate": 0.00022000000000000003,
|
| 9128 |
+
"loss": 2.0619,
|
| 9129 |
+
"step": 1505
|
| 9130 |
+
},
|
| 9131 |
+
{
|
| 9132 |
+
"epoch": 1.14,
|
| 9133 |
+
"learning_rate": 0.00021955555555555555,
|
| 9134 |
+
"loss": 2.1274,
|
| 9135 |
+
"step": 1506
|
| 9136 |
+
},
|
| 9137 |
+
{
|
| 9138 |
+
"epoch": 1.14,
|
| 9139 |
+
"learning_rate": 0.00021911111111111114,
|
| 9140 |
+
"loss": 2.0527,
|
| 9141 |
+
"step": 1507
|
| 9142 |
+
},
|
| 9143 |
+
{
|
| 9144 |
+
"epoch": 1.14,
|
| 9145 |
+
"learning_rate": 0.00021866666666666665,
|
| 9146 |
+
"loss": 2.0104,
|
| 9147 |
+
"step": 1508
|
| 9148 |
+
},
|
| 9149 |
+
{
|
| 9150 |
+
"epoch": 1.14,
|
| 9151 |
+
"learning_rate": 0.00021822222222222224,
|
| 9152 |
+
"loss": 2.1593,
|
| 9153 |
+
"step": 1509
|
| 9154 |
+
},
|
| 9155 |
+
{
|
| 9156 |
+
"epoch": 1.14,
|
| 9157 |
+
"learning_rate": 0.00021777777777777778,
|
| 9158 |
+
"loss": 2.0624,
|
| 9159 |
+
"step": 1510
|
| 9160 |
+
},
|
| 9161 |
+
{
|
| 9162 |
+
"epoch": 1.14,
|
| 9163 |
+
"learning_rate": 0.00021733333333333335,
|
| 9164 |
+
"loss": 2.1088,
|
| 9165 |
+
"step": 1511
|
| 9166 |
+
},
|
| 9167 |
+
{
|
| 9168 |
+
"epoch": 1.14,
|
| 9169 |
+
"learning_rate": 0.00021688888888888891,
|
| 9170 |
+
"loss": 2.0986,
|
| 9171 |
+
"step": 1512
|
| 9172 |
+
},
|
| 9173 |
+
{
|
| 9174 |
+
"epoch": 1.14,
|
| 9175 |
+
"learning_rate": 0.00021644444444444445,
|
| 9176 |
+
"loss": 2.0843,
|
| 9177 |
+
"step": 1513
|
| 9178 |
+
},
|
| 9179 |
+
{
|
| 9180 |
+
"epoch": 1.14,
|
| 9181 |
+
"learning_rate": 0.00021600000000000002,
|
| 9182 |
+
"loss": 2.1256,
|
| 9183 |
+
"step": 1514
|
| 9184 |
+
},
|
| 9185 |
+
{
|
| 9186 |
+
"epoch": 1.14,
|
| 9187 |
+
"learning_rate": 0.00021555555555555556,
|
| 9188 |
+
"loss": 2.1612,
|
| 9189 |
+
"step": 1515
|
| 9190 |
+
},
|
| 9191 |
+
{
|
| 9192 |
+
"epoch": 1.14,
|
| 9193 |
+
"learning_rate": 0.00021511111111111112,
|
| 9194 |
+
"loss": 2.1048,
|
| 9195 |
+
"step": 1516
|
| 9196 |
+
},
|
| 9197 |
+
{
|
| 9198 |
+
"epoch": 1.15,
|
| 9199 |
+
"learning_rate": 0.00021466666666666666,
|
| 9200 |
+
"loss": 2.1749,
|
| 9201 |
+
"step": 1517
|
| 9202 |
+
},
|
| 9203 |
+
{
|
| 9204 |
+
"epoch": 1.15,
|
| 9205 |
+
"learning_rate": 0.00021422222222222223,
|
| 9206 |
+
"loss": 2.13,
|
| 9207 |
+
"step": 1518
|
| 9208 |
+
},
|
| 9209 |
+
{
|
| 9210 |
+
"epoch": 1.15,
|
| 9211 |
+
"learning_rate": 0.00021377777777777782,
|
| 9212 |
+
"loss": 2.1937,
|
| 9213 |
+
"step": 1519
|
| 9214 |
+
},
|
| 9215 |
+
{
|
| 9216 |
+
"epoch": 1.15,
|
| 9217 |
+
"learning_rate": 0.00021333333333333333,
|
| 9218 |
+
"loss": 2.1608,
|
| 9219 |
+
"step": 1520
|
| 9220 |
+
},
|
| 9221 |
+
{
|
| 9222 |
+
"epoch": 1.15,
|
| 9223 |
+
"learning_rate": 0.00021288888888888893,
|
| 9224 |
+
"loss": 2.1284,
|
| 9225 |
+
"step": 1521
|
| 9226 |
+
},
|
| 9227 |
+
{
|
| 9228 |
+
"epoch": 1.15,
|
| 9229 |
+
"learning_rate": 0.00021244444444444444,
|
| 9230 |
+
"loss": 2.1542,
|
| 9231 |
+
"step": 1522
|
| 9232 |
+
},
|
| 9233 |
+
{
|
| 9234 |
+
"epoch": 1.15,
|
| 9235 |
+
"learning_rate": 0.00021200000000000003,
|
| 9236 |
+
"loss": 2.1237,
|
| 9237 |
+
"step": 1523
|
| 9238 |
+
},
|
| 9239 |
+
{
|
| 9240 |
+
"epoch": 1.15,
|
| 9241 |
+
"learning_rate": 0.00021155555555555554,
|
| 9242 |
+
"loss": 2.1512,
|
| 9243 |
+
"step": 1524
|
| 9244 |
+
},
|
| 9245 |
+
{
|
| 9246 |
+
"epoch": 1.15,
|
| 9247 |
+
"learning_rate": 0.00021111111111111113,
|
| 9248 |
+
"loss": 2.0976,
|
| 9249 |
+
"step": 1525
|
| 9250 |
+
},
|
| 9251 |
+
{
|
| 9252 |
+
"epoch": 1.15,
|
| 9253 |
+
"learning_rate": 0.00021066666666666665,
|
| 9254 |
+
"loss": 2.135,
|
| 9255 |
+
"step": 1526
|
| 9256 |
+
},
|
| 9257 |
+
{
|
| 9258 |
+
"epoch": 1.15,
|
| 9259 |
+
"learning_rate": 0.00021022222222222224,
|
| 9260 |
+
"loss": 2.1227,
|
| 9261 |
+
"step": 1527
|
| 9262 |
+
},
|
| 9263 |
+
{
|
| 9264 |
+
"epoch": 1.15,
|
| 9265 |
+
"learning_rate": 0.0002097777777777778,
|
| 9266 |
+
"loss": 2.114,
|
| 9267 |
+
"step": 1528
|
| 9268 |
+
},
|
| 9269 |
+
{
|
| 9270 |
+
"epoch": 1.15,
|
| 9271 |
+
"learning_rate": 0.00020933333333333334,
|
| 9272 |
+
"loss": 2.141,
|
| 9273 |
+
"step": 1529
|
| 9274 |
+
},
|
| 9275 |
+
{
|
| 9276 |
+
"epoch": 1.15,
|
| 9277 |
+
"learning_rate": 0.0002088888888888889,
|
| 9278 |
+
"loss": 2.1855,
|
| 9279 |
+
"step": 1530
|
| 9280 |
+
},
|
| 9281 |
+
{
|
| 9282 |
+
"epoch": 1.15,
|
| 9283 |
+
"learning_rate": 0.00020844444444444445,
|
| 9284 |
+
"loss": 2.1343,
|
| 9285 |
+
"step": 1531
|
| 9286 |
+
},
|
| 9287 |
+
{
|
| 9288 |
+
"epoch": 1.15,
|
| 9289 |
+
"learning_rate": 0.00020800000000000001,
|
| 9290 |
+
"loss": 2.0732,
|
| 9291 |
+
"step": 1532
|
| 9292 |
+
},
|
| 9293 |
+
{
|
| 9294 |
+
"epoch": 1.15,
|
| 9295 |
+
"learning_rate": 0.00020755555555555555,
|
| 9296 |
+
"loss": 2.1155,
|
| 9297 |
+
"step": 1533
|
| 9298 |
+
},
|
| 9299 |
+
{
|
| 9300 |
+
"epoch": 1.15,
|
| 9301 |
+
"learning_rate": 0.00020711111111111112,
|
| 9302 |
+
"loss": 2.1146,
|
| 9303 |
+
"step": 1534
|
| 9304 |
+
},
|
| 9305 |
+
{
|
| 9306 |
+
"epoch": 1.15,
|
| 9307 |
+
"learning_rate": 0.00020666666666666668,
|
| 9308 |
+
"loss": 2.1464,
|
| 9309 |
+
"step": 1535
|
| 9310 |
+
},
|
| 9311 |
+
{
|
| 9312 |
+
"epoch": 1.15,
|
| 9313 |
+
"learning_rate": 0.00020622222222222222,
|
| 9314 |
+
"loss": 2.1327,
|
| 9315 |
+
"step": 1536
|
| 9316 |
+
},
|
| 9317 |
+
{
|
| 9318 |
+
"epoch": 1.16,
|
| 9319 |
+
"learning_rate": 0.0002057777777777778,
|
| 9320 |
+
"loss": 2.1498,
|
| 9321 |
+
"step": 1537
|
| 9322 |
+
},
|
| 9323 |
+
{
|
| 9324 |
+
"epoch": 1.16,
|
| 9325 |
+
"learning_rate": 0.00020533333333333333,
|
| 9326 |
+
"loss": 2.133,
|
| 9327 |
+
"step": 1538
|
| 9328 |
+
},
|
| 9329 |
+
{
|
| 9330 |
+
"epoch": 1.16,
|
| 9331 |
+
"learning_rate": 0.00020488888888888892,
|
| 9332 |
+
"loss": 2.1969,
|
| 9333 |
+
"step": 1539
|
| 9334 |
+
},
|
| 9335 |
+
{
|
| 9336 |
+
"epoch": 1.16,
|
| 9337 |
+
"learning_rate": 0.00020444444444444443,
|
| 9338 |
+
"loss": 2.1602,
|
| 9339 |
+
"step": 1540
|
| 9340 |
+
},
|
| 9341 |
+
{
|
| 9342 |
+
"epoch": 1.16,
|
| 9343 |
+
"learning_rate": 0.00020400000000000003,
|
| 9344 |
+
"loss": 2.1356,
|
| 9345 |
+
"step": 1541
|
| 9346 |
+
},
|
| 9347 |
+
{
|
| 9348 |
+
"epoch": 1.16,
|
| 9349 |
+
"learning_rate": 0.00020355555555555554,
|
| 9350 |
+
"loss": 2.1506,
|
| 9351 |
+
"step": 1542
|
| 9352 |
+
},
|
| 9353 |
+
{
|
| 9354 |
+
"epoch": 1.16,
|
| 9355 |
+
"learning_rate": 0.00020311111111111113,
|
| 9356 |
+
"loss": 2.0696,
|
| 9357 |
+
"step": 1543
|
| 9358 |
+
},
|
| 9359 |
+
{
|
| 9360 |
+
"epoch": 1.16,
|
| 9361 |
+
"learning_rate": 0.0002026666666666667,
|
| 9362 |
+
"loss": 2.1018,
|
| 9363 |
+
"step": 1544
|
| 9364 |
+
},
|
| 9365 |
+
{
|
| 9366 |
+
"epoch": 1.16,
|
| 9367 |
+
"learning_rate": 0.00020222222222222223,
|
| 9368 |
+
"loss": 2.1302,
|
| 9369 |
+
"step": 1545
|
| 9370 |
+
},
|
| 9371 |
+
{
|
| 9372 |
+
"epoch": 1.16,
|
| 9373 |
+
"learning_rate": 0.0002017777777777778,
|
| 9374 |
+
"loss": 2.1233,
|
| 9375 |
+
"step": 1546
|
| 9376 |
+
},
|
| 9377 |
+
{
|
| 9378 |
+
"epoch": 1.16,
|
| 9379 |
+
"learning_rate": 0.00020133333333333334,
|
| 9380 |
+
"loss": 2.1112,
|
| 9381 |
+
"step": 1547
|
| 9382 |
+
},
|
| 9383 |
+
{
|
| 9384 |
+
"epoch": 1.16,
|
| 9385 |
+
"learning_rate": 0.0002008888888888889,
|
| 9386 |
+
"loss": 2.219,
|
| 9387 |
+
"step": 1548
|
| 9388 |
+
},
|
| 9389 |
+
{
|
| 9390 |
+
"epoch": 1.16,
|
| 9391 |
+
"learning_rate": 0.00020044444444444444,
|
| 9392 |
+
"loss": 2.1709,
|
| 9393 |
+
"step": 1549
|
| 9394 |
+
},
|
| 9395 |
+
{
|
| 9396 |
+
"epoch": 1.16,
|
| 9397 |
+
"learning_rate": 0.0002,
|
| 9398 |
+
"loss": 2.077,
|
| 9399 |
+
"step": 1550
|
| 9400 |
+
},
|
| 9401 |
+
{
|
| 9402 |
+
"epoch": 1.16,
|
| 9403 |
+
"learning_rate": 0.00019955555555555558,
|
| 9404 |
+
"loss": 2.167,
|
| 9405 |
+
"step": 1551
|
| 9406 |
+
},
|
| 9407 |
+
{
|
| 9408 |
+
"epoch": 1.16,
|
| 9409 |
+
"learning_rate": 0.00019911111111111111,
|
| 9410 |
+
"loss": 2.0899,
|
| 9411 |
+
"step": 1552
|
| 9412 |
+
},
|
| 9413 |
+
{
|
| 9414 |
+
"epoch": 1.16,
|
| 9415 |
+
"learning_rate": 0.00019866666666666668,
|
| 9416 |
+
"loss": 2.1202,
|
| 9417 |
+
"step": 1553
|
| 9418 |
+
},
|
| 9419 |
+
{
|
| 9420 |
+
"epoch": 1.16,
|
| 9421 |
+
"learning_rate": 0.00019822222222222225,
|
| 9422 |
+
"loss": 2.1057,
|
| 9423 |
+
"step": 1554
|
| 9424 |
+
},
|
| 9425 |
+
{
|
| 9426 |
+
"epoch": 1.16,
|
| 9427 |
+
"learning_rate": 0.00019777777777777778,
|
| 9428 |
+
"loss": 2.0946,
|
| 9429 |
+
"step": 1555
|
| 9430 |
+
},
|
| 9431 |
+
{
|
| 9432 |
+
"epoch": 1.16,
|
| 9433 |
+
"learning_rate": 0.00019733333333333335,
|
| 9434 |
+
"loss": 2.1357,
|
| 9435 |
+
"step": 1556
|
| 9436 |
+
},
|
| 9437 |
+
{
|
| 9438 |
+
"epoch": 1.17,
|
| 9439 |
+
"learning_rate": 0.0001968888888888889,
|
| 9440 |
+
"loss": 2.1182,
|
| 9441 |
+
"step": 1557
|
| 9442 |
+
},
|
| 9443 |
+
{
|
| 9444 |
+
"epoch": 1.17,
|
| 9445 |
+
"learning_rate": 0.00019644444444444445,
|
| 9446 |
+
"loss": 2.112,
|
| 9447 |
+
"step": 1558
|
| 9448 |
+
},
|
| 9449 |
+
{
|
| 9450 |
+
"epoch": 1.17,
|
| 9451 |
+
"learning_rate": 0.000196,
|
| 9452 |
+
"loss": 2.1319,
|
| 9453 |
+
"step": 1559
|
| 9454 |
+
},
|
| 9455 |
+
{
|
| 9456 |
+
"epoch": 1.17,
|
| 9457 |
+
"learning_rate": 0.00019555555555555556,
|
| 9458 |
+
"loss": 2.1783,
|
| 9459 |
+
"step": 1560
|
| 9460 |
+
},
|
| 9461 |
+
{
|
| 9462 |
+
"epoch": 1.17,
|
| 9463 |
+
"learning_rate": 0.0001951111111111111,
|
| 9464 |
+
"loss": 2.1374,
|
| 9465 |
+
"step": 1561
|
| 9466 |
+
},
|
| 9467 |
+
{
|
| 9468 |
+
"epoch": 1.17,
|
| 9469 |
+
"learning_rate": 0.0001946666666666667,
|
| 9470 |
+
"loss": 2.1862,
|
| 9471 |
+
"step": 1562
|
| 9472 |
+
},
|
| 9473 |
+
{
|
| 9474 |
+
"epoch": 1.17,
|
| 9475 |
+
"learning_rate": 0.00019422222222222223,
|
| 9476 |
+
"loss": 2.1993,
|
| 9477 |
+
"step": 1563
|
| 9478 |
+
},
|
| 9479 |
+
{
|
| 9480 |
+
"epoch": 1.17,
|
| 9481 |
+
"learning_rate": 0.0001937777777777778,
|
| 9482 |
+
"loss": 2.0878,
|
| 9483 |
+
"step": 1564
|
| 9484 |
+
},
|
| 9485 |
+
{
|
| 9486 |
+
"epoch": 1.17,
|
| 9487 |
+
"learning_rate": 0.00019333333333333333,
|
| 9488 |
+
"loss": 2.1082,
|
| 9489 |
+
"step": 1565
|
| 9490 |
+
},
|
| 9491 |
+
{
|
| 9492 |
+
"epoch": 1.17,
|
| 9493 |
+
"learning_rate": 0.0001928888888888889,
|
| 9494 |
+
"loss": 2.1007,
|
| 9495 |
+
"step": 1566
|
| 9496 |
+
},
|
| 9497 |
+
{
|
| 9498 |
+
"epoch": 1.17,
|
| 9499 |
+
"learning_rate": 0.00019244444444444444,
|
| 9500 |
+
"loss": 2.1888,
|
| 9501 |
+
"step": 1567
|
| 9502 |
+
},
|
| 9503 |
+
{
|
| 9504 |
+
"epoch": 1.17,
|
| 9505 |
+
"learning_rate": 0.000192,
|
| 9506 |
+
"loss": 2.1479,
|
| 9507 |
+
"step": 1568
|
| 9508 |
+
},
|
| 9509 |
+
{
|
| 9510 |
+
"epoch": 1.17,
|
| 9511 |
+
"learning_rate": 0.00019155555555555554,
|
| 9512 |
+
"loss": 2.235,
|
| 9513 |
+
"step": 1569
|
| 9514 |
+
},
|
| 9515 |
+
{
|
| 9516 |
+
"epoch": 1.17,
|
| 9517 |
+
"learning_rate": 0.00019111111111111114,
|
| 9518 |
+
"loss": 2.1963,
|
| 9519 |
+
"step": 1570
|
| 9520 |
+
},
|
| 9521 |
+
{
|
| 9522 |
+
"epoch": 1.17,
|
| 9523 |
+
"learning_rate": 0.00019066666666666668,
|
| 9524 |
+
"loss": 2.0638,
|
| 9525 |
+
"step": 1571
|
| 9526 |
+
},
|
| 9527 |
+
{
|
| 9528 |
+
"epoch": 1.17,
|
| 9529 |
+
"learning_rate": 0.00019022222222222224,
|
| 9530 |
+
"loss": 2.1389,
|
| 9531 |
+
"step": 1572
|
| 9532 |
+
},
|
| 9533 |
+
{
|
| 9534 |
+
"epoch": 1.17,
|
| 9535 |
+
"learning_rate": 0.00018977777777777778,
|
| 9536 |
+
"loss": 2.0982,
|
| 9537 |
+
"step": 1573
|
| 9538 |
+
},
|
| 9539 |
+
{
|
| 9540 |
+
"epoch": 1.17,
|
| 9541 |
+
"learning_rate": 0.00018933333333333335,
|
| 9542 |
+
"loss": 2.1129,
|
| 9543 |
+
"step": 1574
|
| 9544 |
+
},
|
| 9545 |
+
{
|
| 9546 |
+
"epoch": 1.17,
|
| 9547 |
+
"learning_rate": 0.00018888888888888888,
|
| 9548 |
+
"loss": 2.1199,
|
| 9549 |
+
"step": 1575
|
| 9550 |
+
},
|
| 9551 |
+
{
|
| 9552 |
+
"epoch": 1.17,
|
| 9553 |
+
"learning_rate": 0.00018844444444444445,
|
| 9554 |
+
"loss": 2.0823,
|
| 9555 |
+
"step": 1576
|
| 9556 |
+
},
|
| 9557 |
+
{
|
| 9558 |
+
"epoch": 1.18,
|
| 9559 |
+
"learning_rate": 0.000188,
|
| 9560 |
+
"loss": 2.1113,
|
| 9561 |
+
"step": 1577
|
| 9562 |
+
},
|
| 9563 |
+
{
|
| 9564 |
+
"epoch": 1.18,
|
| 9565 |
+
"learning_rate": 0.00018755555555555558,
|
| 9566 |
+
"loss": 2.1034,
|
| 9567 |
+
"step": 1578
|
| 9568 |
+
},
|
| 9569 |
+
{
|
| 9570 |
+
"epoch": 1.18,
|
| 9571 |
+
"learning_rate": 0.00018711111111111112,
|
| 9572 |
+
"loss": 2.0663,
|
| 9573 |
+
"step": 1579
|
| 9574 |
+
},
|
| 9575 |
+
{
|
| 9576 |
+
"epoch": 1.18,
|
| 9577 |
+
"learning_rate": 0.0001866666666666667,
|
| 9578 |
+
"loss": 2.0847,
|
| 9579 |
+
"step": 1580
|
| 9580 |
+
},
|
| 9581 |
+
{
|
| 9582 |
+
"epoch": 1.18,
|
| 9583 |
+
"learning_rate": 0.00018622222222222223,
|
| 9584 |
+
"loss": 2.155,
|
| 9585 |
+
"step": 1581
|
| 9586 |
+
},
|
| 9587 |
+
{
|
| 9588 |
+
"epoch": 1.18,
|
| 9589 |
+
"learning_rate": 0.0001857777777777778,
|
| 9590 |
+
"loss": 2.1369,
|
| 9591 |
+
"step": 1582
|
| 9592 |
+
},
|
| 9593 |
+
{
|
| 9594 |
+
"epoch": 1.18,
|
| 9595 |
+
"learning_rate": 0.00018533333333333333,
|
| 9596 |
+
"loss": 2.0877,
|
| 9597 |
+
"step": 1583
|
| 9598 |
+
},
|
| 9599 |
+
{
|
| 9600 |
+
"epoch": 1.18,
|
| 9601 |
+
"learning_rate": 0.0001848888888888889,
|
| 9602 |
+
"loss": 2.0629,
|
| 9603 |
+
"step": 1584
|
| 9604 |
+
},
|
| 9605 |
+
{
|
| 9606 |
+
"epoch": 1.18,
|
| 9607 |
+
"learning_rate": 0.00018444444444444446,
|
| 9608 |
+
"loss": 2.1677,
|
| 9609 |
+
"step": 1585
|
| 9610 |
+
},
|
| 9611 |
+
{
|
| 9612 |
+
"epoch": 1.18,
|
| 9613 |
+
"learning_rate": 0.00018400000000000003,
|
| 9614 |
+
"loss": 2.0936,
|
| 9615 |
+
"step": 1586
|
| 9616 |
+
},
|
| 9617 |
+
{
|
| 9618 |
+
"epoch": 1.18,
|
| 9619 |
+
"learning_rate": 0.00018355555555555557,
|
| 9620 |
+
"loss": 2.0795,
|
| 9621 |
+
"step": 1587
|
| 9622 |
+
},
|
| 9623 |
+
{
|
| 9624 |
+
"epoch": 1.18,
|
| 9625 |
+
"learning_rate": 0.00018311111111111113,
|
| 9626 |
+
"loss": 2.0966,
|
| 9627 |
+
"step": 1588
|
| 9628 |
+
},
|
| 9629 |
+
{
|
| 9630 |
+
"epoch": 1.18,
|
| 9631 |
+
"learning_rate": 0.00018266666666666667,
|
| 9632 |
+
"loss": 2.1501,
|
| 9633 |
+
"step": 1589
|
| 9634 |
+
},
|
| 9635 |
+
{
|
| 9636 |
+
"epoch": 1.18,
|
| 9637 |
+
"learning_rate": 0.00018222222222222224,
|
| 9638 |
+
"loss": 2.1039,
|
| 9639 |
+
"step": 1590
|
| 9640 |
+
},
|
| 9641 |
+
{
|
| 9642 |
+
"epoch": 1.18,
|
| 9643 |
+
"learning_rate": 0.00018177777777777778,
|
| 9644 |
+
"loss": 2.1578,
|
| 9645 |
+
"step": 1591
|
| 9646 |
+
},
|
| 9647 |
+
{
|
| 9648 |
+
"epoch": 1.18,
|
| 9649 |
+
"learning_rate": 0.00018133333333333334,
|
| 9650 |
+
"loss": 2.0771,
|
| 9651 |
+
"step": 1592
|
| 9652 |
+
},
|
| 9653 |
+
{
|
| 9654 |
+
"epoch": 1.18,
|
| 9655 |
+
"learning_rate": 0.0001808888888888889,
|
| 9656 |
+
"loss": 2.126,
|
| 9657 |
+
"step": 1593
|
| 9658 |
+
},
|
| 9659 |
+
{
|
| 9660 |
+
"epoch": 1.18,
|
| 9661 |
+
"learning_rate": 0.00018044444444444447,
|
| 9662 |
+
"loss": 2.0727,
|
| 9663 |
+
"step": 1594
|
| 9664 |
+
},
|
| 9665 |
+
{
|
| 9666 |
+
"epoch": 1.18,
|
| 9667 |
+
"learning_rate": 0.00018,
|
| 9668 |
+
"loss": 2.1178,
|
| 9669 |
+
"step": 1595
|
| 9670 |
+
},
|
| 9671 |
+
{
|
| 9672 |
+
"epoch": 1.18,
|
| 9673 |
+
"learning_rate": 0.00017955555555555558,
|
| 9674 |
+
"loss": 2.1327,
|
| 9675 |
+
"step": 1596
|
| 9676 |
+
},
|
| 9677 |
+
{
|
| 9678 |
+
"epoch": 1.19,
|
| 9679 |
+
"learning_rate": 0.00017911111111111112,
|
| 9680 |
+
"loss": 2.1465,
|
| 9681 |
+
"step": 1597
|
| 9682 |
+
},
|
| 9683 |
+
{
|
| 9684 |
+
"epoch": 1.19,
|
| 9685 |
+
"learning_rate": 0.00017866666666666668,
|
| 9686 |
+
"loss": 2.1755,
|
| 9687 |
+
"step": 1598
|
| 9688 |
+
},
|
| 9689 |
+
{
|
| 9690 |
+
"epoch": 1.19,
|
| 9691 |
+
"learning_rate": 0.00017822222222222222,
|
| 9692 |
+
"loss": 2.1052,
|
| 9693 |
+
"step": 1599
|
| 9694 |
+
},
|
| 9695 |
+
{
|
| 9696 |
+
"epoch": 1.19,
|
| 9697 |
+
"learning_rate": 0.00017777777777777779,
|
| 9698 |
+
"loss": 2.1484,
|
| 9699 |
+
"step": 1600
|
| 9700 |
+
},
|
| 9701 |
+
{
|
| 9702 |
+
"epoch": 1.19,
|
| 9703 |
+
"eval_gen_len": 1023.0,
|
| 9704 |
+
"eval_loss": 2.1070616245269775,
|
| 9705 |
+
"eval_rouge1": 14.2289,
|
| 9706 |
+
"eval_rouge2": 4.7872,
|
| 9707 |
+
"eval_rougeL": 8.9958,
|
| 9708 |
+
"eval_rougeLsum": 9.0075,
|
| 9709 |
+
"eval_runtime": 9037.4832,
|
| 9710 |
+
"eval_samples_per_second": 0.055,
|
| 9711 |
+
"eval_steps_per_second": 0.055,
|
| 9712 |
+
"step": 1600
|
| 9713 |
}
|
| 9714 |
],
|
| 9715 |
"max_steps": 2000,
|
| 9716 |
"num_train_epochs": 9223372036854775807,
|
| 9717 |
+
"total_flos": 1.2852794184310794e+18,
|
| 9718 |
"trial_name": null,
|
| 9719 |
"trial_params": null
|
| 9720 |
}
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 2368281769
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:728b6dfaa1c989be32093514baf4c7d88f3ff72d43aa5c614b342b108779fd92
|
| 3 |
size 2368281769
|