Training in progress, step 5945, checkpoint
Browse files- last-checkpoint/optimizer.pt +1 -1
- last-checkpoint/rng_state.pth +1 -1
- last-checkpoint/scheduler.pt +1 -1
- last-checkpoint/trainer_state.json +1467 -4
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 516802328
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:70c3b19bc4eb8be76e3f5a614ad9651dac58cc5c1cf201702fc3087309d23ca7
|
| 3 |
size 516802328
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3662bc0e65237626b783f1e110f32d0f58b50371353139f3380aa21907d63486
|
| 3 |
size 14244
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ee9c30235a00d4e956792d86eabe739f21989b7ae3dfec94f60c7acfc17fd683
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -40159,6 +40159,1469 @@
|
|
| 40159 |
"learning_rate": 3.151421924994513e-07,
|
| 40160 |
"loss": 0.0,
|
| 40161 |
"step": 5736
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 40162 |
}
|
| 40163 |
],
|
| 40164 |
"logging_steps": 1,
|
|
@@ -40173,12 +41636,12 @@
|
|
| 40173 |
"should_evaluate": false,
|
| 40174 |
"should_log": false,
|
| 40175 |
"should_save": true,
|
| 40176 |
-
"should_training_stop":
|
| 40177 |
},
|
| 40178 |
"attributes": {}
|
| 40179 |
}
|
| 40180 |
},
|
| 40181 |
-
"total_flos": 1.
|
| 40182 |
"train_batch_size": 8,
|
| 40183 |
"trial_name": null,
|
| 40184 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9999159027836179,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 5945,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 40159 |
"learning_rate": 3.151421924994513e-07,
|
| 40160 |
"loss": 0.0,
|
| 40161 |
"step": 5736
|
| 40162 |
+
},
|
| 40163 |
+
{
|
| 40164 |
+
"epoch": 0.9649314607686486,
|
| 40165 |
+
"grad_norm": NaN,
|
| 40166 |
+
"learning_rate": 3.121368265136704e-07,
|
| 40167 |
+
"loss": 0.0,
|
| 40168 |
+
"step": 5737
|
| 40169 |
+
},
|
| 40170 |
+
{
|
| 40171 |
+
"epoch": 0.9650996552014128,
|
| 40172 |
+
"grad_norm": NaN,
|
| 40173 |
+
"learning_rate": 3.091458147950255e-07,
|
| 40174 |
+
"loss": 0.0,
|
| 40175 |
+
"step": 5738
|
| 40176 |
+
},
|
| 40177 |
+
{
|
| 40178 |
+
"epoch": 0.9652678496341771,
|
| 40179 |
+
"grad_norm": NaN,
|
| 40180 |
+
"learning_rate": 3.0616915820758095e-07,
|
| 40181 |
+
"loss": 0.0,
|
| 40182 |
+
"step": 5739
|
| 40183 |
+
},
|
| 40184 |
+
{
|
| 40185 |
+
"epoch": 0.9654360440669414,
|
| 40186 |
+
"grad_norm": NaN,
|
| 40187 |
+
"learning_rate": 3.0320685761127123e-07,
|
| 40188 |
+
"loss": 0.0,
|
| 40189 |
+
"step": 5740
|
| 40190 |
+
},
|
| 40191 |
+
{
|
| 40192 |
+
"epoch": 0.9656042384997057,
|
| 40193 |
+
"grad_norm": NaN,
|
| 40194 |
+
"learning_rate": 3.002589138618561e-07,
|
| 40195 |
+
"loss": 0.0,
|
| 40196 |
+
"step": 5741
|
| 40197 |
+
},
|
| 40198 |
+
{
|
| 40199 |
+
"epoch": 0.96577243293247,
|
| 40200 |
+
"grad_norm": NaN,
|
| 40201 |
+
"learning_rate": 2.973253278109767e-07,
|
| 40202 |
+
"loss": 0.0,
|
| 40203 |
+
"step": 5742
|
| 40204 |
+
},
|
| 40205 |
+
{
|
| 40206 |
+
"epoch": 0.9659406273652342,
|
| 40207 |
+
"grad_norm": NaN,
|
| 40208 |
+
"learning_rate": 2.9440610030610494e-07,
|
| 40209 |
+
"loss": 0.0,
|
| 40210 |
+
"step": 5743
|
| 40211 |
+
},
|
| 40212 |
+
{
|
| 40213 |
+
"epoch": 0.9661088217979985,
|
| 40214 |
+
"grad_norm": NaN,
|
| 40215 |
+
"learning_rate": 2.91501232190583e-07,
|
| 40216 |
+
"loss": 0.0,
|
| 40217 |
+
"step": 5744
|
| 40218 |
+
},
|
| 40219 |
+
{
|
| 40220 |
+
"epoch": 0.9662770162307628,
|
| 40221 |
+
"grad_norm": NaN,
|
| 40222 |
+
"learning_rate": 2.8861072430358404e-07,
|
| 40223 |
+
"loss": 0.0,
|
| 40224 |
+
"step": 5745
|
| 40225 |
+
},
|
| 40226 |
+
{
|
| 40227 |
+
"epoch": 0.9664452106635271,
|
| 40228 |
+
"grad_norm": NaN,
|
| 40229 |
+
"learning_rate": 2.8573457748014564e-07,
|
| 40230 |
+
"loss": 0.0,
|
| 40231 |
+
"step": 5746
|
| 40232 |
+
},
|
| 40233 |
+
{
|
| 40234 |
+
"epoch": 0.9666134050962913,
|
| 40235 |
+
"grad_norm": NaN,
|
| 40236 |
+
"learning_rate": 2.8287279255115873e-07,
|
| 40237 |
+
"loss": 0.0,
|
| 40238 |
+
"step": 5747
|
| 40239 |
+
},
|
| 40240 |
+
{
|
| 40241 |
+
"epoch": 0.9667815995290556,
|
| 40242 |
+
"grad_norm": NaN,
|
| 40243 |
+
"learning_rate": 2.800253703433564e-07,
|
| 40244 |
+
"loss": 0.0,
|
| 40245 |
+
"step": 5748
|
| 40246 |
+
},
|
| 40247 |
+
{
|
| 40248 |
+
"epoch": 0.9669497939618199,
|
| 40249 |
+
"grad_norm": NaN,
|
| 40250 |
+
"learning_rate": 2.7719231167933067e-07,
|
| 40251 |
+
"loss": 0.0,
|
| 40252 |
+
"step": 5749
|
| 40253 |
+
},
|
| 40254 |
+
{
|
| 40255 |
+
"epoch": 0.9671179883945842,
|
| 40256 |
+
"grad_norm": NaN,
|
| 40257 |
+
"learning_rate": 2.743736173775213e-07,
|
| 40258 |
+
"loss": 0.0,
|
| 40259 |
+
"step": 5750
|
| 40260 |
+
},
|
| 40261 |
+
{
|
| 40262 |
+
"epoch": 0.9672861828273485,
|
| 40263 |
+
"grad_norm": NaN,
|
| 40264 |
+
"learning_rate": 2.715692882522103e-07,
|
| 40265 |
+
"loss": 0.0,
|
| 40266 |
+
"step": 5751
|
| 40267 |
+
},
|
| 40268 |
+
{
|
| 40269 |
+
"epoch": 0.9674543772601127,
|
| 40270 |
+
"grad_norm": NaN,
|
| 40271 |
+
"learning_rate": 2.687793251135384e-07,
|
| 40272 |
+
"loss": 0.0,
|
| 40273 |
+
"step": 5752
|
| 40274 |
+
},
|
| 40275 |
+
{
|
| 40276 |
+
"epoch": 0.967622571692877,
|
| 40277 |
+
"grad_norm": NaN,
|
| 40278 |
+
"learning_rate": 2.6600372876750544e-07,
|
| 40279 |
+
"loss": 0.0,
|
| 40280 |
+
"step": 5753
|
| 40281 |
+
},
|
| 40282 |
+
{
|
| 40283 |
+
"epoch": 0.9677907661256412,
|
| 40284 |
+
"grad_norm": NaN,
|
| 40285 |
+
"learning_rate": 2.6324250001593664e-07,
|
| 40286 |
+
"loss": 0.0,
|
| 40287 |
+
"step": 5754
|
| 40288 |
+
},
|
| 40289 |
+
{
|
| 40290 |
+
"epoch": 0.9679589605584055,
|
| 40291 |
+
"grad_norm": NaN,
|
| 40292 |
+
"learning_rate": 2.604956396565328e-07,
|
| 40293 |
+
"loss": 0.0,
|
| 40294 |
+
"step": 5755
|
| 40295 |
+
},
|
| 40296 |
+
{
|
| 40297 |
+
"epoch": 0.9681271549911697,
|
| 40298 |
+
"grad_norm": NaN,
|
| 40299 |
+
"learning_rate": 2.577631484828147e-07,
|
| 40300 |
+
"loss": 0.0,
|
| 40301 |
+
"step": 5756
|
| 40302 |
+
},
|
| 40303 |
+
{
|
| 40304 |
+
"epoch": 0.968295349423934,
|
| 40305 |
+
"grad_norm": NaN,
|
| 40306 |
+
"learning_rate": 2.550450272841842e-07,
|
| 40307 |
+
"loss": 0.0,
|
| 40308 |
+
"step": 5757
|
| 40309 |
+
},
|
| 40310 |
+
{
|
| 40311 |
+
"epoch": 0.9684635438566983,
|
| 40312 |
+
"grad_norm": NaN,
|
| 40313 |
+
"learning_rate": 2.523412768458688e-07,
|
| 40314 |
+
"loss": 0.0,
|
| 40315 |
+
"step": 5758
|
| 40316 |
+
},
|
| 40317 |
+
{
|
| 40318 |
+
"epoch": 0.9686317382894626,
|
| 40319 |
+
"grad_norm": NaN,
|
| 40320 |
+
"learning_rate": 2.4965189794895485e-07,
|
| 40321 |
+
"loss": 0.0,
|
| 40322 |
+
"step": 5759
|
| 40323 |
+
},
|
| 40324 |
+
{
|
| 40325 |
+
"epoch": 0.9687999327222269,
|
| 40326 |
+
"grad_norm": NaN,
|
| 40327 |
+
"learning_rate": 2.469768913703707e-07,
|
| 40328 |
+
"loss": 0.0,
|
| 40329 |
+
"step": 5760
|
| 40330 |
+
},
|
| 40331 |
+
{
|
| 40332 |
+
"epoch": 0.9689681271549911,
|
| 40333 |
+
"grad_norm": NaN,
|
| 40334 |
+
"learning_rate": 2.443162578828928e-07,
|
| 40335 |
+
"loss": 0.0,
|
| 40336 |
+
"step": 5761
|
| 40337 |
+
},
|
| 40338 |
+
{
|
| 40339 |
+
"epoch": 0.9691363215877554,
|
| 40340 |
+
"grad_norm": NaN,
|
| 40341 |
+
"learning_rate": 2.4166999825515625e-07,
|
| 40342 |
+
"loss": 0.0,
|
| 40343 |
+
"step": 5762
|
| 40344 |
+
},
|
| 40345 |
+
{
|
| 40346 |
+
"epoch": 0.9693045160205197,
|
| 40347 |
+
"grad_norm": NaN,
|
| 40348 |
+
"learning_rate": 2.3903811325163283e-07,
|
| 40349 |
+
"loss": 0.0,
|
| 40350 |
+
"step": 5763
|
| 40351 |
+
},
|
| 40352 |
+
{
|
| 40353 |
+
"epoch": 0.969472710453284,
|
| 40354 |
+
"grad_norm": NaN,
|
| 40355 |
+
"learning_rate": 2.3642060363264217e-07,
|
| 40356 |
+
"loss": 0.0,
|
| 40357 |
+
"step": 5764
|
| 40358 |
+
},
|
| 40359 |
+
{
|
| 40360 |
+
"epoch": 0.9696409048860483,
|
| 40361 |
+
"grad_norm": NaN,
|
| 40362 |
+
"learning_rate": 2.3381747015435163e-07,
|
| 40363 |
+
"loss": 0.0,
|
| 40364 |
+
"step": 5765
|
| 40365 |
+
},
|
| 40366 |
+
{
|
| 40367 |
+
"epoch": 0.9698090993188125,
|
| 40368 |
+
"grad_norm": NaN,
|
| 40369 |
+
"learning_rate": 2.312287135687874e-07,
|
| 40370 |
+
"loss": 0.0,
|
| 40371 |
+
"step": 5766
|
| 40372 |
+
},
|
| 40373 |
+
{
|
| 40374 |
+
"epoch": 0.9699772937515768,
|
| 40375 |
+
"grad_norm": NaN,
|
| 40376 |
+
"learning_rate": 2.2865433462380125e-07,
|
| 40377 |
+
"loss": 0.0,
|
| 40378 |
+
"step": 5767
|
| 40379 |
+
},
|
| 40380 |
+
{
|
| 40381 |
+
"epoch": 0.9701454881843411,
|
| 40382 |
+
"grad_norm": NaN,
|
| 40383 |
+
"learning_rate": 2.2609433406310941e-07,
|
| 40384 |
+
"loss": 0.0,
|
| 40385 |
+
"step": 5768
|
| 40386 |
+
},
|
| 40387 |
+
{
|
| 40388 |
+
"epoch": 0.9703136826171054,
|
| 40389 |
+
"grad_norm": NaN,
|
| 40390 |
+
"learning_rate": 2.2354871262626477e-07,
|
| 40391 |
+
"loss": 0.0,
|
| 40392 |
+
"step": 5769
|
| 40393 |
+
},
|
| 40394 |
+
{
|
| 40395 |
+
"epoch": 0.9704818770498697,
|
| 40396 |
+
"grad_norm": NaN,
|
| 40397 |
+
"learning_rate": 2.2101747104866788e-07,
|
| 40398 |
+
"loss": 0.0,
|
| 40399 |
+
"step": 5770
|
| 40400 |
+
},
|
| 40401 |
+
{
|
| 40402 |
+
"epoch": 0.9706500714826339,
|
| 40403 |
+
"grad_norm": NaN,
|
| 40404 |
+
"learning_rate": 2.185006100615672e-07,
|
| 40405 |
+
"loss": 0.0,
|
| 40406 |
+
"step": 5771
|
| 40407 |
+
},
|
| 40408 |
+
{
|
| 40409 |
+
"epoch": 0.9708182659153982,
|
| 40410 |
+
"grad_norm": NaN,
|
| 40411 |
+
"learning_rate": 2.159981303920533e-07,
|
| 40412 |
+
"loss": 0.0,
|
| 40413 |
+
"step": 5772
|
| 40414 |
+
},
|
| 40415 |
+
{
|
| 40416 |
+
"epoch": 0.9709864603481625,
|
| 40417 |
+
"grad_norm": NaN,
|
| 40418 |
+
"learning_rate": 2.1351003276307014e-07,
|
| 40419 |
+
"loss": 0.0,
|
| 40420 |
+
"step": 5773
|
| 40421 |
+
},
|
| 40422 |
+
{
|
| 40423 |
+
"epoch": 0.9711546547809268,
|
| 40424 |
+
"grad_norm": NaN,
|
| 40425 |
+
"learning_rate": 2.1103631789339272e-07,
|
| 40426 |
+
"loss": 0.0,
|
| 40427 |
+
"step": 5774
|
| 40428 |
+
},
|
| 40429 |
+
{
|
| 40430 |
+
"epoch": 0.971322849213691,
|
| 40431 |
+
"grad_norm": NaN,
|
| 40432 |
+
"learning_rate": 2.0857698649766055e-07,
|
| 40433 |
+
"loss": 0.0,
|
| 40434 |
+
"step": 5775
|
| 40435 |
+
},
|
| 40436 |
+
{
|
| 40437 |
+
"epoch": 0.9714910436464553,
|
| 40438 |
+
"grad_norm": NaN,
|
| 40439 |
+
"learning_rate": 2.061320392863386e-07,
|
| 40440 |
+
"loss": 0.0,
|
| 40441 |
+
"step": 5776
|
| 40442 |
+
},
|
| 40443 |
+
{
|
| 40444 |
+
"epoch": 0.9716592380792196,
|
| 40445 |
+
"grad_norm": NaN,
|
| 40446 |
+
"learning_rate": 2.0370147696574526e-07,
|
| 40447 |
+
"loss": 0.0,
|
| 40448 |
+
"step": 5777
|
| 40449 |
+
},
|
| 40450 |
+
{
|
| 40451 |
+
"epoch": 0.9718274325119839,
|
| 40452 |
+
"grad_norm": NaN,
|
| 40453 |
+
"learning_rate": 2.012853002380466e-07,
|
| 40454 |
+
"loss": 0.0,
|
| 40455 |
+
"step": 5778
|
| 40456 |
+
},
|
| 40457 |
+
{
|
| 40458 |
+
"epoch": 0.9719956269447482,
|
| 40459 |
+
"grad_norm": NaN,
|
| 40460 |
+
"learning_rate": 1.988835098012509e-07,
|
| 40461 |
+
"loss": 0.0,
|
| 40462 |
+
"step": 5779
|
| 40463 |
+
},
|
| 40464 |
+
{
|
| 40465 |
+
"epoch": 0.9721638213775124,
|
| 40466 |
+
"grad_norm": NaN,
|
| 40467 |
+
"learning_rate": 1.9649610634919767e-07,
|
| 40468 |
+
"loss": 0.0,
|
| 40469 |
+
"step": 5780
|
| 40470 |
+
},
|
| 40471 |
+
{
|
| 40472 |
+
"epoch": 0.9723320158102767,
|
| 40473 |
+
"grad_norm": NaN,
|
| 40474 |
+
"learning_rate": 1.9412309057159073e-07,
|
| 40475 |
+
"loss": 0.0,
|
| 40476 |
+
"step": 5781
|
| 40477 |
+
},
|
| 40478 |
+
{
|
| 40479 |
+
"epoch": 0.972500210243041,
|
| 40480 |
+
"grad_norm": NaN,
|
| 40481 |
+
"learning_rate": 1.9176446315397056e-07,
|
| 40482 |
+
"loss": 0.0,
|
| 40483 |
+
"step": 5782
|
| 40484 |
+
},
|
| 40485 |
+
{
|
| 40486 |
+
"epoch": 0.9726684046758052,
|
| 40487 |
+
"grad_norm": NaN,
|
| 40488 |
+
"learning_rate": 1.894202247777088e-07,
|
| 40489 |
+
"loss": 0.0,
|
| 40490 |
+
"step": 5783
|
| 40491 |
+
},
|
| 40492 |
+
{
|
| 40493 |
+
"epoch": 0.9728365991085695,
|
| 40494 |
+
"grad_norm": NaN,
|
| 40495 |
+
"learning_rate": 1.8709037612003045e-07,
|
| 40496 |
+
"loss": 0.0,
|
| 40497 |
+
"step": 5784
|
| 40498 |
+
},
|
| 40499 |
+
{
|
| 40500 |
+
"epoch": 0.9730047935413337,
|
| 40501 |
+
"grad_norm": NaN,
|
| 40502 |
+
"learning_rate": 1.8477491785400813e-07,
|
| 40503 |
+
"loss": 0.0,
|
| 40504 |
+
"step": 5785
|
| 40505 |
+
},
|
| 40506 |
+
{
|
| 40507 |
+
"epoch": 0.973172987974098,
|
| 40508 |
+
"grad_norm": NaN,
|
| 40509 |
+
"learning_rate": 1.8247385064855127e-07,
|
| 40510 |
+
"loss": 0.0,
|
| 40511 |
+
"step": 5786
|
| 40512 |
+
},
|
| 40513 |
+
{
|
| 40514 |
+
"epoch": 0.9733411824068623,
|
| 40515 |
+
"grad_norm": NaN,
|
| 40516 |
+
"learning_rate": 1.8018717516841143e-07,
|
| 40517 |
+
"loss": 0.0,
|
| 40518 |
+
"step": 5787
|
| 40519 |
+
},
|
| 40520 |
+
{
|
| 40521 |
+
"epoch": 0.9735093768396266,
|
| 40522 |
+
"grad_norm": NaN,
|
| 40523 |
+
"learning_rate": 1.779148920741769e-07,
|
| 40524 |
+
"loss": 0.0,
|
| 40525 |
+
"step": 5788
|
| 40526 |
+
},
|
| 40527 |
+
{
|
| 40528 |
+
"epoch": 0.9736775712723909,
|
| 40529 |
+
"grad_norm": NaN,
|
| 40530 |
+
"learning_rate": 1.7565700202229473e-07,
|
| 40531 |
+
"loss": 0.0,
|
| 40532 |
+
"step": 5789
|
| 40533 |
+
},
|
| 40534 |
+
{
|
| 40535 |
+
"epoch": 0.9738457657051551,
|
| 40536 |
+
"grad_norm": NaN,
|
| 40537 |
+
"learning_rate": 1.7341350566504323e-07,
|
| 40538 |
+
"loss": 0.0,
|
| 40539 |
+
"step": 5790
|
| 40540 |
+
},
|
| 40541 |
+
{
|
| 40542 |
+
"epoch": 0.9740139601379194,
|
| 40543 |
+
"grad_norm": NaN,
|
| 40544 |
+
"learning_rate": 1.7118440365053722e-07,
|
| 40545 |
+
"loss": 0.0,
|
| 40546 |
+
"step": 5791
|
| 40547 |
+
},
|
| 40548 |
+
{
|
| 40549 |
+
"epoch": 0.9741821545706837,
|
| 40550 |
+
"grad_norm": NaN,
|
| 40551 |
+
"learning_rate": 1.6896969662273944e-07,
|
| 40552 |
+
"loss": 0.0,
|
| 40553 |
+
"step": 5792
|
| 40554 |
+
},
|
| 40555 |
+
{
|
| 40556 |
+
"epoch": 0.974350349003448,
|
| 40557 |
+
"grad_norm": NaN,
|
| 40558 |
+
"learning_rate": 1.6676938522146023e-07,
|
| 40559 |
+
"loss": 0.0,
|
| 40560 |
+
"step": 5793
|
| 40561 |
+
},
|
| 40562 |
+
{
|
| 40563 |
+
"epoch": 0.9745185434362122,
|
| 40564 |
+
"grad_norm": NaN,
|
| 40565 |
+
"learning_rate": 1.645834700823412e-07,
|
| 40566 |
+
"loss": 0.0,
|
| 40567 |
+
"step": 5794
|
| 40568 |
+
},
|
| 40569 |
+
{
|
| 40570 |
+
"epoch": 0.9746867378689765,
|
| 40571 |
+
"grad_norm": NaN,
|
| 40572 |
+
"learning_rate": 1.6241195183686608e-07,
|
| 40573 |
+
"loss": 0.0,
|
| 40574 |
+
"step": 5795
|
| 40575 |
+
},
|
| 40576 |
+
{
|
| 40577 |
+
"epoch": 0.9748549323017408,
|
| 40578 |
+
"grad_norm": NaN,
|
| 40579 |
+
"learning_rate": 1.6025483111236638e-07,
|
| 40580 |
+
"loss": 0.0,
|
| 40581 |
+
"step": 5796
|
| 40582 |
+
},
|
| 40583 |
+
{
|
| 40584 |
+
"epoch": 0.9750231267345051,
|
| 40585 |
+
"grad_norm": NaN,
|
| 40586 |
+
"learning_rate": 1.581121085320103e-07,
|
| 40587 |
+
"loss": 0.0,
|
| 40588 |
+
"step": 5797
|
| 40589 |
+
},
|
| 40590 |
+
{
|
| 40591 |
+
"epoch": 0.9751913211672694,
|
| 40592 |
+
"grad_norm": NaN,
|
| 40593 |
+
"learning_rate": 1.5598378471480267e-07,
|
| 40594 |
+
"loss": 0.0,
|
| 40595 |
+
"step": 5798
|
| 40596 |
+
},
|
| 40597 |
+
{
|
| 40598 |
+
"epoch": 0.9753595156000336,
|
| 40599 |
+
"grad_norm": NaN,
|
| 40600 |
+
"learning_rate": 1.5386986027559613e-07,
|
| 40601 |
+
"loss": 0.0,
|
| 40602 |
+
"step": 5799
|
| 40603 |
+
},
|
| 40604 |
+
{
|
| 40605 |
+
"epoch": 0.9755277100327979,
|
| 40606 |
+
"grad_norm": NaN,
|
| 40607 |
+
"learning_rate": 1.5177033582507993e-07,
|
| 40608 |
+
"loss": 0.0,
|
| 40609 |
+
"step": 5800
|
| 40610 |
+
},
|
| 40611 |
+
{
|
| 40612 |
+
"epoch": 0.9756959044655622,
|
| 40613 |
+
"grad_norm": NaN,
|
| 40614 |
+
"learning_rate": 1.4968521196978002e-07,
|
| 40615 |
+
"loss": 0.0,
|
| 40616 |
+
"step": 5801
|
| 40617 |
+
},
|
| 40618 |
+
{
|
| 40619 |
+
"epoch": 0.9758640988983265,
|
| 40620 |
+
"grad_norm": NaN,
|
| 40621 |
+
"learning_rate": 1.4761448931206455e-07,
|
| 40622 |
+
"loss": 0.0,
|
| 40623 |
+
"step": 5802
|
| 40624 |
+
},
|
| 40625 |
+
{
|
| 40626 |
+
"epoch": 0.9760322933310908,
|
| 40627 |
+
"grad_norm": NaN,
|
| 40628 |
+
"learning_rate": 1.4555816845014948e-07,
|
| 40629 |
+
"loss": 0.0,
|
| 40630 |
+
"step": 5803
|
| 40631 |
+
},
|
| 40632 |
+
{
|
| 40633 |
+
"epoch": 0.976200487763855,
|
| 40634 |
+
"grad_norm": NaN,
|
| 40635 |
+
"learning_rate": 1.4351624997807623e-07,
|
| 40636 |
+
"loss": 0.0,
|
| 40637 |
+
"step": 5804
|
| 40638 |
+
},
|
| 40639 |
+
{
|
| 40640 |
+
"epoch": 0.9763686821966193,
|
| 40641 |
+
"grad_norm": NaN,
|
| 40642 |
+
"learning_rate": 1.4148873448573408e-07,
|
| 40643 |
+
"loss": 0.0,
|
| 40644 |
+
"step": 5805
|
| 40645 |
+
},
|
| 40646 |
+
{
|
| 40647 |
+
"epoch": 0.9765368766293836,
|
| 40648 |
+
"grad_norm": NaN,
|
| 40649 |
+
"learning_rate": 1.3947562255884338e-07,
|
| 40650 |
+
"loss": 0.0,
|
| 40651 |
+
"step": 5806
|
| 40652 |
+
},
|
| 40653 |
+
{
|
| 40654 |
+
"epoch": 0.9767050710621479,
|
| 40655 |
+
"grad_norm": NaN,
|
| 40656 |
+
"learning_rate": 1.374769147789834e-07,
|
| 40657 |
+
"loss": 0.0,
|
| 40658 |
+
"step": 5807
|
| 40659 |
+
},
|
| 40660 |
+
{
|
| 40661 |
+
"epoch": 0.9768732654949122,
|
| 40662 |
+
"grad_norm": NaN,
|
| 40663 |
+
"learning_rate": 1.3549261172354777e-07,
|
| 40664 |
+
"loss": 0.0,
|
| 40665 |
+
"step": 5808
|
| 40666 |
+
},
|
| 40667 |
+
{
|
| 40668 |
+
"epoch": 0.9770414599276764,
|
| 40669 |
+
"grad_norm": NaN,
|
| 40670 |
+
"learning_rate": 1.3352271396577798e-07,
|
| 40671 |
+
"loss": 0.0,
|
| 40672 |
+
"step": 5809
|
| 40673 |
+
},
|
| 40674 |
+
{
|
| 40675 |
+
"epoch": 0.9772096543604407,
|
| 40676 |
+
"grad_norm": NaN,
|
| 40677 |
+
"learning_rate": 1.3156722207476324e-07,
|
| 40678 |
+
"loss": 0.0,
|
| 40679 |
+
"step": 5810
|
| 40680 |
+
},
|
| 40681 |
+
{
|
| 40682 |
+
"epoch": 0.977377848793205,
|
| 40683 |
+
"grad_norm": NaN,
|
| 40684 |
+
"learning_rate": 1.2962613661541834e-07,
|
| 40685 |
+
"loss": 0.0,
|
| 40686 |
+
"step": 5811
|
| 40687 |
+
},
|
| 40688 |
+
{
|
| 40689 |
+
"epoch": 0.9775460432259693,
|
| 40690 |
+
"grad_norm": NaN,
|
| 40691 |
+
"learning_rate": 1.2769945814850582e-07,
|
| 40692 |
+
"loss": 0.0,
|
| 40693 |
+
"step": 5812
|
| 40694 |
+
},
|
| 40695 |
+
{
|
| 40696 |
+
"epoch": 0.9777142376587334,
|
| 40697 |
+
"grad_norm": NaN,
|
| 40698 |
+
"learning_rate": 1.2578718723061378e-07,
|
| 40699 |
+
"loss": 0.0,
|
| 40700 |
+
"step": 5813
|
| 40701 |
+
},
|
| 40702 |
+
{
|
| 40703 |
+
"epoch": 0.9778824320914977,
|
| 40704 |
+
"grad_norm": NaN,
|
| 40705 |
+
"learning_rate": 1.2388932441418367e-07,
|
| 40706 |
+
"loss": 0.0,
|
| 40707 |
+
"step": 5814
|
| 40708 |
+
},
|
| 40709 |
+
{
|
| 40710 |
+
"epoch": 0.978050626524262,
|
| 40711 |
+
"grad_norm": NaN,
|
| 40712 |
+
"learning_rate": 1.220058702474769e-07,
|
| 40713 |
+
"loss": 0.0,
|
| 40714 |
+
"step": 5815
|
| 40715 |
+
},
|
| 40716 |
+
{
|
| 40717 |
+
"epoch": 0.9782188209570263,
|
| 40718 |
+
"grad_norm": NaN,
|
| 40719 |
+
"learning_rate": 1.2013682527461379e-07,
|
| 40720 |
+
"loss": 0.0,
|
| 40721 |
+
"step": 5816
|
| 40722 |
+
},
|
| 40723 |
+
{
|
| 40724 |
+
"epoch": 0.9783870153897906,
|
| 40725 |
+
"grad_norm": NaN,
|
| 40726 |
+
"learning_rate": 1.1828219003553465e-07,
|
| 40727 |
+
"loss": 0.0,
|
| 40728 |
+
"step": 5817
|
| 40729 |
+
},
|
| 40730 |
+
{
|
| 40731 |
+
"epoch": 0.9785552098225548,
|
| 40732 |
+
"grad_norm": NaN,
|
| 40733 |
+
"learning_rate": 1.16441965066022e-07,
|
| 40734 |
+
"loss": 0.0,
|
| 40735 |
+
"step": 5818
|
| 40736 |
+
},
|
| 40737 |
+
{
|
| 40738 |
+
"epoch": 0.9787234042553191,
|
| 40739 |
+
"grad_norm": NaN,
|
| 40740 |
+
"learning_rate": 1.1461615089770062e-07,
|
| 40741 |
+
"loss": 0.0,
|
| 40742 |
+
"step": 5819
|
| 40743 |
+
},
|
| 40744 |
+
{
|
| 40745 |
+
"epoch": 0.9788915986880834,
|
| 40746 |
+
"grad_norm": NaN,
|
| 40747 |
+
"learning_rate": 1.1280474805802632e-07,
|
| 40748 |
+
"loss": 0.0,
|
| 40749 |
+
"step": 5820
|
| 40750 |
+
},
|
| 40751 |
+
{
|
| 40752 |
+
"epoch": 0.9790597931208477,
|
| 40753 |
+
"grad_norm": NaN,
|
| 40754 |
+
"learning_rate": 1.110077570702861e-07,
|
| 40755 |
+
"loss": 0.0,
|
| 40756 |
+
"step": 5821
|
| 40757 |
+
},
|
| 40758 |
+
{
|
| 40759 |
+
"epoch": 0.979227987553612,
|
| 40760 |
+
"grad_norm": NaN,
|
| 40761 |
+
"learning_rate": 1.0922517845362023e-07,
|
| 40762 |
+
"loss": 0.0,
|
| 40763 |
+
"step": 5822
|
| 40764 |
+
},
|
| 40765 |
+
{
|
| 40766 |
+
"epoch": 0.9793961819863762,
|
| 40767 |
+
"grad_norm": NaN,
|
| 40768 |
+
"learning_rate": 1.0745701272298902e-07,
|
| 40769 |
+
"loss": 0.0,
|
| 40770 |
+
"step": 5823
|
| 40771 |
+
},
|
| 40772 |
+
{
|
| 40773 |
+
"epoch": 0.9795643764191405,
|
| 40774 |
+
"grad_norm": NaN,
|
| 40775 |
+
"learning_rate": 1.0570326038920053e-07,
|
| 40776 |
+
"loss": 0.0,
|
| 40777 |
+
"step": 5824
|
| 40778 |
+
},
|
| 40779 |
+
{
|
| 40780 |
+
"epoch": 0.9797325708519048,
|
| 40781 |
+
"grad_norm": NaN,
|
| 40782 |
+
"learning_rate": 1.0396392195889393e-07,
|
| 40783 |
+
"loss": 0.0,
|
| 40784 |
+
"step": 5825
|
| 40785 |
+
},
|
| 40786 |
+
{
|
| 40787 |
+
"epoch": 0.9799007652846691,
|
| 40788 |
+
"grad_norm": NaN,
|
| 40789 |
+
"learning_rate": 1.022389979345395e-07,
|
| 40790 |
+
"loss": 0.0,
|
| 40791 |
+
"step": 5826
|
| 40792 |
+
},
|
| 40793 |
+
{
|
| 40794 |
+
"epoch": 0.9800689597174334,
|
| 40795 |
+
"grad_norm": NaN,
|
| 40796 |
+
"learning_rate": 1.0052848881444976e-07,
|
| 40797 |
+
"loss": 0.0,
|
| 40798 |
+
"step": 5827
|
| 40799 |
+
},
|
| 40800 |
+
{
|
| 40801 |
+
"epoch": 0.9802371541501976,
|
| 40802 |
+
"grad_norm": NaN,
|
| 40803 |
+
"learning_rate": 9.883239509277942e-08,
|
| 40804 |
+
"loss": 0.0,
|
| 40805 |
+
"step": 5828
|
| 40806 |
+
},
|
| 40807 |
+
{
|
| 40808 |
+
"epoch": 0.9804053485829619,
|
| 40809 |
+
"grad_norm": NaN,
|
| 40810 |
+
"learning_rate": 9.715071725949765e-08,
|
| 40811 |
+
"loss": 0.0,
|
| 40812 |
+
"step": 5829
|
| 40813 |
+
},
|
| 40814 |
+
{
|
| 40815 |
+
"epoch": 0.9805735430157262,
|
| 40816 |
+
"grad_norm": NaN,
|
| 40817 |
+
"learning_rate": 9.54834558004325e-08,
|
| 40818 |
+
"loss": 0.0,
|
| 40819 |
+
"step": 5830
|
| 40820 |
+
},
|
| 40821 |
+
{
|
| 40822 |
+
"epoch": 0.9807417374484905,
|
| 40823 |
+
"grad_norm": NaN,
|
| 40824 |
+
"learning_rate": 9.383061119723757e-08,
|
| 40825 |
+
"loss": 0.0,
|
| 40826 |
+
"step": 5831
|
| 40827 |
+
},
|
| 40828 |
+
{
|
| 40829 |
+
"epoch": 0.9809099318812547,
|
| 40830 |
+
"grad_norm": NaN,
|
| 40831 |
+
"learning_rate": 9.219218392739759e-08,
|
| 40832 |
+
"loss": 0.0,
|
| 40833 |
+
"step": 5832
|
| 40834 |
+
},
|
| 40835 |
+
{
|
| 40836 |
+
"epoch": 0.981078126314019,
|
| 40837 |
+
"grad_norm": NaN,
|
| 40838 |
+
"learning_rate": 9.056817446422839e-08,
|
| 40839 |
+
"loss": 0.0,
|
| 40840 |
+
"step": 5833
|
| 40841 |
+
},
|
| 40842 |
+
{
|
| 40843 |
+
"epoch": 0.9812463207467833,
|
| 40844 |
+
"grad_norm": NaN,
|
| 40845 |
+
"learning_rate": 8.895858327690464e-08,
|
| 40846 |
+
"loss": 0.0,
|
| 40847 |
+
"step": 5834
|
| 40848 |
+
},
|
| 40849 |
+
{
|
| 40850 |
+
"epoch": 0.9814145151795476,
|
| 40851 |
+
"grad_norm": NaN,
|
| 40852 |
+
"learning_rate": 8.736341083041e-08,
|
| 40853 |
+
"loss": 0.0,
|
| 40854 |
+
"step": 5835
|
| 40855 |
+
},
|
| 40856 |
+
{
|
| 40857 |
+
"epoch": 0.9815827096123119,
|
| 40858 |
+
"grad_norm": NaN,
|
| 40859 |
+
"learning_rate": 8.578265758557024e-08,
|
| 40860 |
+
"loss": 0.0,
|
| 40861 |
+
"step": 5836
|
| 40862 |
+
},
|
| 40863 |
+
{
|
| 40864 |
+
"epoch": 0.9817509040450761,
|
| 40865 |
+
"grad_norm": NaN,
|
| 40866 |
+
"learning_rate": 8.421632399904788e-08,
|
| 40867 |
+
"loss": 0.0,
|
| 40868 |
+
"step": 5837
|
| 40869 |
+
},
|
| 40870 |
+
{
|
| 40871 |
+
"epoch": 0.9819190984778404,
|
| 40872 |
+
"grad_norm": NaN,
|
| 40873 |
+
"learning_rate": 8.266441052334206e-08,
|
| 40874 |
+
"loss": 0.0,
|
| 40875 |
+
"step": 5838
|
| 40876 |
+
},
|
| 40877 |
+
{
|
| 40878 |
+
"epoch": 0.9820872929106047,
|
| 40879 |
+
"grad_norm": NaN,
|
| 40880 |
+
"learning_rate": 8.112691760677749e-08,
|
| 40881 |
+
"loss": 0.0,
|
| 40882 |
+
"step": 5839
|
| 40883 |
+
},
|
| 40884 |
+
{
|
| 40885 |
+
"epoch": 0.982255487343369,
|
| 40886 |
+
"grad_norm": NaN,
|
| 40887 |
+
"learning_rate": 7.960384569353219e-08,
|
| 40888 |
+
"loss": 0.0,
|
| 40889 |
+
"step": 5840
|
| 40890 |
+
},
|
| 40891 |
+
{
|
| 40892 |
+
"epoch": 0.9824236817761333,
|
| 40893 |
+
"grad_norm": NaN,
|
| 40894 |
+
"learning_rate": 7.809519522358755e-08,
|
| 40895 |
+
"loss": 0.0,
|
| 40896 |
+
"step": 5841
|
| 40897 |
+
},
|
| 40898 |
+
{
|
| 40899 |
+
"epoch": 0.9825918762088974,
|
| 40900 |
+
"grad_norm": NaN,
|
| 40901 |
+
"learning_rate": 7.660096663278938e-08,
|
| 40902 |
+
"loss": 0.0,
|
| 40903 |
+
"step": 5842
|
| 40904 |
+
},
|
| 40905 |
+
{
|
| 40906 |
+
"epoch": 0.9827600706416617,
|
| 40907 |
+
"grad_norm": NaN,
|
| 40908 |
+
"learning_rate": 7.512116035279237e-08,
|
| 40909 |
+
"loss": 0.0,
|
| 40910 |
+
"step": 5843
|
| 40911 |
+
},
|
| 40912 |
+
{
|
| 40913 |
+
"epoch": 0.982928265074426,
|
| 40914 |
+
"grad_norm": NaN,
|
| 40915 |
+
"learning_rate": 7.365577681110458e-08,
|
| 40916 |
+
"loss": 0.0,
|
| 40917 |
+
"step": 5844
|
| 40918 |
+
},
|
| 40919 |
+
{
|
| 40920 |
+
"epoch": 0.9830964595071903,
|
| 40921 |
+
"grad_norm": NaN,
|
| 40922 |
+
"learning_rate": 7.220481643105403e-08,
|
| 40923 |
+
"loss": 0.0,
|
| 40924 |
+
"step": 5845
|
| 40925 |
+
},
|
| 40926 |
+
{
|
| 40927 |
+
"epoch": 0.9832646539399545,
|
| 40928 |
+
"grad_norm": NaN,
|
| 40929 |
+
"learning_rate": 7.076827963181099e-08,
|
| 40930 |
+
"loss": 0.0,
|
| 40931 |
+
"step": 5846
|
| 40932 |
+
},
|
| 40933 |
+
{
|
| 40934 |
+
"epoch": 0.9834328483727188,
|
| 40935 |
+
"grad_norm": NaN,
|
| 40936 |
+
"learning_rate": 6.934616682837125e-08,
|
| 40937 |
+
"loss": 0.0,
|
| 40938 |
+
"step": 5847
|
| 40939 |
+
},
|
| 40940 |
+
{
|
| 40941 |
+
"epoch": 0.9836010428054831,
|
| 40942 |
+
"grad_norm": NaN,
|
| 40943 |
+
"learning_rate": 6.79384784315673e-08,
|
| 40944 |
+
"loss": 0.0,
|
| 40945 |
+
"step": 5848
|
| 40946 |
+
},
|
| 40947 |
+
{
|
| 40948 |
+
"epoch": 0.9837692372382474,
|
| 40949 |
+
"grad_norm": NaN,
|
| 40950 |
+
"learning_rate": 6.65452148480683e-08,
|
| 40951 |
+
"loss": 0.0,
|
| 40952 |
+
"step": 5849
|
| 40953 |
+
},
|
| 40954 |
+
{
|
| 40955 |
+
"epoch": 0.9839374316710117,
|
| 40956 |
+
"grad_norm": NaN,
|
| 40957 |
+
"learning_rate": 6.516637648036894e-08,
|
| 40958 |
+
"loss": 0.0,
|
| 40959 |
+
"step": 5850
|
| 40960 |
+
},
|
| 40961 |
+
{
|
| 40962 |
+
"epoch": 0.9841056261037759,
|
| 40963 |
+
"grad_norm": NaN,
|
| 40964 |
+
"learning_rate": 6.380196372680058e-08,
|
| 40965 |
+
"loss": 0.0,
|
| 40966 |
+
"step": 5851
|
| 40967 |
+
},
|
| 40968 |
+
{
|
| 40969 |
+
"epoch": 0.9842738205365402,
|
| 40970 |
+
"grad_norm": NaN,
|
| 40971 |
+
"learning_rate": 6.245197698152571e-08,
|
| 40972 |
+
"loss": 0.0,
|
| 40973 |
+
"step": 5852
|
| 40974 |
+
},
|
| 40975 |
+
{
|
| 40976 |
+
"epoch": 0.9844420149693045,
|
| 40977 |
+
"grad_norm": NaN,
|
| 40978 |
+
"learning_rate": 6.111641663454903e-08,
|
| 40979 |
+
"loss": 0.0,
|
| 40980 |
+
"step": 5853
|
| 40981 |
+
},
|
| 40982 |
+
{
|
| 40983 |
+
"epoch": 0.9846102094020688,
|
| 40984 |
+
"grad_norm": NaN,
|
| 40985 |
+
"learning_rate": 5.979528307168414e-08,
|
| 40986 |
+
"loss": 0.0,
|
| 40987 |
+
"step": 5854
|
| 40988 |
+
},
|
| 40989 |
+
{
|
| 40990 |
+
"epoch": 0.9847784038348331,
|
| 40991 |
+
"grad_norm": NaN,
|
| 40992 |
+
"learning_rate": 5.84885766746035e-08,
|
| 40993 |
+
"loss": 0.0,
|
| 40994 |
+
"step": 5855
|
| 40995 |
+
},
|
| 40996 |
+
{
|
| 40997 |
+
"epoch": 0.9849465982675973,
|
| 40998 |
+
"grad_norm": NaN,
|
| 40999 |
+
"learning_rate": 5.7196297820794054e-08,
|
| 41000 |
+
"loss": 0.0,
|
| 41001 |
+
"step": 5856
|
| 41002 |
+
},
|
| 41003 |
+
{
|
| 41004 |
+
"epoch": 0.9851147927003616,
|
| 41005 |
+
"grad_norm": NaN,
|
| 41006 |
+
"learning_rate": 5.591844688358494e-08,
|
| 41007 |
+
"loss": 0.0,
|
| 41008 |
+
"step": 5857
|
| 41009 |
+
},
|
| 41010 |
+
{
|
| 41011 |
+
"epoch": 0.9852829871331259,
|
| 41012 |
+
"grad_norm": NaN,
|
| 41013 |
+
"learning_rate": 5.465502423213087e-08,
|
| 41014 |
+
"loss": 0.0,
|
| 41015 |
+
"step": 5858
|
| 41016 |
+
},
|
| 41017 |
+
{
|
| 41018 |
+
"epoch": 0.9854511815658902,
|
| 41019 |
+
"grad_norm": NaN,
|
| 41020 |
+
"learning_rate": 5.340603023141766e-08,
|
| 41021 |
+
"loss": 0.0,
|
| 41022 |
+
"step": 5859
|
| 41023 |
+
},
|
| 41024 |
+
{
|
| 41025 |
+
"epoch": 0.9856193759986545,
|
| 41026 |
+
"grad_norm": NaN,
|
| 41027 |
+
"learning_rate": 5.217146524226779e-08,
|
| 41028 |
+
"loss": 0.0,
|
| 41029 |
+
"step": 5860
|
| 41030 |
+
},
|
| 41031 |
+
{
|
| 41032 |
+
"epoch": 0.9857875704314187,
|
| 41033 |
+
"grad_norm": NaN,
|
| 41034 |
+
"learning_rate": 5.0951329621340416e-08,
|
| 41035 |
+
"loss": 0.0,
|
| 41036 |
+
"step": 5861
|
| 41037 |
+
},
|
| 41038 |
+
{
|
| 41039 |
+
"epoch": 0.985955764864183,
|
| 41040 |
+
"grad_norm": NaN,
|
| 41041 |
+
"learning_rate": 4.9745623721109135e-08,
|
| 41042 |
+
"loss": 0.0,
|
| 41043 |
+
"step": 5862
|
| 41044 |
+
},
|
| 41045 |
+
{
|
| 41046 |
+
"epoch": 0.9861239592969473,
|
| 41047 |
+
"grad_norm": NaN,
|
| 41048 |
+
"learning_rate": 4.855434788988977e-08,
|
| 41049 |
+
"loss": 0.0,
|
| 41050 |
+
"step": 5863
|
| 41051 |
+
},
|
| 41052 |
+
{
|
| 41053 |
+
"epoch": 0.9862921537297116,
|
| 41054 |
+
"grad_norm": NaN,
|
| 41055 |
+
"learning_rate": 4.737750247183481e-08,
|
| 41056 |
+
"loss": 0.0,
|
| 41057 |
+
"step": 5864
|
| 41058 |
+
},
|
| 41059 |
+
{
|
| 41060 |
+
"epoch": 0.9864603481624759,
|
| 41061 |
+
"grad_norm": NaN,
|
| 41062 |
+
"learning_rate": 4.621508780691119e-08,
|
| 41063 |
+
"loss": 0.0,
|
| 41064 |
+
"step": 5865
|
| 41065 |
+
},
|
| 41066 |
+
{
|
| 41067 |
+
"epoch": 0.9866285425952401,
|
| 41068 |
+
"grad_norm": NaN,
|
| 41069 |
+
"learning_rate": 4.506710423093918e-08,
|
| 41070 |
+
"loss": 0.0,
|
| 41071 |
+
"step": 5866
|
| 41072 |
+
},
|
| 41073 |
+
{
|
| 41074 |
+
"epoch": 0.9867967370280044,
|
| 41075 |
+
"grad_norm": NaN,
|
| 41076 |
+
"learning_rate": 4.39335520755535e-08,
|
| 41077 |
+
"loss": 0.0,
|
| 41078 |
+
"step": 5867
|
| 41079 |
+
},
|
| 41080 |
+
{
|
| 41081 |
+
"epoch": 0.9869649314607687,
|
| 41082 |
+
"grad_norm": NaN,
|
| 41083 |
+
"learning_rate": 4.281443166822552e-08,
|
| 41084 |
+
"loss": 0.0,
|
| 41085 |
+
"step": 5868
|
| 41086 |
+
},
|
| 41087 |
+
{
|
| 41088 |
+
"epoch": 0.987133125893533,
|
| 41089 |
+
"grad_norm": NaN,
|
| 41090 |
+
"learning_rate": 4.1709743332252196e-08,
|
| 41091 |
+
"loss": 0.0,
|
| 41092 |
+
"step": 5869
|
| 41093 |
+
},
|
| 41094 |
+
{
|
| 41095 |
+
"epoch": 0.9873013203262972,
|
| 41096 |
+
"grad_norm": NaN,
|
| 41097 |
+
"learning_rate": 4.061948738677268e-08,
|
| 41098 |
+
"loss": 0.0,
|
| 41099 |
+
"step": 5870
|
| 41100 |
+
},
|
| 41101 |
+
{
|
| 41102 |
+
"epoch": 0.9874695147590615,
|
| 41103 |
+
"grad_norm": NaN,
|
| 41104 |
+
"learning_rate": 3.9543664146746154e-08,
|
| 41105 |
+
"loss": 0.0,
|
| 41106 |
+
"step": 5871
|
| 41107 |
+
},
|
| 41108 |
+
{
|
| 41109 |
+
"epoch": 0.9876377091918257,
|
| 41110 |
+
"grad_norm": NaN,
|
| 41111 |
+
"learning_rate": 3.8482273922962884e-08,
|
| 41112 |
+
"loss": 0.0,
|
| 41113 |
+
"step": 5872
|
| 41114 |
+
},
|
| 41115 |
+
{
|
| 41116 |
+
"epoch": 0.98780590362459,
|
| 41117 |
+
"grad_norm": NaN,
|
| 41118 |
+
"learning_rate": 3.743531702204983e-08,
|
| 41119 |
+
"loss": 0.0,
|
| 41120 |
+
"step": 5873
|
| 41121 |
+
},
|
| 41122 |
+
{
|
| 41123 |
+
"epoch": 0.9879740980573543,
|
| 41124 |
+
"grad_norm": NaN,
|
| 41125 |
+
"learning_rate": 3.6402793746465045e-08,
|
| 41126 |
+
"loss": 0.0,
|
| 41127 |
+
"step": 5874
|
| 41128 |
+
},
|
| 41129 |
+
{
|
| 41130 |
+
"epoch": 0.9881422924901185,
|
| 41131 |
+
"grad_norm": NaN,
|
| 41132 |
+
"learning_rate": 3.538470439448105e-08,
|
| 41133 |
+
"loss": 0.0,
|
| 41134 |
+
"step": 5875
|
| 41135 |
+
},
|
| 41136 |
+
{
|
| 41137 |
+
"epoch": 0.9883104869228828,
|
| 41138 |
+
"grad_norm": NaN,
|
| 41139 |
+
"learning_rate": 3.438104926022923e-08,
|
| 41140 |
+
"loss": 0.0,
|
| 41141 |
+
"step": 5876
|
| 41142 |
+
},
|
| 41143 |
+
{
|
| 41144 |
+
"epoch": 0.9884786813556471,
|
| 41145 |
+
"grad_norm": NaN,
|
| 41146 |
+
"learning_rate": 3.339182863363877e-08,
|
| 41147 |
+
"loss": 0.0,
|
| 41148 |
+
"step": 5877
|
| 41149 |
+
},
|
| 41150 |
+
{
|
| 41151 |
+
"epoch": 0.9886468757884114,
|
| 41152 |
+
"grad_norm": NaN,
|
| 41153 |
+
"learning_rate": 3.241704280049218e-08,
|
| 41154 |
+
"loss": 0.0,
|
| 41155 |
+
"step": 5878
|
| 41156 |
+
},
|
| 41157 |
+
{
|
| 41158 |
+
"epoch": 0.9888150702211757,
|
| 41159 |
+
"grad_norm": NaN,
|
| 41160 |
+
"learning_rate": 3.145669204239754e-08,
|
| 41161 |
+
"loss": 0.0,
|
| 41162 |
+
"step": 5879
|
| 41163 |
+
},
|
| 41164 |
+
{
|
| 41165 |
+
"epoch": 0.9889832646539399,
|
| 41166 |
+
"grad_norm": NaN,
|
| 41167 |
+
"learning_rate": 3.051077663677737e-08,
|
| 41168 |
+
"loss": 0.0,
|
| 41169 |
+
"step": 5880
|
| 41170 |
+
},
|
| 41171 |
+
{
|
| 41172 |
+
"epoch": 0.9891514590867042,
|
| 41173 |
+
"grad_norm": NaN,
|
| 41174 |
+
"learning_rate": 2.9579296856907523e-08,
|
| 41175 |
+
"loss": 0.0,
|
| 41176 |
+
"step": 5881
|
| 41177 |
+
},
|
| 41178 |
+
{
|
| 41179 |
+
"epoch": 0.9893196535194685,
|
| 41180 |
+
"grad_norm": NaN,
|
| 41181 |
+
"learning_rate": 2.86622529718783e-08,
|
| 41182 |
+
"loss": 0.0,
|
| 41183 |
+
"step": 5882
|
| 41184 |
+
},
|
| 41185 |
+
{
|
| 41186 |
+
"epoch": 0.9894878479522328,
|
| 41187 |
+
"grad_norm": NaN,
|
| 41188 |
+
"learning_rate": 2.775964524661667e-08,
|
| 41189 |
+
"loss": 0.0,
|
| 41190 |
+
"step": 5883
|
| 41191 |
+
},
|
| 41192 |
+
{
|
| 41193 |
+
"epoch": 0.989656042384997,
|
| 41194 |
+
"grad_norm": NaN,
|
| 41195 |
+
"learning_rate": 2.6871473941864067e-08,
|
| 41196 |
+
"loss": 0.0,
|
| 41197 |
+
"step": 5884
|
| 41198 |
+
},
|
| 41199 |
+
{
|
| 41200 |
+
"epoch": 0.9898242368177613,
|
| 41201 |
+
"grad_norm": NaN,
|
| 41202 |
+
"learning_rate": 2.599773931422078e-08,
|
| 41203 |
+
"loss": 0.0,
|
| 41204 |
+
"step": 5885
|
| 41205 |
+
},
|
| 41206 |
+
{
|
| 41207 |
+
"epoch": 0.9899924312505256,
|
| 41208 |
+
"grad_norm": NaN,
|
| 41209 |
+
"learning_rate": 2.5138441616079367e-08,
|
| 41210 |
+
"loss": 0.0,
|
| 41211 |
+
"step": 5886
|
| 41212 |
+
},
|
| 41213 |
+
{
|
| 41214 |
+
"epoch": 0.9901606256832899,
|
| 41215 |
+
"grad_norm": NaN,
|
| 41216 |
+
"learning_rate": 2.4293581095696794e-08,
|
| 41217 |
+
"loss": 0.0,
|
| 41218 |
+
"step": 5887
|
| 41219 |
+
},
|
| 41220 |
+
{
|
| 41221 |
+
"epoch": 0.9903288201160542,
|
| 41222 |
+
"grad_norm": NaN,
|
| 41223 |
+
"learning_rate": 2.346315799713894e-08,
|
| 41224 |
+
"loss": 0.0,
|
| 41225 |
+
"step": 5888
|
| 41226 |
+
},
|
| 41227 |
+
{
|
| 41228 |
+
"epoch": 0.9904970145488184,
|
| 41229 |
+
"grad_norm": NaN,
|
| 41230 |
+
"learning_rate": 2.264717256030835e-08,
|
| 41231 |
+
"loss": 0.0,
|
| 41232 |
+
"step": 5889
|
| 41233 |
+
},
|
| 41234 |
+
{
|
| 41235 |
+
"epoch": 0.9906652089815827,
|
| 41236 |
+
"grad_norm": NaN,
|
| 41237 |
+
"learning_rate": 2.1845625020927572e-08,
|
| 41238 |
+
"loss": 0.0,
|
| 41239 |
+
"step": 5890
|
| 41240 |
+
},
|
| 41241 |
+
{
|
| 41242 |
+
"epoch": 0.990833403414347,
|
| 41243 |
+
"grad_norm": NaN,
|
| 41244 |
+
"learning_rate": 2.105851561056138e-08,
|
| 41245 |
+
"loss": 0.0,
|
| 41246 |
+
"step": 5891
|
| 41247 |
+
},
|
| 41248 |
+
{
|
| 41249 |
+
"epoch": 0.9910015978471113,
|
| 41250 |
+
"grad_norm": NaN,
|
| 41251 |
+
"learning_rate": 2.0285844556588996e-08,
|
| 41252 |
+
"loss": 0.0,
|
| 41253 |
+
"step": 5892
|
| 41254 |
+
},
|
| 41255 |
+
{
|
| 41256 |
+
"epoch": 0.9911697922798756,
|
| 41257 |
+
"grad_norm": NaN,
|
| 41258 |
+
"learning_rate": 1.952761208223186e-08,
|
| 41259 |
+
"loss": 0.0,
|
| 41260 |
+
"step": 5893
|
| 41261 |
+
},
|
| 41262 |
+
{
|
| 41263 |
+
"epoch": 0.9913379867126398,
|
| 41264 |
+
"grad_norm": NaN,
|
| 41265 |
+
"learning_rate": 1.878381840653698e-08,
|
| 41266 |
+
"loss": 0.0,
|
| 41267 |
+
"step": 5894
|
| 41268 |
+
},
|
| 41269 |
+
{
|
| 41270 |
+
"epoch": 0.9915061811454041,
|
| 41271 |
+
"grad_norm": NaN,
|
| 41272 |
+
"learning_rate": 1.8054463744376914e-08,
|
| 41273 |
+
"loss": 0.0,
|
| 41274 |
+
"step": 5895
|
| 41275 |
+
},
|
| 41276 |
+
{
|
| 41277 |
+
"epoch": 0.9916743755781684,
|
| 41278 |
+
"grad_norm": NaN,
|
| 41279 |
+
"learning_rate": 1.7339548306449794e-08,
|
| 41280 |
+
"loss": 0.0,
|
| 41281 |
+
"step": 5896
|
| 41282 |
+
},
|
| 41283 |
+
{
|
| 41284 |
+
"epoch": 0.9918425700109327,
|
| 41285 |
+
"grad_norm": NaN,
|
| 41286 |
+
"learning_rate": 1.6639072299284852e-08,
|
| 41287 |
+
"loss": 0.0,
|
| 41288 |
+
"step": 5897
|
| 41289 |
+
},
|
| 41290 |
+
{
|
| 41291 |
+
"epoch": 0.992010764443697,
|
| 41292 |
+
"grad_norm": NaN,
|
| 41293 |
+
"learning_rate": 1.5953035925253547e-08,
|
| 41294 |
+
"loss": 0.0,
|
| 41295 |
+
"step": 5898
|
| 41296 |
+
},
|
| 41297 |
+
{
|
| 41298 |
+
"epoch": 0.9921789588764612,
|
| 41299 |
+
"grad_norm": NaN,
|
| 41300 |
+
"learning_rate": 1.528143938253068e-08,
|
| 41301 |
+
"loss": 0.0,
|
| 41302 |
+
"step": 5899
|
| 41303 |
+
},
|
| 41304 |
+
{
|
| 41305 |
+
"epoch": 0.9923471533092255,
|
| 41306 |
+
"grad_norm": NaN,
|
| 41307 |
+
"learning_rate": 1.4624282865144389e-08,
|
| 41308 |
+
"loss": 0.0,
|
| 41309 |
+
"step": 5900
|
| 41310 |
+
},
|
| 41311 |
+
{
|
| 41312 |
+
"epoch": 0.9925153477419898,
|
| 41313 |
+
"grad_norm": NaN,
|
| 41314 |
+
"learning_rate": 1.3981566562931702e-08,
|
| 41315 |
+
"loss": 0.0,
|
| 41316 |
+
"step": 5901
|
| 41317 |
+
},
|
| 41318 |
+
{
|
| 41319 |
+
"epoch": 0.992683542174754,
|
| 41320 |
+
"grad_norm": NaN,
|
| 41321 |
+
"learning_rate": 1.3353290661571871e-08,
|
| 41322 |
+
"loss": 0.0,
|
| 41323 |
+
"step": 5902
|
| 41324 |
+
},
|
| 41325 |
+
{
|
| 41326 |
+
"epoch": 0.9928517366075182,
|
| 41327 |
+
"grad_norm": NaN,
|
| 41328 |
+
"learning_rate": 1.2739455342558603e-08,
|
| 41329 |
+
"loss": 0.0,
|
| 41330 |
+
"step": 5903
|
| 41331 |
+
},
|
| 41332 |
+
{
|
| 41333 |
+
"epoch": 0.9930199310402825,
|
| 41334 |
+
"grad_norm": NaN,
|
| 41335 |
+
"learning_rate": 1.2140060783227824e-08,
|
| 41336 |
+
"loss": 0.0,
|
| 41337 |
+
"step": 5904
|
| 41338 |
+
},
|
| 41339 |
+
{
|
| 41340 |
+
"epoch": 0.9931881254730468,
|
| 41341 |
+
"grad_norm": NaN,
|
| 41342 |
+
"learning_rate": 1.155510715674102e-08,
|
| 41343 |
+
"loss": 0.0,
|
| 41344 |
+
"step": 5905
|
| 41345 |
+
},
|
| 41346 |
+
{
|
| 41347 |
+
"epoch": 0.9933563199058111,
|
| 41348 |
+
"grad_norm": NaN,
|
| 41349 |
+
"learning_rate": 1.098459463207968e-08,
|
| 41350 |
+
"loss": 0.0,
|
| 41351 |
+
"step": 5906
|
| 41352 |
+
},
|
| 41353 |
+
{
|
| 41354 |
+
"epoch": 0.9935245143385754,
|
| 41355 |
+
"grad_norm": NaN,
|
| 41356 |
+
"learning_rate": 1.042852337406197e-08,
|
| 41357 |
+
"loss": 0.0,
|
| 41358 |
+
"step": 5907
|
| 41359 |
+
},
|
| 41360 |
+
{
|
| 41361 |
+
"epoch": 0.9936927087713396,
|
| 41362 |
+
"grad_norm": NaN,
|
| 41363 |
+
"learning_rate": 9.88689354332606e-09,
|
| 41364 |
+
"loss": 0.0,
|
| 41365 |
+
"step": 5908
|
| 41366 |
+
},
|
| 41367 |
+
{
|
| 41368 |
+
"epoch": 0.9938609032041039,
|
| 41369 |
+
"grad_norm": NaN,
|
| 41370 |
+
"learning_rate": 9.359705296346776e-09,
|
| 41371 |
+
"loss": 0.0,
|
| 41372 |
+
"step": 5909
|
| 41373 |
+
},
|
| 41374 |
+
{
|
| 41375 |
+
"epoch": 0.9940290976368682,
|
| 41376 |
+
"grad_norm": NaN,
|
| 41377 |
+
"learning_rate": 8.846958785418968e-09,
|
| 41378 |
+
"loss": 0.0,
|
| 41379 |
+
"step": 5910
|
| 41380 |
+
},
|
| 41381 |
+
{
|
| 41382 |
+
"epoch": 0.9941972920696325,
|
| 41383 |
+
"grad_norm": NaN,
|
| 41384 |
+
"learning_rate": 8.34865415867414e-09,
|
| 41385 |
+
"loss": 0.0,
|
| 41386 |
+
"step": 5911
|
| 41387 |
+
},
|
| 41388 |
+
{
|
| 41389 |
+
"epoch": 0.9943654865023968,
|
| 41390 |
+
"grad_norm": NaN,
|
| 41391 |
+
"learning_rate": 7.86479156006381e-09,
|
| 41392 |
+
"loss": 0.0,
|
| 41393 |
+
"step": 5912
|
| 41394 |
+
},
|
| 41395 |
+
{
|
| 41396 |
+
"epoch": 0.994533680935161,
|
| 41397 |
+
"grad_norm": NaN,
|
| 41398 |
+
"learning_rate": 7.3953711293706096e-09,
|
| 41399 |
+
"loss": 0.0,
|
| 41400 |
+
"step": 5913
|
| 41401 |
+
},
|
| 41402 |
+
{
|
| 41403 |
+
"epoch": 0.9947018753679253,
|
| 41404 |
+
"grad_norm": NaN,
|
| 41405 |
+
"learning_rate": 6.940393002202727e-09,
|
| 41406 |
+
"loss": 0.0,
|
| 41407 |
+
"step": 5914
|
| 41408 |
+
},
|
| 41409 |
+
{
|
| 41410 |
+
"epoch": 0.9948700698006896,
|
| 41411 |
+
"grad_norm": NaN,
|
| 41412 |
+
"learning_rate": 6.4998573100050195e-09,
|
| 41413 |
+
"loss": 0.0,
|
| 41414 |
+
"step": 5915
|
| 41415 |
+
},
|
| 41416 |
+
{
|
| 41417 |
+
"epoch": 0.9950382642334539,
|
| 41418 |
+
"grad_norm": NaN,
|
| 41419 |
+
"learning_rate": 6.0737641800368e-09,
|
| 41420 |
+
"loss": 0.0,
|
| 41421 |
+
"step": 5916
|
| 41422 |
+
},
|
| 41423 |
+
{
|
| 41424 |
+
"epoch": 0.9952064586662182,
|
| 41425 |
+
"grad_norm": NaN,
|
| 41426 |
+
"learning_rate": 5.662113735394048e-09,
|
| 41427 |
+
"loss": 0.0,
|
| 41428 |
+
"step": 5917
|
| 41429 |
+
},
|
| 41430 |
+
{
|
| 41431 |
+
"epoch": 0.9953746530989824,
|
| 41432 |
+
"grad_norm": NaN,
|
| 41433 |
+
"learning_rate": 5.264906095003852e-09,
|
| 41434 |
+
"loss": 0.0,
|
| 41435 |
+
"step": 5918
|
| 41436 |
+
},
|
| 41437 |
+
{
|
| 41438 |
+
"epoch": 0.9955428475317467,
|
| 41439 |
+
"grad_norm": NaN,
|
| 41440 |
+
"learning_rate": 4.8821413736022115e-09,
|
| 41441 |
+
"loss": 0.0,
|
| 41442 |
+
"step": 5919
|
| 41443 |
+
},
|
| 41444 |
+
{
|
| 41445 |
+
"epoch": 0.995711041964511,
|
| 41446 |
+
"grad_norm": NaN,
|
| 41447 |
+
"learning_rate": 4.51381968177289e-09,
|
| 41448 |
+
"loss": 0.0,
|
| 41449 |
+
"step": 5920
|
| 41450 |
+
},
|
| 41451 |
+
{
|
| 41452 |
+
"epoch": 0.9958792363972753,
|
| 41453 |
+
"grad_norm": NaN,
|
| 41454 |
+
"learning_rate": 4.159941125925215e-09,
|
| 41455 |
+
"loss": 0.0,
|
| 41456 |
+
"step": 5921
|
| 41457 |
+
},
|
| 41458 |
+
{
|
| 41459 |
+
"epoch": 0.9960474308300395,
|
| 41460 |
+
"grad_norm": NaN,
|
| 41461 |
+
"learning_rate": 3.820505808277419e-09,
|
| 41462 |
+
"loss": 0.0,
|
| 41463 |
+
"step": 5922
|
| 41464 |
+
},
|
| 41465 |
+
{
|
| 41466 |
+
"epoch": 0.9962156252628038,
|
| 41467 |
+
"grad_norm": NaN,
|
| 41468 |
+
"learning_rate": 3.4955138269010534e-09,
|
| 41469 |
+
"loss": 0.0,
|
| 41470 |
+
"step": 5923
|
| 41471 |
+
},
|
| 41472 |
+
{
|
| 41473 |
+
"epoch": 0.9963838196955681,
|
| 41474 |
+
"grad_norm": NaN,
|
| 41475 |
+
"learning_rate": 3.184965275676577e-09,
|
| 41476 |
+
"loss": 0.0,
|
| 41477 |
+
"step": 5924
|
| 41478 |
+
},
|
| 41479 |
+
{
|
| 41480 |
+
"epoch": 0.9965520141283324,
|
| 41481 |
+
"grad_norm": NaN,
|
| 41482 |
+
"learning_rate": 2.8888602443211122e-09,
|
| 41483 |
+
"loss": 0.0,
|
| 41484 |
+
"step": 5925
|
| 41485 |
+
},
|
| 41486 |
+
{
|
| 41487 |
+
"epoch": 0.9967202085610967,
|
| 41488 |
+
"grad_norm": NaN,
|
| 41489 |
+
"learning_rate": 2.607198818371792e-09,
|
| 41490 |
+
"loss": 0.0,
|
| 41491 |
+
"step": 5926
|
| 41492 |
+
},
|
| 41493 |
+
{
|
| 41494 |
+
"epoch": 0.996888402993861,
|
| 41495 |
+
"grad_norm": NaN,
|
| 41496 |
+
"learning_rate": 2.3399810792024133e-09,
|
| 41497 |
+
"loss": 0.0,
|
| 41498 |
+
"step": 5927
|
| 41499 |
+
},
|
| 41500 |
+
{
|
| 41501 |
+
"epoch": 0.9970565974266252,
|
| 41502 |
+
"grad_norm": NaN,
|
| 41503 |
+
"learning_rate": 2.087207104001232e-09,
|
| 41504 |
+
"loss": 0.0,
|
| 41505 |
+
"step": 5928
|
| 41506 |
+
},
|
| 41507 |
+
{
|
| 41508 |
+
"epoch": 0.9972247918593895,
|
| 41509 |
+
"grad_norm": NaN,
|
| 41510 |
+
"learning_rate": 1.8488769658042693e-09,
|
| 41511 |
+
"loss": 0.0,
|
| 41512 |
+
"step": 5929
|
| 41513 |
+
},
|
| 41514 |
+
{
|
| 41515 |
+
"epoch": 0.9973929862921538,
|
| 41516 |
+
"grad_norm": NaN,
|
| 41517 |
+
"learning_rate": 1.624990733450904e-09,
|
| 41518 |
+
"loss": 0.0,
|
| 41519 |
+
"step": 5930
|
| 41520 |
+
},
|
| 41521 |
+
{
|
| 41522 |
+
"epoch": 0.997561180724918,
|
| 41523 |
+
"grad_norm": NaN,
|
| 41524 |
+
"learning_rate": 1.4155484716227296e-09,
|
| 41525 |
+
"loss": 0.0,
|
| 41526 |
+
"step": 5931
|
| 41527 |
+
},
|
| 41528 |
+
{
|
| 41529 |
+
"epoch": 0.9977293751576822,
|
| 41530 |
+
"grad_norm": NaN,
|
| 41531 |
+
"learning_rate": 1.220550240826901e-09,
|
| 41532 |
+
"loss": 0.0,
|
| 41533 |
+
"step": 5932
|
| 41534 |
+
},
|
| 41535 |
+
{
|
| 41536 |
+
"epoch": 0.9978975695904465,
|
| 41537 |
+
"grad_norm": NaN,
|
| 41538 |
+
"learning_rate": 1.0399960974016854e-09,
|
| 41539 |
+
"loss": 0.0,
|
| 41540 |
+
"step": 5933
|
| 41541 |
+
},
|
| 41542 |
+
{
|
| 41543 |
+
"epoch": 0.9980657640232108,
|
| 41544 |
+
"grad_norm": NaN,
|
| 41545 |
+
"learning_rate": 8.738860934942584e-10,
|
| 41546 |
+
"loss": 0.0,
|
| 41547 |
+
"step": 5934
|
| 41548 |
+
},
|
| 41549 |
+
{
|
| 41550 |
+
"epoch": 0.9982339584559751,
|
| 41551 |
+
"grad_norm": NaN,
|
| 41552 |
+
"learning_rate": 7.222202770995612e-10,
|
| 41553 |
+
"loss": 0.0,
|
| 41554 |
+
"step": 5935
|
| 41555 |
+
},
|
| 41556 |
+
{
|
| 41557 |
+
"epoch": 0.9984021528887393,
|
| 41558 |
+
"grad_norm": NaN,
|
| 41559 |
+
"learning_rate": 5.849986920325456e-10,
|
| 41560 |
+
"loss": 0.0,
|
| 41561 |
+
"step": 5936
|
| 41562 |
+
},
|
| 41563 |
+
{
|
| 41564 |
+
"epoch": 0.9985703473215036,
|
| 41565 |
+
"grad_norm": NaN,
|
| 41566 |
+
"learning_rate": 4.622213779392759e-10,
|
| 41567 |
+
"loss": 0.0,
|
| 41568 |
+
"step": 5937
|
| 41569 |
+
},
|
| 41570 |
+
{
|
| 41571 |
+
"epoch": 0.9987385417542679,
|
| 41572 |
+
"grad_norm": NaN,
|
| 41573 |
+
"learning_rate": 3.538883702747242e-10,
|
| 41574 |
+
"loss": 0.0,
|
| 41575 |
+
"step": 5938
|
| 41576 |
+
},
|
| 41577 |
+
{
|
| 41578 |
+
"epoch": 0.9989067361870322,
|
| 41579 |
+
"grad_norm": NaN,
|
| 41580 |
+
"learning_rate": 2.599997003471799e-10,
|
| 41581 |
+
"loss": 0.0,
|
| 41582 |
+
"step": 5939
|
| 41583 |
+
},
|
| 41584 |
+
{
|
| 41585 |
+
"epoch": 0.9990749306197965,
|
| 41586 |
+
"grad_norm": NaN,
|
| 41587 |
+
"learning_rate": 1.8055539527939148e-10,
|
| 41588 |
+
"loss": 0.0,
|
| 41589 |
+
"step": 5940
|
| 41590 |
+
},
|
| 41591 |
+
{
|
| 41592 |
+
"epoch": 0.9992431250525607,
|
| 41593 |
+
"grad_norm": NaN,
|
| 41594 |
+
"learning_rate": 1.155554780141177e-10,
|
| 41595 |
+
"loss": 0.0,
|
| 41596 |
+
"step": 5941
|
| 41597 |
+
},
|
| 41598 |
+
{
|
| 41599 |
+
"epoch": 0.999411319485325,
|
| 41600 |
+
"grad_norm": NaN,
|
| 41601 |
+
"learning_rate": 6.499996733633218e-11,
|
| 41602 |
+
"loss": 0.0,
|
| 41603 |
+
"step": 5942
|
| 41604 |
+
},
|
| 41605 |
+
{
|
| 41606 |
+
"epoch": 0.9995795139180893,
|
| 41607 |
+
"grad_norm": NaN,
|
| 41608 |
+
"learning_rate": 2.8888877851018792e-11,
|
| 41609 |
+
"loss": 0.0,
|
| 41610 |
+
"step": 5943
|
| 41611 |
+
},
|
| 41612 |
+
{
|
| 41613 |
+
"epoch": 0.9997477083508536,
|
| 41614 |
+
"grad_norm": NaN,
|
| 41615 |
+
"learning_rate": 7.2222199831717406e-12,
|
| 41616 |
+
"loss": 0.0,
|
| 41617 |
+
"step": 5944
|
| 41618 |
+
},
|
| 41619 |
+
{
|
| 41620 |
+
"epoch": 0.9999159027836179,
|
| 41621 |
+
"grad_norm": NaN,
|
| 41622 |
+
"learning_rate": 0.0,
|
| 41623 |
+
"loss": 0.0,
|
| 41624 |
+
"step": 5945
|
| 41625 |
}
|
| 41626 |
],
|
| 41627 |
"logging_steps": 1,
|
|
|
|
| 41636 |
"should_evaluate": false,
|
| 41637 |
"should_log": false,
|
| 41638 |
"should_save": true,
|
| 41639 |
+
"should_training_stop": true
|
| 41640 |
},
|
| 41641 |
"attributes": {}
|
| 41642 |
}
|
| 41643 |
},
|
| 41644 |
+
"total_flos": 1.1191100899590144e+17,
|
| 41645 |
"train_batch_size": 8,
|
| 41646 |
"trial_name": null,
|
| 41647 |
"trial_params": null
|