Training in progress, epoch 1, 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 +2629 -4
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1101572914
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8f82829ffee530562250a891ebc2de26fad2f54c5434110c9be18e0716ac6d22
|
| 3 |
size 1101572914
|
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 sha256:8ed244bccd74c18cfe87600cb347ede07968b51d25052de564ed5d70d1359f01
|
| 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 sha256:1380500cb5ce47f6085ce18023f66434aacc2bfc3f43d387430fdbaa4c016dad
|
| 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":
|
| 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,
|
|
@@ -2640,6 +2640,2631 @@
|
|
| 2640 |
"eval_samples_per_second": 7.59,
|
| 2641 |
"eval_steps_per_second": 1.907,
|
| 2642 |
"step": 375
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2643 |
}
|
| 2644 |
],
|
| 2645 |
"logging_steps": 1,
|
|
@@ -2654,12 +5279,12 @@
|
|
| 2654 |
"should_evaluate": false,
|
| 2655 |
"should_log": false,
|
| 2656 |
"should_save": true,
|
| 2657 |
-
"should_training_stop":
|
| 2658 |
},
|
| 2659 |
"attributes": {}
|
| 2660 |
}
|
| 2661 |
},
|
| 2662 |
-
"total_flos":
|
| 2663 |
"train_batch_size": 4,
|
| 2664 |
"trial_name": null,
|
| 2665 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.997006983704689,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 750,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 2640 |
"eval_samples_per_second": 7.59,
|
| 2641 |
"eval_steps_per_second": 1.907,
|
| 2642 |
"step": 375
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"epoch": 1.0019953441968739,
|
| 2646 |
+
"grad_norm": NaN,
|
| 2647 |
+
"learning_rate": 5.084903828741312e-05,
|
| 2648 |
+
"loss": 0.0,
|
| 2649 |
+
"step": 376
|
| 2650 |
+
},
|
| 2651 |
+
{
|
| 2652 |
+
"epoch": 1.0046558031260393,
|
| 2653 |
+
"grad_norm": NaN,
|
| 2654 |
+
"learning_rate": 5.063679210562602e-05,
|
| 2655 |
+
"loss": 0.0,
|
| 2656 |
+
"step": 377
|
| 2657 |
+
},
|
| 2658 |
+
{
|
| 2659 |
+
"epoch": 1.0073162620552045,
|
| 2660 |
+
"grad_norm": NaN,
|
| 2661 |
+
"learning_rate": 5.042453444670828e-05,
|
| 2662 |
+
"loss": 0.0,
|
| 2663 |
+
"step": 378
|
| 2664 |
+
},
|
| 2665 |
+
{
|
| 2666 |
+
"epoch": 1.0099767209843697,
|
| 2667 |
+
"grad_norm": NaN,
|
| 2668 |
+
"learning_rate": 5.021226913625522e-05,
|
| 2669 |
+
"loss": 0.0,
|
| 2670 |
+
"step": 379
|
| 2671 |
+
},
|
| 2672 |
+
{
|
| 2673 |
+
"epoch": 1.0126371799135352,
|
| 2674 |
+
"grad_norm": NaN,
|
| 2675 |
+
"learning_rate": 5e-05,
|
| 2676 |
+
"loss": 0.0,
|
| 2677 |
+
"step": 380
|
| 2678 |
+
},
|
| 2679 |
+
{
|
| 2680 |
+
"epoch": 1.0152976388427004,
|
| 2681 |
+
"grad_norm": NaN,
|
| 2682 |
+
"learning_rate": 4.978773086374479e-05,
|
| 2683 |
+
"loss": 0.0,
|
| 2684 |
+
"step": 381
|
| 2685 |
+
},
|
| 2686 |
+
{
|
| 2687 |
+
"epoch": 1.0179580977718656,
|
| 2688 |
+
"grad_norm": NaN,
|
| 2689 |
+
"learning_rate": 4.957546555329173e-05,
|
| 2690 |
+
"loss": 0.0,
|
| 2691 |
+
"step": 382
|
| 2692 |
+
},
|
| 2693 |
+
{
|
| 2694 |
+
"epoch": 1.0206185567010309,
|
| 2695 |
+
"grad_norm": NaN,
|
| 2696 |
+
"learning_rate": 4.9363207894374e-05,
|
| 2697 |
+
"loss": 0.0,
|
| 2698 |
+
"step": 383
|
| 2699 |
+
},
|
| 2700 |
+
{
|
| 2701 |
+
"epoch": 1.0232790156301963,
|
| 2702 |
+
"grad_norm": NaN,
|
| 2703 |
+
"learning_rate": 4.9150961712586895e-05,
|
| 2704 |
+
"loss": 0.0,
|
| 2705 |
+
"step": 384
|
| 2706 |
+
},
|
| 2707 |
+
{
|
| 2708 |
+
"epoch": 1.0259394745593615,
|
| 2709 |
+
"grad_norm": NaN,
|
| 2710 |
+
"learning_rate": 4.893873083331882e-05,
|
| 2711 |
+
"loss": 0.0,
|
| 2712 |
+
"step": 385
|
| 2713 |
+
},
|
| 2714 |
+
{
|
| 2715 |
+
"epoch": 1.0285999334885267,
|
| 2716 |
+
"grad_norm": NaN,
|
| 2717 |
+
"learning_rate": 4.8726519081682444e-05,
|
| 2718 |
+
"loss": 0.0,
|
| 2719 |
+
"step": 386
|
| 2720 |
+
},
|
| 2721 |
+
{
|
| 2722 |
+
"epoch": 1.031260392417692,
|
| 2723 |
+
"grad_norm": NaN,
|
| 2724 |
+
"learning_rate": 4.851433028244562e-05,
|
| 2725 |
+
"loss": 0.0,
|
| 2726 |
+
"step": 387
|
| 2727 |
+
},
|
| 2728 |
+
{
|
| 2729 |
+
"epoch": 1.0339208513468574,
|
| 2730 |
+
"grad_norm": NaN,
|
| 2731 |
+
"learning_rate": 4.830216825996257e-05,
|
| 2732 |
+
"loss": 0.0,
|
| 2733 |
+
"step": 388
|
| 2734 |
+
},
|
| 2735 |
+
{
|
| 2736 |
+
"epoch": 1.0365813102760226,
|
| 2737 |
+
"grad_norm": NaN,
|
| 2738 |
+
"learning_rate": 4.809003683810486e-05,
|
| 2739 |
+
"loss": 0.0,
|
| 2740 |
+
"step": 389
|
| 2741 |
+
},
|
| 2742 |
+
{
|
| 2743 |
+
"epoch": 1.0392417692051878,
|
| 2744 |
+
"grad_norm": NaN,
|
| 2745 |
+
"learning_rate": 4.78779398401926e-05,
|
| 2746 |
+
"loss": 0.0,
|
| 2747 |
+
"step": 390
|
| 2748 |
+
},
|
| 2749 |
+
{
|
| 2750 |
+
"epoch": 1.041902228134353,
|
| 2751 |
+
"grad_norm": NaN,
|
| 2752 |
+
"learning_rate": 4.7665881088925376e-05,
|
| 2753 |
+
"loss": 0.0,
|
| 2754 |
+
"step": 391
|
| 2755 |
+
},
|
| 2756 |
+
{
|
| 2757 |
+
"epoch": 1.0445626870635185,
|
| 2758 |
+
"grad_norm": NaN,
|
| 2759 |
+
"learning_rate": 4.7453864406313544e-05,
|
| 2760 |
+
"loss": 0.0,
|
| 2761 |
+
"step": 392
|
| 2762 |
+
},
|
| 2763 |
+
{
|
| 2764 |
+
"epoch": 1.0472231459926837,
|
| 2765 |
+
"grad_norm": NaN,
|
| 2766 |
+
"learning_rate": 4.7241893613609126e-05,
|
| 2767 |
+
"loss": 0.0,
|
| 2768 |
+
"step": 393
|
| 2769 |
+
},
|
| 2770 |
+
{
|
| 2771 |
+
"epoch": 1.049883604921849,
|
| 2772 |
+
"grad_norm": NaN,
|
| 2773 |
+
"learning_rate": 4.702997253123716e-05,
|
| 2774 |
+
"loss": 0.0,
|
| 2775 |
+
"step": 394
|
| 2776 |
+
},
|
| 2777 |
+
{
|
| 2778 |
+
"epoch": 1.0525440638510144,
|
| 2779 |
+
"grad_norm": NaN,
|
| 2780 |
+
"learning_rate": 4.6818104978726685e-05,
|
| 2781 |
+
"loss": 0.0,
|
| 2782 |
+
"step": 395
|
| 2783 |
+
},
|
| 2784 |
+
{
|
| 2785 |
+
"epoch": 1.0552045227801796,
|
| 2786 |
+
"grad_norm": NaN,
|
| 2787 |
+
"learning_rate": 4.6606294774641966e-05,
|
| 2788 |
+
"loss": 0.0,
|
| 2789 |
+
"step": 396
|
| 2790 |
+
},
|
| 2791 |
+
{
|
| 2792 |
+
"epoch": 1.0578649817093448,
|
| 2793 |
+
"grad_norm": NaN,
|
| 2794 |
+
"learning_rate": 4.6394545736513634e-05,
|
| 2795 |
+
"loss": 0.0,
|
| 2796 |
+
"step": 397
|
| 2797 |
+
},
|
| 2798 |
+
{
|
| 2799 |
+
"epoch": 1.06052544063851,
|
| 2800 |
+
"grad_norm": NaN,
|
| 2801 |
+
"learning_rate": 4.618286168076993e-05,
|
| 2802 |
+
"loss": 0.0,
|
| 2803 |
+
"step": 398
|
| 2804 |
+
},
|
| 2805 |
+
{
|
| 2806 |
+
"epoch": 1.0631858995676755,
|
| 2807 |
+
"grad_norm": NaN,
|
| 2808 |
+
"learning_rate": 4.597124642266788e-05,
|
| 2809 |
+
"loss": 0.0,
|
| 2810 |
+
"step": 399
|
| 2811 |
+
},
|
| 2812 |
+
{
|
| 2813 |
+
"epoch": 1.0658463584968407,
|
| 2814 |
+
"grad_norm": NaN,
|
| 2815 |
+
"learning_rate": 4.575970377622456e-05,
|
| 2816 |
+
"loss": 0.0,
|
| 2817 |
+
"step": 400
|
| 2818 |
+
},
|
| 2819 |
+
{
|
| 2820 |
+
"epoch": 1.068506817426006,
|
| 2821 |
+
"grad_norm": NaN,
|
| 2822 |
+
"learning_rate": 4.554823755414829e-05,
|
| 2823 |
+
"loss": 0.0,
|
| 2824 |
+
"step": 401
|
| 2825 |
+
},
|
| 2826 |
+
{
|
| 2827 |
+
"epoch": 1.0711672763551712,
|
| 2828 |
+
"grad_norm": NaN,
|
| 2829 |
+
"learning_rate": 4.5336851567770076e-05,
|
| 2830 |
+
"loss": 0.0,
|
| 2831 |
+
"step": 402
|
| 2832 |
+
},
|
| 2833 |
+
{
|
| 2834 |
+
"epoch": 1.0738277352843366,
|
| 2835 |
+
"grad_norm": NaN,
|
| 2836 |
+
"learning_rate": 4.5125549626974696e-05,
|
| 2837 |
+
"loss": 0.0,
|
| 2838 |
+
"step": 403
|
| 2839 |
+
},
|
| 2840 |
+
{
|
| 2841 |
+
"epoch": 1.0764881942135018,
|
| 2842 |
+
"grad_norm": NaN,
|
| 2843 |
+
"learning_rate": 4.491433554013221e-05,
|
| 2844 |
+
"loss": 0.0,
|
| 2845 |
+
"step": 404
|
| 2846 |
+
},
|
| 2847 |
+
{
|
| 2848 |
+
"epoch": 1.079148653142667,
|
| 2849 |
+
"grad_norm": NaN,
|
| 2850 |
+
"learning_rate": 4.47032131140292e-05,
|
| 2851 |
+
"loss": 0.0,
|
| 2852 |
+
"step": 405
|
| 2853 |
+
},
|
| 2854 |
+
{
|
| 2855 |
+
"epoch": 1.0818091120718325,
|
| 2856 |
+
"grad_norm": NaN,
|
| 2857 |
+
"learning_rate": 4.449218615380029e-05,
|
| 2858 |
+
"loss": 0.0,
|
| 2859 |
+
"step": 406
|
| 2860 |
+
},
|
| 2861 |
+
{
|
| 2862 |
+
"epoch": 1.0844695710009977,
|
| 2863 |
+
"grad_norm": NaN,
|
| 2864 |
+
"learning_rate": 4.4281258462859396e-05,
|
| 2865 |
+
"loss": 0.0,
|
| 2866 |
+
"step": 407
|
| 2867 |
+
},
|
| 2868 |
+
{
|
| 2869 |
+
"epoch": 1.087130029930163,
|
| 2870 |
+
"grad_norm": NaN,
|
| 2871 |
+
"learning_rate": 4.407043384283136e-05,
|
| 2872 |
+
"loss": 0.0,
|
| 2873 |
+
"step": 408
|
| 2874 |
+
},
|
| 2875 |
+
{
|
| 2876 |
+
"epoch": 1.0897904888593282,
|
| 2877 |
+
"grad_norm": NaN,
|
| 2878 |
+
"learning_rate": 4.3859716093483245e-05,
|
| 2879 |
+
"loss": 0.0,
|
| 2880 |
+
"step": 409
|
| 2881 |
+
},
|
| 2882 |
+
{
|
| 2883 |
+
"epoch": 1.0924509477884936,
|
| 2884 |
+
"grad_norm": NaN,
|
| 2885 |
+
"learning_rate": 4.364910901265606e-05,
|
| 2886 |
+
"loss": 0.0,
|
| 2887 |
+
"step": 410
|
| 2888 |
+
},
|
| 2889 |
+
{
|
| 2890 |
+
"epoch": 1.0951114067176588,
|
| 2891 |
+
"grad_norm": NaN,
|
| 2892 |
+
"learning_rate": 4.343861639619611e-05,
|
| 2893 |
+
"loss": 0.0,
|
| 2894 |
+
"step": 411
|
| 2895 |
+
},
|
| 2896 |
+
{
|
| 2897 |
+
"epoch": 1.097771865646824,
|
| 2898 |
+
"grad_norm": NaN,
|
| 2899 |
+
"learning_rate": 4.322824203788669e-05,
|
| 2900 |
+
"loss": 0.0,
|
| 2901 |
+
"step": 412
|
| 2902 |
+
},
|
| 2903 |
+
{
|
| 2904 |
+
"epoch": 1.1004323245759893,
|
| 2905 |
+
"grad_norm": NaN,
|
| 2906 |
+
"learning_rate": 4.3017989729379675e-05,
|
| 2907 |
+
"loss": 0.0,
|
| 2908 |
+
"step": 413
|
| 2909 |
+
},
|
| 2910 |
+
{
|
| 2911 |
+
"epoch": 1.1030927835051547,
|
| 2912 |
+
"grad_norm": NaN,
|
| 2913 |
+
"learning_rate": 4.280786326012723e-05,
|
| 2914 |
+
"loss": 0.0,
|
| 2915 |
+
"step": 414
|
| 2916 |
+
},
|
| 2917 |
+
{
|
| 2918 |
+
"epoch": 1.10575324243432,
|
| 2919 |
+
"grad_norm": NaN,
|
| 2920 |
+
"learning_rate": 4.2597866417313436e-05,
|
| 2921 |
+
"loss": 0.0,
|
| 2922 |
+
"step": 415
|
| 2923 |
+
},
|
| 2924 |
+
{
|
| 2925 |
+
"epoch": 1.1084137013634852,
|
| 2926 |
+
"grad_norm": NaN,
|
| 2927 |
+
"learning_rate": 4.23880029857861e-05,
|
| 2928 |
+
"loss": 0.0,
|
| 2929 |
+
"step": 416
|
| 2930 |
+
},
|
| 2931 |
+
{
|
| 2932 |
+
"epoch": 1.1110741602926504,
|
| 2933 |
+
"grad_norm": NaN,
|
| 2934 |
+
"learning_rate": 4.2178276747988446e-05,
|
| 2935 |
+
"loss": 0.0,
|
| 2936 |
+
"step": 417
|
| 2937 |
+
},
|
| 2938 |
+
{
|
| 2939 |
+
"epoch": 1.1137346192218158,
|
| 2940 |
+
"grad_norm": NaN,
|
| 2941 |
+
"learning_rate": 4.196869148389114e-05,
|
| 2942 |
+
"loss": 0.0,
|
| 2943 |
+
"step": 418
|
| 2944 |
+
},
|
| 2945 |
+
{
|
| 2946 |
+
"epoch": 1.116395078150981,
|
| 2947 |
+
"grad_norm": NaN,
|
| 2948 |
+
"learning_rate": 4.175925097092388e-05,
|
| 2949 |
+
"loss": 0.0,
|
| 2950 |
+
"step": 419
|
| 2951 |
+
},
|
| 2952 |
+
{
|
| 2953 |
+
"epoch": 1.1190555370801463,
|
| 2954 |
+
"grad_norm": NaN,
|
| 2955 |
+
"learning_rate": 4.1549958983907555e-05,
|
| 2956 |
+
"loss": 0.0,
|
| 2957 |
+
"step": 420
|
| 2958 |
+
},
|
| 2959 |
+
{
|
| 2960 |
+
"epoch": 1.1217159960093115,
|
| 2961 |
+
"grad_norm": NaN,
|
| 2962 |
+
"learning_rate": 4.1340819294986076e-05,
|
| 2963 |
+
"loss": 0.0,
|
| 2964 |
+
"step": 421
|
| 2965 |
+
},
|
| 2966 |
+
{
|
| 2967 |
+
"epoch": 1.124376454938477,
|
| 2968 |
+
"grad_norm": NaN,
|
| 2969 |
+
"learning_rate": 4.113183567355846e-05,
|
| 2970 |
+
"loss": 0.0,
|
| 2971 |
+
"step": 422
|
| 2972 |
+
},
|
| 2973 |
+
{
|
| 2974 |
+
"epoch": 1.1270369138676422,
|
| 2975 |
+
"grad_norm": NaN,
|
| 2976 |
+
"learning_rate": 4.092301188621084e-05,
|
| 2977 |
+
"loss": 0.0,
|
| 2978 |
+
"step": 423
|
| 2979 |
+
},
|
| 2980 |
+
{
|
| 2981 |
+
"epoch": 1.1296973727968074,
|
| 2982 |
+
"grad_norm": NaN,
|
| 2983 |
+
"learning_rate": 4.0714351696648614e-05,
|
| 2984 |
+
"loss": 0.0,
|
| 2985 |
+
"step": 424
|
| 2986 |
+
},
|
| 2987 |
+
{
|
| 2988 |
+
"epoch": 1.1323578317259728,
|
| 2989 |
+
"grad_norm": NaN,
|
| 2990 |
+
"learning_rate": 4.050585886562858e-05,
|
| 2991 |
+
"loss": 0.0,
|
| 2992 |
+
"step": 425
|
| 2993 |
+
},
|
| 2994 |
+
{
|
| 2995 |
+
"epoch": 1.135018290655138,
|
| 2996 |
+
"grad_norm": NaN,
|
| 2997 |
+
"learning_rate": 4.0297537150891235e-05,
|
| 2998 |
+
"loss": 0.0,
|
| 2999 |
+
"step": 426
|
| 3000 |
+
},
|
| 3001 |
+
{
|
| 3002 |
+
"epoch": 1.1376787495843033,
|
| 3003 |
+
"grad_norm": NaN,
|
| 3004 |
+
"learning_rate": 4.008939030709291e-05,
|
| 3005 |
+
"loss": 0.0,
|
| 3006 |
+
"step": 427
|
| 3007 |
+
},
|
| 3008 |
+
{
|
| 3009 |
+
"epoch": 1.1403392085134685,
|
| 3010 |
+
"grad_norm": NaN,
|
| 3011 |
+
"learning_rate": 3.988142208573822e-05,
|
| 3012 |
+
"loss": 0.0,
|
| 3013 |
+
"step": 428
|
| 3014 |
+
},
|
| 3015 |
+
{
|
| 3016 |
+
"epoch": 1.142999667442634,
|
| 3017 |
+
"grad_norm": NaN,
|
| 3018 |
+
"learning_rate": 3.9673636235112376e-05,
|
| 3019 |
+
"loss": 0.0,
|
| 3020 |
+
"step": 429
|
| 3021 |
+
},
|
| 3022 |
+
{
|
| 3023 |
+
"epoch": 1.1456601263717991,
|
| 3024 |
+
"grad_norm": NaN,
|
| 3025 |
+
"learning_rate": 3.94660365002137e-05,
|
| 3026 |
+
"loss": 0.0,
|
| 3027 |
+
"step": 430
|
| 3028 |
+
},
|
| 3029 |
+
{
|
| 3030 |
+
"epoch": 1.1483205853009644,
|
| 3031 |
+
"grad_norm": NaN,
|
| 3032 |
+
"learning_rate": 3.925862662268602e-05,
|
| 3033 |
+
"loss": 0.0,
|
| 3034 |
+
"step": 431
|
| 3035 |
+
},
|
| 3036 |
+
{
|
| 3037 |
+
"epoch": 1.1509810442301296,
|
| 3038 |
+
"grad_norm": NaN,
|
| 3039 |
+
"learning_rate": 3.905141034075135e-05,
|
| 3040 |
+
"loss": 0.0,
|
| 3041 |
+
"step": 432
|
| 3042 |
+
},
|
| 3043 |
+
{
|
| 3044 |
+
"epoch": 1.153641503159295,
|
| 3045 |
+
"grad_norm": NaN,
|
| 3046 |
+
"learning_rate": 3.884439138914243e-05,
|
| 3047 |
+
"loss": 0.0,
|
| 3048 |
+
"step": 433
|
| 3049 |
+
},
|
| 3050 |
+
{
|
| 3051 |
+
"epoch": 1.1563019620884603,
|
| 3052 |
+
"grad_norm": NaN,
|
| 3053 |
+
"learning_rate": 3.863757349903551e-05,
|
| 3054 |
+
"loss": 0.0,
|
| 3055 |
+
"step": 434
|
| 3056 |
+
},
|
| 3057 |
+
{
|
| 3058 |
+
"epoch": 1.1589624210176255,
|
| 3059 |
+
"grad_norm": NaN,
|
| 3060 |
+
"learning_rate": 3.843096039798293e-05,
|
| 3061 |
+
"loss": 0.0,
|
| 3062 |
+
"step": 435
|
| 3063 |
+
},
|
| 3064 |
+
{
|
| 3065 |
+
"epoch": 1.161622879946791,
|
| 3066 |
+
"grad_norm": NaN,
|
| 3067 |
+
"learning_rate": 3.822455580984613e-05,
|
| 3068 |
+
"loss": 0.0,
|
| 3069 |
+
"step": 436
|
| 3070 |
+
},
|
| 3071 |
+
{
|
| 3072 |
+
"epoch": 1.1642833388759561,
|
| 3073 |
+
"grad_norm": NaN,
|
| 3074 |
+
"learning_rate": 3.801836345472841e-05,
|
| 3075 |
+
"loss": 0.0,
|
| 3076 |
+
"step": 437
|
| 3077 |
+
},
|
| 3078 |
+
{
|
| 3079 |
+
"epoch": 1.1669437978051214,
|
| 3080 |
+
"grad_norm": NaN,
|
| 3081 |
+
"learning_rate": 3.781238704890793e-05,
|
| 3082 |
+
"loss": 0.0,
|
| 3083 |
+
"step": 438
|
| 3084 |
+
},
|
| 3085 |
+
{
|
| 3086 |
+
"epoch": 1.1696042567342866,
|
| 3087 |
+
"grad_norm": NaN,
|
| 3088 |
+
"learning_rate": 3.760663030477072e-05,
|
| 3089 |
+
"loss": 0.0,
|
| 3090 |
+
"step": 439
|
| 3091 |
+
},
|
| 3092 |
+
{
|
| 3093 |
+
"epoch": 1.172264715663452,
|
| 3094 |
+
"grad_norm": NaN,
|
| 3095 |
+
"learning_rate": 3.740109693074375e-05,
|
| 3096 |
+
"loss": 0.0,
|
| 3097 |
+
"step": 440
|
| 3098 |
+
},
|
| 3099 |
+
{
|
| 3100 |
+
"epoch": 1.1749251745926172,
|
| 3101 |
+
"grad_norm": NaN,
|
| 3102 |
+
"learning_rate": 3.7195790631228136e-05,
|
| 3103 |
+
"loss": 0.0,
|
| 3104 |
+
"step": 441
|
| 3105 |
+
},
|
| 3106 |
+
{
|
| 3107 |
+
"epoch": 1.1775856335217825,
|
| 3108 |
+
"grad_norm": NaN,
|
| 3109 |
+
"learning_rate": 3.6990715106532356e-05,
|
| 3110 |
+
"loss": 0.0,
|
| 3111 |
+
"step": 442
|
| 3112 |
+
},
|
| 3113 |
+
{
|
| 3114 |
+
"epoch": 1.1802460924509477,
|
| 3115 |
+
"grad_norm": NaN,
|
| 3116 |
+
"learning_rate": 3.6785874052805516e-05,
|
| 3117 |
+
"loss": 0.0,
|
| 3118 |
+
"step": 443
|
| 3119 |
+
},
|
| 3120 |
+
{
|
| 3121 |
+
"epoch": 1.1829065513801131,
|
| 3122 |
+
"grad_norm": NaN,
|
| 3123 |
+
"learning_rate": 3.658127116197079e-05,
|
| 3124 |
+
"loss": 0.0,
|
| 3125 |
+
"step": 444
|
| 3126 |
+
},
|
| 3127 |
+
{
|
| 3128 |
+
"epoch": 1.1855670103092784,
|
| 3129 |
+
"grad_norm": NaN,
|
| 3130 |
+
"learning_rate": 3.637691012165886e-05,
|
| 3131 |
+
"loss": 0.0,
|
| 3132 |
+
"step": 445
|
| 3133 |
+
},
|
| 3134 |
+
{
|
| 3135 |
+
"epoch": 1.1882274692384436,
|
| 3136 |
+
"grad_norm": NaN,
|
| 3137 |
+
"learning_rate": 3.6172794615141446e-05,
|
| 3138 |
+
"loss": 0.0,
|
| 3139 |
+
"step": 446
|
| 3140 |
+
},
|
| 3141 |
+
{
|
| 3142 |
+
"epoch": 1.190887928167609,
|
| 3143 |
+
"grad_norm": NaN,
|
| 3144 |
+
"learning_rate": 3.596892832126494e-05,
|
| 3145 |
+
"loss": 0.0,
|
| 3146 |
+
"step": 447
|
| 3147 |
+
},
|
| 3148 |
+
{
|
| 3149 |
+
"epoch": 1.1935483870967742,
|
| 3150 |
+
"grad_norm": NaN,
|
| 3151 |
+
"learning_rate": 3.5765314914384026e-05,
|
| 3152 |
+
"loss": 0.0,
|
| 3153 |
+
"step": 448
|
| 3154 |
+
},
|
| 3155 |
+
{
|
| 3156 |
+
"epoch": 1.1962088460259395,
|
| 3157 |
+
"grad_norm": NaN,
|
| 3158 |
+
"learning_rate": 3.556195806429559e-05,
|
| 3159 |
+
"loss": 0.0,
|
| 3160 |
+
"step": 449
|
| 3161 |
+
},
|
| 3162 |
+
{
|
| 3163 |
+
"epoch": 1.1988693049551047,
|
| 3164 |
+
"grad_norm": NaN,
|
| 3165 |
+
"learning_rate": 3.5358861436172485e-05,
|
| 3166 |
+
"loss": 0.0,
|
| 3167 |
+
"step": 450
|
| 3168 |
+
},
|
| 3169 |
+
{
|
| 3170 |
+
"epoch": 1.2015297638842701,
|
| 3171 |
+
"grad_norm": NaN,
|
| 3172 |
+
"learning_rate": 3.5156028690497454e-05,
|
| 3173 |
+
"loss": 0.0,
|
| 3174 |
+
"step": 451
|
| 3175 |
+
},
|
| 3176 |
+
{
|
| 3177 |
+
"epoch": 1.2041902228134354,
|
| 3178 |
+
"grad_norm": NaN,
|
| 3179 |
+
"learning_rate": 3.495346348299724e-05,
|
| 3180 |
+
"loss": 0.0,
|
| 3181 |
+
"step": 452
|
| 3182 |
+
},
|
| 3183 |
+
{
|
| 3184 |
+
"epoch": 1.2068506817426006,
|
| 3185 |
+
"grad_norm": NaN,
|
| 3186 |
+
"learning_rate": 3.4751169464576615e-05,
|
| 3187 |
+
"loss": 0.0,
|
| 3188 |
+
"step": 453
|
| 3189 |
+
},
|
| 3190 |
+
{
|
| 3191 |
+
"epoch": 1.2095111406717658,
|
| 3192 |
+
"grad_norm": NaN,
|
| 3193 |
+
"learning_rate": 3.4549150281252636e-05,
|
| 3194 |
+
"loss": 0.0,
|
| 3195 |
+
"step": 454
|
| 3196 |
+
},
|
| 3197 |
+
{
|
| 3198 |
+
"epoch": 1.2121715996009312,
|
| 3199 |
+
"grad_norm": NaN,
|
| 3200 |
+
"learning_rate": 3.434740957408889e-05,
|
| 3201 |
+
"loss": 0.0,
|
| 3202 |
+
"step": 455
|
| 3203 |
+
},
|
| 3204 |
+
{
|
| 3205 |
+
"epoch": 1.2148320585300965,
|
| 3206 |
+
"grad_norm": NaN,
|
| 3207 |
+
"learning_rate": 3.4145950979129915e-05,
|
| 3208 |
+
"loss": 0.0,
|
| 3209 |
+
"step": 456
|
| 3210 |
+
},
|
| 3211 |
+
{
|
| 3212 |
+
"epoch": 1.2174925174592617,
|
| 3213 |
+
"grad_norm": NaN,
|
| 3214 |
+
"learning_rate": 3.3944778127335595e-05,
|
| 3215 |
+
"loss": 0.0,
|
| 3216 |
+
"step": 457
|
| 3217 |
+
},
|
| 3218 |
+
{
|
| 3219 |
+
"epoch": 1.2201529763884271,
|
| 3220 |
+
"grad_norm": NaN,
|
| 3221 |
+
"learning_rate": 3.374389464451583e-05,
|
| 3222 |
+
"loss": 0.0,
|
| 3223 |
+
"step": 458
|
| 3224 |
+
},
|
| 3225 |
+
{
|
| 3226 |
+
"epoch": 1.2228134353175923,
|
| 3227 |
+
"grad_norm": NaN,
|
| 3228 |
+
"learning_rate": 3.3543304151265074e-05,
|
| 3229 |
+
"loss": 0.0,
|
| 3230 |
+
"step": 459
|
| 3231 |
+
},
|
| 3232 |
+
{
|
| 3233 |
+
"epoch": 1.2254738942467576,
|
| 3234 |
+
"grad_norm": NaN,
|
| 3235 |
+
"learning_rate": 3.334301026289712e-05,
|
| 3236 |
+
"loss": 0.0,
|
| 3237 |
+
"step": 460
|
| 3238 |
+
},
|
| 3239 |
+
{
|
| 3240 |
+
"epoch": 1.2281343531759228,
|
| 3241 |
+
"grad_norm": NaN,
|
| 3242 |
+
"learning_rate": 3.314301658937998e-05,
|
| 3243 |
+
"loss": 0.0,
|
| 3244 |
+
"step": 461
|
| 3245 |
+
},
|
| 3246 |
+
{
|
| 3247 |
+
"epoch": 1.230794812105088,
|
| 3248 |
+
"grad_norm": NaN,
|
| 3249 |
+
"learning_rate": 3.294332673527076e-05,
|
| 3250 |
+
"loss": 0.0,
|
| 3251 |
+
"step": 462
|
| 3252 |
+
},
|
| 3253 |
+
{
|
| 3254 |
+
"epoch": 1.2334552710342535,
|
| 3255 |
+
"grad_norm": NaN,
|
| 3256 |
+
"learning_rate": 3.274394429965073e-05,
|
| 3257 |
+
"loss": 0.0,
|
| 3258 |
+
"step": 463
|
| 3259 |
+
},
|
| 3260 |
+
{
|
| 3261 |
+
"epoch": 1.2361157299634187,
|
| 3262 |
+
"grad_norm": NaN,
|
| 3263 |
+
"learning_rate": 3.254487287606044e-05,
|
| 3264 |
+
"loss": 0.0,
|
| 3265 |
+
"step": 464
|
| 3266 |
+
},
|
| 3267 |
+
{
|
| 3268 |
+
"epoch": 1.238776188892584,
|
| 3269 |
+
"grad_norm": NaN,
|
| 3270 |
+
"learning_rate": 3.234611605243496e-05,
|
| 3271 |
+
"loss": 0.0,
|
| 3272 |
+
"step": 465
|
| 3273 |
+
},
|
| 3274 |
+
{
|
| 3275 |
+
"epoch": 1.2414366478217493,
|
| 3276 |
+
"grad_norm": NaN,
|
| 3277 |
+
"learning_rate": 3.214767741103923e-05,
|
| 3278 |
+
"loss": 0.0,
|
| 3279 |
+
"step": 466
|
| 3280 |
+
},
|
| 3281 |
+
{
|
| 3282 |
+
"epoch": 1.2440971067509146,
|
| 3283 |
+
"grad_norm": NaN,
|
| 3284 |
+
"learning_rate": 3.1949560528403486e-05,
|
| 3285 |
+
"loss": 0.0,
|
| 3286 |
+
"step": 467
|
| 3287 |
+
},
|
| 3288 |
+
{
|
| 3289 |
+
"epoch": 1.2467575656800798,
|
| 3290 |
+
"grad_norm": NaN,
|
| 3291 |
+
"learning_rate": 3.175176897525874e-05,
|
| 3292 |
+
"loss": 0.0,
|
| 3293 |
+
"step": 468
|
| 3294 |
+
},
|
| 3295 |
+
{
|
| 3296 |
+
"epoch": 1.249418024609245,
|
| 3297 |
+
"grad_norm": NaN,
|
| 3298 |
+
"learning_rate": 3.1554306316472536e-05,
|
| 3299 |
+
"loss": 0.0,
|
| 3300 |
+
"step": 469
|
| 3301 |
+
},
|
| 3302 |
+
{
|
| 3303 |
+
"epoch": 1.2520784835384104,
|
| 3304 |
+
"grad_norm": NaN,
|
| 3305 |
+
"learning_rate": 3.135717611098458e-05,
|
| 3306 |
+
"loss": 0.0,
|
| 3307 |
+
"step": 470
|
| 3308 |
+
},
|
| 3309 |
+
{
|
| 3310 |
+
"epoch": 1.2547389424675757,
|
| 3311 |
+
"grad_norm": NaN,
|
| 3312 |
+
"learning_rate": 3.1160381911742695e-05,
|
| 3313 |
+
"loss": 0.0,
|
| 3314 |
+
"step": 471
|
| 3315 |
+
},
|
| 3316 |
+
{
|
| 3317 |
+
"epoch": 1.257399401396741,
|
| 3318 |
+
"grad_norm": NaN,
|
| 3319 |
+
"learning_rate": 3.096392726563874e-05,
|
| 3320 |
+
"loss": 0.0,
|
| 3321 |
+
"step": 472
|
| 3322 |
+
},
|
| 3323 |
+
{
|
| 3324 |
+
"epoch": 1.2600598603259061,
|
| 3325 |
+
"grad_norm": NaN,
|
| 3326 |
+
"learning_rate": 3.076781571344465e-05,
|
| 3327 |
+
"loss": 0.0,
|
| 3328 |
+
"step": 473
|
| 3329 |
+
},
|
| 3330 |
+
{
|
| 3331 |
+
"epoch": 1.2627203192550716,
|
| 3332 |
+
"grad_norm": NaN,
|
| 3333 |
+
"learning_rate": 3.057205078974873e-05,
|
| 3334 |
+
"loss": 0.0,
|
| 3335 |
+
"step": 474
|
| 3336 |
+
},
|
| 3337 |
+
{
|
| 3338 |
+
"epoch": 1.2653807781842368,
|
| 3339 |
+
"grad_norm": NaN,
|
| 3340 |
+
"learning_rate": 3.0376636022891812e-05,
|
| 3341 |
+
"loss": 0.0,
|
| 3342 |
+
"step": 475
|
| 3343 |
+
},
|
| 3344 |
+
{
|
| 3345 |
+
"epoch": 1.268041237113402,
|
| 3346 |
+
"grad_norm": NaN,
|
| 3347 |
+
"learning_rate": 3.0181574934903743e-05,
|
| 3348 |
+
"loss": 0.0,
|
| 3349 |
+
"step": 476
|
| 3350 |
+
},
|
| 3351 |
+
{
|
| 3352 |
+
"epoch": 1.2707016960425674,
|
| 3353 |
+
"grad_norm": NaN,
|
| 3354 |
+
"learning_rate": 2.9986871041439908e-05,
|
| 3355 |
+
"loss": 0.0,
|
| 3356 |
+
"step": 477
|
| 3357 |
+
},
|
| 3358 |
+
{
|
| 3359 |
+
"epoch": 1.2733621549717327,
|
| 3360 |
+
"grad_norm": NaN,
|
| 3361 |
+
"learning_rate": 2.9792527851717805e-05,
|
| 3362 |
+
"loss": 0.0,
|
| 3363 |
+
"step": 478
|
| 3364 |
+
},
|
| 3365 |
+
{
|
| 3366 |
+
"epoch": 1.2760226139008979,
|
| 3367 |
+
"grad_norm": NaN,
|
| 3368 |
+
"learning_rate": 2.95985488684539e-05,
|
| 3369 |
+
"loss": 0.0,
|
| 3370 |
+
"step": 479
|
| 3371 |
+
},
|
| 3372 |
+
{
|
| 3373 |
+
"epoch": 1.278683072830063,
|
| 3374 |
+
"grad_norm": NaN,
|
| 3375 |
+
"learning_rate": 2.9404937587800375e-05,
|
| 3376 |
+
"loss": 0.0,
|
| 3377 |
+
"step": 480
|
| 3378 |
+
},
|
| 3379 |
+
{
|
| 3380 |
+
"epoch": 1.2813435317592285,
|
| 3381 |
+
"grad_norm": NaN,
|
| 3382 |
+
"learning_rate": 2.921169749928222e-05,
|
| 3383 |
+
"loss": 0.0,
|
| 3384 |
+
"step": 481
|
| 3385 |
+
},
|
| 3386 |
+
{
|
| 3387 |
+
"epoch": 1.2840039906883938,
|
| 3388 |
+
"grad_norm": NaN,
|
| 3389 |
+
"learning_rate": 2.9018832085734298e-05,
|
| 3390 |
+
"loss": 0.0,
|
| 3391 |
+
"step": 482
|
| 3392 |
+
},
|
| 3393 |
+
{
|
| 3394 |
+
"epoch": 1.286664449617559,
|
| 3395 |
+
"grad_norm": NaN,
|
| 3396 |
+
"learning_rate": 2.882634482323856e-05,
|
| 3397 |
+
"loss": 0.0,
|
| 3398 |
+
"step": 483
|
| 3399 |
+
},
|
| 3400 |
+
{
|
| 3401 |
+
"epoch": 1.2893249085467242,
|
| 3402 |
+
"grad_norm": NaN,
|
| 3403 |
+
"learning_rate": 2.8634239181061384e-05,
|
| 3404 |
+
"loss": 0.0,
|
| 3405 |
+
"step": 484
|
| 3406 |
+
},
|
| 3407 |
+
{
|
| 3408 |
+
"epoch": 1.2919853674758897,
|
| 3409 |
+
"grad_norm": NaN,
|
| 3410 |
+
"learning_rate": 2.8442518621591086e-05,
|
| 3411 |
+
"loss": 0.0,
|
| 3412 |
+
"step": 485
|
| 3413 |
+
},
|
| 3414 |
+
{
|
| 3415 |
+
"epoch": 1.2946458264050549,
|
| 3416 |
+
"grad_norm": NaN,
|
| 3417 |
+
"learning_rate": 2.8251186600275533e-05,
|
| 3418 |
+
"loss": 0.0,
|
| 3419 |
+
"step": 486
|
| 3420 |
+
},
|
| 3421 |
+
{
|
| 3422 |
+
"epoch": 1.29730628533422,
|
| 3423 |
+
"grad_norm": NaN,
|
| 3424 |
+
"learning_rate": 2.8060246565559778e-05,
|
| 3425 |
+
"loss": 0.0,
|
| 3426 |
+
"step": 487
|
| 3427 |
+
},
|
| 3428 |
+
{
|
| 3429 |
+
"epoch": 1.2999667442633855,
|
| 3430 |
+
"grad_norm": NaN,
|
| 3431 |
+
"learning_rate": 2.786970195882398e-05,
|
| 3432 |
+
"loss": 0.0,
|
| 3433 |
+
"step": 488
|
| 3434 |
+
},
|
| 3435 |
+
{
|
| 3436 |
+
"epoch": 1.3026272031925508,
|
| 3437 |
+
"grad_norm": NaN,
|
| 3438 |
+
"learning_rate": 2.7679556214321357e-05,
|
| 3439 |
+
"loss": 0.0,
|
| 3440 |
+
"step": 489
|
| 3441 |
+
},
|
| 3442 |
+
{
|
| 3443 |
+
"epoch": 1.305287662121716,
|
| 3444 |
+
"grad_norm": NaN,
|
| 3445 |
+
"learning_rate": 2.748981275911633e-05,
|
| 3446 |
+
"loss": 0.0,
|
| 3447 |
+
"step": 490
|
| 3448 |
+
},
|
| 3449 |
+
{
|
| 3450 |
+
"epoch": 1.3079481210508812,
|
| 3451 |
+
"grad_norm": NaN,
|
| 3452 |
+
"learning_rate": 2.7300475013022663e-05,
|
| 3453 |
+
"loss": 0.0,
|
| 3454 |
+
"step": 491
|
| 3455 |
+
},
|
| 3456 |
+
{
|
| 3457 |
+
"epoch": 1.3106085799800464,
|
| 3458 |
+
"grad_norm": NaN,
|
| 3459 |
+
"learning_rate": 2.7111546388541896e-05,
|
| 3460 |
+
"loss": 0.0,
|
| 3461 |
+
"step": 492
|
| 3462 |
+
},
|
| 3463 |
+
{
|
| 3464 |
+
"epoch": 1.3132690389092119,
|
| 3465 |
+
"grad_norm": NaN,
|
| 3466 |
+
"learning_rate": 2.6923030290801815e-05,
|
| 3467 |
+
"loss": 0.0,
|
| 3468 |
+
"step": 493
|
| 3469 |
+
},
|
| 3470 |
+
{
|
| 3471 |
+
"epoch": 1.315929497838377,
|
| 3472 |
+
"grad_norm": NaN,
|
| 3473 |
+
"learning_rate": 2.6734930117495134e-05,
|
| 3474 |
+
"loss": 0.0,
|
| 3475 |
+
"step": 494
|
| 3476 |
+
},
|
| 3477 |
+
{
|
| 3478 |
+
"epoch": 1.3185899567675423,
|
| 3479 |
+
"grad_norm": NaN,
|
| 3480 |
+
"learning_rate": 2.6547249258818164e-05,
|
| 3481 |
+
"loss": 0.0,
|
| 3482 |
+
"step": 495
|
| 3483 |
+
},
|
| 3484 |
+
{
|
| 3485 |
+
"epoch": 1.3212504156967078,
|
| 3486 |
+
"grad_norm": NaN,
|
| 3487 |
+
"learning_rate": 2.6359991097409765e-05,
|
| 3488 |
+
"loss": 0.0,
|
| 3489 |
+
"step": 496
|
| 3490 |
+
},
|
| 3491 |
+
{
|
| 3492 |
+
"epoch": 1.323910874625873,
|
| 3493 |
+
"grad_norm": NaN,
|
| 3494 |
+
"learning_rate": 2.6173159008290406e-05,
|
| 3495 |
+
"loss": 0.0,
|
| 3496 |
+
"step": 497
|
| 3497 |
+
},
|
| 3498 |
+
{
|
| 3499 |
+
"epoch": 1.3265713335550382,
|
| 3500 |
+
"grad_norm": NaN,
|
| 3501 |
+
"learning_rate": 2.598675635880129e-05,
|
| 3502 |
+
"loss": 0.0,
|
| 3503 |
+
"step": 498
|
| 3504 |
+
},
|
| 3505 |
+
{
|
| 3506 |
+
"epoch": 1.3292317924842036,
|
| 3507 |
+
"grad_norm": NaN,
|
| 3508 |
+
"learning_rate": 2.580078650854366e-05,
|
| 3509 |
+
"loss": 0.0,
|
| 3510 |
+
"step": 499
|
| 3511 |
+
},
|
| 3512 |
+
{
|
| 3513 |
+
"epoch": 1.3318922514133689,
|
| 3514 |
+
"grad_norm": NaN,
|
| 3515 |
+
"learning_rate": 2.5615252809318284e-05,
|
| 3516 |
+
"loss": 0.0,
|
| 3517 |
+
"step": 500
|
| 3518 |
+
},
|
| 3519 |
+
{
|
| 3520 |
+
"epoch": 1.334552710342534,
|
| 3521 |
+
"grad_norm": NaN,
|
| 3522 |
+
"learning_rate": 2.543015860506499e-05,
|
| 3523 |
+
"loss": 0.0,
|
| 3524 |
+
"step": 501
|
| 3525 |
+
},
|
| 3526 |
+
{
|
| 3527 |
+
"epoch": 1.3372131692716993,
|
| 3528 |
+
"grad_norm": NaN,
|
| 3529 |
+
"learning_rate": 2.524550723180249e-05,
|
| 3530 |
+
"loss": 0.0,
|
| 3531 |
+
"step": 502
|
| 3532 |
+
},
|
| 3533 |
+
{
|
| 3534 |
+
"epoch": 1.3398736282008645,
|
| 3535 |
+
"grad_norm": NaN,
|
| 3536 |
+
"learning_rate": 2.506130201756815e-05,
|
| 3537 |
+
"loss": 0.0,
|
| 3538 |
+
"step": 503
|
| 3539 |
+
},
|
| 3540 |
+
{
|
| 3541 |
+
"epoch": 1.34253408713003,
|
| 3542 |
+
"grad_norm": NaN,
|
| 3543 |
+
"learning_rate": 2.487754628235805e-05,
|
| 3544 |
+
"loss": 0.0,
|
| 3545 |
+
"step": 504
|
| 3546 |
+
},
|
| 3547 |
+
{
|
| 3548 |
+
"epoch": 1.3451945460591952,
|
| 3549 |
+
"grad_norm": NaN,
|
| 3550 |
+
"learning_rate": 2.469424333806718e-05,
|
| 3551 |
+
"loss": 0.0,
|
| 3552 |
+
"step": 505
|
| 3553 |
+
},
|
| 3554 |
+
{
|
| 3555 |
+
"epoch": 1.3478550049883604,
|
| 3556 |
+
"grad_norm": NaN,
|
| 3557 |
+
"learning_rate": 2.4511396488429727e-05,
|
| 3558 |
+
"loss": 0.0,
|
| 3559 |
+
"step": 506
|
| 3560 |
+
},
|
| 3561 |
+
{
|
| 3562 |
+
"epoch": 1.3505154639175259,
|
| 3563 |
+
"grad_norm": NaN,
|
| 3564 |
+
"learning_rate": 2.4329009028959482e-05,
|
| 3565 |
+
"loss": 0.0,
|
| 3566 |
+
"step": 507
|
| 3567 |
+
},
|
| 3568 |
+
{
|
| 3569 |
+
"epoch": 1.353175922846691,
|
| 3570 |
+
"grad_norm": NaN,
|
| 3571 |
+
"learning_rate": 2.4147084246890478e-05,
|
| 3572 |
+
"loss": 0.0,
|
| 3573 |
+
"step": 508
|
| 3574 |
+
},
|
| 3575 |
+
{
|
| 3576 |
+
"epoch": 1.3558363817758563,
|
| 3577 |
+
"grad_norm": NaN,
|
| 3578 |
+
"learning_rate": 2.396562542111781e-05,
|
| 3579 |
+
"loss": 0.0,
|
| 3580 |
+
"step": 509
|
| 3581 |
+
},
|
| 3582 |
+
{
|
| 3583 |
+
"epoch": 1.3584968407050217,
|
| 3584 |
+
"grad_norm": NaN,
|
| 3585 |
+
"learning_rate": 2.3784635822138424e-05,
|
| 3586 |
+
"loss": 0.0,
|
| 3587 |
+
"step": 510
|
| 3588 |
+
},
|
| 3589 |
+
{
|
| 3590 |
+
"epoch": 1.361157299634187,
|
| 3591 |
+
"grad_norm": NaN,
|
| 3592 |
+
"learning_rate": 2.3604118711992235e-05,
|
| 3593 |
+
"loss": 0.0,
|
| 3594 |
+
"step": 511
|
| 3595 |
+
},
|
| 3596 |
+
{
|
| 3597 |
+
"epoch": 1.3638177585633522,
|
| 3598 |
+
"grad_norm": NaN,
|
| 3599 |
+
"learning_rate": 2.342407734420331e-05,
|
| 3600 |
+
"loss": 0.0,
|
| 3601 |
+
"step": 512
|
| 3602 |
+
},
|
| 3603 |
+
{
|
| 3604 |
+
"epoch": 1.3664782174925174,
|
| 3605 |
+
"grad_norm": NaN,
|
| 3606 |
+
"learning_rate": 2.3244514963721276e-05,
|
| 3607 |
+
"loss": 0.0,
|
| 3608 |
+
"step": 513
|
| 3609 |
+
},
|
| 3610 |
+
{
|
| 3611 |
+
"epoch": 1.3691386764216826,
|
| 3612 |
+
"grad_norm": NaN,
|
| 3613 |
+
"learning_rate": 2.3065434806862807e-05,
|
| 3614 |
+
"loss": 0.0,
|
| 3615 |
+
"step": 514
|
| 3616 |
+
},
|
| 3617 |
+
{
|
| 3618 |
+
"epoch": 1.371799135350848,
|
| 3619 |
+
"grad_norm": NaN,
|
| 3620 |
+
"learning_rate": 2.288684010125325e-05,
|
| 3621 |
+
"loss": 0.0,
|
| 3622 |
+
"step": 515
|
| 3623 |
+
},
|
| 3624 |
+
{
|
| 3625 |
+
"epoch": 1.3744595942800133,
|
| 3626 |
+
"grad_norm": NaN,
|
| 3627 |
+
"learning_rate": 2.2708734065768488e-05,
|
| 3628 |
+
"loss": 0.0,
|
| 3629 |
+
"step": 516
|
| 3630 |
+
},
|
| 3631 |
+
{
|
| 3632 |
+
"epoch": 1.3771200532091785,
|
| 3633 |
+
"grad_norm": NaN,
|
| 3634 |
+
"learning_rate": 2.2531119910476993e-05,
|
| 3635 |
+
"loss": 0.0,
|
| 3636 |
+
"step": 517
|
| 3637 |
+
},
|
| 3638 |
+
{
|
| 3639 |
+
"epoch": 1.379780512138344,
|
| 3640 |
+
"grad_norm": NaN,
|
| 3641 |
+
"learning_rate": 2.2354000836581834e-05,
|
| 3642 |
+
"loss": 0.0,
|
| 3643 |
+
"step": 518
|
| 3644 |
+
},
|
| 3645 |
+
{
|
| 3646 |
+
"epoch": 1.3824409710675092,
|
| 3647 |
+
"grad_norm": NaN,
|
| 3648 |
+
"learning_rate": 2.2177380036363088e-05,
|
| 3649 |
+
"loss": 0.0,
|
| 3650 |
+
"step": 519
|
| 3651 |
+
},
|
| 3652 |
+
{
|
| 3653 |
+
"epoch": 1.3851014299966744,
|
| 3654 |
+
"grad_norm": NaN,
|
| 3655 |
+
"learning_rate": 2.2001260693120233e-05,
|
| 3656 |
+
"loss": 0.0,
|
| 3657 |
+
"step": 520
|
| 3658 |
+
},
|
| 3659 |
+
{
|
| 3660 |
+
"epoch": 1.3877618889258396,
|
| 3661 |
+
"grad_norm": NaN,
|
| 3662 |
+
"learning_rate": 2.182564598111486e-05,
|
| 3663 |
+
"loss": 0.0,
|
| 3664 |
+
"step": 521
|
| 3665 |
+
},
|
| 3666 |
+
{
|
| 3667 |
+
"epoch": 1.390422347855005,
|
| 3668 |
+
"grad_norm": NaN,
|
| 3669 |
+
"learning_rate": 2.1650539065513413e-05,
|
| 3670 |
+
"loss": 0.0,
|
| 3671 |
+
"step": 522
|
| 3672 |
+
},
|
| 3673 |
+
{
|
| 3674 |
+
"epoch": 1.3930828067841703,
|
| 3675 |
+
"grad_norm": NaN,
|
| 3676 |
+
"learning_rate": 2.1475943102330087e-05,
|
| 3677 |
+
"loss": 0.0,
|
| 3678 |
+
"step": 523
|
| 3679 |
+
},
|
| 3680 |
+
{
|
| 3681 |
+
"epoch": 1.3957432657133355,
|
| 3682 |
+
"grad_norm": NaN,
|
| 3683 |
+
"learning_rate": 2.130186123837002e-05,
|
| 3684 |
+
"loss": 0.0,
|
| 3685 |
+
"step": 524
|
| 3686 |
+
},
|
| 3687 |
+
{
|
| 3688 |
+
"epoch": 1.3984037246425007,
|
| 3689 |
+
"grad_norm": NaN,
|
| 3690 |
+
"learning_rate": 2.1128296611172593e-05,
|
| 3691 |
+
"loss": 0.0,
|
| 3692 |
+
"step": 525
|
| 3693 |
+
},
|
| 3694 |
+
{
|
| 3695 |
+
"epoch": 1.4010641835716662,
|
| 3696 |
+
"grad_norm": NaN,
|
| 3697 |
+
"learning_rate": 2.0955252348954807e-05,
|
| 3698 |
+
"loss": 0.0,
|
| 3699 |
+
"step": 526
|
| 3700 |
+
},
|
| 3701 |
+
{
|
| 3702 |
+
"epoch": 1.4037246425008314,
|
| 3703 |
+
"grad_norm": NaN,
|
| 3704 |
+
"learning_rate": 2.0782731570554947e-05,
|
| 3705 |
+
"loss": 0.0,
|
| 3706 |
+
"step": 527
|
| 3707 |
+
},
|
| 3708 |
+
{
|
| 3709 |
+
"epoch": 1.4063851014299966,
|
| 3710 |
+
"grad_norm": NaN,
|
| 3711 |
+
"learning_rate": 2.061073738537635e-05,
|
| 3712 |
+
"loss": 0.0,
|
| 3713 |
+
"step": 528
|
| 3714 |
+
},
|
| 3715 |
+
{
|
| 3716 |
+
"epoch": 1.409045560359162,
|
| 3717 |
+
"grad_norm": NaN,
|
| 3718 |
+
"learning_rate": 2.043927289333141e-05,
|
| 3719 |
+
"loss": 0.0,
|
| 3720 |
+
"step": 529
|
| 3721 |
+
},
|
| 3722 |
+
{
|
| 3723 |
+
"epoch": 1.4117060192883273,
|
| 3724 |
+
"grad_norm": NaN,
|
| 3725 |
+
"learning_rate": 2.026834118478567e-05,
|
| 3726 |
+
"loss": 0.0,
|
| 3727 |
+
"step": 530
|
| 3728 |
+
},
|
| 3729 |
+
{
|
| 3730 |
+
"epoch": 1.4143664782174925,
|
| 3731 |
+
"grad_norm": NaN,
|
| 3732 |
+
"learning_rate": 2.00979453405021e-05,
|
| 3733 |
+
"loss": 0.0,
|
| 3734 |
+
"step": 531
|
| 3735 |
+
},
|
| 3736 |
+
{
|
| 3737 |
+
"epoch": 1.4170269371466577,
|
| 3738 |
+
"grad_norm": NaN,
|
| 3739 |
+
"learning_rate": 1.992808843158559e-05,
|
| 3740 |
+
"loss": 0.0,
|
| 3741 |
+
"step": 532
|
| 3742 |
+
},
|
| 3743 |
+
{
|
| 3744 |
+
"epoch": 1.419687396075823,
|
| 3745 |
+
"grad_norm": NaN,
|
| 3746 |
+
"learning_rate": 1.9758773519427665e-05,
|
| 3747 |
+
"loss": 0.0,
|
| 3748 |
+
"step": 533
|
| 3749 |
+
},
|
| 3750 |
+
{
|
| 3751 |
+
"epoch": 1.4223478550049884,
|
| 3752 |
+
"grad_norm": NaN,
|
| 3753 |
+
"learning_rate": 1.959000365565119e-05,
|
| 3754 |
+
"loss": 0.0,
|
| 3755 |
+
"step": 534
|
| 3756 |
+
},
|
| 3757 |
+
{
|
| 3758 |
+
"epoch": 1.4250083139341536,
|
| 3759 |
+
"grad_norm": NaN,
|
| 3760 |
+
"learning_rate": 1.9421781882055444e-05,
|
| 3761 |
+
"loss": 0.0,
|
| 3762 |
+
"step": 535
|
| 3763 |
+
},
|
| 3764 |
+
{
|
| 3765 |
+
"epoch": 1.4276687728633188,
|
| 3766 |
+
"grad_norm": NaN,
|
| 3767 |
+
"learning_rate": 1.9254111230561283e-05,
|
| 3768 |
+
"loss": 0.0,
|
| 3769 |
+
"step": 536
|
| 3770 |
+
},
|
| 3771 |
+
{
|
| 3772 |
+
"epoch": 1.4303292317924843,
|
| 3773 |
+
"grad_norm": NaN,
|
| 3774 |
+
"learning_rate": 1.908699472315651e-05,
|
| 3775 |
+
"loss": 0.0,
|
| 3776 |
+
"step": 537
|
| 3777 |
+
},
|
| 3778 |
+
{
|
| 3779 |
+
"epoch": 1.4329896907216495,
|
| 3780 |
+
"grad_norm": NaN,
|
| 3781 |
+
"learning_rate": 1.8920435371841394e-05,
|
| 3782 |
+
"loss": 0.0,
|
| 3783 |
+
"step": 538
|
| 3784 |
+
},
|
| 3785 |
+
{
|
| 3786 |
+
"epoch": 1.4356501496508147,
|
| 3787 |
+
"grad_norm": NaN,
|
| 3788 |
+
"learning_rate": 1.875443617857436e-05,
|
| 3789 |
+
"loss": 0.0,
|
| 3790 |
+
"step": 539
|
| 3791 |
+
},
|
| 3792 |
+
{
|
| 3793 |
+
"epoch": 1.4383106085799802,
|
| 3794 |
+
"grad_norm": NaN,
|
| 3795 |
+
"learning_rate": 1.858900013521788e-05,
|
| 3796 |
+
"loss": 0.0,
|
| 3797 |
+
"step": 540
|
| 3798 |
+
},
|
| 3799 |
+
{
|
| 3800 |
+
"epoch": 1.4409710675091454,
|
| 3801 |
+
"grad_norm": NaN,
|
| 3802 |
+
"learning_rate": 1.8424130223484653e-05,
|
| 3803 |
+
"loss": 0.0,
|
| 3804 |
+
"step": 541
|
| 3805 |
+
},
|
| 3806 |
+
{
|
| 3807 |
+
"epoch": 1.4436315264383106,
|
| 3808 |
+
"grad_norm": NaN,
|
| 3809 |
+
"learning_rate": 1.8259829414883728e-05,
|
| 3810 |
+
"loss": 0.0,
|
| 3811 |
+
"step": 542
|
| 3812 |
+
},
|
| 3813 |
+
{
|
| 3814 |
+
"epoch": 1.4462919853674758,
|
| 3815 |
+
"grad_norm": NaN,
|
| 3816 |
+
"learning_rate": 1.809610067066701e-05,
|
| 3817 |
+
"loss": 0.0,
|
| 3818 |
+
"step": 543
|
| 3819 |
+
},
|
| 3820 |
+
{
|
| 3821 |
+
"epoch": 1.448952444296641,
|
| 3822 |
+
"grad_norm": NaN,
|
| 3823 |
+
"learning_rate": 1.793294694177588e-05,
|
| 3824 |
+
"loss": 0.0,
|
| 3825 |
+
"step": 544
|
| 3826 |
+
},
|
| 3827 |
+
{
|
| 3828 |
+
"epoch": 1.4516129032258065,
|
| 3829 |
+
"grad_norm": NaN,
|
| 3830 |
+
"learning_rate": 1.777037116878804e-05,
|
| 3831 |
+
"loss": 0.0,
|
| 3832 |
+
"step": 545
|
| 3833 |
+
},
|
| 3834 |
+
{
|
| 3835 |
+
"epoch": 1.4542733621549717,
|
| 3836 |
+
"grad_norm": NaN,
|
| 3837 |
+
"learning_rate": 1.76083762818645e-05,
|
| 3838 |
+
"loss": 0.0,
|
| 3839 |
+
"step": 546
|
| 3840 |
+
},
|
| 3841 |
+
{
|
| 3842 |
+
"epoch": 1.456933821084137,
|
| 3843 |
+
"grad_norm": NaN,
|
| 3844 |
+
"learning_rate": 1.7446965200696703e-05,
|
| 3845 |
+
"loss": 0.0,
|
| 3846 |
+
"step": 547
|
| 3847 |
+
},
|
| 3848 |
+
{
|
| 3849 |
+
"epoch": 1.4595942800133024,
|
| 3850 |
+
"grad_norm": NaN,
|
| 3851 |
+
"learning_rate": 1.7286140834453956e-05,
|
| 3852 |
+
"loss": 0.0,
|
| 3853 |
+
"step": 548
|
| 3854 |
+
},
|
| 3855 |
+
{
|
| 3856 |
+
"epoch": 1.4622547389424676,
|
| 3857 |
+
"grad_norm": NaN,
|
| 3858 |
+
"learning_rate": 1.7125906081731064e-05,
|
| 3859 |
+
"loss": 0.0,
|
| 3860 |
+
"step": 549
|
| 3861 |
+
},
|
| 3862 |
+
{
|
| 3863 |
+
"epoch": 1.4649151978716328,
|
| 3864 |
+
"grad_norm": NaN,
|
| 3865 |
+
"learning_rate": 1.6966263830495936e-05,
|
| 3866 |
+
"loss": 0.0,
|
| 3867 |
+
"step": 550
|
| 3868 |
+
},
|
| 3869 |
+
{
|
| 3870 |
+
"epoch": 1.4675756568007983,
|
| 3871 |
+
"grad_norm": NaN,
|
| 3872 |
+
"learning_rate": 1.6807216958037663e-05,
|
| 3873 |
+
"loss": 0.0,
|
| 3874 |
+
"step": 551
|
| 3875 |
+
},
|
| 3876 |
+
{
|
| 3877 |
+
"epoch": 1.4702361157299635,
|
| 3878 |
+
"grad_norm": NaN,
|
| 3879 |
+
"learning_rate": 1.6648768330914578e-05,
|
| 3880 |
+
"loss": 0.0,
|
| 3881 |
+
"step": 552
|
| 3882 |
+
},
|
| 3883 |
+
{
|
| 3884 |
+
"epoch": 1.4728965746591287,
|
| 3885 |
+
"grad_norm": NaN,
|
| 3886 |
+
"learning_rate": 1.649092080490266e-05,
|
| 3887 |
+
"loss": 0.0,
|
| 3888 |
+
"step": 553
|
| 3889 |
+
},
|
| 3890 |
+
{
|
| 3891 |
+
"epoch": 1.475557033588294,
|
| 3892 |
+
"grad_norm": NaN,
|
| 3893 |
+
"learning_rate": 1.633367722494404e-05,
|
| 3894 |
+
"loss": 0.0,
|
| 3895 |
+
"step": 554
|
| 3896 |
+
},
|
| 3897 |
+
{
|
| 3898 |
+
"epoch": 1.4782174925174592,
|
| 3899 |
+
"grad_norm": NaN,
|
| 3900 |
+
"learning_rate": 1.6177040425095662e-05,
|
| 3901 |
+
"loss": 0.0,
|
| 3902 |
+
"step": 555
|
| 3903 |
+
},
|
| 3904 |
+
{
|
| 3905 |
+
"epoch": 1.4808779514466246,
|
| 3906 |
+
"grad_norm": NaN,
|
| 3907 |
+
"learning_rate": 1.6021013228478278e-05,
|
| 3908 |
+
"loss": 0.0,
|
| 3909 |
+
"step": 556
|
| 3910 |
+
},
|
| 3911 |
+
{
|
| 3912 |
+
"epoch": 1.4835384103757898,
|
| 3913 |
+
"grad_norm": NaN,
|
| 3914 |
+
"learning_rate": 1.5865598447225577e-05,
|
| 3915 |
+
"loss": 0.0,
|
| 3916 |
+
"step": 557
|
| 3917 |
+
},
|
| 3918 |
+
{
|
| 3919 |
+
"epoch": 1.486198869304955,
|
| 3920 |
+
"grad_norm": NaN,
|
| 3921 |
+
"learning_rate": 1.5710798882433432e-05,
|
| 3922 |
+
"loss": 0.0,
|
| 3923 |
+
"step": 558
|
| 3924 |
+
},
|
| 3925 |
+
{
|
| 3926 |
+
"epoch": 1.4888593282341205,
|
| 3927 |
+
"grad_norm": NaN,
|
| 3928 |
+
"learning_rate": 1.5556617324109442e-05,
|
| 3929 |
+
"loss": 0.0,
|
| 3930 |
+
"step": 559
|
| 3931 |
+
},
|
| 3932 |
+
{
|
| 3933 |
+
"epoch": 1.4915197871632857,
|
| 3934 |
+
"grad_norm": NaN,
|
| 3935 |
+
"learning_rate": 1.5403056551122697e-05,
|
| 3936 |
+
"loss": 0.0,
|
| 3937 |
+
"step": 560
|
| 3938 |
+
},
|
| 3939 |
+
{
|
| 3940 |
+
"epoch": 1.494180246092451,
|
| 3941 |
+
"grad_norm": NaN,
|
| 3942 |
+
"learning_rate": 1.5250119331153606e-05,
|
| 3943 |
+
"loss": 0.0,
|
| 3944 |
+
"step": 561
|
| 3945 |
+
},
|
| 3946 |
+
{
|
| 3947 |
+
"epoch": 1.4968407050216161,
|
| 3948 |
+
"grad_norm": NaN,
|
| 3949 |
+
"learning_rate": 1.5097808420644115e-05,
|
| 3950 |
+
"loss": 0.0,
|
| 3951 |
+
"step": 562
|
| 3952 |
+
},
|
| 3953 |
+
{
|
| 3954 |
+
"epoch": 1.4995011639507816,
|
| 3955 |
+
"grad_norm": NaN,
|
| 3956 |
+
"learning_rate": 1.494612656474792e-05,
|
| 3957 |
+
"loss": 0.0,
|
| 3958 |
+
"step": 563
|
| 3959 |
+
},
|
| 3960 |
+
{
|
| 3961 |
+
"epoch": 1.5021616228799468,
|
| 3962 |
+
"grad_norm": NaN,
|
| 3963 |
+
"learning_rate": 1.4795076497281052e-05,
|
| 3964 |
+
"loss": 0.0,
|
| 3965 |
+
"step": 564
|
| 3966 |
+
},
|
| 3967 |
+
{
|
| 3968 |
+
"epoch": 1.504822081809112,
|
| 3969 |
+
"grad_norm": NaN,
|
| 3970 |
+
"learning_rate": 1.4644660940672627e-05,
|
| 3971 |
+
"loss": 0.0,
|
| 3972 |
+
"step": 565
|
| 3973 |
+
},
|
| 3974 |
+
{
|
| 3975 |
+
"epoch": 1.5074825407382773,
|
| 3976 |
+
"grad_norm": NaN,
|
| 3977 |
+
"learning_rate": 1.4494882605915717e-05,
|
| 3978 |
+
"loss": 0.0,
|
| 3979 |
+
"step": 566
|
| 3980 |
+
},
|
| 3981 |
+
{
|
| 3982 |
+
"epoch": 1.5101429996674427,
|
| 3983 |
+
"grad_norm": NaN,
|
| 3984 |
+
"learning_rate": 1.4345744192518506e-05,
|
| 3985 |
+
"loss": 0.0,
|
| 3986 |
+
"step": 567
|
| 3987 |
+
},
|
| 3988 |
+
{
|
| 3989 |
+
"epoch": 1.512803458596608,
|
| 3990 |
+
"grad_norm": NaN,
|
| 3991 |
+
"learning_rate": 1.4197248388455691e-05,
|
| 3992 |
+
"loss": 0.0,
|
| 3993 |
+
"step": 568
|
| 3994 |
+
},
|
| 3995 |
+
{
|
| 3996 |
+
"epoch": 1.5154639175257731,
|
| 3997 |
+
"grad_norm": NaN,
|
| 3998 |
+
"learning_rate": 1.4049397870119962e-05,
|
| 3999 |
+
"loss": 0.0,
|
| 4000 |
+
"step": 569
|
| 4001 |
+
},
|
| 4002 |
+
{
|
| 4003 |
+
"epoch": 1.5181243764549386,
|
| 4004 |
+
"grad_norm": NaN,
|
| 4005 |
+
"learning_rate": 1.3902195302273779e-05,
|
| 4006 |
+
"loss": 0.0,
|
| 4007 |
+
"step": 570
|
| 4008 |
+
},
|
| 4009 |
+
{
|
| 4010 |
+
"epoch": 1.5207848353841038,
|
| 4011 |
+
"grad_norm": NaN,
|
| 4012 |
+
"learning_rate": 1.3755643338001412e-05,
|
| 4013 |
+
"loss": 0.0,
|
| 4014 |
+
"step": 571
|
| 4015 |
+
},
|
| 4016 |
+
{
|
| 4017 |
+
"epoch": 1.523445294313269,
|
| 4018 |
+
"grad_norm": NaN,
|
| 4019 |
+
"learning_rate": 1.3609744618661013e-05,
|
| 4020 |
+
"loss": 0.0,
|
| 4021 |
+
"step": 572
|
| 4022 |
+
},
|
| 4023 |
+
{
|
| 4024 |
+
"epoch": 1.5261057532424345,
|
| 4025 |
+
"grad_norm": NaN,
|
| 4026 |
+
"learning_rate": 1.3464501773837124e-05,
|
| 4027 |
+
"loss": 0.0,
|
| 4028 |
+
"step": 573
|
| 4029 |
+
},
|
| 4030 |
+
{
|
| 4031 |
+
"epoch": 1.5287662121715995,
|
| 4032 |
+
"grad_norm": NaN,
|
| 4033 |
+
"learning_rate": 1.3319917421293182e-05,
|
| 4034 |
+
"loss": 0.0,
|
| 4035 |
+
"step": 574
|
| 4036 |
+
},
|
| 4037 |
+
{
|
| 4038 |
+
"epoch": 1.531426671100765,
|
| 4039 |
+
"grad_norm": NaN,
|
| 4040 |
+
"learning_rate": 1.3175994166924394e-05,
|
| 4041 |
+
"loss": 0.0,
|
| 4042 |
+
"step": 575
|
| 4043 |
+
},
|
| 4044 |
+
{
|
| 4045 |
+
"epoch": 1.5340871300299301,
|
| 4046 |
+
"grad_norm": NaN,
|
| 4047 |
+
"learning_rate": 1.3032734604710783e-05,
|
| 4048 |
+
"loss": 0.0,
|
| 4049 |
+
"step": 576
|
| 4050 |
+
},
|
| 4051 |
+
{
|
| 4052 |
+
"epoch": 1.5367475889590954,
|
| 4053 |
+
"grad_norm": NaN,
|
| 4054 |
+
"learning_rate": 1.289014131667039e-05,
|
| 4055 |
+
"loss": 0.0,
|
| 4056 |
+
"step": 577
|
| 4057 |
+
},
|
| 4058 |
+
{
|
| 4059 |
+
"epoch": 1.5394080478882608,
|
| 4060 |
+
"grad_norm": NaN,
|
| 4061 |
+
"learning_rate": 1.2748216872812745e-05,
|
| 4062 |
+
"loss": 0.0,
|
| 4063 |
+
"step": 578
|
| 4064 |
+
},
|
| 4065 |
+
{
|
| 4066 |
+
"epoch": 1.542068506817426,
|
| 4067 |
+
"grad_norm": NaN,
|
| 4068 |
+
"learning_rate": 1.26069638310926e-05,
|
| 4069 |
+
"loss": 0.0,
|
| 4070 |
+
"step": 579
|
| 4071 |
+
},
|
| 4072 |
+
{
|
| 4073 |
+
"epoch": 1.5447289657465912,
|
| 4074 |
+
"grad_norm": NaN,
|
| 4075 |
+
"learning_rate": 1.246638473736378e-05,
|
| 4076 |
+
"loss": 0.0,
|
| 4077 |
+
"step": 580
|
| 4078 |
+
},
|
| 4079 |
+
{
|
| 4080 |
+
"epoch": 1.5473894246757567,
|
| 4081 |
+
"grad_norm": NaN,
|
| 4082 |
+
"learning_rate": 1.2326482125333284e-05,
|
| 4083 |
+
"loss": 0.0,
|
| 4084 |
+
"step": 581
|
| 4085 |
+
},
|
| 4086 |
+
{
|
| 4087 |
+
"epoch": 1.5500498836049217,
|
| 4088 |
+
"grad_norm": NaN,
|
| 4089 |
+
"learning_rate": 1.2187258516515642e-05,
|
| 4090 |
+
"loss": 0.0,
|
| 4091 |
+
"step": 582
|
| 4092 |
+
},
|
| 4093 |
+
{
|
| 4094 |
+
"epoch": 1.5527103425340871,
|
| 4095 |
+
"grad_norm": NaN,
|
| 4096 |
+
"learning_rate": 1.2048716420187472e-05,
|
| 4097 |
+
"loss": 0.0,
|
| 4098 |
+
"step": 583
|
| 4099 |
+
},
|
| 4100 |
+
{
|
| 4101 |
+
"epoch": 1.5553708014632526,
|
| 4102 |
+
"grad_norm": NaN,
|
| 4103 |
+
"learning_rate": 1.191085833334228e-05,
|
| 4104 |
+
"loss": 0.0,
|
| 4105 |
+
"step": 584
|
| 4106 |
+
},
|
| 4107 |
+
{
|
| 4108 |
+
"epoch": 1.5580312603924176,
|
| 4109 |
+
"grad_norm": NaN,
|
| 4110 |
+
"learning_rate": 1.1773686740645384e-05,
|
| 4111 |
+
"loss": 0.0,
|
| 4112 |
+
"step": 585
|
| 4113 |
+
},
|
| 4114 |
+
{
|
| 4115 |
+
"epoch": 1.560691719321583,
|
| 4116 |
+
"grad_norm": NaN,
|
| 4117 |
+
"learning_rate": 1.1637204114389177e-05,
|
| 4118 |
+
"loss": 0.0,
|
| 4119 |
+
"step": 586
|
| 4120 |
+
},
|
| 4121 |
+
{
|
| 4122 |
+
"epoch": 1.5633521782507482,
|
| 4123 |
+
"grad_norm": NaN,
|
| 4124 |
+
"learning_rate": 1.1501412914448595e-05,
|
| 4125 |
+
"loss": 0.0,
|
| 4126 |
+
"step": 587
|
| 4127 |
+
},
|
| 4128 |
+
{
|
| 4129 |
+
"epoch": 1.5660126371799135,
|
| 4130 |
+
"grad_norm": NaN,
|
| 4131 |
+
"learning_rate": 1.1366315588236742e-05,
|
| 4132 |
+
"loss": 0.0,
|
| 4133 |
+
"step": 588
|
| 4134 |
+
},
|
| 4135 |
+
{
|
| 4136 |
+
"epoch": 1.568673096109079,
|
| 4137 |
+
"grad_norm": NaN,
|
| 4138 |
+
"learning_rate": 1.1231914570660774e-05,
|
| 4139 |
+
"loss": 0.0,
|
| 4140 |
+
"step": 589
|
| 4141 |
+
},
|
| 4142 |
+
{
|
| 4143 |
+
"epoch": 1.5713335550382441,
|
| 4144 |
+
"grad_norm": NaN,
|
| 4145 |
+
"learning_rate": 1.1098212284078036e-05,
|
| 4146 |
+
"loss": 0.0,
|
| 4147 |
+
"step": 590
|
| 4148 |
+
},
|
| 4149 |
+
{
|
| 4150 |
+
"epoch": 1.5739940139674093,
|
| 4151 |
+
"grad_norm": NaN,
|
| 4152 |
+
"learning_rate": 1.0965211138252373e-05,
|
| 4153 |
+
"loss": 0.0,
|
| 4154 |
+
"step": 591
|
| 4155 |
+
},
|
| 4156 |
+
{
|
| 4157 |
+
"epoch": 1.5766544728965748,
|
| 4158 |
+
"grad_norm": NaN,
|
| 4159 |
+
"learning_rate": 1.0832913530310784e-05,
|
| 4160 |
+
"loss": 0.0,
|
| 4161 |
+
"step": 592
|
| 4162 |
+
},
|
| 4163 |
+
{
|
| 4164 |
+
"epoch": 1.5793149318257398,
|
| 4165 |
+
"grad_norm": NaN,
|
| 4166 |
+
"learning_rate": 1.0701321844700096e-05,
|
| 4167 |
+
"loss": 0.0,
|
| 4168 |
+
"step": 593
|
| 4169 |
+
},
|
| 4170 |
+
{
|
| 4171 |
+
"epoch": 1.5819753907549052,
|
| 4172 |
+
"grad_norm": NaN,
|
| 4173 |
+
"learning_rate": 1.0570438453144043e-05,
|
| 4174 |
+
"loss": 0.0,
|
| 4175 |
+
"step": 594
|
| 4176 |
+
},
|
| 4177 |
+
{
|
| 4178 |
+
"epoch": 1.5846358496840705,
|
| 4179 |
+
"grad_norm": NaN,
|
| 4180 |
+
"learning_rate": 1.0440265714600572e-05,
|
| 4181 |
+
"loss": 0.0,
|
| 4182 |
+
"step": 595
|
| 4183 |
+
},
|
| 4184 |
+
{
|
| 4185 |
+
"epoch": 1.5872963086132357,
|
| 4186 |
+
"grad_norm": NaN,
|
| 4187 |
+
"learning_rate": 1.0310805975219256e-05,
|
| 4188 |
+
"loss": 0.0,
|
| 4189 |
+
"step": 596
|
| 4190 |
+
},
|
| 4191 |
+
{
|
| 4192 |
+
"epoch": 1.5899567675424011,
|
| 4193 |
+
"grad_norm": NaN,
|
| 4194 |
+
"learning_rate": 1.0182061568299018e-05,
|
| 4195 |
+
"loss": 0.0,
|
| 4196 |
+
"step": 597
|
| 4197 |
+
},
|
| 4198 |
+
{
|
| 4199 |
+
"epoch": 1.5926172264715663,
|
| 4200 |
+
"grad_norm": NaN,
|
| 4201 |
+
"learning_rate": 1.0054034814246094e-05,
|
| 4202 |
+
"loss": 0.0,
|
| 4203 |
+
"step": 598
|
| 4204 |
+
},
|
| 4205 |
+
{
|
| 4206 |
+
"epoch": 1.5952776854007316,
|
| 4207 |
+
"grad_norm": NaN,
|
| 4208 |
+
"learning_rate": 9.926728020532195e-06,
|
| 4209 |
+
"loss": 0.0,
|
| 4210 |
+
"step": 599
|
| 4211 |
+
},
|
| 4212 |
+
{
|
| 4213 |
+
"epoch": 1.597938144329897,
|
| 4214 |
+
"grad_norm": NaN,
|
| 4215 |
+
"learning_rate": 9.800143481652979e-06,
|
| 4216 |
+
"loss": 0.0,
|
| 4217 |
+
"step": 600
|
| 4218 |
+
},
|
| 4219 |
+
{
|
| 4220 |
+
"epoch": 1.6005986032590622,
|
| 4221 |
+
"grad_norm": NaN,
|
| 4222 |
+
"learning_rate": 9.674283479086599e-06,
|
| 4223 |
+
"loss": 0.0,
|
| 4224 |
+
"step": 601
|
| 4225 |
+
},
|
| 4226 |
+
{
|
| 4227 |
+
"epoch": 1.6032590621882274,
|
| 4228 |
+
"grad_norm": NaN,
|
| 4229 |
+
"learning_rate": 9.549150281252633e-06,
|
| 4230 |
+
"loss": 0.0,
|
| 4231 |
+
"step": 602
|
| 4232 |
+
},
|
| 4233 |
+
{
|
| 4234 |
+
"epoch": 1.605919521117393,
|
| 4235 |
+
"grad_norm": NaN,
|
| 4236 |
+
"learning_rate": 9.42474614347123e-06,
|
| 4237 |
+
"loss": 0.0,
|
| 4238 |
+
"step": 603
|
| 4239 |
+
},
|
| 4240 |
+
{
|
| 4241 |
+
"epoch": 1.608579980046558,
|
| 4242 |
+
"grad_norm": NaN,
|
| 4243 |
+
"learning_rate": 9.30107330792243e-06,
|
| 4244 |
+
"loss": 0.0,
|
| 4245 |
+
"step": 604
|
| 4246 |
+
},
|
| 4247 |
+
{
|
| 4248 |
+
"epoch": 1.6112404389757233,
|
| 4249 |
+
"grad_norm": NaN,
|
| 4250 |
+
"learning_rate": 9.17813400360572e-06,
|
| 4251 |
+
"loss": 0.0,
|
| 4252 |
+
"step": 605
|
| 4253 |
+
},
|
| 4254 |
+
{
|
| 4255 |
+
"epoch": 1.6139008979048886,
|
| 4256 |
+
"grad_norm": NaN,
|
| 4257 |
+
"learning_rate": 9.055930446299915e-06,
|
| 4258 |
+
"loss": 0.0,
|
| 4259 |
+
"step": 606
|
| 4260 |
+
},
|
| 4261 |
+
{
|
| 4262 |
+
"epoch": 1.6165613568340538,
|
| 4263 |
+
"grad_norm": NaN,
|
| 4264 |
+
"learning_rate": 8.934464838523182e-06,
|
| 4265 |
+
"loss": 0.0,
|
| 4266 |
+
"step": 607
|
| 4267 |
+
},
|
| 4268 |
+
{
|
| 4269 |
+
"epoch": 1.6192218157632192,
|
| 4270 |
+
"grad_norm": NaN,
|
| 4271 |
+
"learning_rate": 8.813739369493395e-06,
|
| 4272 |
+
"loss": 0.0,
|
| 4273 |
+
"step": 608
|
| 4274 |
+
},
|
| 4275 |
+
{
|
| 4276 |
+
"epoch": 1.6218822746923844,
|
| 4277 |
+
"grad_norm": NaN,
|
| 4278 |
+
"learning_rate": 8.693756215088617e-06,
|
| 4279 |
+
"loss": 0.0,
|
| 4280 |
+
"step": 609
|
| 4281 |
+
},
|
| 4282 |
+
{
|
| 4283 |
+
"epoch": 1.6245427336215497,
|
| 4284 |
+
"grad_norm": NaN,
|
| 4285 |
+
"learning_rate": 8.574517537807897e-06,
|
| 4286 |
+
"loss": 0.0,
|
| 4287 |
+
"step": 610
|
| 4288 |
+
},
|
| 4289 |
+
{
|
| 4290 |
+
"epoch": 1.627203192550715,
|
| 4291 |
+
"grad_norm": NaN,
|
| 4292 |
+
"learning_rate": 8.45602548673235e-06,
|
| 4293 |
+
"loss": 0.0,
|
| 4294 |
+
"step": 611
|
| 4295 |
+
},
|
| 4296 |
+
{
|
| 4297 |
+
"epoch": 1.6298636514798803,
|
| 4298 |
+
"grad_norm": NaN,
|
| 4299 |
+
"learning_rate": 8.338282197486364e-06,
|
| 4300 |
+
"loss": 0.0,
|
| 4301 |
+
"step": 612
|
| 4302 |
+
},
|
| 4303 |
+
{
|
| 4304 |
+
"epoch": 1.6325241104090455,
|
| 4305 |
+
"grad_norm": NaN,
|
| 4306 |
+
"learning_rate": 8.221289792199116e-06,
|
| 4307 |
+
"loss": 0.0,
|
| 4308 |
+
"step": 613
|
| 4309 |
+
},
|
| 4310 |
+
{
|
| 4311 |
+
"epoch": 1.635184569338211,
|
| 4312 |
+
"grad_norm": NaN,
|
| 4313 |
+
"learning_rate": 8.105050379466333e-06,
|
| 4314 |
+
"loss": 0.0,
|
| 4315 |
+
"step": 614
|
| 4316 |
+
},
|
| 4317 |
+
{
|
| 4318 |
+
"epoch": 1.637845028267376,
|
| 4319 |
+
"grad_norm": NaN,
|
| 4320 |
+
"learning_rate": 7.989566054312287e-06,
|
| 4321 |
+
"loss": 0.0,
|
| 4322 |
+
"step": 615
|
| 4323 |
+
},
|
| 4324 |
+
{
|
| 4325 |
+
"epoch": 1.6405054871965414,
|
| 4326 |
+
"grad_norm": NaN,
|
| 4327 |
+
"learning_rate": 7.87483889815207e-06,
|
| 4328 |
+
"loss": 0.0,
|
| 4329 |
+
"step": 616
|
| 4330 |
+
},
|
| 4331 |
+
{
|
| 4332 |
+
"epoch": 1.6431659461257067,
|
| 4333 |
+
"grad_norm": NaN,
|
| 4334 |
+
"learning_rate": 7.760870978754014e-06,
|
| 4335 |
+
"loss": 0.0,
|
| 4336 |
+
"step": 617
|
| 4337 |
+
},
|
| 4338 |
+
{
|
| 4339 |
+
"epoch": 1.6458264050548719,
|
| 4340 |
+
"grad_norm": NaN,
|
| 4341 |
+
"learning_rate": 7.64766435020246e-06,
|
| 4342 |
+
"loss": 0.0,
|
| 4343 |
+
"step": 618
|
| 4344 |
+
},
|
| 4345 |
+
{
|
| 4346 |
+
"epoch": 1.6484868639840373,
|
| 4347 |
+
"grad_norm": NaN,
|
| 4348 |
+
"learning_rate": 7.535221052860747e-06,
|
| 4349 |
+
"loss": 0.0,
|
| 4350 |
+
"step": 619
|
| 4351 |
+
},
|
| 4352 |
+
{
|
| 4353 |
+
"epoch": 1.6511473229132025,
|
| 4354 |
+
"grad_norm": NaN,
|
| 4355 |
+
"learning_rate": 7.423543113334436e-06,
|
| 4356 |
+
"loss": 0.0,
|
| 4357 |
+
"step": 620
|
| 4358 |
+
},
|
| 4359 |
+
{
|
| 4360 |
+
"epoch": 1.6538077818423678,
|
| 4361 |
+
"grad_norm": NaN,
|
| 4362 |
+
"learning_rate": 7.312632544434738e-06,
|
| 4363 |
+
"loss": 0.0,
|
| 4364 |
+
"step": 621
|
| 4365 |
+
},
|
| 4366 |
+
{
|
| 4367 |
+
"epoch": 1.6564682407715332,
|
| 4368 |
+
"grad_norm": NaN,
|
| 4369 |
+
"learning_rate": 7.2024913451422875e-06,
|
| 4370 |
+
"loss": 0.0,
|
| 4371 |
+
"step": 622
|
| 4372 |
+
},
|
| 4373 |
+
{
|
| 4374 |
+
"epoch": 1.6591286997006982,
|
| 4375 |
+
"grad_norm": NaN,
|
| 4376 |
+
"learning_rate": 7.093121500571082e-06,
|
| 4377 |
+
"loss": 0.0,
|
| 4378 |
+
"step": 623
|
| 4379 |
+
},
|
| 4380 |
+
{
|
| 4381 |
+
"epoch": 1.6617891586298636,
|
| 4382 |
+
"grad_norm": NaN,
|
| 4383 |
+
"learning_rate": 6.984524981932755e-06,
|
| 4384 |
+
"loss": 0.0,
|
| 4385 |
+
"step": 624
|
| 4386 |
+
},
|
| 4387 |
+
{
|
| 4388 |
+
"epoch": 1.664449617559029,
|
| 4389 |
+
"grad_norm": NaN,
|
| 4390 |
+
"learning_rate": 6.876703746500984e-06,
|
| 4391 |
+
"loss": 0.0,
|
| 4392 |
+
"step": 625
|
| 4393 |
+
},
|
| 4394 |
+
{
|
| 4395 |
+
"epoch": 1.667110076488194,
|
| 4396 |
+
"grad_norm": NaN,
|
| 4397 |
+
"learning_rate": 6.769659737576229e-06,
|
| 4398 |
+
"loss": 0.0,
|
| 4399 |
+
"step": 626
|
| 4400 |
+
},
|
| 4401 |
+
{
|
| 4402 |
+
"epoch": 1.6697705354173595,
|
| 4403 |
+
"grad_norm": NaN,
|
| 4404 |
+
"learning_rate": 6.663394884450752e-06,
|
| 4405 |
+
"loss": 0.0,
|
| 4406 |
+
"step": 627
|
| 4407 |
+
},
|
| 4408 |
+
{
|
| 4409 |
+
"epoch": 1.6724309943465248,
|
| 4410 |
+
"grad_norm": NaN,
|
| 4411 |
+
"learning_rate": 6.557911102373809e-06,
|
| 4412 |
+
"loss": 0.0,
|
| 4413 |
+
"step": 628
|
| 4414 |
+
},
|
| 4415 |
+
{
|
| 4416 |
+
"epoch": 1.67509145327569,
|
| 4417 |
+
"grad_norm": NaN,
|
| 4418 |
+
"learning_rate": 6.453210292517114e-06,
|
| 4419 |
+
"loss": 0.0,
|
| 4420 |
+
"step": 629
|
| 4421 |
+
},
|
| 4422 |
+
{
|
| 4423 |
+
"epoch": 1.6777519122048554,
|
| 4424 |
+
"grad_norm": NaN,
|
| 4425 |
+
"learning_rate": 6.349294341940593e-06,
|
| 4426 |
+
"loss": 0.0,
|
| 4427 |
+
"step": 630
|
| 4428 |
+
},
|
| 4429 |
+
{
|
| 4430 |
+
"epoch": 1.6804123711340206,
|
| 4431 |
+
"grad_norm": NaN,
|
| 4432 |
+
"learning_rate": 6.246165123558401e-06,
|
| 4433 |
+
"loss": 0.0,
|
| 4434 |
+
"step": 631
|
| 4435 |
+
},
|
| 4436 |
+
{
|
| 4437 |
+
"epoch": 1.6830728300631859,
|
| 4438 |
+
"grad_norm": NaN,
|
| 4439 |
+
"learning_rate": 6.143824496105121e-06,
|
| 4440 |
+
"loss": 0.0,
|
| 4441 |
+
"step": 632
|
| 4442 |
+
},
|
| 4443 |
+
{
|
| 4444 |
+
"epoch": 1.6857332889923513,
|
| 4445 |
+
"grad_norm": NaN,
|
| 4446 |
+
"learning_rate": 6.04227430410228e-06,
|
| 4447 |
+
"loss": 0.0,
|
| 4448 |
+
"step": 633
|
| 4449 |
+
},
|
| 4450 |
+
{
|
| 4451 |
+
"epoch": 1.6883937479215163,
|
| 4452 |
+
"grad_norm": NaN,
|
| 4453 |
+
"learning_rate": 5.941516377825102e-06,
|
| 4454 |
+
"loss": 0.0,
|
| 4455 |
+
"step": 634
|
| 4456 |
+
},
|
| 4457 |
+
{
|
| 4458 |
+
"epoch": 1.6910542068506818,
|
| 4459 |
+
"grad_norm": NaN,
|
| 4460 |
+
"learning_rate": 5.841552533269534e-06,
|
| 4461 |
+
"loss": 0.0,
|
| 4462 |
+
"step": 635
|
| 4463 |
+
},
|
| 4464 |
+
{
|
| 4465 |
+
"epoch": 1.693714665779847,
|
| 4466 |
+
"grad_norm": NaN,
|
| 4467 |
+
"learning_rate": 5.742384572119519e-06,
|
| 4468 |
+
"loss": 0.0,
|
| 4469 |
+
"step": 636
|
| 4470 |
+
},
|
| 4471 |
+
{
|
| 4472 |
+
"epoch": 1.6963751247090122,
|
| 4473 |
+
"grad_norm": NaN,
|
| 4474 |
+
"learning_rate": 5.6440142817144826e-06,
|
| 4475 |
+
"loss": 0.0,
|
| 4476 |
+
"step": 637
|
| 4477 |
+
},
|
| 4478 |
+
{
|
| 4479 |
+
"epoch": 1.6990355836381776,
|
| 4480 |
+
"grad_norm": NaN,
|
| 4481 |
+
"learning_rate": 5.546443435017146e-06,
|
| 4482 |
+
"loss": 0.0,
|
| 4483 |
+
"step": 638
|
| 4484 |
+
},
|
| 4485 |
+
{
|
| 4486 |
+
"epoch": 1.7016960425673429,
|
| 4487 |
+
"grad_norm": NaN,
|
| 4488 |
+
"learning_rate": 5.449673790581611e-06,
|
| 4489 |
+
"loss": 0.0,
|
| 4490 |
+
"step": 639
|
| 4491 |
+
},
|
| 4492 |
+
{
|
| 4493 |
+
"epoch": 1.704356501496508,
|
| 4494 |
+
"grad_norm": NaN,
|
| 4495 |
+
"learning_rate": 5.353707092521582e-06,
|
| 4496 |
+
"loss": 0.0,
|
| 4497 |
+
"step": 640
|
| 4498 |
+
},
|
| 4499 |
+
{
|
| 4500 |
+
"epoch": 1.7070169604256735,
|
| 4501 |
+
"grad_norm": NaN,
|
| 4502 |
+
"learning_rate": 5.258545070478999e-06,
|
| 4503 |
+
"loss": 0.0,
|
| 4504 |
+
"step": 641
|
| 4505 |
+
},
|
| 4506 |
+
{
|
| 4507 |
+
"epoch": 1.7096774193548387,
|
| 4508 |
+
"grad_norm": NaN,
|
| 4509 |
+
"learning_rate": 5.1641894395928205e-06,
|
| 4510 |
+
"loss": 0.0,
|
| 4511 |
+
"step": 642
|
| 4512 |
+
},
|
| 4513 |
+
{
|
| 4514 |
+
"epoch": 1.712337878284004,
|
| 4515 |
+
"grad_norm": NaN,
|
| 4516 |
+
"learning_rate": 5.0706419004681485e-06,
|
| 4517 |
+
"loss": 0.0,
|
| 4518 |
+
"step": 643
|
| 4519 |
+
},
|
| 4520 |
+
{
|
| 4521 |
+
"epoch": 1.7149983372131694,
|
| 4522 |
+
"grad_norm": NaN,
|
| 4523 |
+
"learning_rate": 4.977904139145578e-06,
|
| 4524 |
+
"loss": 0.0,
|
| 4525 |
+
"step": 644
|
| 4526 |
+
},
|
| 4527 |
+
{
|
| 4528 |
+
"epoch": 1.7176587961423344,
|
| 4529 |
+
"grad_norm": NaN,
|
| 4530 |
+
"learning_rate": 4.885977827070748e-06,
|
| 4531 |
+
"loss": 0.0,
|
| 4532 |
+
"step": 645
|
| 4533 |
+
},
|
| 4534 |
+
{
|
| 4535 |
+
"epoch": 1.7203192550714999,
|
| 4536 |
+
"grad_norm": NaN,
|
| 4537 |
+
"learning_rate": 4.794864621064266e-06,
|
| 4538 |
+
"loss": 0.0,
|
| 4539 |
+
"step": 646
|
| 4540 |
+
},
|
| 4541 |
+
{
|
| 4542 |
+
"epoch": 1.722979714000665,
|
| 4543 |
+
"grad_norm": NaN,
|
| 4544 |
+
"learning_rate": 4.704566163291879e-06,
|
| 4545 |
+
"loss": 0.0,
|
| 4546 |
+
"step": 647
|
| 4547 |
+
},
|
| 4548 |
+
{
|
| 4549 |
+
"epoch": 1.7256401729298303,
|
| 4550 |
+
"grad_norm": NaN,
|
| 4551 |
+
"learning_rate": 4.6150840812347995e-06,
|
| 4552 |
+
"loss": 0.0,
|
| 4553 |
+
"step": 648
|
| 4554 |
+
},
|
| 4555 |
+
{
|
| 4556 |
+
"epoch": 1.7283006318589957,
|
| 4557 |
+
"grad_norm": NaN,
|
| 4558 |
+
"learning_rate": 4.526419987660418e-06,
|
| 4559 |
+
"loss": 0.0,
|
| 4560 |
+
"step": 649
|
| 4561 |
+
},
|
| 4562 |
+
{
|
| 4563 |
+
"epoch": 1.730961090788161,
|
| 4564 |
+
"grad_norm": NaN,
|
| 4565 |
+
"learning_rate": 4.43857548059321e-06,
|
| 4566 |
+
"loss": 0.0,
|
| 4567 |
+
"step": 650
|
| 4568 |
+
},
|
| 4569 |
+
{
|
| 4570 |
+
"epoch": 1.7336215497173262,
|
| 4571 |
+
"grad_norm": NaN,
|
| 4572 |
+
"learning_rate": 4.351552143286014e-06,
|
| 4573 |
+
"loss": 0.0,
|
| 4574 |
+
"step": 651
|
| 4575 |
+
},
|
| 4576 |
+
{
|
| 4577 |
+
"epoch": 1.7362820086464916,
|
| 4578 |
+
"grad_norm": NaN,
|
| 4579 |
+
"learning_rate": 4.265351544191365e-06,
|
| 4580 |
+
"loss": 0.0,
|
| 4581 |
+
"step": 652
|
| 4582 |
+
},
|
| 4583 |
+
{
|
| 4584 |
+
"epoch": 1.7389424675756568,
|
| 4585 |
+
"grad_norm": NaN,
|
| 4586 |
+
"learning_rate": 4.179975236933331e-06,
|
| 4587 |
+
"loss": 0.0,
|
| 4588 |
+
"step": 653
|
| 4589 |
+
},
|
| 4590 |
+
{
|
| 4591 |
+
"epoch": 1.741602926504822,
|
| 4592 |
+
"grad_norm": NaN,
|
| 4593 |
+
"learning_rate": 4.095424760279454e-06,
|
| 4594 |
+
"loss": 0.0,
|
| 4595 |
+
"step": 654
|
| 4596 |
+
},
|
| 4597 |
+
{
|
| 4598 |
+
"epoch": 1.7442633854339875,
|
| 4599 |
+
"grad_norm": NaN,
|
| 4600 |
+
"learning_rate": 4.011701638113063e-06,
|
| 4601 |
+
"loss": 0.0,
|
| 4602 |
+
"step": 655
|
| 4603 |
+
},
|
| 4604 |
+
{
|
| 4605 |
+
"epoch": 1.7469238443631525,
|
| 4606 |
+
"grad_norm": NaN,
|
| 4607 |
+
"learning_rate": 3.928807379405764e-06,
|
| 4608 |
+
"loss": 0.0,
|
| 4609 |
+
"step": 656
|
| 4610 |
+
},
|
| 4611 |
+
{
|
| 4612 |
+
"epoch": 1.749584303292318,
|
| 4613 |
+
"grad_norm": NaN,
|
| 4614 |
+
"learning_rate": 3.8467434781902626e-06,
|
| 4615 |
+
"loss": 0.0,
|
| 4616 |
+
"step": 657
|
| 4617 |
+
},
|
| 4618 |
+
{
|
| 4619 |
+
"epoch": 1.7522447622214832,
|
| 4620 |
+
"grad_norm": NaN,
|
| 4621 |
+
"learning_rate": 3.765511413533429e-06,
|
| 4622 |
+
"loss": 0.0,
|
| 4623 |
+
"step": 658
|
| 4624 |
+
},
|
| 4625 |
+
{
|
| 4626 |
+
"epoch": 1.7549052211506484,
|
| 4627 |
+
"grad_norm": NaN,
|
| 4628 |
+
"learning_rate": 3.685112649509681e-06,
|
| 4629 |
+
"loss": 0.0,
|
| 4630 |
+
"step": 659
|
| 4631 |
+
},
|
| 4632 |
+
{
|
| 4633 |
+
"epoch": 1.7575656800798138,
|
| 4634 |
+
"grad_norm": NaN,
|
| 4635 |
+
"learning_rate": 3.605548635174533e-06,
|
| 4636 |
+
"loss": 0.0,
|
| 4637 |
+
"step": 660
|
| 4638 |
+
},
|
| 4639 |
+
{
|
| 4640 |
+
"epoch": 1.760226139008979,
|
| 4641 |
+
"grad_norm": NaN,
|
| 4642 |
+
"learning_rate": 3.5268208045385164e-06,
|
| 4643 |
+
"loss": 0.0,
|
| 4644 |
+
"step": 661
|
| 4645 |
+
},
|
| 4646 |
+
{
|
| 4647 |
+
"epoch": 1.7628865979381443,
|
| 4648 |
+
"grad_norm": NaN,
|
| 4649 |
+
"learning_rate": 3.448930576541309e-06,
|
| 4650 |
+
"loss": 0.0,
|
| 4651 |
+
"step": 662
|
| 4652 |
+
},
|
| 4653 |
+
{
|
| 4654 |
+
"epoch": 1.7655470568673097,
|
| 4655 |
+
"grad_norm": NaN,
|
| 4656 |
+
"learning_rate": 3.3718793550262195e-06,
|
| 4657 |
+
"loss": 0.0,
|
| 4658 |
+
"step": 663
|
| 4659 |
+
},
|
| 4660 |
+
{
|
| 4661 |
+
"epoch": 1.7682075157964747,
|
| 4662 |
+
"grad_norm": NaN,
|
| 4663 |
+
"learning_rate": 3.2956685287148016e-06,
|
| 4664 |
+
"loss": 0.0,
|
| 4665 |
+
"step": 664
|
| 4666 |
+
},
|
| 4667 |
+
{
|
| 4668 |
+
"epoch": 1.7708679747256402,
|
| 4669 |
+
"grad_norm": NaN,
|
| 4670 |
+
"learning_rate": 3.220299471181898e-06,
|
| 4671 |
+
"loss": 0.0,
|
| 4672 |
+
"step": 665
|
| 4673 |
+
},
|
| 4674 |
+
{
|
| 4675 |
+
"epoch": 1.7735284336548056,
|
| 4676 |
+
"grad_norm": NaN,
|
| 4677 |
+
"learning_rate": 3.1457735408308154e-06,
|
| 4678 |
+
"loss": 0.0,
|
| 4679 |
+
"step": 666
|
| 4680 |
+
},
|
| 4681 |
+
{
|
| 4682 |
+
"epoch": 1.7761888925839706,
|
| 4683 |
+
"grad_norm": NaN,
|
| 4684 |
+
"learning_rate": 3.0720920808689434e-06,
|
| 4685 |
+
"loss": 0.0,
|
| 4686 |
+
"step": 667
|
| 4687 |
+
},
|
| 4688 |
+
{
|
| 4689 |
+
"epoch": 1.778849351513136,
|
| 4690 |
+
"grad_norm": NaN,
|
| 4691 |
+
"learning_rate": 2.999256419283425e-06,
|
| 4692 |
+
"loss": 0.0,
|
| 4693 |
+
"step": 668
|
| 4694 |
+
},
|
| 4695 |
+
{
|
| 4696 |
+
"epoch": 1.7815098104423013,
|
| 4697 |
+
"grad_norm": NaN,
|
| 4698 |
+
"learning_rate": 2.9272678688172905e-06,
|
| 4699 |
+
"loss": 0.0,
|
| 4700 |
+
"step": 669
|
| 4701 |
+
},
|
| 4702 |
+
{
|
| 4703 |
+
"epoch": 1.7841702693714665,
|
| 4704 |
+
"grad_norm": NaN,
|
| 4705 |
+
"learning_rate": 2.85612772694579e-06,
|
| 4706 |
+
"loss": 0.0,
|
| 4707 |
+
"step": 670
|
| 4708 |
+
},
|
| 4709 |
+
{
|
| 4710 |
+
"epoch": 1.786830728300632,
|
| 4711 |
+
"grad_norm": NaN,
|
| 4712 |
+
"learning_rate": 2.7858372758530205e-06,
|
| 4713 |
+
"loss": 0.0,
|
| 4714 |
+
"step": 671
|
| 4715 |
+
},
|
| 4716 |
+
{
|
| 4717 |
+
"epoch": 1.7894911872297972,
|
| 4718 |
+
"grad_norm": NaN,
|
| 4719 |
+
"learning_rate": 2.7163977824087694e-06,
|
| 4720 |
+
"loss": 0.0,
|
| 4721 |
+
"step": 672
|
| 4722 |
+
},
|
| 4723 |
+
{
|
| 4724 |
+
"epoch": 1.7921516461589624,
|
| 4725 |
+
"grad_norm": NaN,
|
| 4726 |
+
"learning_rate": 2.6478104981457253e-06,
|
| 4727 |
+
"loss": 0.0,
|
| 4728 |
+
"step": 673
|
| 4729 |
+
},
|
| 4730 |
+
{
|
| 4731 |
+
"epoch": 1.7948121050881278,
|
| 4732 |
+
"grad_norm": NaN,
|
| 4733 |
+
"learning_rate": 2.5800766592369073e-06,
|
| 4734 |
+
"loss": 0.0,
|
| 4735 |
+
"step": 674
|
| 4736 |
+
},
|
| 4737 |
+
{
|
| 4738 |
+
"epoch": 1.7974725640172928,
|
| 4739 |
+
"grad_norm": NaN,
|
| 4740 |
+
"learning_rate": 2.5131974864734066e-06,
|
| 4741 |
+
"loss": 0.0,
|
| 4742 |
+
"step": 675
|
| 4743 |
+
},
|
| 4744 |
+
{
|
| 4745 |
+
"epoch": 1.8001330229464583,
|
| 4746 |
+
"grad_norm": NaN,
|
| 4747 |
+
"learning_rate": 2.4471741852423237e-06,
|
| 4748 |
+
"loss": 0.0,
|
| 4749 |
+
"step": 676
|
| 4750 |
+
},
|
| 4751 |
+
{
|
| 4752 |
+
"epoch": 1.8027934818756235,
|
| 4753 |
+
"grad_norm": NaN,
|
| 4754 |
+
"learning_rate": 2.382007945505105e-06,
|
| 4755 |
+
"loss": 0.0,
|
| 4756 |
+
"step": 677
|
| 4757 |
+
},
|
| 4758 |
+
{
|
| 4759 |
+
"epoch": 1.8054539408047887,
|
| 4760 |
+
"grad_norm": NaN,
|
| 4761 |
+
"learning_rate": 2.317699941776064e-06,
|
| 4762 |
+
"loss": 0.0,
|
| 4763 |
+
"step": 678
|
| 4764 |
+
},
|
| 4765 |
+
{
|
| 4766 |
+
"epoch": 1.8081143997339542,
|
| 4767 |
+
"grad_norm": NaN,
|
| 4768 |
+
"learning_rate": 2.254251333101237e-06,
|
| 4769 |
+
"loss": 0.0,
|
| 4770 |
+
"step": 679
|
| 4771 |
+
},
|
| 4772 |
+
{
|
| 4773 |
+
"epoch": 1.8107748586631194,
|
| 4774 |
+
"grad_norm": NaN,
|
| 4775 |
+
"learning_rate": 2.191663263037458e-06,
|
| 4776 |
+
"loss": 0.0,
|
| 4777 |
+
"step": 680
|
| 4778 |
+
},
|
| 4779 |
+
{
|
| 4780 |
+
"epoch": 1.8134353175922846,
|
| 4781 |
+
"grad_norm": NaN,
|
| 4782 |
+
"learning_rate": 2.1299368596317702e-06,
|
| 4783 |
+
"loss": 0.0,
|
| 4784 |
+
"step": 681
|
| 4785 |
+
},
|
| 4786 |
+
{
|
| 4787 |
+
"epoch": 1.81609577652145,
|
| 4788 |
+
"grad_norm": NaN,
|
| 4789 |
+
"learning_rate": 2.069073235401109e-06,
|
| 4790 |
+
"loss": 0.0,
|
| 4791 |
+
"step": 682
|
| 4792 |
+
},
|
| 4793 |
+
{
|
| 4794 |
+
"epoch": 1.8187562354506153,
|
| 4795 |
+
"grad_norm": NaN,
|
| 4796 |
+
"learning_rate": 2.009073487312224e-06,
|
| 4797 |
+
"loss": 0.0,
|
| 4798 |
+
"step": 683
|
| 4799 |
+
},
|
| 4800 |
+
{
|
| 4801 |
+
"epoch": 1.8214166943797805,
|
| 4802 |
+
"grad_norm": NaN,
|
| 4803 |
+
"learning_rate": 1.9499386967619103e-06,
|
| 4804 |
+
"loss": 0.0,
|
| 4805 |
+
"step": 684
|
| 4806 |
+
},
|
| 4807 |
+
{
|
| 4808 |
+
"epoch": 1.824077153308946,
|
| 4809 |
+
"grad_norm": NaN,
|
| 4810 |
+
"learning_rate": 1.8916699295575324e-06,
|
| 4811 |
+
"loss": 0.0,
|
| 4812 |
+
"step": 685
|
| 4813 |
+
},
|
| 4814 |
+
{
|
| 4815 |
+
"epoch": 1.826737612238111,
|
| 4816 |
+
"grad_norm": NaN,
|
| 4817 |
+
"learning_rate": 1.8342682358978069e-06,
|
| 4818 |
+
"loss": 0.0,
|
| 4819 |
+
"step": 686
|
| 4820 |
+
},
|
| 4821 |
+
{
|
| 4822 |
+
"epoch": 1.8293980711672764,
|
| 4823 |
+
"grad_norm": NaN,
|
| 4824 |
+
"learning_rate": 1.7777346503538794e-06,
|
| 4825 |
+
"loss": 0.0,
|
| 4826 |
+
"step": 687
|
| 4827 |
+
},
|
| 4828 |
+
{
|
| 4829 |
+
"epoch": 1.8320585300964416,
|
| 4830 |
+
"grad_norm": NaN,
|
| 4831 |
+
"learning_rate": 1.7220701918506665e-06,
|
| 4832 |
+
"loss": 0.0,
|
| 4833 |
+
"step": 688
|
| 4834 |
+
},
|
| 4835 |
+
{
|
| 4836 |
+
"epoch": 1.8347189890256068,
|
| 4837 |
+
"grad_norm": NaN,
|
| 4838 |
+
"learning_rate": 1.6672758636485042e-06,
|
| 4839 |
+
"loss": 0.0,
|
| 4840 |
+
"step": 689
|
| 4841 |
+
},
|
| 4842 |
+
{
|
| 4843 |
+
"epoch": 1.8373794479547723,
|
| 4844 |
+
"grad_norm": NaN,
|
| 4845 |
+
"learning_rate": 1.6133526533250565e-06,
|
| 4846 |
+
"loss": 0.0,
|
| 4847 |
+
"step": 690
|
| 4848 |
+
},
|
| 4849 |
+
{
|
| 4850 |
+
"epoch": 1.8400399068839375,
|
| 4851 |
+
"grad_norm": NaN,
|
| 4852 |
+
"learning_rate": 1.5603015327575354e-06,
|
| 4853 |
+
"loss": 0.0,
|
| 4854 |
+
"step": 691
|
| 4855 |
+
},
|
| 4856 |
+
{
|
| 4857 |
+
"epoch": 1.8427003658131027,
|
| 4858 |
+
"grad_norm": NaN,
|
| 4859 |
+
"learning_rate": 1.5081234581051485e-06,
|
| 4860 |
+
"loss": 0.0,
|
| 4861 |
+
"step": 692
|
| 4862 |
+
},
|
| 4863 |
+
{
|
| 4864 |
+
"epoch": 1.8453608247422681,
|
| 4865 |
+
"grad_norm": NaN,
|
| 4866 |
+
"learning_rate": 1.4568193697918841e-06,
|
| 4867 |
+
"loss": 0.0,
|
| 4868 |
+
"step": 693
|
| 4869 |
+
},
|
| 4870 |
+
{
|
| 4871 |
+
"epoch": 1.8480212836714334,
|
| 4872 |
+
"grad_norm": NaN,
|
| 4873 |
+
"learning_rate": 1.4063901924895984e-06,
|
| 4874 |
+
"loss": 0.0,
|
| 4875 |
+
"step": 694
|
| 4876 |
+
},
|
| 4877 |
+
{
|
| 4878 |
+
"epoch": 1.8506817426005986,
|
| 4879 |
+
"grad_norm": NaN,
|
| 4880 |
+
"learning_rate": 1.3568368351012717e-06,
|
| 4881 |
+
"loss": 0.0,
|
| 4882 |
+
"step": 695
|
| 4883 |
+
},
|
| 4884 |
+
{
|
| 4885 |
+
"epoch": 1.853342201529764,
|
| 4886 |
+
"grad_norm": NaN,
|
| 4887 |
+
"learning_rate": 1.3081601907447006e-06,
|
| 4888 |
+
"loss": 0.0,
|
| 4889 |
+
"step": 696
|
| 4890 |
+
},
|
| 4891 |
+
{
|
| 4892 |
+
"epoch": 1.856002660458929,
|
| 4893 |
+
"grad_norm": NaN,
|
| 4894 |
+
"learning_rate": 1.260361136736349e-06,
|
| 4895 |
+
"loss": 0.0,
|
| 4896 |
+
"step": 697
|
| 4897 |
+
},
|
| 4898 |
+
{
|
| 4899 |
+
"epoch": 1.8586631193880945,
|
| 4900 |
+
"grad_norm": NaN,
|
| 4901 |
+
"learning_rate": 1.2134405345755773e-06,
|
| 4902 |
+
"loss": 0.0,
|
| 4903 |
+
"step": 698
|
| 4904 |
+
},
|
| 4905 |
+
{
|
| 4906 |
+
"epoch": 1.8613235783172597,
|
| 4907 |
+
"grad_norm": NaN,
|
| 4908 |
+
"learning_rate": 1.1673992299290892e-06,
|
| 4909 |
+
"loss": 0.0,
|
| 4910 |
+
"step": 699
|
| 4911 |
+
},
|
| 4912 |
+
{
|
| 4913 |
+
"epoch": 1.863984037246425,
|
| 4914 |
+
"grad_norm": NaN,
|
| 4915 |
+
"learning_rate": 1.1222380526156928e-06,
|
| 4916 |
+
"loss": 0.0,
|
| 4917 |
+
"step": 700
|
| 4918 |
+
},
|
| 4919 |
+
{
|
| 4920 |
+
"epoch": 1.8666444961755904,
|
| 4921 |
+
"grad_norm": NaN,
|
| 4922 |
+
"learning_rate": 1.0779578165913463e-06,
|
| 4923 |
+
"loss": 0.0,
|
| 4924 |
+
"step": 701
|
| 4925 |
+
},
|
| 4926 |
+
{
|
| 4927 |
+
"epoch": 1.8693049551047556,
|
| 4928 |
+
"grad_norm": NaN,
|
| 4929 |
+
"learning_rate": 1.034559319934497e-06,
|
| 4930 |
+
"loss": 0.0,
|
| 4931 |
+
"step": 702
|
| 4932 |
+
},
|
| 4933 |
+
{
|
| 4934 |
+
"epoch": 1.8719654140339208,
|
| 4935 |
+
"grad_norm": NaN,
|
| 4936 |
+
"learning_rate": 9.92043344831689e-07,
|
| 4937 |
+
"loss": 0.0,
|
| 4938 |
+
"step": 703
|
| 4939 |
+
},
|
| 4940 |
+
{
|
| 4941 |
+
"epoch": 1.8746258729630862,
|
| 4942 |
+
"grad_norm": NaN,
|
| 4943 |
+
"learning_rate": 9.504106575634664e-07,
|
| 4944 |
+
"loss": 0.0,
|
| 4945 |
+
"step": 704
|
| 4946 |
+
},
|
| 4947 |
+
{
|
| 4948 |
+
"epoch": 1.8772863318922512,
|
| 4949 |
+
"grad_norm": NaN,
|
| 4950 |
+
"learning_rate": 9.096620084905472e-07,
|
| 4951 |
+
"loss": 0.0,
|
| 4952 |
+
"step": 705
|
| 4953 |
+
},
|
| 4954 |
+
{
|
| 4955 |
+
"epoch": 1.8799467908214167,
|
| 4956 |
+
"grad_norm": NaN,
|
| 4957 |
+
"learning_rate": 8.697981320403337e-07,
|
| 4958 |
+
"loss": 0.0,
|
| 4959 |
+
"step": 706
|
| 4960 |
+
},
|
| 4961 |
+
{
|
| 4962 |
+
"epoch": 1.8826072497505821,
|
| 4963 |
+
"grad_norm": NaN,
|
| 4964 |
+
"learning_rate": 8.308197466936618e-07,
|
| 4965 |
+
"loss": 0.0,
|
| 4966 |
+
"step": 707
|
| 4967 |
+
},
|
| 4968 |
+
{
|
| 4969 |
+
"epoch": 1.8852677086797471,
|
| 4970 |
+
"grad_norm": NaN,
|
| 4971 |
+
"learning_rate": 7.927275549718228e-07,
|
| 4972 |
+
"loss": 0.0,
|
| 4973 |
+
"step": 708
|
| 4974 |
+
},
|
| 4975 |
+
{
|
| 4976 |
+
"epoch": 1.8879281676089126,
|
| 4977 |
+
"grad_norm": NaN,
|
| 4978 |
+
"learning_rate": 7.555222434239395e-07,
|
| 4979 |
+
"loss": 0.0,
|
| 4980 |
+
"step": 709
|
| 4981 |
+
},
|
| 4982 |
+
{
|
| 4983 |
+
"epoch": 1.8905886265380778,
|
| 4984 |
+
"grad_norm": NaN,
|
| 4985 |
+
"learning_rate": 7.192044826145771e-07,
|
| 4986 |
+
"loss": 0.0,
|
| 4987 |
+
"step": 710
|
| 4988 |
+
},
|
| 4989 |
+
{
|
| 4990 |
+
"epoch": 1.893249085467243,
|
| 4991 |
+
"grad_norm": NaN,
|
| 4992 |
+
"learning_rate": 6.837749271116578e-07,
|
| 4993 |
+
"loss": 0.0,
|
| 4994 |
+
"step": 711
|
| 4995 |
+
},
|
| 4996 |
+
{
|
| 4997 |
+
"epoch": 1.8959095443964085,
|
| 4998 |
+
"grad_norm": NaN,
|
| 4999 |
+
"learning_rate": 6.492342154746589e-07,
|
| 5000 |
+
"loss": 0.0,
|
| 5001 |
+
"step": 712
|
| 5002 |
+
},
|
| 5003 |
+
{
|
| 5004 |
+
"epoch": 1.8985700033255737,
|
| 5005 |
+
"grad_norm": NaN,
|
| 5006 |
+
"learning_rate": 6.15582970243117e-07,
|
| 5007 |
+
"loss": 0.0,
|
| 5008 |
+
"step": 713
|
| 5009 |
+
},
|
| 5010 |
+
{
|
| 5011 |
+
"epoch": 1.901230462254739,
|
| 5012 |
+
"grad_norm": NaN,
|
| 5013 |
+
"learning_rate": 5.828217979253869e-07,
|
| 5014 |
+
"loss": 0.0,
|
| 5015 |
+
"step": 714
|
| 5016 |
+
},
|
| 5017 |
+
{
|
| 5018 |
+
"epoch": 1.9038909211839044,
|
| 5019 |
+
"grad_norm": NaN,
|
| 5020 |
+
"learning_rate": 5.509512889877333e-07,
|
| 5021 |
+
"loss": 0.0,
|
| 5022 |
+
"step": 715
|
| 5023 |
+
},
|
| 5024 |
+
{
|
| 5025 |
+
"epoch": 1.9065513801130693,
|
| 5026 |
+
"grad_norm": NaN,
|
| 5027 |
+
"learning_rate": 5.19972017843684e-07,
|
| 5028 |
+
"loss": 0.0,
|
| 5029 |
+
"step": 716
|
| 5030 |
+
},
|
| 5031 |
+
{
|
| 5032 |
+
"epoch": 1.9092118390422348,
|
| 5033 |
+
"grad_norm": NaN,
|
| 5034 |
+
"learning_rate": 4.898845428436494e-07,
|
| 5035 |
+
"loss": 0.0,
|
| 5036 |
+
"step": 717
|
| 5037 |
+
},
|
| 5038 |
+
{
|
| 5039 |
+
"epoch": 1.9118722979714,
|
| 5040 |
+
"grad_norm": NaN,
|
| 5041 |
+
"learning_rate": 4.60689406264897e-07,
|
| 5042 |
+
"loss": 0.0,
|
| 5043 |
+
"step": 718
|
| 5044 |
+
},
|
| 5045 |
+
{
|
| 5046 |
+
"epoch": 1.9145327569005652,
|
| 5047 |
+
"grad_norm": NaN,
|
| 5048 |
+
"learning_rate": 4.323871343017649e-07,
|
| 5049 |
+
"loss": 0.0,
|
| 5050 |
+
"step": 719
|
| 5051 |
+
},
|
| 5052 |
+
{
|
| 5053 |
+
"epoch": 1.9171932158297307,
|
| 5054 |
+
"grad_norm": NaN,
|
| 5055 |
+
"learning_rate": 4.049782370561583e-07,
|
| 5056 |
+
"loss": 0.0,
|
| 5057 |
+
"step": 720
|
| 5058 |
+
},
|
| 5059 |
+
{
|
| 5060 |
+
"epoch": 1.919853674758896,
|
| 5061 |
+
"grad_norm": NaN,
|
| 5062 |
+
"learning_rate": 3.7846320852838456e-07,
|
| 5063 |
+
"loss": 0.0,
|
| 5064 |
+
"step": 721
|
| 5065 |
+
},
|
| 5066 |
+
{
|
| 5067 |
+
"epoch": 1.9225141336880611,
|
| 5068 |
+
"grad_norm": NaN,
|
| 5069 |
+
"learning_rate": 3.528425266082325e-07,
|
| 5070 |
+
"loss": 0.0,
|
| 5071 |
+
"step": 722
|
| 5072 |
+
},
|
| 5073 |
+
{
|
| 5074 |
+
"epoch": 1.9251745926172266,
|
| 5075 |
+
"grad_norm": NaN,
|
| 5076 |
+
"learning_rate": 3.281166530663682e-07,
|
| 5077 |
+
"loss": 0.0,
|
| 5078 |
+
"step": 723
|
| 5079 |
+
},
|
| 5080 |
+
{
|
| 5081 |
+
"epoch": 1.9278350515463918,
|
| 5082 |
+
"grad_norm": NaN,
|
| 5083 |
+
"learning_rate": 3.042860335460085e-07,
|
| 5084 |
+
"loss": 0.0,
|
| 5085 |
+
"step": 724
|
| 5086 |
+
},
|
| 5087 |
+
{
|
| 5088 |
+
"epoch": 1.930495510475557,
|
| 5089 |
+
"grad_norm": NaN,
|
| 5090 |
+
"learning_rate": 2.813510975548772e-07,
|
| 5091 |
+
"loss": 0.0,
|
| 5092 |
+
"step": 725
|
| 5093 |
+
},
|
| 5094 |
+
{
|
| 5095 |
+
"epoch": 1.9331559694047225,
|
| 5096 |
+
"grad_norm": NaN,
|
| 5097 |
+
"learning_rate": 2.593122584574892e-07,
|
| 5098 |
+
"loss": 0.0,
|
| 5099 |
+
"step": 726
|
| 5100 |
+
},
|
| 5101 |
+
{
|
| 5102 |
+
"epoch": 1.9358164283338875,
|
| 5103 |
+
"grad_norm": NaN,
|
| 5104 |
+
"learning_rate": 2.3816991346767847e-07,
|
| 5105 |
+
"loss": 0.0,
|
| 5106 |
+
"step": 727
|
| 5107 |
+
},
|
| 5108 |
+
{
|
| 5109 |
+
"epoch": 1.938476887263053,
|
| 5110 |
+
"grad_norm": NaN,
|
| 5111 |
+
"learning_rate": 2.1792444364144847e-07,
|
| 5112 |
+
"loss": 0.0,
|
| 5113 |
+
"step": 728
|
| 5114 |
+
},
|
| 5115 |
+
{
|
| 5116 |
+
"epoch": 1.9411373461922181,
|
| 5117 |
+
"grad_norm": NaN,
|
| 5118 |
+
"learning_rate": 1.9857621387010527e-07,
|
| 5119 |
+
"loss": 0.0,
|
| 5120 |
+
"step": 729
|
| 5121 |
+
},
|
| 5122 |
+
{
|
| 5123 |
+
"epoch": 1.9437978051213833,
|
| 5124 |
+
"grad_norm": NaN,
|
| 5125 |
+
"learning_rate": 1.8012557287367392e-07,
|
| 5126 |
+
"loss": 0.0,
|
| 5127 |
+
"step": 730
|
| 5128 |
+
},
|
| 5129 |
+
{
|
| 5130 |
+
"epoch": 1.9464582640505488,
|
| 5131 |
+
"grad_norm": NaN,
|
| 5132 |
+
"learning_rate": 1.6257285319462578e-07,
|
| 5133 |
+
"loss": 0.0,
|
| 5134 |
+
"step": 731
|
| 5135 |
+
},
|
| 5136 |
+
{
|
| 5137 |
+
"epoch": 1.949118722979714,
|
| 5138 |
+
"grad_norm": NaN,
|
| 5139 |
+
"learning_rate": 1.4591837119186103e-07,
|
| 5140 |
+
"loss": 0.0,
|
| 5141 |
+
"step": 732
|
| 5142 |
+
},
|
| 5143 |
+
{
|
| 5144 |
+
"epoch": 1.9517791819088792,
|
| 5145 |
+
"grad_norm": NaN,
|
| 5146 |
+
"learning_rate": 1.3016242703503544e-07,
|
| 5147 |
+
"loss": 0.0,
|
| 5148 |
+
"step": 733
|
| 5149 |
+
},
|
| 5150 |
+
{
|
| 5151 |
+
"epoch": 1.9544396408380447,
|
| 5152 |
+
"grad_norm": NaN,
|
| 5153 |
+
"learning_rate": 1.1530530469914259e-07,
|
| 5154 |
+
"loss": 0.0,
|
| 5155 |
+
"step": 734
|
| 5156 |
+
},
|
| 5157 |
+
{
|
| 5158 |
+
"epoch": 1.9571000997672099,
|
| 5159 |
+
"grad_norm": NaN,
|
| 5160 |
+
"learning_rate": 1.0134727195937333e-07,
|
| 5161 |
+
"loss": 0.0,
|
| 5162 |
+
"step": 735
|
| 5163 |
+
},
|
| 5164 |
+
{
|
| 5165 |
+
"epoch": 1.959760558696375,
|
| 5166 |
+
"grad_norm": NaN,
|
| 5167 |
+
"learning_rate": 8.828858038632536e-08,
|
| 5168 |
+
"loss": 0.0,
|
| 5169 |
+
"step": 736
|
| 5170 |
+
},
|
| 5171 |
+
{
|
| 5172 |
+
"epoch": 1.9624210176255406,
|
| 5173 |
+
"grad_norm": NaN,
|
| 5174 |
+
"learning_rate": 7.612946534143461e-08,
|
| 5175 |
+
"loss": 0.0,
|
| 5176 |
+
"step": 737
|
| 5177 |
+
},
|
| 5178 |
+
{
|
| 5179 |
+
"epoch": 1.9650814765547056,
|
| 5180 |
+
"grad_norm": NaN,
|
| 5181 |
+
"learning_rate": 6.487014597275631e-08,
|
| 5182 |
+
"loss": 0.0,
|
| 5183 |
+
"step": 738
|
| 5184 |
+
},
|
| 5185 |
+
{
|
| 5186 |
+
"epoch": 1.967741935483871,
|
| 5187 |
+
"grad_norm": NaN,
|
| 5188 |
+
"learning_rate": 5.4510825211012694e-08,
|
| 5189 |
+
"loss": 0.0,
|
| 5190 |
+
"step": 739
|
| 5191 |
+
},
|
| 5192 |
+
{
|
| 5193 |
+
"epoch": 1.9704023944130362,
|
| 5194 |
+
"grad_norm": NaN,
|
| 5195 |
+
"learning_rate": 4.5051689765929214e-08,
|
| 5196 |
+
"loss": 0.0,
|
| 5197 |
+
"step": 740
|
| 5198 |
+
},
|
| 5199 |
+
{
|
| 5200 |
+
"epoch": 1.9730628533422014,
|
| 5201 |
+
"grad_norm": NaN,
|
| 5202 |
+
"learning_rate": 3.6492910122859494e-08,
|
| 5203 |
+
"loss": 0.0,
|
| 5204 |
+
"step": 741
|
| 5205 |
+
},
|
| 5206 |
+
{
|
| 5207 |
+
"epoch": 1.9757233122713669,
|
| 5208 |
+
"grad_norm": NaN,
|
| 5209 |
+
"learning_rate": 2.8834640539737723e-08,
|
| 5210 |
+
"loss": 0.0,
|
| 5211 |
+
"step": 742
|
| 5212 |
+
},
|
| 5213 |
+
{
|
| 5214 |
+
"epoch": 1.978383771200532,
|
| 5215 |
+
"grad_norm": NaN,
|
| 5216 |
+
"learning_rate": 2.2077019044280943e-08,
|
| 5217 |
+
"loss": 0.0,
|
| 5218 |
+
"step": 743
|
| 5219 |
+
},
|
| 5220 |
+
{
|
| 5221 |
+
"epoch": 1.9810442301296973,
|
| 5222 |
+
"grad_norm": NaN,
|
| 5223 |
+
"learning_rate": 1.622016743150212e-08,
|
| 5224 |
+
"loss": 0.0,
|
| 5225 |
+
"step": 744
|
| 5226 |
+
},
|
| 5227 |
+
{
|
| 5228 |
+
"epoch": 1.9837046890588628,
|
| 5229 |
+
"grad_norm": NaN,
|
| 5230 |
+
"learning_rate": 1.1264191261528557e-08,
|
| 5231 |
+
"loss": 0.0,
|
| 5232 |
+
"step": 745
|
| 5233 |
+
},
|
| 5234 |
+
{
|
| 5235 |
+
"epoch": 1.9863651479880278,
|
| 5236 |
+
"grad_norm": NaN,
|
| 5237 |
+
"learning_rate": 7.209179857675663e-09,
|
| 5238 |
+
"loss": 0.0,
|
| 5239 |
+
"step": 746
|
| 5240 |
+
},
|
| 5241 |
+
{
|
| 5242 |
+
"epoch": 1.9890256069171932,
|
| 5243 |
+
"grad_norm": NaN,
|
| 5244 |
+
"learning_rate": 4.055206304859338e-09,
|
| 5245 |
+
"loss": 0.0,
|
| 5246 |
+
"step": 747
|
| 5247 |
+
},
|
| 5248 |
+
{
|
| 5249 |
+
"epoch": 1.9916860658463587,
|
| 5250 |
+
"grad_norm": NaN,
|
| 5251 |
+
"learning_rate": 1.8023274482636965e-09,
|
| 5252 |
+
"loss": 0.0,
|
| 5253 |
+
"step": 748
|
| 5254 |
+
},
|
| 5255 |
+
{
|
| 5256 |
+
"epoch": 1.9943465247755237,
|
| 5257 |
+
"grad_norm": NaN,
|
| 5258 |
+
"learning_rate": 4.5058389232521816e-10,
|
| 5259 |
+
"loss": 0.0,
|
| 5260 |
+
"step": 749
|
| 5261 |
+
},
|
| 5262 |
+
{
|
| 5263 |
+
"epoch": 1.997006983704689,
|
| 5264 |
+
"grad_norm": NaN,
|
| 5265 |
+
"learning_rate": 0.0,
|
| 5266 |
+
"loss": 0.0,
|
| 5267 |
+
"step": 750
|
| 5268 |
}
|
| 5269 |
],
|
| 5270 |
"logging_steps": 1,
|
|
|
|
| 5279 |
"should_evaluate": false,
|
| 5280 |
"should_log": false,
|
| 5281 |
"should_save": true,
|
| 5282 |
+
"should_training_stop": true
|
| 5283 |
},
|
| 5284 |
"attributes": {}
|
| 5285 |
}
|
| 5286 |
},
|
| 5287 |
+
"total_flos": 2.083419328610304e+18,
|
| 5288 |
"train_batch_size": 4,
|
| 5289 |
"trial_name": null,
|
| 5290 |
"trial_params": null
|