Training in progress, step 200, 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 +1503 -3
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 453156026
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:81927c416d0ed83e63ba43e8296226e165456da2baba9aaa71e305105d0dcf81
|
| 3 |
size 453156026
|
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 sha256:7a81d5d35ff0c1cfc3cd4e03db6ce33aa9d62187b2759f2277441bd67c8ec4cd
|
| 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 sha256:1e6e2c3bcff0341bfddccfdfe85ee6efb7f42535e1af3e46983f5243cc0444e8
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -1507,6 +1507,1506 @@
|
|
| 1507 |
"rewards/margins": NaN,
|
| 1508 |
"rewards/rejected": NaN,
|
| 1509 |
"step": 100
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1510 |
}
|
| 1511 |
],
|
| 1512 |
"logging_steps": 1,
|
|
@@ -1521,7 +3021,7 @@
|
|
| 1521 |
"should_evaluate": false,
|
| 1522 |
"should_log": false,
|
| 1523 |
"should_save": true,
|
| 1524 |
-
"should_training_stop":
|
| 1525 |
},
|
| 1526 |
"attributes": {}
|
| 1527 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.05603418085031869,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 200,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 1507 |
"rewards/margins": NaN,
|
| 1508 |
"rewards/rejected": NaN,
|
| 1509 |
"step": 100
|
| 1510 |
+
},
|
| 1511 |
+
{
|
| 1512 |
+
"epoch": 0.02829726132941094,
|
| 1513 |
+
"grad_norm": NaN,
|
| 1514 |
+
"learning_rate": 5.050000000000001e-06,
|
| 1515 |
+
"logits/chosen": NaN,
|
| 1516 |
+
"logits/rejected": NaN,
|
| 1517 |
+
"logps/chosen": NaN,
|
| 1518 |
+
"logps/rejected": NaN,
|
| 1519 |
+
"loss": 0.0,
|
| 1520 |
+
"rewards/accuracies": 0.0,
|
| 1521 |
+
"rewards/chosen": NaN,
|
| 1522 |
+
"rewards/margins": NaN,
|
| 1523 |
+
"rewards/rejected": NaN,
|
| 1524 |
+
"step": 101
|
| 1525 |
+
},
|
| 1526 |
+
{
|
| 1527 |
+
"epoch": 0.028577432233662534,
|
| 1528 |
+
"grad_norm": NaN,
|
| 1529 |
+
"learning_rate": 5.1e-06,
|
| 1530 |
+
"logits/chosen": NaN,
|
| 1531 |
+
"logits/rejected": NaN,
|
| 1532 |
+
"logps/chosen": NaN,
|
| 1533 |
+
"logps/rejected": NaN,
|
| 1534 |
+
"loss": 0.0,
|
| 1535 |
+
"rewards/accuracies": 0.0,
|
| 1536 |
+
"rewards/chosen": NaN,
|
| 1537 |
+
"rewards/margins": NaN,
|
| 1538 |
+
"rewards/rejected": NaN,
|
| 1539 |
+
"step": 102
|
| 1540 |
+
},
|
| 1541 |
+
{
|
| 1542 |
+
"epoch": 0.028857603137914127,
|
| 1543 |
+
"grad_norm": NaN,
|
| 1544 |
+
"learning_rate": 5.150000000000001e-06,
|
| 1545 |
+
"logits/chosen": NaN,
|
| 1546 |
+
"logits/rejected": NaN,
|
| 1547 |
+
"logps/chosen": NaN,
|
| 1548 |
+
"logps/rejected": NaN,
|
| 1549 |
+
"loss": 0.0,
|
| 1550 |
+
"rewards/accuracies": 0.0,
|
| 1551 |
+
"rewards/chosen": NaN,
|
| 1552 |
+
"rewards/margins": NaN,
|
| 1553 |
+
"rewards/rejected": NaN,
|
| 1554 |
+
"step": 103
|
| 1555 |
+
},
|
| 1556 |
+
{
|
| 1557 |
+
"epoch": 0.02913777404216572,
|
| 1558 |
+
"grad_norm": NaN,
|
| 1559 |
+
"learning_rate": 5.2e-06,
|
| 1560 |
+
"logits/chosen": NaN,
|
| 1561 |
+
"logits/rejected": NaN,
|
| 1562 |
+
"logps/chosen": NaN,
|
| 1563 |
+
"logps/rejected": NaN,
|
| 1564 |
+
"loss": 0.0,
|
| 1565 |
+
"rewards/accuracies": 0.0,
|
| 1566 |
+
"rewards/chosen": NaN,
|
| 1567 |
+
"rewards/margins": NaN,
|
| 1568 |
+
"rewards/rejected": NaN,
|
| 1569 |
+
"step": 104
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 0.029417944946417314,
|
| 1573 |
+
"grad_norm": NaN,
|
| 1574 |
+
"learning_rate": 5.2500000000000006e-06,
|
| 1575 |
+
"logits/chosen": NaN,
|
| 1576 |
+
"logits/rejected": NaN,
|
| 1577 |
+
"logps/chosen": NaN,
|
| 1578 |
+
"logps/rejected": NaN,
|
| 1579 |
+
"loss": 0.0,
|
| 1580 |
+
"rewards/accuracies": 0.0,
|
| 1581 |
+
"rewards/chosen": NaN,
|
| 1582 |
+
"rewards/margins": NaN,
|
| 1583 |
+
"rewards/rejected": NaN,
|
| 1584 |
+
"step": 105
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 0.029698115850668907,
|
| 1588 |
+
"grad_norm": NaN,
|
| 1589 |
+
"learning_rate": 5.300000000000001e-06,
|
| 1590 |
+
"logits/chosen": NaN,
|
| 1591 |
+
"logits/rejected": NaN,
|
| 1592 |
+
"logps/chosen": NaN,
|
| 1593 |
+
"logps/rejected": NaN,
|
| 1594 |
+
"loss": 0.0,
|
| 1595 |
+
"rewards/accuracies": 0.0,
|
| 1596 |
+
"rewards/chosen": NaN,
|
| 1597 |
+
"rewards/margins": NaN,
|
| 1598 |
+
"rewards/rejected": NaN,
|
| 1599 |
+
"step": 106
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 0.0299782867549205,
|
| 1603 |
+
"grad_norm": NaN,
|
| 1604 |
+
"learning_rate": 5.3500000000000004e-06,
|
| 1605 |
+
"logits/chosen": NaN,
|
| 1606 |
+
"logits/rejected": NaN,
|
| 1607 |
+
"logps/chosen": NaN,
|
| 1608 |
+
"logps/rejected": NaN,
|
| 1609 |
+
"loss": 0.0,
|
| 1610 |
+
"rewards/accuracies": 0.0,
|
| 1611 |
+
"rewards/chosen": NaN,
|
| 1612 |
+
"rewards/margins": NaN,
|
| 1613 |
+
"rewards/rejected": NaN,
|
| 1614 |
+
"step": 107
|
| 1615 |
+
},
|
| 1616 |
+
{
|
| 1617 |
+
"epoch": 0.030258457659172094,
|
| 1618 |
+
"grad_norm": NaN,
|
| 1619 |
+
"learning_rate": 5.400000000000001e-06,
|
| 1620 |
+
"logits/chosen": NaN,
|
| 1621 |
+
"logits/rejected": NaN,
|
| 1622 |
+
"logps/chosen": NaN,
|
| 1623 |
+
"logps/rejected": NaN,
|
| 1624 |
+
"loss": 0.0,
|
| 1625 |
+
"rewards/accuracies": 0.0,
|
| 1626 |
+
"rewards/chosen": NaN,
|
| 1627 |
+
"rewards/margins": NaN,
|
| 1628 |
+
"rewards/rejected": NaN,
|
| 1629 |
+
"step": 108
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 0.030538628563423688,
|
| 1633 |
+
"grad_norm": NaN,
|
| 1634 |
+
"learning_rate": 5.450000000000001e-06,
|
| 1635 |
+
"logits/chosen": NaN,
|
| 1636 |
+
"logits/rejected": NaN,
|
| 1637 |
+
"logps/chosen": NaN,
|
| 1638 |
+
"logps/rejected": NaN,
|
| 1639 |
+
"loss": 0.0,
|
| 1640 |
+
"rewards/accuracies": 0.0,
|
| 1641 |
+
"rewards/chosen": NaN,
|
| 1642 |
+
"rewards/margins": NaN,
|
| 1643 |
+
"rewards/rejected": NaN,
|
| 1644 |
+
"step": 109
|
| 1645 |
+
},
|
| 1646 |
+
{
|
| 1647 |
+
"epoch": 0.03081879946767528,
|
| 1648 |
+
"grad_norm": NaN,
|
| 1649 |
+
"learning_rate": 5.500000000000001e-06,
|
| 1650 |
+
"logits/chosen": NaN,
|
| 1651 |
+
"logits/rejected": NaN,
|
| 1652 |
+
"logps/chosen": NaN,
|
| 1653 |
+
"logps/rejected": NaN,
|
| 1654 |
+
"loss": 0.0,
|
| 1655 |
+
"rewards/accuracies": 0.0,
|
| 1656 |
+
"rewards/chosen": NaN,
|
| 1657 |
+
"rewards/margins": NaN,
|
| 1658 |
+
"rewards/rejected": NaN,
|
| 1659 |
+
"step": 110
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 0.031098970371926875,
|
| 1663 |
+
"grad_norm": NaN,
|
| 1664 |
+
"learning_rate": 5.550000000000001e-06,
|
| 1665 |
+
"logits/chosen": NaN,
|
| 1666 |
+
"logits/rejected": NaN,
|
| 1667 |
+
"logps/chosen": NaN,
|
| 1668 |
+
"logps/rejected": NaN,
|
| 1669 |
+
"loss": 0.0,
|
| 1670 |
+
"rewards/accuracies": 0.0,
|
| 1671 |
+
"rewards/chosen": NaN,
|
| 1672 |
+
"rewards/margins": NaN,
|
| 1673 |
+
"rewards/rejected": NaN,
|
| 1674 |
+
"step": 111
|
| 1675 |
+
},
|
| 1676 |
+
{
|
| 1677 |
+
"epoch": 0.03137914127617847,
|
| 1678 |
+
"grad_norm": NaN,
|
| 1679 |
+
"learning_rate": 5.600000000000001e-06,
|
| 1680 |
+
"logits/chosen": NaN,
|
| 1681 |
+
"logits/rejected": NaN,
|
| 1682 |
+
"logps/chosen": NaN,
|
| 1683 |
+
"logps/rejected": NaN,
|
| 1684 |
+
"loss": 0.0,
|
| 1685 |
+
"rewards/accuracies": 0.0,
|
| 1686 |
+
"rewards/chosen": NaN,
|
| 1687 |
+
"rewards/margins": NaN,
|
| 1688 |
+
"rewards/rejected": NaN,
|
| 1689 |
+
"step": 112
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 0.03165931218043006,
|
| 1693 |
+
"grad_norm": NaN,
|
| 1694 |
+
"learning_rate": 5.65e-06,
|
| 1695 |
+
"logits/chosen": NaN,
|
| 1696 |
+
"logits/rejected": NaN,
|
| 1697 |
+
"logps/chosen": NaN,
|
| 1698 |
+
"logps/rejected": NaN,
|
| 1699 |
+
"loss": 0.0,
|
| 1700 |
+
"rewards/accuracies": 0.0,
|
| 1701 |
+
"rewards/chosen": NaN,
|
| 1702 |
+
"rewards/margins": NaN,
|
| 1703 |
+
"rewards/rejected": NaN,
|
| 1704 |
+
"step": 113
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 0.031939483084681655,
|
| 1708 |
+
"grad_norm": NaN,
|
| 1709 |
+
"learning_rate": 5.7e-06,
|
| 1710 |
+
"logits/chosen": NaN,
|
| 1711 |
+
"logits/rejected": NaN,
|
| 1712 |
+
"logps/chosen": NaN,
|
| 1713 |
+
"logps/rejected": NaN,
|
| 1714 |
+
"loss": 0.0,
|
| 1715 |
+
"rewards/accuracies": 0.0,
|
| 1716 |
+
"rewards/chosen": NaN,
|
| 1717 |
+
"rewards/margins": NaN,
|
| 1718 |
+
"rewards/rejected": NaN,
|
| 1719 |
+
"step": 114
|
| 1720 |
+
},
|
| 1721 |
+
{
|
| 1722 |
+
"epoch": 0.03221965398893325,
|
| 1723 |
+
"grad_norm": NaN,
|
| 1724 |
+
"learning_rate": 5.75e-06,
|
| 1725 |
+
"logits/chosen": NaN,
|
| 1726 |
+
"logits/rejected": NaN,
|
| 1727 |
+
"logps/chosen": NaN,
|
| 1728 |
+
"logps/rejected": NaN,
|
| 1729 |
+
"loss": 0.0,
|
| 1730 |
+
"rewards/accuracies": 0.0,
|
| 1731 |
+
"rewards/chosen": NaN,
|
| 1732 |
+
"rewards/margins": NaN,
|
| 1733 |
+
"rewards/rejected": NaN,
|
| 1734 |
+
"step": 115
|
| 1735 |
+
},
|
| 1736 |
+
{
|
| 1737 |
+
"epoch": 0.03249982489318484,
|
| 1738 |
+
"grad_norm": NaN,
|
| 1739 |
+
"learning_rate": 5.8e-06,
|
| 1740 |
+
"logits/chosen": NaN,
|
| 1741 |
+
"logits/rejected": NaN,
|
| 1742 |
+
"logps/chosen": NaN,
|
| 1743 |
+
"logps/rejected": NaN,
|
| 1744 |
+
"loss": 0.0,
|
| 1745 |
+
"rewards/accuracies": 0.0,
|
| 1746 |
+
"rewards/chosen": NaN,
|
| 1747 |
+
"rewards/margins": NaN,
|
| 1748 |
+
"rewards/rejected": NaN,
|
| 1749 |
+
"step": 116
|
| 1750 |
+
},
|
| 1751 |
+
{
|
| 1752 |
+
"epoch": 0.032779995797436436,
|
| 1753 |
+
"grad_norm": NaN,
|
| 1754 |
+
"learning_rate": 5.85e-06,
|
| 1755 |
+
"logits/chosen": NaN,
|
| 1756 |
+
"logits/rejected": NaN,
|
| 1757 |
+
"logps/chosen": NaN,
|
| 1758 |
+
"logps/rejected": NaN,
|
| 1759 |
+
"loss": 0.0,
|
| 1760 |
+
"rewards/accuracies": 0.0,
|
| 1761 |
+
"rewards/chosen": NaN,
|
| 1762 |
+
"rewards/margins": NaN,
|
| 1763 |
+
"rewards/rejected": NaN,
|
| 1764 |
+
"step": 117
|
| 1765 |
+
},
|
| 1766 |
+
{
|
| 1767 |
+
"epoch": 0.03306016670168803,
|
| 1768 |
+
"grad_norm": NaN,
|
| 1769 |
+
"learning_rate": 5.9e-06,
|
| 1770 |
+
"logits/chosen": NaN,
|
| 1771 |
+
"logits/rejected": NaN,
|
| 1772 |
+
"logps/chosen": NaN,
|
| 1773 |
+
"logps/rejected": NaN,
|
| 1774 |
+
"loss": 0.0,
|
| 1775 |
+
"rewards/accuracies": 0.0,
|
| 1776 |
+
"rewards/chosen": NaN,
|
| 1777 |
+
"rewards/margins": NaN,
|
| 1778 |
+
"rewards/rejected": NaN,
|
| 1779 |
+
"step": 118
|
| 1780 |
+
},
|
| 1781 |
+
{
|
| 1782 |
+
"epoch": 0.03334033760593962,
|
| 1783 |
+
"grad_norm": NaN,
|
| 1784 |
+
"learning_rate": 5.950000000000001e-06,
|
| 1785 |
+
"logits/chosen": NaN,
|
| 1786 |
+
"logits/rejected": NaN,
|
| 1787 |
+
"logps/chosen": NaN,
|
| 1788 |
+
"logps/rejected": NaN,
|
| 1789 |
+
"loss": 0.0,
|
| 1790 |
+
"rewards/accuracies": 0.0,
|
| 1791 |
+
"rewards/chosen": NaN,
|
| 1792 |
+
"rewards/margins": NaN,
|
| 1793 |
+
"rewards/rejected": NaN,
|
| 1794 |
+
"step": 119
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 0.033620508510191216,
|
| 1798 |
+
"grad_norm": NaN,
|
| 1799 |
+
"learning_rate": 6e-06,
|
| 1800 |
+
"logits/chosen": NaN,
|
| 1801 |
+
"logits/rejected": NaN,
|
| 1802 |
+
"logps/chosen": NaN,
|
| 1803 |
+
"logps/rejected": NaN,
|
| 1804 |
+
"loss": 0.0,
|
| 1805 |
+
"rewards/accuracies": 0.0,
|
| 1806 |
+
"rewards/chosen": NaN,
|
| 1807 |
+
"rewards/margins": NaN,
|
| 1808 |
+
"rewards/rejected": NaN,
|
| 1809 |
+
"step": 120
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 0.03390067941444281,
|
| 1813 |
+
"grad_norm": NaN,
|
| 1814 |
+
"learning_rate": 6.0500000000000005e-06,
|
| 1815 |
+
"logits/chosen": NaN,
|
| 1816 |
+
"logits/rejected": NaN,
|
| 1817 |
+
"logps/chosen": NaN,
|
| 1818 |
+
"logps/rejected": NaN,
|
| 1819 |
+
"loss": 0.0,
|
| 1820 |
+
"rewards/accuracies": 0.0,
|
| 1821 |
+
"rewards/chosen": NaN,
|
| 1822 |
+
"rewards/margins": NaN,
|
| 1823 |
+
"rewards/rejected": NaN,
|
| 1824 |
+
"step": 121
|
| 1825 |
+
},
|
| 1826 |
+
{
|
| 1827 |
+
"epoch": 0.0341808503186944,
|
| 1828 |
+
"grad_norm": NaN,
|
| 1829 |
+
"learning_rate": 6.1e-06,
|
| 1830 |
+
"logits/chosen": NaN,
|
| 1831 |
+
"logits/rejected": NaN,
|
| 1832 |
+
"logps/chosen": NaN,
|
| 1833 |
+
"logps/rejected": NaN,
|
| 1834 |
+
"loss": 0.0,
|
| 1835 |
+
"rewards/accuracies": 0.0,
|
| 1836 |
+
"rewards/chosen": NaN,
|
| 1837 |
+
"rewards/margins": NaN,
|
| 1838 |
+
"rewards/rejected": NaN,
|
| 1839 |
+
"step": 122
|
| 1840 |
+
},
|
| 1841 |
+
{
|
| 1842 |
+
"epoch": 0.034461021222945996,
|
| 1843 |
+
"grad_norm": NaN,
|
| 1844 |
+
"learning_rate": 6.15e-06,
|
| 1845 |
+
"logits/chosen": NaN,
|
| 1846 |
+
"logits/rejected": NaN,
|
| 1847 |
+
"logps/chosen": NaN,
|
| 1848 |
+
"logps/rejected": NaN,
|
| 1849 |
+
"loss": 0.0,
|
| 1850 |
+
"rewards/accuracies": 0.0,
|
| 1851 |
+
"rewards/chosen": NaN,
|
| 1852 |
+
"rewards/margins": NaN,
|
| 1853 |
+
"rewards/rejected": NaN,
|
| 1854 |
+
"step": 123
|
| 1855 |
+
},
|
| 1856 |
+
{
|
| 1857 |
+
"epoch": 0.03474119212719759,
|
| 1858 |
+
"grad_norm": NaN,
|
| 1859 |
+
"learning_rate": 6.200000000000001e-06,
|
| 1860 |
+
"logits/chosen": NaN,
|
| 1861 |
+
"logits/rejected": NaN,
|
| 1862 |
+
"logps/chosen": NaN,
|
| 1863 |
+
"logps/rejected": NaN,
|
| 1864 |
+
"loss": 0.0,
|
| 1865 |
+
"rewards/accuracies": 0.0,
|
| 1866 |
+
"rewards/chosen": NaN,
|
| 1867 |
+
"rewards/margins": NaN,
|
| 1868 |
+
"rewards/rejected": NaN,
|
| 1869 |
+
"step": 124
|
| 1870 |
+
},
|
| 1871 |
+
{
|
| 1872 |
+
"epoch": 0.03502136303144918,
|
| 1873 |
+
"grad_norm": NaN,
|
| 1874 |
+
"learning_rate": 6.25e-06,
|
| 1875 |
+
"logits/chosen": NaN,
|
| 1876 |
+
"logits/rejected": NaN,
|
| 1877 |
+
"logps/chosen": NaN,
|
| 1878 |
+
"logps/rejected": NaN,
|
| 1879 |
+
"loss": 0.0,
|
| 1880 |
+
"rewards/accuracies": 0.0,
|
| 1881 |
+
"rewards/chosen": NaN,
|
| 1882 |
+
"rewards/margins": NaN,
|
| 1883 |
+
"rewards/rejected": NaN,
|
| 1884 |
+
"step": 125
|
| 1885 |
+
},
|
| 1886 |
+
{
|
| 1887 |
+
"epoch": 0.03530153393570078,
|
| 1888 |
+
"grad_norm": NaN,
|
| 1889 |
+
"learning_rate": 6.300000000000001e-06,
|
| 1890 |
+
"logits/chosen": NaN,
|
| 1891 |
+
"logits/rejected": NaN,
|
| 1892 |
+
"logps/chosen": NaN,
|
| 1893 |
+
"logps/rejected": NaN,
|
| 1894 |
+
"loss": 0.0,
|
| 1895 |
+
"rewards/accuracies": 0.0,
|
| 1896 |
+
"rewards/chosen": NaN,
|
| 1897 |
+
"rewards/margins": NaN,
|
| 1898 |
+
"rewards/rejected": NaN,
|
| 1899 |
+
"step": 126
|
| 1900 |
+
},
|
| 1901 |
+
{
|
| 1902 |
+
"epoch": 0.03558170483995237,
|
| 1903 |
+
"grad_norm": NaN,
|
| 1904 |
+
"learning_rate": 6.35e-06,
|
| 1905 |
+
"logits/chosen": NaN,
|
| 1906 |
+
"logits/rejected": NaN,
|
| 1907 |
+
"logps/chosen": NaN,
|
| 1908 |
+
"logps/rejected": NaN,
|
| 1909 |
+
"loss": 0.0,
|
| 1910 |
+
"rewards/accuracies": 0.0,
|
| 1911 |
+
"rewards/chosen": NaN,
|
| 1912 |
+
"rewards/margins": NaN,
|
| 1913 |
+
"rewards/rejected": NaN,
|
| 1914 |
+
"step": 127
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 0.035861875744203964,
|
| 1918 |
+
"grad_norm": NaN,
|
| 1919 |
+
"learning_rate": 6.4000000000000006e-06,
|
| 1920 |
+
"logits/chosen": NaN,
|
| 1921 |
+
"logits/rejected": NaN,
|
| 1922 |
+
"logps/chosen": NaN,
|
| 1923 |
+
"logps/rejected": NaN,
|
| 1924 |
+
"loss": 0.0,
|
| 1925 |
+
"rewards/accuracies": 0.0,
|
| 1926 |
+
"rewards/chosen": NaN,
|
| 1927 |
+
"rewards/margins": NaN,
|
| 1928 |
+
"rewards/rejected": NaN,
|
| 1929 |
+
"step": 128
|
| 1930 |
+
},
|
| 1931 |
+
{
|
| 1932 |
+
"epoch": 0.03614204664845556,
|
| 1933 |
+
"grad_norm": NaN,
|
| 1934 |
+
"learning_rate": 6.450000000000001e-06,
|
| 1935 |
+
"logits/chosen": NaN,
|
| 1936 |
+
"logits/rejected": NaN,
|
| 1937 |
+
"logps/chosen": NaN,
|
| 1938 |
+
"logps/rejected": NaN,
|
| 1939 |
+
"loss": 0.0,
|
| 1940 |
+
"rewards/accuracies": 0.0,
|
| 1941 |
+
"rewards/chosen": NaN,
|
| 1942 |
+
"rewards/margins": NaN,
|
| 1943 |
+
"rewards/rejected": NaN,
|
| 1944 |
+
"step": 129
|
| 1945 |
+
},
|
| 1946 |
+
{
|
| 1947 |
+
"epoch": 0.03642221755270715,
|
| 1948 |
+
"grad_norm": NaN,
|
| 1949 |
+
"learning_rate": 6.5000000000000004e-06,
|
| 1950 |
+
"logits/chosen": NaN,
|
| 1951 |
+
"logits/rejected": NaN,
|
| 1952 |
+
"logps/chosen": NaN,
|
| 1953 |
+
"logps/rejected": NaN,
|
| 1954 |
+
"loss": 0.0,
|
| 1955 |
+
"rewards/accuracies": 0.0,
|
| 1956 |
+
"rewards/chosen": NaN,
|
| 1957 |
+
"rewards/margins": NaN,
|
| 1958 |
+
"rewards/rejected": NaN,
|
| 1959 |
+
"step": 130
|
| 1960 |
+
},
|
| 1961 |
+
{
|
| 1962 |
+
"epoch": 0.036702388456958744,
|
| 1963 |
+
"grad_norm": NaN,
|
| 1964 |
+
"learning_rate": 6.550000000000001e-06,
|
| 1965 |
+
"logits/chosen": NaN,
|
| 1966 |
+
"logits/rejected": NaN,
|
| 1967 |
+
"logps/chosen": NaN,
|
| 1968 |
+
"logps/rejected": NaN,
|
| 1969 |
+
"loss": 0.0,
|
| 1970 |
+
"rewards/accuracies": 0.0,
|
| 1971 |
+
"rewards/chosen": NaN,
|
| 1972 |
+
"rewards/margins": NaN,
|
| 1973 |
+
"rewards/rejected": NaN,
|
| 1974 |
+
"step": 131
|
| 1975 |
+
},
|
| 1976 |
+
{
|
| 1977 |
+
"epoch": 0.03698255936121034,
|
| 1978 |
+
"grad_norm": NaN,
|
| 1979 |
+
"learning_rate": 6.600000000000001e-06,
|
| 1980 |
+
"logits/chosen": NaN,
|
| 1981 |
+
"logits/rejected": NaN,
|
| 1982 |
+
"logps/chosen": NaN,
|
| 1983 |
+
"logps/rejected": NaN,
|
| 1984 |
+
"loss": 0.0,
|
| 1985 |
+
"rewards/accuracies": 0.0,
|
| 1986 |
+
"rewards/chosen": NaN,
|
| 1987 |
+
"rewards/margins": NaN,
|
| 1988 |
+
"rewards/rejected": NaN,
|
| 1989 |
+
"step": 132
|
| 1990 |
+
},
|
| 1991 |
+
{
|
| 1992 |
+
"epoch": 0.03726273026546193,
|
| 1993 |
+
"grad_norm": NaN,
|
| 1994 |
+
"learning_rate": 6.650000000000001e-06,
|
| 1995 |
+
"logits/chosen": NaN,
|
| 1996 |
+
"logits/rejected": NaN,
|
| 1997 |
+
"logps/chosen": NaN,
|
| 1998 |
+
"logps/rejected": NaN,
|
| 1999 |
+
"loss": 0.0,
|
| 2000 |
+
"rewards/accuracies": 0.0,
|
| 2001 |
+
"rewards/chosen": NaN,
|
| 2002 |
+
"rewards/margins": NaN,
|
| 2003 |
+
"rewards/rejected": NaN,
|
| 2004 |
+
"step": 133
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 0.037542901169713525,
|
| 2008 |
+
"grad_norm": NaN,
|
| 2009 |
+
"learning_rate": 6.700000000000001e-06,
|
| 2010 |
+
"logits/chosen": NaN,
|
| 2011 |
+
"logits/rejected": NaN,
|
| 2012 |
+
"logps/chosen": NaN,
|
| 2013 |
+
"logps/rejected": NaN,
|
| 2014 |
+
"loss": 0.0,
|
| 2015 |
+
"rewards/accuracies": 0.0,
|
| 2016 |
+
"rewards/chosen": NaN,
|
| 2017 |
+
"rewards/margins": NaN,
|
| 2018 |
+
"rewards/rejected": NaN,
|
| 2019 |
+
"step": 134
|
| 2020 |
+
},
|
| 2021 |
+
{
|
| 2022 |
+
"epoch": 0.03782307207396512,
|
| 2023 |
+
"grad_norm": NaN,
|
| 2024 |
+
"learning_rate": 6.750000000000001e-06,
|
| 2025 |
+
"logits/chosen": NaN,
|
| 2026 |
+
"logits/rejected": NaN,
|
| 2027 |
+
"logps/chosen": NaN,
|
| 2028 |
+
"logps/rejected": NaN,
|
| 2029 |
+
"loss": 0.0,
|
| 2030 |
+
"rewards/accuracies": 0.0,
|
| 2031 |
+
"rewards/chosen": NaN,
|
| 2032 |
+
"rewards/margins": NaN,
|
| 2033 |
+
"rewards/rejected": NaN,
|
| 2034 |
+
"step": 135
|
| 2035 |
+
},
|
| 2036 |
+
{
|
| 2037 |
+
"epoch": 0.03810324297821671,
|
| 2038 |
+
"grad_norm": NaN,
|
| 2039 |
+
"learning_rate": 6.800000000000001e-06,
|
| 2040 |
+
"logits/chosen": NaN,
|
| 2041 |
+
"logits/rejected": NaN,
|
| 2042 |
+
"logps/chosen": NaN,
|
| 2043 |
+
"logps/rejected": NaN,
|
| 2044 |
+
"loss": 0.0,
|
| 2045 |
+
"rewards/accuracies": 0.0,
|
| 2046 |
+
"rewards/chosen": NaN,
|
| 2047 |
+
"rewards/margins": NaN,
|
| 2048 |
+
"rewards/rejected": NaN,
|
| 2049 |
+
"step": 136
|
| 2050 |
+
},
|
| 2051 |
+
{
|
| 2052 |
+
"epoch": 0.038383413882468305,
|
| 2053 |
+
"grad_norm": NaN,
|
| 2054 |
+
"learning_rate": 6.850000000000001e-06,
|
| 2055 |
+
"logits/chosen": NaN,
|
| 2056 |
+
"logits/rejected": NaN,
|
| 2057 |
+
"logps/chosen": NaN,
|
| 2058 |
+
"logps/rejected": NaN,
|
| 2059 |
+
"loss": 0.0,
|
| 2060 |
+
"rewards/accuracies": 0.0,
|
| 2061 |
+
"rewards/chosen": NaN,
|
| 2062 |
+
"rewards/margins": NaN,
|
| 2063 |
+
"rewards/rejected": NaN,
|
| 2064 |
+
"step": 137
|
| 2065 |
+
},
|
| 2066 |
+
{
|
| 2067 |
+
"epoch": 0.0386635847867199,
|
| 2068 |
+
"grad_norm": NaN,
|
| 2069 |
+
"learning_rate": 6.9e-06,
|
| 2070 |
+
"logits/chosen": NaN,
|
| 2071 |
+
"logits/rejected": NaN,
|
| 2072 |
+
"logps/chosen": NaN,
|
| 2073 |
+
"logps/rejected": NaN,
|
| 2074 |
+
"loss": 0.0,
|
| 2075 |
+
"rewards/accuracies": 0.0,
|
| 2076 |
+
"rewards/chosen": NaN,
|
| 2077 |
+
"rewards/margins": NaN,
|
| 2078 |
+
"rewards/rejected": NaN,
|
| 2079 |
+
"step": 138
|
| 2080 |
+
},
|
| 2081 |
+
{
|
| 2082 |
+
"epoch": 0.03894375569097149,
|
| 2083 |
+
"grad_norm": NaN,
|
| 2084 |
+
"learning_rate": 6.95e-06,
|
| 2085 |
+
"logits/chosen": NaN,
|
| 2086 |
+
"logits/rejected": NaN,
|
| 2087 |
+
"logps/chosen": NaN,
|
| 2088 |
+
"logps/rejected": NaN,
|
| 2089 |
+
"loss": 0.0,
|
| 2090 |
+
"rewards/accuracies": 0.0,
|
| 2091 |
+
"rewards/chosen": NaN,
|
| 2092 |
+
"rewards/margins": NaN,
|
| 2093 |
+
"rewards/rejected": NaN,
|
| 2094 |
+
"step": 139
|
| 2095 |
+
},
|
| 2096 |
+
{
|
| 2097 |
+
"epoch": 0.039223926595223085,
|
| 2098 |
+
"grad_norm": NaN,
|
| 2099 |
+
"learning_rate": 7e-06,
|
| 2100 |
+
"logits/chosen": NaN,
|
| 2101 |
+
"logits/rejected": NaN,
|
| 2102 |
+
"logps/chosen": NaN,
|
| 2103 |
+
"logps/rejected": NaN,
|
| 2104 |
+
"loss": 0.0,
|
| 2105 |
+
"rewards/accuracies": 0.0,
|
| 2106 |
+
"rewards/chosen": NaN,
|
| 2107 |
+
"rewards/margins": NaN,
|
| 2108 |
+
"rewards/rejected": NaN,
|
| 2109 |
+
"step": 140
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 0.03950409749947468,
|
| 2113 |
+
"grad_norm": NaN,
|
| 2114 |
+
"learning_rate": 7.05e-06,
|
| 2115 |
+
"logits/chosen": NaN,
|
| 2116 |
+
"logits/rejected": NaN,
|
| 2117 |
+
"logps/chosen": NaN,
|
| 2118 |
+
"logps/rejected": NaN,
|
| 2119 |
+
"loss": 0.0,
|
| 2120 |
+
"rewards/accuracies": 0.0,
|
| 2121 |
+
"rewards/chosen": NaN,
|
| 2122 |
+
"rewards/margins": NaN,
|
| 2123 |
+
"rewards/rejected": NaN,
|
| 2124 |
+
"step": 141
|
| 2125 |
+
},
|
| 2126 |
+
{
|
| 2127 |
+
"epoch": 0.03978426840372627,
|
| 2128 |
+
"grad_norm": NaN,
|
| 2129 |
+
"learning_rate": 7.100000000000001e-06,
|
| 2130 |
+
"logits/chosen": NaN,
|
| 2131 |
+
"logits/rejected": NaN,
|
| 2132 |
+
"logps/chosen": NaN,
|
| 2133 |
+
"logps/rejected": NaN,
|
| 2134 |
+
"loss": 0.0,
|
| 2135 |
+
"rewards/accuracies": 0.0,
|
| 2136 |
+
"rewards/chosen": NaN,
|
| 2137 |
+
"rewards/margins": NaN,
|
| 2138 |
+
"rewards/rejected": NaN,
|
| 2139 |
+
"step": 142
|
| 2140 |
+
},
|
| 2141 |
+
{
|
| 2142 |
+
"epoch": 0.040064439307977866,
|
| 2143 |
+
"grad_norm": NaN,
|
| 2144 |
+
"learning_rate": 7.15e-06,
|
| 2145 |
+
"logits/chosen": NaN,
|
| 2146 |
+
"logits/rejected": NaN,
|
| 2147 |
+
"logps/chosen": NaN,
|
| 2148 |
+
"logps/rejected": NaN,
|
| 2149 |
+
"loss": 0.0,
|
| 2150 |
+
"rewards/accuracies": 0.0,
|
| 2151 |
+
"rewards/chosen": NaN,
|
| 2152 |
+
"rewards/margins": NaN,
|
| 2153 |
+
"rewards/rejected": NaN,
|
| 2154 |
+
"step": 143
|
| 2155 |
+
},
|
| 2156 |
+
{
|
| 2157 |
+
"epoch": 0.04034461021222946,
|
| 2158 |
+
"grad_norm": NaN,
|
| 2159 |
+
"learning_rate": 7.2000000000000005e-06,
|
| 2160 |
+
"logits/chosen": NaN,
|
| 2161 |
+
"logits/rejected": NaN,
|
| 2162 |
+
"logps/chosen": NaN,
|
| 2163 |
+
"logps/rejected": NaN,
|
| 2164 |
+
"loss": 0.0,
|
| 2165 |
+
"rewards/accuracies": 0.0,
|
| 2166 |
+
"rewards/chosen": NaN,
|
| 2167 |
+
"rewards/margins": NaN,
|
| 2168 |
+
"rewards/rejected": NaN,
|
| 2169 |
+
"step": 144
|
| 2170 |
+
},
|
| 2171 |
+
{
|
| 2172 |
+
"epoch": 0.04062478111648105,
|
| 2173 |
+
"grad_norm": NaN,
|
| 2174 |
+
"learning_rate": 7.25e-06,
|
| 2175 |
+
"logits/chosen": NaN,
|
| 2176 |
+
"logits/rejected": NaN,
|
| 2177 |
+
"logps/chosen": NaN,
|
| 2178 |
+
"logps/rejected": NaN,
|
| 2179 |
+
"loss": 0.0,
|
| 2180 |
+
"rewards/accuracies": 0.0,
|
| 2181 |
+
"rewards/chosen": NaN,
|
| 2182 |
+
"rewards/margins": NaN,
|
| 2183 |
+
"rewards/rejected": NaN,
|
| 2184 |
+
"step": 145
|
| 2185 |
+
},
|
| 2186 |
+
{
|
| 2187 |
+
"epoch": 0.040904952020732646,
|
| 2188 |
+
"grad_norm": NaN,
|
| 2189 |
+
"learning_rate": 7.3e-06,
|
| 2190 |
+
"logits/chosen": NaN,
|
| 2191 |
+
"logits/rejected": NaN,
|
| 2192 |
+
"logps/chosen": NaN,
|
| 2193 |
+
"logps/rejected": NaN,
|
| 2194 |
+
"loss": 0.0,
|
| 2195 |
+
"rewards/accuracies": 0.0,
|
| 2196 |
+
"rewards/chosen": NaN,
|
| 2197 |
+
"rewards/margins": NaN,
|
| 2198 |
+
"rewards/rejected": NaN,
|
| 2199 |
+
"step": 146
|
| 2200 |
+
},
|
| 2201 |
+
{
|
| 2202 |
+
"epoch": 0.04118512292498424,
|
| 2203 |
+
"grad_norm": NaN,
|
| 2204 |
+
"learning_rate": 7.350000000000001e-06,
|
| 2205 |
+
"logits/chosen": NaN,
|
| 2206 |
+
"logits/rejected": NaN,
|
| 2207 |
+
"logps/chosen": NaN,
|
| 2208 |
+
"logps/rejected": NaN,
|
| 2209 |
+
"loss": 0.0,
|
| 2210 |
+
"rewards/accuracies": 0.0,
|
| 2211 |
+
"rewards/chosen": NaN,
|
| 2212 |
+
"rewards/margins": NaN,
|
| 2213 |
+
"rewards/rejected": NaN,
|
| 2214 |
+
"step": 147
|
| 2215 |
+
},
|
| 2216 |
+
{
|
| 2217 |
+
"epoch": 0.04146529382923583,
|
| 2218 |
+
"grad_norm": NaN,
|
| 2219 |
+
"learning_rate": 7.4e-06,
|
| 2220 |
+
"logits/chosen": NaN,
|
| 2221 |
+
"logits/rejected": NaN,
|
| 2222 |
+
"logps/chosen": NaN,
|
| 2223 |
+
"logps/rejected": NaN,
|
| 2224 |
+
"loss": 0.0,
|
| 2225 |
+
"rewards/accuracies": 0.0,
|
| 2226 |
+
"rewards/chosen": NaN,
|
| 2227 |
+
"rewards/margins": NaN,
|
| 2228 |
+
"rewards/rejected": NaN,
|
| 2229 |
+
"step": 148
|
| 2230 |
+
},
|
| 2231 |
+
{
|
| 2232 |
+
"epoch": 0.04174546473348743,
|
| 2233 |
+
"grad_norm": NaN,
|
| 2234 |
+
"learning_rate": 7.450000000000001e-06,
|
| 2235 |
+
"logits/chosen": NaN,
|
| 2236 |
+
"logits/rejected": NaN,
|
| 2237 |
+
"logps/chosen": NaN,
|
| 2238 |
+
"logps/rejected": NaN,
|
| 2239 |
+
"loss": 0.0,
|
| 2240 |
+
"rewards/accuracies": 0.0,
|
| 2241 |
+
"rewards/chosen": NaN,
|
| 2242 |
+
"rewards/margins": NaN,
|
| 2243 |
+
"rewards/rejected": NaN,
|
| 2244 |
+
"step": 149
|
| 2245 |
+
},
|
| 2246 |
+
{
|
| 2247 |
+
"epoch": 0.04202563563773902,
|
| 2248 |
+
"grad_norm": NaN,
|
| 2249 |
+
"learning_rate": 7.500000000000001e-06,
|
| 2250 |
+
"logits/chosen": NaN,
|
| 2251 |
+
"logits/rejected": NaN,
|
| 2252 |
+
"logps/chosen": NaN,
|
| 2253 |
+
"logps/rejected": NaN,
|
| 2254 |
+
"loss": 0.0,
|
| 2255 |
+
"rewards/accuracies": 0.0,
|
| 2256 |
+
"rewards/chosen": NaN,
|
| 2257 |
+
"rewards/margins": NaN,
|
| 2258 |
+
"rewards/rejected": NaN,
|
| 2259 |
+
"step": 150
|
| 2260 |
+
},
|
| 2261 |
+
{
|
| 2262 |
+
"epoch": 0.04230580654199061,
|
| 2263 |
+
"grad_norm": NaN,
|
| 2264 |
+
"learning_rate": 7.5500000000000006e-06,
|
| 2265 |
+
"logits/chosen": NaN,
|
| 2266 |
+
"logits/rejected": NaN,
|
| 2267 |
+
"logps/chosen": NaN,
|
| 2268 |
+
"logps/rejected": NaN,
|
| 2269 |
+
"loss": 0.0,
|
| 2270 |
+
"rewards/accuracies": 0.0,
|
| 2271 |
+
"rewards/chosen": NaN,
|
| 2272 |
+
"rewards/margins": NaN,
|
| 2273 |
+
"rewards/rejected": NaN,
|
| 2274 |
+
"step": 151
|
| 2275 |
+
},
|
| 2276 |
+
{
|
| 2277 |
+
"epoch": 0.04258597744624221,
|
| 2278 |
+
"grad_norm": NaN,
|
| 2279 |
+
"learning_rate": 7.600000000000001e-06,
|
| 2280 |
+
"logits/chosen": NaN,
|
| 2281 |
+
"logits/rejected": NaN,
|
| 2282 |
+
"logps/chosen": NaN,
|
| 2283 |
+
"logps/rejected": NaN,
|
| 2284 |
+
"loss": 0.0,
|
| 2285 |
+
"rewards/accuracies": 0.0,
|
| 2286 |
+
"rewards/chosen": NaN,
|
| 2287 |
+
"rewards/margins": NaN,
|
| 2288 |
+
"rewards/rejected": NaN,
|
| 2289 |
+
"step": 152
|
| 2290 |
+
},
|
| 2291 |
+
{
|
| 2292 |
+
"epoch": 0.0428661483504938,
|
| 2293 |
+
"grad_norm": NaN,
|
| 2294 |
+
"learning_rate": 7.650000000000001e-06,
|
| 2295 |
+
"logits/chosen": NaN,
|
| 2296 |
+
"logits/rejected": NaN,
|
| 2297 |
+
"logps/chosen": NaN,
|
| 2298 |
+
"logps/rejected": NaN,
|
| 2299 |
+
"loss": 0.0,
|
| 2300 |
+
"rewards/accuracies": 0.0,
|
| 2301 |
+
"rewards/chosen": NaN,
|
| 2302 |
+
"rewards/margins": NaN,
|
| 2303 |
+
"rewards/rejected": NaN,
|
| 2304 |
+
"step": 153
|
| 2305 |
+
},
|
| 2306 |
+
{
|
| 2307 |
+
"epoch": 0.043146319254745394,
|
| 2308 |
+
"grad_norm": NaN,
|
| 2309 |
+
"learning_rate": 7.7e-06,
|
| 2310 |
+
"logits/chosen": NaN,
|
| 2311 |
+
"logits/rejected": NaN,
|
| 2312 |
+
"logps/chosen": NaN,
|
| 2313 |
+
"logps/rejected": NaN,
|
| 2314 |
+
"loss": 0.0,
|
| 2315 |
+
"rewards/accuracies": 0.0,
|
| 2316 |
+
"rewards/chosen": NaN,
|
| 2317 |
+
"rewards/margins": NaN,
|
| 2318 |
+
"rewards/rejected": NaN,
|
| 2319 |
+
"step": 154
|
| 2320 |
+
},
|
| 2321 |
+
{
|
| 2322 |
+
"epoch": 0.04342649015899699,
|
| 2323 |
+
"grad_norm": NaN,
|
| 2324 |
+
"learning_rate": 7.75e-06,
|
| 2325 |
+
"logits/chosen": NaN,
|
| 2326 |
+
"logits/rejected": NaN,
|
| 2327 |
+
"logps/chosen": NaN,
|
| 2328 |
+
"logps/rejected": NaN,
|
| 2329 |
+
"loss": 0.0,
|
| 2330 |
+
"rewards/accuracies": 0.0,
|
| 2331 |
+
"rewards/chosen": NaN,
|
| 2332 |
+
"rewards/margins": NaN,
|
| 2333 |
+
"rewards/rejected": NaN,
|
| 2334 |
+
"step": 155
|
| 2335 |
+
},
|
| 2336 |
+
{
|
| 2337 |
+
"epoch": 0.04370666106324858,
|
| 2338 |
+
"grad_norm": NaN,
|
| 2339 |
+
"learning_rate": 7.800000000000002e-06,
|
| 2340 |
+
"logits/chosen": NaN,
|
| 2341 |
+
"logits/rejected": NaN,
|
| 2342 |
+
"logps/chosen": NaN,
|
| 2343 |
+
"logps/rejected": NaN,
|
| 2344 |
+
"loss": 0.0,
|
| 2345 |
+
"rewards/accuracies": 0.0,
|
| 2346 |
+
"rewards/chosen": NaN,
|
| 2347 |
+
"rewards/margins": NaN,
|
| 2348 |
+
"rewards/rejected": NaN,
|
| 2349 |
+
"step": 156
|
| 2350 |
+
},
|
| 2351 |
+
{
|
| 2352 |
+
"epoch": 0.043986831967500174,
|
| 2353 |
+
"grad_norm": NaN,
|
| 2354 |
+
"learning_rate": 7.850000000000001e-06,
|
| 2355 |
+
"logits/chosen": NaN,
|
| 2356 |
+
"logits/rejected": NaN,
|
| 2357 |
+
"logps/chosen": NaN,
|
| 2358 |
+
"logps/rejected": NaN,
|
| 2359 |
+
"loss": 0.0,
|
| 2360 |
+
"rewards/accuracies": 0.0,
|
| 2361 |
+
"rewards/chosen": NaN,
|
| 2362 |
+
"rewards/margins": NaN,
|
| 2363 |
+
"rewards/rejected": NaN,
|
| 2364 |
+
"step": 157
|
| 2365 |
+
},
|
| 2366 |
+
{
|
| 2367 |
+
"epoch": 0.04426700287175177,
|
| 2368 |
+
"grad_norm": NaN,
|
| 2369 |
+
"learning_rate": 7.9e-06,
|
| 2370 |
+
"logits/chosen": NaN,
|
| 2371 |
+
"logits/rejected": NaN,
|
| 2372 |
+
"logps/chosen": NaN,
|
| 2373 |
+
"logps/rejected": NaN,
|
| 2374 |
+
"loss": 0.0,
|
| 2375 |
+
"rewards/accuracies": 0.0,
|
| 2376 |
+
"rewards/chosen": NaN,
|
| 2377 |
+
"rewards/margins": NaN,
|
| 2378 |
+
"rewards/rejected": NaN,
|
| 2379 |
+
"step": 158
|
| 2380 |
+
},
|
| 2381 |
+
{
|
| 2382 |
+
"epoch": 0.04454717377600336,
|
| 2383 |
+
"grad_norm": NaN,
|
| 2384 |
+
"learning_rate": 7.950000000000002e-06,
|
| 2385 |
+
"logits/chosen": NaN,
|
| 2386 |
+
"logits/rejected": NaN,
|
| 2387 |
+
"logps/chosen": NaN,
|
| 2388 |
+
"logps/rejected": NaN,
|
| 2389 |
+
"loss": 0.0,
|
| 2390 |
+
"rewards/accuracies": 0.0,
|
| 2391 |
+
"rewards/chosen": NaN,
|
| 2392 |
+
"rewards/margins": NaN,
|
| 2393 |
+
"rewards/rejected": NaN,
|
| 2394 |
+
"step": 159
|
| 2395 |
+
},
|
| 2396 |
+
{
|
| 2397 |
+
"epoch": 0.044827344680254955,
|
| 2398 |
+
"grad_norm": NaN,
|
| 2399 |
+
"learning_rate": 8.000000000000001e-06,
|
| 2400 |
+
"logits/chosen": NaN,
|
| 2401 |
+
"logits/rejected": NaN,
|
| 2402 |
+
"logps/chosen": NaN,
|
| 2403 |
+
"logps/rejected": NaN,
|
| 2404 |
+
"loss": 0.0,
|
| 2405 |
+
"rewards/accuracies": 0.0,
|
| 2406 |
+
"rewards/chosen": NaN,
|
| 2407 |
+
"rewards/margins": NaN,
|
| 2408 |
+
"rewards/rejected": NaN,
|
| 2409 |
+
"step": 160
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 0.04510751558450655,
|
| 2413 |
+
"grad_norm": NaN,
|
| 2414 |
+
"learning_rate": 8.050000000000001e-06,
|
| 2415 |
+
"logits/chosen": NaN,
|
| 2416 |
+
"logits/rejected": NaN,
|
| 2417 |
+
"logps/chosen": NaN,
|
| 2418 |
+
"logps/rejected": NaN,
|
| 2419 |
+
"loss": 0.0,
|
| 2420 |
+
"rewards/accuracies": 0.0,
|
| 2421 |
+
"rewards/chosen": NaN,
|
| 2422 |
+
"rewards/margins": NaN,
|
| 2423 |
+
"rewards/rejected": NaN,
|
| 2424 |
+
"step": 161
|
| 2425 |
+
},
|
| 2426 |
+
{
|
| 2427 |
+
"epoch": 0.04538768648875814,
|
| 2428 |
+
"grad_norm": NaN,
|
| 2429 |
+
"learning_rate": 8.1e-06,
|
| 2430 |
+
"logits/chosen": NaN,
|
| 2431 |
+
"logits/rejected": NaN,
|
| 2432 |
+
"logps/chosen": NaN,
|
| 2433 |
+
"logps/rejected": NaN,
|
| 2434 |
+
"loss": 0.0,
|
| 2435 |
+
"rewards/accuracies": 0.0,
|
| 2436 |
+
"rewards/chosen": NaN,
|
| 2437 |
+
"rewards/margins": NaN,
|
| 2438 |
+
"rewards/rejected": NaN,
|
| 2439 |
+
"step": 162
|
| 2440 |
+
},
|
| 2441 |
+
{
|
| 2442 |
+
"epoch": 0.045667857393009735,
|
| 2443 |
+
"grad_norm": NaN,
|
| 2444 |
+
"learning_rate": 8.15e-06,
|
| 2445 |
+
"logits/chosen": NaN,
|
| 2446 |
+
"logits/rejected": NaN,
|
| 2447 |
+
"logps/chosen": NaN,
|
| 2448 |
+
"logps/rejected": NaN,
|
| 2449 |
+
"loss": 0.0,
|
| 2450 |
+
"rewards/accuracies": 0.0,
|
| 2451 |
+
"rewards/chosen": NaN,
|
| 2452 |
+
"rewards/margins": NaN,
|
| 2453 |
+
"rewards/rejected": NaN,
|
| 2454 |
+
"step": 163
|
| 2455 |
+
},
|
| 2456 |
+
{
|
| 2457 |
+
"epoch": 0.04594802829726133,
|
| 2458 |
+
"grad_norm": NaN,
|
| 2459 |
+
"learning_rate": 8.2e-06,
|
| 2460 |
+
"logits/chosen": NaN,
|
| 2461 |
+
"logits/rejected": NaN,
|
| 2462 |
+
"logps/chosen": NaN,
|
| 2463 |
+
"logps/rejected": NaN,
|
| 2464 |
+
"loss": 0.0,
|
| 2465 |
+
"rewards/accuracies": 0.0,
|
| 2466 |
+
"rewards/chosen": NaN,
|
| 2467 |
+
"rewards/margins": NaN,
|
| 2468 |
+
"rewards/rejected": NaN,
|
| 2469 |
+
"step": 164
|
| 2470 |
+
},
|
| 2471 |
+
{
|
| 2472 |
+
"epoch": 0.04622819920151292,
|
| 2473 |
+
"grad_norm": NaN,
|
| 2474 |
+
"learning_rate": 8.25e-06,
|
| 2475 |
+
"logits/chosen": NaN,
|
| 2476 |
+
"logits/rejected": NaN,
|
| 2477 |
+
"logps/chosen": NaN,
|
| 2478 |
+
"logps/rejected": NaN,
|
| 2479 |
+
"loss": 0.0,
|
| 2480 |
+
"rewards/accuracies": 0.0,
|
| 2481 |
+
"rewards/chosen": NaN,
|
| 2482 |
+
"rewards/margins": NaN,
|
| 2483 |
+
"rewards/rejected": NaN,
|
| 2484 |
+
"step": 165
|
| 2485 |
+
},
|
| 2486 |
+
{
|
| 2487 |
+
"epoch": 0.046508370105764515,
|
| 2488 |
+
"grad_norm": NaN,
|
| 2489 |
+
"learning_rate": 8.3e-06,
|
| 2490 |
+
"logits/chosen": NaN,
|
| 2491 |
+
"logits/rejected": NaN,
|
| 2492 |
+
"logps/chosen": NaN,
|
| 2493 |
+
"logps/rejected": NaN,
|
| 2494 |
+
"loss": 0.0,
|
| 2495 |
+
"rewards/accuracies": 0.0,
|
| 2496 |
+
"rewards/chosen": NaN,
|
| 2497 |
+
"rewards/margins": NaN,
|
| 2498 |
+
"rewards/rejected": NaN,
|
| 2499 |
+
"step": 166
|
| 2500 |
+
},
|
| 2501 |
+
{
|
| 2502 |
+
"epoch": 0.04678854101001611,
|
| 2503 |
+
"grad_norm": NaN,
|
| 2504 |
+
"learning_rate": 8.35e-06,
|
| 2505 |
+
"logits/chosen": NaN,
|
| 2506 |
+
"logits/rejected": NaN,
|
| 2507 |
+
"logps/chosen": NaN,
|
| 2508 |
+
"logps/rejected": NaN,
|
| 2509 |
+
"loss": 0.0,
|
| 2510 |
+
"rewards/accuracies": 0.0,
|
| 2511 |
+
"rewards/chosen": NaN,
|
| 2512 |
+
"rewards/margins": NaN,
|
| 2513 |
+
"rewards/rejected": NaN,
|
| 2514 |
+
"step": 167
|
| 2515 |
+
},
|
| 2516 |
+
{
|
| 2517 |
+
"epoch": 0.0470687119142677,
|
| 2518 |
+
"grad_norm": NaN,
|
| 2519 |
+
"learning_rate": 8.400000000000001e-06,
|
| 2520 |
+
"logits/chosen": NaN,
|
| 2521 |
+
"logits/rejected": NaN,
|
| 2522 |
+
"logps/chosen": NaN,
|
| 2523 |
+
"logps/rejected": NaN,
|
| 2524 |
+
"loss": 0.0,
|
| 2525 |
+
"rewards/accuracies": 0.0,
|
| 2526 |
+
"rewards/chosen": NaN,
|
| 2527 |
+
"rewards/margins": NaN,
|
| 2528 |
+
"rewards/rejected": NaN,
|
| 2529 |
+
"step": 168
|
| 2530 |
+
},
|
| 2531 |
+
{
|
| 2532 |
+
"epoch": 0.047348882818519296,
|
| 2533 |
+
"grad_norm": NaN,
|
| 2534 |
+
"learning_rate": 8.45e-06,
|
| 2535 |
+
"logits/chosen": NaN,
|
| 2536 |
+
"logits/rejected": NaN,
|
| 2537 |
+
"logps/chosen": NaN,
|
| 2538 |
+
"logps/rejected": NaN,
|
| 2539 |
+
"loss": 0.0,
|
| 2540 |
+
"rewards/accuracies": 0.0,
|
| 2541 |
+
"rewards/chosen": NaN,
|
| 2542 |
+
"rewards/margins": NaN,
|
| 2543 |
+
"rewards/rejected": NaN,
|
| 2544 |
+
"step": 169
|
| 2545 |
+
},
|
| 2546 |
+
{
|
| 2547 |
+
"epoch": 0.04762905372277089,
|
| 2548 |
+
"grad_norm": NaN,
|
| 2549 |
+
"learning_rate": 8.5e-06,
|
| 2550 |
+
"logits/chosen": NaN,
|
| 2551 |
+
"logits/rejected": NaN,
|
| 2552 |
+
"logps/chosen": NaN,
|
| 2553 |
+
"logps/rejected": NaN,
|
| 2554 |
+
"loss": 0.0,
|
| 2555 |
+
"rewards/accuracies": 0.0,
|
| 2556 |
+
"rewards/chosen": NaN,
|
| 2557 |
+
"rewards/margins": NaN,
|
| 2558 |
+
"rewards/rejected": NaN,
|
| 2559 |
+
"step": 170
|
| 2560 |
+
},
|
| 2561 |
+
{
|
| 2562 |
+
"epoch": 0.04790922462702248,
|
| 2563 |
+
"grad_norm": NaN,
|
| 2564 |
+
"learning_rate": 8.550000000000001e-06,
|
| 2565 |
+
"logits/chosen": NaN,
|
| 2566 |
+
"logits/rejected": NaN,
|
| 2567 |
+
"logps/chosen": NaN,
|
| 2568 |
+
"logps/rejected": NaN,
|
| 2569 |
+
"loss": 0.0,
|
| 2570 |
+
"rewards/accuracies": 0.0,
|
| 2571 |
+
"rewards/chosen": NaN,
|
| 2572 |
+
"rewards/margins": NaN,
|
| 2573 |
+
"rewards/rejected": NaN,
|
| 2574 |
+
"step": 171
|
| 2575 |
+
},
|
| 2576 |
+
{
|
| 2577 |
+
"epoch": 0.048189395531274076,
|
| 2578 |
+
"grad_norm": NaN,
|
| 2579 |
+
"learning_rate": 8.6e-06,
|
| 2580 |
+
"logits/chosen": NaN,
|
| 2581 |
+
"logits/rejected": NaN,
|
| 2582 |
+
"logps/chosen": NaN,
|
| 2583 |
+
"logps/rejected": NaN,
|
| 2584 |
+
"loss": 0.0,
|
| 2585 |
+
"rewards/accuracies": 0.0,
|
| 2586 |
+
"rewards/chosen": NaN,
|
| 2587 |
+
"rewards/margins": NaN,
|
| 2588 |
+
"rewards/rejected": NaN,
|
| 2589 |
+
"step": 172
|
| 2590 |
+
},
|
| 2591 |
+
{
|
| 2592 |
+
"epoch": 0.04846956643552567,
|
| 2593 |
+
"grad_norm": NaN,
|
| 2594 |
+
"learning_rate": 8.65e-06,
|
| 2595 |
+
"logits/chosen": NaN,
|
| 2596 |
+
"logits/rejected": NaN,
|
| 2597 |
+
"logps/chosen": NaN,
|
| 2598 |
+
"logps/rejected": NaN,
|
| 2599 |
+
"loss": 0.0,
|
| 2600 |
+
"rewards/accuracies": 0.0,
|
| 2601 |
+
"rewards/chosen": NaN,
|
| 2602 |
+
"rewards/margins": NaN,
|
| 2603 |
+
"rewards/rejected": NaN,
|
| 2604 |
+
"step": 173
|
| 2605 |
+
},
|
| 2606 |
+
{
|
| 2607 |
+
"epoch": 0.04874973733977726,
|
| 2608 |
+
"grad_norm": NaN,
|
| 2609 |
+
"learning_rate": 8.700000000000001e-06,
|
| 2610 |
+
"logits/chosen": NaN,
|
| 2611 |
+
"logits/rejected": NaN,
|
| 2612 |
+
"logps/chosen": NaN,
|
| 2613 |
+
"logps/rejected": NaN,
|
| 2614 |
+
"loss": 0.0,
|
| 2615 |
+
"rewards/accuracies": 0.0,
|
| 2616 |
+
"rewards/chosen": NaN,
|
| 2617 |
+
"rewards/margins": NaN,
|
| 2618 |
+
"rewards/rejected": NaN,
|
| 2619 |
+
"step": 174
|
| 2620 |
+
},
|
| 2621 |
+
{
|
| 2622 |
+
"epoch": 0.04902990824402886,
|
| 2623 |
+
"grad_norm": NaN,
|
| 2624 |
+
"learning_rate": 8.750000000000001e-06,
|
| 2625 |
+
"logits/chosen": NaN,
|
| 2626 |
+
"logits/rejected": NaN,
|
| 2627 |
+
"logps/chosen": NaN,
|
| 2628 |
+
"logps/rejected": NaN,
|
| 2629 |
+
"loss": 0.0,
|
| 2630 |
+
"rewards/accuracies": 0.0,
|
| 2631 |
+
"rewards/chosen": NaN,
|
| 2632 |
+
"rewards/margins": NaN,
|
| 2633 |
+
"rewards/rejected": NaN,
|
| 2634 |
+
"step": 175
|
| 2635 |
+
},
|
| 2636 |
+
{
|
| 2637 |
+
"epoch": 0.04931007914828045,
|
| 2638 |
+
"grad_norm": NaN,
|
| 2639 |
+
"learning_rate": 8.8e-06,
|
| 2640 |
+
"logits/chosen": NaN,
|
| 2641 |
+
"logits/rejected": NaN,
|
| 2642 |
+
"logps/chosen": NaN,
|
| 2643 |
+
"logps/rejected": NaN,
|
| 2644 |
+
"loss": 0.0,
|
| 2645 |
+
"rewards/accuracies": 0.0,
|
| 2646 |
+
"rewards/chosen": NaN,
|
| 2647 |
+
"rewards/margins": NaN,
|
| 2648 |
+
"rewards/rejected": NaN,
|
| 2649 |
+
"step": 176
|
| 2650 |
+
},
|
| 2651 |
+
{
|
| 2652 |
+
"epoch": 0.049590250052532044,
|
| 2653 |
+
"grad_norm": NaN,
|
| 2654 |
+
"learning_rate": 8.85e-06,
|
| 2655 |
+
"logits/chosen": NaN,
|
| 2656 |
+
"logits/rejected": NaN,
|
| 2657 |
+
"logps/chosen": NaN,
|
| 2658 |
+
"logps/rejected": NaN,
|
| 2659 |
+
"loss": 0.0,
|
| 2660 |
+
"rewards/accuracies": 0.0,
|
| 2661 |
+
"rewards/chosen": NaN,
|
| 2662 |
+
"rewards/margins": NaN,
|
| 2663 |
+
"rewards/rejected": NaN,
|
| 2664 |
+
"step": 177
|
| 2665 |
+
},
|
| 2666 |
+
{
|
| 2667 |
+
"epoch": 0.04987042095678364,
|
| 2668 |
+
"grad_norm": NaN,
|
| 2669 |
+
"learning_rate": 8.900000000000001e-06,
|
| 2670 |
+
"logits/chosen": NaN,
|
| 2671 |
+
"logits/rejected": NaN,
|
| 2672 |
+
"logps/chosen": NaN,
|
| 2673 |
+
"logps/rejected": NaN,
|
| 2674 |
+
"loss": 0.0,
|
| 2675 |
+
"rewards/accuracies": 0.0,
|
| 2676 |
+
"rewards/chosen": NaN,
|
| 2677 |
+
"rewards/margins": NaN,
|
| 2678 |
+
"rewards/rejected": NaN,
|
| 2679 |
+
"step": 178
|
| 2680 |
+
},
|
| 2681 |
+
{
|
| 2682 |
+
"epoch": 0.05015059186103523,
|
| 2683 |
+
"grad_norm": NaN,
|
| 2684 |
+
"learning_rate": 8.95e-06,
|
| 2685 |
+
"logits/chosen": NaN,
|
| 2686 |
+
"logits/rejected": NaN,
|
| 2687 |
+
"logps/chosen": NaN,
|
| 2688 |
+
"logps/rejected": NaN,
|
| 2689 |
+
"loss": 0.0,
|
| 2690 |
+
"rewards/accuracies": 0.0,
|
| 2691 |
+
"rewards/chosen": NaN,
|
| 2692 |
+
"rewards/margins": NaN,
|
| 2693 |
+
"rewards/rejected": NaN,
|
| 2694 |
+
"step": 179
|
| 2695 |
+
},
|
| 2696 |
+
{
|
| 2697 |
+
"epoch": 0.050430762765286824,
|
| 2698 |
+
"grad_norm": NaN,
|
| 2699 |
+
"learning_rate": 9e-06,
|
| 2700 |
+
"logits/chosen": NaN,
|
| 2701 |
+
"logits/rejected": NaN,
|
| 2702 |
+
"logps/chosen": NaN,
|
| 2703 |
+
"logps/rejected": NaN,
|
| 2704 |
+
"loss": 0.0,
|
| 2705 |
+
"rewards/accuracies": 0.0,
|
| 2706 |
+
"rewards/chosen": NaN,
|
| 2707 |
+
"rewards/margins": NaN,
|
| 2708 |
+
"rewards/rejected": NaN,
|
| 2709 |
+
"step": 180
|
| 2710 |
+
},
|
| 2711 |
+
{
|
| 2712 |
+
"epoch": 0.05071093366953842,
|
| 2713 |
+
"grad_norm": NaN,
|
| 2714 |
+
"learning_rate": 9.050000000000001e-06,
|
| 2715 |
+
"logits/chosen": NaN,
|
| 2716 |
+
"logits/rejected": NaN,
|
| 2717 |
+
"logps/chosen": NaN,
|
| 2718 |
+
"logps/rejected": NaN,
|
| 2719 |
+
"loss": 0.0,
|
| 2720 |
+
"rewards/accuracies": 0.0,
|
| 2721 |
+
"rewards/chosen": NaN,
|
| 2722 |
+
"rewards/margins": NaN,
|
| 2723 |
+
"rewards/rejected": NaN,
|
| 2724 |
+
"step": 181
|
| 2725 |
+
},
|
| 2726 |
+
{
|
| 2727 |
+
"epoch": 0.05099110457379001,
|
| 2728 |
+
"grad_norm": NaN,
|
| 2729 |
+
"learning_rate": 9.100000000000001e-06,
|
| 2730 |
+
"logits/chosen": NaN,
|
| 2731 |
+
"logits/rejected": NaN,
|
| 2732 |
+
"logps/chosen": NaN,
|
| 2733 |
+
"logps/rejected": NaN,
|
| 2734 |
+
"loss": 0.0,
|
| 2735 |
+
"rewards/accuracies": 0.0,
|
| 2736 |
+
"rewards/chosen": NaN,
|
| 2737 |
+
"rewards/margins": NaN,
|
| 2738 |
+
"rewards/rejected": NaN,
|
| 2739 |
+
"step": 182
|
| 2740 |
+
},
|
| 2741 |
+
{
|
| 2742 |
+
"epoch": 0.051271275478041604,
|
| 2743 |
+
"grad_norm": NaN,
|
| 2744 |
+
"learning_rate": 9.15e-06,
|
| 2745 |
+
"logits/chosen": NaN,
|
| 2746 |
+
"logits/rejected": NaN,
|
| 2747 |
+
"logps/chosen": NaN,
|
| 2748 |
+
"logps/rejected": NaN,
|
| 2749 |
+
"loss": 0.0,
|
| 2750 |
+
"rewards/accuracies": 0.0,
|
| 2751 |
+
"rewards/chosen": NaN,
|
| 2752 |
+
"rewards/margins": NaN,
|
| 2753 |
+
"rewards/rejected": NaN,
|
| 2754 |
+
"step": 183
|
| 2755 |
+
},
|
| 2756 |
+
{
|
| 2757 |
+
"epoch": 0.0515514463822932,
|
| 2758 |
+
"grad_norm": NaN,
|
| 2759 |
+
"learning_rate": 9.200000000000002e-06,
|
| 2760 |
+
"logits/chosen": NaN,
|
| 2761 |
+
"logits/rejected": NaN,
|
| 2762 |
+
"logps/chosen": NaN,
|
| 2763 |
+
"logps/rejected": NaN,
|
| 2764 |
+
"loss": 0.0,
|
| 2765 |
+
"rewards/accuracies": 0.0,
|
| 2766 |
+
"rewards/chosen": NaN,
|
| 2767 |
+
"rewards/margins": NaN,
|
| 2768 |
+
"rewards/rejected": NaN,
|
| 2769 |
+
"step": 184
|
| 2770 |
+
},
|
| 2771 |
+
{
|
| 2772 |
+
"epoch": 0.05183161728654479,
|
| 2773 |
+
"grad_norm": NaN,
|
| 2774 |
+
"learning_rate": 9.250000000000001e-06,
|
| 2775 |
+
"logits/chosen": NaN,
|
| 2776 |
+
"logits/rejected": NaN,
|
| 2777 |
+
"logps/chosen": NaN,
|
| 2778 |
+
"logps/rejected": NaN,
|
| 2779 |
+
"loss": 0.0,
|
| 2780 |
+
"rewards/accuracies": 0.0,
|
| 2781 |
+
"rewards/chosen": NaN,
|
| 2782 |
+
"rewards/margins": NaN,
|
| 2783 |
+
"rewards/rejected": NaN,
|
| 2784 |
+
"step": 185
|
| 2785 |
+
},
|
| 2786 |
+
{
|
| 2787 |
+
"epoch": 0.052111788190796385,
|
| 2788 |
+
"grad_norm": NaN,
|
| 2789 |
+
"learning_rate": 9.3e-06,
|
| 2790 |
+
"logits/chosen": NaN,
|
| 2791 |
+
"logits/rejected": NaN,
|
| 2792 |
+
"logps/chosen": NaN,
|
| 2793 |
+
"logps/rejected": NaN,
|
| 2794 |
+
"loss": 0.0,
|
| 2795 |
+
"rewards/accuracies": 0.0,
|
| 2796 |
+
"rewards/chosen": NaN,
|
| 2797 |
+
"rewards/margins": NaN,
|
| 2798 |
+
"rewards/rejected": NaN,
|
| 2799 |
+
"step": 186
|
| 2800 |
+
},
|
| 2801 |
+
{
|
| 2802 |
+
"epoch": 0.05239195909504798,
|
| 2803 |
+
"grad_norm": NaN,
|
| 2804 |
+
"learning_rate": 9.350000000000002e-06,
|
| 2805 |
+
"logits/chosen": NaN,
|
| 2806 |
+
"logits/rejected": NaN,
|
| 2807 |
+
"logps/chosen": NaN,
|
| 2808 |
+
"logps/rejected": NaN,
|
| 2809 |
+
"loss": 0.0,
|
| 2810 |
+
"rewards/accuracies": 0.0,
|
| 2811 |
+
"rewards/chosen": NaN,
|
| 2812 |
+
"rewards/margins": NaN,
|
| 2813 |
+
"rewards/rejected": NaN,
|
| 2814 |
+
"step": 187
|
| 2815 |
+
},
|
| 2816 |
+
{
|
| 2817 |
+
"epoch": 0.05267212999929957,
|
| 2818 |
+
"grad_norm": NaN,
|
| 2819 |
+
"learning_rate": 9.4e-06,
|
| 2820 |
+
"logits/chosen": NaN,
|
| 2821 |
+
"logits/rejected": NaN,
|
| 2822 |
+
"logps/chosen": NaN,
|
| 2823 |
+
"logps/rejected": NaN,
|
| 2824 |
+
"loss": 0.0,
|
| 2825 |
+
"rewards/accuracies": 0.0,
|
| 2826 |
+
"rewards/chosen": NaN,
|
| 2827 |
+
"rewards/margins": NaN,
|
| 2828 |
+
"rewards/rejected": NaN,
|
| 2829 |
+
"step": 188
|
| 2830 |
+
},
|
| 2831 |
+
{
|
| 2832 |
+
"epoch": 0.052952300903551165,
|
| 2833 |
+
"grad_norm": NaN,
|
| 2834 |
+
"learning_rate": 9.450000000000001e-06,
|
| 2835 |
+
"logits/chosen": NaN,
|
| 2836 |
+
"logits/rejected": NaN,
|
| 2837 |
+
"logps/chosen": NaN,
|
| 2838 |
+
"logps/rejected": NaN,
|
| 2839 |
+
"loss": 0.0,
|
| 2840 |
+
"rewards/accuracies": 0.0,
|
| 2841 |
+
"rewards/chosen": NaN,
|
| 2842 |
+
"rewards/margins": NaN,
|
| 2843 |
+
"rewards/rejected": NaN,
|
| 2844 |
+
"step": 189
|
| 2845 |
+
},
|
| 2846 |
+
{
|
| 2847 |
+
"epoch": 0.05323247180780276,
|
| 2848 |
+
"grad_norm": NaN,
|
| 2849 |
+
"learning_rate": 9.5e-06,
|
| 2850 |
+
"logits/chosen": NaN,
|
| 2851 |
+
"logits/rejected": NaN,
|
| 2852 |
+
"logps/chosen": NaN,
|
| 2853 |
+
"logps/rejected": NaN,
|
| 2854 |
+
"loss": 0.0,
|
| 2855 |
+
"rewards/accuracies": 0.0,
|
| 2856 |
+
"rewards/chosen": NaN,
|
| 2857 |
+
"rewards/margins": NaN,
|
| 2858 |
+
"rewards/rejected": NaN,
|
| 2859 |
+
"step": 190
|
| 2860 |
+
},
|
| 2861 |
+
{
|
| 2862 |
+
"epoch": 0.05351264271205435,
|
| 2863 |
+
"grad_norm": NaN,
|
| 2864 |
+
"learning_rate": 9.55e-06,
|
| 2865 |
+
"logits/chosen": NaN,
|
| 2866 |
+
"logits/rejected": NaN,
|
| 2867 |
+
"logps/chosen": NaN,
|
| 2868 |
+
"logps/rejected": NaN,
|
| 2869 |
+
"loss": 0.0,
|
| 2870 |
+
"rewards/accuracies": 0.0,
|
| 2871 |
+
"rewards/chosen": NaN,
|
| 2872 |
+
"rewards/margins": NaN,
|
| 2873 |
+
"rewards/rejected": NaN,
|
| 2874 |
+
"step": 191
|
| 2875 |
+
},
|
| 2876 |
+
{
|
| 2877 |
+
"epoch": 0.053792813616305946,
|
| 2878 |
+
"grad_norm": NaN,
|
| 2879 |
+
"learning_rate": 9.600000000000001e-06,
|
| 2880 |
+
"logits/chosen": NaN,
|
| 2881 |
+
"logits/rejected": NaN,
|
| 2882 |
+
"logps/chosen": NaN,
|
| 2883 |
+
"logps/rejected": NaN,
|
| 2884 |
+
"loss": 0.0,
|
| 2885 |
+
"rewards/accuracies": 0.0,
|
| 2886 |
+
"rewards/chosen": NaN,
|
| 2887 |
+
"rewards/margins": NaN,
|
| 2888 |
+
"rewards/rejected": NaN,
|
| 2889 |
+
"step": 192
|
| 2890 |
+
},
|
| 2891 |
+
{
|
| 2892 |
+
"epoch": 0.05407298452055754,
|
| 2893 |
+
"grad_norm": NaN,
|
| 2894 |
+
"learning_rate": 9.65e-06,
|
| 2895 |
+
"logits/chosen": NaN,
|
| 2896 |
+
"logits/rejected": NaN,
|
| 2897 |
+
"logps/chosen": NaN,
|
| 2898 |
+
"logps/rejected": NaN,
|
| 2899 |
+
"loss": 0.0,
|
| 2900 |
+
"rewards/accuracies": 0.0,
|
| 2901 |
+
"rewards/chosen": NaN,
|
| 2902 |
+
"rewards/margins": NaN,
|
| 2903 |
+
"rewards/rejected": NaN,
|
| 2904 |
+
"step": 193
|
| 2905 |
+
},
|
| 2906 |
+
{
|
| 2907 |
+
"epoch": 0.05435315542480913,
|
| 2908 |
+
"grad_norm": NaN,
|
| 2909 |
+
"learning_rate": 9.7e-06,
|
| 2910 |
+
"logits/chosen": NaN,
|
| 2911 |
+
"logits/rejected": NaN,
|
| 2912 |
+
"logps/chosen": NaN,
|
| 2913 |
+
"logps/rejected": NaN,
|
| 2914 |
+
"loss": 0.0,
|
| 2915 |
+
"rewards/accuracies": 0.0,
|
| 2916 |
+
"rewards/chosen": NaN,
|
| 2917 |
+
"rewards/margins": NaN,
|
| 2918 |
+
"rewards/rejected": NaN,
|
| 2919 |
+
"step": 194
|
| 2920 |
+
},
|
| 2921 |
+
{
|
| 2922 |
+
"epoch": 0.054633326329060726,
|
| 2923 |
+
"grad_norm": NaN,
|
| 2924 |
+
"learning_rate": 9.75e-06,
|
| 2925 |
+
"logits/chosen": NaN,
|
| 2926 |
+
"logits/rejected": NaN,
|
| 2927 |
+
"logps/chosen": NaN,
|
| 2928 |
+
"logps/rejected": NaN,
|
| 2929 |
+
"loss": 0.0,
|
| 2930 |
+
"rewards/accuracies": 0.0,
|
| 2931 |
+
"rewards/chosen": NaN,
|
| 2932 |
+
"rewards/margins": NaN,
|
| 2933 |
+
"rewards/rejected": NaN,
|
| 2934 |
+
"step": 195
|
| 2935 |
+
},
|
| 2936 |
+
{
|
| 2937 |
+
"epoch": 0.05491349723331232,
|
| 2938 |
+
"grad_norm": NaN,
|
| 2939 |
+
"learning_rate": 9.800000000000001e-06,
|
| 2940 |
+
"logits/chosen": NaN,
|
| 2941 |
+
"logits/rejected": NaN,
|
| 2942 |
+
"logps/chosen": NaN,
|
| 2943 |
+
"logps/rejected": NaN,
|
| 2944 |
+
"loss": 0.0,
|
| 2945 |
+
"rewards/accuracies": 0.0,
|
| 2946 |
+
"rewards/chosen": NaN,
|
| 2947 |
+
"rewards/margins": NaN,
|
| 2948 |
+
"rewards/rejected": NaN,
|
| 2949 |
+
"step": 196
|
| 2950 |
+
},
|
| 2951 |
+
{
|
| 2952 |
+
"epoch": 0.05519366813756391,
|
| 2953 |
+
"grad_norm": NaN,
|
| 2954 |
+
"learning_rate": 9.85e-06,
|
| 2955 |
+
"logits/chosen": NaN,
|
| 2956 |
+
"logits/rejected": NaN,
|
| 2957 |
+
"logps/chosen": NaN,
|
| 2958 |
+
"logps/rejected": NaN,
|
| 2959 |
+
"loss": 0.0,
|
| 2960 |
+
"rewards/accuracies": 0.0,
|
| 2961 |
+
"rewards/chosen": NaN,
|
| 2962 |
+
"rewards/margins": NaN,
|
| 2963 |
+
"rewards/rejected": NaN,
|
| 2964 |
+
"step": 197
|
| 2965 |
+
},
|
| 2966 |
+
{
|
| 2967 |
+
"epoch": 0.055473839041815506,
|
| 2968 |
+
"grad_norm": NaN,
|
| 2969 |
+
"learning_rate": 9.9e-06,
|
| 2970 |
+
"logits/chosen": NaN,
|
| 2971 |
+
"logits/rejected": NaN,
|
| 2972 |
+
"logps/chosen": NaN,
|
| 2973 |
+
"logps/rejected": NaN,
|
| 2974 |
+
"loss": 0.0,
|
| 2975 |
+
"rewards/accuracies": 0.0,
|
| 2976 |
+
"rewards/chosen": NaN,
|
| 2977 |
+
"rewards/margins": NaN,
|
| 2978 |
+
"rewards/rejected": NaN,
|
| 2979 |
+
"step": 198
|
| 2980 |
+
},
|
| 2981 |
+
{
|
| 2982 |
+
"epoch": 0.0557540099460671,
|
| 2983 |
+
"grad_norm": NaN,
|
| 2984 |
+
"learning_rate": 9.950000000000001e-06,
|
| 2985 |
+
"logits/chosen": NaN,
|
| 2986 |
+
"logits/rejected": NaN,
|
| 2987 |
+
"logps/chosen": NaN,
|
| 2988 |
+
"logps/rejected": NaN,
|
| 2989 |
+
"loss": 0.0,
|
| 2990 |
+
"rewards/accuracies": 0.0,
|
| 2991 |
+
"rewards/chosen": NaN,
|
| 2992 |
+
"rewards/margins": NaN,
|
| 2993 |
+
"rewards/rejected": NaN,
|
| 2994 |
+
"step": 199
|
| 2995 |
+
},
|
| 2996 |
+
{
|
| 2997 |
+
"epoch": 0.05603418085031869,
|
| 2998 |
+
"grad_norm": NaN,
|
| 2999 |
+
"learning_rate": 1e-05,
|
| 3000 |
+
"logits/chosen": NaN,
|
| 3001 |
+
"logits/rejected": NaN,
|
| 3002 |
+
"logps/chosen": NaN,
|
| 3003 |
+
"logps/rejected": NaN,
|
| 3004 |
+
"loss": 0.0,
|
| 3005 |
+
"rewards/accuracies": 0.0,
|
| 3006 |
+
"rewards/chosen": NaN,
|
| 3007 |
+
"rewards/margins": NaN,
|
| 3008 |
+
"rewards/rejected": NaN,
|
| 3009 |
+
"step": 200
|
| 3010 |
}
|
| 3011 |
],
|
| 3012 |
"logging_steps": 1,
|
|
|
|
| 3021 |
"should_evaluate": false,
|
| 3022 |
"should_log": false,
|
| 3023 |
"should_save": true,
|
| 3024 |
+
"should_training_stop": true
|
| 3025 |
},
|
| 3026 |
"attributes": {}
|
| 3027 |
}
|