Uploaded checkpoint-27500
Browse files- model.safetensors +1 -1
- optimizer.pt +1 -1
- rng_state.pth +1 -1
- scheduler.pt +1 -1
- trainer_state.json +1793 -3
model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 2692969128
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:40a3f22374b10f1dacc2052bbe1eeb9d3ec51c4d0215210d58e218ca693293da
|
3 |
size 2692969128
|
optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 5386075202
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:dae1cc318abadbe97655b2e89cba3a93d2fadb650d801744bdfc7c6a4ccca5c0
|
3 |
size 5386075202
|
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 sha256:38fbaa4aaa427747240a6b65afd267f1edcb968fa67c3bf21b881737ad1b8da3
|
3 |
size 14244
|
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 sha256:30c1646ffd4f2e4e86a7c5c87af0949f3be46b7539d2a0137b1bb01bf3e8bbe5
|
3 |
size 1064
|
trainer_state.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"best_metric": 0.6027244925498962,
|
3 |
"best_model_checkpoint": "runs/deepseek_CMU-AIR2/math-deepseek_FULL_HardArith_Interm_20240424-065814/checkpoint-5000",
|
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,
|
@@ -3587,6 +3587,1796 @@
|
|
3587 |
"eval_samples_per_second": 26.216,
|
3588 |
"eval_steps_per_second": 26.216,
|
3589 |
"step": 5000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
3590 |
}
|
3591 |
],
|
3592 |
"logging_steps": 10,
|
@@ -3594,7 +5384,7 @@
|
|
3594 |
"num_input_tokens_seen": 0,
|
3595 |
"num_train_epochs": 1,
|
3596 |
"save_steps": 2500,
|
3597 |
-
"total_flos":
|
3598 |
"train_batch_size": 1,
|
3599 |
"trial_name": null,
|
3600 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": 0.6027244925498962,
|
3 |
"best_model_checkpoint": "runs/deepseek_CMU-AIR2/math-deepseek_FULL_HardArith_Interm_20240424-065814/checkpoint-5000",
|
4 |
+
"epoch": 0.1875,
|
5 |
"eval_steps": 500,
|
6 |
+
"global_step": 7500,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
3587 |
"eval_samples_per_second": 26.216,
|
3588 |
"eval_steps_per_second": 26.216,
|
3589 |
"step": 5000
|
3590 |
+
},
|
3591 |
+
{
|
3592 |
+
"epoch": 0.13,
|
3593 |
+
"grad_norm": 4.53125,
|
3594 |
+
"learning_rate": 1.0505263157894739e-05,
|
3595 |
+
"loss": 0.7573,
|
3596 |
+
"step": 5010
|
3597 |
+
},
|
3598 |
+
{
|
3599 |
+
"epoch": 0.13,
|
3600 |
+
"grad_norm": 1.8515625,
|
3601 |
+
"learning_rate": 1.048421052631579e-05,
|
3602 |
+
"loss": 0.5092,
|
3603 |
+
"step": 5020
|
3604 |
+
},
|
3605 |
+
{
|
3606 |
+
"epoch": 0.13,
|
3607 |
+
"grad_norm": 14.9375,
|
3608 |
+
"learning_rate": 1.0463157894736844e-05,
|
3609 |
+
"loss": 0.642,
|
3610 |
+
"step": 5030
|
3611 |
+
},
|
3612 |
+
{
|
3613 |
+
"epoch": 0.13,
|
3614 |
+
"grad_norm": 4.90625,
|
3615 |
+
"learning_rate": 1.0442105263157895e-05,
|
3616 |
+
"loss": 0.6185,
|
3617 |
+
"step": 5040
|
3618 |
+
},
|
3619 |
+
{
|
3620 |
+
"epoch": 0.13,
|
3621 |
+
"grad_norm": 4.5625,
|
3622 |
+
"learning_rate": 1.0421052631578948e-05,
|
3623 |
+
"loss": 0.6183,
|
3624 |
+
"step": 5050
|
3625 |
+
},
|
3626 |
+
{
|
3627 |
+
"epoch": 0.13,
|
3628 |
+
"grad_norm": 6.125,
|
3629 |
+
"learning_rate": 1.04e-05,
|
3630 |
+
"loss": 0.5496,
|
3631 |
+
"step": 5060
|
3632 |
+
},
|
3633 |
+
{
|
3634 |
+
"epoch": 0.13,
|
3635 |
+
"grad_norm": 4.5625,
|
3636 |
+
"learning_rate": 1.0378947368421053e-05,
|
3637 |
+
"loss": 0.5861,
|
3638 |
+
"step": 5070
|
3639 |
+
},
|
3640 |
+
{
|
3641 |
+
"epoch": 0.13,
|
3642 |
+
"grad_norm": 3.515625,
|
3643 |
+
"learning_rate": 1.0357894736842107e-05,
|
3644 |
+
"loss": 0.6365,
|
3645 |
+
"step": 5080
|
3646 |
+
},
|
3647 |
+
{
|
3648 |
+
"epoch": 0.13,
|
3649 |
+
"grad_norm": 4.75,
|
3650 |
+
"learning_rate": 1.0336842105263158e-05,
|
3651 |
+
"loss": 0.6688,
|
3652 |
+
"step": 5090
|
3653 |
+
},
|
3654 |
+
{
|
3655 |
+
"epoch": 0.13,
|
3656 |
+
"grad_norm": 2.703125,
|
3657 |
+
"learning_rate": 1.0315789473684213e-05,
|
3658 |
+
"loss": 0.5361,
|
3659 |
+
"step": 5100
|
3660 |
+
},
|
3661 |
+
{
|
3662 |
+
"epoch": 0.13,
|
3663 |
+
"grad_norm": 4.125,
|
3664 |
+
"learning_rate": 1.0294736842105264e-05,
|
3665 |
+
"loss": 0.6006,
|
3666 |
+
"step": 5110
|
3667 |
+
},
|
3668 |
+
{
|
3669 |
+
"epoch": 0.13,
|
3670 |
+
"grad_norm": 3.890625,
|
3671 |
+
"learning_rate": 1.0273684210526316e-05,
|
3672 |
+
"loss": 0.6353,
|
3673 |
+
"step": 5120
|
3674 |
+
},
|
3675 |
+
{
|
3676 |
+
"epoch": 0.13,
|
3677 |
+
"grad_norm": 6.25,
|
3678 |
+
"learning_rate": 1.0252631578947369e-05,
|
3679 |
+
"loss": 0.6475,
|
3680 |
+
"step": 5130
|
3681 |
+
},
|
3682 |
+
{
|
3683 |
+
"epoch": 0.13,
|
3684 |
+
"grad_norm": 4.625,
|
3685 |
+
"learning_rate": 1.0231578947368422e-05,
|
3686 |
+
"loss": 0.498,
|
3687 |
+
"step": 5140
|
3688 |
+
},
|
3689 |
+
{
|
3690 |
+
"epoch": 0.13,
|
3691 |
+
"grad_norm": 4.75,
|
3692 |
+
"learning_rate": 1.0210526315789476e-05,
|
3693 |
+
"loss": 0.6467,
|
3694 |
+
"step": 5150
|
3695 |
+
},
|
3696 |
+
{
|
3697 |
+
"epoch": 0.13,
|
3698 |
+
"grad_norm": 4.1875,
|
3699 |
+
"learning_rate": 1.0189473684210527e-05,
|
3700 |
+
"loss": 0.611,
|
3701 |
+
"step": 5160
|
3702 |
+
},
|
3703 |
+
{
|
3704 |
+
"epoch": 0.13,
|
3705 |
+
"grad_norm": 9.75,
|
3706 |
+
"learning_rate": 1.0168421052631581e-05,
|
3707 |
+
"loss": 0.5392,
|
3708 |
+
"step": 5170
|
3709 |
+
},
|
3710 |
+
{
|
3711 |
+
"epoch": 0.13,
|
3712 |
+
"grad_norm": 13.25,
|
3713 |
+
"learning_rate": 1.0147368421052632e-05,
|
3714 |
+
"loss": 0.645,
|
3715 |
+
"step": 5180
|
3716 |
+
},
|
3717 |
+
{
|
3718 |
+
"epoch": 0.13,
|
3719 |
+
"grad_norm": 4.125,
|
3720 |
+
"learning_rate": 1.0126315789473685e-05,
|
3721 |
+
"loss": 0.6287,
|
3722 |
+
"step": 5190
|
3723 |
+
},
|
3724 |
+
{
|
3725 |
+
"epoch": 0.13,
|
3726 |
+
"grad_norm": 4.21875,
|
3727 |
+
"learning_rate": 1.0105263157894738e-05,
|
3728 |
+
"loss": 0.5719,
|
3729 |
+
"step": 5200
|
3730 |
+
},
|
3731 |
+
{
|
3732 |
+
"epoch": 0.13,
|
3733 |
+
"grad_norm": 2.671875,
|
3734 |
+
"learning_rate": 1.008421052631579e-05,
|
3735 |
+
"loss": 0.621,
|
3736 |
+
"step": 5210
|
3737 |
+
},
|
3738 |
+
{
|
3739 |
+
"epoch": 0.13,
|
3740 |
+
"grad_norm": 2.984375,
|
3741 |
+
"learning_rate": 1.0063157894736843e-05,
|
3742 |
+
"loss": 0.5169,
|
3743 |
+
"step": 5220
|
3744 |
+
},
|
3745 |
+
{
|
3746 |
+
"epoch": 0.13,
|
3747 |
+
"grad_norm": 8.375,
|
3748 |
+
"learning_rate": 1.0042105263157896e-05,
|
3749 |
+
"loss": 0.6793,
|
3750 |
+
"step": 5230
|
3751 |
+
},
|
3752 |
+
{
|
3753 |
+
"epoch": 0.13,
|
3754 |
+
"grad_norm": 117.0,
|
3755 |
+
"learning_rate": 1.002105263157895e-05,
|
3756 |
+
"loss": 0.6005,
|
3757 |
+
"step": 5240
|
3758 |
+
},
|
3759 |
+
{
|
3760 |
+
"epoch": 0.13,
|
3761 |
+
"grad_norm": 3.5625,
|
3762 |
+
"learning_rate": 1e-05,
|
3763 |
+
"loss": 0.6674,
|
3764 |
+
"step": 5250
|
3765 |
+
},
|
3766 |
+
{
|
3767 |
+
"epoch": 0.13,
|
3768 |
+
"grad_norm": 3.9375,
|
3769 |
+
"learning_rate": 9.978947368421053e-06,
|
3770 |
+
"loss": 0.6105,
|
3771 |
+
"step": 5260
|
3772 |
+
},
|
3773 |
+
{
|
3774 |
+
"epoch": 0.13,
|
3775 |
+
"grad_norm": 4.75,
|
3776 |
+
"learning_rate": 9.957894736842106e-06,
|
3777 |
+
"loss": 0.5876,
|
3778 |
+
"step": 5270
|
3779 |
+
},
|
3780 |
+
{
|
3781 |
+
"epoch": 0.13,
|
3782 |
+
"grad_norm": 3.203125,
|
3783 |
+
"learning_rate": 9.936842105263159e-06,
|
3784 |
+
"loss": 0.5551,
|
3785 |
+
"step": 5280
|
3786 |
+
},
|
3787 |
+
{
|
3788 |
+
"epoch": 0.13,
|
3789 |
+
"grad_norm": 4.84375,
|
3790 |
+
"learning_rate": 9.915789473684211e-06,
|
3791 |
+
"loss": 0.5699,
|
3792 |
+
"step": 5290
|
3793 |
+
},
|
3794 |
+
{
|
3795 |
+
"epoch": 0.13,
|
3796 |
+
"grad_norm": 3.5625,
|
3797 |
+
"learning_rate": 9.894736842105264e-06,
|
3798 |
+
"loss": 0.5652,
|
3799 |
+
"step": 5300
|
3800 |
+
},
|
3801 |
+
{
|
3802 |
+
"epoch": 0.13,
|
3803 |
+
"grad_norm": 7.59375,
|
3804 |
+
"learning_rate": 9.873684210526317e-06,
|
3805 |
+
"loss": 0.6494,
|
3806 |
+
"step": 5310
|
3807 |
+
},
|
3808 |
+
{
|
3809 |
+
"epoch": 0.13,
|
3810 |
+
"grad_norm": 6.03125,
|
3811 |
+
"learning_rate": 9.85263157894737e-06,
|
3812 |
+
"loss": 0.6,
|
3813 |
+
"step": 5320
|
3814 |
+
},
|
3815 |
+
{
|
3816 |
+
"epoch": 0.13,
|
3817 |
+
"grad_norm": 3.3125,
|
3818 |
+
"learning_rate": 9.831578947368422e-06,
|
3819 |
+
"loss": 0.603,
|
3820 |
+
"step": 5330
|
3821 |
+
},
|
3822 |
+
{
|
3823 |
+
"epoch": 0.13,
|
3824 |
+
"grad_norm": 3.546875,
|
3825 |
+
"learning_rate": 9.810526315789475e-06,
|
3826 |
+
"loss": 0.6117,
|
3827 |
+
"step": 5340
|
3828 |
+
},
|
3829 |
+
{
|
3830 |
+
"epoch": 0.13,
|
3831 |
+
"grad_norm": 4.625,
|
3832 |
+
"learning_rate": 9.789473684210527e-06,
|
3833 |
+
"loss": 0.5883,
|
3834 |
+
"step": 5350
|
3835 |
+
},
|
3836 |
+
{
|
3837 |
+
"epoch": 0.13,
|
3838 |
+
"grad_norm": 3.234375,
|
3839 |
+
"learning_rate": 9.76842105263158e-06,
|
3840 |
+
"loss": 0.5722,
|
3841 |
+
"step": 5360
|
3842 |
+
},
|
3843 |
+
{
|
3844 |
+
"epoch": 0.13,
|
3845 |
+
"grad_norm": 3.171875,
|
3846 |
+
"learning_rate": 9.747368421052633e-06,
|
3847 |
+
"loss": 0.5692,
|
3848 |
+
"step": 5370
|
3849 |
+
},
|
3850 |
+
{
|
3851 |
+
"epoch": 0.13,
|
3852 |
+
"grad_norm": 3.515625,
|
3853 |
+
"learning_rate": 9.726315789473685e-06,
|
3854 |
+
"loss": 0.7724,
|
3855 |
+
"step": 5380
|
3856 |
+
},
|
3857 |
+
{
|
3858 |
+
"epoch": 0.13,
|
3859 |
+
"grad_norm": 4.25,
|
3860 |
+
"learning_rate": 9.705263157894738e-06,
|
3861 |
+
"loss": 0.6017,
|
3862 |
+
"step": 5390
|
3863 |
+
},
|
3864 |
+
{
|
3865 |
+
"epoch": 0.14,
|
3866 |
+
"grad_norm": 6.53125,
|
3867 |
+
"learning_rate": 9.68421052631579e-06,
|
3868 |
+
"loss": 0.6049,
|
3869 |
+
"step": 5400
|
3870 |
+
},
|
3871 |
+
{
|
3872 |
+
"epoch": 0.14,
|
3873 |
+
"grad_norm": 4.34375,
|
3874 |
+
"learning_rate": 9.663157894736843e-06,
|
3875 |
+
"loss": 0.5158,
|
3876 |
+
"step": 5410
|
3877 |
+
},
|
3878 |
+
{
|
3879 |
+
"epoch": 0.14,
|
3880 |
+
"grad_norm": 5.34375,
|
3881 |
+
"learning_rate": 9.642105263157896e-06,
|
3882 |
+
"loss": 0.6487,
|
3883 |
+
"step": 5420
|
3884 |
+
},
|
3885 |
+
{
|
3886 |
+
"epoch": 0.14,
|
3887 |
+
"grad_norm": 5.71875,
|
3888 |
+
"learning_rate": 9.621052631578947e-06,
|
3889 |
+
"loss": 0.5599,
|
3890 |
+
"step": 5430
|
3891 |
+
},
|
3892 |
+
{
|
3893 |
+
"epoch": 0.14,
|
3894 |
+
"grad_norm": 5.34375,
|
3895 |
+
"learning_rate": 9.600000000000001e-06,
|
3896 |
+
"loss": 0.5035,
|
3897 |
+
"step": 5440
|
3898 |
+
},
|
3899 |
+
{
|
3900 |
+
"epoch": 0.14,
|
3901 |
+
"grad_norm": 4.4375,
|
3902 |
+
"learning_rate": 9.578947368421054e-06,
|
3903 |
+
"loss": 0.5768,
|
3904 |
+
"step": 5450
|
3905 |
+
},
|
3906 |
+
{
|
3907 |
+
"epoch": 0.14,
|
3908 |
+
"grad_norm": 48.25,
|
3909 |
+
"learning_rate": 9.557894736842107e-06,
|
3910 |
+
"loss": 0.5815,
|
3911 |
+
"step": 5460
|
3912 |
+
},
|
3913 |
+
{
|
3914 |
+
"epoch": 0.14,
|
3915 |
+
"grad_norm": 3.40625,
|
3916 |
+
"learning_rate": 9.53684210526316e-06,
|
3917 |
+
"loss": 0.6133,
|
3918 |
+
"step": 5470
|
3919 |
+
},
|
3920 |
+
{
|
3921 |
+
"epoch": 0.14,
|
3922 |
+
"grad_norm": 2.40625,
|
3923 |
+
"learning_rate": 9.515789473684212e-06,
|
3924 |
+
"loss": 0.5997,
|
3925 |
+
"step": 5480
|
3926 |
+
},
|
3927 |
+
{
|
3928 |
+
"epoch": 0.14,
|
3929 |
+
"grad_norm": 3.578125,
|
3930 |
+
"learning_rate": 9.494736842105265e-06,
|
3931 |
+
"loss": 0.5689,
|
3932 |
+
"step": 5490
|
3933 |
+
},
|
3934 |
+
{
|
3935 |
+
"epoch": 0.14,
|
3936 |
+
"grad_norm": 7.0625,
|
3937 |
+
"learning_rate": 9.473684210526315e-06,
|
3938 |
+
"loss": 0.6511,
|
3939 |
+
"step": 5500
|
3940 |
+
},
|
3941 |
+
{
|
3942 |
+
"epoch": 0.14,
|
3943 |
+
"eval_loss": 0.6153059005737305,
|
3944 |
+
"eval_runtime": 38.2125,
|
3945 |
+
"eval_samples_per_second": 26.169,
|
3946 |
+
"eval_steps_per_second": 26.169,
|
3947 |
+
"step": 5500
|
3948 |
+
},
|
3949 |
+
{
|
3950 |
+
"epoch": 0.14,
|
3951 |
+
"grad_norm": 3.1875,
|
3952 |
+
"learning_rate": 9.452631578947368e-06,
|
3953 |
+
"loss": 0.5449,
|
3954 |
+
"step": 5510
|
3955 |
+
},
|
3956 |
+
{
|
3957 |
+
"epoch": 0.14,
|
3958 |
+
"grad_norm": 4.40625,
|
3959 |
+
"learning_rate": 9.43157894736842e-06,
|
3960 |
+
"loss": 0.6634,
|
3961 |
+
"step": 5520
|
3962 |
+
},
|
3963 |
+
{
|
3964 |
+
"epoch": 0.14,
|
3965 |
+
"grad_norm": 7.875,
|
3966 |
+
"learning_rate": 9.410526315789475e-06,
|
3967 |
+
"loss": 0.6222,
|
3968 |
+
"step": 5530
|
3969 |
+
},
|
3970 |
+
{
|
3971 |
+
"epoch": 0.14,
|
3972 |
+
"grad_norm": 5.4375,
|
3973 |
+
"learning_rate": 9.389473684210528e-06,
|
3974 |
+
"loss": 0.5707,
|
3975 |
+
"step": 5540
|
3976 |
+
},
|
3977 |
+
{
|
3978 |
+
"epoch": 0.14,
|
3979 |
+
"grad_norm": 3.828125,
|
3980 |
+
"learning_rate": 9.36842105263158e-06,
|
3981 |
+
"loss": 0.4959,
|
3982 |
+
"step": 5550
|
3983 |
+
},
|
3984 |
+
{
|
3985 |
+
"epoch": 0.14,
|
3986 |
+
"grad_norm": 6.0625,
|
3987 |
+
"learning_rate": 9.347368421052633e-06,
|
3988 |
+
"loss": 0.5941,
|
3989 |
+
"step": 5560
|
3990 |
+
},
|
3991 |
+
{
|
3992 |
+
"epoch": 0.14,
|
3993 |
+
"grad_norm": 1.6015625,
|
3994 |
+
"learning_rate": 9.326315789473684e-06,
|
3995 |
+
"loss": 0.6101,
|
3996 |
+
"step": 5570
|
3997 |
+
},
|
3998 |
+
{
|
3999 |
+
"epoch": 0.14,
|
4000 |
+
"grad_norm": 4.4375,
|
4001 |
+
"learning_rate": 9.305263157894737e-06,
|
4002 |
+
"loss": 0.5916,
|
4003 |
+
"step": 5580
|
4004 |
+
},
|
4005 |
+
{
|
4006 |
+
"epoch": 0.14,
|
4007 |
+
"grad_norm": 4.34375,
|
4008 |
+
"learning_rate": 9.28421052631579e-06,
|
4009 |
+
"loss": 0.5927,
|
4010 |
+
"step": 5590
|
4011 |
+
},
|
4012 |
+
{
|
4013 |
+
"epoch": 0.14,
|
4014 |
+
"grad_norm": 5.90625,
|
4015 |
+
"learning_rate": 9.263157894736842e-06,
|
4016 |
+
"loss": 0.5575,
|
4017 |
+
"step": 5600
|
4018 |
+
},
|
4019 |
+
{
|
4020 |
+
"epoch": 0.14,
|
4021 |
+
"grad_norm": 8.5,
|
4022 |
+
"learning_rate": 9.242105263157896e-06,
|
4023 |
+
"loss": 0.5709,
|
4024 |
+
"step": 5610
|
4025 |
+
},
|
4026 |
+
{
|
4027 |
+
"epoch": 0.14,
|
4028 |
+
"grad_norm": 3.5,
|
4029 |
+
"learning_rate": 9.221052631578949e-06,
|
4030 |
+
"loss": 0.6237,
|
4031 |
+
"step": 5620
|
4032 |
+
},
|
4033 |
+
{
|
4034 |
+
"epoch": 0.14,
|
4035 |
+
"grad_norm": 5.25,
|
4036 |
+
"learning_rate": 9.200000000000002e-06,
|
4037 |
+
"loss": 0.6183,
|
4038 |
+
"step": 5630
|
4039 |
+
},
|
4040 |
+
{
|
4041 |
+
"epoch": 0.14,
|
4042 |
+
"grad_norm": 6.3125,
|
4043 |
+
"learning_rate": 9.178947368421053e-06,
|
4044 |
+
"loss": 0.579,
|
4045 |
+
"step": 5640
|
4046 |
+
},
|
4047 |
+
{
|
4048 |
+
"epoch": 0.14,
|
4049 |
+
"grad_norm": 4.8125,
|
4050 |
+
"learning_rate": 9.157894736842105e-06,
|
4051 |
+
"loss": 0.5801,
|
4052 |
+
"step": 5650
|
4053 |
+
},
|
4054 |
+
{
|
4055 |
+
"epoch": 0.14,
|
4056 |
+
"grad_norm": 3.71875,
|
4057 |
+
"learning_rate": 9.136842105263158e-06,
|
4058 |
+
"loss": 0.5568,
|
4059 |
+
"step": 5660
|
4060 |
+
},
|
4061 |
+
{
|
4062 |
+
"epoch": 0.14,
|
4063 |
+
"grad_norm": 2.84375,
|
4064 |
+
"learning_rate": 9.11578947368421e-06,
|
4065 |
+
"loss": 0.6277,
|
4066 |
+
"step": 5670
|
4067 |
+
},
|
4068 |
+
{
|
4069 |
+
"epoch": 0.14,
|
4070 |
+
"grad_norm": 3.125,
|
4071 |
+
"learning_rate": 9.094736842105263e-06,
|
4072 |
+
"loss": 0.5189,
|
4073 |
+
"step": 5680
|
4074 |
+
},
|
4075 |
+
{
|
4076 |
+
"epoch": 0.14,
|
4077 |
+
"grad_norm": 2.3125,
|
4078 |
+
"learning_rate": 9.073684210526316e-06,
|
4079 |
+
"loss": 0.6304,
|
4080 |
+
"step": 5690
|
4081 |
+
},
|
4082 |
+
{
|
4083 |
+
"epoch": 0.14,
|
4084 |
+
"grad_norm": 2.359375,
|
4085 |
+
"learning_rate": 9.05263157894737e-06,
|
4086 |
+
"loss": 0.5956,
|
4087 |
+
"step": 5700
|
4088 |
+
},
|
4089 |
+
{
|
4090 |
+
"epoch": 0.14,
|
4091 |
+
"grad_norm": 11.9375,
|
4092 |
+
"learning_rate": 9.031578947368423e-06,
|
4093 |
+
"loss": 0.6483,
|
4094 |
+
"step": 5710
|
4095 |
+
},
|
4096 |
+
{
|
4097 |
+
"epoch": 0.14,
|
4098 |
+
"grad_norm": 5.09375,
|
4099 |
+
"learning_rate": 9.010526315789474e-06,
|
4100 |
+
"loss": 0.601,
|
4101 |
+
"step": 5720
|
4102 |
+
},
|
4103 |
+
{
|
4104 |
+
"epoch": 0.14,
|
4105 |
+
"grad_norm": 3.5,
|
4106 |
+
"learning_rate": 8.989473684210527e-06,
|
4107 |
+
"loss": 0.5887,
|
4108 |
+
"step": 5730
|
4109 |
+
},
|
4110 |
+
{
|
4111 |
+
"epoch": 0.14,
|
4112 |
+
"grad_norm": 3.09375,
|
4113 |
+
"learning_rate": 8.96842105263158e-06,
|
4114 |
+
"loss": 0.63,
|
4115 |
+
"step": 5740
|
4116 |
+
},
|
4117 |
+
{
|
4118 |
+
"epoch": 0.14,
|
4119 |
+
"grad_norm": 4.4375,
|
4120 |
+
"learning_rate": 8.947368421052632e-06,
|
4121 |
+
"loss": 0.5988,
|
4122 |
+
"step": 5750
|
4123 |
+
},
|
4124 |
+
{
|
4125 |
+
"epoch": 0.14,
|
4126 |
+
"grad_norm": 3.265625,
|
4127 |
+
"learning_rate": 8.926315789473685e-06,
|
4128 |
+
"loss": 0.5692,
|
4129 |
+
"step": 5760
|
4130 |
+
},
|
4131 |
+
{
|
4132 |
+
"epoch": 0.14,
|
4133 |
+
"grad_norm": 3.015625,
|
4134 |
+
"learning_rate": 8.905263157894737e-06,
|
4135 |
+
"loss": 0.5917,
|
4136 |
+
"step": 5770
|
4137 |
+
},
|
4138 |
+
{
|
4139 |
+
"epoch": 0.14,
|
4140 |
+
"grad_norm": 4.40625,
|
4141 |
+
"learning_rate": 8.884210526315792e-06,
|
4142 |
+
"loss": 0.565,
|
4143 |
+
"step": 5780
|
4144 |
+
},
|
4145 |
+
{
|
4146 |
+
"epoch": 0.14,
|
4147 |
+
"grad_norm": 4.15625,
|
4148 |
+
"learning_rate": 8.863157894736842e-06,
|
4149 |
+
"loss": 0.5388,
|
4150 |
+
"step": 5790
|
4151 |
+
},
|
4152 |
+
{
|
4153 |
+
"epoch": 0.14,
|
4154 |
+
"grad_norm": 7.875,
|
4155 |
+
"learning_rate": 8.842105263157895e-06,
|
4156 |
+
"loss": 0.6622,
|
4157 |
+
"step": 5800
|
4158 |
+
},
|
4159 |
+
{
|
4160 |
+
"epoch": 0.15,
|
4161 |
+
"grad_norm": 6.25,
|
4162 |
+
"learning_rate": 8.821052631578948e-06,
|
4163 |
+
"loss": 0.6043,
|
4164 |
+
"step": 5810
|
4165 |
+
},
|
4166 |
+
{
|
4167 |
+
"epoch": 0.15,
|
4168 |
+
"grad_norm": 3.953125,
|
4169 |
+
"learning_rate": 8.8e-06,
|
4170 |
+
"loss": 0.6431,
|
4171 |
+
"step": 5820
|
4172 |
+
},
|
4173 |
+
{
|
4174 |
+
"epoch": 0.15,
|
4175 |
+
"grad_norm": 2.421875,
|
4176 |
+
"learning_rate": 8.778947368421053e-06,
|
4177 |
+
"loss": 0.563,
|
4178 |
+
"step": 5830
|
4179 |
+
},
|
4180 |
+
{
|
4181 |
+
"epoch": 0.15,
|
4182 |
+
"grad_norm": 7.4375,
|
4183 |
+
"learning_rate": 8.757894736842106e-06,
|
4184 |
+
"loss": 0.6342,
|
4185 |
+
"step": 5840
|
4186 |
+
},
|
4187 |
+
{
|
4188 |
+
"epoch": 0.15,
|
4189 |
+
"grad_norm": 4.46875,
|
4190 |
+
"learning_rate": 8.736842105263158e-06,
|
4191 |
+
"loss": 0.6206,
|
4192 |
+
"step": 5850
|
4193 |
+
},
|
4194 |
+
{
|
4195 |
+
"epoch": 0.15,
|
4196 |
+
"grad_norm": 2.828125,
|
4197 |
+
"learning_rate": 8.715789473684211e-06,
|
4198 |
+
"loss": 0.5046,
|
4199 |
+
"step": 5860
|
4200 |
+
},
|
4201 |
+
{
|
4202 |
+
"epoch": 0.15,
|
4203 |
+
"grad_norm": 13.75,
|
4204 |
+
"learning_rate": 8.694736842105264e-06,
|
4205 |
+
"loss": 0.5405,
|
4206 |
+
"step": 5870
|
4207 |
+
},
|
4208 |
+
{
|
4209 |
+
"epoch": 0.15,
|
4210 |
+
"grad_norm": 4.6875,
|
4211 |
+
"learning_rate": 8.673684210526316e-06,
|
4212 |
+
"loss": 0.6021,
|
4213 |
+
"step": 5880
|
4214 |
+
},
|
4215 |
+
{
|
4216 |
+
"epoch": 0.15,
|
4217 |
+
"grad_norm": 4.0,
|
4218 |
+
"learning_rate": 8.652631578947369e-06,
|
4219 |
+
"loss": 0.7437,
|
4220 |
+
"step": 5890
|
4221 |
+
},
|
4222 |
+
{
|
4223 |
+
"epoch": 0.15,
|
4224 |
+
"grad_norm": 3.4375,
|
4225 |
+
"learning_rate": 8.631578947368422e-06,
|
4226 |
+
"loss": 0.6618,
|
4227 |
+
"step": 5900
|
4228 |
+
},
|
4229 |
+
{
|
4230 |
+
"epoch": 0.15,
|
4231 |
+
"grad_norm": 3.34375,
|
4232 |
+
"learning_rate": 8.610526315789474e-06,
|
4233 |
+
"loss": 0.6088,
|
4234 |
+
"step": 5910
|
4235 |
+
},
|
4236 |
+
{
|
4237 |
+
"epoch": 0.15,
|
4238 |
+
"grad_norm": 4.90625,
|
4239 |
+
"learning_rate": 8.589473684210527e-06,
|
4240 |
+
"loss": 0.5808,
|
4241 |
+
"step": 5920
|
4242 |
+
},
|
4243 |
+
{
|
4244 |
+
"epoch": 0.15,
|
4245 |
+
"grad_norm": 4.25,
|
4246 |
+
"learning_rate": 8.56842105263158e-06,
|
4247 |
+
"loss": 0.6744,
|
4248 |
+
"step": 5930
|
4249 |
+
},
|
4250 |
+
{
|
4251 |
+
"epoch": 0.15,
|
4252 |
+
"grad_norm": 4.4375,
|
4253 |
+
"learning_rate": 8.547368421052632e-06,
|
4254 |
+
"loss": 0.5432,
|
4255 |
+
"step": 5940
|
4256 |
+
},
|
4257 |
+
{
|
4258 |
+
"epoch": 0.15,
|
4259 |
+
"grad_norm": 6.65625,
|
4260 |
+
"learning_rate": 8.526315789473685e-06,
|
4261 |
+
"loss": 0.6601,
|
4262 |
+
"step": 5950
|
4263 |
+
},
|
4264 |
+
{
|
4265 |
+
"epoch": 0.15,
|
4266 |
+
"grad_norm": 4.3125,
|
4267 |
+
"learning_rate": 8.505263157894738e-06,
|
4268 |
+
"loss": 0.4648,
|
4269 |
+
"step": 5960
|
4270 |
+
},
|
4271 |
+
{
|
4272 |
+
"epoch": 0.15,
|
4273 |
+
"grad_norm": 2.953125,
|
4274 |
+
"learning_rate": 8.48421052631579e-06,
|
4275 |
+
"loss": 0.5452,
|
4276 |
+
"step": 5970
|
4277 |
+
},
|
4278 |
+
{
|
4279 |
+
"epoch": 0.15,
|
4280 |
+
"grad_norm": 5.09375,
|
4281 |
+
"learning_rate": 8.463157894736843e-06,
|
4282 |
+
"loss": 0.5761,
|
4283 |
+
"step": 5980
|
4284 |
+
},
|
4285 |
+
{
|
4286 |
+
"epoch": 0.15,
|
4287 |
+
"grad_norm": 2.828125,
|
4288 |
+
"learning_rate": 8.442105263157896e-06,
|
4289 |
+
"loss": 0.5394,
|
4290 |
+
"step": 5990
|
4291 |
+
},
|
4292 |
+
{
|
4293 |
+
"epoch": 0.15,
|
4294 |
+
"grad_norm": 5.65625,
|
4295 |
+
"learning_rate": 8.421052631578948e-06,
|
4296 |
+
"loss": 0.6339,
|
4297 |
+
"step": 6000
|
4298 |
+
},
|
4299 |
+
{
|
4300 |
+
"epoch": 0.15,
|
4301 |
+
"eval_loss": 0.6167545318603516,
|
4302 |
+
"eval_runtime": 38.2022,
|
4303 |
+
"eval_samples_per_second": 26.177,
|
4304 |
+
"eval_steps_per_second": 26.177,
|
4305 |
+
"step": 6000
|
4306 |
+
},
|
4307 |
+
{
|
4308 |
+
"epoch": 0.15,
|
4309 |
+
"grad_norm": 3.609375,
|
4310 |
+
"learning_rate": 8.400000000000001e-06,
|
4311 |
+
"loss": 0.5043,
|
4312 |
+
"step": 6010
|
4313 |
+
},
|
4314 |
+
{
|
4315 |
+
"epoch": 0.15,
|
4316 |
+
"grad_norm": 3.171875,
|
4317 |
+
"learning_rate": 8.378947368421054e-06,
|
4318 |
+
"loss": 0.6012,
|
4319 |
+
"step": 6020
|
4320 |
+
},
|
4321 |
+
{
|
4322 |
+
"epoch": 0.15,
|
4323 |
+
"grad_norm": 7.03125,
|
4324 |
+
"learning_rate": 8.357894736842106e-06,
|
4325 |
+
"loss": 0.5963,
|
4326 |
+
"step": 6030
|
4327 |
+
},
|
4328 |
+
{
|
4329 |
+
"epoch": 0.15,
|
4330 |
+
"grad_norm": 5.9375,
|
4331 |
+
"learning_rate": 8.336842105263159e-06,
|
4332 |
+
"loss": 0.6281,
|
4333 |
+
"step": 6040
|
4334 |
+
},
|
4335 |
+
{
|
4336 |
+
"epoch": 0.15,
|
4337 |
+
"grad_norm": 2.59375,
|
4338 |
+
"learning_rate": 8.315789473684212e-06,
|
4339 |
+
"loss": 0.4643,
|
4340 |
+
"step": 6050
|
4341 |
+
},
|
4342 |
+
{
|
4343 |
+
"epoch": 0.15,
|
4344 |
+
"grad_norm": 7.4375,
|
4345 |
+
"learning_rate": 8.294736842105264e-06,
|
4346 |
+
"loss": 0.6371,
|
4347 |
+
"step": 6060
|
4348 |
+
},
|
4349 |
+
{
|
4350 |
+
"epoch": 0.15,
|
4351 |
+
"grad_norm": 2.421875,
|
4352 |
+
"learning_rate": 8.273684210526317e-06,
|
4353 |
+
"loss": 0.5769,
|
4354 |
+
"step": 6070
|
4355 |
+
},
|
4356 |
+
{
|
4357 |
+
"epoch": 0.15,
|
4358 |
+
"grad_norm": 3.703125,
|
4359 |
+
"learning_rate": 8.25263157894737e-06,
|
4360 |
+
"loss": 0.7226,
|
4361 |
+
"step": 6080
|
4362 |
+
},
|
4363 |
+
{
|
4364 |
+
"epoch": 0.15,
|
4365 |
+
"grad_norm": 4.375,
|
4366 |
+
"learning_rate": 8.231578947368422e-06,
|
4367 |
+
"loss": 0.541,
|
4368 |
+
"step": 6090
|
4369 |
+
},
|
4370 |
+
{
|
4371 |
+
"epoch": 0.15,
|
4372 |
+
"grad_norm": 4.46875,
|
4373 |
+
"learning_rate": 8.210526315789475e-06,
|
4374 |
+
"loss": 0.5965,
|
4375 |
+
"step": 6100
|
4376 |
+
},
|
4377 |
+
{
|
4378 |
+
"epoch": 0.15,
|
4379 |
+
"grad_norm": 6.28125,
|
4380 |
+
"learning_rate": 8.189473684210527e-06,
|
4381 |
+
"loss": 0.5999,
|
4382 |
+
"step": 6110
|
4383 |
+
},
|
4384 |
+
{
|
4385 |
+
"epoch": 0.15,
|
4386 |
+
"grad_norm": 6.34375,
|
4387 |
+
"learning_rate": 8.16842105263158e-06,
|
4388 |
+
"loss": 0.6388,
|
4389 |
+
"step": 6120
|
4390 |
+
},
|
4391 |
+
{
|
4392 |
+
"epoch": 0.15,
|
4393 |
+
"grad_norm": 5.6875,
|
4394 |
+
"learning_rate": 8.147368421052633e-06,
|
4395 |
+
"loss": 0.5696,
|
4396 |
+
"step": 6130
|
4397 |
+
},
|
4398 |
+
{
|
4399 |
+
"epoch": 0.15,
|
4400 |
+
"grad_norm": 2.4375,
|
4401 |
+
"learning_rate": 8.126315789473684e-06,
|
4402 |
+
"loss": 0.4886,
|
4403 |
+
"step": 6140
|
4404 |
+
},
|
4405 |
+
{
|
4406 |
+
"epoch": 0.15,
|
4407 |
+
"grad_norm": 5.59375,
|
4408 |
+
"learning_rate": 8.105263157894736e-06,
|
4409 |
+
"loss": 0.5406,
|
4410 |
+
"step": 6150
|
4411 |
+
},
|
4412 |
+
{
|
4413 |
+
"epoch": 0.15,
|
4414 |
+
"grad_norm": 6.9375,
|
4415 |
+
"learning_rate": 8.08421052631579e-06,
|
4416 |
+
"loss": 0.6134,
|
4417 |
+
"step": 6160
|
4418 |
+
},
|
4419 |
+
{
|
4420 |
+
"epoch": 0.15,
|
4421 |
+
"grad_norm": 4.875,
|
4422 |
+
"learning_rate": 8.063157894736843e-06,
|
4423 |
+
"loss": 0.6944,
|
4424 |
+
"step": 6170
|
4425 |
+
},
|
4426 |
+
{
|
4427 |
+
"epoch": 0.15,
|
4428 |
+
"grad_norm": 7.4375,
|
4429 |
+
"learning_rate": 8.042105263157896e-06,
|
4430 |
+
"loss": 0.5848,
|
4431 |
+
"step": 6180
|
4432 |
+
},
|
4433 |
+
{
|
4434 |
+
"epoch": 0.15,
|
4435 |
+
"grad_norm": 3.40625,
|
4436 |
+
"learning_rate": 8.021052631578949e-06,
|
4437 |
+
"loss": 0.5441,
|
4438 |
+
"step": 6190
|
4439 |
+
},
|
4440 |
+
{
|
4441 |
+
"epoch": 0.15,
|
4442 |
+
"grad_norm": 3.6875,
|
4443 |
+
"learning_rate": 8.000000000000001e-06,
|
4444 |
+
"loss": 0.678,
|
4445 |
+
"step": 6200
|
4446 |
+
},
|
4447 |
+
{
|
4448 |
+
"epoch": 0.16,
|
4449 |
+
"grad_norm": 2.625,
|
4450 |
+
"learning_rate": 7.978947368421052e-06,
|
4451 |
+
"loss": 0.639,
|
4452 |
+
"step": 6210
|
4453 |
+
},
|
4454 |
+
{
|
4455 |
+
"epoch": 0.16,
|
4456 |
+
"grad_norm": 6.09375,
|
4457 |
+
"learning_rate": 7.957894736842105e-06,
|
4458 |
+
"loss": 0.6824,
|
4459 |
+
"step": 6220
|
4460 |
+
},
|
4461 |
+
{
|
4462 |
+
"epoch": 0.16,
|
4463 |
+
"grad_norm": 4.4375,
|
4464 |
+
"learning_rate": 7.936842105263158e-06,
|
4465 |
+
"loss": 0.6219,
|
4466 |
+
"step": 6230
|
4467 |
+
},
|
4468 |
+
{
|
4469 |
+
"epoch": 0.16,
|
4470 |
+
"grad_norm": 3.796875,
|
4471 |
+
"learning_rate": 7.915789473684212e-06,
|
4472 |
+
"loss": 0.6267,
|
4473 |
+
"step": 6240
|
4474 |
+
},
|
4475 |
+
{
|
4476 |
+
"epoch": 0.16,
|
4477 |
+
"grad_norm": 3.453125,
|
4478 |
+
"learning_rate": 7.894736842105265e-06,
|
4479 |
+
"loss": 0.5348,
|
4480 |
+
"step": 6250
|
4481 |
+
},
|
4482 |
+
{
|
4483 |
+
"epoch": 0.16,
|
4484 |
+
"grad_norm": 5.90625,
|
4485 |
+
"learning_rate": 7.873684210526317e-06,
|
4486 |
+
"loss": 0.6523,
|
4487 |
+
"step": 6260
|
4488 |
+
},
|
4489 |
+
{
|
4490 |
+
"epoch": 0.16,
|
4491 |
+
"grad_norm": 8.25,
|
4492 |
+
"learning_rate": 7.85263157894737e-06,
|
4493 |
+
"loss": 0.6024,
|
4494 |
+
"step": 6270
|
4495 |
+
},
|
4496 |
+
{
|
4497 |
+
"epoch": 0.16,
|
4498 |
+
"grad_norm": 2.875,
|
4499 |
+
"learning_rate": 7.831578947368421e-06,
|
4500 |
+
"loss": 0.5896,
|
4501 |
+
"step": 6280
|
4502 |
+
},
|
4503 |
+
{
|
4504 |
+
"epoch": 0.16,
|
4505 |
+
"grad_norm": 7.40625,
|
4506 |
+
"learning_rate": 7.810526315789474e-06,
|
4507 |
+
"loss": 0.7023,
|
4508 |
+
"step": 6290
|
4509 |
+
},
|
4510 |
+
{
|
4511 |
+
"epoch": 0.16,
|
4512 |
+
"grad_norm": 4.09375,
|
4513 |
+
"learning_rate": 7.789473684210526e-06,
|
4514 |
+
"loss": 0.5793,
|
4515 |
+
"step": 6300
|
4516 |
+
},
|
4517 |
+
{
|
4518 |
+
"epoch": 0.16,
|
4519 |
+
"grad_norm": 6.375,
|
4520 |
+
"learning_rate": 7.768421052631579e-06,
|
4521 |
+
"loss": 0.6332,
|
4522 |
+
"step": 6310
|
4523 |
+
},
|
4524 |
+
{
|
4525 |
+
"epoch": 0.16,
|
4526 |
+
"grad_norm": 5.3125,
|
4527 |
+
"learning_rate": 7.747368421052631e-06,
|
4528 |
+
"loss": 0.5124,
|
4529 |
+
"step": 6320
|
4530 |
+
},
|
4531 |
+
{
|
4532 |
+
"epoch": 0.16,
|
4533 |
+
"grad_norm": 2.203125,
|
4534 |
+
"learning_rate": 7.726315789473686e-06,
|
4535 |
+
"loss": 0.6329,
|
4536 |
+
"step": 6330
|
4537 |
+
},
|
4538 |
+
{
|
4539 |
+
"epoch": 0.16,
|
4540 |
+
"grad_norm": 9.0,
|
4541 |
+
"learning_rate": 7.705263157894738e-06,
|
4542 |
+
"loss": 0.7025,
|
4543 |
+
"step": 6340
|
4544 |
+
},
|
4545 |
+
{
|
4546 |
+
"epoch": 0.16,
|
4547 |
+
"grad_norm": 7.09375,
|
4548 |
+
"learning_rate": 7.68421052631579e-06,
|
4549 |
+
"loss": 0.5197,
|
4550 |
+
"step": 6350
|
4551 |
+
},
|
4552 |
+
{
|
4553 |
+
"epoch": 0.16,
|
4554 |
+
"grad_norm": 3.78125,
|
4555 |
+
"learning_rate": 7.663157894736842e-06,
|
4556 |
+
"loss": 0.6173,
|
4557 |
+
"step": 6360
|
4558 |
+
},
|
4559 |
+
{
|
4560 |
+
"epoch": 0.16,
|
4561 |
+
"grad_norm": 2.578125,
|
4562 |
+
"learning_rate": 7.642105263157895e-06,
|
4563 |
+
"loss": 0.5852,
|
4564 |
+
"step": 6370
|
4565 |
+
},
|
4566 |
+
{
|
4567 |
+
"epoch": 0.16,
|
4568 |
+
"grad_norm": 3.109375,
|
4569 |
+
"learning_rate": 7.621052631578948e-06,
|
4570 |
+
"loss": 0.5688,
|
4571 |
+
"step": 6380
|
4572 |
+
},
|
4573 |
+
{
|
4574 |
+
"epoch": 0.16,
|
4575 |
+
"grad_norm": 9.0625,
|
4576 |
+
"learning_rate": 7.600000000000001e-06,
|
4577 |
+
"loss": 0.5457,
|
4578 |
+
"step": 6390
|
4579 |
+
},
|
4580 |
+
{
|
4581 |
+
"epoch": 0.16,
|
4582 |
+
"grad_norm": 1.9296875,
|
4583 |
+
"learning_rate": 7.578947368421054e-06,
|
4584 |
+
"loss": 0.6564,
|
4585 |
+
"step": 6400
|
4586 |
+
},
|
4587 |
+
{
|
4588 |
+
"epoch": 0.16,
|
4589 |
+
"grad_norm": 2.15625,
|
4590 |
+
"learning_rate": 7.557894736842106e-06,
|
4591 |
+
"loss": 0.5177,
|
4592 |
+
"step": 6410
|
4593 |
+
},
|
4594 |
+
{
|
4595 |
+
"epoch": 0.16,
|
4596 |
+
"grad_norm": 3.6875,
|
4597 |
+
"learning_rate": 7.536842105263158e-06,
|
4598 |
+
"loss": 0.6287,
|
4599 |
+
"step": 6420
|
4600 |
+
},
|
4601 |
+
{
|
4602 |
+
"epoch": 0.16,
|
4603 |
+
"grad_norm": 3.578125,
|
4604 |
+
"learning_rate": 7.515789473684211e-06,
|
4605 |
+
"loss": 0.6026,
|
4606 |
+
"step": 6430
|
4607 |
+
},
|
4608 |
+
{
|
4609 |
+
"epoch": 0.16,
|
4610 |
+
"grad_norm": 2.8125,
|
4611 |
+
"learning_rate": 7.494736842105263e-06,
|
4612 |
+
"loss": 0.5795,
|
4613 |
+
"step": 6440
|
4614 |
+
},
|
4615 |
+
{
|
4616 |
+
"epoch": 0.16,
|
4617 |
+
"grad_norm": 3.09375,
|
4618 |
+
"learning_rate": 7.473684210526316e-06,
|
4619 |
+
"loss": 0.6848,
|
4620 |
+
"step": 6450
|
4621 |
+
},
|
4622 |
+
{
|
4623 |
+
"epoch": 0.16,
|
4624 |
+
"grad_norm": 3.421875,
|
4625 |
+
"learning_rate": 7.4526315789473695e-06,
|
4626 |
+
"loss": 0.4139,
|
4627 |
+
"step": 6460
|
4628 |
+
},
|
4629 |
+
{
|
4630 |
+
"epoch": 0.16,
|
4631 |
+
"grad_norm": 5.0,
|
4632 |
+
"learning_rate": 7.431578947368422e-06,
|
4633 |
+
"loss": 0.5696,
|
4634 |
+
"step": 6470
|
4635 |
+
},
|
4636 |
+
{
|
4637 |
+
"epoch": 0.16,
|
4638 |
+
"grad_norm": 3.953125,
|
4639 |
+
"learning_rate": 7.410526315789475e-06,
|
4640 |
+
"loss": 0.6651,
|
4641 |
+
"step": 6480
|
4642 |
+
},
|
4643 |
+
{
|
4644 |
+
"epoch": 0.16,
|
4645 |
+
"grad_norm": 2.40625,
|
4646 |
+
"learning_rate": 7.3894736842105275e-06,
|
4647 |
+
"loss": 0.6557,
|
4648 |
+
"step": 6490
|
4649 |
+
},
|
4650 |
+
{
|
4651 |
+
"epoch": 0.16,
|
4652 |
+
"grad_norm": 8.5625,
|
4653 |
+
"learning_rate": 7.368421052631579e-06,
|
4654 |
+
"loss": 0.5214,
|
4655 |
+
"step": 6500
|
4656 |
+
},
|
4657 |
+
{
|
4658 |
+
"epoch": 0.16,
|
4659 |
+
"eval_loss": 0.6133315563201904,
|
4660 |
+
"eval_runtime": 38.1535,
|
4661 |
+
"eval_samples_per_second": 26.21,
|
4662 |
+
"eval_steps_per_second": 26.21,
|
4663 |
+
"step": 6500
|
4664 |
+
},
|
4665 |
+
{
|
4666 |
+
"epoch": 0.16,
|
4667 |
+
"grad_norm": 2.3125,
|
4668 |
+
"learning_rate": 7.347368421052632e-06,
|
4669 |
+
"loss": 0.6287,
|
4670 |
+
"step": 6510
|
4671 |
+
},
|
4672 |
+
{
|
4673 |
+
"epoch": 0.16,
|
4674 |
+
"grad_norm": 6.75,
|
4675 |
+
"learning_rate": 7.326315789473685e-06,
|
4676 |
+
"loss": 0.6126,
|
4677 |
+
"step": 6520
|
4678 |
+
},
|
4679 |
+
{
|
4680 |
+
"epoch": 0.16,
|
4681 |
+
"grad_norm": 4.90625,
|
4682 |
+
"learning_rate": 7.305263157894737e-06,
|
4683 |
+
"loss": 0.68,
|
4684 |
+
"step": 6530
|
4685 |
+
},
|
4686 |
+
{
|
4687 |
+
"epoch": 0.16,
|
4688 |
+
"grad_norm": 4.96875,
|
4689 |
+
"learning_rate": 7.28421052631579e-06,
|
4690 |
+
"loss": 0.4831,
|
4691 |
+
"step": 6540
|
4692 |
+
},
|
4693 |
+
{
|
4694 |
+
"epoch": 0.16,
|
4695 |
+
"grad_norm": 3.328125,
|
4696 |
+
"learning_rate": 7.263157894736843e-06,
|
4697 |
+
"loss": 0.6983,
|
4698 |
+
"step": 6550
|
4699 |
+
},
|
4700 |
+
{
|
4701 |
+
"epoch": 0.16,
|
4702 |
+
"grad_norm": 2.640625,
|
4703 |
+
"learning_rate": 7.242105263157896e-06,
|
4704 |
+
"loss": 0.5214,
|
4705 |
+
"step": 6560
|
4706 |
+
},
|
4707 |
+
{
|
4708 |
+
"epoch": 0.16,
|
4709 |
+
"grad_norm": 4.5625,
|
4710 |
+
"learning_rate": 7.221052631578948e-06,
|
4711 |
+
"loss": 0.5789,
|
4712 |
+
"step": 6570
|
4713 |
+
},
|
4714 |
+
{
|
4715 |
+
"epoch": 0.16,
|
4716 |
+
"grad_norm": 11.3125,
|
4717 |
+
"learning_rate": 7.2000000000000005e-06,
|
4718 |
+
"loss": 0.6369,
|
4719 |
+
"step": 6580
|
4720 |
+
},
|
4721 |
+
{
|
4722 |
+
"epoch": 0.16,
|
4723 |
+
"grad_norm": 4.5625,
|
4724 |
+
"learning_rate": 7.178947368421053e-06,
|
4725 |
+
"loss": 0.608,
|
4726 |
+
"step": 6590
|
4727 |
+
},
|
4728 |
+
{
|
4729 |
+
"epoch": 0.17,
|
4730 |
+
"grad_norm": 3.25,
|
4731 |
+
"learning_rate": 7.157894736842106e-06,
|
4732 |
+
"loss": 0.6514,
|
4733 |
+
"step": 6600
|
4734 |
+
},
|
4735 |
+
{
|
4736 |
+
"epoch": 0.17,
|
4737 |
+
"grad_norm": 8.5625,
|
4738 |
+
"learning_rate": 7.1368421052631585e-06,
|
4739 |
+
"loss": 0.4964,
|
4740 |
+
"step": 6610
|
4741 |
+
},
|
4742 |
+
{
|
4743 |
+
"epoch": 0.17,
|
4744 |
+
"grad_norm": 5.40625,
|
4745 |
+
"learning_rate": 7.115789473684211e-06,
|
4746 |
+
"loss": 0.6726,
|
4747 |
+
"step": 6620
|
4748 |
+
},
|
4749 |
+
{
|
4750 |
+
"epoch": 0.17,
|
4751 |
+
"grad_norm": 2.65625,
|
4752 |
+
"learning_rate": 7.094736842105265e-06,
|
4753 |
+
"loss": 0.636,
|
4754 |
+
"step": 6630
|
4755 |
+
},
|
4756 |
+
{
|
4757 |
+
"epoch": 0.17,
|
4758 |
+
"grad_norm": 5.5625,
|
4759 |
+
"learning_rate": 7.073684210526316e-06,
|
4760 |
+
"loss": 0.69,
|
4761 |
+
"step": 6640
|
4762 |
+
},
|
4763 |
+
{
|
4764 |
+
"epoch": 0.17,
|
4765 |
+
"grad_norm": 15.375,
|
4766 |
+
"learning_rate": 7.052631578947369e-06,
|
4767 |
+
"loss": 0.5622,
|
4768 |
+
"step": 6650
|
4769 |
+
},
|
4770 |
+
{
|
4771 |
+
"epoch": 0.17,
|
4772 |
+
"grad_norm": 6.5625,
|
4773 |
+
"learning_rate": 7.031578947368422e-06,
|
4774 |
+
"loss": 0.5597,
|
4775 |
+
"step": 6660
|
4776 |
+
},
|
4777 |
+
{
|
4778 |
+
"epoch": 0.17,
|
4779 |
+
"grad_norm": 3.46875,
|
4780 |
+
"learning_rate": 7.010526315789474e-06,
|
4781 |
+
"loss": 0.6246,
|
4782 |
+
"step": 6670
|
4783 |
+
},
|
4784 |
+
{
|
4785 |
+
"epoch": 0.17,
|
4786 |
+
"grad_norm": 3.515625,
|
4787 |
+
"learning_rate": 6.989473684210527e-06,
|
4788 |
+
"loss": 0.5107,
|
4789 |
+
"step": 6680
|
4790 |
+
},
|
4791 |
+
{
|
4792 |
+
"epoch": 0.17,
|
4793 |
+
"grad_norm": 3.84375,
|
4794 |
+
"learning_rate": 6.96842105263158e-06,
|
4795 |
+
"loss": 0.646,
|
4796 |
+
"step": 6690
|
4797 |
+
},
|
4798 |
+
{
|
4799 |
+
"epoch": 0.17,
|
4800 |
+
"grad_norm": 3.5625,
|
4801 |
+
"learning_rate": 6.947368421052632e-06,
|
4802 |
+
"loss": 0.659,
|
4803 |
+
"step": 6700
|
4804 |
+
},
|
4805 |
+
{
|
4806 |
+
"epoch": 0.17,
|
4807 |
+
"grad_norm": 2.703125,
|
4808 |
+
"learning_rate": 6.926315789473684e-06,
|
4809 |
+
"loss": 0.4536,
|
4810 |
+
"step": 6710
|
4811 |
+
},
|
4812 |
+
{
|
4813 |
+
"epoch": 0.17,
|
4814 |
+
"grad_norm": 2.703125,
|
4815 |
+
"learning_rate": 6.905263157894737e-06,
|
4816 |
+
"loss": 0.6416,
|
4817 |
+
"step": 6720
|
4818 |
+
},
|
4819 |
+
{
|
4820 |
+
"epoch": 0.17,
|
4821 |
+
"grad_norm": 2.34375,
|
4822 |
+
"learning_rate": 6.8842105263157895e-06,
|
4823 |
+
"loss": 0.6546,
|
4824 |
+
"step": 6730
|
4825 |
+
},
|
4826 |
+
{
|
4827 |
+
"epoch": 0.17,
|
4828 |
+
"grad_norm": 3.046875,
|
4829 |
+
"learning_rate": 6.863157894736843e-06,
|
4830 |
+
"loss": 0.6834,
|
4831 |
+
"step": 6740
|
4832 |
+
},
|
4833 |
+
{
|
4834 |
+
"epoch": 0.17,
|
4835 |
+
"grad_norm": 5.53125,
|
4836 |
+
"learning_rate": 6.842105263157896e-06,
|
4837 |
+
"loss": 0.6329,
|
4838 |
+
"step": 6750
|
4839 |
+
},
|
4840 |
+
{
|
4841 |
+
"epoch": 0.17,
|
4842 |
+
"grad_norm": 4.75,
|
4843 |
+
"learning_rate": 6.821052631578948e-06,
|
4844 |
+
"loss": 0.5559,
|
4845 |
+
"step": 6760
|
4846 |
+
},
|
4847 |
+
{
|
4848 |
+
"epoch": 0.17,
|
4849 |
+
"grad_norm": 3.484375,
|
4850 |
+
"learning_rate": 6.800000000000001e-06,
|
4851 |
+
"loss": 0.5685,
|
4852 |
+
"step": 6770
|
4853 |
+
},
|
4854 |
+
{
|
4855 |
+
"epoch": 0.17,
|
4856 |
+
"grad_norm": 5.03125,
|
4857 |
+
"learning_rate": 6.778947368421053e-06,
|
4858 |
+
"loss": 0.6151,
|
4859 |
+
"step": 6780
|
4860 |
+
},
|
4861 |
+
{
|
4862 |
+
"epoch": 0.17,
|
4863 |
+
"grad_norm": 3.296875,
|
4864 |
+
"learning_rate": 6.7578947368421054e-06,
|
4865 |
+
"loss": 0.6197,
|
4866 |
+
"step": 6790
|
4867 |
+
},
|
4868 |
+
{
|
4869 |
+
"epoch": 0.17,
|
4870 |
+
"grad_norm": 3.578125,
|
4871 |
+
"learning_rate": 6.736842105263158e-06,
|
4872 |
+
"loss": 0.565,
|
4873 |
+
"step": 6800
|
4874 |
+
},
|
4875 |
+
{
|
4876 |
+
"epoch": 0.17,
|
4877 |
+
"grad_norm": 4.09375,
|
4878 |
+
"learning_rate": 6.715789473684211e-06,
|
4879 |
+
"loss": 0.5766,
|
4880 |
+
"step": 6810
|
4881 |
+
},
|
4882 |
+
{
|
4883 |
+
"epoch": 0.17,
|
4884 |
+
"grad_norm": 2.5625,
|
4885 |
+
"learning_rate": 6.694736842105264e-06,
|
4886 |
+
"loss": 0.5622,
|
4887 |
+
"step": 6820
|
4888 |
+
},
|
4889 |
+
{
|
4890 |
+
"epoch": 0.17,
|
4891 |
+
"grad_norm": 13.125,
|
4892 |
+
"learning_rate": 6.673684210526317e-06,
|
4893 |
+
"loss": 0.6691,
|
4894 |
+
"step": 6830
|
4895 |
+
},
|
4896 |
+
{
|
4897 |
+
"epoch": 0.17,
|
4898 |
+
"grad_norm": 9.375,
|
4899 |
+
"learning_rate": 6.6526315789473695e-06,
|
4900 |
+
"loss": 0.5872,
|
4901 |
+
"step": 6840
|
4902 |
+
},
|
4903 |
+
{
|
4904 |
+
"epoch": 0.17,
|
4905 |
+
"grad_norm": 3.296875,
|
4906 |
+
"learning_rate": 6.631578947368421e-06,
|
4907 |
+
"loss": 0.5883,
|
4908 |
+
"step": 6850
|
4909 |
+
},
|
4910 |
+
{
|
4911 |
+
"epoch": 0.17,
|
4912 |
+
"grad_norm": 12.0625,
|
4913 |
+
"learning_rate": 6.610526315789474e-06,
|
4914 |
+
"loss": 0.6339,
|
4915 |
+
"step": 6860
|
4916 |
+
},
|
4917 |
+
{
|
4918 |
+
"epoch": 0.17,
|
4919 |
+
"grad_norm": 2.96875,
|
4920 |
+
"learning_rate": 6.589473684210527e-06,
|
4921 |
+
"loss": 0.612,
|
4922 |
+
"step": 6870
|
4923 |
+
},
|
4924 |
+
{
|
4925 |
+
"epoch": 0.17,
|
4926 |
+
"grad_norm": 3.34375,
|
4927 |
+
"learning_rate": 6.568421052631579e-06,
|
4928 |
+
"loss": 0.5187,
|
4929 |
+
"step": 6880
|
4930 |
+
},
|
4931 |
+
{
|
4932 |
+
"epoch": 0.17,
|
4933 |
+
"grad_norm": 2.828125,
|
4934 |
+
"learning_rate": 6.547368421052632e-06,
|
4935 |
+
"loss": 0.5306,
|
4936 |
+
"step": 6890
|
4937 |
+
},
|
4938 |
+
{
|
4939 |
+
"epoch": 0.17,
|
4940 |
+
"grad_norm": 3.671875,
|
4941 |
+
"learning_rate": 6.526315789473685e-06,
|
4942 |
+
"loss": 0.5485,
|
4943 |
+
"step": 6900
|
4944 |
+
},
|
4945 |
+
{
|
4946 |
+
"epoch": 0.17,
|
4947 |
+
"grad_norm": 9.5625,
|
4948 |
+
"learning_rate": 6.505263157894738e-06,
|
4949 |
+
"loss": 0.6115,
|
4950 |
+
"step": 6910
|
4951 |
+
},
|
4952 |
+
{
|
4953 |
+
"epoch": 0.17,
|
4954 |
+
"grad_norm": 3.9375,
|
4955 |
+
"learning_rate": 6.484210526315789e-06,
|
4956 |
+
"loss": 0.5446,
|
4957 |
+
"step": 6920
|
4958 |
+
},
|
4959 |
+
{
|
4960 |
+
"epoch": 0.17,
|
4961 |
+
"grad_norm": 1.7109375,
|
4962 |
+
"learning_rate": 6.463157894736843e-06,
|
4963 |
+
"loss": 0.6284,
|
4964 |
+
"step": 6930
|
4965 |
+
},
|
4966 |
+
{
|
4967 |
+
"epoch": 0.17,
|
4968 |
+
"grad_norm": 8.5625,
|
4969 |
+
"learning_rate": 6.442105263157895e-06,
|
4970 |
+
"loss": 0.526,
|
4971 |
+
"step": 6940
|
4972 |
+
},
|
4973 |
+
{
|
4974 |
+
"epoch": 0.17,
|
4975 |
+
"grad_norm": 4.34375,
|
4976 |
+
"learning_rate": 6.421052631578948e-06,
|
4977 |
+
"loss": 0.6324,
|
4978 |
+
"step": 6950
|
4979 |
+
},
|
4980 |
+
{
|
4981 |
+
"epoch": 0.17,
|
4982 |
+
"grad_norm": 5.28125,
|
4983 |
+
"learning_rate": 6.4000000000000006e-06,
|
4984 |
+
"loss": 0.575,
|
4985 |
+
"step": 6960
|
4986 |
+
},
|
4987 |
+
{
|
4988 |
+
"epoch": 0.17,
|
4989 |
+
"grad_norm": 3.09375,
|
4990 |
+
"learning_rate": 6.378947368421053e-06,
|
4991 |
+
"loss": 0.5664,
|
4992 |
+
"step": 6970
|
4993 |
+
},
|
4994 |
+
{
|
4995 |
+
"epoch": 0.17,
|
4996 |
+
"grad_norm": 3.75,
|
4997 |
+
"learning_rate": 6.357894736842106e-06,
|
4998 |
+
"loss": 0.5779,
|
4999 |
+
"step": 6980
|
5000 |
+
},
|
5001 |
+
{
|
5002 |
+
"epoch": 0.17,
|
5003 |
+
"grad_norm": 5.15625,
|
5004 |
+
"learning_rate": 6.336842105263158e-06,
|
5005 |
+
"loss": 0.5258,
|
5006 |
+
"step": 6990
|
5007 |
+
},
|
5008 |
+
{
|
5009 |
+
"epoch": 0.17,
|
5010 |
+
"grad_norm": 5.0625,
|
5011 |
+
"learning_rate": 6.31578947368421e-06,
|
5012 |
+
"loss": 0.5891,
|
5013 |
+
"step": 7000
|
5014 |
+
},
|
5015 |
+
{
|
5016 |
+
"epoch": 0.17,
|
5017 |
+
"eval_loss": 0.6335042119026184,
|
5018 |
+
"eval_runtime": 38.1328,
|
5019 |
+
"eval_samples_per_second": 26.224,
|
5020 |
+
"eval_steps_per_second": 26.224,
|
5021 |
+
"step": 7000
|
5022 |
+
},
|
5023 |
+
{
|
5024 |
+
"epoch": 0.18,
|
5025 |
+
"grad_norm": 5.53125,
|
5026 |
+
"learning_rate": 6.294736842105264e-06,
|
5027 |
+
"loss": 0.524,
|
5028 |
+
"step": 7010
|
5029 |
+
},
|
5030 |
+
{
|
5031 |
+
"epoch": 0.18,
|
5032 |
+
"grad_norm": 4.46875,
|
5033 |
+
"learning_rate": 6.2736842105263165e-06,
|
5034 |
+
"loss": 0.6672,
|
5035 |
+
"step": 7020
|
5036 |
+
},
|
5037 |
+
{
|
5038 |
+
"epoch": 0.18,
|
5039 |
+
"grad_norm": 2.4375,
|
5040 |
+
"learning_rate": 6.252631578947369e-06,
|
5041 |
+
"loss": 0.5906,
|
5042 |
+
"step": 7030
|
5043 |
+
},
|
5044 |
+
{
|
5045 |
+
"epoch": 0.18,
|
5046 |
+
"grad_norm": 3.703125,
|
5047 |
+
"learning_rate": 6.231578947368422e-06,
|
5048 |
+
"loss": 0.6795,
|
5049 |
+
"step": 7040
|
5050 |
+
},
|
5051 |
+
{
|
5052 |
+
"epoch": 0.18,
|
5053 |
+
"grad_norm": 5.09375,
|
5054 |
+
"learning_rate": 6.2105263157894745e-06,
|
5055 |
+
"loss": 0.574,
|
5056 |
+
"step": 7050
|
5057 |
+
},
|
5058 |
+
{
|
5059 |
+
"epoch": 0.18,
|
5060 |
+
"grad_norm": 3.921875,
|
5061 |
+
"learning_rate": 6.189473684210526e-06,
|
5062 |
+
"loss": 0.6008,
|
5063 |
+
"step": 7060
|
5064 |
+
},
|
5065 |
+
{
|
5066 |
+
"epoch": 0.18,
|
5067 |
+
"grad_norm": 5.84375,
|
5068 |
+
"learning_rate": 6.168421052631579e-06,
|
5069 |
+
"loss": 0.5116,
|
5070 |
+
"step": 7070
|
5071 |
+
},
|
5072 |
+
{
|
5073 |
+
"epoch": 0.18,
|
5074 |
+
"grad_norm": 3.203125,
|
5075 |
+
"learning_rate": 6.1473684210526316e-06,
|
5076 |
+
"loss": 0.6047,
|
5077 |
+
"step": 7080
|
5078 |
+
},
|
5079 |
+
{
|
5080 |
+
"epoch": 0.18,
|
5081 |
+
"grad_norm": 2.15625,
|
5082 |
+
"learning_rate": 6.126315789473685e-06,
|
5083 |
+
"loss": 0.3864,
|
5084 |
+
"step": 7090
|
5085 |
+
},
|
5086 |
+
{
|
5087 |
+
"epoch": 0.18,
|
5088 |
+
"grad_norm": 3.65625,
|
5089 |
+
"learning_rate": 6.105263157894738e-06,
|
5090 |
+
"loss": 0.5869,
|
5091 |
+
"step": 7100
|
5092 |
+
},
|
5093 |
+
{
|
5094 |
+
"epoch": 0.18,
|
5095 |
+
"grad_norm": 9.875,
|
5096 |
+
"learning_rate": 6.08421052631579e-06,
|
5097 |
+
"loss": 0.5377,
|
5098 |
+
"step": 7110
|
5099 |
+
},
|
5100 |
+
{
|
5101 |
+
"epoch": 0.18,
|
5102 |
+
"grad_norm": 3.078125,
|
5103 |
+
"learning_rate": 6.063157894736843e-06,
|
5104 |
+
"loss": 0.5649,
|
5105 |
+
"step": 7120
|
5106 |
+
},
|
5107 |
+
{
|
5108 |
+
"epoch": 0.18,
|
5109 |
+
"grad_norm": 4.875,
|
5110 |
+
"learning_rate": 6.042105263157895e-06,
|
5111 |
+
"loss": 0.5955,
|
5112 |
+
"step": 7130
|
5113 |
+
},
|
5114 |
+
{
|
5115 |
+
"epoch": 0.18,
|
5116 |
+
"grad_norm": 3.53125,
|
5117 |
+
"learning_rate": 6.0210526315789475e-06,
|
5118 |
+
"loss": 0.5352,
|
5119 |
+
"step": 7140
|
5120 |
+
},
|
5121 |
+
{
|
5122 |
+
"epoch": 0.18,
|
5123 |
+
"grad_norm": 9.625,
|
5124 |
+
"learning_rate": 6e-06,
|
5125 |
+
"loss": 0.5801,
|
5126 |
+
"step": 7150
|
5127 |
+
},
|
5128 |
+
{
|
5129 |
+
"epoch": 0.18,
|
5130 |
+
"grad_norm": 4.375,
|
5131 |
+
"learning_rate": 5.978947368421053e-06,
|
5132 |
+
"loss": 0.4599,
|
5133 |
+
"step": 7160
|
5134 |
+
},
|
5135 |
+
{
|
5136 |
+
"epoch": 0.18,
|
5137 |
+
"grad_norm": 3.3125,
|
5138 |
+
"learning_rate": 5.9578947368421055e-06,
|
5139 |
+
"loss": 0.5337,
|
5140 |
+
"step": 7170
|
5141 |
+
},
|
5142 |
+
{
|
5143 |
+
"epoch": 0.18,
|
5144 |
+
"grad_norm": 3.40625,
|
5145 |
+
"learning_rate": 5.936842105263159e-06,
|
5146 |
+
"loss": 0.5677,
|
5147 |
+
"step": 7180
|
5148 |
+
},
|
5149 |
+
{
|
5150 |
+
"epoch": 0.18,
|
5151 |
+
"grad_norm": 2.90625,
|
5152 |
+
"learning_rate": 5.915789473684212e-06,
|
5153 |
+
"loss": 0.6926,
|
5154 |
+
"step": 7190
|
5155 |
+
},
|
5156 |
+
{
|
5157 |
+
"epoch": 0.18,
|
5158 |
+
"grad_norm": 3.15625,
|
5159 |
+
"learning_rate": 5.8947368421052634e-06,
|
5160 |
+
"loss": 0.6243,
|
5161 |
+
"step": 7200
|
5162 |
+
},
|
5163 |
+
{
|
5164 |
+
"epoch": 0.18,
|
5165 |
+
"grad_norm": 5.5,
|
5166 |
+
"learning_rate": 5.873684210526316e-06,
|
5167 |
+
"loss": 0.5837,
|
5168 |
+
"step": 7210
|
5169 |
+
},
|
5170 |
+
{
|
5171 |
+
"epoch": 0.18,
|
5172 |
+
"grad_norm": 4.53125,
|
5173 |
+
"learning_rate": 5.852631578947369e-06,
|
5174 |
+
"loss": 0.6096,
|
5175 |
+
"step": 7220
|
5176 |
+
},
|
5177 |
+
{
|
5178 |
+
"epoch": 0.18,
|
5179 |
+
"grad_norm": 3.984375,
|
5180 |
+
"learning_rate": 5.831578947368421e-06,
|
5181 |
+
"loss": 0.5598,
|
5182 |
+
"step": 7230
|
5183 |
+
},
|
5184 |
+
{
|
5185 |
+
"epoch": 0.18,
|
5186 |
+
"grad_norm": 4.40625,
|
5187 |
+
"learning_rate": 5.810526315789474e-06,
|
5188 |
+
"loss": 0.5448,
|
5189 |
+
"step": 7240
|
5190 |
+
},
|
5191 |
+
{
|
5192 |
+
"epoch": 0.18,
|
5193 |
+
"grad_norm": 3.671875,
|
5194 |
+
"learning_rate": 5.789473684210527e-06,
|
5195 |
+
"loss": 0.5487,
|
5196 |
+
"step": 7250
|
5197 |
+
},
|
5198 |
+
{
|
5199 |
+
"epoch": 0.18,
|
5200 |
+
"grad_norm": 10.5625,
|
5201 |
+
"learning_rate": 5.76842105263158e-06,
|
5202 |
+
"loss": 0.5739,
|
5203 |
+
"step": 7260
|
5204 |
+
},
|
5205 |
+
{
|
5206 |
+
"epoch": 0.18,
|
5207 |
+
"grad_norm": 9.4375,
|
5208 |
+
"learning_rate": 5.747368421052633e-06,
|
5209 |
+
"loss": 0.6278,
|
5210 |
+
"step": 7270
|
5211 |
+
},
|
5212 |
+
{
|
5213 |
+
"epoch": 0.18,
|
5214 |
+
"grad_norm": 3.578125,
|
5215 |
+
"learning_rate": 5.726315789473685e-06,
|
5216 |
+
"loss": 0.5801,
|
5217 |
+
"step": 7280
|
5218 |
+
},
|
5219 |
+
{
|
5220 |
+
"epoch": 0.18,
|
5221 |
+
"grad_norm": 3.796875,
|
5222 |
+
"learning_rate": 5.705263157894737e-06,
|
5223 |
+
"loss": 0.4992,
|
5224 |
+
"step": 7290
|
5225 |
+
},
|
5226 |
+
{
|
5227 |
+
"epoch": 0.18,
|
5228 |
+
"grad_norm": 4.8125,
|
5229 |
+
"learning_rate": 5.68421052631579e-06,
|
5230 |
+
"loss": 0.607,
|
5231 |
+
"step": 7300
|
5232 |
+
},
|
5233 |
+
{
|
5234 |
+
"epoch": 0.18,
|
5235 |
+
"grad_norm": 5.0,
|
5236 |
+
"learning_rate": 5.663157894736843e-06,
|
5237 |
+
"loss": 0.5189,
|
5238 |
+
"step": 7310
|
5239 |
+
},
|
5240 |
+
{
|
5241 |
+
"epoch": 0.18,
|
5242 |
+
"grad_norm": 5.78125,
|
5243 |
+
"learning_rate": 5.642105263157895e-06,
|
5244 |
+
"loss": 0.6886,
|
5245 |
+
"step": 7320
|
5246 |
+
},
|
5247 |
+
{
|
5248 |
+
"epoch": 0.18,
|
5249 |
+
"grad_norm": 3.46875,
|
5250 |
+
"learning_rate": 5.621052631578948e-06,
|
5251 |
+
"loss": 0.6127,
|
5252 |
+
"step": 7330
|
5253 |
+
},
|
5254 |
+
{
|
5255 |
+
"epoch": 0.18,
|
5256 |
+
"grad_norm": 3.875,
|
5257 |
+
"learning_rate": 5.600000000000001e-06,
|
5258 |
+
"loss": 0.625,
|
5259 |
+
"step": 7340
|
5260 |
+
},
|
5261 |
+
{
|
5262 |
+
"epoch": 0.18,
|
5263 |
+
"grad_norm": 16.25,
|
5264 |
+
"learning_rate": 5.578947368421052e-06,
|
5265 |
+
"loss": 0.537,
|
5266 |
+
"step": 7350
|
5267 |
+
},
|
5268 |
+
{
|
5269 |
+
"epoch": 0.18,
|
5270 |
+
"grad_norm": 3.90625,
|
5271 |
+
"learning_rate": 5.557894736842105e-06,
|
5272 |
+
"loss": 0.5933,
|
5273 |
+
"step": 7360
|
5274 |
+
},
|
5275 |
+
{
|
5276 |
+
"epoch": 0.18,
|
5277 |
+
"grad_norm": 13.375,
|
5278 |
+
"learning_rate": 5.5368421052631586e-06,
|
5279 |
+
"loss": 0.6001,
|
5280 |
+
"step": 7370
|
5281 |
+
},
|
5282 |
+
{
|
5283 |
+
"epoch": 0.18,
|
5284 |
+
"grad_norm": 14.75,
|
5285 |
+
"learning_rate": 5.515789473684211e-06,
|
5286 |
+
"loss": 0.6809,
|
5287 |
+
"step": 7380
|
5288 |
+
},
|
5289 |
+
{
|
5290 |
+
"epoch": 0.18,
|
5291 |
+
"grad_norm": 5.90625,
|
5292 |
+
"learning_rate": 5.494736842105264e-06,
|
5293 |
+
"loss": 0.5997,
|
5294 |
+
"step": 7390
|
5295 |
+
},
|
5296 |
+
{
|
5297 |
+
"epoch": 0.18,
|
5298 |
+
"grad_norm": 4.03125,
|
5299 |
+
"learning_rate": 5.4736842105263165e-06,
|
5300 |
+
"loss": 0.6418,
|
5301 |
+
"step": 7400
|
5302 |
+
},
|
5303 |
+
{
|
5304 |
+
"epoch": 0.19,
|
5305 |
+
"grad_norm": 3.65625,
|
5306 |
+
"learning_rate": 5.452631578947369e-06,
|
5307 |
+
"loss": 0.5285,
|
5308 |
+
"step": 7410
|
5309 |
+
},
|
5310 |
+
{
|
5311 |
+
"epoch": 0.19,
|
5312 |
+
"grad_norm": 10.0625,
|
5313 |
+
"learning_rate": 5.431578947368421e-06,
|
5314 |
+
"loss": 0.6465,
|
5315 |
+
"step": 7420
|
5316 |
+
},
|
5317 |
+
{
|
5318 |
+
"epoch": 0.19,
|
5319 |
+
"grad_norm": 3.359375,
|
5320 |
+
"learning_rate": 5.410526315789474e-06,
|
5321 |
+
"loss": 0.6697,
|
5322 |
+
"step": 7430
|
5323 |
+
},
|
5324 |
+
{
|
5325 |
+
"epoch": 0.19,
|
5326 |
+
"grad_norm": 4.21875,
|
5327 |
+
"learning_rate": 5.389473684210526e-06,
|
5328 |
+
"loss": 0.5943,
|
5329 |
+
"step": 7440
|
5330 |
+
},
|
5331 |
+
{
|
5332 |
+
"epoch": 0.19,
|
5333 |
+
"grad_norm": 5.875,
|
5334 |
+
"learning_rate": 5.36842105263158e-06,
|
5335 |
+
"loss": 0.6122,
|
5336 |
+
"step": 7450
|
5337 |
+
},
|
5338 |
+
{
|
5339 |
+
"epoch": 0.19,
|
5340 |
+
"grad_norm": 2.4375,
|
5341 |
+
"learning_rate": 5.3473684210526325e-06,
|
5342 |
+
"loss": 0.5825,
|
5343 |
+
"step": 7460
|
5344 |
+
},
|
5345 |
+
{
|
5346 |
+
"epoch": 0.19,
|
5347 |
+
"grad_norm": 27.75,
|
5348 |
+
"learning_rate": 5.326315789473685e-06,
|
5349 |
+
"loss": 0.6151,
|
5350 |
+
"step": 7470
|
5351 |
+
},
|
5352 |
+
{
|
5353 |
+
"epoch": 0.19,
|
5354 |
+
"grad_norm": 3.28125,
|
5355 |
+
"learning_rate": 5.305263157894738e-06,
|
5356 |
+
"loss": 0.621,
|
5357 |
+
"step": 7480
|
5358 |
+
},
|
5359 |
+
{
|
5360 |
+
"epoch": 0.19,
|
5361 |
+
"grad_norm": 3.265625,
|
5362 |
+
"learning_rate": 5.2842105263157896e-06,
|
5363 |
+
"loss": 0.5397,
|
5364 |
+
"step": 7490
|
5365 |
+
},
|
5366 |
+
{
|
5367 |
+
"epoch": 0.19,
|
5368 |
+
"grad_norm": 4.53125,
|
5369 |
+
"learning_rate": 5.263157894736842e-06,
|
5370 |
+
"loss": 0.6428,
|
5371 |
+
"step": 7500
|
5372 |
+
},
|
5373 |
+
{
|
5374 |
+
"epoch": 0.19,
|
5375 |
+
"eval_loss": 0.6119253039360046,
|
5376 |
+
"eval_runtime": 38.1677,
|
5377 |
+
"eval_samples_per_second": 26.2,
|
5378 |
+
"eval_steps_per_second": 26.2,
|
5379 |
+
"step": 7500
|
5380 |
}
|
5381 |
],
|
5382 |
"logging_steps": 10,
|
|
|
5384 |
"num_input_tokens_seen": 0,
|
5385 |
"num_train_epochs": 1,
|
5386 |
"save_steps": 2500,
|
5387 |
+
"total_flos": 1.1800273747968e+17,
|
5388 |
"train_batch_size": 1,
|
5389 |
"trial_name": null,
|
5390 |
"trial_params": null
|