Training in progress, step 1370, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14696
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8a563552bfd5ec7b5ec8bb8ec684f39e32cfd1630c09237883cfbced9bd38c6c
|
| 3 |
size 14696
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 39462
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ce4e92a5ef647defcdc3284ef96dd7be021f6bb7e5a11e35da67409f9db47af0
|
| 3 |
size 39462
|
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 sha256:573bf4a51d609bacbfa177d4b7a66f413284f6d5d46714074842fcd03298cb36
|
| 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:acf4544362da4658ccf0a8d711112b5fd046dc2382e9eb8522b18bbf6c179ba6
|
| 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": 343,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -7242,6 +7242,2393 @@
|
|
| 7242 |
"eval_samples_per_second": 171.924,
|
| 7243 |
"eval_steps_per_second": 85.968,
|
| 7244 |
"step": 1029
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7245 |
}
|
| 7246 |
],
|
| 7247 |
"logging_steps": 1,
|
|
@@ -7256,12 +9643,12 @@
|
|
| 7256 |
"should_evaluate": false,
|
| 7257 |
"should_log": false,
|
| 7258 |
"should_save": true,
|
| 7259 |
-
"should_training_stop":
|
| 7260 |
},
|
| 7261 |
"attributes": {}
|
| 7262 |
}
|
| 7263 |
},
|
| 7264 |
-
"total_flos":
|
| 7265 |
"train_batch_size": 2,
|
| 7266 |
"trial_name": null,
|
| 7267 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.03700427439884935,
|
| 5 |
"eval_steps": 343,
|
| 6 |
+
"global_step": 1370,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 7242 |
"eval_samples_per_second": 171.924,
|
| 7243 |
"eval_steps_per_second": 85.968,
|
| 7244 |
"step": 1029
|
| 7245 |
+
},
|
| 7246 |
+
{
|
| 7247 |
+
"epoch": 0.0278207318473101,
|
| 7248 |
+
"grad_norm": 0.031977105885744095,
|
| 7249 |
+
"learning_rate": 2.9289321881345254e-05,
|
| 7250 |
+
"loss": 11.9146,
|
| 7251 |
+
"step": 1030
|
| 7252 |
+
},
|
| 7253 |
+
{
|
| 7254 |
+
"epoch": 0.027847742266579333,
|
| 7255 |
+
"grad_norm": 0.026009000837802887,
|
| 7256 |
+
"learning_rate": 2.9126169400779535e-05,
|
| 7257 |
+
"loss": 11.917,
|
| 7258 |
+
"step": 1031
|
| 7259 |
+
},
|
| 7260 |
+
{
|
| 7261 |
+
"epoch": 0.027874752685848565,
|
| 7262 |
+
"grad_norm": 0.031822383403778076,
|
| 7263 |
+
"learning_rate": 2.896339510812436e-05,
|
| 7264 |
+
"loss": 11.9205,
|
| 7265 |
+
"step": 1032
|
| 7266 |
+
},
|
| 7267 |
+
{
|
| 7268 |
+
"epoch": 0.0279017631051178,
|
| 7269 |
+
"grad_norm": 0.03649403899908066,
|
| 7270 |
+
"learning_rate": 2.880099987195516e-05,
|
| 7271 |
+
"loss": 11.9184,
|
| 7272 |
+
"step": 1033
|
| 7273 |
+
},
|
| 7274 |
+
{
|
| 7275 |
+
"epoch": 0.02792877352438703,
|
| 7276 |
+
"grad_norm": 0.03650156781077385,
|
| 7277 |
+
"learning_rate": 2.8638984558824777e-05,
|
| 7278 |
+
"loss": 11.9134,
|
| 7279 |
+
"step": 1034
|
| 7280 |
+
},
|
| 7281 |
+
{
|
| 7282 |
+
"epoch": 0.027955783943656266,
|
| 7283 |
+
"grad_norm": 0.03782755509018898,
|
| 7284 |
+
"learning_rate": 2.8477350033258677e-05,
|
| 7285 |
+
"loss": 11.915,
|
| 7286 |
+
"step": 1035
|
| 7287 |
+
},
|
| 7288 |
+
{
|
| 7289 |
+
"epoch": 0.027982794362925497,
|
| 7290 |
+
"grad_norm": 0.027834778651595116,
|
| 7291 |
+
"learning_rate": 2.831609715775042e-05,
|
| 7292 |
+
"loss": 11.9188,
|
| 7293 |
+
"step": 1036
|
| 7294 |
+
},
|
| 7295 |
+
{
|
| 7296 |
+
"epoch": 0.028009804782194732,
|
| 7297 |
+
"grad_norm": 0.02846688963472843,
|
| 7298 |
+
"learning_rate": 2.815522679275704e-05,
|
| 7299 |
+
"loss": 11.9144,
|
| 7300 |
+
"step": 1037
|
| 7301 |
+
},
|
| 7302 |
+
{
|
| 7303 |
+
"epoch": 0.028036815201463963,
|
| 7304 |
+
"grad_norm": 0.03040280006825924,
|
| 7305 |
+
"learning_rate": 2.7994739796694557e-05,
|
| 7306 |
+
"loss": 11.9162,
|
| 7307 |
+
"step": 1038
|
| 7308 |
+
},
|
| 7309 |
+
{
|
| 7310 |
+
"epoch": 0.028063825620733198,
|
| 7311 |
+
"grad_norm": 0.020829658955335617,
|
| 7312 |
+
"learning_rate": 2.783463702593322e-05,
|
| 7313 |
+
"loss": 11.9178,
|
| 7314 |
+
"step": 1039
|
| 7315 |
+
},
|
| 7316 |
+
{
|
| 7317 |
+
"epoch": 0.02809083604000243,
|
| 7318 |
+
"grad_norm": 0.024987973272800446,
|
| 7319 |
+
"learning_rate": 2.7674919334793035e-05,
|
| 7320 |
+
"loss": 11.9133,
|
| 7321 |
+
"step": 1040
|
| 7322 |
+
},
|
| 7323 |
+
{
|
| 7324 |
+
"epoch": 0.028117846459271664,
|
| 7325 |
+
"grad_norm": 0.029873451218008995,
|
| 7326 |
+
"learning_rate": 2.751558757553919e-05,
|
| 7327 |
+
"loss": 11.9143,
|
| 7328 |
+
"step": 1041
|
| 7329 |
+
},
|
| 7330 |
+
{
|
| 7331 |
+
"epoch": 0.028144856878540896,
|
| 7332 |
+
"grad_norm": 0.027596384286880493,
|
| 7333 |
+
"learning_rate": 2.7356642598377603e-05,
|
| 7334 |
+
"loss": 11.9192,
|
| 7335 |
+
"step": 1042
|
| 7336 |
+
},
|
| 7337 |
+
{
|
| 7338 |
+
"epoch": 0.02817186729781013,
|
| 7339 |
+
"grad_norm": 0.04160171374678612,
|
| 7340 |
+
"learning_rate": 2.7198085251450165e-05,
|
| 7341 |
+
"loss": 11.9093,
|
| 7342 |
+
"step": 1043
|
| 7343 |
+
},
|
| 7344 |
+
{
|
| 7345 |
+
"epoch": 0.028198877717079362,
|
| 7346 |
+
"grad_norm": 0.046739108860492706,
|
| 7347 |
+
"learning_rate": 2.703991638083042e-05,
|
| 7348 |
+
"loss": 11.9167,
|
| 7349 |
+
"step": 1044
|
| 7350 |
+
},
|
| 7351 |
+
{
|
| 7352 |
+
"epoch": 0.028225888136348597,
|
| 7353 |
+
"grad_norm": 0.029877908527851105,
|
| 7354 |
+
"learning_rate": 2.688213683051892e-05,
|
| 7355 |
+
"loss": 11.9135,
|
| 7356 |
+
"step": 1045
|
| 7357 |
+
},
|
| 7358 |
+
{
|
| 7359 |
+
"epoch": 0.028252898555617828,
|
| 7360 |
+
"grad_norm": 0.036895956844091415,
|
| 7361 |
+
"learning_rate": 2.672474744243888e-05,
|
| 7362 |
+
"loss": 11.91,
|
| 7363 |
+
"step": 1046
|
| 7364 |
+
},
|
| 7365 |
+
{
|
| 7366 |
+
"epoch": 0.028279908974887063,
|
| 7367 |
+
"grad_norm": 0.0331883430480957,
|
| 7368 |
+
"learning_rate": 2.6567749056431467e-05,
|
| 7369 |
+
"loss": 11.9119,
|
| 7370 |
+
"step": 1047
|
| 7371 |
+
},
|
| 7372 |
+
{
|
| 7373 |
+
"epoch": 0.028306919394156294,
|
| 7374 |
+
"grad_norm": 0.03258933126926422,
|
| 7375 |
+
"learning_rate": 2.6411142510251496e-05,
|
| 7376 |
+
"loss": 11.9141,
|
| 7377 |
+
"step": 1048
|
| 7378 |
+
},
|
| 7379 |
+
{
|
| 7380 |
+
"epoch": 0.02833392981342553,
|
| 7381 |
+
"grad_norm": 0.024799838662147522,
|
| 7382 |
+
"learning_rate": 2.6254928639562826e-05,
|
| 7383 |
+
"loss": 11.916,
|
| 7384 |
+
"step": 1049
|
| 7385 |
+
},
|
| 7386 |
+
{
|
| 7387 |
+
"epoch": 0.02836094023269476,
|
| 7388 |
+
"grad_norm": 0.021943412721157074,
|
| 7389 |
+
"learning_rate": 2.6099108277934103e-05,
|
| 7390 |
+
"loss": 11.9177,
|
| 7391 |
+
"step": 1050
|
| 7392 |
+
},
|
| 7393 |
+
{
|
| 7394 |
+
"epoch": 0.028387950651963995,
|
| 7395 |
+
"grad_norm": 0.048662442713975906,
|
| 7396 |
+
"learning_rate": 2.594368225683407e-05,
|
| 7397 |
+
"loss": 11.9082,
|
| 7398 |
+
"step": 1051
|
| 7399 |
+
},
|
| 7400 |
+
{
|
| 7401 |
+
"epoch": 0.028414961071233227,
|
| 7402 |
+
"grad_norm": 0.03284497186541557,
|
| 7403 |
+
"learning_rate": 2.5788651405627218e-05,
|
| 7404 |
+
"loss": 11.912,
|
| 7405 |
+
"step": 1052
|
| 7406 |
+
},
|
| 7407 |
+
{
|
| 7408 |
+
"epoch": 0.02844197149050246,
|
| 7409 |
+
"grad_norm": 0.032096017152071,
|
| 7410 |
+
"learning_rate": 2.5634016551569518e-05,
|
| 7411 |
+
"loss": 11.9112,
|
| 7412 |
+
"step": 1053
|
| 7413 |
+
},
|
| 7414 |
+
{
|
| 7415 |
+
"epoch": 0.028468981909771693,
|
| 7416 |
+
"grad_norm": 0.03855212405323982,
|
| 7417 |
+
"learning_rate": 2.5479778519803733e-05,
|
| 7418 |
+
"loss": 11.9183,
|
| 7419 |
+
"step": 1054
|
| 7420 |
+
},
|
| 7421 |
+
{
|
| 7422 |
+
"epoch": 0.028495992329040928,
|
| 7423 |
+
"grad_norm": 0.04152005538344383,
|
| 7424 |
+
"learning_rate": 2.5325938133355242e-05,
|
| 7425 |
+
"loss": 11.9143,
|
| 7426 |
+
"step": 1055
|
| 7427 |
+
},
|
| 7428 |
+
{
|
| 7429 |
+
"epoch": 0.02852300274831016,
|
| 7430 |
+
"grad_norm": 0.03331034630537033,
|
| 7431 |
+
"learning_rate": 2.5172496213127515e-05,
|
| 7432 |
+
"loss": 11.9142,
|
| 7433 |
+
"step": 1056
|
| 7434 |
+
},
|
| 7435 |
+
{
|
| 7436 |
+
"epoch": 0.028550013167579394,
|
| 7437 |
+
"grad_norm": 0.026922961696982384,
|
| 7438 |
+
"learning_rate": 2.5019453577897788e-05,
|
| 7439 |
+
"loss": 11.9152,
|
| 7440 |
+
"step": 1057
|
| 7441 |
+
},
|
| 7442 |
+
{
|
| 7443 |
+
"epoch": 0.028577023586848625,
|
| 7444 |
+
"grad_norm": 0.023443300276994705,
|
| 7445 |
+
"learning_rate": 2.4866811044312665e-05,
|
| 7446 |
+
"loss": 11.9119,
|
| 7447 |
+
"step": 1058
|
| 7448 |
+
},
|
| 7449 |
+
{
|
| 7450 |
+
"epoch": 0.02860403400611786,
|
| 7451 |
+
"grad_norm": 0.04499503970146179,
|
| 7452 |
+
"learning_rate": 2.4714569426883837e-05,
|
| 7453 |
+
"loss": 11.9107,
|
| 7454 |
+
"step": 1059
|
| 7455 |
+
},
|
| 7456 |
+
{
|
| 7457 |
+
"epoch": 0.02863104442538709,
|
| 7458 |
+
"grad_norm": 0.03397069498896599,
|
| 7459 |
+
"learning_rate": 2.456272953798361e-05,
|
| 7460 |
+
"loss": 11.9186,
|
| 7461 |
+
"step": 1060
|
| 7462 |
+
},
|
| 7463 |
+
{
|
| 7464 |
+
"epoch": 0.028658054844656326,
|
| 7465 |
+
"grad_norm": 0.02813122048974037,
|
| 7466 |
+
"learning_rate": 2.4411292187840685e-05,
|
| 7467 |
+
"loss": 11.9148,
|
| 7468 |
+
"step": 1061
|
| 7469 |
+
},
|
| 7470 |
+
{
|
| 7471 |
+
"epoch": 0.028685065263925558,
|
| 7472 |
+
"grad_norm": 0.031681545078754425,
|
| 7473 |
+
"learning_rate": 2.4260258184535723e-05,
|
| 7474 |
+
"loss": 11.9102,
|
| 7475 |
+
"step": 1062
|
| 7476 |
+
},
|
| 7477 |
+
{
|
| 7478 |
+
"epoch": 0.028712075683194793,
|
| 7479 |
+
"grad_norm": 0.02378995344042778,
|
| 7480 |
+
"learning_rate": 2.410962833399719e-05,
|
| 7481 |
+
"loss": 11.9172,
|
| 7482 |
+
"step": 1063
|
| 7483 |
+
},
|
| 7484 |
+
{
|
| 7485 |
+
"epoch": 0.028739086102464024,
|
| 7486 |
+
"grad_norm": 0.028632421046495438,
|
| 7487 |
+
"learning_rate": 2.3959403439996907e-05,
|
| 7488 |
+
"loss": 11.9148,
|
| 7489 |
+
"step": 1064
|
| 7490 |
+
},
|
| 7491 |
+
{
|
| 7492 |
+
"epoch": 0.02876609652173326,
|
| 7493 |
+
"grad_norm": 0.033645711839199066,
|
| 7494 |
+
"learning_rate": 2.3809584304145827e-05,
|
| 7495 |
+
"loss": 11.9178,
|
| 7496 |
+
"step": 1065
|
| 7497 |
+
},
|
| 7498 |
+
{
|
| 7499 |
+
"epoch": 0.02879310694100249,
|
| 7500 |
+
"grad_norm": 0.02937261573970318,
|
| 7501 |
+
"learning_rate": 2.36601717258897e-05,
|
| 7502 |
+
"loss": 11.9163,
|
| 7503 |
+
"step": 1066
|
| 7504 |
+
},
|
| 7505 |
+
{
|
| 7506 |
+
"epoch": 0.028820117360271725,
|
| 7507 |
+
"grad_norm": 0.02876032330095768,
|
| 7508 |
+
"learning_rate": 2.3511166502504966e-05,
|
| 7509 |
+
"loss": 11.9136,
|
| 7510 |
+
"step": 1067
|
| 7511 |
+
},
|
| 7512 |
+
{
|
| 7513 |
+
"epoch": 0.028847127779540956,
|
| 7514 |
+
"grad_norm": 0.03480491042137146,
|
| 7515 |
+
"learning_rate": 2.3362569429094295e-05,
|
| 7516 |
+
"loss": 11.9131,
|
| 7517 |
+
"step": 1068
|
| 7518 |
+
},
|
| 7519 |
+
{
|
| 7520 |
+
"epoch": 0.02887413819881019,
|
| 7521 |
+
"grad_norm": 0.05046708509325981,
|
| 7522 |
+
"learning_rate": 2.3214381298582476e-05,
|
| 7523 |
+
"loss": 11.9113,
|
| 7524 |
+
"step": 1069
|
| 7525 |
+
},
|
| 7526 |
+
{
|
| 7527 |
+
"epoch": 0.028901148618079422,
|
| 7528 |
+
"grad_norm": 0.03661332651972771,
|
| 7529 |
+
"learning_rate": 2.3066602901712108e-05,
|
| 7530 |
+
"loss": 11.9109,
|
| 7531 |
+
"step": 1070
|
| 7532 |
+
},
|
| 7533 |
+
{
|
| 7534 |
+
"epoch": 0.028928159037348657,
|
| 7535 |
+
"grad_norm": 0.036345239728689194,
|
| 7536 |
+
"learning_rate": 2.291923502703951e-05,
|
| 7537 |
+
"loss": 11.9152,
|
| 7538 |
+
"step": 1071
|
| 7539 |
+
},
|
| 7540 |
+
{
|
| 7541 |
+
"epoch": 0.02895516945661789,
|
| 7542 |
+
"grad_norm": 0.03236823529005051,
|
| 7543 |
+
"learning_rate": 2.277227846093035e-05,
|
| 7544 |
+
"loss": 11.9159,
|
| 7545 |
+
"step": 1072
|
| 7546 |
+
},
|
| 7547 |
+
{
|
| 7548 |
+
"epoch": 0.028982179875887124,
|
| 7549 |
+
"grad_norm": 0.024910030886530876,
|
| 7550 |
+
"learning_rate": 2.2625733987555543e-05,
|
| 7551 |
+
"loss": 11.9158,
|
| 7552 |
+
"step": 1073
|
| 7553 |
+
},
|
| 7554 |
+
{
|
| 7555 |
+
"epoch": 0.029009190295156355,
|
| 7556 |
+
"grad_norm": 0.02995794452726841,
|
| 7557 |
+
"learning_rate": 2.2479602388887012e-05,
|
| 7558 |
+
"loss": 11.9131,
|
| 7559 |
+
"step": 1074
|
| 7560 |
+
},
|
| 7561 |
+
{
|
| 7562 |
+
"epoch": 0.02903620071442559,
|
| 7563 |
+
"grad_norm": 0.028216613456606865,
|
| 7564 |
+
"learning_rate": 2.2333884444693652e-05,
|
| 7565 |
+
"loss": 11.914,
|
| 7566 |
+
"step": 1075
|
| 7567 |
+
},
|
| 7568 |
+
{
|
| 7569 |
+
"epoch": 0.02906321113369482,
|
| 7570 |
+
"grad_norm": 0.024038277566432953,
|
| 7571 |
+
"learning_rate": 2.2188580932536984e-05,
|
| 7572 |
+
"loss": 11.9083,
|
| 7573 |
+
"step": 1076
|
| 7574 |
+
},
|
| 7575 |
+
{
|
| 7576 |
+
"epoch": 0.029090221552964056,
|
| 7577 |
+
"grad_norm": 0.03442312404513359,
|
| 7578 |
+
"learning_rate": 2.2043692627767076e-05,
|
| 7579 |
+
"loss": 11.9213,
|
| 7580 |
+
"step": 1077
|
| 7581 |
+
},
|
| 7582 |
+
{
|
| 7583 |
+
"epoch": 0.029117231972233287,
|
| 7584 |
+
"grad_norm": 0.020812995731830597,
|
| 7585 |
+
"learning_rate": 2.1899220303518465e-05,
|
| 7586 |
+
"loss": 11.9112,
|
| 7587 |
+
"step": 1078
|
| 7588 |
+
},
|
| 7589 |
+
{
|
| 7590 |
+
"epoch": 0.029144242391502522,
|
| 7591 |
+
"grad_norm": 0.03081013821065426,
|
| 7592 |
+
"learning_rate": 2.175516473070599e-05,
|
| 7593 |
+
"loss": 11.9138,
|
| 7594 |
+
"step": 1079
|
| 7595 |
+
},
|
| 7596 |
+
{
|
| 7597 |
+
"epoch": 0.029171252810771753,
|
| 7598 |
+
"grad_norm": 0.0224771685898304,
|
| 7599 |
+
"learning_rate": 2.1611526678020654e-05,
|
| 7600 |
+
"loss": 11.9211,
|
| 7601 |
+
"step": 1080
|
| 7602 |
+
},
|
| 7603 |
+
{
|
| 7604 |
+
"epoch": 0.02919826323004099,
|
| 7605 |
+
"grad_norm": 0.041392020881175995,
|
| 7606 |
+
"learning_rate": 2.146830691192553e-05,
|
| 7607 |
+
"loss": 11.9109,
|
| 7608 |
+
"step": 1081
|
| 7609 |
+
},
|
| 7610 |
+
{
|
| 7611 |
+
"epoch": 0.02922527364931022,
|
| 7612 |
+
"grad_norm": 0.04150688648223877,
|
| 7613 |
+
"learning_rate": 2.132550619665168e-05,
|
| 7614 |
+
"loss": 11.9145,
|
| 7615 |
+
"step": 1082
|
| 7616 |
+
},
|
| 7617 |
+
{
|
| 7618 |
+
"epoch": 0.029252284068579455,
|
| 7619 |
+
"grad_norm": 0.03591577708721161,
|
| 7620 |
+
"learning_rate": 2.118312529419414e-05,
|
| 7621 |
+
"loss": 11.9155,
|
| 7622 |
+
"step": 1083
|
| 7623 |
+
},
|
| 7624 |
+
{
|
| 7625 |
+
"epoch": 0.029279294487848686,
|
| 7626 |
+
"grad_norm": 0.03724061697721481,
|
| 7627 |
+
"learning_rate": 2.1041164964307748e-05,
|
| 7628 |
+
"loss": 11.914,
|
| 7629 |
+
"step": 1084
|
| 7630 |
+
},
|
| 7631 |
+
{
|
| 7632 |
+
"epoch": 0.02930630490711792,
|
| 7633 |
+
"grad_norm": 0.0344524085521698,
|
| 7634 |
+
"learning_rate": 2.0899625964503113e-05,
|
| 7635 |
+
"loss": 11.915,
|
| 7636 |
+
"step": 1085
|
| 7637 |
+
},
|
| 7638 |
+
{
|
| 7639 |
+
"epoch": 0.029333315326387152,
|
| 7640 |
+
"grad_norm": 0.028263026848435402,
|
| 7641 |
+
"learning_rate": 2.0758509050042618e-05,
|
| 7642 |
+
"loss": 11.9162,
|
| 7643 |
+
"step": 1086
|
| 7644 |
+
},
|
| 7645 |
+
{
|
| 7646 |
+
"epoch": 0.029360325745656387,
|
| 7647 |
+
"grad_norm": 0.028055129572749138,
|
| 7648 |
+
"learning_rate": 2.0617814973936422e-05,
|
| 7649 |
+
"loss": 11.9133,
|
| 7650 |
+
"step": 1087
|
| 7651 |
+
},
|
| 7652 |
+
{
|
| 7653 |
+
"epoch": 0.029387336164925622,
|
| 7654 |
+
"grad_norm": 0.05751683562994003,
|
| 7655 |
+
"learning_rate": 2.0477544486938305e-05,
|
| 7656 |
+
"loss": 11.9149,
|
| 7657 |
+
"step": 1088
|
| 7658 |
+
},
|
| 7659 |
+
{
|
| 7660 |
+
"epoch": 0.029414346584194853,
|
| 7661 |
+
"grad_norm": 0.03942582756280899,
|
| 7662 |
+
"learning_rate": 2.0337698337541787e-05,
|
| 7663 |
+
"loss": 11.9161,
|
| 7664 |
+
"step": 1089
|
| 7665 |
+
},
|
| 7666 |
+
{
|
| 7667 |
+
"epoch": 0.029441357003464088,
|
| 7668 |
+
"grad_norm": 0.02929741144180298,
|
| 7669 |
+
"learning_rate": 2.0198277271976052e-05,
|
| 7670 |
+
"loss": 11.9055,
|
| 7671 |
+
"step": 1090
|
| 7672 |
+
},
|
| 7673 |
+
{
|
| 7674 |
+
"epoch": 0.02946836742273332,
|
| 7675 |
+
"grad_norm": 0.04245249181985855,
|
| 7676 |
+
"learning_rate": 2.00592820342021e-05,
|
| 7677 |
+
"loss": 11.9179,
|
| 7678 |
+
"step": 1091
|
| 7679 |
+
},
|
| 7680 |
+
{
|
| 7681 |
+
"epoch": 0.029495377842002554,
|
| 7682 |
+
"grad_norm": 0.033552706241607666,
|
| 7683 |
+
"learning_rate": 1.9920713365908583e-05,
|
| 7684 |
+
"loss": 11.918,
|
| 7685 |
+
"step": 1092
|
| 7686 |
+
},
|
| 7687 |
+
{
|
| 7688 |
+
"epoch": 0.029522388261271786,
|
| 7689 |
+
"grad_norm": 0.02807275578379631,
|
| 7690 |
+
"learning_rate": 1.9782572006507994e-05,
|
| 7691 |
+
"loss": 11.9161,
|
| 7692 |
+
"step": 1093
|
| 7693 |
+
},
|
| 7694 |
+
{
|
| 7695 |
+
"epoch": 0.02954939868054102,
|
| 7696 |
+
"grad_norm": 0.0388345792889595,
|
| 7697 |
+
"learning_rate": 1.9644858693132628e-05,
|
| 7698 |
+
"loss": 11.9169,
|
| 7699 |
+
"step": 1094
|
| 7700 |
+
},
|
| 7701 |
+
{
|
| 7702 |
+
"epoch": 0.029576409099810252,
|
| 7703 |
+
"grad_norm": 0.023157969117164612,
|
| 7704 |
+
"learning_rate": 1.950757416063077e-05,
|
| 7705 |
+
"loss": 11.9135,
|
| 7706 |
+
"step": 1095
|
| 7707 |
+
},
|
| 7708 |
+
{
|
| 7709 |
+
"epoch": 0.029603419519079487,
|
| 7710 |
+
"grad_norm": 0.030544571578502655,
|
| 7711 |
+
"learning_rate": 1.9370719141562688e-05,
|
| 7712 |
+
"loss": 11.9075,
|
| 7713 |
+
"step": 1096
|
| 7714 |
+
},
|
| 7715 |
+
{
|
| 7716 |
+
"epoch": 0.029630429938348718,
|
| 7717 |
+
"grad_norm": 0.0371607206761837,
|
| 7718 |
+
"learning_rate": 1.923429436619668e-05,
|
| 7719 |
+
"loss": 11.9131,
|
| 7720 |
+
"step": 1097
|
| 7721 |
+
},
|
| 7722 |
+
{
|
| 7723 |
+
"epoch": 0.029657440357617953,
|
| 7724 |
+
"grad_norm": 0.025847923010587692,
|
| 7725 |
+
"learning_rate": 1.9098300562505266e-05,
|
| 7726 |
+
"loss": 11.9153,
|
| 7727 |
+
"step": 1098
|
| 7728 |
+
},
|
| 7729 |
+
{
|
| 7730 |
+
"epoch": 0.029684450776887184,
|
| 7731 |
+
"grad_norm": 0.04432433098554611,
|
| 7732 |
+
"learning_rate": 1.8962738456161223e-05,
|
| 7733 |
+
"loss": 11.9086,
|
| 7734 |
+
"step": 1099
|
| 7735 |
+
},
|
| 7736 |
+
{
|
| 7737 |
+
"epoch": 0.02971146119615642,
|
| 7738 |
+
"grad_norm": 0.03393876925110817,
|
| 7739 |
+
"learning_rate": 1.882760877053388e-05,
|
| 7740 |
+
"loss": 11.9142,
|
| 7741 |
+
"step": 1100
|
| 7742 |
+
},
|
| 7743 |
+
{
|
| 7744 |
+
"epoch": 0.02973847161542565,
|
| 7745 |
+
"grad_norm": 0.03972648084163666,
|
| 7746 |
+
"learning_rate": 1.8692912226685012e-05,
|
| 7747 |
+
"loss": 11.9116,
|
| 7748 |
+
"step": 1101
|
| 7749 |
+
},
|
| 7750 |
+
{
|
| 7751 |
+
"epoch": 0.029765482034694885,
|
| 7752 |
+
"grad_norm": 0.038004882633686066,
|
| 7753 |
+
"learning_rate": 1.855864954336517e-05,
|
| 7754 |
+
"loss": 11.9143,
|
| 7755 |
+
"step": 1102
|
| 7756 |
+
},
|
| 7757 |
+
{
|
| 7758 |
+
"epoch": 0.029792492453964117,
|
| 7759 |
+
"grad_norm": 0.039161406457424164,
|
| 7760 |
+
"learning_rate": 1.8424821437009764e-05,
|
| 7761 |
+
"loss": 11.9147,
|
| 7762 |
+
"step": 1103
|
| 7763 |
+
},
|
| 7764 |
+
{
|
| 7765 |
+
"epoch": 0.02981950287323335,
|
| 7766 |
+
"grad_norm": 0.03622069209814072,
|
| 7767 |
+
"learning_rate": 1.8291428621735352e-05,
|
| 7768 |
+
"loss": 11.9149,
|
| 7769 |
+
"step": 1104
|
| 7770 |
+
},
|
| 7771 |
+
{
|
| 7772 |
+
"epoch": 0.029846513292502583,
|
| 7773 |
+
"grad_norm": 0.025561653077602386,
|
| 7774 |
+
"learning_rate": 1.8158471809335654e-05,
|
| 7775 |
+
"loss": 11.9152,
|
| 7776 |
+
"step": 1105
|
| 7777 |
+
},
|
| 7778 |
+
{
|
| 7779 |
+
"epoch": 0.029873523711771818,
|
| 7780 |
+
"grad_norm": 0.05868985503911972,
|
| 7781 |
+
"learning_rate": 1.8025951709277898e-05,
|
| 7782 |
+
"loss": 11.9149,
|
| 7783 |
+
"step": 1106
|
| 7784 |
+
},
|
| 7785 |
+
{
|
| 7786 |
+
"epoch": 0.02990053413104105,
|
| 7787 |
+
"grad_norm": 0.031001469120383263,
|
| 7788 |
+
"learning_rate": 1.789386902869893e-05,
|
| 7789 |
+
"loss": 11.9139,
|
| 7790 |
+
"step": 1107
|
| 7791 |
+
},
|
| 7792 |
+
{
|
| 7793 |
+
"epoch": 0.029927544550310284,
|
| 7794 |
+
"grad_norm": 0.034948207437992096,
|
| 7795 |
+
"learning_rate": 1.776222447240159e-05,
|
| 7796 |
+
"loss": 11.9145,
|
| 7797 |
+
"step": 1108
|
| 7798 |
+
},
|
| 7799 |
+
{
|
| 7800 |
+
"epoch": 0.029954554969579515,
|
| 7801 |
+
"grad_norm": 0.03424816578626633,
|
| 7802 |
+
"learning_rate": 1.7631018742850768e-05,
|
| 7803 |
+
"loss": 11.9133,
|
| 7804 |
+
"step": 1109
|
| 7805 |
+
},
|
| 7806 |
+
{
|
| 7807 |
+
"epoch": 0.02998156538884875,
|
| 7808 |
+
"grad_norm": 0.03631990775465965,
|
| 7809 |
+
"learning_rate": 1.750025254016978e-05,
|
| 7810 |
+
"loss": 11.9162,
|
| 7811 |
+
"step": 1110
|
| 7812 |
+
},
|
| 7813 |
+
{
|
| 7814 |
+
"epoch": 0.03000857580811798,
|
| 7815 |
+
"grad_norm": 0.025225920602679253,
|
| 7816 |
+
"learning_rate": 1.7369926562136552e-05,
|
| 7817 |
+
"loss": 11.9097,
|
| 7818 |
+
"step": 1111
|
| 7819 |
+
},
|
| 7820 |
+
{
|
| 7821 |
+
"epoch": 0.030035586227387216,
|
| 7822 |
+
"grad_norm": 0.028935188427567482,
|
| 7823 |
+
"learning_rate": 1.7240041504180017e-05,
|
| 7824 |
+
"loss": 11.9097,
|
| 7825 |
+
"step": 1112
|
| 7826 |
+
},
|
| 7827 |
+
{
|
| 7828 |
+
"epoch": 0.030062596646656448,
|
| 7829 |
+
"grad_norm": 0.04426940903067589,
|
| 7830 |
+
"learning_rate": 1.711059805937628e-05,
|
| 7831 |
+
"loss": 11.9172,
|
| 7832 |
+
"step": 1113
|
| 7833 |
+
},
|
| 7834 |
+
{
|
| 7835 |
+
"epoch": 0.030089607065925682,
|
| 7836 |
+
"grad_norm": 0.03160575404763222,
|
| 7837 |
+
"learning_rate": 1.6981596918444953e-05,
|
| 7838 |
+
"loss": 11.9206,
|
| 7839 |
+
"step": 1114
|
| 7840 |
+
},
|
| 7841 |
+
{
|
| 7842 |
+
"epoch": 0.030116617485194914,
|
| 7843 |
+
"grad_norm": 0.03350413218140602,
|
| 7844 |
+
"learning_rate": 1.6853038769745467e-05,
|
| 7845 |
+
"loss": 11.9142,
|
| 7846 |
+
"step": 1115
|
| 7847 |
+
},
|
| 7848 |
+
{
|
| 7849 |
+
"epoch": 0.03014362790446415,
|
| 7850 |
+
"grad_norm": 0.02894110232591629,
|
| 7851 |
+
"learning_rate": 1.6724924299273514e-05,
|
| 7852 |
+
"loss": 11.9201,
|
| 7853 |
+
"step": 1116
|
| 7854 |
+
},
|
| 7855 |
+
{
|
| 7856 |
+
"epoch": 0.03017063832373338,
|
| 7857 |
+
"grad_norm": 0.02642958238720894,
|
| 7858 |
+
"learning_rate": 1.6597254190657184e-05,
|
| 7859 |
+
"loss": 11.9156,
|
| 7860 |
+
"step": 1117
|
| 7861 |
+
},
|
| 7862 |
+
{
|
| 7863 |
+
"epoch": 0.030197648743002615,
|
| 7864 |
+
"grad_norm": 0.043042778968811035,
|
| 7865 |
+
"learning_rate": 1.6470029125153464e-05,
|
| 7866 |
+
"loss": 11.9121,
|
| 7867 |
+
"step": 1118
|
| 7868 |
+
},
|
| 7869 |
+
{
|
| 7870 |
+
"epoch": 0.030224659162271846,
|
| 7871 |
+
"grad_norm": 0.03481782600283623,
|
| 7872 |
+
"learning_rate": 1.634324978164453e-05,
|
| 7873 |
+
"loss": 11.9122,
|
| 7874 |
+
"step": 1119
|
| 7875 |
+
},
|
| 7876 |
+
{
|
| 7877 |
+
"epoch": 0.03025166958154108,
|
| 7878 |
+
"grad_norm": 0.03108305297791958,
|
| 7879 |
+
"learning_rate": 1.621691683663418e-05,
|
| 7880 |
+
"loss": 11.9152,
|
| 7881 |
+
"step": 1120
|
| 7882 |
+
},
|
| 7883 |
+
{
|
| 7884 |
+
"epoch": 0.030278680000810312,
|
| 7885 |
+
"grad_norm": 0.025693703442811966,
|
| 7886 |
+
"learning_rate": 1.6091030964244192e-05,
|
| 7887 |
+
"loss": 11.919,
|
| 7888 |
+
"step": 1121
|
| 7889 |
+
},
|
| 7890 |
+
{
|
| 7891 |
+
"epoch": 0.030305690420079547,
|
| 7892 |
+
"grad_norm": 0.026021650061011314,
|
| 7893 |
+
"learning_rate": 1.5965592836210743e-05,
|
| 7894 |
+
"loss": 11.9156,
|
| 7895 |
+
"step": 1122
|
| 7896 |
+
},
|
| 7897 |
+
{
|
| 7898 |
+
"epoch": 0.03033270083934878,
|
| 7899 |
+
"grad_norm": 0.026449911296367645,
|
| 7900 |
+
"learning_rate": 1.584060312188078e-05,
|
| 7901 |
+
"loss": 11.9165,
|
| 7902 |
+
"step": 1123
|
| 7903 |
+
},
|
| 7904 |
+
{
|
| 7905 |
+
"epoch": 0.030359711258618013,
|
| 7906 |
+
"grad_norm": 0.028504110872745514,
|
| 7907 |
+
"learning_rate": 1.5716062488208493e-05,
|
| 7908 |
+
"loss": 11.9125,
|
| 7909 |
+
"step": 1124
|
| 7910 |
+
},
|
| 7911 |
+
{
|
| 7912 |
+
"epoch": 0.030386721677887245,
|
| 7913 |
+
"grad_norm": 0.03680235892534256,
|
| 7914 |
+
"learning_rate": 1.55919715997518e-05,
|
| 7915 |
+
"loss": 11.9106,
|
| 7916 |
+
"step": 1125
|
| 7917 |
+
},
|
| 7918 |
+
{
|
| 7919 |
+
"epoch": 0.03041373209715648,
|
| 7920 |
+
"grad_norm": 0.028458967804908752,
|
| 7921 |
+
"learning_rate": 1.5468331118668656e-05,
|
| 7922 |
+
"loss": 11.9174,
|
| 7923 |
+
"step": 1126
|
| 7924 |
+
},
|
| 7925 |
+
{
|
| 7926 |
+
"epoch": 0.03044074251642571,
|
| 7927 |
+
"grad_norm": 0.026577237993478775,
|
| 7928 |
+
"learning_rate": 1.5345141704713674e-05,
|
| 7929 |
+
"loss": 11.914,
|
| 7930 |
+
"step": 1127
|
| 7931 |
+
},
|
| 7932 |
+
{
|
| 7933 |
+
"epoch": 0.030467752935694946,
|
| 7934 |
+
"grad_norm": 0.028341144323349,
|
| 7935 |
+
"learning_rate": 1.5222404015234482e-05,
|
| 7936 |
+
"loss": 11.9119,
|
| 7937 |
+
"step": 1128
|
| 7938 |
+
},
|
| 7939 |
+
{
|
| 7940 |
+
"epoch": 0.030494763354964177,
|
| 7941 |
+
"grad_norm": 0.015995662659406662,
|
| 7942 |
+
"learning_rate": 1.5100118705168365e-05,
|
| 7943 |
+
"loss": 11.9122,
|
| 7944 |
+
"step": 1129
|
| 7945 |
+
},
|
| 7946 |
+
{
|
| 7947 |
+
"epoch": 0.030521773774233412,
|
| 7948 |
+
"grad_norm": 0.023735560476779938,
|
| 7949 |
+
"learning_rate": 1.4978286427038601e-05,
|
| 7950 |
+
"loss": 11.9165,
|
| 7951 |
+
"step": 1130
|
| 7952 |
+
},
|
| 7953 |
+
{
|
| 7954 |
+
"epoch": 0.030548784193502643,
|
| 7955 |
+
"grad_norm": 0.02409549616277218,
|
| 7956 |
+
"learning_rate": 1.4856907830951084e-05,
|
| 7957 |
+
"loss": 11.9162,
|
| 7958 |
+
"step": 1131
|
| 7959 |
+
},
|
| 7960 |
+
{
|
| 7961 |
+
"epoch": 0.03057579461277188,
|
| 7962 |
+
"grad_norm": 0.02909088507294655,
|
| 7963 |
+
"learning_rate": 1.4735983564590783e-05,
|
| 7964 |
+
"loss": 11.9112,
|
| 7965 |
+
"step": 1132
|
| 7966 |
+
},
|
| 7967 |
+
{
|
| 7968 |
+
"epoch": 0.03060280503204111,
|
| 7969 |
+
"grad_norm": 0.030376920476555824,
|
| 7970 |
+
"learning_rate": 1.4615514273218433e-05,
|
| 7971 |
+
"loss": 11.9207,
|
| 7972 |
+
"step": 1133
|
| 7973 |
+
},
|
| 7974 |
+
{
|
| 7975 |
+
"epoch": 0.030629815451310344,
|
| 7976 |
+
"grad_norm": 0.0357181541621685,
|
| 7977 |
+
"learning_rate": 1.4495500599666901e-05,
|
| 7978 |
+
"loss": 11.9153,
|
| 7979 |
+
"step": 1134
|
| 7980 |
+
},
|
| 7981 |
+
{
|
| 7982 |
+
"epoch": 0.030656825870579576,
|
| 7983 |
+
"grad_norm": 0.02690085954964161,
|
| 7984 |
+
"learning_rate": 1.437594318433787e-05,
|
| 7985 |
+
"loss": 11.9123,
|
| 7986 |
+
"step": 1135
|
| 7987 |
+
},
|
| 7988 |
+
{
|
| 7989 |
+
"epoch": 0.03068383628984881,
|
| 7990 |
+
"grad_norm": 0.021294275298714638,
|
| 7991 |
+
"learning_rate": 1.4256842665198377e-05,
|
| 7992 |
+
"loss": 11.9101,
|
| 7993 |
+
"step": 1136
|
| 7994 |
+
},
|
| 7995 |
+
{
|
| 7996 |
+
"epoch": 0.030710846709118042,
|
| 7997 |
+
"grad_norm": 0.030441630631685257,
|
| 7998 |
+
"learning_rate": 1.4138199677777464e-05,
|
| 7999 |
+
"loss": 11.9173,
|
| 8000 |
+
"step": 1137
|
| 8001 |
+
},
|
| 8002 |
+
{
|
| 8003 |
+
"epoch": 0.030737857128387277,
|
| 8004 |
+
"grad_norm": 0.021473079919815063,
|
| 8005 |
+
"learning_rate": 1.4020014855162755e-05,
|
| 8006 |
+
"loss": 11.9148,
|
| 8007 |
+
"step": 1138
|
| 8008 |
+
},
|
| 8009 |
+
{
|
| 8010 |
+
"epoch": 0.030764867547656508,
|
| 8011 |
+
"grad_norm": 0.02123764157295227,
|
| 8012 |
+
"learning_rate": 1.3902288827997035e-05,
|
| 8013 |
+
"loss": 11.9168,
|
| 8014 |
+
"step": 1139
|
| 8015 |
+
},
|
| 8016 |
+
{
|
| 8017 |
+
"epoch": 0.030791877966925743,
|
| 8018 |
+
"grad_norm": 0.040539633482694626,
|
| 8019 |
+
"learning_rate": 1.3785022224474942e-05,
|
| 8020 |
+
"loss": 11.9106,
|
| 8021 |
+
"step": 1140
|
| 8022 |
+
},
|
| 8023 |
+
{
|
| 8024 |
+
"epoch": 0.030818888386194974,
|
| 8025 |
+
"grad_norm": 0.03444276005029678,
|
| 8026 |
+
"learning_rate": 1.3668215670339569e-05,
|
| 8027 |
+
"loss": 11.9142,
|
| 8028 |
+
"step": 1141
|
| 8029 |
+
},
|
| 8030 |
+
{
|
| 8031 |
+
"epoch": 0.03084589880546421,
|
| 8032 |
+
"grad_norm": 0.03476952388882637,
|
| 8033 |
+
"learning_rate": 1.3551869788879212e-05,
|
| 8034 |
+
"loss": 11.9189,
|
| 8035 |
+
"step": 1142
|
| 8036 |
+
},
|
| 8037 |
+
{
|
| 8038 |
+
"epoch": 0.03087290922473344,
|
| 8039 |
+
"grad_norm": 0.038158733397722244,
|
| 8040 |
+
"learning_rate": 1.3435985200923939e-05,
|
| 8041 |
+
"loss": 11.912,
|
| 8042 |
+
"step": 1143
|
| 8043 |
+
},
|
| 8044 |
+
{
|
| 8045 |
+
"epoch": 0.030899919644002675,
|
| 8046 |
+
"grad_norm": 0.02733631245791912,
|
| 8047 |
+
"learning_rate": 1.332056252484234e-05,
|
| 8048 |
+
"loss": 11.9163,
|
| 8049 |
+
"step": 1144
|
| 8050 |
+
},
|
| 8051 |
+
{
|
| 8052 |
+
"epoch": 0.030926930063271907,
|
| 8053 |
+
"grad_norm": 0.03728555142879486,
|
| 8054 |
+
"learning_rate": 1.3205602376538163e-05,
|
| 8055 |
+
"loss": 11.9198,
|
| 8056 |
+
"step": 1145
|
| 8057 |
+
},
|
| 8058 |
+
{
|
| 8059 |
+
"epoch": 0.03095394048254114,
|
| 8060 |
+
"grad_norm": 0.030959341675043106,
|
| 8061 |
+
"learning_rate": 1.3091105369447165e-05,
|
| 8062 |
+
"loss": 11.9133,
|
| 8063 |
+
"step": 1146
|
| 8064 |
+
},
|
| 8065 |
+
{
|
| 8066 |
+
"epoch": 0.030980950901810373,
|
| 8067 |
+
"grad_norm": 0.02419600822031498,
|
| 8068 |
+
"learning_rate": 1.2977072114533684e-05,
|
| 8069 |
+
"loss": 11.9113,
|
| 8070 |
+
"step": 1147
|
| 8071 |
+
},
|
| 8072 |
+
{
|
| 8073 |
+
"epoch": 0.031007961321079608,
|
| 8074 |
+
"grad_norm": 0.024129154160618782,
|
| 8075 |
+
"learning_rate": 1.2863503220287431e-05,
|
| 8076 |
+
"loss": 11.9131,
|
| 8077 |
+
"step": 1148
|
| 8078 |
+
},
|
| 8079 |
+
{
|
| 8080 |
+
"epoch": 0.03103497174034884,
|
| 8081 |
+
"grad_norm": 0.03664977848529816,
|
| 8082 |
+
"learning_rate": 1.2750399292720283e-05,
|
| 8083 |
+
"loss": 11.9169,
|
| 8084 |
+
"step": 1149
|
| 8085 |
+
},
|
| 8086 |
+
{
|
| 8087 |
+
"epoch": 0.031061982159618074,
|
| 8088 |
+
"grad_norm": 0.02940729632973671,
|
| 8089 |
+
"learning_rate": 1.2637760935363053e-05,
|
| 8090 |
+
"loss": 11.9168,
|
| 8091 |
+
"step": 1150
|
| 8092 |
+
},
|
| 8093 |
+
{
|
| 8094 |
+
"epoch": 0.031088992578887305,
|
| 8095 |
+
"grad_norm": 0.032761119306087494,
|
| 8096 |
+
"learning_rate": 1.2525588749262163e-05,
|
| 8097 |
+
"loss": 11.9123,
|
| 8098 |
+
"step": 1151
|
| 8099 |
+
},
|
| 8100 |
+
{
|
| 8101 |
+
"epoch": 0.03111600299815654,
|
| 8102 |
+
"grad_norm": 0.028623409569263458,
|
| 8103 |
+
"learning_rate": 1.2413883332976573e-05,
|
| 8104 |
+
"loss": 11.9169,
|
| 8105 |
+
"step": 1152
|
| 8106 |
+
},
|
| 8107 |
+
{
|
| 8108 |
+
"epoch": 0.03114301341742577,
|
| 8109 |
+
"grad_norm": 0.021913375705480576,
|
| 8110 |
+
"learning_rate": 1.2302645282574465e-05,
|
| 8111 |
+
"loss": 11.9111,
|
| 8112 |
+
"step": 1153
|
| 8113 |
+
},
|
| 8114 |
+
{
|
| 8115 |
+
"epoch": 0.031170023836695007,
|
| 8116 |
+
"grad_norm": 0.03011070378124714,
|
| 8117 |
+
"learning_rate": 1.2191875191630209e-05,
|
| 8118 |
+
"loss": 11.9135,
|
| 8119 |
+
"step": 1154
|
| 8120 |
+
},
|
| 8121 |
+
{
|
| 8122 |
+
"epoch": 0.031197034255964238,
|
| 8123 |
+
"grad_norm": 0.02495531737804413,
|
| 8124 |
+
"learning_rate": 1.2081573651221035e-05,
|
| 8125 |
+
"loss": 11.9133,
|
| 8126 |
+
"step": 1155
|
| 8127 |
+
},
|
| 8128 |
+
{
|
| 8129 |
+
"epoch": 0.031224044675233473,
|
| 8130 |
+
"grad_norm": 0.03857101872563362,
|
| 8131 |
+
"learning_rate": 1.1971741249923984e-05,
|
| 8132 |
+
"loss": 11.9193,
|
| 8133 |
+
"step": 1156
|
| 8134 |
+
},
|
| 8135 |
+
{
|
| 8136 |
+
"epoch": 0.031251055094502704,
|
| 8137 |
+
"grad_norm": 0.036937810480594635,
|
| 8138 |
+
"learning_rate": 1.1862378573812716e-05,
|
| 8139 |
+
"loss": 11.9148,
|
| 8140 |
+
"step": 1157
|
| 8141 |
+
},
|
| 8142 |
+
{
|
| 8143 |
+
"epoch": 0.031278065513771935,
|
| 8144 |
+
"grad_norm": 0.03026803955435753,
|
| 8145 |
+
"learning_rate": 1.1753486206454434e-05,
|
| 8146 |
+
"loss": 11.9121,
|
| 8147 |
+
"step": 1158
|
| 8148 |
+
},
|
| 8149 |
+
{
|
| 8150 |
+
"epoch": 0.031305075933041174,
|
| 8151 |
+
"grad_norm": 0.05029657855629921,
|
| 8152 |
+
"learning_rate": 1.164506472890673e-05,
|
| 8153 |
+
"loss": 11.9099,
|
| 8154 |
+
"step": 1159
|
| 8155 |
+
},
|
| 8156 |
+
{
|
| 8157 |
+
"epoch": 0.031332086352310405,
|
| 8158 |
+
"grad_norm": 0.02796155959367752,
|
| 8159 |
+
"learning_rate": 1.1537114719714481e-05,
|
| 8160 |
+
"loss": 11.9079,
|
| 8161 |
+
"step": 1160
|
| 8162 |
+
},
|
| 8163 |
+
{
|
| 8164 |
+
"epoch": 0.031359096771579636,
|
| 8165 |
+
"grad_norm": 0.03133730962872505,
|
| 8166 |
+
"learning_rate": 1.1429636754906747e-05,
|
| 8167 |
+
"loss": 11.9188,
|
| 8168 |
+
"step": 1161
|
| 8169 |
+
},
|
| 8170 |
+
{
|
| 8171 |
+
"epoch": 0.03138610719084887,
|
| 8172 |
+
"grad_norm": 0.02825941890478134,
|
| 8173 |
+
"learning_rate": 1.1322631407993811e-05,
|
| 8174 |
+
"loss": 11.9119,
|
| 8175 |
+
"step": 1162
|
| 8176 |
+
},
|
| 8177 |
+
{
|
| 8178 |
+
"epoch": 0.031413117610118106,
|
| 8179 |
+
"grad_norm": 0.027236824855208397,
|
| 8180 |
+
"learning_rate": 1.1216099249963963e-05,
|
| 8181 |
+
"loss": 11.914,
|
| 8182 |
+
"step": 1163
|
| 8183 |
+
},
|
| 8184 |
+
{
|
| 8185 |
+
"epoch": 0.03144012802938734,
|
| 8186 |
+
"grad_norm": 0.0481816902756691,
|
| 8187 |
+
"learning_rate": 1.1110040849280534e-05,
|
| 8188 |
+
"loss": 11.9197,
|
| 8189 |
+
"step": 1164
|
| 8190 |
+
},
|
| 8191 |
+
{
|
| 8192 |
+
"epoch": 0.03146713844865657,
|
| 8193 |
+
"grad_norm": 0.03225933760404587,
|
| 8194 |
+
"learning_rate": 1.1004456771878835e-05,
|
| 8195 |
+
"loss": 11.9128,
|
| 8196 |
+
"step": 1165
|
| 8197 |
+
},
|
| 8198 |
+
{
|
| 8199 |
+
"epoch": 0.0314941488679258,
|
| 8200 |
+
"grad_norm": 0.04117142781615257,
|
| 8201 |
+
"learning_rate": 1.0899347581163221e-05,
|
| 8202 |
+
"loss": 11.9139,
|
| 8203 |
+
"step": 1166
|
| 8204 |
+
},
|
| 8205 |
+
{
|
| 8206 |
+
"epoch": 0.03152115928719504,
|
| 8207 |
+
"grad_norm": 0.029993971809744835,
|
| 8208 |
+
"learning_rate": 1.0794713838003944e-05,
|
| 8209 |
+
"loss": 11.9166,
|
| 8210 |
+
"step": 1167
|
| 8211 |
+
},
|
| 8212 |
+
{
|
| 8213 |
+
"epoch": 0.03154816970646427,
|
| 8214 |
+
"grad_norm": 0.025391219183802605,
|
| 8215 |
+
"learning_rate": 1.0690556100734283e-05,
|
| 8216 |
+
"loss": 11.9163,
|
| 8217 |
+
"step": 1168
|
| 8218 |
+
},
|
| 8219 |
+
{
|
| 8220 |
+
"epoch": 0.0315751801257335,
|
| 8221 |
+
"grad_norm": 0.03363854065537453,
|
| 8222 |
+
"learning_rate": 1.058687492514745e-05,
|
| 8223 |
+
"loss": 11.9194,
|
| 8224 |
+
"step": 1169
|
| 8225 |
+
},
|
| 8226 |
+
{
|
| 8227 |
+
"epoch": 0.03160219054500273,
|
| 8228 |
+
"grad_norm": 0.031209127977490425,
|
| 8229 |
+
"learning_rate": 1.0483670864493778e-05,
|
| 8230 |
+
"loss": 11.9158,
|
| 8231 |
+
"step": 1170
|
| 8232 |
+
},
|
| 8233 |
+
{
|
| 8234 |
+
"epoch": 0.03162920096427197,
|
| 8235 |
+
"grad_norm": 0.02980896271765232,
|
| 8236 |
+
"learning_rate": 1.0380944469477616e-05,
|
| 8237 |
+
"loss": 11.9153,
|
| 8238 |
+
"step": 1171
|
| 8239 |
+
},
|
| 8240 |
+
{
|
| 8241 |
+
"epoch": 0.0316562113835412,
|
| 8242 |
+
"grad_norm": 0.027297070249915123,
|
| 8243 |
+
"learning_rate": 1.0278696288254475e-05,
|
| 8244 |
+
"loss": 11.9157,
|
| 8245 |
+
"step": 1172
|
| 8246 |
+
},
|
| 8247 |
+
{
|
| 8248 |
+
"epoch": 0.031683221802810434,
|
| 8249 |
+
"grad_norm": 0.035955507308244705,
|
| 8250 |
+
"learning_rate": 1.017692686642806e-05,
|
| 8251 |
+
"loss": 11.9219,
|
| 8252 |
+
"step": 1173
|
| 8253 |
+
},
|
| 8254 |
+
{
|
| 8255 |
+
"epoch": 0.031710232222079665,
|
| 8256 |
+
"grad_norm": 0.03889148682355881,
|
| 8257 |
+
"learning_rate": 1.0075636747047446e-05,
|
| 8258 |
+
"loss": 11.9155,
|
| 8259 |
+
"step": 1174
|
| 8260 |
+
},
|
| 8261 |
+
{
|
| 8262 |
+
"epoch": 0.0317372426413489,
|
| 8263 |
+
"grad_norm": 0.048469457775354385,
|
| 8264 |
+
"learning_rate": 9.974826470604049e-06,
|
| 8265 |
+
"loss": 11.9167,
|
| 8266 |
+
"step": 1175
|
| 8267 |
+
},
|
| 8268 |
+
{
|
| 8269 |
+
"epoch": 0.031764253060618135,
|
| 8270 |
+
"grad_norm": 0.02912323549389839,
|
| 8271 |
+
"learning_rate": 9.874496575028814e-06,
|
| 8272 |
+
"loss": 11.9148,
|
| 8273 |
+
"step": 1176
|
| 8274 |
+
},
|
| 8275 |
+
{
|
| 8276 |
+
"epoch": 0.031791263479887366,
|
| 8277 |
+
"grad_norm": 0.03220577538013458,
|
| 8278 |
+
"learning_rate": 9.774647595689357e-06,
|
| 8279 |
+
"loss": 11.9114,
|
| 8280 |
+
"step": 1177
|
| 8281 |
+
},
|
| 8282 |
+
{
|
| 8283 |
+
"epoch": 0.0318182738991566,
|
| 8284 |
+
"grad_norm": 0.039429958909749985,
|
| 8285 |
+
"learning_rate": 9.675280065387116e-06,
|
| 8286 |
+
"loss": 11.9146,
|
| 8287 |
+
"step": 1178
|
| 8288 |
+
},
|
| 8289 |
+
{
|
| 8290 |
+
"epoch": 0.031845284318425836,
|
| 8291 |
+
"grad_norm": 0.02698604390025139,
|
| 8292 |
+
"learning_rate": 9.576394514354425e-06,
|
| 8293 |
+
"loss": 11.9155,
|
| 8294 |
+
"step": 1179
|
| 8295 |
+
},
|
| 8296 |
+
{
|
| 8297 |
+
"epoch": 0.03187229473769507,
|
| 8298 |
+
"grad_norm": 0.04509074240922928,
|
| 8299 |
+
"learning_rate": 9.477991470251791e-06,
|
| 8300 |
+
"loss": 11.9149,
|
| 8301 |
+
"step": 1180
|
| 8302 |
+
},
|
| 8303 |
+
{
|
| 8304 |
+
"epoch": 0.0318993051569643,
|
| 8305 |
+
"grad_norm": 0.03135255351662636,
|
| 8306 |
+
"learning_rate": 9.380071458165008e-06,
|
| 8307 |
+
"loss": 11.9121,
|
| 8308 |
+
"step": 1181
|
| 8309 |
+
},
|
| 8310 |
+
{
|
| 8311 |
+
"epoch": 0.03192631557623353,
|
| 8312 |
+
"grad_norm": 0.041827112436294556,
|
| 8313 |
+
"learning_rate": 9.282635000602346e-06,
|
| 8314 |
+
"loss": 11.9127,
|
| 8315 |
+
"step": 1182
|
| 8316 |
+
},
|
| 8317 |
+
{
|
| 8318 |
+
"epoch": 0.03195332599550277,
|
| 8319 |
+
"grad_norm": 0.03321888670325279,
|
| 8320 |
+
"learning_rate": 9.185682617491863e-06,
|
| 8321 |
+
"loss": 11.9151,
|
| 8322 |
+
"step": 1183
|
| 8323 |
+
},
|
| 8324 |
+
{
|
| 8325 |
+
"epoch": 0.031980336414772,
|
| 8326 |
+
"grad_norm": 0.027054108679294586,
|
| 8327 |
+
"learning_rate": 9.089214826178504e-06,
|
| 8328 |
+
"loss": 11.9158,
|
| 8329 |
+
"step": 1184
|
| 8330 |
+
},
|
| 8331 |
+
{
|
| 8332 |
+
"epoch": 0.03200734683404123,
|
| 8333 |
+
"grad_norm": 0.04247848317027092,
|
| 8334 |
+
"learning_rate": 8.993232141421415e-06,
|
| 8335 |
+
"loss": 11.9157,
|
| 8336 |
+
"step": 1185
|
| 8337 |
+
},
|
| 8338 |
+
{
|
| 8339 |
+
"epoch": 0.03203435725331046,
|
| 8340 |
+
"grad_norm": 0.026086729019880295,
|
| 8341 |
+
"learning_rate": 8.897735075391155e-06,
|
| 8342 |
+
"loss": 11.9173,
|
| 8343 |
+
"step": 1186
|
| 8344 |
+
},
|
| 8345 |
+
{
|
| 8346 |
+
"epoch": 0.0320613676725797,
|
| 8347 |
+
"grad_norm": 0.025407347828149796,
|
| 8348 |
+
"learning_rate": 8.80272413766705e-06,
|
| 8349 |
+
"loss": 11.9129,
|
| 8350 |
+
"step": 1187
|
| 8351 |
+
},
|
| 8352 |
+
{
|
| 8353 |
+
"epoch": 0.03208837809184893,
|
| 8354 |
+
"grad_norm": 0.04205694794654846,
|
| 8355 |
+
"learning_rate": 8.708199835234344e-06,
|
| 8356 |
+
"loss": 11.9094,
|
| 8357 |
+
"step": 1188
|
| 8358 |
+
},
|
| 8359 |
+
{
|
| 8360 |
+
"epoch": 0.03211538851111816,
|
| 8361 |
+
"grad_norm": 0.03572431951761246,
|
| 8362 |
+
"learning_rate": 8.614162672481585e-06,
|
| 8363 |
+
"loss": 11.9159,
|
| 8364 |
+
"step": 1189
|
| 8365 |
+
},
|
| 8366 |
+
{
|
| 8367 |
+
"epoch": 0.032142398930387395,
|
| 8368 |
+
"grad_norm": 0.026628505438566208,
|
| 8369 |
+
"learning_rate": 8.520613151197898e-06,
|
| 8370 |
+
"loss": 11.9148,
|
| 8371 |
+
"step": 1190
|
| 8372 |
+
},
|
| 8373 |
+
{
|
| 8374 |
+
"epoch": 0.03216940934965663,
|
| 8375 |
+
"grad_norm": 0.016602449119091034,
|
| 8376 |
+
"learning_rate": 8.427551770570353e-06,
|
| 8377 |
+
"loss": 11.9139,
|
| 8378 |
+
"step": 1191
|
| 8379 |
+
},
|
| 8380 |
+
{
|
| 8381 |
+
"epoch": 0.032196419768925864,
|
| 8382 |
+
"grad_norm": 0.024417880922555923,
|
| 8383 |
+
"learning_rate": 8.334979027181223e-06,
|
| 8384 |
+
"loss": 11.9109,
|
| 8385 |
+
"step": 1192
|
| 8386 |
+
},
|
| 8387 |
+
{
|
| 8388 |
+
"epoch": 0.032223430188195096,
|
| 8389 |
+
"grad_norm": 0.02814560942351818,
|
| 8390 |
+
"learning_rate": 8.242895415005392e-06,
|
| 8391 |
+
"loss": 11.9093,
|
| 8392 |
+
"step": 1193
|
| 8393 |
+
},
|
| 8394 |
+
{
|
| 8395 |
+
"epoch": 0.03225044060746433,
|
| 8396 |
+
"grad_norm": 0.03642592579126358,
|
| 8397 |
+
"learning_rate": 8.151301425407699e-06,
|
| 8398 |
+
"loss": 11.9159,
|
| 8399 |
+
"step": 1194
|
| 8400 |
+
},
|
| 8401 |
+
{
|
| 8402 |
+
"epoch": 0.032277451026733565,
|
| 8403 |
+
"grad_norm": 0.0355217419564724,
|
| 8404 |
+
"learning_rate": 8.060197547140347e-06,
|
| 8405 |
+
"loss": 11.9137,
|
| 8406 |
+
"step": 1195
|
| 8407 |
+
},
|
| 8408 |
+
{
|
| 8409 |
+
"epoch": 0.0323044614460028,
|
| 8410 |
+
"grad_norm": 0.04367974027991295,
|
| 8411 |
+
"learning_rate": 7.969584266340258e-06,
|
| 8412 |
+
"loss": 11.9159,
|
| 8413 |
+
"step": 1196
|
| 8414 |
+
},
|
| 8415 |
+
{
|
| 8416 |
+
"epoch": 0.03233147186527203,
|
| 8417 |
+
"grad_norm": 0.03913813829421997,
|
| 8418 |
+
"learning_rate": 7.879462066526455e-06,
|
| 8419 |
+
"loss": 11.916,
|
| 8420 |
+
"step": 1197
|
| 8421 |
+
},
|
| 8422 |
+
{
|
| 8423 |
+
"epoch": 0.03235848228454126,
|
| 8424 |
+
"grad_norm": 0.04448280110955238,
|
| 8425 |
+
"learning_rate": 7.789831428597549e-06,
|
| 8426 |
+
"loss": 11.9118,
|
| 8427 |
+
"step": 1198
|
| 8428 |
+
},
|
| 8429 |
+
{
|
| 8430 |
+
"epoch": 0.0323854927038105,
|
| 8431 |
+
"grad_norm": 0.03358766809105873,
|
| 8432 |
+
"learning_rate": 7.70069283082917e-06,
|
| 8433 |
+
"loss": 11.9152,
|
| 8434 |
+
"step": 1199
|
| 8435 |
+
},
|
| 8436 |
+
{
|
| 8437 |
+
"epoch": 0.03241250312307973,
|
| 8438 |
+
"grad_norm": 0.030266188085079193,
|
| 8439 |
+
"learning_rate": 7.612046748871327e-06,
|
| 8440 |
+
"loss": 11.916,
|
| 8441 |
+
"step": 1200
|
| 8442 |
+
},
|
| 8443 |
+
{
|
| 8444 |
+
"epoch": 0.03243951354234896,
|
| 8445 |
+
"grad_norm": 0.02374487929046154,
|
| 8446 |
+
"learning_rate": 7.523893655745961e-06,
|
| 8447 |
+
"loss": 11.9186,
|
| 8448 |
+
"step": 1201
|
| 8449 |
+
},
|
| 8450 |
+
{
|
| 8451 |
+
"epoch": 0.03246652396161819,
|
| 8452 |
+
"grad_norm": 0.03596069663763046,
|
| 8453 |
+
"learning_rate": 7.43623402184438e-06,
|
| 8454 |
+
"loss": 11.9123,
|
| 8455 |
+
"step": 1202
|
| 8456 |
+
},
|
| 8457 |
+
{
|
| 8458 |
+
"epoch": 0.03249353438088743,
|
| 8459 |
+
"grad_norm": 0.039939962327480316,
|
| 8460 |
+
"learning_rate": 7.349068314924756e-06,
|
| 8461 |
+
"loss": 11.9128,
|
| 8462 |
+
"step": 1203
|
| 8463 |
+
},
|
| 8464 |
+
{
|
| 8465 |
+
"epoch": 0.03252054480015666,
|
| 8466 |
+
"grad_norm": 0.03782733157277107,
|
| 8467 |
+
"learning_rate": 7.2623970001096444e-06,
|
| 8468 |
+
"loss": 11.915,
|
| 8469 |
+
"step": 1204
|
| 8470 |
+
},
|
| 8471 |
+
{
|
| 8472 |
+
"epoch": 0.03254755521942589,
|
| 8473 |
+
"grad_norm": 0.023838946595788002,
|
| 8474 |
+
"learning_rate": 7.1762205398834936e-06,
|
| 8475 |
+
"loss": 11.9141,
|
| 8476 |
+
"step": 1205
|
| 8477 |
+
},
|
| 8478 |
+
{
|
| 8479 |
+
"epoch": 0.032574565638695124,
|
| 8480 |
+
"grad_norm": 0.04269898682832718,
|
| 8481 |
+
"learning_rate": 7.090539394090134e-06,
|
| 8482 |
+
"loss": 11.919,
|
| 8483 |
+
"step": 1206
|
| 8484 |
+
},
|
| 8485 |
+
{
|
| 8486 |
+
"epoch": 0.03260157605796436,
|
| 8487 |
+
"grad_norm": 0.027135029435157776,
|
| 8488 |
+
"learning_rate": 7.005354019930377e-06,
|
| 8489 |
+
"loss": 11.9166,
|
| 8490 |
+
"step": 1207
|
| 8491 |
+
},
|
| 8492 |
+
{
|
| 8493 |
+
"epoch": 0.032628586477233594,
|
| 8494 |
+
"grad_norm": 0.035581089556217194,
|
| 8495 |
+
"learning_rate": 6.9206648719596034e-06,
|
| 8496 |
+
"loss": 11.9127,
|
| 8497 |
+
"step": 1208
|
| 8498 |
+
},
|
| 8499 |
+
{
|
| 8500 |
+
"epoch": 0.032655596896502825,
|
| 8501 |
+
"grad_norm": 0.03042713738977909,
|
| 8502 |
+
"learning_rate": 6.836472402085236e-06,
|
| 8503 |
+
"loss": 11.9149,
|
| 8504 |
+
"step": 1209
|
| 8505 |
+
},
|
| 8506 |
+
{
|
| 8507 |
+
"epoch": 0.03268260731577206,
|
| 8508 |
+
"grad_norm": 0.035642750561237335,
|
| 8509 |
+
"learning_rate": 6.75277705956443e-06,
|
| 8510 |
+
"loss": 11.9135,
|
| 8511 |
+
"step": 1210
|
| 8512 |
+
},
|
| 8513 |
+
{
|
| 8514 |
+
"epoch": 0.032709617735041295,
|
| 8515 |
+
"grad_norm": 0.03757479786872864,
|
| 8516 |
+
"learning_rate": 6.669579291001593e-06,
|
| 8517 |
+
"loss": 11.9127,
|
| 8518 |
+
"step": 1211
|
| 8519 |
+
},
|
| 8520 |
+
{
|
| 8521 |
+
"epoch": 0.032736628154310526,
|
| 8522 |
+
"grad_norm": 0.026762204244732857,
|
| 8523 |
+
"learning_rate": 6.586879540346091e-06,
|
| 8524 |
+
"loss": 11.9134,
|
| 8525 |
+
"step": 1212
|
| 8526 |
+
},
|
| 8527 |
+
{
|
| 8528 |
+
"epoch": 0.03276363857357976,
|
| 8529 |
+
"grad_norm": 0.03260150924324989,
|
| 8530 |
+
"learning_rate": 6.504678248889784e-06,
|
| 8531 |
+
"loss": 11.9117,
|
| 8532 |
+
"step": 1213
|
| 8533 |
+
},
|
| 8534 |
+
{
|
| 8535 |
+
"epoch": 0.03279064899284899,
|
| 8536 |
+
"grad_norm": 0.04702017083764076,
|
| 8537 |
+
"learning_rate": 6.422975855264757e-06,
|
| 8538 |
+
"loss": 11.9161,
|
| 8539 |
+
"step": 1214
|
| 8540 |
+
},
|
| 8541 |
+
{
|
| 8542 |
+
"epoch": 0.03281765941211823,
|
| 8543 |
+
"grad_norm": 0.04338684678077698,
|
| 8544 |
+
"learning_rate": 6.34177279544087e-06,
|
| 8545 |
+
"loss": 11.916,
|
| 8546 |
+
"step": 1215
|
| 8547 |
+
},
|
| 8548 |
+
{
|
| 8549 |
+
"epoch": 0.03284466983138746,
|
| 8550 |
+
"grad_norm": 0.04819529503583908,
|
| 8551 |
+
"learning_rate": 6.261069502723616e-06,
|
| 8552 |
+
"loss": 11.9085,
|
| 8553 |
+
"step": 1216
|
| 8554 |
+
},
|
| 8555 |
+
{
|
| 8556 |
+
"epoch": 0.03287168025065669,
|
| 8557 |
+
"grad_norm": 0.03627943620085716,
|
| 8558 |
+
"learning_rate": 6.180866407751595e-06,
|
| 8559 |
+
"loss": 11.9158,
|
| 8560 |
+
"step": 1217
|
| 8561 |
+
},
|
| 8562 |
+
{
|
| 8563 |
+
"epoch": 0.03289869066992592,
|
| 8564 |
+
"grad_norm": 0.02766910195350647,
|
| 8565 |
+
"learning_rate": 6.1011639384943585e-06,
|
| 8566 |
+
"loss": 11.9122,
|
| 8567 |
+
"step": 1218
|
| 8568 |
+
},
|
| 8569 |
+
{
|
| 8570 |
+
"epoch": 0.03292570108919516,
|
| 8571 |
+
"grad_norm": 0.02208135463297367,
|
| 8572 |
+
"learning_rate": 6.021962520250057e-06,
|
| 8573 |
+
"loss": 11.9129,
|
| 8574 |
+
"step": 1219
|
| 8575 |
+
},
|
| 8576 |
+
{
|
| 8577 |
+
"epoch": 0.03295271150846439,
|
| 8578 |
+
"grad_norm": 0.02409740723669529,
|
| 8579 |
+
"learning_rate": 5.943262575643238e-06,
|
| 8580 |
+
"loss": 11.9122,
|
| 8581 |
+
"step": 1220
|
| 8582 |
+
},
|
| 8583 |
+
{
|
| 8584 |
+
"epoch": 0.03297972192773362,
|
| 8585 |
+
"grad_norm": 0.024469245225191116,
|
| 8586 |
+
"learning_rate": 5.865064524622521e-06,
|
| 8587 |
+
"loss": 11.9149,
|
| 8588 |
+
"step": 1221
|
| 8589 |
+
},
|
| 8590 |
+
{
|
| 8591 |
+
"epoch": 0.033006732347002854,
|
| 8592 |
+
"grad_norm": 0.04426274076104164,
|
| 8593 |
+
"learning_rate": 5.7873687844583695e-06,
|
| 8594 |
+
"loss": 11.9125,
|
| 8595 |
+
"step": 1222
|
| 8596 |
+
},
|
| 8597 |
+
{
|
| 8598 |
+
"epoch": 0.03303374276627209,
|
| 8599 |
+
"grad_norm": 0.025476930662989616,
|
| 8600 |
+
"learning_rate": 5.710175769740933e-06,
|
| 8601 |
+
"loss": 11.9158,
|
| 8602 |
+
"step": 1223
|
| 8603 |
+
},
|
| 8604 |
+
{
|
| 8605 |
+
"epoch": 0.033060753185541324,
|
| 8606 |
+
"grad_norm": 0.037165865302085876,
|
| 8607 |
+
"learning_rate": 5.633485892377699e-06,
|
| 8608 |
+
"loss": 11.9168,
|
| 8609 |
+
"step": 1224
|
| 8610 |
+
},
|
| 8611 |
+
{
|
| 8612 |
+
"epoch": 0.033087763604810555,
|
| 8613 |
+
"grad_norm": 0.043615397065877914,
|
| 8614 |
+
"learning_rate": 5.557299561591478e-06,
|
| 8615 |
+
"loss": 11.9173,
|
| 8616 |
+
"step": 1225
|
| 8617 |
+
},
|
| 8618 |
+
{
|
| 8619 |
+
"epoch": 0.033114774024079786,
|
| 8620 |
+
"grad_norm": 0.029055394232273102,
|
| 8621 |
+
"learning_rate": 5.481617183918053e-06,
|
| 8622 |
+
"loss": 11.9177,
|
| 8623 |
+
"step": 1226
|
| 8624 |
+
},
|
| 8625 |
+
{
|
| 8626 |
+
"epoch": 0.033141784443349025,
|
| 8627 |
+
"grad_norm": 0.026653952896595,
|
| 8628 |
+
"learning_rate": 5.406439163204069e-06,
|
| 8629 |
+
"loss": 11.9149,
|
| 8630 |
+
"step": 1227
|
| 8631 |
+
},
|
| 8632 |
+
{
|
| 8633 |
+
"epoch": 0.033168794862618256,
|
| 8634 |
+
"grad_norm": 0.06095029041171074,
|
| 8635 |
+
"learning_rate": 5.331765900604913e-06,
|
| 8636 |
+
"loss": 11.9175,
|
| 8637 |
+
"step": 1228
|
| 8638 |
+
},
|
| 8639 |
+
{
|
| 8640 |
+
"epoch": 0.03319580528188749,
|
| 8641 |
+
"grad_norm": 0.02985543943941593,
|
| 8642 |
+
"learning_rate": 5.2575977945825695e-06,
|
| 8643 |
+
"loss": 11.9084,
|
| 8644 |
+
"step": 1229
|
| 8645 |
+
},
|
| 8646 |
+
{
|
| 8647 |
+
"epoch": 0.03322281570115672,
|
| 8648 |
+
"grad_norm": 0.035458605736494064,
|
| 8649 |
+
"learning_rate": 5.183935240903414e-06,
|
| 8650 |
+
"loss": 11.9126,
|
| 8651 |
+
"step": 1230
|
| 8652 |
+
},
|
| 8653 |
+
{
|
| 8654 |
+
"epoch": 0.03324982612042596,
|
| 8655 |
+
"grad_norm": 0.04084185138344765,
|
| 8656 |
+
"learning_rate": 5.110778632636204e-06,
|
| 8657 |
+
"loss": 11.9117,
|
| 8658 |
+
"step": 1231
|
| 8659 |
+
},
|
| 8660 |
+
{
|
| 8661 |
+
"epoch": 0.03327683653969519,
|
| 8662 |
+
"grad_norm": 0.0327073372900486,
|
| 8663 |
+
"learning_rate": 5.0381283601498855e-06,
|
| 8664 |
+
"loss": 11.9109,
|
| 8665 |
+
"step": 1232
|
| 8666 |
+
},
|
| 8667 |
+
{
|
| 8668 |
+
"epoch": 0.03330384695896442,
|
| 8669 |
+
"grad_norm": 0.03445550426840782,
|
| 8670 |
+
"learning_rate": 4.965984811111634e-06,
|
| 8671 |
+
"loss": 11.9216,
|
| 8672 |
+
"step": 1233
|
| 8673 |
+
},
|
| 8674 |
+
{
|
| 8675 |
+
"epoch": 0.03333085737823365,
|
| 8676 |
+
"grad_norm": 0.03623393177986145,
|
| 8677 |
+
"learning_rate": 4.8943483704846475e-06,
|
| 8678 |
+
"loss": 11.9113,
|
| 8679 |
+
"step": 1234
|
| 8680 |
+
},
|
| 8681 |
+
{
|
| 8682 |
+
"epoch": 0.03335786779750289,
|
| 8683 |
+
"grad_norm": 0.03152257949113846,
|
| 8684 |
+
"learning_rate": 4.8232194205261815e-06,
|
| 8685 |
+
"loss": 11.9186,
|
| 8686 |
+
"step": 1235
|
| 8687 |
+
},
|
| 8688 |
+
{
|
| 8689 |
+
"epoch": 0.03338487821677212,
|
| 8690 |
+
"grad_norm": 0.024906009435653687,
|
| 8691 |
+
"learning_rate": 4.752598340785474e-06,
|
| 8692 |
+
"loss": 11.9158,
|
| 8693 |
+
"step": 1236
|
| 8694 |
+
},
|
| 8695 |
+
{
|
| 8696 |
+
"epoch": 0.03341188863604135,
|
| 8697 |
+
"grad_norm": 0.037356190383434296,
|
| 8698 |
+
"learning_rate": 4.682485508101752e-06,
|
| 8699 |
+
"loss": 11.9145,
|
| 8700 |
+
"step": 1237
|
| 8701 |
+
},
|
| 8702 |
+
{
|
| 8703 |
+
"epoch": 0.033438899055310584,
|
| 8704 |
+
"grad_norm": 0.030487028881907463,
|
| 8705 |
+
"learning_rate": 4.61288129660219e-06,
|
| 8706 |
+
"loss": 11.9116,
|
| 8707 |
+
"step": 1238
|
| 8708 |
+
},
|
| 8709 |
+
{
|
| 8710 |
+
"epoch": 0.03346590947457982,
|
| 8711 |
+
"grad_norm": 0.035832881927490234,
|
| 8712 |
+
"learning_rate": 4.543786077699907e-06,
|
| 8713 |
+
"loss": 11.913,
|
| 8714 |
+
"step": 1239
|
| 8715 |
+
},
|
| 8716 |
+
{
|
| 8717 |
+
"epoch": 0.03349291989384905,
|
| 8718 |
+
"grad_norm": 0.028598813340067863,
|
| 8719 |
+
"learning_rate": 4.475200220092002e-06,
|
| 8720 |
+
"loss": 11.9135,
|
| 8721 |
+
"step": 1240
|
| 8722 |
+
},
|
| 8723 |
+
{
|
| 8724 |
+
"epoch": 0.033519930313118285,
|
| 8725 |
+
"grad_norm": 0.03946138173341751,
|
| 8726 |
+
"learning_rate": 4.4071240897576195e-06,
|
| 8727 |
+
"loss": 11.9129,
|
| 8728 |
+
"step": 1241
|
| 8729 |
+
},
|
| 8730 |
+
{
|
| 8731 |
+
"epoch": 0.033546940732387516,
|
| 8732 |
+
"grad_norm": 0.037372782826423645,
|
| 8733 |
+
"learning_rate": 4.339558049955927e-06,
|
| 8734 |
+
"loss": 11.9156,
|
| 8735 |
+
"step": 1242
|
| 8736 |
+
},
|
| 8737 |
+
{
|
| 8738 |
+
"epoch": 0.033573951151656754,
|
| 8739 |
+
"grad_norm": 0.029304128140211105,
|
| 8740 |
+
"learning_rate": 4.272502461224226e-06,
|
| 8741 |
+
"loss": 11.9152,
|
| 8742 |
+
"step": 1243
|
| 8743 |
+
},
|
| 8744 |
+
{
|
| 8745 |
+
"epoch": 0.033600961570925986,
|
| 8746 |
+
"grad_norm": 0.042041193693876266,
|
| 8747 |
+
"learning_rate": 4.205957681375994e-06,
|
| 8748 |
+
"loss": 11.9085,
|
| 8749 |
+
"step": 1244
|
| 8750 |
+
},
|
| 8751 |
+
{
|
| 8752 |
+
"epoch": 0.03362797199019522,
|
| 8753 |
+
"grad_norm": 0.027062848210334778,
|
| 8754 |
+
"learning_rate": 4.139924065499035e-06,
|
| 8755 |
+
"loss": 11.9168,
|
| 8756 |
+
"step": 1245
|
| 8757 |
+
},
|
| 8758 |
+
{
|
| 8759 |
+
"epoch": 0.03365498240946445,
|
| 8760 |
+
"grad_norm": 0.03303879499435425,
|
| 8761 |
+
"learning_rate": 4.0744019659535115e-06,
|
| 8762 |
+
"loss": 11.9175,
|
| 8763 |
+
"step": 1246
|
| 8764 |
+
},
|
| 8765 |
+
{
|
| 8766 |
+
"epoch": 0.03368199282873369,
|
| 8767 |
+
"grad_norm": 0.041767872869968414,
|
| 8768 |
+
"learning_rate": 4.0093917323701155e-06,
|
| 8769 |
+
"loss": 11.9143,
|
| 8770 |
+
"step": 1247
|
| 8771 |
+
},
|
| 8772 |
+
{
|
| 8773 |
+
"epoch": 0.03370900324800292,
|
| 8774 |
+
"grad_norm": 0.03279326483607292,
|
| 8775 |
+
"learning_rate": 3.944893711648167e-06,
|
| 8776 |
+
"loss": 11.9166,
|
| 8777 |
+
"step": 1248
|
| 8778 |
+
},
|
| 8779 |
+
{
|
| 8780 |
+
"epoch": 0.03373601366727215,
|
| 8781 |
+
"grad_norm": 0.04130978509783745,
|
| 8782 |
+
"learning_rate": 3.880908247953796e-06,
|
| 8783 |
+
"loss": 11.919,
|
| 8784 |
+
"step": 1249
|
| 8785 |
+
},
|
| 8786 |
+
{
|
| 8787 |
+
"epoch": 0.03376302408654138,
|
| 8788 |
+
"grad_norm": 0.03301661089062691,
|
| 8789 |
+
"learning_rate": 3.817435682718096e-06,
|
| 8790 |
+
"loss": 11.916,
|
| 8791 |
+
"step": 1250
|
| 8792 |
+
},
|
| 8793 |
+
{
|
| 8794 |
+
"epoch": 0.03379003450581062,
|
| 8795 |
+
"grad_norm": 0.036643002182245255,
|
| 8796 |
+
"learning_rate": 3.7544763546352834e-06,
|
| 8797 |
+
"loss": 11.9127,
|
| 8798 |
+
"step": 1251
|
| 8799 |
+
},
|
| 8800 |
+
{
|
| 8801 |
+
"epoch": 0.03381704492507985,
|
| 8802 |
+
"grad_norm": 0.0306033156812191,
|
| 8803 |
+
"learning_rate": 3.6920305996608785e-06,
|
| 8804 |
+
"loss": 11.916,
|
| 8805 |
+
"step": 1252
|
| 8806 |
+
},
|
| 8807 |
+
{
|
| 8808 |
+
"epoch": 0.03384405534434908,
|
| 8809 |
+
"grad_norm": 0.019081279635429382,
|
| 8810 |
+
"learning_rate": 3.6300987510100136e-06,
|
| 8811 |
+
"loss": 11.9135,
|
| 8812 |
+
"step": 1253
|
| 8813 |
+
},
|
| 8814 |
+
{
|
| 8815 |
+
"epoch": 0.03387106576361831,
|
| 8816 |
+
"grad_norm": 0.03544707968831062,
|
| 8817 |
+
"learning_rate": 3.568681139155516e-06,
|
| 8818 |
+
"loss": 11.914,
|
| 8819 |
+
"step": 1254
|
| 8820 |
+
},
|
| 8821 |
+
{
|
| 8822 |
+
"epoch": 0.03389807618288755,
|
| 8823 |
+
"grad_norm": 0.04367835447192192,
|
| 8824 |
+
"learning_rate": 3.507778091826219e-06,
|
| 8825 |
+
"loss": 11.9149,
|
| 8826 |
+
"step": 1255
|
| 8827 |
+
},
|
| 8828 |
+
{
|
| 8829 |
+
"epoch": 0.03392508660215678,
|
| 8830 |
+
"grad_norm": 0.03611910715699196,
|
| 8831 |
+
"learning_rate": 3.4473899340052075e-06,
|
| 8832 |
+
"loss": 11.9093,
|
| 8833 |
+
"step": 1256
|
| 8834 |
+
},
|
| 8835 |
+
{
|
| 8836 |
+
"epoch": 0.033952097021426014,
|
| 8837 |
+
"grad_norm": 0.030203288421034813,
|
| 8838 |
+
"learning_rate": 3.3875169879280966e-06,
|
| 8839 |
+
"loss": 11.9111,
|
| 8840 |
+
"step": 1257
|
| 8841 |
+
},
|
| 8842 |
+
{
|
| 8843 |
+
"epoch": 0.033979107440695246,
|
| 8844 |
+
"grad_norm": 0.04575571045279503,
|
| 8845 |
+
"learning_rate": 3.3281595730812575e-06,
|
| 8846 |
+
"loss": 11.9104,
|
| 8847 |
+
"step": 1258
|
| 8848 |
+
},
|
| 8849 |
+
{
|
| 8850 |
+
"epoch": 0.034006117859964484,
|
| 8851 |
+
"grad_norm": 0.03389758616685867,
|
| 8852 |
+
"learning_rate": 3.269318006200195e-06,
|
| 8853 |
+
"loss": 11.9157,
|
| 8854 |
+
"step": 1259
|
| 8855 |
+
},
|
| 8856 |
+
{
|
| 8857 |
+
"epoch": 0.034033128279233715,
|
| 8858 |
+
"grad_norm": 0.02616419829428196,
|
| 8859 |
+
"learning_rate": 3.2109926012677484e-06,
|
| 8860 |
+
"loss": 11.9161,
|
| 8861 |
+
"step": 1260
|
| 8862 |
+
},
|
| 8863 |
+
{
|
| 8864 |
+
"epoch": 0.03406013869850295,
|
| 8865 |
+
"grad_norm": 0.04158425331115723,
|
| 8866 |
+
"learning_rate": 3.1531836695125495e-06,
|
| 8867 |
+
"loss": 11.9192,
|
| 8868 |
+
"step": 1261
|
| 8869 |
+
},
|
| 8870 |
+
{
|
| 8871 |
+
"epoch": 0.03408714911777218,
|
| 8872 |
+
"grad_norm": 0.048009276390075684,
|
| 8873 |
+
"learning_rate": 3.095891519407246e-06,
|
| 8874 |
+
"loss": 11.9204,
|
| 8875 |
+
"step": 1262
|
| 8876 |
+
},
|
| 8877 |
+
{
|
| 8878 |
+
"epoch": 0.034114159537041416,
|
| 8879 |
+
"grad_norm": 0.027957575395703316,
|
| 8880 |
+
"learning_rate": 3.039116456666924e-06,
|
| 8881 |
+
"loss": 11.9122,
|
| 8882 |
+
"step": 1263
|
| 8883 |
+
},
|
| 8884 |
+
{
|
| 8885 |
+
"epoch": 0.03414116995631065,
|
| 8886 |
+
"grad_norm": 0.03297886624932289,
|
| 8887 |
+
"learning_rate": 2.982858784247422e-06,
|
| 8888 |
+
"loss": 11.914,
|
| 8889 |
+
"step": 1264
|
| 8890 |
+
},
|
| 8891 |
+
{
|
| 8892 |
+
"epoch": 0.03416818037557988,
|
| 8893 |
+
"grad_norm": 0.028546515852212906,
|
| 8894 |
+
"learning_rate": 2.9271188023437867e-06,
|
| 8895 |
+
"loss": 11.9121,
|
| 8896 |
+
"step": 1265
|
| 8897 |
+
},
|
| 8898 |
+
{
|
| 8899 |
+
"epoch": 0.03419519079484911,
|
| 8900 |
+
"grad_norm": 0.03843577578663826,
|
| 8901 |
+
"learning_rate": 2.8718968083886075e-06,
|
| 8902 |
+
"loss": 11.9144,
|
| 8903 |
+
"step": 1266
|
| 8904 |
+
},
|
| 8905 |
+
{
|
| 8906 |
+
"epoch": 0.03422220121411835,
|
| 8907 |
+
"grad_norm": 0.04053690657019615,
|
| 8908 |
+
"learning_rate": 2.8171930970504744e-06,
|
| 8909 |
+
"loss": 11.9118,
|
| 8910 |
+
"step": 1267
|
| 8911 |
+
},
|
| 8912 |
+
{
|
| 8913 |
+
"epoch": 0.03424921163338758,
|
| 8914 |
+
"grad_norm": 0.03197146952152252,
|
| 8915 |
+
"learning_rate": 2.7630079602323442e-06,
|
| 8916 |
+
"loss": 11.9128,
|
| 8917 |
+
"step": 1268
|
| 8918 |
+
},
|
| 8919 |
+
{
|
| 8920 |
+
"epoch": 0.03427622205265681,
|
| 8921 |
+
"grad_norm": 0.03896786645054817,
|
| 8922 |
+
"learning_rate": 2.709341687070044e-06,
|
| 8923 |
+
"loss": 11.9158,
|
| 8924 |
+
"step": 1269
|
| 8925 |
+
},
|
| 8926 |
+
{
|
| 8927 |
+
"epoch": 0.03430323247192604,
|
| 8928 |
+
"grad_norm": 0.018603939563035965,
|
| 8929 |
+
"learning_rate": 2.656194563930714e-06,
|
| 8930 |
+
"loss": 11.9154,
|
| 8931 |
+
"step": 1270
|
| 8932 |
+
},
|
| 8933 |
+
{
|
| 8934 |
+
"epoch": 0.03433024289119528,
|
| 8935 |
+
"grad_norm": 0.036858513951301575,
|
| 8936 |
+
"learning_rate": 2.6035668744112784e-06,
|
| 8937 |
+
"loss": 11.9161,
|
| 8938 |
+
"step": 1271
|
| 8939 |
+
},
|
| 8940 |
+
{
|
| 8941 |
+
"epoch": 0.03435725331046451,
|
| 8942 |
+
"grad_norm": 0.0390339121222496,
|
| 8943 |
+
"learning_rate": 2.5514588993368893e-06,
|
| 8944 |
+
"loss": 11.9118,
|
| 8945 |
+
"step": 1272
|
| 8946 |
+
},
|
| 8947 |
+
{
|
| 8948 |
+
"epoch": 0.034384263729733744,
|
| 8949 |
+
"grad_norm": 0.03321358561515808,
|
| 8950 |
+
"learning_rate": 2.4998709167594946e-06,
|
| 8951 |
+
"loss": 11.913,
|
| 8952 |
+
"step": 1273
|
| 8953 |
+
},
|
| 8954 |
+
{
|
| 8955 |
+
"epoch": 0.034411274149002975,
|
| 8956 |
+
"grad_norm": 0.037293024361133575,
|
| 8957 |
+
"learning_rate": 2.4488032019563402e-06,
|
| 8958 |
+
"loss": 11.9166,
|
| 8959 |
+
"step": 1274
|
| 8960 |
+
},
|
| 8961 |
+
{
|
| 8962 |
+
"epoch": 0.034438284568272214,
|
| 8963 |
+
"grad_norm": 0.018514040857553482,
|
| 8964 |
+
"learning_rate": 2.398256027428436e-06,
|
| 8965 |
+
"loss": 11.9198,
|
| 8966 |
+
"step": 1275
|
| 8967 |
+
},
|
| 8968 |
+
{
|
| 8969 |
+
"epoch": 0.034465294987541445,
|
| 8970 |
+
"grad_norm": 0.024408234283328056,
|
| 8971 |
+
"learning_rate": 2.348229662899193e-06,
|
| 8972 |
+
"loss": 11.9143,
|
| 8973 |
+
"step": 1276
|
| 8974 |
+
},
|
| 8975 |
+
{
|
| 8976 |
+
"epoch": 0.034492305406810676,
|
| 8977 |
+
"grad_norm": 0.03678836300969124,
|
| 8978 |
+
"learning_rate": 2.2987243753129106e-06,
|
| 8979 |
+
"loss": 11.9151,
|
| 8980 |
+
"step": 1277
|
| 8981 |
+
},
|
| 8982 |
+
{
|
| 8983 |
+
"epoch": 0.03451931582607991,
|
| 8984 |
+
"grad_norm": 0.02599814161658287,
|
| 8985 |
+
"learning_rate": 2.2497404288334245e-06,
|
| 8986 |
+
"loss": 11.9117,
|
| 8987 |
+
"step": 1278
|
| 8988 |
+
},
|
| 8989 |
+
{
|
| 8990 |
+
"epoch": 0.034546326245349146,
|
| 8991 |
+
"grad_norm": 0.04199979081749916,
|
| 8992 |
+
"learning_rate": 2.2012780848426285e-06,
|
| 8993 |
+
"loss": 11.9198,
|
| 8994 |
+
"step": 1279
|
| 8995 |
+
},
|
| 8996 |
+
{
|
| 8997 |
+
"epoch": 0.03457333666461838,
|
| 8998 |
+
"grad_norm": 0.0232488252222538,
|
| 8999 |
+
"learning_rate": 2.1533376019391094e-06,
|
| 9000 |
+
"loss": 11.914,
|
| 9001 |
+
"step": 1280
|
| 9002 |
+
},
|
| 9003 |
+
{
|
| 9004 |
+
"epoch": 0.03460034708388761,
|
| 9005 |
+
"grad_norm": 0.023775804787874222,
|
| 9006 |
+
"learning_rate": 2.1059192359367486e-06,
|
| 9007 |
+
"loss": 11.9162,
|
| 9008 |
+
"step": 1281
|
| 9009 |
+
},
|
| 9010 |
+
{
|
| 9011 |
+
"epoch": 0.03462735750315684,
|
| 9012 |
+
"grad_norm": 0.034829165786504745,
|
| 9013 |
+
"learning_rate": 2.0590232398634114e-06,
|
| 9014 |
+
"loss": 11.9107,
|
| 9015 |
+
"step": 1282
|
| 9016 |
+
},
|
| 9017 |
+
{
|
| 9018 |
+
"epoch": 0.03465436792242608,
|
| 9019 |
+
"grad_norm": 0.04222894087433815,
|
| 9020 |
+
"learning_rate": 2.0126498639595483e-06,
|
| 9021 |
+
"loss": 11.916,
|
| 9022 |
+
"step": 1283
|
| 9023 |
+
},
|
| 9024 |
+
{
|
| 9025 |
+
"epoch": 0.03468137834169531,
|
| 9026 |
+
"grad_norm": 0.045567966997623444,
|
| 9027 |
+
"learning_rate": 1.9667993556768516e-06,
|
| 9028 |
+
"loss": 11.9099,
|
| 9029 |
+
"step": 1284
|
| 9030 |
+
},
|
| 9031 |
+
{
|
| 9032 |
+
"epoch": 0.03470838876096454,
|
| 9033 |
+
"grad_norm": 0.03558320179581642,
|
| 9034 |
+
"learning_rate": 1.921471959676957e-06,
|
| 9035 |
+
"loss": 11.9142,
|
| 9036 |
+
"step": 1285
|
| 9037 |
+
},
|
| 9038 |
+
{
|
| 9039 |
+
"epoch": 0.03473539918023377,
|
| 9040 |
+
"grad_norm": 0.02498921938240528,
|
| 9041 |
+
"learning_rate": 1.876667917830155e-06,
|
| 9042 |
+
"loss": 11.9097,
|
| 9043 |
+
"step": 1286
|
| 9044 |
+
},
|
| 9045 |
+
{
|
| 9046 |
+
"epoch": 0.03476240959950301,
|
| 9047 |
+
"grad_norm": 0.03918744996190071,
|
| 9048 |
+
"learning_rate": 1.8323874692140808e-06,
|
| 9049 |
+
"loss": 11.9157,
|
| 9050 |
+
"step": 1287
|
| 9051 |
+
},
|
| 9052 |
+
{
|
| 9053 |
+
"epoch": 0.03478942001877224,
|
| 9054 |
+
"grad_norm": 0.03008473478257656,
|
| 9055 |
+
"learning_rate": 1.7886308501124049e-06,
|
| 9056 |
+
"loss": 11.9202,
|
| 9057 |
+
"step": 1288
|
| 9058 |
+
},
|
| 9059 |
+
{
|
| 9060 |
+
"epoch": 0.034816430438041474,
|
| 9061 |
+
"grad_norm": 0.020178912207484245,
|
| 9062 |
+
"learning_rate": 1.7453982940136338e-06,
|
| 9063 |
+
"loss": 11.916,
|
| 9064 |
+
"step": 1289
|
| 9065 |
+
},
|
| 9066 |
+
{
|
| 9067 |
+
"epoch": 0.034843440857310705,
|
| 9068 |
+
"grad_norm": 0.021205449476838112,
|
| 9069 |
+
"learning_rate": 1.7026900316098215e-06,
|
| 9070 |
+
"loss": 11.9163,
|
| 9071 |
+
"step": 1290
|
| 9072 |
+
},
|
| 9073 |
+
{
|
| 9074 |
+
"epoch": 0.03487045127657994,
|
| 9075 |
+
"grad_norm": 0.03665931150317192,
|
| 9076 |
+
"learning_rate": 1.6605062907953828e-06,
|
| 9077 |
+
"loss": 11.9151,
|
| 9078 |
+
"step": 1291
|
| 9079 |
+
},
|
| 9080 |
+
{
|
| 9081 |
+
"epoch": 0.034897461695849175,
|
| 9082 |
+
"grad_norm": 0.03289037197828293,
|
| 9083 |
+
"learning_rate": 1.6188472966658042e-06,
|
| 9084 |
+
"loss": 11.9163,
|
| 9085 |
+
"step": 1292
|
| 9086 |
+
},
|
| 9087 |
+
{
|
| 9088 |
+
"epoch": 0.034924472115118406,
|
| 9089 |
+
"grad_norm": 0.040927328169345856,
|
| 9090 |
+
"learning_rate": 1.5777132715165011e-06,
|
| 9091 |
+
"loss": 11.9203,
|
| 9092 |
+
"step": 1293
|
| 9093 |
+
},
|
| 9094 |
+
{
|
| 9095 |
+
"epoch": 0.03495148253438764,
|
| 9096 |
+
"grad_norm": 0.026337536051869392,
|
| 9097 |
+
"learning_rate": 1.5371044348416408e-06,
|
| 9098 |
+
"loss": 11.9134,
|
| 9099 |
+
"step": 1294
|
| 9100 |
+
},
|
| 9101 |
+
{
|
| 9102 |
+
"epoch": 0.034978492953656876,
|
| 9103 |
+
"grad_norm": 0.024075064808130264,
|
| 9104 |
+
"learning_rate": 1.49702100333291e-06,
|
| 9105 |
+
"loss": 11.9119,
|
| 9106 |
+
"step": 1295
|
| 9107 |
+
},
|
| 9108 |
+
{
|
| 9109 |
+
"epoch": 0.03500550337292611,
|
| 9110 |
+
"grad_norm": 0.021006707102060318,
|
| 9111 |
+
"learning_rate": 1.4574631908784275e-06,
|
| 9112 |
+
"loss": 11.9183,
|
| 9113 |
+
"step": 1296
|
| 9114 |
+
},
|
| 9115 |
+
{
|
| 9116 |
+
"epoch": 0.03503251379219534,
|
| 9117 |
+
"grad_norm": 0.03662874922156334,
|
| 9118 |
+
"learning_rate": 1.4184312085615436e-06,
|
| 9119 |
+
"loss": 11.9168,
|
| 9120 |
+
"step": 1297
|
| 9121 |
+
},
|
| 9122 |
+
{
|
| 9123 |
+
"epoch": 0.03505952421146457,
|
| 9124 |
+
"grad_norm": 0.02508450672030449,
|
| 9125 |
+
"learning_rate": 1.3799252646597426e-06,
|
| 9126 |
+
"loss": 11.9116,
|
| 9127 |
+
"step": 1298
|
| 9128 |
+
},
|
| 9129 |
+
{
|
| 9130 |
+
"epoch": 0.03508653463073381,
|
| 9131 |
+
"grad_norm": 0.027173347771167755,
|
| 9132 |
+
"learning_rate": 1.3419455646435652e-06,
|
| 9133 |
+
"loss": 11.9194,
|
| 9134 |
+
"step": 1299
|
| 9135 |
+
},
|
| 9136 |
+
{
|
| 9137 |
+
"epoch": 0.03511354505000304,
|
| 9138 |
+
"grad_norm": 0.02922755852341652,
|
| 9139 |
+
"learning_rate": 1.3044923111754427e-06,
|
| 9140 |
+
"loss": 11.9091,
|
| 9141 |
+
"step": 1300
|
| 9142 |
+
},
|
| 9143 |
+
{
|
| 9144 |
+
"epoch": 0.03514055546927227,
|
| 9145 |
+
"grad_norm": 0.02054756134748459,
|
| 9146 |
+
"learning_rate": 1.267565704108642e-06,
|
| 9147 |
+
"loss": 11.9152,
|
| 9148 |
+
"step": 1301
|
| 9149 |
+
},
|
| 9150 |
+
{
|
| 9151 |
+
"epoch": 0.0351675658885415,
|
| 9152 |
+
"grad_norm": 0.02704966813325882,
|
| 9153 |
+
"learning_rate": 1.231165940486234e-06,
|
| 9154 |
+
"loss": 11.9156,
|
| 9155 |
+
"step": 1302
|
| 9156 |
+
},
|
| 9157 |
+
{
|
| 9158 |
+
"epoch": 0.03519457630781074,
|
| 9159 |
+
"grad_norm": 0.024654177948832512,
|
| 9160 |
+
"learning_rate": 1.1952932145399943e-06,
|
| 9161 |
+
"loss": 11.9216,
|
| 9162 |
+
"step": 1303
|
| 9163 |
+
},
|
| 9164 |
+
{
|
| 9165 |
+
"epoch": 0.03522158672707997,
|
| 9166 |
+
"grad_norm": 0.03191498667001724,
|
| 9167 |
+
"learning_rate": 1.1599477176894135e-06,
|
| 9168 |
+
"loss": 11.9125,
|
| 9169 |
+
"step": 1304
|
| 9170 |
+
},
|
| 9171 |
+
{
|
| 9172 |
+
"epoch": 0.0352485971463492,
|
| 9173 |
+
"grad_norm": 0.044170890003442764,
|
| 9174 |
+
"learning_rate": 1.125129638540623e-06,
|
| 9175 |
+
"loss": 11.9205,
|
| 9176 |
+
"step": 1305
|
| 9177 |
+
},
|
| 9178 |
+
{
|
| 9179 |
+
"epoch": 0.035275607565618435,
|
| 9180 |
+
"grad_norm": 0.024596545845270157,
|
| 9181 |
+
"learning_rate": 1.0908391628854041e-06,
|
| 9182 |
+
"loss": 11.9156,
|
| 9183 |
+
"step": 1306
|
| 9184 |
+
},
|
| 9185 |
+
{
|
| 9186 |
+
"epoch": 0.03530261798488767,
|
| 9187 |
+
"grad_norm": 0.02692142128944397,
|
| 9188 |
+
"learning_rate": 1.057076473700247e-06,
|
| 9189 |
+
"loss": 11.9133,
|
| 9190 |
+
"step": 1307
|
| 9191 |
+
},
|
| 9192 |
+
{
|
| 9193 |
+
"epoch": 0.035329628404156904,
|
| 9194 |
+
"grad_norm": 0.02368812821805477,
|
| 9195 |
+
"learning_rate": 1.0238417511453157e-06,
|
| 9196 |
+
"loss": 11.9158,
|
| 9197 |
+
"step": 1308
|
| 9198 |
+
},
|
| 9199 |
+
{
|
| 9200 |
+
"epoch": 0.035356638823426136,
|
| 9201 |
+
"grad_norm": 0.027224168181419373,
|
| 9202 |
+
"learning_rate": 9.911351725635065e-07,
|
| 9203 |
+
"loss": 11.9143,
|
| 9204 |
+
"step": 1309
|
| 9205 |
+
},
|
| 9206 |
+
{
|
| 9207 |
+
"epoch": 0.03538364924269537,
|
| 9208 |
+
"grad_norm": 0.03143531084060669,
|
| 9209 |
+
"learning_rate": 9.589569124794916e-07,
|
| 9210 |
+
"loss": 11.91,
|
| 9211 |
+
"step": 1310
|
| 9212 |
+
},
|
| 9213 |
+
{
|
| 9214 |
+
"epoch": 0.035410659661964605,
|
| 9215 |
+
"grad_norm": 0.04223516955971718,
|
| 9216 |
+
"learning_rate": 9.273071425987878e-07,
|
| 9217 |
+
"loss": 11.9146,
|
| 9218 |
+
"step": 1311
|
| 9219 |
+
},
|
| 9220 |
+
{
|
| 9221 |
+
"epoch": 0.03543767008123384,
|
| 9222 |
+
"grad_norm": 0.03618738427758217,
|
| 9223 |
+
"learning_rate": 8.961860318069115e-07,
|
| 9224 |
+
"loss": 11.9113,
|
| 9225 |
+
"step": 1312
|
| 9226 |
+
},
|
| 9227 |
+
{
|
| 9228 |
+
"epoch": 0.03546468050050307,
|
| 9229 |
+
"grad_norm": 0.04262852296233177,
|
| 9230 |
+
"learning_rate": 8.655937461683361e-07,
|
| 9231 |
+
"loss": 11.9149,
|
| 9232 |
+
"step": 1313
|
| 9233 |
+
},
|
| 9234 |
+
{
|
| 9235 |
+
"epoch": 0.0354916909197723,
|
| 9236 |
+
"grad_norm": 0.0435975007712841,
|
| 9237 |
+
"learning_rate": 8.355304489257254e-07,
|
| 9238 |
+
"loss": 11.9074,
|
| 9239 |
+
"step": 1314
|
| 9240 |
+
},
|
| 9241 |
+
{
|
| 9242 |
+
"epoch": 0.03551870133904154,
|
| 9243 |
+
"grad_norm": 0.037201475352048874,
|
| 9244 |
+
"learning_rate": 8.059963004990234e-07,
|
| 9245 |
+
"loss": 11.9133,
|
| 9246 |
+
"step": 1315
|
| 9247 |
+
},
|
| 9248 |
+
{
|
| 9249 |
+
"epoch": 0.03554571175831077,
|
| 9250 |
+
"grad_norm": 0.037290722131729126,
|
| 9251 |
+
"learning_rate": 7.769914584845994e-07,
|
| 9252 |
+
"loss": 11.9123,
|
| 9253 |
+
"step": 1316
|
| 9254 |
+
},
|
| 9255 |
+
{
|
| 9256 |
+
"epoch": 0.03557272217758,
|
| 9257 |
+
"grad_norm": 0.03923411667346954,
|
| 9258 |
+
"learning_rate": 7.485160776543931e-07,
|
| 9259 |
+
"loss": 11.9101,
|
| 9260 |
+
"step": 1317
|
| 9261 |
+
},
|
| 9262 |
+
{
|
| 9263 |
+
"epoch": 0.03559973259684923,
|
| 9264 |
+
"grad_norm": 0.025412987917661667,
|
| 9265 |
+
"learning_rate": 7.205703099551042e-07,
|
| 9266 |
+
"loss": 11.9154,
|
| 9267 |
+
"step": 1318
|
| 9268 |
+
},
|
| 9269 |
+
{
|
| 9270 |
+
"epoch": 0.03562674301611847,
|
| 9271 |
+
"grad_norm": 0.04205279424786568,
|
| 9272 |
+
"learning_rate": 6.931543045073708e-07,
|
| 9273 |
+
"loss": 11.9172,
|
| 9274 |
+
"step": 1319
|
| 9275 |
+
},
|
| 9276 |
+
{
|
| 9277 |
+
"epoch": 0.0356537534353877,
|
| 9278 |
+
"grad_norm": 0.030700096860527992,
|
| 9279 |
+
"learning_rate": 6.662682076050031e-07,
|
| 9280 |
+
"loss": 11.9168,
|
| 9281 |
+
"step": 1320
|
| 9282 |
+
},
|
| 9283 |
+
{
|
| 9284 |
+
"epoch": 0.03568076385465693,
|
| 9285 |
+
"grad_norm": 0.03404088318347931,
|
| 9286 |
+
"learning_rate": 6.399121627141736e-07,
|
| 9287 |
+
"loss": 11.9157,
|
| 9288 |
+
"step": 1321
|
| 9289 |
+
},
|
| 9290 |
+
{
|
| 9291 |
+
"epoch": 0.035707774273926164,
|
| 9292 |
+
"grad_norm": 0.025373445823788643,
|
| 9293 |
+
"learning_rate": 6.140863104726391e-07,
|
| 9294 |
+
"loss": 11.9136,
|
| 9295 |
+
"step": 1322
|
| 9296 |
+
},
|
| 9297 |
+
{
|
| 9298 |
+
"epoch": 0.0357347846931954,
|
| 9299 |
+
"grad_norm": 0.028424931690096855,
|
| 9300 |
+
"learning_rate": 5.887907886890198e-07,
|
| 9301 |
+
"loss": 11.9154,
|
| 9302 |
+
"step": 1323
|
| 9303 |
+
},
|
| 9304 |
+
{
|
| 9305 |
+
"epoch": 0.035761795112464634,
|
| 9306 |
+
"grad_norm": 0.044030915945768356,
|
| 9307 |
+
"learning_rate": 5.640257323420772e-07,
|
| 9308 |
+
"loss": 11.9125,
|
| 9309 |
+
"step": 1324
|
| 9310 |
+
},
|
| 9311 |
+
{
|
| 9312 |
+
"epoch": 0.035788805531733865,
|
| 9313 |
+
"grad_norm": 0.03599056974053383,
|
| 9314 |
+
"learning_rate": 5.397912735799371e-07,
|
| 9315 |
+
"loss": 11.9142,
|
| 9316 |
+
"step": 1325
|
| 9317 |
+
},
|
| 9318 |
+
{
|
| 9319 |
+
"epoch": 0.0358158159510031,
|
| 9320 |
+
"grad_norm": 0.04653029888868332,
|
| 9321 |
+
"learning_rate": 5.160875417194455e-07,
|
| 9322 |
+
"loss": 11.9164,
|
| 9323 |
+
"step": 1326
|
| 9324 |
+
},
|
| 9325 |
+
{
|
| 9326 |
+
"epoch": 0.035842826370272335,
|
| 9327 |
+
"grad_norm": 0.03172197937965393,
|
| 9328 |
+
"learning_rate": 4.929146632454252e-07,
|
| 9329 |
+
"loss": 11.9179,
|
| 9330 |
+
"step": 1327
|
| 9331 |
+
},
|
| 9332 |
+
{
|
| 9333 |
+
"epoch": 0.035869836789541566,
|
| 9334 |
+
"grad_norm": 0.027123969048261642,
|
| 9335 |
+
"learning_rate": 4.702727618100422e-07,
|
| 9336 |
+
"loss": 11.9156,
|
| 9337 |
+
"step": 1328
|
| 9338 |
+
},
|
| 9339 |
+
{
|
| 9340 |
+
"epoch": 0.0358968472088108,
|
| 9341 |
+
"grad_norm": 0.03513197973370552,
|
| 9342 |
+
"learning_rate": 4.4816195823212945e-07,
|
| 9343 |
+
"loss": 11.9111,
|
| 9344 |
+
"step": 1329
|
| 9345 |
+
},
|
| 9346 |
+
{
|
| 9347 |
+
"epoch": 0.03592385762808003,
|
| 9348 |
+
"grad_norm": 0.03823890537023544,
|
| 9349 |
+
"learning_rate": 4.2658237049655323e-07,
|
| 9350 |
+
"loss": 11.9079,
|
| 9351 |
+
"step": 1330
|
| 9352 |
+
},
|
| 9353 |
+
{
|
| 9354 |
+
"epoch": 0.03595086804734927,
|
| 9355 |
+
"grad_norm": 0.034188203513622284,
|
| 9356 |
+
"learning_rate": 4.0553411375353623e-07,
|
| 9357 |
+
"loss": 11.9137,
|
| 9358 |
+
"step": 1331
|
| 9359 |
+
},
|
| 9360 |
+
{
|
| 9361 |
+
"epoch": 0.0359778784666185,
|
| 9362 |
+
"grad_norm": 0.03273677080869675,
|
| 9363 |
+
"learning_rate": 3.8501730031813566e-07,
|
| 9364 |
+
"loss": 11.9154,
|
| 9365 |
+
"step": 1332
|
| 9366 |
+
},
|
| 9367 |
+
{
|
| 9368 |
+
"epoch": 0.03600488888588773,
|
| 9369 |
+
"grad_norm": 0.049734581261873245,
|
| 9370 |
+
"learning_rate": 3.650320396695328e-07,
|
| 9371 |
+
"loss": 11.9163,
|
| 9372 |
+
"step": 1333
|
| 9373 |
+
},
|
| 9374 |
+
{
|
| 9375 |
+
"epoch": 0.03603189930515696,
|
| 9376 |
+
"grad_norm": 0.03474251553416252,
|
| 9377 |
+
"learning_rate": 3.455784384505445e-07,
|
| 9378 |
+
"loss": 11.9091,
|
| 9379 |
+
"step": 1334
|
| 9380 |
+
},
|
| 9381 |
+
{
|
| 9382 |
+
"epoch": 0.0360589097244262,
|
| 9383 |
+
"grad_norm": 0.0435771606862545,
|
| 9384 |
+
"learning_rate": 3.266566004670013e-07,
|
| 9385 |
+
"loss": 11.9124,
|
| 9386 |
+
"step": 1335
|
| 9387 |
+
},
|
| 9388 |
+
{
|
| 9389 |
+
"epoch": 0.03608592014369543,
|
| 9390 |
+
"grad_norm": 0.0319393053650856,
|
| 9391 |
+
"learning_rate": 3.0826662668720364e-07,
|
| 9392 |
+
"loss": 11.9126,
|
| 9393 |
+
"step": 1336
|
| 9394 |
+
},
|
| 9395 |
+
{
|
| 9396 |
+
"epoch": 0.03611293056296466,
|
| 9397 |
+
"grad_norm": 0.02378806099295616,
|
| 9398 |
+
"learning_rate": 2.9040861524138874e-07,
|
| 9399 |
+
"loss": 11.9153,
|
| 9400 |
+
"step": 1337
|
| 9401 |
+
},
|
| 9402 |
+
{
|
| 9403 |
+
"epoch": 0.036139940982233894,
|
| 9404 |
+
"grad_norm": 0.030763791874051094,
|
| 9405 |
+
"learning_rate": 2.7308266142119785e-07,
|
| 9406 |
+
"loss": 11.9087,
|
| 9407 |
+
"step": 1338
|
| 9408 |
+
},
|
| 9409 |
+
{
|
| 9410 |
+
"epoch": 0.03616695140150313,
|
| 9411 |
+
"grad_norm": 0.032679032534360886,
|
| 9412 |
+
"learning_rate": 2.5628885767918777e-07,
|
| 9413 |
+
"loss": 11.9118,
|
| 9414 |
+
"step": 1339
|
| 9415 |
+
},
|
| 9416 |
+
{
|
| 9417 |
+
"epoch": 0.03619396182077236,
|
| 9418 |
+
"grad_norm": 0.031153904274106026,
|
| 9419 |
+
"learning_rate": 2.400272936283088e-07,
|
| 9420 |
+
"loss": 11.9108,
|
| 9421 |
+
"step": 1340
|
| 9422 |
+
},
|
| 9423 |
+
{
|
| 9424 |
+
"epoch": 0.036220972240041595,
|
| 9425 |
+
"grad_norm": 0.02682630345225334,
|
| 9426 |
+
"learning_rate": 2.2429805604144983e-07,
|
| 9427 |
+
"loss": 11.9122,
|
| 9428 |
+
"step": 1341
|
| 9429 |
+
},
|
| 9430 |
+
{
|
| 9431 |
+
"epoch": 0.036247982659310826,
|
| 9432 |
+
"grad_norm": 0.029447071254253387,
|
| 9433 |
+
"learning_rate": 2.0910122885097194e-07,
|
| 9434 |
+
"loss": 11.9118,
|
| 9435 |
+
"step": 1342
|
| 9436 |
+
},
|
| 9437 |
+
{
|
| 9438 |
+
"epoch": 0.036274993078580064,
|
| 9439 |
+
"grad_norm": 0.029255639761686325,
|
| 9440 |
+
"learning_rate": 1.94436893148231e-07,
|
| 9441 |
+
"loss": 11.9144,
|
| 9442 |
+
"step": 1343
|
| 9443 |
+
},
|
| 9444 |
+
{
|
| 9445 |
+
"epoch": 0.036302003497849296,
|
| 9446 |
+
"grad_norm": 0.03713764622807503,
|
| 9447 |
+
"learning_rate": 1.8030512718322234e-07,
|
| 9448 |
+
"loss": 11.9159,
|
| 9449 |
+
"step": 1344
|
| 9450 |
+
},
|
| 9451 |
+
{
|
| 9452 |
+
"epoch": 0.03632901391711853,
|
| 9453 |
+
"grad_norm": 0.032858531922101974,
|
| 9454 |
+
"learning_rate": 1.6670600636403687e-07,
|
| 9455 |
+
"loss": 11.9196,
|
| 9456 |
+
"step": 1345
|
| 9457 |
+
},
|
| 9458 |
+
{
|
| 9459 |
+
"epoch": 0.03635602433638776,
|
| 9460 |
+
"grad_norm": 0.01837058551609516,
|
| 9461 |
+
"learning_rate": 1.5363960325660565e-07,
|
| 9462 |
+
"loss": 11.9125,
|
| 9463 |
+
"step": 1346
|
| 9464 |
+
},
|
| 9465 |
+
{
|
| 9466 |
+
"epoch": 0.036383034755657,
|
| 9467 |
+
"grad_norm": 0.023487647995352745,
|
| 9468 |
+
"learning_rate": 1.4110598758417803e-07,
|
| 9469 |
+
"loss": 11.9123,
|
| 9470 |
+
"step": 1347
|
| 9471 |
+
},
|
| 9472 |
+
{
|
| 9473 |
+
"epoch": 0.03641004517492623,
|
| 9474 |
+
"grad_norm": 0.04510617256164551,
|
| 9475 |
+
"learning_rate": 1.2910522622705535e-07,
|
| 9476 |
+
"loss": 11.9179,
|
| 9477 |
+
"step": 1348
|
| 9478 |
+
},
|
| 9479 |
+
{
|
| 9480 |
+
"epoch": 0.03643705559419546,
|
| 9481 |
+
"grad_norm": 0.03857404366135597,
|
| 9482 |
+
"learning_rate": 1.1763738322216889e-07,
|
| 9483 |
+
"loss": 11.9141,
|
| 9484 |
+
"step": 1349
|
| 9485 |
+
},
|
| 9486 |
+
{
|
| 9487 |
+
"epoch": 0.03646406601346469,
|
| 9488 |
+
"grad_norm": 0.02398962341248989,
|
| 9489 |
+
"learning_rate": 1.0670251976275803e-07,
|
| 9490 |
+
"loss": 11.9184,
|
| 9491 |
+
"step": 1350
|
| 9492 |
+
},
|
| 9493 |
+
{
|
| 9494 |
+
"epoch": 0.03649107643273393,
|
| 9495 |
+
"grad_norm": 0.021815018728375435,
|
| 9496 |
+
"learning_rate": 9.630069419804821e-08,
|
| 9497 |
+
"loss": 11.9147,
|
| 9498 |
+
"step": 1351
|
| 9499 |
+
},
|
| 9500 |
+
{
|
| 9501 |
+
"epoch": 0.03651808685200316,
|
| 9502 |
+
"grad_norm": 0.03804771974682808,
|
| 9503 |
+
"learning_rate": 8.643196203294013e-08,
|
| 9504 |
+
"loss": 11.9196,
|
| 9505 |
+
"step": 1352
|
| 9506 |
+
},
|
| 9507 |
+
{
|
| 9508 |
+
"epoch": 0.03654509727127239,
|
| 9509 |
+
"grad_norm": 0.04712639003992081,
|
| 9510 |
+
"learning_rate": 7.709637592770991e-08,
|
| 9511 |
+
"loss": 11.9113,
|
| 9512 |
+
"step": 1353
|
| 9513 |
+
},
|
| 9514 |
+
{
|
| 9515 |
+
"epoch": 0.03657210769054162,
|
| 9516 |
+
"grad_norm": 0.026805303990840912,
|
| 9517 |
+
"learning_rate": 6.829398569770939e-08,
|
| 9518 |
+
"loss": 11.9147,
|
| 9519 |
+
"step": 1354
|
| 9520 |
+
},
|
| 9521 |
+
{
|
| 9522 |
+
"epoch": 0.03659911810981086,
|
| 9523 |
+
"grad_norm": 0.025319278240203857,
|
| 9524 |
+
"learning_rate": 6.002483831314409e-08,
|
| 9525 |
+
"loss": 11.9099,
|
| 9526 |
+
"step": 1355
|
| 9527 |
+
},
|
| 9528 |
+
{
|
| 9529 |
+
"epoch": 0.03662612852908009,
|
| 9530 |
+
"grad_norm": 0.049907177686691284,
|
| 9531 |
+
"learning_rate": 5.2288977898784505e-08,
|
| 9532 |
+
"loss": 11.9144,
|
| 9533 |
+
"step": 1356
|
| 9534 |
+
},
|
| 9535 |
+
{
|
| 9536 |
+
"epoch": 0.036653138948349324,
|
| 9537 |
+
"grad_norm": 0.026024073362350464,
|
| 9538 |
+
"learning_rate": 4.50864457337441e-08,
|
| 9539 |
+
"loss": 11.9172,
|
| 9540 |
+
"step": 1357
|
| 9541 |
+
},
|
| 9542 |
+
{
|
| 9543 |
+
"epoch": 0.036680149367618556,
|
| 9544 |
+
"grad_norm": 0.03450091555714607,
|
| 9545 |
+
"learning_rate": 3.841728025125724e-08,
|
| 9546 |
+
"loss": 11.91,
|
| 9547 |
+
"step": 1358
|
| 9548 |
+
},
|
| 9549 |
+
{
|
| 9550 |
+
"epoch": 0.036707159786887794,
|
| 9551 |
+
"grad_norm": 0.040472082793712616,
|
| 9552 |
+
"learning_rate": 3.228151703847937e-08,
|
| 9553 |
+
"loss": 11.9134,
|
| 9554 |
+
"step": 1359
|
| 9555 |
+
},
|
| 9556 |
+
{
|
| 9557 |
+
"epoch": 0.036734170206157025,
|
| 9558 |
+
"grad_norm": 0.03463630750775337,
|
| 9559 |
+
"learning_rate": 2.667918883627607e-08,
|
| 9560 |
+
"loss": 11.9122,
|
| 9561 |
+
"step": 1360
|
| 9562 |
+
},
|
| 9563 |
+
{
|
| 9564 |
+
"epoch": 0.03676118062542626,
|
| 9565 |
+
"grad_norm": 0.029601003974676132,
|
| 9566 |
+
"learning_rate": 2.1610325539089815e-08,
|
| 9567 |
+
"loss": 11.9121,
|
| 9568 |
+
"step": 1361
|
| 9569 |
+
},
|
| 9570 |
+
{
|
| 9571 |
+
"epoch": 0.03678819104469549,
|
| 9572 |
+
"grad_norm": 0.034978847950696945,
|
| 9573 |
+
"learning_rate": 1.7074954194729044e-08,
|
| 9574 |
+
"loss": 11.9169,
|
| 9575 |
+
"step": 1362
|
| 9576 |
+
},
|
| 9577 |
+
{
|
| 9578 |
+
"epoch": 0.03681520146396473,
|
| 9579 |
+
"grad_norm": 0.023429909721016884,
|
| 9580 |
+
"learning_rate": 1.3073099004290434e-08,
|
| 9581 |
+
"loss": 11.9144,
|
| 9582 |
+
"step": 1363
|
| 9583 |
+
},
|
| 9584 |
+
{
|
| 9585 |
+
"epoch": 0.03684221188323396,
|
| 9586 |
+
"grad_norm": 0.0353400856256485,
|
| 9587 |
+
"learning_rate": 9.604781321936873e-09,
|
| 9588 |
+
"loss": 11.9116,
|
| 9589 |
+
"step": 1364
|
| 9590 |
+
},
|
| 9591 |
+
{
|
| 9592 |
+
"epoch": 0.03686922230250319,
|
| 9593 |
+
"grad_norm": 0.03302585333585739,
|
| 9594 |
+
"learning_rate": 6.670019654875237e-09,
|
| 9595 |
+
"loss": 11.9149,
|
| 9596 |
+
"step": 1365
|
| 9597 |
+
},
|
| 9598 |
+
{
|
| 9599 |
+
"epoch": 0.03689623272177242,
|
| 9600 |
+
"grad_norm": 0.05004185065627098,
|
| 9601 |
+
"learning_rate": 4.268829663212071e-09,
|
| 9602 |
+
"loss": 11.9105,
|
| 9603 |
+
"step": 1366
|
| 9604 |
+
},
|
| 9605 |
+
{
|
| 9606 |
+
"epoch": 0.03692324314104166,
|
| 9607 |
+
"grad_norm": 0.02512243203818798,
|
| 9608 |
+
"learning_rate": 2.40122415987587e-09,
|
| 9609 |
+
"loss": 11.9121,
|
| 9610 |
+
"step": 1367
|
| 9611 |
+
},
|
| 9612 |
+
{
|
| 9613 |
+
"epoch": 0.03695025356031089,
|
| 9614 |
+
"grad_norm": 0.03573236241936684,
|
| 9615 |
+
"learning_rate": 1.0672131105282645e-09,
|
| 9616 |
+
"loss": 11.9171,
|
| 9617 |
+
"step": 1368
|
| 9618 |
+
},
|
| 9619 |
+
{
|
| 9620 |
+
"epoch": 0.03697726397958012,
|
| 9621 |
+
"grad_norm": 0.0429963655769825,
|
| 9622 |
+
"learning_rate": 2.668036335529145e-10,
|
| 9623 |
+
"loss": 11.9096,
|
| 9624 |
+
"step": 1369
|
| 9625 |
+
},
|
| 9626 |
+
{
|
| 9627 |
+
"epoch": 0.03700427439884935,
|
| 9628 |
+
"grad_norm": 0.03545090928673744,
|
| 9629 |
+
"learning_rate": 0.0,
|
| 9630 |
+
"loss": 11.9181,
|
| 9631 |
+
"step": 1370
|
| 9632 |
}
|
| 9633 |
],
|
| 9634 |
"logging_steps": 1,
|
|
|
|
| 9643 |
"should_evaluate": false,
|
| 9644 |
"should_log": false,
|
| 9645 |
"should_save": true,
|
| 9646 |
+
"should_training_stop": true
|
| 9647 |
},
|
| 9648 |
"attributes": {}
|
| 9649 |
}
|
| 9650 |
},
|
| 9651 |
+
"total_flos": 319951552512.0,
|
| 9652 |
"train_batch_size": 2,
|
| 9653 |
"trial_name": null,
|
| 9654 |
"trial_params": null
|