Training in progress, step 1417, 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 125248064
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6e2899bb851d328f8adadfe28b64c90197cdc9df4b928bb70bd0e66cba8ca687
|
3 |
size 125248064
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 64220436
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c8d32cc82193045cb6b5f11899152c8381df0be76e908991130af4d2b118f6b2
|
3 |
size 64220436
|
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:ac894c64cbcf4aad35d02cddc8ead9ae4ce5c3d53a4f512a7bf525555fbded5b
|
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 sha256:938af08430819ad570ba94ba21d2a6e8af7d460eebba19f4f25f8727d34c0a08
|
3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
-
"epoch": 0.
|
5 |
"eval_steps": 500,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -7462,6 +7462,2470 @@
|
|
7462 |
"learning_rate": 1.4566085573529874e-05,
|
7463 |
"loss": 1.3304,
|
7464 |
"step": 1065
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
7465 |
}
|
7466 |
],
|
7467 |
"logging_steps": 1,
|
@@ -7476,12 +9940,12 @@
|
|
7476 |
"should_evaluate": false,
|
7477 |
"should_log": false,
|
7478 |
"should_save": true,
|
7479 |
-
"should_training_stop":
|
7480 |
},
|
7481 |
"attributes": {}
|
7482 |
}
|
7483 |
},
|
7484 |
-
"total_flos": 1.
|
7485 |
"train_batch_size": 4,
|
7486 |
"trial_name": null,
|
7487 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.242387957577831,
|
5 |
"eval_steps": 500,
|
6 |
+
"global_step": 1417,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
7462 |
"learning_rate": 1.4566085573529874e-05,
|
7463 |
"loss": 1.3304,
|
7464 |
"step": 1065
|
7465 |
+
},
|
7466 |
+
{
|
7467 |
+
"epoch": 0.1823469038658912,
|
7468 |
+
"grad_norm": 0.6401252746582031,
|
7469 |
+
"learning_rate": 1.4487685614349999e-05,
|
7470 |
+
"loss": 1.4265,
|
7471 |
+
"step": 1066
|
7472 |
+
},
|
7473 |
+
{
|
7474 |
+
"epoch": 0.18251796099897366,
|
7475 |
+
"grad_norm": 0.6446531414985657,
|
7476 |
+
"learning_rate": 1.4409461451235068e-05,
|
7477 |
+
"loss": 1.3344,
|
7478 |
+
"step": 1067
|
7479 |
+
},
|
7480 |
+
{
|
7481 |
+
"epoch": 0.1826890181320561,
|
7482 |
+
"grad_norm": 0.5877986550331116,
|
7483 |
+
"learning_rate": 1.4331413471416999e-05,
|
7484 |
+
"loss": 1.1895,
|
7485 |
+
"step": 1068
|
7486 |
+
},
|
7487 |
+
{
|
7488 |
+
"epoch": 0.18286007526513856,
|
7489 |
+
"grad_norm": 0.6732580065727234,
|
7490 |
+
"learning_rate": 1.4253542061255515e-05,
|
7491 |
+
"loss": 1.2624,
|
7492 |
+
"step": 1069
|
7493 |
+
},
|
7494 |
+
{
|
7495 |
+
"epoch": 0.183031132398221,
|
7496 |
+
"grad_norm": 0.639057993888855,
|
7497 |
+
"learning_rate": 1.4175847606236225e-05,
|
7498 |
+
"loss": 1.2844,
|
7499 |
+
"step": 1070
|
7500 |
+
},
|
7501 |
+
{
|
7502 |
+
"epoch": 0.18320218953130346,
|
7503 |
+
"grad_norm": 0.6967434287071228,
|
7504 |
+
"learning_rate": 1.409833049096882e-05,
|
7505 |
+
"loss": 1.3516,
|
7506 |
+
"step": 1071
|
7507 |
+
},
|
7508 |
+
{
|
7509 |
+
"epoch": 0.1833732466643859,
|
7510 |
+
"grad_norm": 0.6904391646385193,
|
7511 |
+
"learning_rate": 1.4020991099185077e-05,
|
7512 |
+
"loss": 1.3984,
|
7513 |
+
"step": 1072
|
7514 |
+
},
|
7515 |
+
{
|
7516 |
+
"epoch": 0.18354430379746836,
|
7517 |
+
"grad_norm": 0.6713119745254517,
|
7518 |
+
"learning_rate": 1.3943829813737008e-05,
|
7519 |
+
"loss": 1.1528,
|
7520 |
+
"step": 1073
|
7521 |
+
},
|
7522 |
+
{
|
7523 |
+
"epoch": 0.1837153609305508,
|
7524 |
+
"grad_norm": 0.7345438599586487,
|
7525 |
+
"learning_rate": 1.3866847016594952e-05,
|
7526 |
+
"loss": 1.234,
|
7527 |
+
"step": 1074
|
7528 |
+
},
|
7529 |
+
{
|
7530 |
+
"epoch": 0.18388641806363326,
|
7531 |
+
"grad_norm": 0.716822624206543,
|
7532 |
+
"learning_rate": 1.3790043088845644e-05,
|
7533 |
+
"loss": 1.0562,
|
7534 |
+
"step": 1075
|
7535 |
+
},
|
7536 |
+
{
|
7537 |
+
"epoch": 0.1840574751967157,
|
7538 |
+
"grad_norm": 0.7419716119766235,
|
7539 |
+
"learning_rate": 1.3713418410690398e-05,
|
7540 |
+
"loss": 1.0896,
|
7541 |
+
"step": 1076
|
7542 |
+
},
|
7543 |
+
{
|
7544 |
+
"epoch": 0.18422853232979816,
|
7545 |
+
"grad_norm": 0.7504467964172363,
|
7546 |
+
"learning_rate": 1.3636973361443183e-05,
|
7547 |
+
"loss": 1.0736,
|
7548 |
+
"step": 1077
|
7549 |
+
},
|
7550 |
+
{
|
7551 |
+
"epoch": 0.1843995894628806,
|
7552 |
+
"grad_norm": 0.7508487701416016,
|
7553 |
+
"learning_rate": 1.3560708319528753e-05,
|
7554 |
+
"loss": 1.1895,
|
7555 |
+
"step": 1078
|
7556 |
+
},
|
7557 |
+
{
|
7558 |
+
"epoch": 0.18457064659596306,
|
7559 |
+
"grad_norm": 0.7630247473716736,
|
7560 |
+
"learning_rate": 1.3484623662480778e-05,
|
7561 |
+
"loss": 1.0333,
|
7562 |
+
"step": 1079
|
7563 |
+
},
|
7564 |
+
{
|
7565 |
+
"epoch": 0.1847417037290455,
|
7566 |
+
"grad_norm": 0.8874102830886841,
|
7567 |
+
"learning_rate": 1.3408719766939981e-05,
|
7568 |
+
"loss": 1.2025,
|
7569 |
+
"step": 1080
|
7570 |
+
},
|
7571 |
+
{
|
7572 |
+
"epoch": 0.18491276086212796,
|
7573 |
+
"grad_norm": 0.8376559615135193,
|
7574 |
+
"learning_rate": 1.3332997008652204e-05,
|
7575 |
+
"loss": 1.2985,
|
7576 |
+
"step": 1081
|
7577 |
+
},
|
7578 |
+
{
|
7579 |
+
"epoch": 0.1850838179952104,
|
7580 |
+
"grad_norm": 0.9795891046524048,
|
7581 |
+
"learning_rate": 1.325745576246672e-05,
|
7582 |
+
"loss": 1.2767,
|
7583 |
+
"step": 1082
|
7584 |
+
},
|
7585 |
+
{
|
7586 |
+
"epoch": 0.18525487512829286,
|
7587 |
+
"grad_norm": 0.9145246744155884,
|
7588 |
+
"learning_rate": 1.3182096402334154e-05,
|
7589 |
+
"loss": 0.9915,
|
7590 |
+
"step": 1083
|
7591 |
+
},
|
7592 |
+
{
|
7593 |
+
"epoch": 0.1854259322613753,
|
7594 |
+
"grad_norm": 0.921216607093811,
|
7595 |
+
"learning_rate": 1.3106919301304803e-05,
|
7596 |
+
"loss": 1.2161,
|
7597 |
+
"step": 1084
|
7598 |
+
},
|
7599 |
+
{
|
7600 |
+
"epoch": 0.18559698939445776,
|
7601 |
+
"grad_norm": 1.0113946199417114,
|
7602 |
+
"learning_rate": 1.3031924831526737e-05,
|
7603 |
+
"loss": 1.2265,
|
7604 |
+
"step": 1085
|
7605 |
+
},
|
7606 |
+
{
|
7607 |
+
"epoch": 0.1857680465275402,
|
7608 |
+
"grad_norm": 0.9975959062576294,
|
7609 |
+
"learning_rate": 1.2957113364243883e-05,
|
7610 |
+
"loss": 0.9407,
|
7611 |
+
"step": 1086
|
7612 |
+
},
|
7613 |
+
{
|
7614 |
+
"epoch": 0.18593910366062263,
|
7615 |
+
"grad_norm": 1.016904592514038,
|
7616 |
+
"learning_rate": 1.2882485269794337e-05,
|
7617 |
+
"loss": 1.0962,
|
7618 |
+
"step": 1087
|
7619 |
+
},
|
7620 |
+
{
|
7621 |
+
"epoch": 0.1861101607937051,
|
7622 |
+
"grad_norm": 1.4570573568344116,
|
7623 |
+
"learning_rate": 1.2808040917608416e-05,
|
7624 |
+
"loss": 1.2726,
|
7625 |
+
"step": 1088
|
7626 |
+
},
|
7627 |
+
{
|
7628 |
+
"epoch": 0.18628121792678753,
|
7629 |
+
"grad_norm": 1.337405800819397,
|
7630 |
+
"learning_rate": 1.2733780676206824e-05,
|
7631 |
+
"loss": 1.4358,
|
7632 |
+
"step": 1089
|
7633 |
+
},
|
7634 |
+
{
|
7635 |
+
"epoch": 0.18645227505987,
|
7636 |
+
"grad_norm": 1.4514150619506836,
|
7637 |
+
"learning_rate": 1.2659704913198912e-05,
|
7638 |
+
"loss": 1.3006,
|
7639 |
+
"step": 1090
|
7640 |
+
},
|
7641 |
+
{
|
7642 |
+
"epoch": 0.18662333219295243,
|
7643 |
+
"grad_norm": 1.6620286703109741,
|
7644 |
+
"learning_rate": 1.2585813995280782e-05,
|
7645 |
+
"loss": 1.3035,
|
7646 |
+
"step": 1091
|
7647 |
+
},
|
7648 |
+
{
|
7649 |
+
"epoch": 0.1867943893260349,
|
7650 |
+
"grad_norm": 1.6806087493896484,
|
7651 |
+
"learning_rate": 1.2512108288233527e-05,
|
7652 |
+
"loss": 1.2647,
|
7653 |
+
"step": 1092
|
7654 |
+
},
|
7655 |
+
{
|
7656 |
+
"epoch": 0.18696544645911733,
|
7657 |
+
"grad_norm": 1.809899926185608,
|
7658 |
+
"learning_rate": 1.2438588156921377e-05,
|
7659 |
+
"loss": 1.2139,
|
7660 |
+
"step": 1093
|
7661 |
+
},
|
7662 |
+
{
|
7663 |
+
"epoch": 0.1871365035921998,
|
7664 |
+
"grad_norm": 1.948121428489685,
|
7665 |
+
"learning_rate": 1.2365253965289892e-05,
|
7666 |
+
"loss": 1.2671,
|
7667 |
+
"step": 1094
|
7668 |
+
},
|
7669 |
+
{
|
7670 |
+
"epoch": 0.18730756072528223,
|
7671 |
+
"grad_norm": 2.194424629211426,
|
7672 |
+
"learning_rate": 1.2292106076364201e-05,
|
7673 |
+
"loss": 1.6186,
|
7674 |
+
"step": 1095
|
7675 |
+
},
|
7676 |
+
{
|
7677 |
+
"epoch": 0.1874786178583647,
|
7678 |
+
"grad_norm": 2.3575432300567627,
|
7679 |
+
"learning_rate": 1.2219144852247189e-05,
|
7680 |
+
"loss": 1.352,
|
7681 |
+
"step": 1096
|
7682 |
+
},
|
7683 |
+
{
|
7684 |
+
"epoch": 0.18764967499144714,
|
7685 |
+
"grad_norm": 2.576568841934204,
|
7686 |
+
"learning_rate": 1.2146370654117672e-05,
|
7687 |
+
"loss": 1.444,
|
7688 |
+
"step": 1097
|
7689 |
+
},
|
7690 |
+
{
|
7691 |
+
"epoch": 0.1878207321245296,
|
7692 |
+
"grad_norm": 3.0673329830169678,
|
7693 |
+
"learning_rate": 1.2073783842228647e-05,
|
7694 |
+
"loss": 1.6478,
|
7695 |
+
"step": 1098
|
7696 |
+
},
|
7697 |
+
{
|
7698 |
+
"epoch": 0.18799178925761204,
|
7699 |
+
"grad_norm": 3.8215575218200684,
|
7700 |
+
"learning_rate": 1.2001384775905517e-05,
|
7701 |
+
"loss": 1.9244,
|
7702 |
+
"step": 1099
|
7703 |
+
},
|
7704 |
+
{
|
7705 |
+
"epoch": 0.1881628463906945,
|
7706 |
+
"grad_norm": 3.5268056392669678,
|
7707 |
+
"learning_rate": 1.192917381354422e-05,
|
7708 |
+
"loss": 1.1216,
|
7709 |
+
"step": 1100
|
7710 |
+
},
|
7711 |
+
{
|
7712 |
+
"epoch": 0.18833390352377694,
|
7713 |
+
"grad_norm": 0.4581390619277954,
|
7714 |
+
"learning_rate": 1.1857151312609622e-05,
|
7715 |
+
"loss": 1.1206,
|
7716 |
+
"step": 1101
|
7717 |
+
},
|
7718 |
+
{
|
7719 |
+
"epoch": 0.1885049606568594,
|
7720 |
+
"grad_norm": 0.5717481374740601,
|
7721 |
+
"learning_rate": 1.1785317629633602e-05,
|
7722 |
+
"loss": 1.1875,
|
7723 |
+
"step": 1102
|
7724 |
+
},
|
7725 |
+
{
|
7726 |
+
"epoch": 0.18867601778994184,
|
7727 |
+
"grad_norm": 0.5163336992263794,
|
7728 |
+
"learning_rate": 1.1713673120213308e-05,
|
7729 |
+
"loss": 1.1397,
|
7730 |
+
"step": 1103
|
7731 |
+
},
|
7732 |
+
{
|
7733 |
+
"epoch": 0.1888470749230243,
|
7734 |
+
"grad_norm": 0.52544105052948,
|
7735 |
+
"learning_rate": 1.1642218139009487e-05,
|
7736 |
+
"loss": 1.1623,
|
7737 |
+
"step": 1104
|
7738 |
+
},
|
7739 |
+
{
|
7740 |
+
"epoch": 0.18901813205610674,
|
7741 |
+
"grad_norm": 0.520453155040741,
|
7742 |
+
"learning_rate": 1.1570953039744591e-05,
|
7743 |
+
"loss": 1.1673,
|
7744 |
+
"step": 1105
|
7745 |
+
},
|
7746 |
+
{
|
7747 |
+
"epoch": 0.1891891891891892,
|
7748 |
+
"grad_norm": 0.5462867021560669,
|
7749 |
+
"learning_rate": 1.1499878175201189e-05,
|
7750 |
+
"loss": 1.083,
|
7751 |
+
"step": 1106
|
7752 |
+
},
|
7753 |
+
{
|
7754 |
+
"epoch": 0.18936024632227164,
|
7755 |
+
"grad_norm": 0.5481846928596497,
|
7756 |
+
"learning_rate": 1.1428993897220103e-05,
|
7757 |
+
"loss": 1.1401,
|
7758 |
+
"step": 1107
|
7759 |
+
},
|
7760 |
+
{
|
7761 |
+
"epoch": 0.1895313034553541,
|
7762 |
+
"grad_norm": 0.5269501805305481,
|
7763 |
+
"learning_rate": 1.1358300556698643e-05,
|
7764 |
+
"loss": 1.2787,
|
7765 |
+
"step": 1108
|
7766 |
+
},
|
7767 |
+
{
|
7768 |
+
"epoch": 0.18970236058843654,
|
7769 |
+
"grad_norm": 0.5522426962852478,
|
7770 |
+
"learning_rate": 1.1287798503588992e-05,
|
7771 |
+
"loss": 1.2732,
|
7772 |
+
"step": 1109
|
7773 |
+
},
|
7774 |
+
{
|
7775 |
+
"epoch": 0.189873417721519,
|
7776 |
+
"grad_norm": 0.5485796332359314,
|
7777 |
+
"learning_rate": 1.1217488086896378e-05,
|
7778 |
+
"loss": 1.3491,
|
7779 |
+
"step": 1110
|
7780 |
+
},
|
7781 |
+
{
|
7782 |
+
"epoch": 0.19004447485460144,
|
7783 |
+
"grad_norm": 0.5756316184997559,
|
7784 |
+
"learning_rate": 1.1147369654677375e-05,
|
7785 |
+
"loss": 1.3665,
|
7786 |
+
"step": 1111
|
7787 |
+
},
|
7788 |
+
{
|
7789 |
+
"epoch": 0.1902155319876839,
|
7790 |
+
"grad_norm": 0.5507016777992249,
|
7791 |
+
"learning_rate": 1.1077443554038192e-05,
|
7792 |
+
"loss": 1.4174,
|
7793 |
+
"step": 1112
|
7794 |
+
},
|
7795 |
+
{
|
7796 |
+
"epoch": 0.19038658912076634,
|
7797 |
+
"grad_norm": 0.5559039115905762,
|
7798 |
+
"learning_rate": 1.1007710131132932e-05,
|
7799 |
+
"loss": 1.3666,
|
7800 |
+
"step": 1113
|
7801 |
+
},
|
7802 |
+
{
|
7803 |
+
"epoch": 0.19055764625384877,
|
7804 |
+
"grad_norm": 0.5902049541473389,
|
7805 |
+
"learning_rate": 1.0938169731161874e-05,
|
7806 |
+
"loss": 1.2883,
|
7807 |
+
"step": 1114
|
7808 |
+
},
|
7809 |
+
{
|
7810 |
+
"epoch": 0.19072870338693124,
|
7811 |
+
"grad_norm": 0.6311355233192444,
|
7812 |
+
"learning_rate": 1.0868822698369785e-05,
|
7813 |
+
"loss": 1.4242,
|
7814 |
+
"step": 1115
|
7815 |
+
},
|
7816 |
+
{
|
7817 |
+
"epoch": 0.19089976052001367,
|
7818 |
+
"grad_norm": 0.5926405787467957,
|
7819 |
+
"learning_rate": 1.0799669376044257e-05,
|
7820 |
+
"loss": 1.2911,
|
7821 |
+
"step": 1116
|
7822 |
+
},
|
7823 |
+
{
|
7824 |
+
"epoch": 0.19107081765309614,
|
7825 |
+
"grad_norm": 0.6782661080360413,
|
7826 |
+
"learning_rate": 1.0730710106513886e-05,
|
7827 |
+
"loss": 1.3423,
|
7828 |
+
"step": 1117
|
7829 |
+
},
|
7830 |
+
{
|
7831 |
+
"epoch": 0.19124187478617857,
|
7832 |
+
"grad_norm": 0.6430454850196838,
|
7833 |
+
"learning_rate": 1.0661945231146714e-05,
|
7834 |
+
"loss": 1.2164,
|
7835 |
+
"step": 1118
|
7836 |
+
},
|
7837 |
+
{
|
7838 |
+
"epoch": 0.19141293191926104,
|
7839 |
+
"grad_norm": 0.6636273264884949,
|
7840 |
+
"learning_rate": 1.0593375090348396e-05,
|
7841 |
+
"loss": 1.2754,
|
7842 |
+
"step": 1119
|
7843 |
+
},
|
7844 |
+
{
|
7845 |
+
"epoch": 0.19158398905234347,
|
7846 |
+
"grad_norm": 0.6418951153755188,
|
7847 |
+
"learning_rate": 1.05250000235607e-05,
|
7848 |
+
"loss": 1.1653,
|
7849 |
+
"step": 1120
|
7850 |
+
},
|
7851 |
+
{
|
7852 |
+
"epoch": 0.19175504618542594,
|
7853 |
+
"grad_norm": 0.7251722812652588,
|
7854 |
+
"learning_rate": 1.0456820369259667e-05,
|
7855 |
+
"loss": 1.1221,
|
7856 |
+
"step": 1121
|
7857 |
+
},
|
7858 |
+
{
|
7859 |
+
"epoch": 0.19192610331850837,
|
7860 |
+
"grad_norm": 0.6972784996032715,
|
7861 |
+
"learning_rate": 1.038883646495396e-05,
|
7862 |
+
"loss": 1.126,
|
7863 |
+
"step": 1122
|
7864 |
+
},
|
7865 |
+
{
|
7866 |
+
"epoch": 0.19209716045159084,
|
7867 |
+
"grad_norm": 0.7468768954277039,
|
7868 |
+
"learning_rate": 1.0321048647183278e-05,
|
7869 |
+
"loss": 1.1605,
|
7870 |
+
"step": 1123
|
7871 |
+
},
|
7872 |
+
{
|
7873 |
+
"epoch": 0.19226821758467327,
|
7874 |
+
"grad_norm": 0.7913068532943726,
|
7875 |
+
"learning_rate": 1.0253457251516618e-05,
|
7876 |
+
"loss": 1.1912,
|
7877 |
+
"step": 1124
|
7878 |
+
},
|
7879 |
+
{
|
7880 |
+
"epoch": 0.19243927471775574,
|
7881 |
+
"grad_norm": 0.7409959435462952,
|
7882 |
+
"learning_rate": 1.0186062612550617e-05,
|
7883 |
+
"loss": 1.2058,
|
7884 |
+
"step": 1125
|
7885 |
+
},
|
7886 |
+
{
|
7887 |
+
"epoch": 0.19261033185083817,
|
7888 |
+
"grad_norm": 0.7198361754417419,
|
7889 |
+
"learning_rate": 1.0118865063907934e-05,
|
7890 |
+
"loss": 1.0854,
|
7891 |
+
"step": 1126
|
7892 |
+
},
|
7893 |
+
{
|
7894 |
+
"epoch": 0.19278138898392064,
|
7895 |
+
"grad_norm": 0.7689154148101807,
|
7896 |
+
"learning_rate": 1.0051864938235572e-05,
|
7897 |
+
"loss": 1.0835,
|
7898 |
+
"step": 1127
|
7899 |
+
},
|
7900 |
+
{
|
7901 |
+
"epoch": 0.19295244611700307,
|
7902 |
+
"grad_norm": 0.8822424411773682,
|
7903 |
+
"learning_rate": 9.985062567203195e-06,
|
7904 |
+
"loss": 1.2179,
|
7905 |
+
"step": 1128
|
7906 |
+
},
|
7907 |
+
{
|
7908 |
+
"epoch": 0.19312350325008554,
|
7909 |
+
"grad_norm": 0.9055442810058594,
|
7910 |
+
"learning_rate": 9.91845828150157e-06,
|
7911 |
+
"loss": 1.2174,
|
7912 |
+
"step": 1129
|
7913 |
+
},
|
7914 |
+
{
|
7915 |
+
"epoch": 0.19329456038316797,
|
7916 |
+
"grad_norm": 0.8545695543289185,
|
7917 |
+
"learning_rate": 9.852052410840873e-06,
|
7918 |
+
"loss": 1.0819,
|
7919 |
+
"step": 1130
|
7920 |
+
},
|
7921 |
+
{
|
7922 |
+
"epoch": 0.19346561751625044,
|
7923 |
+
"grad_norm": 0.9511587619781494,
|
7924 |
+
"learning_rate": 9.78584528394907e-06,
|
7925 |
+
"loss": 1.1901,
|
7926 |
+
"step": 1131
|
7927 |
+
},
|
7928 |
+
{
|
7929 |
+
"epoch": 0.19363667464933287,
|
7930 |
+
"grad_norm": 1.0662909746170044,
|
7931 |
+
"learning_rate": 9.719837228570288e-06,
|
7932 |
+
"loss": 1.159,
|
7933 |
+
"step": 1132
|
7934 |
+
},
|
7935 |
+
{
|
7936 |
+
"epoch": 0.19380773178241534,
|
7937 |
+
"grad_norm": 1.1093270778656006,
|
7938 |
+
"learning_rate": 9.65402857146317e-06,
|
7939 |
+
"loss": 1.3553,
|
7940 |
+
"step": 1133
|
7941 |
+
},
|
7942 |
+
{
|
7943 |
+
"epoch": 0.19397878891549777,
|
7944 |
+
"grad_norm": 1.0241601467132568,
|
7945 |
+
"learning_rate": 9.588419638399305e-06,
|
7946 |
+
"loss": 1.1343,
|
7947 |
+
"step": 1134
|
7948 |
+
},
|
7949 |
+
{
|
7950 |
+
"epoch": 0.19414984604858024,
|
7951 |
+
"grad_norm": 1.1049851179122925,
|
7952 |
+
"learning_rate": 9.523010754161621e-06,
|
7953 |
+
"loss": 1.231,
|
7954 |
+
"step": 1135
|
7955 |
+
},
|
7956 |
+
{
|
7957 |
+
"epoch": 0.19432090318166267,
|
7958 |
+
"grad_norm": 1.1771842241287231,
|
7959 |
+
"learning_rate": 9.457802242542679e-06,
|
7960 |
+
"loss": 1.0904,
|
7961 |
+
"step": 1136
|
7962 |
+
},
|
7963 |
+
{
|
7964 |
+
"epoch": 0.19449196031474514,
|
7965 |
+
"grad_norm": 1.3713312149047852,
|
7966 |
+
"learning_rate": 9.392794426343188e-06,
|
7967 |
+
"loss": 1.3375,
|
7968 |
+
"step": 1137
|
7969 |
+
},
|
7970 |
+
{
|
7971 |
+
"epoch": 0.19466301744782757,
|
7972 |
+
"grad_norm": 1.395498514175415,
|
7973 |
+
"learning_rate": 9.327987627370333e-06,
|
7974 |
+
"loss": 1.2486,
|
7975 |
+
"step": 1138
|
7976 |
+
},
|
7977 |
+
{
|
7978 |
+
"epoch": 0.19483407458091,
|
7979 |
+
"grad_norm": 1.2937313318252563,
|
7980 |
+
"learning_rate": 9.263382166436224e-06,
|
7981 |
+
"loss": 1.3116,
|
7982 |
+
"step": 1139
|
7983 |
+
},
|
7984 |
+
{
|
7985 |
+
"epoch": 0.19500513171399247,
|
7986 |
+
"grad_norm": 1.4684338569641113,
|
7987 |
+
"learning_rate": 9.198978363356298e-06,
|
7988 |
+
"loss": 1.3451,
|
7989 |
+
"step": 1140
|
7990 |
+
},
|
7991 |
+
{
|
7992 |
+
"epoch": 0.1951761888470749,
|
7993 |
+
"grad_norm": 1.8545329570770264,
|
7994 |
+
"learning_rate": 9.134776536947687e-06,
|
7995 |
+
"loss": 1.3249,
|
7996 |
+
"step": 1141
|
7997 |
+
},
|
7998 |
+
{
|
7999 |
+
"epoch": 0.19534724598015737,
|
8000 |
+
"grad_norm": 2.3291664123535156,
|
8001 |
+
"learning_rate": 9.070777005027708e-06,
|
8002 |
+
"loss": 1.5226,
|
8003 |
+
"step": 1142
|
8004 |
+
},
|
8005 |
+
{
|
8006 |
+
"epoch": 0.1955183031132398,
|
8007 |
+
"grad_norm": 2.192391872406006,
|
8008 |
+
"learning_rate": 9.00698008441228e-06,
|
8009 |
+
"loss": 1.6328,
|
8010 |
+
"step": 1143
|
8011 |
+
},
|
8012 |
+
{
|
8013 |
+
"epoch": 0.19568936024632227,
|
8014 |
+
"grad_norm": 1.8201500177383423,
|
8015 |
+
"learning_rate": 8.943386090914308e-06,
|
8016 |
+
"loss": 1.1462,
|
8017 |
+
"step": 1144
|
8018 |
+
},
|
8019 |
+
{
|
8020 |
+
"epoch": 0.1958604173794047,
|
8021 |
+
"grad_norm": 2.1104891300201416,
|
8022 |
+
"learning_rate": 8.879995339342167e-06,
|
8023 |
+
"loss": 1.1711,
|
8024 |
+
"step": 1145
|
8025 |
+
},
|
8026 |
+
{
|
8027 |
+
"epoch": 0.19603147451248717,
|
8028 |
+
"grad_norm": 2.287696123123169,
|
8029 |
+
"learning_rate": 8.816808143498134e-06,
|
8030 |
+
"loss": 1.1776,
|
8031 |
+
"step": 1146
|
8032 |
+
},
|
8033 |
+
{
|
8034 |
+
"epoch": 0.1962025316455696,
|
8035 |
+
"grad_norm": 2.593168020248413,
|
8036 |
+
"learning_rate": 8.753824816176787e-06,
|
8037 |
+
"loss": 1.4058,
|
8038 |
+
"step": 1147
|
8039 |
+
},
|
8040 |
+
{
|
8041 |
+
"epoch": 0.19637358877865208,
|
8042 |
+
"grad_norm": 2.805091142654419,
|
8043 |
+
"learning_rate": 8.691045669163523e-06,
|
8044 |
+
"loss": 1.4726,
|
8045 |
+
"step": 1148
|
8046 |
+
},
|
8047 |
+
{
|
8048 |
+
"epoch": 0.1965446459117345,
|
8049 |
+
"grad_norm": 2.7710683345794678,
|
8050 |
+
"learning_rate": 8.628471013233036e-06,
|
8051 |
+
"loss": 1.4785,
|
8052 |
+
"step": 1149
|
8053 |
+
},
|
8054 |
+
{
|
8055 |
+
"epoch": 0.19671570304481698,
|
8056 |
+
"grad_norm": 4.310181617736816,
|
8057 |
+
"learning_rate": 8.566101158147643e-06,
|
8058 |
+
"loss": 2.0281,
|
8059 |
+
"step": 1150
|
8060 |
+
},
|
8061 |
+
{
|
8062 |
+
"epoch": 0.1968867601778994,
|
8063 |
+
"grad_norm": 0.5353257656097412,
|
8064 |
+
"learning_rate": 8.5039364126559e-06,
|
8065 |
+
"loss": 1.3306,
|
8066 |
+
"step": 1151
|
8067 |
+
},
|
8068 |
+
{
|
8069 |
+
"epoch": 0.19705781731098188,
|
8070 |
+
"grad_norm": 0.5203657746315002,
|
8071 |
+
"learning_rate": 8.441977084490976e-06,
|
8072 |
+
"loss": 1.2153,
|
8073 |
+
"step": 1152
|
8074 |
+
},
|
8075 |
+
{
|
8076 |
+
"epoch": 0.1972288744440643,
|
8077 |
+
"grad_norm": 0.5467506051063538,
|
8078 |
+
"learning_rate": 8.380223480369182e-06,
|
8079 |
+
"loss": 1.2193,
|
8080 |
+
"step": 1153
|
8081 |
+
},
|
8082 |
+
{
|
8083 |
+
"epoch": 0.19739993157714678,
|
8084 |
+
"grad_norm": 0.5195339918136597,
|
8085 |
+
"learning_rate": 8.31867590598847e-06,
|
8086 |
+
"loss": 1.2195,
|
8087 |
+
"step": 1154
|
8088 |
+
},
|
8089 |
+
{
|
8090 |
+
"epoch": 0.1975709887102292,
|
8091 |
+
"grad_norm": 0.5019032955169678,
|
8092 |
+
"learning_rate": 8.257334666026817e-06,
|
8093 |
+
"loss": 1.1177,
|
8094 |
+
"step": 1155
|
8095 |
+
},
|
8096 |
+
{
|
8097 |
+
"epoch": 0.19774204584331168,
|
8098 |
+
"grad_norm": 0.5388235449790955,
|
8099 |
+
"learning_rate": 8.196200064140836e-06,
|
8100 |
+
"loss": 1.2386,
|
8101 |
+
"step": 1156
|
8102 |
+
},
|
8103 |
+
{
|
8104 |
+
"epoch": 0.1979131029763941,
|
8105 |
+
"grad_norm": 0.5507181882858276,
|
8106 |
+
"learning_rate": 8.135272402964212e-06,
|
8107 |
+
"loss": 1.3414,
|
8108 |
+
"step": 1157
|
8109 |
+
},
|
8110 |
+
{
|
8111 |
+
"epoch": 0.19808416010947658,
|
8112 |
+
"grad_norm": 0.5381900668144226,
|
8113 |
+
"learning_rate": 8.074551984106198e-06,
|
8114 |
+
"loss": 1.2364,
|
8115 |
+
"step": 1158
|
8116 |
+
},
|
8117 |
+
{
|
8118 |
+
"epoch": 0.198255217242559,
|
8119 |
+
"grad_norm": 0.5386906266212463,
|
8120 |
+
"learning_rate": 8.014039108150168e-06,
|
8121 |
+
"loss": 1.1367,
|
8122 |
+
"step": 1159
|
8123 |
+
},
|
8124 |
+
{
|
8125 |
+
"epoch": 0.19842627437564148,
|
8126 |
+
"grad_norm": 0.5798367261886597,
|
8127 |
+
"learning_rate": 7.953734074652081e-06,
|
8128 |
+
"loss": 1.3284,
|
8129 |
+
"step": 1160
|
8130 |
+
},
|
8131 |
+
{
|
8132 |
+
"epoch": 0.1985973315087239,
|
8133 |
+
"grad_norm": 0.5907285213470459,
|
8134 |
+
"learning_rate": 7.893637182139002e-06,
|
8135 |
+
"loss": 1.3656,
|
8136 |
+
"step": 1161
|
8137 |
+
},
|
8138 |
+
{
|
8139 |
+
"epoch": 0.19876838864180638,
|
8140 |
+
"grad_norm": 0.576878011226654,
|
8141 |
+
"learning_rate": 7.83374872810767e-06,
|
8142 |
+
"loss": 1.3606,
|
8143 |
+
"step": 1162
|
8144 |
+
},
|
8145 |
+
{
|
8146 |
+
"epoch": 0.1989394457748888,
|
8147 |
+
"grad_norm": 0.6249831318855286,
|
8148 |
+
"learning_rate": 7.774069009022982e-06,
|
8149 |
+
"loss": 1.3086,
|
8150 |
+
"step": 1163
|
8151 |
+
},
|
8152 |
+
{
|
8153 |
+
"epoch": 0.19911050290797128,
|
8154 |
+
"grad_norm": 0.5776302814483643,
|
8155 |
+
"learning_rate": 7.714598320316524e-06,
|
8156 |
+
"loss": 1.4029,
|
8157 |
+
"step": 1164
|
8158 |
+
},
|
8159 |
+
{
|
8160 |
+
"epoch": 0.1992815600410537,
|
8161 |
+
"grad_norm": 0.6269991993904114,
|
8162 |
+
"learning_rate": 7.655336956385156e-06,
|
8163 |
+
"loss": 1.4726,
|
8164 |
+
"step": 1165
|
8165 |
+
},
|
8166 |
+
{
|
8167 |
+
"epoch": 0.19945261717413615,
|
8168 |
+
"grad_norm": 0.6286412477493286,
|
8169 |
+
"learning_rate": 7.596285210589471e-06,
|
8170 |
+
"loss": 1.3788,
|
8171 |
+
"step": 1166
|
8172 |
+
},
|
8173 |
+
{
|
8174 |
+
"epoch": 0.1996236743072186,
|
8175 |
+
"grad_norm": 0.6275160908699036,
|
8176 |
+
"learning_rate": 7.537443375252423e-06,
|
8177 |
+
"loss": 1.3076,
|
8178 |
+
"step": 1167
|
8179 |
+
},
|
8180 |
+
{
|
8181 |
+
"epoch": 0.19979473144030105,
|
8182 |
+
"grad_norm": 0.6251853704452515,
|
8183 |
+
"learning_rate": 7.4788117416578765e-06,
|
8184 |
+
"loss": 1.2833,
|
8185 |
+
"step": 1168
|
8186 |
+
},
|
8187 |
+
{
|
8188 |
+
"epoch": 0.1999657885733835,
|
8189 |
+
"grad_norm": 0.6521373987197876,
|
8190 |
+
"learning_rate": 7.420390600049076e-06,
|
8191 |
+
"loss": 1.1192,
|
8192 |
+
"step": 1169
|
8193 |
+
},
|
8194 |
+
{
|
8195 |
+
"epoch": 0.20013684570646595,
|
8196 |
+
"grad_norm": 0.6653075814247131,
|
8197 |
+
"learning_rate": 7.362180239627314e-06,
|
8198 |
+
"loss": 1.3508,
|
8199 |
+
"step": 1170
|
8200 |
+
},
|
8201 |
+
{
|
8202 |
+
"epoch": 0.2003079028395484,
|
8203 |
+
"grad_norm": 0.691904604434967,
|
8204 |
+
"learning_rate": 7.304180948550454e-06,
|
8205 |
+
"loss": 1.1852,
|
8206 |
+
"step": 1171
|
8207 |
+
},
|
8208 |
+
{
|
8209 |
+
"epoch": 0.20047895997263085,
|
8210 |
+
"grad_norm": 0.6476536989212036,
|
8211 |
+
"learning_rate": 7.246393013931463e-06,
|
8212 |
+
"loss": 1.2014,
|
8213 |
+
"step": 1172
|
8214 |
+
},
|
8215 |
+
{
|
8216 |
+
"epoch": 0.2006500171057133,
|
8217 |
+
"grad_norm": 0.7105785608291626,
|
8218 |
+
"learning_rate": 7.188816721837116e-06,
|
8219 |
+
"loss": 1.2,
|
8220 |
+
"step": 1173
|
8221 |
+
},
|
8222 |
+
{
|
8223 |
+
"epoch": 0.20082107423879575,
|
8224 |
+
"grad_norm": 0.701448380947113,
|
8225 |
+
"learning_rate": 7.131452357286411e-06,
|
8226 |
+
"loss": 1.1777,
|
8227 |
+
"step": 1174
|
8228 |
+
},
|
8229 |
+
{
|
8230 |
+
"epoch": 0.2009921313718782,
|
8231 |
+
"grad_norm": 0.7755523324012756,
|
8232 |
+
"learning_rate": 7.074300204249296e-06,
|
8233 |
+
"loss": 1.1084,
|
8234 |
+
"step": 1175
|
8235 |
+
},
|
8236 |
+
{
|
8237 |
+
"epoch": 0.20116318850496065,
|
8238 |
+
"grad_norm": 0.7777191400527954,
|
8239 |
+
"learning_rate": 7.01736054564519e-06,
|
8240 |
+
"loss": 1.1224,
|
8241 |
+
"step": 1176
|
8242 |
+
},
|
8243 |
+
{
|
8244 |
+
"epoch": 0.2013342456380431,
|
8245 |
+
"grad_norm": 0.8104934692382812,
|
8246 |
+
"learning_rate": 6.960633663341615e-06,
|
8247 |
+
"loss": 1.146,
|
8248 |
+
"step": 1177
|
8249 |
+
},
|
8250 |
+
{
|
8251 |
+
"epoch": 0.20150530277112555,
|
8252 |
+
"grad_norm": 0.8348276615142822,
|
8253 |
+
"learning_rate": 6.904119838152789e-06,
|
8254 |
+
"loss": 1.0867,
|
8255 |
+
"step": 1178
|
8256 |
+
},
|
8257 |
+
{
|
8258 |
+
"epoch": 0.201676359904208,
|
8259 |
+
"grad_norm": 0.8821502327919006,
|
8260 |
+
"learning_rate": 6.847819349838247e-06,
|
8261 |
+
"loss": 1.1303,
|
8262 |
+
"step": 1179
|
8263 |
+
},
|
8264 |
+
{
|
8265 |
+
"epoch": 0.20184741703729045,
|
8266 |
+
"grad_norm": 0.930512011051178,
|
8267 |
+
"learning_rate": 6.791732477101398e-06,
|
8268 |
+
"loss": 1.1057,
|
8269 |
+
"step": 1180
|
8270 |
+
},
|
8271 |
+
{
|
8272 |
+
"epoch": 0.2020184741703729,
|
8273 |
+
"grad_norm": 0.9049108624458313,
|
8274 |
+
"learning_rate": 6.7358594975882535e-06,
|
8275 |
+
"loss": 1.204,
|
8276 |
+
"step": 1181
|
8277 |
+
},
|
8278 |
+
{
|
8279 |
+
"epoch": 0.20218953130345535,
|
8280 |
+
"grad_norm": 0.9623181223869324,
|
8281 |
+
"learning_rate": 6.680200687885957e-06,
|
8282 |
+
"loss": 1.3074,
|
8283 |
+
"step": 1182
|
8284 |
+
},
|
8285 |
+
{
|
8286 |
+
"epoch": 0.2023605884365378,
|
8287 |
+
"grad_norm": 1.026506781578064,
|
8288 |
+
"learning_rate": 6.624756323521459e-06,
|
8289 |
+
"loss": 1.445,
|
8290 |
+
"step": 1183
|
8291 |
+
},
|
8292 |
+
{
|
8293 |
+
"epoch": 0.20253164556962025,
|
8294 |
+
"grad_norm": 0.9811666011810303,
|
8295 |
+
"learning_rate": 6.569526678960164e-06,
|
8296 |
+
"loss": 1.1632,
|
8297 |
+
"step": 1184
|
8298 |
+
},
|
8299 |
+
{
|
8300 |
+
"epoch": 0.20270270270270271,
|
8301 |
+
"grad_norm": 1.0132756233215332,
|
8302 |
+
"learning_rate": 6.514512027604508e-06,
|
8303 |
+
"loss": 1.1912,
|
8304 |
+
"step": 1185
|
8305 |
+
},
|
8306 |
+
{
|
8307 |
+
"epoch": 0.20287375983578515,
|
8308 |
+
"grad_norm": 1.1950069665908813,
|
8309 |
+
"learning_rate": 6.459712641792676e-06,
|
8310 |
+
"loss": 1.3509,
|
8311 |
+
"step": 1186
|
8312 |
+
},
|
8313 |
+
{
|
8314 |
+
"epoch": 0.20304481696886761,
|
8315 |
+
"grad_norm": 1.2882367372512817,
|
8316 |
+
"learning_rate": 6.405128792797255e-06,
|
8317 |
+
"loss": 1.1653,
|
8318 |
+
"step": 1187
|
8319 |
+
},
|
8320 |
+
{
|
8321 |
+
"epoch": 0.20321587410195005,
|
8322 |
+
"grad_norm": 1.1003586053848267,
|
8323 |
+
"learning_rate": 6.3507607508238085e-06,
|
8324 |
+
"loss": 1.0812,
|
8325 |
+
"step": 1188
|
8326 |
+
},
|
8327 |
+
{
|
8328 |
+
"epoch": 0.20338693123503251,
|
8329 |
+
"grad_norm": 1.3947727680206299,
|
8330 |
+
"learning_rate": 6.296608785009633e-06,
|
8331 |
+
"loss": 1.1218,
|
8332 |
+
"step": 1189
|
8333 |
+
},
|
8334 |
+
{
|
8335 |
+
"epoch": 0.20355798836811495,
|
8336 |
+
"grad_norm": 1.2513878345489502,
|
8337 |
+
"learning_rate": 6.242673163422374e-06,
|
8338 |
+
"loss": 1.1447,
|
8339 |
+
"step": 1190
|
8340 |
+
},
|
8341 |
+
{
|
8342 |
+
"epoch": 0.2037290455011974,
|
8343 |
+
"grad_norm": 1.4557979106903076,
|
8344 |
+
"learning_rate": 6.1889541530586955e-06,
|
8345 |
+
"loss": 1.395,
|
8346 |
+
"step": 1191
|
8347 |
+
},
|
8348 |
+
{
|
8349 |
+
"epoch": 0.20390010263427985,
|
8350 |
+
"grad_norm": 1.6186139583587646,
|
8351 |
+
"learning_rate": 6.1354520198430075e-06,
|
8352 |
+
"loss": 1.3386,
|
8353 |
+
"step": 1192
|
8354 |
+
},
|
8355 |
+
{
|
8356 |
+
"epoch": 0.2040711597673623,
|
8357 |
+
"grad_norm": 1.6991312503814697,
|
8358 |
+
"learning_rate": 6.0821670286261016e-06,
|
8359 |
+
"loss": 1.4036,
|
8360 |
+
"step": 1193
|
8361 |
+
},
|
8362 |
+
{
|
8363 |
+
"epoch": 0.20424221690044475,
|
8364 |
+
"grad_norm": 1.7909969091415405,
|
8365 |
+
"learning_rate": 6.0290994431838385e-06,
|
8366 |
+
"loss": 1.4398,
|
8367 |
+
"step": 1194
|
8368 |
+
},
|
8369 |
+
{
|
8370 |
+
"epoch": 0.2044132740335272,
|
8371 |
+
"grad_norm": 2.124344825744629,
|
8372 |
+
"learning_rate": 5.9762495262158764e-06,
|
8373 |
+
"loss": 1.1204,
|
8374 |
+
"step": 1195
|
8375 |
+
},
|
8376 |
+
{
|
8377 |
+
"epoch": 0.20458433116660965,
|
8378 |
+
"grad_norm": 2.179216146469116,
|
8379 |
+
"learning_rate": 5.923617539344351e-06,
|
8380 |
+
"loss": 1.2495,
|
8381 |
+
"step": 1196
|
8382 |
+
},
|
8383 |
+
{
|
8384 |
+
"epoch": 0.2047553882996921,
|
8385 |
+
"grad_norm": 2.3889334201812744,
|
8386 |
+
"learning_rate": 5.87120374311258e-06,
|
8387 |
+
"loss": 1.4048,
|
8388 |
+
"step": 1197
|
8389 |
+
},
|
8390 |
+
{
|
8391 |
+
"epoch": 0.20492644543277455,
|
8392 |
+
"grad_norm": 2.6185591220855713,
|
8393 |
+
"learning_rate": 5.819008396983783e-06,
|
8394 |
+
"loss": 1.4209,
|
8395 |
+
"step": 1198
|
8396 |
+
},
|
8397 |
+
{
|
8398 |
+
"epoch": 0.205097502565857,
|
8399 |
+
"grad_norm": 2.963503360748291,
|
8400 |
+
"learning_rate": 5.767031759339758e-06,
|
8401 |
+
"loss": 1.4406,
|
8402 |
+
"step": 1199
|
8403 |
+
},
|
8404 |
+
{
|
8405 |
+
"epoch": 0.20526855969893945,
|
8406 |
+
"grad_norm": 3.8163952827453613,
|
8407 |
+
"learning_rate": 5.715274087479672e-06,
|
8408 |
+
"loss": 1.4222,
|
8409 |
+
"step": 1200
|
8410 |
+
},
|
8411 |
+
{
|
8412 |
+
"epoch": 0.2054396168320219,
|
8413 |
+
"grad_norm": 0.49252671003341675,
|
8414 |
+
"learning_rate": 5.663735637618728e-06,
|
8415 |
+
"loss": 1.1811,
|
8416 |
+
"step": 1201
|
8417 |
+
},
|
8418 |
+
{
|
8419 |
+
"epoch": 0.20561067396510435,
|
8420 |
+
"grad_norm": 0.5381333827972412,
|
8421 |
+
"learning_rate": 5.6124166648869326e-06,
|
8422 |
+
"loss": 1.1084,
|
8423 |
+
"step": 1202
|
8424 |
+
},
|
8425 |
+
{
|
8426 |
+
"epoch": 0.2057817310981868,
|
8427 |
+
"grad_norm": 0.4863746166229248,
|
8428 |
+
"learning_rate": 5.5613174233278e-06,
|
8429 |
+
"loss": 1.2013,
|
8430 |
+
"step": 1203
|
8431 |
+
},
|
8432 |
+
{
|
8433 |
+
"epoch": 0.20595278823126925,
|
8434 |
+
"grad_norm": 0.5375758409500122,
|
8435 |
+
"learning_rate": 5.510438165897131e-06,
|
8436 |
+
"loss": 1.178,
|
8437 |
+
"step": 1204
|
8438 |
+
},
|
8439 |
+
{
|
8440 |
+
"epoch": 0.2061238453643517,
|
8441 |
+
"grad_norm": 0.5861601829528809,
|
8442 |
+
"learning_rate": 5.459779144461713e-06,
|
8443 |
+
"loss": 1.2813,
|
8444 |
+
"step": 1205
|
8445 |
+
},
|
8446 |
+
{
|
8447 |
+
"epoch": 0.20629490249743415,
|
8448 |
+
"grad_norm": 0.5158637166023254,
|
8449 |
+
"learning_rate": 5.409340609798147e-06,
|
8450 |
+
"loss": 1.276,
|
8451 |
+
"step": 1206
|
8452 |
+
},
|
8453 |
+
{
|
8454 |
+
"epoch": 0.2064659596305166,
|
8455 |
+
"grad_norm": 0.5211811065673828,
|
8456 |
+
"learning_rate": 5.359122811591533e-06,
|
8457 |
+
"loss": 1.2581,
|
8458 |
+
"step": 1207
|
8459 |
+
},
|
8460 |
+
{
|
8461 |
+
"epoch": 0.20663701676359905,
|
8462 |
+
"grad_norm": 0.5310105681419373,
|
8463 |
+
"learning_rate": 5.30912599843425e-06,
|
8464 |
+
"loss": 1.1793,
|
8465 |
+
"step": 1208
|
8466 |
+
},
|
8467 |
+
{
|
8468 |
+
"epoch": 0.2068080738966815,
|
8469 |
+
"grad_norm": 0.5542389154434204,
|
8470 |
+
"learning_rate": 5.259350417824776e-06,
|
8471 |
+
"loss": 1.2673,
|
8472 |
+
"step": 1209
|
8473 |
+
},
|
8474 |
+
{
|
8475 |
+
"epoch": 0.20697913102976395,
|
8476 |
+
"grad_norm": 0.5369300842285156,
|
8477 |
+
"learning_rate": 5.209796316166376e-06,
|
8478 |
+
"loss": 1.2781,
|
8479 |
+
"step": 1210
|
8480 |
+
},
|
8481 |
+
{
|
8482 |
+
"epoch": 0.2071501881628464,
|
8483 |
+
"grad_norm": 0.5323848128318787,
|
8484 |
+
"learning_rate": 5.160463938765986e-06,
|
8485 |
+
"loss": 1.3192,
|
8486 |
+
"step": 1211
|
8487 |
+
},
|
8488 |
+
{
|
8489 |
+
"epoch": 0.20732124529592885,
|
8490 |
+
"grad_norm": 0.5648958086967468,
|
8491 |
+
"learning_rate": 5.111353529832913e-06,
|
8492 |
+
"loss": 1.3,
|
8493 |
+
"step": 1212
|
8494 |
+
},
|
8495 |
+
{
|
8496 |
+
"epoch": 0.2074923024290113,
|
8497 |
+
"grad_norm": 0.5956771373748779,
|
8498 |
+
"learning_rate": 5.062465332477651e-06,
|
8499 |
+
"loss": 1.2465,
|
8500 |
+
"step": 1213
|
8501 |
+
},
|
8502 |
+
{
|
8503 |
+
"epoch": 0.20766335956209375,
|
8504 |
+
"grad_norm": 0.5839723348617554,
|
8505 |
+
"learning_rate": 5.013799588710694e-06,
|
8506 |
+
"loss": 1.316,
|
8507 |
+
"step": 1214
|
8508 |
+
},
|
8509 |
+
{
|
8510 |
+
"epoch": 0.2078344166951762,
|
8511 |
+
"grad_norm": 0.5952252149581909,
|
8512 |
+
"learning_rate": 4.9653565394413295e-06,
|
8513 |
+
"loss": 1.3178,
|
8514 |
+
"step": 1215
|
8515 |
+
},
|
8516 |
+
{
|
8517 |
+
"epoch": 0.20800547382825862,
|
8518 |
+
"grad_norm": 0.6331438422203064,
|
8519 |
+
"learning_rate": 4.917136424476437e-06,
|
8520 |
+
"loss": 1.4325,
|
8521 |
+
"step": 1216
|
8522 |
+
},
|
8523 |
+
{
|
8524 |
+
"epoch": 0.2081765309613411,
|
8525 |
+
"grad_norm": 0.6285284161567688,
|
8526 |
+
"learning_rate": 4.869139482519325e-06,
|
8527 |
+
"loss": 1.2226,
|
8528 |
+
"step": 1217
|
8529 |
+
},
|
8530 |
+
{
|
8531 |
+
"epoch": 0.20834758809442352,
|
8532 |
+
"grad_norm": 0.6116199493408203,
|
8533 |
+
"learning_rate": 4.821365951168516e-06,
|
8534 |
+
"loss": 1.1924,
|
8535 |
+
"step": 1218
|
8536 |
+
},
|
8537 |
+
{
|
8538 |
+
"epoch": 0.208518645227506,
|
8539 |
+
"grad_norm": 0.5923437476158142,
|
8540 |
+
"learning_rate": 4.77381606691657e-06,
|
8541 |
+
"loss": 1.2561,
|
8542 |
+
"step": 1219
|
8543 |
+
},
|
8544 |
+
{
|
8545 |
+
"epoch": 0.20868970236058842,
|
8546 |
+
"grad_norm": 0.6754952073097229,
|
8547 |
+
"learning_rate": 4.726490065148964e-06,
|
8548 |
+
"loss": 1.3237,
|
8549 |
+
"step": 1220
|
8550 |
+
},
|
8551 |
+
{
|
8552 |
+
"epoch": 0.2088607594936709,
|
8553 |
+
"grad_norm": 0.6788840293884277,
|
8554 |
+
"learning_rate": 4.6793881801428796e-06,
|
8555 |
+
"loss": 1.3166,
|
8556 |
+
"step": 1221
|
8557 |
+
},
|
8558 |
+
{
|
8559 |
+
"epoch": 0.20903181662675333,
|
8560 |
+
"grad_norm": 0.7086222767829895,
|
8561 |
+
"learning_rate": 4.632510645066046e-06,
|
8562 |
+
"loss": 1.2834,
|
8563 |
+
"step": 1222
|
8564 |
+
},
|
8565 |
+
{
|
8566 |
+
"epoch": 0.2092028737598358,
|
8567 |
+
"grad_norm": 0.7086116671562195,
|
8568 |
+
"learning_rate": 4.585857691975626e-06,
|
8569 |
+
"loss": 1.1047,
|
8570 |
+
"step": 1223
|
8571 |
+
},
|
8572 |
+
{
|
8573 |
+
"epoch": 0.20937393089291823,
|
8574 |
+
"grad_norm": 0.7346565127372742,
|
8575 |
+
"learning_rate": 4.539429551816981e-06,
|
8576 |
+
"loss": 1.2146,
|
8577 |
+
"step": 1224
|
8578 |
+
},
|
8579 |
+
{
|
8580 |
+
"epoch": 0.2095449880260007,
|
8581 |
+
"grad_norm": 0.7705104351043701,
|
8582 |
+
"learning_rate": 4.493226454422661e-06,
|
8583 |
+
"loss": 1.2224,
|
8584 |
+
"step": 1225
|
8585 |
+
},
|
8586 |
+
{
|
8587 |
+
"epoch": 0.20971604515908313,
|
8588 |
+
"grad_norm": 0.6990749835968018,
|
8589 |
+
"learning_rate": 4.447248628511136e-06,
|
8590 |
+
"loss": 1.0229,
|
8591 |
+
"step": 1226
|
8592 |
+
},
|
8593 |
+
{
|
8594 |
+
"epoch": 0.2098871022921656,
|
8595 |
+
"grad_norm": 0.7617723345756531,
|
8596 |
+
"learning_rate": 4.40149630168572e-06,
|
8597 |
+
"loss": 1.0909,
|
8598 |
+
"step": 1227
|
8599 |
+
},
|
8600 |
+
{
|
8601 |
+
"epoch": 0.21005815942524803,
|
8602 |
+
"grad_norm": 0.7978994846343994,
|
8603 |
+
"learning_rate": 4.35596970043346e-06,
|
8604 |
+
"loss": 1.1296,
|
8605 |
+
"step": 1228
|
8606 |
+
},
|
8607 |
+
{
|
8608 |
+
"epoch": 0.2102292165583305,
|
8609 |
+
"grad_norm": 0.8365187644958496,
|
8610 |
+
"learning_rate": 4.310669050123994e-06,
|
8611 |
+
"loss": 1.1557,
|
8612 |
+
"step": 1229
|
8613 |
+
},
|
8614 |
+
{
|
8615 |
+
"epoch": 0.21040027369141293,
|
8616 |
+
"grad_norm": 0.9419746994972229,
|
8617 |
+
"learning_rate": 4.265594575008441e-06,
|
8618 |
+
"loss": 1.292,
|
8619 |
+
"step": 1230
|
8620 |
+
},
|
8621 |
+
{
|
8622 |
+
"epoch": 0.2105713308244954,
|
8623 |
+
"grad_norm": 0.9970366358757019,
|
8624 |
+
"learning_rate": 4.220746498218286e-06,
|
8625 |
+
"loss": 1.2974,
|
8626 |
+
"step": 1231
|
8627 |
+
},
|
8628 |
+
{
|
8629 |
+
"epoch": 0.21074238795757783,
|
8630 |
+
"grad_norm": 0.9173380136489868,
|
8631 |
+
"learning_rate": 4.176125041764262e-06,
|
8632 |
+
"loss": 1.0588,
|
8633 |
+
"step": 1232
|
8634 |
+
},
|
8635 |
+
{
|
8636 |
+
"epoch": 0.2109134450906603,
|
8637 |
+
"grad_norm": 0.9308356046676636,
|
8638 |
+
"learning_rate": 4.131730426535285e-06,
|
8639 |
+
"loss": 1.1292,
|
8640 |
+
"step": 1233
|
8641 |
+
},
|
8642 |
+
{
|
8643 |
+
"epoch": 0.21108450222374273,
|
8644 |
+
"grad_norm": 1.0378395318984985,
|
8645 |
+
"learning_rate": 4.08756287229734e-06,
|
8646 |
+
"loss": 1.0853,
|
8647 |
+
"step": 1234
|
8648 |
+
},
|
8649 |
+
{
|
8650 |
+
"epoch": 0.2112555593568252,
|
8651 |
+
"grad_norm": 1.171376347541809,
|
8652 |
+
"learning_rate": 4.043622597692387e-06,
|
8653 |
+
"loss": 1.222,
|
8654 |
+
"step": 1235
|
8655 |
+
},
|
8656 |
+
{
|
8657 |
+
"epoch": 0.21142661648990763,
|
8658 |
+
"grad_norm": 1.2011628150939941,
|
8659 |
+
"learning_rate": 3.999909820237302e-06,
|
8660 |
+
"loss": 1.2648,
|
8661 |
+
"step": 1236
|
8662 |
+
},
|
8663 |
+
{
|
8664 |
+
"epoch": 0.2115976736229901,
|
8665 |
+
"grad_norm": 1.2246147394180298,
|
8666 |
+
"learning_rate": 3.956424756322785e-06,
|
8667 |
+
"loss": 1.4853,
|
8668 |
+
"step": 1237
|
8669 |
+
},
|
8670 |
+
{
|
8671 |
+
"epoch": 0.21176873075607253,
|
8672 |
+
"grad_norm": 1.2781755924224854,
|
8673 |
+
"learning_rate": 3.913167621212255e-06,
|
8674 |
+
"loss": 1.2316,
|
8675 |
+
"step": 1238
|
8676 |
+
},
|
8677 |
+
{
|
8678 |
+
"epoch": 0.211939787889155,
|
8679 |
+
"grad_norm": 1.4460945129394531,
|
8680 |
+
"learning_rate": 3.870138629040848e-06,
|
8681 |
+
"loss": 1.0442,
|
8682 |
+
"step": 1239
|
8683 |
+
},
|
8684 |
+
{
|
8685 |
+
"epoch": 0.21211084502223743,
|
8686 |
+
"grad_norm": 1.3829147815704346,
|
8687 |
+
"learning_rate": 3.827337992814356e-06,
|
8688 |
+
"loss": 1.3402,
|
8689 |
+
"step": 1240
|
8690 |
+
},
|
8691 |
+
{
|
8692 |
+
"epoch": 0.2122819021553199,
|
8693 |
+
"grad_norm": 1.6668599843978882,
|
8694 |
+
"learning_rate": 3.7847659244080736e-06,
|
8695 |
+
"loss": 1.277,
|
8696 |
+
"step": 1241
|
8697 |
+
},
|
8698 |
+
{
|
8699 |
+
"epoch": 0.21245295928840233,
|
8700 |
+
"grad_norm": 1.8887028694152832,
|
8701 |
+
"learning_rate": 3.742422634565884e-06,
|
8702 |
+
"loss": 1.3339,
|
8703 |
+
"step": 1242
|
8704 |
+
},
|
8705 |
+
{
|
8706 |
+
"epoch": 0.21262401642148476,
|
8707 |
+
"grad_norm": 1.7078804969787598,
|
8708 |
+
"learning_rate": 3.7003083328990993e-06,
|
8709 |
+
"loss": 1.2418,
|
8710 |
+
"step": 1243
|
8711 |
+
},
|
8712 |
+
{
|
8713 |
+
"epoch": 0.21279507355456723,
|
8714 |
+
"grad_norm": 1.9773505926132202,
|
8715 |
+
"learning_rate": 3.658423227885527e-06,
|
8716 |
+
"loss": 1.2476,
|
8717 |
+
"step": 1244
|
8718 |
+
},
|
8719 |
+
{
|
8720 |
+
"epoch": 0.21296613068764966,
|
8721 |
+
"grad_norm": 1.880808711051941,
|
8722 |
+
"learning_rate": 3.616767526868353e-06,
|
8723 |
+
"loss": 1.0132,
|
8724 |
+
"step": 1245
|
8725 |
+
},
|
8726 |
+
{
|
8727 |
+
"epoch": 0.21313718782073213,
|
8728 |
+
"grad_norm": 3.0526063442230225,
|
8729 |
+
"learning_rate": 3.575341436055141e-06,
|
8730 |
+
"loss": 1.6674,
|
8731 |
+
"step": 1246
|
8732 |
+
},
|
8733 |
+
{
|
8734 |
+
"epoch": 0.21330824495381456,
|
8735 |
+
"grad_norm": 2.4462673664093018,
|
8736 |
+
"learning_rate": 3.534145160516844e-06,
|
8737 |
+
"loss": 1.5029,
|
8738 |
+
"step": 1247
|
8739 |
+
},
|
8740 |
+
{
|
8741 |
+
"epoch": 0.21347930208689703,
|
8742 |
+
"grad_norm": 2.669107437133789,
|
8743 |
+
"learning_rate": 3.49317890418675e-06,
|
8744 |
+
"loss": 1.3254,
|
8745 |
+
"step": 1248
|
8746 |
+
},
|
8747 |
+
{
|
8748 |
+
"epoch": 0.21365035921997946,
|
8749 |
+
"grad_norm": 3.09505558013916,
|
8750 |
+
"learning_rate": 3.45244286985949e-06,
|
8751 |
+
"loss": 1.5135,
|
8752 |
+
"step": 1249
|
8753 |
+
},
|
8754 |
+
{
|
8755 |
+
"epoch": 0.21382141635306193,
|
8756 |
+
"grad_norm": 4.035844326019287,
|
8757 |
+
"learning_rate": 3.411937259190029e-06,
|
8758 |
+
"loss": 1.7973,
|
8759 |
+
"step": 1250
|
8760 |
+
},
|
8761 |
+
{
|
8762 |
+
"epoch": 0.21399247348614436,
|
8763 |
+
"grad_norm": 0.505690336227417,
|
8764 |
+
"learning_rate": 3.371662272692677e-06,
|
8765 |
+
"loss": 1.1879,
|
8766 |
+
"step": 1251
|
8767 |
+
},
|
8768 |
+
{
|
8769 |
+
"epoch": 0.21416353061922683,
|
8770 |
+
"grad_norm": 0.5308461785316467,
|
8771 |
+
"learning_rate": 3.33161810974007e-06,
|
8772 |
+
"loss": 1.2725,
|
8773 |
+
"step": 1252
|
8774 |
+
},
|
8775 |
+
{
|
8776 |
+
"epoch": 0.21433458775230926,
|
8777 |
+
"grad_norm": 0.5629502534866333,
|
8778 |
+
"learning_rate": 3.2918049685622175e-06,
|
8779 |
+
"loss": 1.1357,
|
8780 |
+
"step": 1253
|
8781 |
+
},
|
8782 |
+
{
|
8783 |
+
"epoch": 0.21450564488539173,
|
8784 |
+
"grad_norm": 0.547167181968689,
|
8785 |
+
"learning_rate": 3.2522230462455006e-06,
|
8786 |
+
"loss": 1.1892,
|
8787 |
+
"step": 1254
|
8788 |
+
},
|
8789 |
+
{
|
8790 |
+
"epoch": 0.21467670201847416,
|
8791 |
+
"grad_norm": 0.5315078496932983,
|
8792 |
+
"learning_rate": 3.212872538731704e-06,
|
8793 |
+
"loss": 1.2391,
|
8794 |
+
"step": 1255
|
8795 |
+
},
|
8796 |
+
{
|
8797 |
+
"epoch": 0.21484775915155663,
|
8798 |
+
"grad_norm": 0.5446400046348572,
|
8799 |
+
"learning_rate": 3.173753640817051e-06,
|
8800 |
+
"loss": 1.2884,
|
8801 |
+
"step": 1256
|
8802 |
+
},
|
8803 |
+
{
|
8804 |
+
"epoch": 0.21501881628463906,
|
8805 |
+
"grad_norm": 0.5265285968780518,
|
8806 |
+
"learning_rate": 3.1348665461511994e-06,
|
8807 |
+
"loss": 1.2992,
|
8808 |
+
"step": 1257
|
8809 |
+
},
|
8810 |
+
{
|
8811 |
+
"epoch": 0.21518987341772153,
|
8812 |
+
"grad_norm": 0.5416175723075867,
|
8813 |
+
"learning_rate": 3.096211447236347e-06,
|
8814 |
+
"loss": 1.1872,
|
8815 |
+
"step": 1258
|
8816 |
+
},
|
8817 |
+
{
|
8818 |
+
"epoch": 0.21536093055080396,
|
8819 |
+
"grad_norm": 0.5405793190002441,
|
8820 |
+
"learning_rate": 3.0577885354262534e-06,
|
8821 |
+
"loss": 1.2496,
|
8822 |
+
"step": 1259
|
8823 |
+
},
|
8824 |
+
{
|
8825 |
+
"epoch": 0.21553198768388643,
|
8826 |
+
"grad_norm": 0.5459663271903992,
|
8827 |
+
"learning_rate": 3.0195980009252444e-06,
|
8828 |
+
"loss": 1.2037,
|
8829 |
+
"step": 1260
|
8830 |
+
},
|
8831 |
+
{
|
8832 |
+
"epoch": 0.21570304481696886,
|
8833 |
+
"grad_norm": 0.5698009133338928,
|
8834 |
+
"learning_rate": 2.981640032787347e-06,
|
8835 |
+
"loss": 1.2923,
|
8836 |
+
"step": 1261
|
8837 |
+
},
|
8838 |
+
{
|
8839 |
+
"epoch": 0.21587410195005133,
|
8840 |
+
"grad_norm": 0.5835180878639221,
|
8841 |
+
"learning_rate": 2.9439148189153166e-06,
|
8842 |
+
"loss": 1.3814,
|
8843 |
+
"step": 1262
|
8844 |
+
},
|
8845 |
+
{
|
8846 |
+
"epoch": 0.21604515908313376,
|
8847 |
+
"grad_norm": 0.6166893839836121,
|
8848 |
+
"learning_rate": 2.9064225460596837e-06,
|
8849 |
+
"loss": 1.3029,
|
8850 |
+
"step": 1263
|
8851 |
+
},
|
8852 |
+
{
|
8853 |
+
"epoch": 0.21621621621621623,
|
8854 |
+
"grad_norm": 0.5921556353569031,
|
8855 |
+
"learning_rate": 2.8691633998178924e-06,
|
8856 |
+
"loss": 1.3265,
|
8857 |
+
"step": 1264
|
8858 |
+
},
|
8859 |
+
{
|
8860 |
+
"epoch": 0.21638727334929866,
|
8861 |
+
"grad_norm": 0.61456298828125,
|
8862 |
+
"learning_rate": 2.8321375646333027e-06,
|
8863 |
+
"loss": 1.2458,
|
8864 |
+
"step": 1265
|
8865 |
+
},
|
8866 |
+
{
|
8867 |
+
"epoch": 0.21655833048238113,
|
8868 |
+
"grad_norm": 0.603403627872467,
|
8869 |
+
"learning_rate": 2.7953452237943488e-06,
|
8870 |
+
"loss": 1.4179,
|
8871 |
+
"step": 1266
|
8872 |
+
},
|
8873 |
+
{
|
8874 |
+
"epoch": 0.21672938761546356,
|
8875 |
+
"grad_norm": 0.6556985378265381,
|
8876 |
+
"learning_rate": 2.758786559433585e-06,
|
8877 |
+
"loss": 1.4252,
|
8878 |
+
"step": 1267
|
8879 |
+
},
|
8880 |
+
{
|
8881 |
+
"epoch": 0.216900444748546,
|
8882 |
+
"grad_norm": 0.6773498058319092,
|
8883 |
+
"learning_rate": 2.7224617525268117e-06,
|
8884 |
+
"loss": 1.3483,
|
8885 |
+
"step": 1268
|
8886 |
+
},
|
8887 |
+
{
|
8888 |
+
"epoch": 0.21707150188162846,
|
8889 |
+
"grad_norm": 0.6843062043190002,
|
8890 |
+
"learning_rate": 2.6863709828921614e-06,
|
8891 |
+
"loss": 1.4204,
|
8892 |
+
"step": 1269
|
8893 |
+
},
|
8894 |
+
{
|
8895 |
+
"epoch": 0.2172425590147109,
|
8896 |
+
"grad_norm": 0.6551604270935059,
|
8897 |
+
"learning_rate": 2.650514429189227e-06,
|
8898 |
+
"loss": 1.2844,
|
8899 |
+
"step": 1270
|
8900 |
+
},
|
8901 |
+
{
|
8902 |
+
"epoch": 0.21741361614779336,
|
8903 |
+
"grad_norm": 0.6590097546577454,
|
8904 |
+
"learning_rate": 2.6148922689181397e-06,
|
8905 |
+
"loss": 0.9632,
|
8906 |
+
"step": 1271
|
8907 |
+
},
|
8908 |
+
{
|
8909 |
+
"epoch": 0.2175846732808758,
|
8910 |
+
"grad_norm": 0.693517804145813,
|
8911 |
+
"learning_rate": 2.579504678418726e-06,
|
8912 |
+
"loss": 1.2156,
|
8913 |
+
"step": 1272
|
8914 |
+
},
|
8915 |
+
{
|
8916 |
+
"epoch": 0.21775573041395827,
|
8917 |
+
"grad_norm": 0.7429909706115723,
|
8918 |
+
"learning_rate": 2.544351832869668e-06,
|
8919 |
+
"loss": 1.1741,
|
8920 |
+
"step": 1273
|
8921 |
+
},
|
8922 |
+
{
|
8923 |
+
"epoch": 0.2179267875470407,
|
8924 |
+
"grad_norm": 0.7458788156509399,
|
8925 |
+
"learning_rate": 2.509433906287523e-06,
|
8926 |
+
"loss": 1.1041,
|
8927 |
+
"step": 1274
|
8928 |
+
},
|
8929 |
+
{
|
8930 |
+
"epoch": 0.21809784468012317,
|
8931 |
+
"grad_norm": 0.7648807168006897,
|
8932 |
+
"learning_rate": 2.474751071525988e-06,
|
8933 |
+
"loss": 1.2219,
|
8934 |
+
"step": 1275
|
8935 |
+
},
|
8936 |
+
{
|
8937 |
+
"epoch": 0.2182689018132056,
|
8938 |
+
"grad_norm": 0.806402862071991,
|
8939 |
+
"learning_rate": 2.4403035002749543e-06,
|
8940 |
+
"loss": 1.2273,
|
8941 |
+
"step": 1276
|
8942 |
+
},
|
8943 |
+
{
|
8944 |
+
"epoch": 0.21843995894628807,
|
8945 |
+
"grad_norm": 0.8961294889450073,
|
8946 |
+
"learning_rate": 2.4060913630596983e-06,
|
8947 |
+
"loss": 1.2363,
|
8948 |
+
"step": 1277
|
8949 |
+
},
|
8950 |
+
{
|
8951 |
+
"epoch": 0.2186110160793705,
|
8952 |
+
"grad_norm": 0.8457828760147095,
|
8953 |
+
"learning_rate": 2.3721148292400686e-06,
|
8954 |
+
"loss": 1.0838,
|
8955 |
+
"step": 1278
|
8956 |
+
},
|
8957 |
+
{
|
8958 |
+
"epoch": 0.21878207321245297,
|
8959 |
+
"grad_norm": 0.8772481679916382,
|
8960 |
+
"learning_rate": 2.3383740670095524e-06,
|
8961 |
+
"loss": 1.1298,
|
8962 |
+
"step": 1279
|
8963 |
+
},
|
8964 |
+
{
|
8965 |
+
"epoch": 0.2189531303455354,
|
8966 |
+
"grad_norm": 0.8471772074699402,
|
8967 |
+
"learning_rate": 2.3048692433945307e-06,
|
8968 |
+
"loss": 1.0732,
|
8969 |
+
"step": 1280
|
8970 |
+
},
|
8971 |
+
{
|
8972 |
+
"epoch": 0.21912418747861787,
|
8973 |
+
"grad_norm": 0.900969386100769,
|
8974 |
+
"learning_rate": 2.2716005242534257e-06,
|
8975 |
+
"loss": 1.0379,
|
8976 |
+
"step": 1281
|
8977 |
+
},
|
8978 |
+
{
|
8979 |
+
"epoch": 0.2192952446117003,
|
8980 |
+
"grad_norm": 0.9625850915908813,
|
8981 |
+
"learning_rate": 2.238568074275865e-06,
|
8982 |
+
"loss": 0.9685,
|
8983 |
+
"step": 1282
|
8984 |
+
},
|
8985 |
+
{
|
8986 |
+
"epoch": 0.21946630174478277,
|
8987 |
+
"grad_norm": 0.9194417595863342,
|
8988 |
+
"learning_rate": 2.205772056981875e-06,
|
8989 |
+
"loss": 1.119,
|
8990 |
+
"step": 1283
|
8991 |
+
},
|
8992 |
+
{
|
8993 |
+
"epoch": 0.2196373588778652,
|
8994 |
+
"grad_norm": 1.0923161506652832,
|
8995 |
+
"learning_rate": 2.173212634721089e-06,
|
8996 |
+
"loss": 1.2946,
|
8997 |
+
"step": 1284
|
8998 |
+
},
|
8999 |
+
{
|
9000 |
+
"epoch": 0.21980841601094767,
|
9001 |
+
"grad_norm": 1.2934038639068604,
|
9002 |
+
"learning_rate": 2.1408899686718997e-06,
|
9003 |
+
"loss": 1.1849,
|
9004 |
+
"step": 1285
|
9005 |
+
},
|
9006 |
+
{
|
9007 |
+
"epoch": 0.2199794731440301,
|
9008 |
+
"grad_norm": 1.11680269241333,
|
9009 |
+
"learning_rate": 2.1088042188407097e-06,
|
9010 |
+
"loss": 1.3148,
|
9011 |
+
"step": 1286
|
9012 |
+
},
|
9013 |
+
{
|
9014 |
+
"epoch": 0.22015053027711257,
|
9015 |
+
"grad_norm": 1.1829283237457275,
|
9016 |
+
"learning_rate": 2.0769555440611154e-06,
|
9017 |
+
"loss": 1.2888,
|
9018 |
+
"step": 1287
|
9019 |
+
},
|
9020 |
+
{
|
9021 |
+
"epoch": 0.220321587410195,
|
9022 |
+
"grad_norm": 1.3382573127746582,
|
9023 |
+
"learning_rate": 2.045344101993124e-06,
|
9024 |
+
"loss": 1.4106,
|
9025 |
+
"step": 1288
|
9026 |
+
},
|
9027 |
+
{
|
9028 |
+
"epoch": 0.22049264454327747,
|
9029 |
+
"grad_norm": 1.2632192373275757,
|
9030 |
+
"learning_rate": 2.0139700491223766e-06,
|
9031 |
+
"loss": 1.0973,
|
9032 |
+
"step": 1289
|
9033 |
+
},
|
9034 |
+
{
|
9035 |
+
"epoch": 0.2206637016763599,
|
9036 |
+
"grad_norm": 1.4644615650177002,
|
9037 |
+
"learning_rate": 1.982833540759349e-06,
|
9038 |
+
"loss": 1.2564,
|
9039 |
+
"step": 1290
|
9040 |
+
},
|
9041 |
+
{
|
9042 |
+
"epoch": 0.22083475880944237,
|
9043 |
+
"grad_norm": 1.6283206939697266,
|
9044 |
+
"learning_rate": 1.951934731038635e-06,
|
9045 |
+
"loss": 1.2009,
|
9046 |
+
"step": 1291
|
9047 |
+
},
|
9048 |
+
{
|
9049 |
+
"epoch": 0.2210058159425248,
|
9050 |
+
"grad_norm": 1.6942142248153687,
|
9051 |
+
"learning_rate": 1.9212737729181417e-06,
|
9052 |
+
"loss": 1.148,
|
9053 |
+
"step": 1292
|
9054 |
+
},
|
9055 |
+
{
|
9056 |
+
"epoch": 0.22117687307560727,
|
9057 |
+
"grad_norm": 1.6936148405075073,
|
9058 |
+
"learning_rate": 1.8908508181783247e-06,
|
9059 |
+
"loss": 1.2038,
|
9060 |
+
"step": 1293
|
9061 |
+
},
|
9062 |
+
{
|
9063 |
+
"epoch": 0.2213479302086897,
|
9064 |
+
"grad_norm": 2.2340569496154785,
|
9065 |
+
"learning_rate": 1.8606660174214807e-06,
|
9066 |
+
"loss": 1.4037,
|
9067 |
+
"step": 1294
|
9068 |
+
},
|
9069 |
+
{
|
9070 |
+
"epoch": 0.22151898734177214,
|
9071 |
+
"grad_norm": 2.1160356998443604,
|
9072 |
+
"learning_rate": 1.8307195200709727e-06,
|
9073 |
+
"loss": 1.2712,
|
9074 |
+
"step": 1295
|
9075 |
+
},
|
9076 |
+
{
|
9077 |
+
"epoch": 0.2216900444748546,
|
9078 |
+
"grad_norm": 2.3536376953125,
|
9079 |
+
"learning_rate": 1.801011474370462e-06,
|
9080 |
+
"loss": 1.4199,
|
9081 |
+
"step": 1296
|
9082 |
+
},
|
9083 |
+
{
|
9084 |
+
"epoch": 0.22186110160793704,
|
9085 |
+
"grad_norm": 2.409590482711792,
|
9086 |
+
"learning_rate": 1.771542027383255e-06,
|
9087 |
+
"loss": 1.0671,
|
9088 |
+
"step": 1297
|
9089 |
+
},
|
9090 |
+
{
|
9091 |
+
"epoch": 0.2220321587410195,
|
9092 |
+
"grad_norm": 2.8295600414276123,
|
9093 |
+
"learning_rate": 1.7423113249914968e-06,
|
9094 |
+
"loss": 1.6143,
|
9095 |
+
"step": 1298
|
9096 |
+
},
|
9097 |
+
{
|
9098 |
+
"epoch": 0.22220321587410194,
|
9099 |
+
"grad_norm": 3.7062695026397705,
|
9100 |
+
"learning_rate": 1.7133195118954838e-06,
|
9101 |
+
"loss": 1.6594,
|
9102 |
+
"step": 1299
|
9103 |
+
},
|
9104 |
+
{
|
9105 |
+
"epoch": 0.2223742730071844,
|
9106 |
+
"grad_norm": 4.451565742492676,
|
9107 |
+
"learning_rate": 1.6845667316129353e-06,
|
9108 |
+
"loss": 1.6766,
|
9109 |
+
"step": 1300
|
9110 |
+
},
|
9111 |
+
{
|
9112 |
+
"epoch": 0.22254533014026684,
|
9113 |
+
"grad_norm": 0.4803614318370819,
|
9114 |
+
"learning_rate": 1.656053126478313e-06,
|
9115 |
+
"loss": 1.0114,
|
9116 |
+
"step": 1301
|
9117 |
+
},
|
9118 |
+
{
|
9119 |
+
"epoch": 0.2227163872733493,
|
9120 |
+
"grad_norm": 0.5046311020851135,
|
9121 |
+
"learning_rate": 1.6277788376420634e-06,
|
9122 |
+
"loss": 1.1565,
|
9123 |
+
"step": 1302
|
9124 |
+
},
|
9125 |
+
{
|
9126 |
+
"epoch": 0.22288744440643174,
|
9127 |
+
"grad_norm": 0.507754385471344,
|
9128 |
+
"learning_rate": 1.5997440050699863e-06,
|
9129 |
+
"loss": 1.1134,
|
9130 |
+
"step": 1303
|
9131 |
+
},
|
9132 |
+
{
|
9133 |
+
"epoch": 0.2230585015395142,
|
9134 |
+
"grad_norm": 0.5594250559806824,
|
9135 |
+
"learning_rate": 1.5719487675424582e-06,
|
9136 |
+
"loss": 1.2747,
|
9137 |
+
"step": 1304
|
9138 |
+
},
|
9139 |
+
{
|
9140 |
+
"epoch": 0.22322955867259664,
|
9141 |
+
"grad_norm": 0.5371953248977661,
|
9142 |
+
"learning_rate": 1.5443932626538316e-06,
|
9143 |
+
"loss": 1.2004,
|
9144 |
+
"step": 1305
|
9145 |
+
},
|
9146 |
+
{
|
9147 |
+
"epoch": 0.2234006158056791,
|
9148 |
+
"grad_norm": 0.5366808772087097,
|
9149 |
+
"learning_rate": 1.517077626811697e-06,
|
9150 |
+
"loss": 1.2323,
|
9151 |
+
"step": 1306
|
9152 |
+
},
|
9153 |
+
{
|
9154 |
+
"epoch": 0.22357167293876154,
|
9155 |
+
"grad_norm": 0.5580090880393982,
|
9156 |
+
"learning_rate": 1.4900019952362232e-06,
|
9157 |
+
"loss": 1.2679,
|
9158 |
+
"step": 1307
|
9159 |
+
},
|
9160 |
+
{
|
9161 |
+
"epoch": 0.223742730071844,
|
9162 |
+
"grad_norm": 0.5684660077095032,
|
9163 |
+
"learning_rate": 1.4631665019595009e-06,
|
9164 |
+
"loss": 1.303,
|
9165 |
+
"step": 1308
|
9166 |
+
},
|
9167 |
+
{
|
9168 |
+
"epoch": 0.22391378720492644,
|
9169 |
+
"grad_norm": 0.5655079483985901,
|
9170 |
+
"learning_rate": 1.436571279824861e-06,
|
9171 |
+
"loss": 1.3581,
|
9172 |
+
"step": 1309
|
9173 |
+
},
|
9174 |
+
{
|
9175 |
+
"epoch": 0.2240848443380089,
|
9176 |
+
"grad_norm": 0.582984447479248,
|
9177 |
+
"learning_rate": 1.4102164604862078e-06,
|
9178 |
+
"loss": 1.3664,
|
9179 |
+
"step": 1310
|
9180 |
+
},
|
9181 |
+
{
|
9182 |
+
"epoch": 0.22425590147109134,
|
9183 |
+
"grad_norm": 0.5544376969337463,
|
9184 |
+
"learning_rate": 1.3841021744074257e-06,
|
9185 |
+
"loss": 1.2845,
|
9186 |
+
"step": 1311
|
9187 |
+
},
|
9188 |
+
{
|
9189 |
+
"epoch": 0.2244269586041738,
|
9190 |
+
"grad_norm": 0.565316379070282,
|
9191 |
+
"learning_rate": 1.35822855086164e-06,
|
9192 |
+
"loss": 1.2483,
|
9193 |
+
"step": 1312
|
9194 |
+
},
|
9195 |
+
{
|
9196 |
+
"epoch": 0.22459801573725624,
|
9197 |
+
"grad_norm": 0.5738152861595154,
|
9198 |
+
"learning_rate": 1.3325957179306626e-06,
|
9199 |
+
"loss": 1.2566,
|
9200 |
+
"step": 1313
|
9201 |
+
},
|
9202 |
+
{
|
9203 |
+
"epoch": 0.2247690728703387,
|
9204 |
+
"grad_norm": 0.6058394908905029,
|
9205 |
+
"learning_rate": 1.3072038025043198e-06,
|
9206 |
+
"loss": 1.3314,
|
9207 |
+
"step": 1314
|
9208 |
+
},
|
9209 |
+
{
|
9210 |
+
"epoch": 0.22494013000342114,
|
9211 |
+
"grad_norm": 0.618689775466919,
|
9212 |
+
"learning_rate": 1.2820529302798035e-06,
|
9213 |
+
"loss": 1.3883,
|
9214 |
+
"step": 1315
|
9215 |
+
},
|
9216 |
+
{
|
9217 |
+
"epoch": 0.2251111871365036,
|
9218 |
+
"grad_norm": 0.621052622795105,
|
9219 |
+
"learning_rate": 1.2571432257610983e-06,
|
9220 |
+
"loss": 1.2904,
|
9221 |
+
"step": 1316
|
9222 |
+
},
|
9223 |
+
{
|
9224 |
+
"epoch": 0.22528224426958604,
|
9225 |
+
"grad_norm": 0.6590416431427002,
|
9226 |
+
"learning_rate": 1.232474812258344e-06,
|
9227 |
+
"loss": 1.4642,
|
9228 |
+
"step": 1317
|
9229 |
+
},
|
9230 |
+
{
|
9231 |
+
"epoch": 0.2254533014026685,
|
9232 |
+
"grad_norm": 0.6238689422607422,
|
9233 |
+
"learning_rate": 1.2080478118871974e-06,
|
9234 |
+
"loss": 1.2775,
|
9235 |
+
"step": 1318
|
9236 |
+
},
|
9237 |
+
{
|
9238 |
+
"epoch": 0.22562435853575094,
|
9239 |
+
"grad_norm": 0.6821351647377014,
|
9240 |
+
"learning_rate": 1.1838623455682652e-06,
|
9241 |
+
"loss": 1.2017,
|
9242 |
+
"step": 1319
|
9243 |
+
},
|
9244 |
+
{
|
9245 |
+
"epoch": 0.22579541566883338,
|
9246 |
+
"grad_norm": 0.6280372738838196,
|
9247 |
+
"learning_rate": 1.1599185330264883e-06,
|
9248 |
+
"loss": 1.0048,
|
9249 |
+
"step": 1320
|
9250 |
+
},
|
9251 |
+
{
|
9252 |
+
"epoch": 0.22596647280191584,
|
9253 |
+
"grad_norm": 0.6800519227981567,
|
9254 |
+
"learning_rate": 1.1362164927905595e-06,
|
9255 |
+
"loss": 1.4072,
|
9256 |
+
"step": 1321
|
9257 |
+
},
|
9258 |
+
{
|
9259 |
+
"epoch": 0.22613752993499828,
|
9260 |
+
"grad_norm": 0.7017683386802673,
|
9261 |
+
"learning_rate": 1.1127563421923226e-06,
|
9262 |
+
"loss": 1.0213,
|
9263 |
+
"step": 1322
|
9264 |
+
},
|
9265 |
+
{
|
9266 |
+
"epoch": 0.22630858706808074,
|
9267 |
+
"grad_norm": 0.7159509062767029,
|
9268 |
+
"learning_rate": 1.0895381973662023e-06,
|
9269 |
+
"loss": 1.1265,
|
9270 |
+
"step": 1323
|
9271 |
+
},
|
9272 |
+
{
|
9273 |
+
"epoch": 0.22647964420116318,
|
9274 |
+
"grad_norm": 0.7772588729858398,
|
9275 |
+
"learning_rate": 1.0665621732486198e-06,
|
9276 |
+
"loss": 1.293,
|
9277 |
+
"step": 1324
|
9278 |
+
},
|
9279 |
+
{
|
9280 |
+
"epoch": 0.22665070133424564,
|
9281 |
+
"grad_norm": 0.7362921237945557,
|
9282 |
+
"learning_rate": 1.0438283835774387e-06,
|
9283 |
+
"loss": 1.0704,
|
9284 |
+
"step": 1325
|
9285 |
+
},
|
9286 |
+
{
|
9287 |
+
"epoch": 0.22682175846732808,
|
9288 |
+
"grad_norm": 0.7900636792182922,
|
9289 |
+
"learning_rate": 1.0213369408913932e-06,
|
9290 |
+
"loss": 1.0694,
|
9291 |
+
"step": 1326
|
9292 |
+
},
|
9293 |
+
{
|
9294 |
+
"epoch": 0.22699281560041054,
|
9295 |
+
"grad_norm": 0.810276448726654,
|
9296 |
+
"learning_rate": 9.990879565295153e-07,
|
9297 |
+
"loss": 1.2045,
|
9298 |
+
"step": 1327
|
9299 |
+
},
|
9300 |
+
{
|
9301 |
+
"epoch": 0.22716387273349298,
|
9302 |
+
"grad_norm": 0.8556481003761292,
|
9303 |
+
"learning_rate": 9.770815406306312e-07,
|
9304 |
+
"loss": 1.2036,
|
9305 |
+
"step": 1328
|
9306 |
+
},
|
9307 |
+
{
|
9308 |
+
"epoch": 0.22733492986657544,
|
9309 |
+
"grad_norm": 0.9373174905776978,
|
9310 |
+
"learning_rate": 9.553178021327326e-07,
|
9311 |
+
"loss": 1.4073,
|
9312 |
+
"step": 1329
|
9313 |
+
},
|
9314 |
+
{
|
9315 |
+
"epoch": 0.22750598699965788,
|
9316 |
+
"grad_norm": 0.8940708637237549,
|
9317 |
+
"learning_rate": 9.337968487725501e-07,
|
9318 |
+
"loss": 1.2526,
|
9319 |
+
"step": 1330
|
9320 |
+
},
|
9321 |
+
{
|
9322 |
+
"epoch": 0.22767704413274034,
|
9323 |
+
"grad_norm": 0.9707661867141724,
|
9324 |
+
"learning_rate": 9.12518787084915e-07,
|
9325 |
+
"loss": 1.2184,
|
9326 |
+
"step": 1331
|
9327 |
+
},
|
9328 |
+
{
|
9329 |
+
"epoch": 0.22784810126582278,
|
9330 |
+
"grad_norm": 0.844813346862793,
|
9331 |
+
"learning_rate": 8.91483722402292e-07,
|
9332 |
+
"loss": 1.0718,
|
9333 |
+
"step": 1332
|
9334 |
+
},
|
9335 |
+
{
|
9336 |
+
"epoch": 0.22801915839890524,
|
9337 |
+
"grad_norm": 1.115906834602356,
|
9338 |
+
"learning_rate": 8.706917588542418e-07,
|
9339 |
+
"loss": 1.3152,
|
9340 |
+
"step": 1333
|
9341 |
+
},
|
9342 |
+
{
|
9343 |
+
"epoch": 0.22819021553198768,
|
9344 |
+
"grad_norm": 0.9982916116714478,
|
9345 |
+
"learning_rate": 8.50142999366893e-07,
|
9346 |
+
"loss": 1.1193,
|
9347 |
+
"step": 1334
|
9348 |
+
},
|
9349 |
+
{
|
9350 |
+
"epoch": 0.22836127266507014,
|
9351 |
+
"grad_norm": 1.2234420776367188,
|
9352 |
+
"learning_rate": 8.298375456624652e-07,
|
9353 |
+
"loss": 1.2037,
|
9354 |
+
"step": 1335
|
9355 |
+
},
|
9356 |
+
{
|
9357 |
+
"epoch": 0.22853232979815258,
|
9358 |
+
"grad_norm": 1.1149396896362305,
|
9359 |
+
"learning_rate": 8.097754982587303e-07,
|
9360 |
+
"loss": 1.0903,
|
9361 |
+
"step": 1336
|
9362 |
+
},
|
9363 |
+
{
|
9364 |
+
"epoch": 0.22870338693123504,
|
9365 |
+
"grad_norm": 1.2126588821411133,
|
9366 |
+
"learning_rate": 7.899569564685294e-07,
|
9367 |
+
"loss": 1.2339,
|
9368 |
+
"step": 1337
|
9369 |
+
},
|
9370 |
+
{
|
9371 |
+
"epoch": 0.22887444406431748,
|
9372 |
+
"grad_norm": 1.2695289850234985,
|
9373 |
+
"learning_rate": 7.70382018399296e-07,
|
9374 |
+
"loss": 1.1751,
|
9375 |
+
"step": 1338
|
9376 |
+
},
|
9377 |
+
{
|
9378 |
+
"epoch": 0.22904550119739994,
|
9379 |
+
"grad_norm": 1.532913327217102,
|
9380 |
+
"learning_rate": 7.510507809525335e-07,
|
9381 |
+
"loss": 1.4258,
|
9382 |
+
"step": 1339
|
9383 |
+
},
|
9384 |
+
{
|
9385 |
+
"epoch": 0.22921655833048238,
|
9386 |
+
"grad_norm": 1.5996785163879395,
|
9387 |
+
"learning_rate": 7.319633398233772e-07,
|
9388 |
+
"loss": 1.2917,
|
9389 |
+
"step": 1340
|
9390 |
+
},
|
9391 |
+
{
|
9392 |
+
"epoch": 0.22938761546356484,
|
9393 |
+
"grad_norm": 1.3822323083877563,
|
9394 |
+
"learning_rate": 7.131197895000941e-07,
|
9395 |
+
"loss": 1.1636,
|
9396 |
+
"step": 1341
|
9397 |
+
},
|
9398 |
+
{
|
9399 |
+
"epoch": 0.22955867259664728,
|
9400 |
+
"grad_norm": 1.7828960418701172,
|
9401 |
+
"learning_rate": 6.945202232636228e-07,
|
9402 |
+
"loss": 1.3946,
|
9403 |
+
"step": 1342
|
9404 |
+
},
|
9405 |
+
{
|
9406 |
+
"epoch": 0.22972972972972974,
|
9407 |
+
"grad_norm": 1.8755525350570679,
|
9408 |
+
"learning_rate": 6.761647331871013e-07,
|
9409 |
+
"loss": 1.3885,
|
9410 |
+
"step": 1343
|
9411 |
+
},
|
9412 |
+
{
|
9413 |
+
"epoch": 0.22990078686281218,
|
9414 |
+
"grad_norm": 2.543682098388672,
|
9415 |
+
"learning_rate": 6.580534101354452e-07,
|
9416 |
+
"loss": 1.7416,
|
9417 |
+
"step": 1344
|
9418 |
+
},
|
9419 |
+
{
|
9420 |
+
"epoch": 0.23007184399589464,
|
9421 |
+
"grad_norm": 1.908764362335205,
|
9422 |
+
"learning_rate": 6.401863437648481e-07,
|
9423 |
+
"loss": 1.1909,
|
9424 |
+
"step": 1345
|
9425 |
+
},
|
9426 |
+
{
|
9427 |
+
"epoch": 0.23024290112897708,
|
9428 |
+
"grad_norm": 2.3398802280426025,
|
9429 |
+
"learning_rate": 6.225636225223763e-07,
|
9430 |
+
"loss": 1.4931,
|
9431 |
+
"step": 1346
|
9432 |
+
},
|
9433 |
+
{
|
9434 |
+
"epoch": 0.23041395826205952,
|
9435 |
+
"grad_norm": 2.357182264328003,
|
9436 |
+
"learning_rate": 6.051853336455249e-07,
|
9437 |
+
"loss": 1.4752,
|
9438 |
+
"step": 1347
|
9439 |
+
},
|
9440 |
+
{
|
9441 |
+
"epoch": 0.23058501539514198,
|
9442 |
+
"grad_norm": 2.3424172401428223,
|
9443 |
+
"learning_rate": 5.880515631617567e-07,
|
9444 |
+
"loss": 1.0249,
|
9445 |
+
"step": 1348
|
9446 |
+
},
|
9447 |
+
{
|
9448 |
+
"epoch": 0.23075607252822442,
|
9449 |
+
"grad_norm": 2.934676170349121,
|
9450 |
+
"learning_rate": 5.711623958881195e-07,
|
9451 |
+
"loss": 1.5519,
|
9452 |
+
"step": 1349
|
9453 |
+
},
|
9454 |
+
{
|
9455 |
+
"epoch": 0.23092712966130688,
|
9456 |
+
"grad_norm": 4.101928234100342,
|
9457 |
+
"learning_rate": 5.545179154308078e-07,
|
9458 |
+
"loss": 1.8836,
|
9459 |
+
"step": 1350
|
9460 |
+
},
|
9461 |
+
{
|
9462 |
+
"epoch": 0.23109818679438932,
|
9463 |
+
"grad_norm": 0.532934844493866,
|
9464 |
+
"learning_rate": 5.381182041847232e-07,
|
9465 |
+
"loss": 1.1214,
|
9466 |
+
"step": 1351
|
9467 |
+
},
|
9468 |
+
{
|
9469 |
+
"epoch": 0.23126924392747178,
|
9470 |
+
"grad_norm": 0.47591790556907654,
|
9471 |
+
"learning_rate": 5.21963343333115e-07,
|
9472 |
+
"loss": 1.1511,
|
9473 |
+
"step": 1352
|
9474 |
+
},
|
9475 |
+
{
|
9476 |
+
"epoch": 0.23144030106055422,
|
9477 |
+
"grad_norm": 0.4995153546333313,
|
9478 |
+
"learning_rate": 5.06053412847135e-07,
|
9479 |
+
"loss": 1.1037,
|
9480 |
+
"step": 1353
|
9481 |
+
},
|
9482 |
+
{
|
9483 |
+
"epoch": 0.23161135819363668,
|
9484 |
+
"grad_norm": 0.5946906805038452,
|
9485 |
+
"learning_rate": 4.903884914854717e-07,
|
9486 |
+
"loss": 1.1168,
|
9487 |
+
"step": 1354
|
9488 |
+
},
|
9489 |
+
{
|
9490 |
+
"epoch": 0.23178241532671912,
|
9491 |
+
"grad_norm": 0.4970369338989258,
|
9492 |
+
"learning_rate": 4.749686567939504e-07,
|
9493 |
+
"loss": 1.0871,
|
9494 |
+
"step": 1355
|
9495 |
+
},
|
9496 |
+
{
|
9497 |
+
"epoch": 0.23195347245980158,
|
9498 |
+
"grad_norm": 0.5578677654266357,
|
9499 |
+
"learning_rate": 4.597939851051336e-07,
|
9500 |
+
"loss": 1.3034,
|
9501 |
+
"step": 1356
|
9502 |
+
},
|
9503 |
+
{
|
9504 |
+
"epoch": 0.23212452959288402,
|
9505 |
+
"grad_norm": 0.5544520616531372,
|
9506 |
+
"learning_rate": 4.448645515379657e-07,
|
9507 |
+
"loss": 1.2657,
|
9508 |
+
"step": 1357
|
9509 |
+
},
|
9510 |
+
{
|
9511 |
+
"epoch": 0.23229558672596648,
|
9512 |
+
"grad_norm": 0.529667317867279,
|
9513 |
+
"learning_rate": 4.301804299973955e-07,
|
9514 |
+
"loss": 1.2368,
|
9515 |
+
"step": 1358
|
9516 |
+
},
|
9517 |
+
{
|
9518 |
+
"epoch": 0.23246664385904892,
|
9519 |
+
"grad_norm": 0.5463925004005432,
|
9520 |
+
"learning_rate": 4.157416931740099e-07,
|
9521 |
+
"loss": 1.2884,
|
9522 |
+
"step": 1359
|
9523 |
+
},
|
9524 |
+
{
|
9525 |
+
"epoch": 0.23263770099213138,
|
9526 |
+
"grad_norm": 0.5573237538337708,
|
9527 |
+
"learning_rate": 4.015484125436564e-07,
|
9528 |
+
"loss": 1.2477,
|
9529 |
+
"step": 1360
|
9530 |
+
},
|
9531 |
+
{
|
9532 |
+
"epoch": 0.23280875812521382,
|
9533 |
+
"grad_norm": 0.5883082151412964,
|
9534 |
+
"learning_rate": 3.8760065836712653e-07,
|
9535 |
+
"loss": 1.1823,
|
9536 |
+
"step": 1361
|
9537 |
+
},
|
9538 |
+
{
|
9539 |
+
"epoch": 0.23297981525829628,
|
9540 |
+
"grad_norm": 0.5314277410507202,
|
9541 |
+
"learning_rate": 3.738984996897621e-07,
|
9542 |
+
"loss": 1.1647,
|
9543 |
+
"step": 1362
|
9544 |
+
},
|
9545 |
+
{
|
9546 |
+
"epoch": 0.23315087239137872,
|
9547 |
+
"grad_norm": 0.6144003868103027,
|
9548 |
+
"learning_rate": 3.604420043411494e-07,
|
9549 |
+
"loss": 1.4237,
|
9550 |
+
"step": 1363
|
9551 |
+
},
|
9552 |
+
{
|
9553 |
+
"epoch": 0.23332192952446118,
|
9554 |
+
"grad_norm": 0.5864267349243164,
|
9555 |
+
"learning_rate": 3.472312389347809e-07,
|
9556 |
+
"loss": 1.4627,
|
9557 |
+
"step": 1364
|
9558 |
+
},
|
9559 |
+
{
|
9560 |
+
"epoch": 0.23349298665754362,
|
9561 |
+
"grad_norm": 0.5905508995056152,
|
9562 |
+
"learning_rate": 3.342662688676945e-07,
|
9563 |
+
"loss": 1.2693,
|
9564 |
+
"step": 1365
|
9565 |
+
},
|
9566 |
+
{
|
9567 |
+
"epoch": 0.23366404379062608,
|
9568 |
+
"grad_norm": 0.6206147074699402,
|
9569 |
+
"learning_rate": 3.2154715832019013e-07,
|
9570 |
+
"loss": 1.33,
|
9571 |
+
"step": 1366
|
9572 |
+
},
|
9573 |
+
{
|
9574 |
+
"epoch": 0.23383510092370852,
|
9575 |
+
"grad_norm": 0.594439685344696,
|
9576 |
+
"learning_rate": 3.090739702554801e-07,
|
9577 |
+
"loss": 1.3374,
|
9578 |
+
"step": 1367
|
9579 |
+
},
|
9580 |
+
{
|
9581 |
+
"epoch": 0.23400615805679098,
|
9582 |
+
"grad_norm": 0.5832518935203552,
|
9583 |
+
"learning_rate": 2.968467664194008e-07,
|
9584 |
+
"loss": 1.2305,
|
9585 |
+
"step": 1368
|
9586 |
+
},
|
9587 |
+
{
|
9588 |
+
"epoch": 0.23417721518987342,
|
9589 |
+
"grad_norm": 0.6082518696784973,
|
9590 |
+
"learning_rate": 2.8486560734009573e-07,
|
9591 |
+
"loss": 1.1496,
|
9592 |
+
"step": 1369
|
9593 |
+
},
|
9594 |
+
{
|
9595 |
+
"epoch": 0.23434827232295588,
|
9596 |
+
"grad_norm": 0.6523653864860535,
|
9597 |
+
"learning_rate": 2.7313055232769947e-07,
|
9598 |
+
"loss": 1.3162,
|
9599 |
+
"step": 1370
|
9600 |
+
},
|
9601 |
+
{
|
9602 |
+
"epoch": 0.23451932945603832,
|
9603 |
+
"grad_norm": 0.7001534700393677,
|
9604 |
+
"learning_rate": 2.6164165947408226e-07,
|
9605 |
+
"loss": 1.2608,
|
9606 |
+
"step": 1371
|
9607 |
+
},
|
9608 |
+
{
|
9609 |
+
"epoch": 0.23469038658912075,
|
9610 |
+
"grad_norm": 0.6603479385375977,
|
9611 |
+
"learning_rate": 2.5039898565253903e-07,
|
9612 |
+
"loss": 1.054,
|
9613 |
+
"step": 1372
|
9614 |
+
},
|
9615 |
+
{
|
9616 |
+
"epoch": 0.23486144372220322,
|
9617 |
+
"grad_norm": 0.7289173007011414,
|
9618 |
+
"learning_rate": 2.3940258651749536e-07,
|
9619 |
+
"loss": 0.9877,
|
9620 |
+
"step": 1373
|
9621 |
+
},
|
9622 |
+
{
|
9623 |
+
"epoch": 0.23503250085528565,
|
9624 |
+
"grad_norm": 0.7646000385284424,
|
9625 |
+
"learning_rate": 2.2865251650426857e-07,
|
9626 |
+
"loss": 1.262,
|
9627 |
+
"step": 1374
|
9628 |
+
},
|
9629 |
+
{
|
9630 |
+
"epoch": 0.23520355798836812,
|
9631 |
+
"grad_norm": 0.6895982027053833,
|
9632 |
+
"learning_rate": 2.1814882882875698e-07,
|
9633 |
+
"loss": 1.0696,
|
9634 |
+
"step": 1375
|
9635 |
+
},
|
9636 |
+
{
|
9637 |
+
"epoch": 0.23537461512145055,
|
9638 |
+
"grad_norm": 0.7753621339797974,
|
9639 |
+
"learning_rate": 2.078915754872013e-07,
|
9640 |
+
"loss": 1.1467,
|
9641 |
+
"step": 1376
|
9642 |
+
},
|
9643 |
+
{
|
9644 |
+
"epoch": 0.23554567225453302,
|
9645 |
+
"grad_norm": 0.7920718193054199,
|
9646 |
+
"learning_rate": 1.9788080725593462e-07,
|
9647 |
+
"loss": 1.1686,
|
9648 |
+
"step": 1377
|
9649 |
+
},
|
9650 |
+
{
|
9651 |
+
"epoch": 0.23571672938761545,
|
9652 |
+
"grad_norm": 0.93338543176651,
|
9653 |
+
"learning_rate": 1.8811657369110504e-07,
|
9654 |
+
"loss": 1.2003,
|
9655 |
+
"step": 1378
|
9656 |
+
},
|
9657 |
+
{
|
9658 |
+
"epoch": 0.23588778652069792,
|
9659 |
+
"grad_norm": 0.8253167867660522,
|
9660 |
+
"learning_rate": 1.7859892312844795e-07,
|
9661 |
+
"loss": 1.1813,
|
9662 |
+
"step": 1379
|
9663 |
+
},
|
9664 |
+
{
|
9665 |
+
"epoch": 0.23605884365378035,
|
9666 |
+
"grad_norm": 0.9050092697143555,
|
9667 |
+
"learning_rate": 1.6932790268304743e-07,
|
9668 |
+
"loss": 1.0717,
|
9669 |
+
"step": 1380
|
9670 |
+
},
|
9671 |
+
{
|
9672 |
+
"epoch": 0.23622990078686282,
|
9673 |
+
"grad_norm": 0.9545491933822632,
|
9674 |
+
"learning_rate": 1.6030355824908639e-07,
|
9675 |
+
"loss": 1.1853,
|
9676 |
+
"step": 1381
|
9677 |
+
},
|
9678 |
+
{
|
9679 |
+
"epoch": 0.23640095791994525,
|
9680 |
+
"grad_norm": 0.9502745866775513,
|
9681 |
+
"learning_rate": 1.5152593449964115e-07,
|
9682 |
+
"loss": 1.1989,
|
9683 |
+
"step": 1382
|
9684 |
+
},
|
9685 |
+
{
|
9686 |
+
"epoch": 0.23657201505302772,
|
9687 |
+
"grad_norm": 1.1270335912704468,
|
9688 |
+
"learning_rate": 1.4299507488644838e-07,
|
9689 |
+
"loss": 1.4991,
|
9690 |
+
"step": 1383
|
9691 |
+
},
|
9692 |
+
{
|
9693 |
+
"epoch": 0.23674307218611015,
|
9694 |
+
"grad_norm": 1.0703458786010742,
|
9695 |
+
"learning_rate": 1.3471102163969407e-07,
|
9696 |
+
"loss": 1.0992,
|
9697 |
+
"step": 1384
|
9698 |
+
},
|
9699 |
+
{
|
9700 |
+
"epoch": 0.23691412931919262,
|
9701 |
+
"grad_norm": 1.2064449787139893,
|
9702 |
+
"learning_rate": 1.2667381576779712e-07,
|
9703 |
+
"loss": 1.2553,
|
9704 |
+
"step": 1385
|
9705 |
+
},
|
9706 |
+
{
|
9707 |
+
"epoch": 0.23708518645227505,
|
9708 |
+
"grad_norm": 1.2369565963745117,
|
9709 |
+
"learning_rate": 1.1888349705721502e-07,
|
9710 |
+
"loss": 1.1318,
|
9711 |
+
"step": 1386
|
9712 |
+
},
|
9713 |
+
{
|
9714 |
+
"epoch": 0.23725624358535752,
|
9715 |
+
"grad_norm": 1.4282811880111694,
|
9716 |
+
"learning_rate": 1.1134010407223838e-07,
|
9717 |
+
"loss": 1.4422,
|
9718 |
+
"step": 1387
|
9719 |
+
},
|
9720 |
+
{
|
9721 |
+
"epoch": 0.23742730071843995,
|
9722 |
+
"grad_norm": 1.3976672887802124,
|
9723 |
+
"learning_rate": 1.0404367415481897e-07,
|
9724 |
+
"loss": 1.4074,
|
9725 |
+
"step": 1388
|
9726 |
+
},
|
9727 |
+
{
|
9728 |
+
"epoch": 0.23759835785152242,
|
9729 |
+
"grad_norm": 1.4005348682403564,
|
9730 |
+
"learning_rate": 9.69942434243587e-08,
|
9731 |
+
"loss": 1.2234,
|
9732 |
+
"step": 1389
|
9733 |
+
},
|
9734 |
+
{
|
9735 |
+
"epoch": 0.23776941498460485,
|
9736 |
+
"grad_norm": 2.1577513217926025,
|
9737 |
+
"learning_rate": 9.019184677753756e-08,
|
9738 |
+
"loss": 1.6203,
|
9739 |
+
"step": 1390
|
9740 |
+
},
|
9741 |
+
{
|
9742 |
+
"epoch": 0.23794047211768732,
|
9743 |
+
"grad_norm": 1.7010389566421509,
|
9744 |
+
"learning_rate": 8.363651788816929e-08,
|
9745 |
+
"loss": 1.4602,
|
9746 |
+
"step": 1391
|
9747 |
+
},
|
9748 |
+
{
|
9749 |
+
"epoch": 0.23811152925076975,
|
9750 |
+
"grad_norm": 1.6112992763519287,
|
9751 |
+
"learning_rate": 7.732828920698487e-08,
|
9752 |
+
"loss": 1.2554,
|
9753 |
+
"step": 1392
|
9754 |
+
},
|
9755 |
+
{
|
9756 |
+
"epoch": 0.23828258638385222,
|
9757 |
+
"grad_norm": 1.6530187129974365,
|
9758 |
+
"learning_rate": 7.126719196152154e-08,
|
9759 |
+
"loss": 0.9881,
|
9760 |
+
"step": 1393
|
9761 |
+
},
|
9762 |
+
{
|
9763 |
+
"epoch": 0.23845364351693465,
|
9764 |
+
"grad_norm": 2.1780290603637695,
|
9765 |
+
"learning_rate": 6.5453256155934e-08,
|
9766 |
+
"loss": 1.6389,
|
9767 |
+
"step": 1394
|
9768 |
+
},
|
9769 |
+
{
|
9770 |
+
"epoch": 0.23862470065001712,
|
9771 |
+
"grad_norm": 1.7187129259109497,
|
9772 |
+
"learning_rate": 5.98865105708557e-08,
|
9773 |
+
"loss": 1.0483,
|
9774 |
+
"step": 1395
|
9775 |
+
},
|
9776 |
+
{
|
9777 |
+
"epoch": 0.23879575778309955,
|
9778 |
+
"grad_norm": 2.0782599449157715,
|
9779 |
+
"learning_rate": 5.4566982763265554e-08,
|
9780 |
+
"loss": 1.4907,
|
9781 |
+
"step": 1396
|
9782 |
+
},
|
9783 |
+
{
|
9784 |
+
"epoch": 0.238966814916182,
|
9785 |
+
"grad_norm": 2.34914231300354,
|
9786 |
+
"learning_rate": 4.949469906633808e-08,
|
9787 |
+
"loss": 1.3049,
|
9788 |
+
"step": 1397
|
9789 |
+
},
|
9790 |
+
{
|
9791 |
+
"epoch": 0.23913787204926445,
|
9792 |
+
"grad_norm": 2.8132481575012207,
|
9793 |
+
"learning_rate": 4.466968458932685e-08,
|
9794 |
+
"loss": 1.291,
|
9795 |
+
"step": 1398
|
9796 |
+
},
|
9797 |
+
{
|
9798 |
+
"epoch": 0.2393089291823469,
|
9799 |
+
"grad_norm": 2.93622088432312,
|
9800 |
+
"learning_rate": 4.0091963217420147e-08,
|
9801 |
+
"loss": 1.335,
|
9802 |
+
"step": 1399
|
9803 |
+
},
|
9804 |
+
{
|
9805 |
+
"epoch": 0.23947998631542936,
|
9806 |
+
"grad_norm": 4.360016345977783,
|
9807 |
+
"learning_rate": 3.5761557611641016e-08,
|
9808 |
+
"loss": 1.7978,
|
9809 |
+
"step": 1400
|
9810 |
+
},
|
9811 |
+
{
|
9812 |
+
"epoch": 0.2396510434485118,
|
9813 |
+
"grad_norm": 0.5031181573867798,
|
9814 |
+
"learning_rate": 3.167848920873628e-08,
|
9815 |
+
"loss": 1.1583,
|
9816 |
+
"step": 1401
|
9817 |
+
},
|
9818 |
+
{
|
9819 |
+
"epoch": 0.23982210058159426,
|
9820 |
+
"grad_norm": 0.5267295241355896,
|
9821 |
+
"learning_rate": 2.784277822104886e-08,
|
9822 |
+
"loss": 1.0758,
|
9823 |
+
"step": 1402
|
9824 |
+
},
|
9825 |
+
{
|
9826 |
+
"epoch": 0.2399931577146767,
|
9827 |
+
"grad_norm": 0.5213353633880615,
|
9828 |
+
"learning_rate": 2.425444363645113e-08,
|
9829 |
+
"loss": 1.1879,
|
9830 |
+
"step": 1403
|
9831 |
+
},
|
9832 |
+
{
|
9833 |
+
"epoch": 0.24016421484775916,
|
9834 |
+
"grad_norm": 0.5531721115112305,
|
9835 |
+
"learning_rate": 2.0913503218211726e-08,
|
9836 |
+
"loss": 1.2648,
|
9837 |
+
"step": 1404
|
9838 |
+
},
|
9839 |
+
{
|
9840 |
+
"epoch": 0.2403352719808416,
|
9841 |
+
"grad_norm": 0.5778745412826538,
|
9842 |
+
"learning_rate": 1.7819973504940025e-08,
|
9843 |
+
"loss": 1.1453,
|
9844 |
+
"step": 1405
|
9845 |
+
},
|
9846 |
+
{
|
9847 |
+
"epoch": 0.24050632911392406,
|
9848 |
+
"grad_norm": 0.5460866689682007,
|
9849 |
+
"learning_rate": 1.497386981048621e-08,
|
9850 |
+
"loss": 1.2523,
|
9851 |
+
"step": 1406
|
9852 |
+
},
|
9853 |
+
{
|
9854 |
+
"epoch": 0.2406773862470065,
|
9855 |
+
"grad_norm": 0.5462162494659424,
|
9856 |
+
"learning_rate": 1.2375206223874669e-08,
|
9857 |
+
"loss": 1.2443,
|
9858 |
+
"step": 1407
|
9859 |
+
},
|
9860 |
+
{
|
9861 |
+
"epoch": 0.24084844338008896,
|
9862 |
+
"grad_norm": 0.5448779463768005,
|
9863 |
+
"learning_rate": 1.002399560923184e-08,
|
9864 |
+
"loss": 1.1947,
|
9865 |
+
"step": 1408
|
9866 |
+
},
|
9867 |
+
{
|
9868 |
+
"epoch": 0.2410195005131714,
|
9869 |
+
"grad_norm": 0.5699012875556946,
|
9870 |
+
"learning_rate": 7.920249605719576e-09,
|
9871 |
+
"loss": 1.362,
|
9872 |
+
"step": 1409
|
9873 |
+
},
|
9874 |
+
{
|
9875 |
+
"epoch": 0.24119055764625386,
|
9876 |
+
"grad_norm": 0.563005268573761,
|
9877 |
+
"learning_rate": 6.063978627479649e-09,
|
9878 |
+
"loss": 1.2649,
|
9879 |
+
"step": 1410
|
9880 |
+
},
|
9881 |
+
{
|
9882 |
+
"epoch": 0.2413616147793363,
|
9883 |
+
"grad_norm": 0.5439906120300293,
|
9884 |
+
"learning_rate": 4.455191863578234e-09,
|
9885 |
+
"loss": 1.2617,
|
9886 |
+
"step": 1411
|
9887 |
+
},
|
9888 |
+
{
|
9889 |
+
"epoch": 0.24153267191241876,
|
9890 |
+
"grad_norm": 0.6309058666229248,
|
9891 |
+
"learning_rate": 3.0938972779726015e-09,
|
9892 |
+
"loss": 1.483,
|
9893 |
+
"step": 1412
|
9894 |
+
},
|
9895 |
+
{
|
9896 |
+
"epoch": 0.2417037290455012,
|
9897 |
+
"grad_norm": 0.5740118622779846,
|
9898 |
+
"learning_rate": 1.980101609455609e-09,
|
9899 |
+
"loss": 1.324,
|
9900 |
+
"step": 1413
|
9901 |
+
},
|
9902 |
+
{
|
9903 |
+
"epoch": 0.24187478617858366,
|
9904 |
+
"grad_norm": 0.598046064376831,
|
9905 |
+
"learning_rate": 1.1138103716279435e-09,
|
9906 |
+
"loss": 1.3101,
|
9907 |
+
"step": 1414
|
9908 |
+
},
|
9909 |
+
{
|
9910 |
+
"epoch": 0.2420458433116661,
|
9911 |
+
"grad_norm": 0.6388869881629944,
|
9912 |
+
"learning_rate": 4.9502785288702e-10,
|
9913 |
+
"loss": 1.3599,
|
9914 |
+
"step": 1415
|
9915 |
+
},
|
9916 |
+
{
|
9917 |
+
"epoch": 0.24221690044474856,
|
9918 |
+
"grad_norm": 0.5941327214241028,
|
9919 |
+
"learning_rate": 1.2375711638257236e-10,
|
9920 |
+
"loss": 1.1234,
|
9921 |
+
"step": 1416
|
9922 |
+
},
|
9923 |
+
{
|
9924 |
+
"epoch": 0.242387957577831,
|
9925 |
+
"grad_norm": 0.5998250246047974,
|
9926 |
+
"learning_rate": 0.0,
|
9927 |
+
"loss": 1.2886,
|
9928 |
+
"step": 1417
|
9929 |
}
|
9930 |
],
|
9931 |
"logging_steps": 1,
|
|
|
9940 |
"should_evaluate": false,
|
9941 |
"should_log": false,
|
9942 |
"should_save": true,
|
9943 |
+
"should_training_stop": true
|
9944 |
},
|
9945 |
"attributes": {}
|
9946 |
}
|
9947 |
},
|
9948 |
+
"total_flos": 1.7989385934864384e+18,
|
9949 |
"train_batch_size": 4,
|
9950 |
"trial_name": null,
|
9951 |
"trial_params": null
|