Training in progress, step 1113, 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 59933632
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a70ff39eb45330b1ddbae0149c139b83df07041d2a2d17a82a8769548637cf05
|
| 3 |
size 59933632
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 31823460
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:49b98ec65f885b1f8233988bab550c3af663d282c2f346bee0e514d624f3a5bd
|
| 3 |
size 31823460
|
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 sha256:2f99bb828da8c7c17d7f02755c7a92aa763a6dcb6d8cb0eb635b68af4ff58eb8
|
| 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 sha256:3ed33cf7a576c85290ba9fb49df664a3aa86d2fd4fa421aec86f72d269d77c0a
|
| 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,
|
|
@@ -5201,6 +5201,2603 @@
|
|
| 5201 |
"learning_rate": 5.0265515372575956e-05,
|
| 5202 |
"loss": 0.0005,
|
| 5203 |
"step": 742
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5204 |
}
|
| 5205 |
],
|
| 5206 |
"logging_steps": 1,
|
|
@@ -5220,7 +7817,7 @@
|
|
| 5220 |
"attributes": {}
|
| 5221 |
}
|
| 5222 |
},
|
| 5223 |
-
"total_flos":
|
| 5224 |
"train_batch_size": 4,
|
| 5225 |
"trial_name": null,
|
| 5226 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.43089430894308944,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 1113,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 5201 |
"learning_rate": 5.0265515372575956e-05,
|
| 5202 |
"loss": 0.0005,
|
| 5203 |
"step": 742
|
| 5204 |
+
},
|
| 5205 |
+
{
|
| 5206 |
+
"epoch": 0.2876500193573364,
|
| 5207 |
+
"grad_norm": 0.001839546370320022,
|
| 5208 |
+
"learning_rate": 5.015930970274277e-05,
|
| 5209 |
+
"loss": 0.0,
|
| 5210 |
+
"step": 743
|
| 5211 |
+
},
|
| 5212 |
+
{
|
| 5213 |
+
"epoch": 0.2880371660859466,
|
| 5214 |
+
"grad_norm": 0.0016663463320583105,
|
| 5215 |
+
"learning_rate": 5.00531033141139e-05,
|
| 5216 |
+
"loss": 0.0,
|
| 5217 |
+
"step": 744
|
| 5218 |
+
},
|
| 5219 |
+
{
|
| 5220 |
+
"epoch": 0.2884243128145567,
|
| 5221 |
+
"grad_norm": 0.0012716924538835883,
|
| 5222 |
+
"learning_rate": 4.99468966858861e-05,
|
| 5223 |
+
"loss": 0.0,
|
| 5224 |
+
"step": 745
|
| 5225 |
+
},
|
| 5226 |
+
{
|
| 5227 |
+
"epoch": 0.28881145954316684,
|
| 5228 |
+
"grad_norm": 0.0038500288501381874,
|
| 5229 |
+
"learning_rate": 4.984069029725723e-05,
|
| 5230 |
+
"loss": 0.0,
|
| 5231 |
+
"step": 746
|
| 5232 |
+
},
|
| 5233 |
+
{
|
| 5234 |
+
"epoch": 0.289198606271777,
|
| 5235 |
+
"grad_norm": 0.3896702229976654,
|
| 5236 |
+
"learning_rate": 4.973448462742405e-05,
|
| 5237 |
+
"loss": 0.0008,
|
| 5238 |
+
"step": 747
|
| 5239 |
+
},
|
| 5240 |
+
{
|
| 5241 |
+
"epoch": 0.28958575300038714,
|
| 5242 |
+
"grad_norm": 0.0010479100747033954,
|
| 5243 |
+
"learning_rate": 4.962828015558006e-05,
|
| 5244 |
+
"loss": 0.0,
|
| 5245 |
+
"step": 748
|
| 5246 |
+
},
|
| 5247 |
+
{
|
| 5248 |
+
"epoch": 0.2899728997289973,
|
| 5249 |
+
"grad_norm": 0.011884471401572227,
|
| 5250 |
+
"learning_rate": 4.952207736091341e-05,
|
| 5251 |
+
"loss": 0.0001,
|
| 5252 |
+
"step": 749
|
| 5253 |
+
},
|
| 5254 |
+
{
|
| 5255 |
+
"epoch": 0.29036004645760743,
|
| 5256 |
+
"grad_norm": 0.2021235078573227,
|
| 5257 |
+
"learning_rate": 4.941587672260461e-05,
|
| 5258 |
+
"loss": 0.0033,
|
| 5259 |
+
"step": 750
|
| 5260 |
+
},
|
| 5261 |
+
{
|
| 5262 |
+
"epoch": 0.29074719318621756,
|
| 5263 |
+
"grad_norm": 0.045500051230192184,
|
| 5264 |
+
"learning_rate": 4.930967871982451e-05,
|
| 5265 |
+
"loss": 0.0055,
|
| 5266 |
+
"step": 751
|
| 5267 |
+
},
|
| 5268 |
+
{
|
| 5269 |
+
"epoch": 0.29113433991482773,
|
| 5270 |
+
"grad_norm": 0.0288743507117033,
|
| 5271 |
+
"learning_rate": 4.9203483831732006e-05,
|
| 5272 |
+
"loss": 0.0007,
|
| 5273 |
+
"step": 752
|
| 5274 |
+
},
|
| 5275 |
+
{
|
| 5276 |
+
"epoch": 0.29152148664343785,
|
| 5277 |
+
"grad_norm": 0.020009569823741913,
|
| 5278 |
+
"learning_rate": 4.909729253747197e-05,
|
| 5279 |
+
"loss": 0.0002,
|
| 5280 |
+
"step": 753
|
| 5281 |
+
},
|
| 5282 |
+
{
|
| 5283 |
+
"epoch": 0.29190863337204803,
|
| 5284 |
+
"grad_norm": 0.006925898138433695,
|
| 5285 |
+
"learning_rate": 4.899110531617311e-05,
|
| 5286 |
+
"loss": 0.0001,
|
| 5287 |
+
"step": 754
|
| 5288 |
+
},
|
| 5289 |
+
{
|
| 5290 |
+
"epoch": 0.29229578010065815,
|
| 5291 |
+
"grad_norm": 0.03738551586866379,
|
| 5292 |
+
"learning_rate": 4.8884922646945656e-05,
|
| 5293 |
+
"loss": 0.0006,
|
| 5294 |
+
"step": 755
|
| 5295 |
+
},
|
| 5296 |
+
{
|
| 5297 |
+
"epoch": 0.2926829268292683,
|
| 5298 |
+
"grad_norm": 0.0020933786872774363,
|
| 5299 |
+
"learning_rate": 4.877874500887938e-05,
|
| 5300 |
+
"loss": 0.0001,
|
| 5301 |
+
"step": 756
|
| 5302 |
+
},
|
| 5303 |
+
{
|
| 5304 |
+
"epoch": 0.29307007355787845,
|
| 5305 |
+
"grad_norm": 0.002445946214720607,
|
| 5306 |
+
"learning_rate": 4.867257288104131e-05,
|
| 5307 |
+
"loss": 0.0,
|
| 5308 |
+
"step": 757
|
| 5309 |
+
},
|
| 5310 |
+
{
|
| 5311 |
+
"epoch": 0.2934572202864886,
|
| 5312 |
+
"grad_norm": 0.01913662999868393,
|
| 5313 |
+
"learning_rate": 4.856640674247363e-05,
|
| 5314 |
+
"loss": 0.0002,
|
| 5315 |
+
"step": 758
|
| 5316 |
+
},
|
| 5317 |
+
{
|
| 5318 |
+
"epoch": 0.2938443670150987,
|
| 5319 |
+
"grad_norm": 0.0019830456003546715,
|
| 5320 |
+
"learning_rate": 4.8460247072191496e-05,
|
| 5321 |
+
"loss": 0.0,
|
| 5322 |
+
"step": 759
|
| 5323 |
+
},
|
| 5324 |
+
{
|
| 5325 |
+
"epoch": 0.2942315137437089,
|
| 5326 |
+
"grad_norm": 0.0007783794426359236,
|
| 5327 |
+
"learning_rate": 4.8354094349180885e-05,
|
| 5328 |
+
"loss": 0.0,
|
| 5329 |
+
"step": 760
|
| 5330 |
+
},
|
| 5331 |
+
{
|
| 5332 |
+
"epoch": 0.294618660472319,
|
| 5333 |
+
"grad_norm": 0.0003296427894383669,
|
| 5334 |
+
"learning_rate": 4.82479490523964e-05,
|
| 5335 |
+
"loss": 0.0,
|
| 5336 |
+
"step": 761
|
| 5337 |
+
},
|
| 5338 |
+
{
|
| 5339 |
+
"epoch": 0.29500580720092917,
|
| 5340 |
+
"grad_norm": 0.0009307400323450565,
|
| 5341 |
+
"learning_rate": 4.8141811660759185e-05,
|
| 5342 |
+
"loss": 0.0,
|
| 5343 |
+
"step": 762
|
| 5344 |
+
},
|
| 5345 |
+
{
|
| 5346 |
+
"epoch": 0.2953929539295393,
|
| 5347 |
+
"grad_norm": 0.0011843329994007945,
|
| 5348 |
+
"learning_rate": 4.8035682653154704e-05,
|
| 5349 |
+
"loss": 0.0,
|
| 5350 |
+
"step": 763
|
| 5351 |
+
},
|
| 5352 |
+
{
|
| 5353 |
+
"epoch": 0.2957801006581494,
|
| 5354 |
+
"grad_norm": 0.001043383264914155,
|
| 5355 |
+
"learning_rate": 4.7929562508430556e-05,
|
| 5356 |
+
"loss": 0.0,
|
| 5357 |
+
"step": 764
|
| 5358 |
+
},
|
| 5359 |
+
{
|
| 5360 |
+
"epoch": 0.2961672473867596,
|
| 5361 |
+
"grad_norm": 0.00028426575590856373,
|
| 5362 |
+
"learning_rate": 4.782345170539441e-05,
|
| 5363 |
+
"loss": 0.0,
|
| 5364 |
+
"step": 765
|
| 5365 |
+
},
|
| 5366 |
+
{
|
| 5367 |
+
"epoch": 0.2965543941153697,
|
| 5368 |
+
"grad_norm": 0.17862246930599213,
|
| 5369 |
+
"learning_rate": 4.7717350722811705e-05,
|
| 5370 |
+
"loss": 0.0013,
|
| 5371 |
+
"step": 766
|
| 5372 |
+
},
|
| 5373 |
+
{
|
| 5374 |
+
"epoch": 0.2969415408439799,
|
| 5375 |
+
"grad_norm": 0.0018550584791228175,
|
| 5376 |
+
"learning_rate": 4.761126003940366e-05,
|
| 5377 |
+
"loss": 0.0,
|
| 5378 |
+
"step": 767
|
| 5379 |
+
},
|
| 5380 |
+
{
|
| 5381 |
+
"epoch": 0.29732868757259,
|
| 5382 |
+
"grad_norm": 0.0009288343135267496,
|
| 5383 |
+
"learning_rate": 4.7505180133844954e-05,
|
| 5384 |
+
"loss": 0.0,
|
| 5385 |
+
"step": 768
|
| 5386 |
+
},
|
| 5387 |
+
{
|
| 5388 |
+
"epoch": 0.29771583430120013,
|
| 5389 |
+
"grad_norm": 0.00047417895984835923,
|
| 5390 |
+
"learning_rate": 4.7399111484761674e-05,
|
| 5391 |
+
"loss": 0.0,
|
| 5392 |
+
"step": 769
|
| 5393 |
+
},
|
| 5394 |
+
{
|
| 5395 |
+
"epoch": 0.2981029810298103,
|
| 5396 |
+
"grad_norm": 0.0017211599042639136,
|
| 5397 |
+
"learning_rate": 4.729305457072913e-05,
|
| 5398 |
+
"loss": 0.0,
|
| 5399 |
+
"step": 770
|
| 5400 |
+
},
|
| 5401 |
+
{
|
| 5402 |
+
"epoch": 0.29849012775842043,
|
| 5403 |
+
"grad_norm": 0.0004953122115693986,
|
| 5404 |
+
"learning_rate": 4.718700987026964e-05,
|
| 5405 |
+
"loss": 0.0,
|
| 5406 |
+
"step": 771
|
| 5407 |
+
},
|
| 5408 |
+
{
|
| 5409 |
+
"epoch": 0.2988772744870306,
|
| 5410 |
+
"grad_norm": 0.0006206472753547132,
|
| 5411 |
+
"learning_rate": 4.708097786185046e-05,
|
| 5412 |
+
"loss": 0.0,
|
| 5413 |
+
"step": 772
|
| 5414 |
+
},
|
| 5415 |
+
{
|
| 5416 |
+
"epoch": 0.29926442121564073,
|
| 5417 |
+
"grad_norm": 0.00035984007990919054,
|
| 5418 |
+
"learning_rate": 4.697495902388154e-05,
|
| 5419 |
+
"loss": 0.0,
|
| 5420 |
+
"step": 773
|
| 5421 |
+
},
|
| 5422 |
+
{
|
| 5423 |
+
"epoch": 0.29965156794425085,
|
| 5424 |
+
"grad_norm": 0.0028558659832924604,
|
| 5425 |
+
"learning_rate": 4.6868953834713456e-05,
|
| 5426 |
+
"loss": 0.0,
|
| 5427 |
+
"step": 774
|
| 5428 |
+
},
|
| 5429 |
+
{
|
| 5430 |
+
"epoch": 0.30003871467286103,
|
| 5431 |
+
"grad_norm": 0.0013832409167662263,
|
| 5432 |
+
"learning_rate": 4.6762962772635125e-05,
|
| 5433 |
+
"loss": 0.0,
|
| 5434 |
+
"step": 775
|
| 5435 |
+
},
|
| 5436 |
+
{
|
| 5437 |
+
"epoch": 0.30042586140147115,
|
| 5438 |
+
"grad_norm": 0.0032838445622473955,
|
| 5439 |
+
"learning_rate": 4.665698631587182e-05,
|
| 5440 |
+
"loss": 0.0,
|
| 5441 |
+
"step": 776
|
| 5442 |
+
},
|
| 5443 |
+
{
|
| 5444 |
+
"epoch": 0.3008130081300813,
|
| 5445 |
+
"grad_norm": 0.0006759579991921782,
|
| 5446 |
+
"learning_rate": 4.655102494258281e-05,
|
| 5447 |
+
"loss": 0.0,
|
| 5448 |
+
"step": 777
|
| 5449 |
+
},
|
| 5450 |
+
{
|
| 5451 |
+
"epoch": 0.30120015485869145,
|
| 5452 |
+
"grad_norm": 0.0005392953753471375,
|
| 5453 |
+
"learning_rate": 4.644507913085942e-05,
|
| 5454 |
+
"loss": 0.0,
|
| 5455 |
+
"step": 778
|
| 5456 |
+
},
|
| 5457 |
+
{
|
| 5458 |
+
"epoch": 0.30158730158730157,
|
| 5459 |
+
"grad_norm": 0.005940568167716265,
|
| 5460 |
+
"learning_rate": 4.633914935872268e-05,
|
| 5461 |
+
"loss": 0.0,
|
| 5462 |
+
"step": 779
|
| 5463 |
+
},
|
| 5464 |
+
{
|
| 5465 |
+
"epoch": 0.30197444831591175,
|
| 5466 |
+
"grad_norm": 0.005116292275488377,
|
| 5467 |
+
"learning_rate": 4.623323610412126e-05,
|
| 5468 |
+
"loss": 0.0,
|
| 5469 |
+
"step": 780
|
| 5470 |
+
},
|
| 5471 |
+
{
|
| 5472 |
+
"epoch": 0.30236159504452187,
|
| 5473 |
+
"grad_norm": 0.0008851262973621488,
|
| 5474 |
+
"learning_rate": 4.612733984492936e-05,
|
| 5475 |
+
"loss": 0.0,
|
| 5476 |
+
"step": 781
|
| 5477 |
+
},
|
| 5478 |
+
{
|
| 5479 |
+
"epoch": 0.302748741773132,
|
| 5480 |
+
"grad_norm": 0.000602389860432595,
|
| 5481 |
+
"learning_rate": 4.602146105894442e-05,
|
| 5482 |
+
"loss": 0.0,
|
| 5483 |
+
"step": 782
|
| 5484 |
+
},
|
| 5485 |
+
{
|
| 5486 |
+
"epoch": 0.30313588850174217,
|
| 5487 |
+
"grad_norm": 0.0011187748750671744,
|
| 5488 |
+
"learning_rate": 4.591560022388511e-05,
|
| 5489 |
+
"loss": 0.0,
|
| 5490 |
+
"step": 783
|
| 5491 |
+
},
|
| 5492 |
+
{
|
| 5493 |
+
"epoch": 0.3035230352303523,
|
| 5494 |
+
"grad_norm": 0.0007747854106128216,
|
| 5495 |
+
"learning_rate": 4.580975781738905e-05,
|
| 5496 |
+
"loss": 0.0,
|
| 5497 |
+
"step": 784
|
| 5498 |
+
},
|
| 5499 |
+
{
|
| 5500 |
+
"epoch": 0.30391018195896247,
|
| 5501 |
+
"grad_norm": 0.021920830011367798,
|
| 5502 |
+
"learning_rate": 4.5703934317010735e-05,
|
| 5503 |
+
"loss": 0.0003,
|
| 5504 |
+
"step": 785
|
| 5505 |
+
},
|
| 5506 |
+
{
|
| 5507 |
+
"epoch": 0.3042973286875726,
|
| 5508 |
+
"grad_norm": 0.0004085338150616735,
|
| 5509 |
+
"learning_rate": 4.559813020021939e-05,
|
| 5510 |
+
"loss": 0.0,
|
| 5511 |
+
"step": 786
|
| 5512 |
+
},
|
| 5513 |
+
{
|
| 5514 |
+
"epoch": 0.3046844754161827,
|
| 5515 |
+
"grad_norm": 0.48235246539115906,
|
| 5516 |
+
"learning_rate": 4.549234594439674e-05,
|
| 5517 |
+
"loss": 0.0031,
|
| 5518 |
+
"step": 787
|
| 5519 |
+
},
|
| 5520 |
+
{
|
| 5521 |
+
"epoch": 0.3050716221447929,
|
| 5522 |
+
"grad_norm": 0.00026683625765144825,
|
| 5523 |
+
"learning_rate": 4.5386582026834906e-05,
|
| 5524 |
+
"loss": 0.0,
|
| 5525 |
+
"step": 788
|
| 5526 |
+
},
|
| 5527 |
+
{
|
| 5528 |
+
"epoch": 0.305458768873403,
|
| 5529 |
+
"grad_norm": 0.007353804539889097,
|
| 5530 |
+
"learning_rate": 4.528083892473424e-05,
|
| 5531 |
+
"loss": 0.0001,
|
| 5532 |
+
"step": 789
|
| 5533 |
+
},
|
| 5534 |
+
{
|
| 5535 |
+
"epoch": 0.3058459156020132,
|
| 5536 |
+
"grad_norm": 0.0011645752238109708,
|
| 5537 |
+
"learning_rate": 4.517511711520121e-05,
|
| 5538 |
+
"loss": 0.0,
|
| 5539 |
+
"step": 790
|
| 5540 |
+
},
|
| 5541 |
+
{
|
| 5542 |
+
"epoch": 0.3062330623306233,
|
| 5543 |
+
"grad_norm": 0.2283429056406021,
|
| 5544 |
+
"learning_rate": 4.506941707524615e-05,
|
| 5545 |
+
"loss": 0.0007,
|
| 5546 |
+
"step": 791
|
| 5547 |
+
},
|
| 5548 |
+
{
|
| 5549 |
+
"epoch": 0.30662020905923343,
|
| 5550 |
+
"grad_norm": 0.0008667742949910462,
|
| 5551 |
+
"learning_rate": 4.496373928178125e-05,
|
| 5552 |
+
"loss": 0.0,
|
| 5553 |
+
"step": 792
|
| 5554 |
+
},
|
| 5555 |
+
{
|
| 5556 |
+
"epoch": 0.3070073557878436,
|
| 5557 |
+
"grad_norm": 0.0012296324130147696,
|
| 5558 |
+
"learning_rate": 4.485808421161823e-05,
|
| 5559 |
+
"loss": 0.0,
|
| 5560 |
+
"step": 793
|
| 5561 |
+
},
|
| 5562 |
+
{
|
| 5563 |
+
"epoch": 0.3073945025164537,
|
| 5564 |
+
"grad_norm": 0.0036023801658302546,
|
| 5565 |
+
"learning_rate": 4.4752452341466395e-05,
|
| 5566 |
+
"loss": 0.0,
|
| 5567 |
+
"step": 794
|
| 5568 |
+
},
|
| 5569 |
+
{
|
| 5570 |
+
"epoch": 0.3077816492450639,
|
| 5571 |
+
"grad_norm": 0.001286810147576034,
|
| 5572 |
+
"learning_rate": 4.4646844147930294e-05,
|
| 5573 |
+
"loss": 0.0,
|
| 5574 |
+
"step": 795
|
| 5575 |
+
},
|
| 5576 |
+
{
|
| 5577 |
+
"epoch": 0.308168795973674,
|
| 5578 |
+
"grad_norm": 0.0008486775914207101,
|
| 5579 |
+
"learning_rate": 4.454126010750766e-05,
|
| 5580 |
+
"loss": 0.0,
|
| 5581 |
+
"step": 796
|
| 5582 |
+
},
|
| 5583 |
+
{
|
| 5584 |
+
"epoch": 0.30855594270228415,
|
| 5585 |
+
"grad_norm": 0.0011814788449555635,
|
| 5586 |
+
"learning_rate": 4.443570069658727e-05,
|
| 5587 |
+
"loss": 0.0,
|
| 5588 |
+
"step": 797
|
| 5589 |
+
},
|
| 5590 |
+
{
|
| 5591 |
+
"epoch": 0.3089430894308943,
|
| 5592 |
+
"grad_norm": 0.011497690342366695,
|
| 5593 |
+
"learning_rate": 4.433016639144676e-05,
|
| 5594 |
+
"loss": 0.0002,
|
| 5595 |
+
"step": 798
|
| 5596 |
+
},
|
| 5597 |
+
{
|
| 5598 |
+
"epoch": 0.30933023615950445,
|
| 5599 |
+
"grad_norm": 0.0019194751512259245,
|
| 5600 |
+
"learning_rate": 4.4224657668250495e-05,
|
| 5601 |
+
"loss": 0.0,
|
| 5602 |
+
"step": 799
|
| 5603 |
+
},
|
| 5604 |
+
{
|
| 5605 |
+
"epoch": 0.3097173828881146,
|
| 5606 |
+
"grad_norm": 1.5372258424758911,
|
| 5607 |
+
"learning_rate": 4.411917500304741e-05,
|
| 5608 |
+
"loss": 0.0304,
|
| 5609 |
+
"step": 800
|
| 5610 |
+
},
|
| 5611 |
+
{
|
| 5612 |
+
"epoch": 0.31010452961672474,
|
| 5613 |
+
"grad_norm": 0.04230941832065582,
|
| 5614 |
+
"learning_rate": 4.4013718871768855e-05,
|
| 5615 |
+
"loss": 0.0087,
|
| 5616 |
+
"step": 801
|
| 5617 |
+
},
|
| 5618 |
+
{
|
| 5619 |
+
"epoch": 0.31049167634533487,
|
| 5620 |
+
"grad_norm": 0.0006967206718400121,
|
| 5621 |
+
"learning_rate": 4.390828975022652e-05,
|
| 5622 |
+
"loss": 0.0,
|
| 5623 |
+
"step": 802
|
| 5624 |
+
},
|
| 5625 |
+
{
|
| 5626 |
+
"epoch": 0.31087882307394504,
|
| 5627 |
+
"grad_norm": 0.011427176184952259,
|
| 5628 |
+
"learning_rate": 4.380288811411015e-05,
|
| 5629 |
+
"loss": 0.0001,
|
| 5630 |
+
"step": 803
|
| 5631 |
+
},
|
| 5632 |
+
{
|
| 5633 |
+
"epoch": 0.31126596980255516,
|
| 5634 |
+
"grad_norm": 0.007503742817789316,
|
| 5635 |
+
"learning_rate": 4.3697514438985536e-05,
|
| 5636 |
+
"loss": 0.0,
|
| 5637 |
+
"step": 804
|
| 5638 |
+
},
|
| 5639 |
+
{
|
| 5640 |
+
"epoch": 0.3116531165311653,
|
| 5641 |
+
"grad_norm": 0.005879178177565336,
|
| 5642 |
+
"learning_rate": 4.359216920029227e-05,
|
| 5643 |
+
"loss": 0.0001,
|
| 5644 |
+
"step": 805
|
| 5645 |
+
},
|
| 5646 |
+
{
|
| 5647 |
+
"epoch": 0.31204026325977546,
|
| 5648 |
+
"grad_norm": 0.006521875504404306,
|
| 5649 |
+
"learning_rate": 4.3486852873341676e-05,
|
| 5650 |
+
"loss": 0.0001,
|
| 5651 |
+
"step": 806
|
| 5652 |
+
},
|
| 5653 |
+
{
|
| 5654 |
+
"epoch": 0.3124274099883856,
|
| 5655 |
+
"grad_norm": 0.0018767932197079062,
|
| 5656 |
+
"learning_rate": 4.33815659333146e-05,
|
| 5657 |
+
"loss": 0.0,
|
| 5658 |
+
"step": 807
|
| 5659 |
+
},
|
| 5660 |
+
{
|
| 5661 |
+
"epoch": 0.31281455671699576,
|
| 5662 |
+
"grad_norm": 0.006302808877080679,
|
| 5663 |
+
"learning_rate": 4.327630885525932e-05,
|
| 5664 |
+
"loss": 0.0001,
|
| 5665 |
+
"step": 808
|
| 5666 |
+
},
|
| 5667 |
+
{
|
| 5668 |
+
"epoch": 0.3132017034456059,
|
| 5669 |
+
"grad_norm": 0.0005288548418320715,
|
| 5670 |
+
"learning_rate": 4.3171082114089336e-05,
|
| 5671 |
+
"loss": 0.0,
|
| 5672 |
+
"step": 809
|
| 5673 |
+
},
|
| 5674 |
+
{
|
| 5675 |
+
"epoch": 0.313588850174216,
|
| 5676 |
+
"grad_norm": 0.000991316745057702,
|
| 5677 |
+
"learning_rate": 4.306588618458134e-05,
|
| 5678 |
+
"loss": 0.0,
|
| 5679 |
+
"step": 810
|
| 5680 |
+
},
|
| 5681 |
+
{
|
| 5682 |
+
"epoch": 0.3139759969028262,
|
| 5683 |
+
"grad_norm": 0.00023478269577026367,
|
| 5684 |
+
"learning_rate": 4.2960721541372964e-05,
|
| 5685 |
+
"loss": 0.0,
|
| 5686 |
+
"step": 811
|
| 5687 |
+
},
|
| 5688 |
+
{
|
| 5689 |
+
"epoch": 0.3143631436314363,
|
| 5690 |
+
"grad_norm": 0.007332727778702974,
|
| 5691 |
+
"learning_rate": 4.285558865896065e-05,
|
| 5692 |
+
"loss": 0.0,
|
| 5693 |
+
"step": 812
|
| 5694 |
+
},
|
| 5695 |
+
{
|
| 5696 |
+
"epoch": 0.3147502903600465,
|
| 5697 |
+
"grad_norm": 0.0019596214406192303,
|
| 5698 |
+
"learning_rate": 4.275048801169759e-05,
|
| 5699 |
+
"loss": 0.0,
|
| 5700 |
+
"step": 813
|
| 5701 |
+
},
|
| 5702 |
+
{
|
| 5703 |
+
"epoch": 0.3151374370886566,
|
| 5704 |
+
"grad_norm": 0.0015021011931821704,
|
| 5705 |
+
"learning_rate": 4.2645420073791496e-05,
|
| 5706 |
+
"loss": 0.0,
|
| 5707 |
+
"step": 814
|
| 5708 |
+
},
|
| 5709 |
+
{
|
| 5710 |
+
"epoch": 0.3155245838172667,
|
| 5711 |
+
"grad_norm": 0.0009233247837983072,
|
| 5712 |
+
"learning_rate": 4.254038531930253e-05,
|
| 5713 |
+
"loss": 0.0,
|
| 5714 |
+
"step": 815
|
| 5715 |
+
},
|
| 5716 |
+
{
|
| 5717 |
+
"epoch": 0.3159117305458769,
|
| 5718 |
+
"grad_norm": 0.00019637109653558582,
|
| 5719 |
+
"learning_rate": 4.2435384222141085e-05,
|
| 5720 |
+
"loss": 0.0,
|
| 5721 |
+
"step": 816
|
| 5722 |
+
},
|
| 5723 |
+
{
|
| 5724 |
+
"epoch": 0.316298877274487,
|
| 5725 |
+
"grad_norm": 0.00023914426856208593,
|
| 5726 |
+
"learning_rate": 4.233041725606572e-05,
|
| 5727 |
+
"loss": 0.0,
|
| 5728 |
+
"step": 817
|
| 5729 |
+
},
|
| 5730 |
+
{
|
| 5731 |
+
"epoch": 0.3166860240030972,
|
| 5732 |
+
"grad_norm": 0.00042257452150806785,
|
| 5733 |
+
"learning_rate": 4.222548489468105e-05,
|
| 5734 |
+
"loss": 0.0,
|
| 5735 |
+
"step": 818
|
| 5736 |
+
},
|
| 5737 |
+
{
|
| 5738 |
+
"epoch": 0.3170731707317073,
|
| 5739 |
+
"grad_norm": 0.0009111497784033418,
|
| 5740 |
+
"learning_rate": 4.2120587611435445e-05,
|
| 5741 |
+
"loss": 0.0,
|
| 5742 |
+
"step": 819
|
| 5743 |
+
},
|
| 5744 |
+
{
|
| 5745 |
+
"epoch": 0.31746031746031744,
|
| 5746 |
+
"grad_norm": 0.0006815852248109877,
|
| 5747 |
+
"learning_rate": 4.201572587961911e-05,
|
| 5748 |
+
"loss": 0.0,
|
| 5749 |
+
"step": 820
|
| 5750 |
+
},
|
| 5751 |
+
{
|
| 5752 |
+
"epoch": 0.3178474641889276,
|
| 5753 |
+
"grad_norm": 0.0004083358508069068,
|
| 5754 |
+
"learning_rate": 4.1910900172361764e-05,
|
| 5755 |
+
"loss": 0.0,
|
| 5756 |
+
"step": 821
|
| 5757 |
+
},
|
| 5758 |
+
{
|
| 5759 |
+
"epoch": 0.31823461091753774,
|
| 5760 |
+
"grad_norm": 0.001000471180304885,
|
| 5761 |
+
"learning_rate": 4.1806110962630644e-05,
|
| 5762 |
+
"loss": 0.0,
|
| 5763 |
+
"step": 822
|
| 5764 |
+
},
|
| 5765 |
+
{
|
| 5766 |
+
"epoch": 0.31862175764614786,
|
| 5767 |
+
"grad_norm": 0.0013421968324109912,
|
| 5768 |
+
"learning_rate": 4.1701358723228266e-05,
|
| 5769 |
+
"loss": 0.0,
|
| 5770 |
+
"step": 823
|
| 5771 |
+
},
|
| 5772 |
+
{
|
| 5773 |
+
"epoch": 0.31900890437475804,
|
| 5774 |
+
"grad_norm": 0.0023856195621192455,
|
| 5775 |
+
"learning_rate": 4.159664392679039e-05,
|
| 5776 |
+
"loss": 0.0,
|
| 5777 |
+
"step": 824
|
| 5778 |
+
},
|
| 5779 |
+
{
|
| 5780 |
+
"epoch": 0.31939605110336816,
|
| 5781 |
+
"grad_norm": 0.000275084690656513,
|
| 5782 |
+
"learning_rate": 4.1491967045783755e-05,
|
| 5783 |
+
"loss": 0.0,
|
| 5784 |
+
"step": 825
|
| 5785 |
+
},
|
| 5786 |
+
{
|
| 5787 |
+
"epoch": 0.31978319783197834,
|
| 5788 |
+
"grad_norm": 0.000643594772554934,
|
| 5789 |
+
"learning_rate": 4.138732855250412e-05,
|
| 5790 |
+
"loss": 0.0,
|
| 5791 |
+
"step": 826
|
| 5792 |
+
},
|
| 5793 |
+
{
|
| 5794 |
+
"epoch": 0.32017034456058846,
|
| 5795 |
+
"grad_norm": 0.00018649610865395516,
|
| 5796 |
+
"learning_rate": 4.128272891907401e-05,
|
| 5797 |
+
"loss": 0.0,
|
| 5798 |
+
"step": 827
|
| 5799 |
+
},
|
| 5800 |
+
{
|
| 5801 |
+
"epoch": 0.3205574912891986,
|
| 5802 |
+
"grad_norm": 0.004007772076874971,
|
| 5803 |
+
"learning_rate": 4.117816861744057e-05,
|
| 5804 |
+
"loss": 0.0,
|
| 5805 |
+
"step": 828
|
| 5806 |
+
},
|
| 5807 |
+
{
|
| 5808 |
+
"epoch": 0.32094463801780876,
|
| 5809 |
+
"grad_norm": 0.000278328952845186,
|
| 5810 |
+
"learning_rate": 4.107364811937355e-05,
|
| 5811 |
+
"loss": 0.0,
|
| 5812 |
+
"step": 829
|
| 5813 |
+
},
|
| 5814 |
+
{
|
| 5815 |
+
"epoch": 0.3213317847464189,
|
| 5816 |
+
"grad_norm": 0.0021861272398382425,
|
| 5817 |
+
"learning_rate": 4.096916789646305e-05,
|
| 5818 |
+
"loss": 0.0,
|
| 5819 |
+
"step": 830
|
| 5820 |
+
},
|
| 5821 |
+
{
|
| 5822 |
+
"epoch": 0.32171893147502906,
|
| 5823 |
+
"grad_norm": 0.00044286204501986504,
|
| 5824 |
+
"learning_rate": 4.0864728420117505e-05,
|
| 5825 |
+
"loss": 0.0,
|
| 5826 |
+
"step": 831
|
| 5827 |
+
},
|
| 5828 |
+
{
|
| 5829 |
+
"epoch": 0.3221060782036392,
|
| 5830 |
+
"grad_norm": 0.00034147806582041085,
|
| 5831 |
+
"learning_rate": 4.0760330161561464e-05,
|
| 5832 |
+
"loss": 0.0,
|
| 5833 |
+
"step": 832
|
| 5834 |
+
},
|
| 5835 |
+
{
|
| 5836 |
+
"epoch": 0.3224932249322493,
|
| 5837 |
+
"grad_norm": 0.001119753229431808,
|
| 5838 |
+
"learning_rate": 4.065597359183348e-05,
|
| 5839 |
+
"loss": 0.0,
|
| 5840 |
+
"step": 833
|
| 5841 |
+
},
|
| 5842 |
+
{
|
| 5843 |
+
"epoch": 0.3228803716608595,
|
| 5844 |
+
"grad_norm": 0.0005893710185773671,
|
| 5845 |
+
"learning_rate": 4.055165918178412e-05,
|
| 5846 |
+
"loss": 0.0,
|
| 5847 |
+
"step": 834
|
| 5848 |
+
},
|
| 5849 |
+
{
|
| 5850 |
+
"epoch": 0.3232675183894696,
|
| 5851 |
+
"grad_norm": 0.013795977458357811,
|
| 5852 |
+
"learning_rate": 4.04473874020736e-05,
|
| 5853 |
+
"loss": 0.0001,
|
| 5854 |
+
"step": 835
|
| 5855 |
+
},
|
| 5856 |
+
{
|
| 5857 |
+
"epoch": 0.3236546651180798,
|
| 5858 |
+
"grad_norm": 0.000739909999538213,
|
| 5859 |
+
"learning_rate": 4.034315872316988e-05,
|
| 5860 |
+
"loss": 0.0,
|
| 5861 |
+
"step": 836
|
| 5862 |
+
},
|
| 5863 |
+
{
|
| 5864 |
+
"epoch": 0.3240418118466899,
|
| 5865 |
+
"grad_norm": 0.0003141966590192169,
|
| 5866 |
+
"learning_rate": 4.0238973615346395e-05,
|
| 5867 |
+
"loss": 0.0,
|
| 5868 |
+
"step": 837
|
| 5869 |
+
},
|
| 5870 |
+
{
|
| 5871 |
+
"epoch": 0.3244289585753,
|
| 5872 |
+
"grad_norm": 0.0013074502348899841,
|
| 5873 |
+
"learning_rate": 4.0134832548680006e-05,
|
| 5874 |
+
"loss": 0.0,
|
| 5875 |
+
"step": 838
|
| 5876 |
+
},
|
| 5877 |
+
{
|
| 5878 |
+
"epoch": 0.3248161053039102,
|
| 5879 |
+
"grad_norm": 0.0003308141021989286,
|
| 5880 |
+
"learning_rate": 4.003073599304889e-05,
|
| 5881 |
+
"loss": 0.0,
|
| 5882 |
+
"step": 839
|
| 5883 |
+
},
|
| 5884 |
+
{
|
| 5885 |
+
"epoch": 0.3252032520325203,
|
| 5886 |
+
"grad_norm": 0.0002786907716654241,
|
| 5887 |
+
"learning_rate": 3.9926684418130355e-05,
|
| 5888 |
+
"loss": 0.0,
|
| 5889 |
+
"step": 840
|
| 5890 |
+
},
|
| 5891 |
+
{
|
| 5892 |
+
"epoch": 0.3255903987611305,
|
| 5893 |
+
"grad_norm": 0.008617997169494629,
|
| 5894 |
+
"learning_rate": 3.982267829339877e-05,
|
| 5895 |
+
"loss": 0.0,
|
| 5896 |
+
"step": 841
|
| 5897 |
+
},
|
| 5898 |
+
{
|
| 5899 |
+
"epoch": 0.3259775454897406,
|
| 5900 |
+
"grad_norm": 0.004956826567649841,
|
| 5901 |
+
"learning_rate": 3.971871808812348e-05,
|
| 5902 |
+
"loss": 0.0001,
|
| 5903 |
+
"step": 842
|
| 5904 |
+
},
|
| 5905 |
+
{
|
| 5906 |
+
"epoch": 0.32636469221835074,
|
| 5907 |
+
"grad_norm": 0.0003836003306787461,
|
| 5908 |
+
"learning_rate": 3.961480427136656e-05,
|
| 5909 |
+
"loss": 0.0,
|
| 5910 |
+
"step": 843
|
| 5911 |
+
},
|
| 5912 |
+
{
|
| 5913 |
+
"epoch": 0.3267518389469609,
|
| 5914 |
+
"grad_norm": 0.0010854278225451708,
|
| 5915 |
+
"learning_rate": 3.951093731198086e-05,
|
| 5916 |
+
"loss": 0.0,
|
| 5917 |
+
"step": 844
|
| 5918 |
+
},
|
| 5919 |
+
{
|
| 5920 |
+
"epoch": 0.32713898567557104,
|
| 5921 |
+
"grad_norm": 0.021701844409108162,
|
| 5922 |
+
"learning_rate": 3.9407117678607755e-05,
|
| 5923 |
+
"loss": 0.0002,
|
| 5924 |
+
"step": 845
|
| 5925 |
+
},
|
| 5926 |
+
{
|
| 5927 |
+
"epoch": 0.32752613240418116,
|
| 5928 |
+
"grad_norm": 0.0007447968819178641,
|
| 5929 |
+
"learning_rate": 3.930334583967514e-05,
|
| 5930 |
+
"loss": 0.0,
|
| 5931 |
+
"step": 846
|
| 5932 |
+
},
|
| 5933 |
+
{
|
| 5934 |
+
"epoch": 0.32791327913279134,
|
| 5935 |
+
"grad_norm": 0.00031065259827300906,
|
| 5936 |
+
"learning_rate": 3.919962226339521e-05,
|
| 5937 |
+
"loss": 0.0,
|
| 5938 |
+
"step": 847
|
| 5939 |
+
},
|
| 5940 |
+
{
|
| 5941 |
+
"epoch": 0.32830042586140146,
|
| 5942 |
+
"grad_norm": 0.003032098524272442,
|
| 5943 |
+
"learning_rate": 3.9095947417762465e-05,
|
| 5944 |
+
"loss": 0.0,
|
| 5945 |
+
"step": 848
|
| 5946 |
+
},
|
| 5947 |
+
{
|
| 5948 |
+
"epoch": 0.32868757259001163,
|
| 5949 |
+
"grad_norm": 0.0012839855626225471,
|
| 5950 |
+
"learning_rate": 3.899232177055145e-05,
|
| 5951 |
+
"loss": 0.0,
|
| 5952 |
+
"step": 849
|
| 5953 |
+
},
|
| 5954 |
+
{
|
| 5955 |
+
"epoch": 0.32907471931862176,
|
| 5956 |
+
"grad_norm": 0.00046888040378689766,
|
| 5957 |
+
"learning_rate": 3.888874578931482e-05,
|
| 5958 |
+
"loss": 0.0,
|
| 5959 |
+
"step": 850
|
| 5960 |
+
},
|
| 5961 |
+
{
|
| 5962 |
+
"epoch": 0.3294618660472319,
|
| 5963 |
+
"grad_norm": 0.03994712233543396,
|
| 5964 |
+
"learning_rate": 3.87852199413811e-05,
|
| 5965 |
+
"loss": 0.0035,
|
| 5966 |
+
"step": 851
|
| 5967 |
+
},
|
| 5968 |
+
{
|
| 5969 |
+
"epoch": 0.32984901277584205,
|
| 5970 |
+
"grad_norm": 0.00314146070741117,
|
| 5971 |
+
"learning_rate": 3.8681744693852595e-05,
|
| 5972 |
+
"loss": 0.0001,
|
| 5973 |
+
"step": 852
|
| 5974 |
+
},
|
| 5975 |
+
{
|
| 5976 |
+
"epoch": 0.3302361595044522,
|
| 5977 |
+
"grad_norm": 0.013279520906507969,
|
| 5978 |
+
"learning_rate": 3.857832051360335e-05,
|
| 5979 |
+
"loss": 0.0003,
|
| 5980 |
+
"step": 853
|
| 5981 |
+
},
|
| 5982 |
+
{
|
| 5983 |
+
"epoch": 0.33062330623306235,
|
| 5984 |
+
"grad_norm": 0.028156403452157974,
|
| 5985 |
+
"learning_rate": 3.847494786727694e-05,
|
| 5986 |
+
"loss": 0.0005,
|
| 5987 |
+
"step": 854
|
| 5988 |
+
},
|
| 5989 |
+
{
|
| 5990 |
+
"epoch": 0.3310104529616725,
|
| 5991 |
+
"grad_norm": 0.038008369505405426,
|
| 5992 |
+
"learning_rate": 3.8371627221284495e-05,
|
| 5993 |
+
"loss": 0.0006,
|
| 5994 |
+
"step": 855
|
| 5995 |
+
},
|
| 5996 |
+
{
|
| 5997 |
+
"epoch": 0.3313975996902826,
|
| 5998 |
+
"grad_norm": 0.006248411722481251,
|
| 5999 |
+
"learning_rate": 3.826835904180244e-05,
|
| 6000 |
+
"loss": 0.0001,
|
| 6001 |
+
"step": 856
|
| 6002 |
+
},
|
| 6003 |
+
{
|
| 6004 |
+
"epoch": 0.3317847464188928,
|
| 6005 |
+
"grad_norm": 0.00033054646337404847,
|
| 6006 |
+
"learning_rate": 3.816514379477054e-05,
|
| 6007 |
+
"loss": 0.0,
|
| 6008 |
+
"step": 857
|
| 6009 |
+
},
|
| 6010 |
+
{
|
| 6011 |
+
"epoch": 0.3321718931475029,
|
| 6012 |
+
"grad_norm": 0.0008822673698887229,
|
| 6013 |
+
"learning_rate": 3.8061981945889726e-05,
|
| 6014 |
+
"loss": 0.0,
|
| 6015 |
+
"step": 858
|
| 6016 |
+
},
|
| 6017 |
+
{
|
| 6018 |
+
"epoch": 0.33255903987611307,
|
| 6019 |
+
"grad_norm": 0.03951540216803551,
|
| 6020 |
+
"learning_rate": 3.795887396061995e-05,
|
| 6021 |
+
"loss": 0.0004,
|
| 6022 |
+
"step": 859
|
| 6023 |
+
},
|
| 6024 |
+
{
|
| 6025 |
+
"epoch": 0.3329461866047232,
|
| 6026 |
+
"grad_norm": 0.0011091261403635144,
|
| 6027 |
+
"learning_rate": 3.7855820304178204e-05,
|
| 6028 |
+
"loss": 0.0,
|
| 6029 |
+
"step": 860
|
| 6030 |
+
},
|
| 6031 |
+
{
|
| 6032 |
+
"epoch": 0.3333333333333333,
|
| 6033 |
+
"grad_norm": 0.00019632387557066977,
|
| 6034 |
+
"learning_rate": 3.775282144153627e-05,
|
| 6035 |
+
"loss": 0.0,
|
| 6036 |
+
"step": 861
|
| 6037 |
+
},
|
| 6038 |
+
{
|
| 6039 |
+
"epoch": 0.3337204800619435,
|
| 6040 |
+
"grad_norm": 0.015141610987484455,
|
| 6041 |
+
"learning_rate": 3.764987783741879e-05,
|
| 6042 |
+
"loss": 0.0001,
|
| 6043 |
+
"step": 862
|
| 6044 |
+
},
|
| 6045 |
+
{
|
| 6046 |
+
"epoch": 0.3341076267905536,
|
| 6047 |
+
"grad_norm": 0.0024289898574352264,
|
| 6048 |
+
"learning_rate": 3.754698995630101e-05,
|
| 6049 |
+
"loss": 0.0,
|
| 6050 |
+
"step": 863
|
| 6051 |
+
},
|
| 6052 |
+
{
|
| 6053 |
+
"epoch": 0.3344947735191638,
|
| 6054 |
+
"grad_norm": 0.030151624232530594,
|
| 6055 |
+
"learning_rate": 3.7444158262406824e-05,
|
| 6056 |
+
"loss": 0.0001,
|
| 6057 |
+
"step": 864
|
| 6058 |
+
},
|
| 6059 |
+
{
|
| 6060 |
+
"epoch": 0.3348819202477739,
|
| 6061 |
+
"grad_norm": 0.00045968167250975966,
|
| 6062 |
+
"learning_rate": 3.734138321970653e-05,
|
| 6063 |
+
"loss": 0.0,
|
| 6064 |
+
"step": 865
|
| 6065 |
+
},
|
| 6066 |
+
{
|
| 6067 |
+
"epoch": 0.33526906697638403,
|
| 6068 |
+
"grad_norm": 0.0013369093649089336,
|
| 6069 |
+
"learning_rate": 3.723866529191491e-05,
|
| 6070 |
+
"loss": 0.0,
|
| 6071 |
+
"step": 866
|
| 6072 |
+
},
|
| 6073 |
+
{
|
| 6074 |
+
"epoch": 0.3356562137049942,
|
| 6075 |
+
"grad_norm": 0.00015746480494271964,
|
| 6076 |
+
"learning_rate": 3.7136004942489e-05,
|
| 6077 |
+
"loss": 0.0,
|
| 6078 |
+
"step": 867
|
| 6079 |
+
},
|
| 6080 |
+
{
|
| 6081 |
+
"epoch": 0.33604336043360433,
|
| 6082 |
+
"grad_norm": 0.0004616810765583068,
|
| 6083 |
+
"learning_rate": 3.703340263462602e-05,
|
| 6084 |
+
"loss": 0.0,
|
| 6085 |
+
"step": 868
|
| 6086 |
+
},
|
| 6087 |
+
{
|
| 6088 |
+
"epoch": 0.33643050716221445,
|
| 6089 |
+
"grad_norm": 0.000538064050488174,
|
| 6090 |
+
"learning_rate": 3.693085883126137e-05,
|
| 6091 |
+
"loss": 0.0,
|
| 6092 |
+
"step": 869
|
| 6093 |
+
},
|
| 6094 |
+
{
|
| 6095 |
+
"epoch": 0.33681765389082463,
|
| 6096 |
+
"grad_norm": 0.0004215772496536374,
|
| 6097 |
+
"learning_rate": 3.6828373995066436e-05,
|
| 6098 |
+
"loss": 0.0,
|
| 6099 |
+
"step": 870
|
| 6100 |
+
},
|
| 6101 |
+
{
|
| 6102 |
+
"epoch": 0.33720480061943475,
|
| 6103 |
+
"grad_norm": 0.00042134601972065866,
|
| 6104 |
+
"learning_rate": 3.672594858844658e-05,
|
| 6105 |
+
"loss": 0.0,
|
| 6106 |
+
"step": 871
|
| 6107 |
+
},
|
| 6108 |
+
{
|
| 6109 |
+
"epoch": 0.33759194734804493,
|
| 6110 |
+
"grad_norm": 0.004986919928342104,
|
| 6111 |
+
"learning_rate": 3.6623583073538966e-05,
|
| 6112 |
+
"loss": 0.0001,
|
| 6113 |
+
"step": 872
|
| 6114 |
+
},
|
| 6115 |
+
{
|
| 6116 |
+
"epoch": 0.33797909407665505,
|
| 6117 |
+
"grad_norm": 0.0002799661597236991,
|
| 6118 |
+
"learning_rate": 3.65212779122106e-05,
|
| 6119 |
+
"loss": 0.0,
|
| 6120 |
+
"step": 873
|
| 6121 |
+
},
|
| 6122 |
+
{
|
| 6123 |
+
"epoch": 0.3383662408052652,
|
| 6124 |
+
"grad_norm": 0.0004925346001982689,
|
| 6125 |
+
"learning_rate": 3.641903356605614e-05,
|
| 6126 |
+
"loss": 0.0,
|
| 6127 |
+
"step": 874
|
| 6128 |
+
},
|
| 6129 |
+
{
|
| 6130 |
+
"epoch": 0.33875338753387535,
|
| 6131 |
+
"grad_norm": 0.001601546653546393,
|
| 6132 |
+
"learning_rate": 3.631685049639586e-05,
|
| 6133 |
+
"loss": 0.0,
|
| 6134 |
+
"step": 875
|
| 6135 |
+
},
|
| 6136 |
+
{
|
| 6137 |
+
"epoch": 0.33914053426248547,
|
| 6138 |
+
"grad_norm": 0.00030690181301906705,
|
| 6139 |
+
"learning_rate": 3.621472916427354e-05,
|
| 6140 |
+
"loss": 0.0,
|
| 6141 |
+
"step": 876
|
| 6142 |
+
},
|
| 6143 |
+
{
|
| 6144 |
+
"epoch": 0.33952768099109565,
|
| 6145 |
+
"grad_norm": 0.00021768384613096714,
|
| 6146 |
+
"learning_rate": 3.611267003045443e-05,
|
| 6147 |
+
"loss": 0.0,
|
| 6148 |
+
"step": 877
|
| 6149 |
+
},
|
| 6150 |
+
{
|
| 6151 |
+
"epoch": 0.33991482771970577,
|
| 6152 |
+
"grad_norm": 0.020688867196440697,
|
| 6153 |
+
"learning_rate": 3.601067355542312e-05,
|
| 6154 |
+
"loss": 0.0002,
|
| 6155 |
+
"step": 878
|
| 6156 |
+
},
|
| 6157 |
+
{
|
| 6158 |
+
"epoch": 0.3403019744483159,
|
| 6159 |
+
"grad_norm": 0.0002073254290735349,
|
| 6160 |
+
"learning_rate": 3.59087401993815e-05,
|
| 6161 |
+
"loss": 0.0,
|
| 6162 |
+
"step": 879
|
| 6163 |
+
},
|
| 6164 |
+
{
|
| 6165 |
+
"epoch": 0.34068912117692607,
|
| 6166 |
+
"grad_norm": 0.0022969204001128674,
|
| 6167 |
+
"learning_rate": 3.580687042224667e-05,
|
| 6168 |
+
"loss": 0.0,
|
| 6169 |
+
"step": 880
|
| 6170 |
+
},
|
| 6171 |
+
{
|
| 6172 |
+
"epoch": 0.3410762679055362,
|
| 6173 |
+
"grad_norm": 0.000548949814401567,
|
| 6174 |
+
"learning_rate": 3.5705064683648856e-05,
|
| 6175 |
+
"loss": 0.0,
|
| 6176 |
+
"step": 881
|
| 6177 |
+
},
|
| 6178 |
+
{
|
| 6179 |
+
"epoch": 0.34146341463414637,
|
| 6180 |
+
"grad_norm": 0.0054088798351585865,
|
| 6181 |
+
"learning_rate": 3.560332344292938e-05,
|
| 6182 |
+
"loss": 0.0,
|
| 6183 |
+
"step": 882
|
| 6184 |
+
},
|
| 6185 |
+
{
|
| 6186 |
+
"epoch": 0.3418505613627565,
|
| 6187 |
+
"grad_norm": 0.00040069175884127617,
|
| 6188 |
+
"learning_rate": 3.550164715913853e-05,
|
| 6189 |
+
"loss": 0.0,
|
| 6190 |
+
"step": 883
|
| 6191 |
+
},
|
| 6192 |
+
{
|
| 6193 |
+
"epoch": 0.3422377080913666,
|
| 6194 |
+
"grad_norm": 0.0004740517179016024,
|
| 6195 |
+
"learning_rate": 3.540003629103349e-05,
|
| 6196 |
+
"loss": 0.0,
|
| 6197 |
+
"step": 884
|
| 6198 |
+
},
|
| 6199 |
+
{
|
| 6200 |
+
"epoch": 0.3426248548199768,
|
| 6201 |
+
"grad_norm": 0.0020350434351712465,
|
| 6202 |
+
"learning_rate": 3.529849129707633e-05,
|
| 6203 |
+
"loss": 0.0,
|
| 6204 |
+
"step": 885
|
| 6205 |
+
},
|
| 6206 |
+
{
|
| 6207 |
+
"epoch": 0.3430120015485869,
|
| 6208 |
+
"grad_norm": 0.0003776460944209248,
|
| 6209 |
+
"learning_rate": 3.519701263543187e-05,
|
| 6210 |
+
"loss": 0.0,
|
| 6211 |
+
"step": 886
|
| 6212 |
+
},
|
| 6213 |
+
{
|
| 6214 |
+
"epoch": 0.34339914827719703,
|
| 6215 |
+
"grad_norm": 0.001164233428426087,
|
| 6216 |
+
"learning_rate": 3.5095600763965676e-05,
|
| 6217 |
+
"loss": 0.0,
|
| 6218 |
+
"step": 887
|
| 6219 |
+
},
|
| 6220 |
+
{
|
| 6221 |
+
"epoch": 0.3437862950058072,
|
| 6222 |
+
"grad_norm": 0.0004077065095771104,
|
| 6223 |
+
"learning_rate": 3.499425614024191e-05,
|
| 6224 |
+
"loss": 0.0,
|
| 6225 |
+
"step": 888
|
| 6226 |
+
},
|
| 6227 |
+
{
|
| 6228 |
+
"epoch": 0.34417344173441733,
|
| 6229 |
+
"grad_norm": 0.002573886886239052,
|
| 6230 |
+
"learning_rate": 3.489297922152136e-05,
|
| 6231 |
+
"loss": 0.0,
|
| 6232 |
+
"step": 889
|
| 6233 |
+
},
|
| 6234 |
+
{
|
| 6235 |
+
"epoch": 0.3445605884630275,
|
| 6236 |
+
"grad_norm": 0.02411019243299961,
|
| 6237 |
+
"learning_rate": 3.479177046475935e-05,
|
| 6238 |
+
"loss": 0.0002,
|
| 6239 |
+
"step": 890
|
| 6240 |
+
},
|
| 6241 |
+
{
|
| 6242 |
+
"epoch": 0.34494773519163763,
|
| 6243 |
+
"grad_norm": 0.001541433739475906,
|
| 6244 |
+
"learning_rate": 3.4690630326603604e-05,
|
| 6245 |
+
"loss": 0.0,
|
| 6246 |
+
"step": 891
|
| 6247 |
+
},
|
| 6248 |
+
{
|
| 6249 |
+
"epoch": 0.34533488192024775,
|
| 6250 |
+
"grad_norm": 0.05096811428666115,
|
| 6251 |
+
"learning_rate": 3.458955926339228e-05,
|
| 6252 |
+
"loss": 0.0003,
|
| 6253 |
+
"step": 892
|
| 6254 |
+
},
|
| 6255 |
+
{
|
| 6256 |
+
"epoch": 0.3457220286488579,
|
| 6257 |
+
"grad_norm": 0.0001714549434836954,
|
| 6258 |
+
"learning_rate": 3.448855773115185e-05,
|
| 6259 |
+
"loss": 0.0,
|
| 6260 |
+
"step": 893
|
| 6261 |
+
},
|
| 6262 |
+
{
|
| 6263 |
+
"epoch": 0.34610917537746805,
|
| 6264 |
+
"grad_norm": 0.0007654880173504353,
|
| 6265 |
+
"learning_rate": 3.438762618559511e-05,
|
| 6266 |
+
"loss": 0.0,
|
| 6267 |
+
"step": 894
|
| 6268 |
+
},
|
| 6269 |
+
{
|
| 6270 |
+
"epoch": 0.3464963221060782,
|
| 6271 |
+
"grad_norm": 0.34326171875,
|
| 6272 |
+
"learning_rate": 3.428676508211902e-05,
|
| 6273 |
+
"loss": 0.0018,
|
| 6274 |
+
"step": 895
|
| 6275 |
+
},
|
| 6276 |
+
{
|
| 6277 |
+
"epoch": 0.34688346883468835,
|
| 6278 |
+
"grad_norm": 0.003775347489863634,
|
| 6279 |
+
"learning_rate": 3.418597487580277e-05,
|
| 6280 |
+
"loss": 0.0001,
|
| 6281 |
+
"step": 896
|
| 6282 |
+
},
|
| 6283 |
+
{
|
| 6284 |
+
"epoch": 0.34727061556329847,
|
| 6285 |
+
"grad_norm": 0.0011271066032350063,
|
| 6286 |
+
"learning_rate": 3.408525602140561e-05,
|
| 6287 |
+
"loss": 0.0,
|
| 6288 |
+
"step": 897
|
| 6289 |
+
},
|
| 6290 |
+
{
|
| 6291 |
+
"epoch": 0.34765776229190865,
|
| 6292 |
+
"grad_norm": 0.00141782162245363,
|
| 6293 |
+
"learning_rate": 3.3984608973364904e-05,
|
| 6294 |
+
"loss": 0.0,
|
| 6295 |
+
"step": 898
|
| 6296 |
+
},
|
| 6297 |
+
{
|
| 6298 |
+
"epoch": 0.34804490902051877,
|
| 6299 |
+
"grad_norm": 0.006298588123172522,
|
| 6300 |
+
"learning_rate": 3.388403418579401e-05,
|
| 6301 |
+
"loss": 0.0001,
|
| 6302 |
+
"step": 899
|
| 6303 |
+
},
|
| 6304 |
+
{
|
| 6305 |
+
"epoch": 0.34843205574912894,
|
| 6306 |
+
"grad_norm": 0.057084690779447556,
|
| 6307 |
+
"learning_rate": 3.378353211248024e-05,
|
| 6308 |
+
"loss": 0.0006,
|
| 6309 |
+
"step": 900
|
| 6310 |
+
},
|
| 6311 |
+
{
|
| 6312 |
+
"epoch": 0.34881920247773907,
|
| 6313 |
+
"grad_norm": 0.03301633149385452,
|
| 6314 |
+
"learning_rate": 3.368310320688284e-05,
|
| 6315 |
+
"loss": 0.0017,
|
| 6316 |
+
"step": 901
|
| 6317 |
+
},
|
| 6318 |
+
{
|
| 6319 |
+
"epoch": 0.3492063492063492,
|
| 6320 |
+
"grad_norm": 0.03648944944143295,
|
| 6321 |
+
"learning_rate": 3.358274792213091e-05,
|
| 6322 |
+
"loss": 0.001,
|
| 6323 |
+
"step": 902
|
| 6324 |
+
},
|
| 6325 |
+
{
|
| 6326 |
+
"epoch": 0.34959349593495936,
|
| 6327 |
+
"grad_norm": 0.045417048037052155,
|
| 6328 |
+
"learning_rate": 3.34824667110214e-05,
|
| 6329 |
+
"loss": 0.0002,
|
| 6330 |
+
"step": 903
|
| 6331 |
+
},
|
| 6332 |
+
{
|
| 6333 |
+
"epoch": 0.3499806426635695,
|
| 6334 |
+
"grad_norm": 0.0014944113790988922,
|
| 6335 |
+
"learning_rate": 3.338226002601703e-05,
|
| 6336 |
+
"loss": 0.0,
|
| 6337 |
+
"step": 904
|
| 6338 |
+
},
|
| 6339 |
+
{
|
| 6340 |
+
"epoch": 0.35036778939217966,
|
| 6341 |
+
"grad_norm": 0.0017919113161042333,
|
| 6342 |
+
"learning_rate": 3.328212831924424e-05,
|
| 6343 |
+
"loss": 0.0,
|
| 6344 |
+
"step": 905
|
| 6345 |
+
},
|
| 6346 |
+
{
|
| 6347 |
+
"epoch": 0.3507549361207898,
|
| 6348 |
+
"grad_norm": 0.0006082098698243499,
|
| 6349 |
+
"learning_rate": 3.3182072042491244e-05,
|
| 6350 |
+
"loss": 0.0,
|
| 6351 |
+
"step": 906
|
| 6352 |
+
},
|
| 6353 |
+
{
|
| 6354 |
+
"epoch": 0.3511420828493999,
|
| 6355 |
+
"grad_norm": 0.010628577321767807,
|
| 6356 |
+
"learning_rate": 3.308209164720584e-05,
|
| 6357 |
+
"loss": 0.0001,
|
| 6358 |
+
"step": 907
|
| 6359 |
+
},
|
| 6360 |
+
{
|
| 6361 |
+
"epoch": 0.3515292295780101,
|
| 6362 |
+
"grad_norm": 0.08203943073749542,
|
| 6363 |
+
"learning_rate": 3.2982187584493515e-05,
|
| 6364 |
+
"loss": 0.0019,
|
| 6365 |
+
"step": 908
|
| 6366 |
+
},
|
| 6367 |
+
{
|
| 6368 |
+
"epoch": 0.3519163763066202,
|
| 6369 |
+
"grad_norm": 0.0009694083128124475,
|
| 6370 |
+
"learning_rate": 3.2882360305115294e-05,
|
| 6371 |
+
"loss": 0.0,
|
| 6372 |
+
"step": 909
|
| 6373 |
+
},
|
| 6374 |
+
{
|
| 6375 |
+
"epoch": 0.3523035230352303,
|
| 6376 |
+
"grad_norm": 0.00442282622680068,
|
| 6377 |
+
"learning_rate": 3.2782610259485813e-05,
|
| 6378 |
+
"loss": 0.0001,
|
| 6379 |
+
"step": 910
|
| 6380 |
+
},
|
| 6381 |
+
{
|
| 6382 |
+
"epoch": 0.3526906697638405,
|
| 6383 |
+
"grad_norm": 0.08702477067708969,
|
| 6384 |
+
"learning_rate": 3.268293789767118e-05,
|
| 6385 |
+
"loss": 0.0014,
|
| 6386 |
+
"step": 911
|
| 6387 |
+
},
|
| 6388 |
+
{
|
| 6389 |
+
"epoch": 0.3530778164924506,
|
| 6390 |
+
"grad_norm": 0.02207796275615692,
|
| 6391 |
+
"learning_rate": 3.2583343669387046e-05,
|
| 6392 |
+
"loss": 0.0001,
|
| 6393 |
+
"step": 912
|
| 6394 |
+
},
|
| 6395 |
+
{
|
| 6396 |
+
"epoch": 0.3534649632210608,
|
| 6397 |
+
"grad_norm": 0.0006780868279747665,
|
| 6398 |
+
"learning_rate": 3.248382802399649e-05,
|
| 6399 |
+
"loss": 0.0,
|
| 6400 |
+
"step": 913
|
| 6401 |
+
},
|
| 6402 |
+
{
|
| 6403 |
+
"epoch": 0.3538521099496709,
|
| 6404 |
+
"grad_norm": 0.02841506339609623,
|
| 6405 |
+
"learning_rate": 3.238439141050807e-05,
|
| 6406 |
+
"loss": 0.0003,
|
| 6407 |
+
"step": 914
|
| 6408 |
+
},
|
| 6409 |
+
{
|
| 6410 |
+
"epoch": 0.35423925667828104,
|
| 6411 |
+
"grad_norm": 0.00042213572305627167,
|
| 6412 |
+
"learning_rate": 3.228503427757374e-05,
|
| 6413 |
+
"loss": 0.0,
|
| 6414 |
+
"step": 915
|
| 6415 |
+
},
|
| 6416 |
+
{
|
| 6417 |
+
"epoch": 0.3546264034068912,
|
| 6418 |
+
"grad_norm": 0.0002454633649904281,
|
| 6419 |
+
"learning_rate": 3.218575707348681e-05,
|
| 6420 |
+
"loss": 0.0,
|
| 6421 |
+
"step": 916
|
| 6422 |
+
},
|
| 6423 |
+
{
|
| 6424 |
+
"epoch": 0.35501355013550134,
|
| 6425 |
+
"grad_norm": 0.0027822977863252163,
|
| 6426 |
+
"learning_rate": 3.208656024618002e-05,
|
| 6427 |
+
"loss": 0.0,
|
| 6428 |
+
"step": 917
|
| 6429 |
+
},
|
| 6430 |
+
{
|
| 6431 |
+
"epoch": 0.3554006968641115,
|
| 6432 |
+
"grad_norm": 0.020471172407269478,
|
| 6433 |
+
"learning_rate": 3.198744424322338e-05,
|
| 6434 |
+
"loss": 0.0004,
|
| 6435 |
+
"step": 918
|
| 6436 |
+
},
|
| 6437 |
+
{
|
| 6438 |
+
"epoch": 0.35578784359272164,
|
| 6439 |
+
"grad_norm": 0.0007699733250774443,
|
| 6440 |
+
"learning_rate": 3.188840951182229e-05,
|
| 6441 |
+
"loss": 0.0,
|
| 6442 |
+
"step": 919
|
| 6443 |
+
},
|
| 6444 |
+
{
|
| 6445 |
+
"epoch": 0.35617499032133176,
|
| 6446 |
+
"grad_norm": 0.00015266439004335552,
|
| 6447 |
+
"learning_rate": 3.178945649881543e-05,
|
| 6448 |
+
"loss": 0.0,
|
| 6449 |
+
"step": 920
|
| 6450 |
+
},
|
| 6451 |
+
{
|
| 6452 |
+
"epoch": 0.35656213704994194,
|
| 6453 |
+
"grad_norm": 0.00025495782028883696,
|
| 6454 |
+
"learning_rate": 3.169058565067274e-05,
|
| 6455 |
+
"loss": 0.0,
|
| 6456 |
+
"step": 921
|
| 6457 |
+
},
|
| 6458 |
+
{
|
| 6459 |
+
"epoch": 0.35694928377855206,
|
| 6460 |
+
"grad_norm": 8.566460019210353e-05,
|
| 6461 |
+
"learning_rate": 3.159179741349354e-05,
|
| 6462 |
+
"loss": 0.0,
|
| 6463 |
+
"step": 922
|
| 6464 |
+
},
|
| 6465 |
+
{
|
| 6466 |
+
"epoch": 0.35733643050716224,
|
| 6467 |
+
"grad_norm": 0.00040574869490228593,
|
| 6468 |
+
"learning_rate": 3.149309223300428e-05,
|
| 6469 |
+
"loss": 0.0,
|
| 6470 |
+
"step": 923
|
| 6471 |
+
},
|
| 6472 |
+
{
|
| 6473 |
+
"epoch": 0.35772357723577236,
|
| 6474 |
+
"grad_norm": 0.00045708048855885863,
|
| 6475 |
+
"learning_rate": 3.1394470554556764e-05,
|
| 6476 |
+
"loss": 0.0,
|
| 6477 |
+
"step": 924
|
| 6478 |
+
},
|
| 6479 |
+
{
|
| 6480 |
+
"epoch": 0.3581107239643825,
|
| 6481 |
+
"grad_norm": 0.0041862414218485355,
|
| 6482 |
+
"learning_rate": 3.1295932823125985e-05,
|
| 6483 |
+
"loss": 0.0,
|
| 6484 |
+
"step": 925
|
| 6485 |
+
},
|
| 6486 |
+
{
|
| 6487 |
+
"epoch": 0.35849787069299266,
|
| 6488 |
+
"grad_norm": 0.000689912005327642,
|
| 6489 |
+
"learning_rate": 3.1197479483308214e-05,
|
| 6490 |
+
"loss": 0.0,
|
| 6491 |
+
"step": 926
|
| 6492 |
+
},
|
| 6493 |
+
{
|
| 6494 |
+
"epoch": 0.3588850174216028,
|
| 6495 |
+
"grad_norm": 0.003948573488742113,
|
| 6496 |
+
"learning_rate": 3.109911097931891e-05,
|
| 6497 |
+
"loss": 0.0,
|
| 6498 |
+
"step": 927
|
| 6499 |
+
},
|
| 6500 |
+
{
|
| 6501 |
+
"epoch": 0.35927216415021296,
|
| 6502 |
+
"grad_norm": 0.00029136036755517125,
|
| 6503 |
+
"learning_rate": 3.1000827754990766e-05,
|
| 6504 |
+
"loss": 0.0,
|
| 6505 |
+
"step": 928
|
| 6506 |
+
},
|
| 6507 |
+
{
|
| 6508 |
+
"epoch": 0.3596593108788231,
|
| 6509 |
+
"grad_norm": 0.0020344445947557688,
|
| 6510 |
+
"learning_rate": 3.090263025377173e-05,
|
| 6511 |
+
"loss": 0.0,
|
| 6512 |
+
"step": 929
|
| 6513 |
+
},
|
| 6514 |
+
{
|
| 6515 |
+
"epoch": 0.3600464576074332,
|
| 6516 |
+
"grad_norm": 0.0002035750512732193,
|
| 6517 |
+
"learning_rate": 3.080451891872295e-05,
|
| 6518 |
+
"loss": 0.0,
|
| 6519 |
+
"step": 930
|
| 6520 |
+
},
|
| 6521 |
+
{
|
| 6522 |
+
"epoch": 0.3604336043360434,
|
| 6523 |
+
"grad_norm": 0.0010056099854409695,
|
| 6524 |
+
"learning_rate": 3.070649419251681e-05,
|
| 6525 |
+
"loss": 0.0,
|
| 6526 |
+
"step": 931
|
| 6527 |
+
},
|
| 6528 |
+
{
|
| 6529 |
+
"epoch": 0.3608207510646535,
|
| 6530 |
+
"grad_norm": 0.000317671918310225,
|
| 6531 |
+
"learning_rate": 3.06085565174349e-05,
|
| 6532 |
+
"loss": 0.0,
|
| 6533 |
+
"step": 932
|
| 6534 |
+
},
|
| 6535 |
+
{
|
| 6536 |
+
"epoch": 0.3612078977932636,
|
| 6537 |
+
"grad_norm": 0.0009133410057984293,
|
| 6538 |
+
"learning_rate": 3.0510706335366035e-05,
|
| 6539 |
+
"loss": 0.0,
|
| 6540 |
+
"step": 933
|
| 6541 |
+
},
|
| 6542 |
+
{
|
| 6543 |
+
"epoch": 0.3615950445218738,
|
| 6544 |
+
"grad_norm": 0.06602532416582108,
|
| 6545 |
+
"learning_rate": 3.041294408780431e-05,
|
| 6546 |
+
"loss": 0.0003,
|
| 6547 |
+
"step": 934
|
| 6548 |
+
},
|
| 6549 |
+
{
|
| 6550 |
+
"epoch": 0.3619821912504839,
|
| 6551 |
+
"grad_norm": 0.00024213208234868944,
|
| 6552 |
+
"learning_rate": 3.0315270215847013e-05,
|
| 6553 |
+
"loss": 0.0,
|
| 6554 |
+
"step": 935
|
| 6555 |
+
},
|
| 6556 |
+
{
|
| 6557 |
+
"epoch": 0.3623693379790941,
|
| 6558 |
+
"grad_norm": 0.0014147679321467876,
|
| 6559 |
+
"learning_rate": 3.0217685160192734e-05,
|
| 6560 |
+
"loss": 0.0,
|
| 6561 |
+
"step": 936
|
| 6562 |
+
},
|
| 6563 |
+
{
|
| 6564 |
+
"epoch": 0.3627564847077042,
|
| 6565 |
+
"grad_norm": 0.0003020058502443135,
|
| 6566 |
+
"learning_rate": 3.0120189361139273e-05,
|
| 6567 |
+
"loss": 0.0,
|
| 6568 |
+
"step": 937
|
| 6569 |
+
},
|
| 6570 |
+
{
|
| 6571 |
+
"epoch": 0.36314363143631434,
|
| 6572 |
+
"grad_norm": 0.003151725744828582,
|
| 6573 |
+
"learning_rate": 3.002278325858177e-05,
|
| 6574 |
+
"loss": 0.0,
|
| 6575 |
+
"step": 938
|
| 6576 |
+
},
|
| 6577 |
+
{
|
| 6578 |
+
"epoch": 0.3635307781649245,
|
| 6579 |
+
"grad_norm": 0.002396926749497652,
|
| 6580 |
+
"learning_rate": 2.9925467292010644e-05,
|
| 6581 |
+
"loss": 0.0,
|
| 6582 |
+
"step": 939
|
| 6583 |
+
},
|
| 6584 |
+
{
|
| 6585 |
+
"epoch": 0.36391792489353464,
|
| 6586 |
+
"grad_norm": 0.0011632463429123163,
|
| 6587 |
+
"learning_rate": 2.982824190050958e-05,
|
| 6588 |
+
"loss": 0.0,
|
| 6589 |
+
"step": 940
|
| 6590 |
+
},
|
| 6591 |
+
{
|
| 6592 |
+
"epoch": 0.3643050716221448,
|
| 6593 |
+
"grad_norm": 0.2228887379169464,
|
| 6594 |
+
"learning_rate": 2.973110752275366e-05,
|
| 6595 |
+
"loss": 0.001,
|
| 6596 |
+
"step": 941
|
| 6597 |
+
},
|
| 6598 |
+
{
|
| 6599 |
+
"epoch": 0.36469221835075494,
|
| 6600 |
+
"grad_norm": 0.0011429457226768136,
|
| 6601 |
+
"learning_rate": 2.963406459700726e-05,
|
| 6602 |
+
"loss": 0.0,
|
| 6603 |
+
"step": 942
|
| 6604 |
+
},
|
| 6605 |
+
{
|
| 6606 |
+
"epoch": 0.36507936507936506,
|
| 6607 |
+
"grad_norm": 0.00012259121285751462,
|
| 6608 |
+
"learning_rate": 2.9537113561122178e-05,
|
| 6609 |
+
"loss": 0.0,
|
| 6610 |
+
"step": 943
|
| 6611 |
+
},
|
| 6612 |
+
{
|
| 6613 |
+
"epoch": 0.36546651180797524,
|
| 6614 |
+
"grad_norm": 0.000644800195004791,
|
| 6615 |
+
"learning_rate": 2.944025485253557e-05,
|
| 6616 |
+
"loss": 0.0,
|
| 6617 |
+
"step": 944
|
| 6618 |
+
},
|
| 6619 |
+
{
|
| 6620 |
+
"epoch": 0.36585365853658536,
|
| 6621 |
+
"grad_norm": 0.00017125460726674646,
|
| 6622 |
+
"learning_rate": 2.9343488908268034e-05,
|
| 6623 |
+
"loss": 0.0,
|
| 6624 |
+
"step": 945
|
| 6625 |
+
},
|
| 6626 |
+
{
|
| 6627 |
+
"epoch": 0.36624080526519553,
|
| 6628 |
+
"grad_norm": 0.00020638681598939002,
|
| 6629 |
+
"learning_rate": 2.9246816164921655e-05,
|
| 6630 |
+
"loss": 0.0,
|
| 6631 |
+
"step": 946
|
| 6632 |
+
},
|
| 6633 |
+
{
|
| 6634 |
+
"epoch": 0.36662795199380566,
|
| 6635 |
+
"grad_norm": 0.0005838845972903073,
|
| 6636 |
+
"learning_rate": 2.915023705867794e-05,
|
| 6637 |
+
"loss": 0.0,
|
| 6638 |
+
"step": 947
|
| 6639 |
+
},
|
| 6640 |
+
{
|
| 6641 |
+
"epoch": 0.3670150987224158,
|
| 6642 |
+
"grad_norm": 0.000993402791209519,
|
| 6643 |
+
"learning_rate": 2.905375202529594e-05,
|
| 6644 |
+
"loss": 0.0,
|
| 6645 |
+
"step": 948
|
| 6646 |
+
},
|
| 6647 |
+
{
|
| 6648 |
+
"epoch": 0.36740224545102595,
|
| 6649 |
+
"grad_norm": 0.00029620109125971794,
|
| 6650 |
+
"learning_rate": 2.895736150011026e-05,
|
| 6651 |
+
"loss": 0.0,
|
| 6652 |
+
"step": 949
|
| 6653 |
+
},
|
| 6654 |
+
{
|
| 6655 |
+
"epoch": 0.3677893921796361,
|
| 6656 |
+
"grad_norm": 2.658259153366089,
|
| 6657 |
+
"learning_rate": 2.8861065918029085e-05,
|
| 6658 |
+
"loss": 0.0567,
|
| 6659 |
+
"step": 950
|
| 6660 |
+
},
|
| 6661 |
+
{
|
| 6662 |
+
"epoch": 0.3681765389082462,
|
| 6663 |
+
"grad_norm": 0.02834871970117092,
|
| 6664 |
+
"learning_rate": 2.8764865713532234e-05,
|
| 6665 |
+
"loss": 0.0021,
|
| 6666 |
+
"step": 951
|
| 6667 |
+
},
|
| 6668 |
+
{
|
| 6669 |
+
"epoch": 0.3685636856368564,
|
| 6670 |
+
"grad_norm": 0.03441345691680908,
|
| 6671 |
+
"learning_rate": 2.8668761320669156e-05,
|
| 6672 |
+
"loss": 0.0008,
|
| 6673 |
+
"step": 952
|
| 6674 |
+
},
|
| 6675 |
+
{
|
| 6676 |
+
"epoch": 0.3689508323654665,
|
| 6677 |
+
"grad_norm": 0.07143665105104446,
|
| 6678 |
+
"learning_rate": 2.8572753173057003e-05,
|
| 6679 |
+
"loss": 0.0004,
|
| 6680 |
+
"step": 953
|
| 6681 |
+
},
|
| 6682 |
+
{
|
| 6683 |
+
"epoch": 0.3693379790940767,
|
| 6684 |
+
"grad_norm": 0.06644494086503983,
|
| 6685 |
+
"learning_rate": 2.8476841703878772e-05,
|
| 6686 |
+
"loss": 0.0007,
|
| 6687 |
+
"step": 954
|
| 6688 |
+
},
|
| 6689 |
+
{
|
| 6690 |
+
"epoch": 0.3697251258226868,
|
| 6691 |
+
"grad_norm": 0.005595757160335779,
|
| 6692 |
+
"learning_rate": 2.8381027345881102e-05,
|
| 6693 |
+
"loss": 0.0001,
|
| 6694 |
+
"step": 955
|
| 6695 |
+
},
|
| 6696 |
+
{
|
| 6697 |
+
"epoch": 0.3701122725512969,
|
| 6698 |
+
"grad_norm": 0.13141833245754242,
|
| 6699 |
+
"learning_rate": 2.8285310531372573e-05,
|
| 6700 |
+
"loss": 0.0014,
|
| 6701 |
+
"step": 956
|
| 6702 |
+
},
|
| 6703 |
+
{
|
| 6704 |
+
"epoch": 0.3704994192799071,
|
| 6705 |
+
"grad_norm": 0.007661540526896715,
|
| 6706 |
+
"learning_rate": 2.8189691692221627e-05,
|
| 6707 |
+
"loss": 0.0,
|
| 6708 |
+
"step": 957
|
| 6709 |
+
},
|
| 6710 |
+
{
|
| 6711 |
+
"epoch": 0.3708865660085172,
|
| 6712 |
+
"grad_norm": 0.0015809281030669808,
|
| 6713 |
+
"learning_rate": 2.809417125985467e-05,
|
| 6714 |
+
"loss": 0.0001,
|
| 6715 |
+
"step": 958
|
| 6716 |
+
},
|
| 6717 |
+
{
|
| 6718 |
+
"epoch": 0.3712737127371274,
|
| 6719 |
+
"grad_norm": 0.007457795087248087,
|
| 6720 |
+
"learning_rate": 2.7998749665254033e-05,
|
| 6721 |
+
"loss": 0.0001,
|
| 6722 |
+
"step": 959
|
| 6723 |
+
},
|
| 6724 |
+
{
|
| 6725 |
+
"epoch": 0.3716608594657375,
|
| 6726 |
+
"grad_norm": 0.0005789248389191926,
|
| 6727 |
+
"learning_rate": 2.7903427338956177e-05,
|
| 6728 |
+
"loss": 0.0,
|
| 6729 |
+
"step": 960
|
| 6730 |
+
},
|
| 6731 |
+
{
|
| 6732 |
+
"epoch": 0.37204800619434764,
|
| 6733 |
+
"grad_norm": 0.00037478163721971214,
|
| 6734 |
+
"learning_rate": 2.7808204711049624e-05,
|
| 6735 |
+
"loss": 0.0,
|
| 6736 |
+
"step": 961
|
| 6737 |
+
},
|
| 6738 |
+
{
|
| 6739 |
+
"epoch": 0.3724351529229578,
|
| 6740 |
+
"grad_norm": 0.003515935968607664,
|
| 6741 |
+
"learning_rate": 2.771308221117309e-05,
|
| 6742 |
+
"loss": 0.0,
|
| 6743 |
+
"step": 962
|
| 6744 |
+
},
|
| 6745 |
+
{
|
| 6746 |
+
"epoch": 0.37282229965156793,
|
| 6747 |
+
"grad_norm": 0.0019339972641319036,
|
| 6748 |
+
"learning_rate": 2.7618060268513503e-05,
|
| 6749 |
+
"loss": 0.0,
|
| 6750 |
+
"step": 963
|
| 6751 |
+
},
|
| 6752 |
+
{
|
| 6753 |
+
"epoch": 0.3732094463801781,
|
| 6754 |
+
"grad_norm": 0.00024095486151054502,
|
| 6755 |
+
"learning_rate": 2.7523139311804074e-05,
|
| 6756 |
+
"loss": 0.0,
|
| 6757 |
+
"step": 964
|
| 6758 |
+
},
|
| 6759 |
+
{
|
| 6760 |
+
"epoch": 0.37359659310878823,
|
| 6761 |
+
"grad_norm": 0.0007112721796147525,
|
| 6762 |
+
"learning_rate": 2.742831976932242e-05,
|
| 6763 |
+
"loss": 0.0,
|
| 6764 |
+
"step": 965
|
| 6765 |
+
},
|
| 6766 |
+
{
|
| 6767 |
+
"epoch": 0.37398373983739835,
|
| 6768 |
+
"grad_norm": 0.0007284593302756548,
|
| 6769 |
+
"learning_rate": 2.7333602068888496e-05,
|
| 6770 |
+
"loss": 0.0,
|
| 6771 |
+
"step": 966
|
| 6772 |
+
},
|
| 6773 |
+
{
|
| 6774 |
+
"epoch": 0.37437088656600853,
|
| 6775 |
+
"grad_norm": 0.006084626540541649,
|
| 6776 |
+
"learning_rate": 2.7238986637862833e-05,
|
| 6777 |
+
"loss": 0.0,
|
| 6778 |
+
"step": 967
|
| 6779 |
+
},
|
| 6780 |
+
{
|
| 6781 |
+
"epoch": 0.37475803329461865,
|
| 6782 |
+
"grad_norm": 0.000507039949297905,
|
| 6783 |
+
"learning_rate": 2.7144473903144496e-05,
|
| 6784 |
+
"loss": 0.0,
|
| 6785 |
+
"step": 968
|
| 6786 |
+
},
|
| 6787 |
+
{
|
| 6788 |
+
"epoch": 0.37514518002322883,
|
| 6789 |
+
"grad_norm": 0.0019885883666574955,
|
| 6790 |
+
"learning_rate": 2.7050064291169187e-05,
|
| 6791 |
+
"loss": 0.0,
|
| 6792 |
+
"step": 969
|
| 6793 |
+
},
|
| 6794 |
+
{
|
| 6795 |
+
"epoch": 0.37553232675183895,
|
| 6796 |
+
"grad_norm": 0.04670470952987671,
|
| 6797 |
+
"learning_rate": 2.6955758227907335e-05,
|
| 6798 |
+
"loss": 0.0002,
|
| 6799 |
+
"step": 970
|
| 6800 |
+
},
|
| 6801 |
+
{
|
| 6802 |
+
"epoch": 0.3759194734804491,
|
| 6803 |
+
"grad_norm": 0.00021539766748901457,
|
| 6804 |
+
"learning_rate": 2.6861556138862155e-05,
|
| 6805 |
+
"loss": 0.0,
|
| 6806 |
+
"step": 971
|
| 6807 |
+
},
|
| 6808 |
+
{
|
| 6809 |
+
"epoch": 0.37630662020905925,
|
| 6810 |
+
"grad_norm": 0.0003146221279166639,
|
| 6811 |
+
"learning_rate": 2.676745844906776e-05,
|
| 6812 |
+
"loss": 0.0,
|
| 6813 |
+
"step": 972
|
| 6814 |
+
},
|
| 6815 |
+
{
|
| 6816 |
+
"epoch": 0.37669376693766937,
|
| 6817 |
+
"grad_norm": 0.0013595115160569549,
|
| 6818 |
+
"learning_rate": 2.6673465583087176e-05,
|
| 6819 |
+
"loss": 0.0,
|
| 6820 |
+
"step": 973
|
| 6821 |
+
},
|
| 6822 |
+
{
|
| 6823 |
+
"epoch": 0.3770809136662795,
|
| 6824 |
+
"grad_norm": 0.004096793942153454,
|
| 6825 |
+
"learning_rate": 2.65795779650105e-05,
|
| 6826 |
+
"loss": 0.0,
|
| 6827 |
+
"step": 974
|
| 6828 |
+
},
|
| 6829 |
+
{
|
| 6830 |
+
"epoch": 0.37746806039488967,
|
| 6831 |
+
"grad_norm": 0.0003931306710001081,
|
| 6832 |
+
"learning_rate": 2.648579601845295e-05,
|
| 6833 |
+
"loss": 0.0,
|
| 6834 |
+
"step": 975
|
| 6835 |
+
},
|
| 6836 |
+
{
|
| 6837 |
+
"epoch": 0.3778552071234998,
|
| 6838 |
+
"grad_norm": 0.00021391961490735412,
|
| 6839 |
+
"learning_rate": 2.6392120166552992e-05,
|
| 6840 |
+
"loss": 0.0,
|
| 6841 |
+
"step": 976
|
| 6842 |
+
},
|
| 6843 |
+
{
|
| 6844 |
+
"epoch": 0.37824235385210997,
|
| 6845 |
+
"grad_norm": 0.00013036343443673104,
|
| 6846 |
+
"learning_rate": 2.629855083197031e-05,
|
| 6847 |
+
"loss": 0.0,
|
| 6848 |
+
"step": 977
|
| 6849 |
+
},
|
| 6850 |
+
{
|
| 6851 |
+
"epoch": 0.3786295005807201,
|
| 6852 |
+
"grad_norm": 0.00223020208068192,
|
| 6853 |
+
"learning_rate": 2.6205088436884095e-05,
|
| 6854 |
+
"loss": 0.0,
|
| 6855 |
+
"step": 978
|
| 6856 |
+
},
|
| 6857 |
+
{
|
| 6858 |
+
"epoch": 0.3790166473093302,
|
| 6859 |
+
"grad_norm": 0.0004039799969177693,
|
| 6860 |
+
"learning_rate": 2.6111733402991e-05,
|
| 6861 |
+
"loss": 0.0,
|
| 6862 |
+
"step": 979
|
| 6863 |
+
},
|
| 6864 |
+
{
|
| 6865 |
+
"epoch": 0.3794037940379404,
|
| 6866 |
+
"grad_norm": 0.004602463450282812,
|
| 6867 |
+
"learning_rate": 2.6018486151503213e-05,
|
| 6868 |
+
"loss": 0.0001,
|
| 6869 |
+
"step": 980
|
| 6870 |
+
},
|
| 6871 |
+
{
|
| 6872 |
+
"epoch": 0.3797909407665505,
|
| 6873 |
+
"grad_norm": 0.00020700599998235703,
|
| 6874 |
+
"learning_rate": 2.592534710314668e-05,
|
| 6875 |
+
"loss": 0.0,
|
| 6876 |
+
"step": 981
|
| 6877 |
+
},
|
| 6878 |
+
{
|
| 6879 |
+
"epoch": 0.3801780874951607,
|
| 6880 |
+
"grad_norm": 0.0005082901916466653,
|
| 6881 |
+
"learning_rate": 2.5832316678159118e-05,
|
| 6882 |
+
"loss": 0.0,
|
| 6883 |
+
"step": 982
|
| 6884 |
+
},
|
| 6885 |
+
{
|
| 6886 |
+
"epoch": 0.3805652342237708,
|
| 6887 |
+
"grad_norm": 0.00023574243823532015,
|
| 6888 |
+
"learning_rate": 2.573939529628816e-05,
|
| 6889 |
+
"loss": 0.0,
|
| 6890 |
+
"step": 983
|
| 6891 |
+
},
|
| 6892 |
+
{
|
| 6893 |
+
"epoch": 0.38095238095238093,
|
| 6894 |
+
"grad_norm": 0.00044681079452857375,
|
| 6895 |
+
"learning_rate": 2.564658337678939e-05,
|
| 6896 |
+
"loss": 0.0,
|
| 6897 |
+
"step": 984
|
| 6898 |
+
},
|
| 6899 |
+
{
|
| 6900 |
+
"epoch": 0.3813395276809911,
|
| 6901 |
+
"grad_norm": 0.00034449860686436296,
|
| 6902 |
+
"learning_rate": 2.555388133842455e-05,
|
| 6903 |
+
"loss": 0.0,
|
| 6904 |
+
"step": 985
|
| 6905 |
+
},
|
| 6906 |
+
{
|
| 6907 |
+
"epoch": 0.38172667440960123,
|
| 6908 |
+
"grad_norm": 0.00018269941210746765,
|
| 6909 |
+
"learning_rate": 2.5461289599459646e-05,
|
| 6910 |
+
"loss": 0.0,
|
| 6911 |
+
"step": 986
|
| 6912 |
+
},
|
| 6913 |
+
{
|
| 6914 |
+
"epoch": 0.3821138211382114,
|
| 6915 |
+
"grad_norm": 0.00023150882043410093,
|
| 6916 |
+
"learning_rate": 2.5368808577662913e-05,
|
| 6917 |
+
"loss": 0.0,
|
| 6918 |
+
"step": 987
|
| 6919 |
+
},
|
| 6920 |
+
{
|
| 6921 |
+
"epoch": 0.38250096786682153,
|
| 6922 |
+
"grad_norm": 0.00035011590807698667,
|
| 6923 |
+
"learning_rate": 2.5276438690303112e-05,
|
| 6924 |
+
"loss": 0.0,
|
| 6925 |
+
"step": 988
|
| 6926 |
+
},
|
| 6927 |
+
{
|
| 6928 |
+
"epoch": 0.38288811459543165,
|
| 6929 |
+
"grad_norm": 0.0003966555814258754,
|
| 6930 |
+
"learning_rate": 2.518418035414756e-05,
|
| 6931 |
+
"loss": 0.0,
|
| 6932 |
+
"step": 989
|
| 6933 |
+
},
|
| 6934 |
+
{
|
| 6935 |
+
"epoch": 0.3832752613240418,
|
| 6936 |
+
"grad_norm": 0.002989868400618434,
|
| 6937 |
+
"learning_rate": 2.5092033985460265e-05,
|
| 6938 |
+
"loss": 0.0,
|
| 6939 |
+
"step": 990
|
| 6940 |
+
},
|
| 6941 |
+
{
|
| 6942 |
+
"epoch": 0.38366240805265195,
|
| 6943 |
+
"grad_norm": 0.0004271006619092077,
|
| 6944 |
+
"learning_rate": 2.500000000000001e-05,
|
| 6945 |
+
"loss": 0.0,
|
| 6946 |
+
"step": 991
|
| 6947 |
+
},
|
| 6948 |
+
{
|
| 6949 |
+
"epoch": 0.3840495547812621,
|
| 6950 |
+
"grad_norm": 0.0008473137859255075,
|
| 6951 |
+
"learning_rate": 2.490807881301855e-05,
|
| 6952 |
+
"loss": 0.0,
|
| 6953 |
+
"step": 992
|
| 6954 |
+
},
|
| 6955 |
+
{
|
| 6956 |
+
"epoch": 0.38443670150987225,
|
| 6957 |
+
"grad_norm": 1.8476226329803467,
|
| 6958 |
+
"learning_rate": 2.481627083925869e-05,
|
| 6959 |
+
"loss": 0.0797,
|
| 6960 |
+
"step": 993
|
| 6961 |
+
},
|
| 6962 |
+
{
|
| 6963 |
+
"epoch": 0.38482384823848237,
|
| 6964 |
+
"grad_norm": 0.00014391196600627154,
|
| 6965 |
+
"learning_rate": 2.472457649295241e-05,
|
| 6966 |
+
"loss": 0.0,
|
| 6967 |
+
"step": 994
|
| 6968 |
+
},
|
| 6969 |
+
{
|
| 6970 |
+
"epoch": 0.38521099496709255,
|
| 6971 |
+
"grad_norm": 0.0013244638685137033,
|
| 6972 |
+
"learning_rate": 2.4632996187819034e-05,
|
| 6973 |
+
"loss": 0.0,
|
| 6974 |
+
"step": 995
|
| 6975 |
+
},
|
| 6976 |
+
{
|
| 6977 |
+
"epoch": 0.38559814169570267,
|
| 6978 |
+
"grad_norm": 0.006371075287461281,
|
| 6979 |
+
"learning_rate": 2.4541530337063325e-05,
|
| 6980 |
+
"loss": 0.0001,
|
| 6981 |
+
"step": 996
|
| 6982 |
+
},
|
| 6983 |
+
{
|
| 6984 |
+
"epoch": 0.3859852884243128,
|
| 6985 |
+
"grad_norm": 0.0009385327575728297,
|
| 6986 |
+
"learning_rate": 2.445017935337365e-05,
|
| 6987 |
+
"loss": 0.0,
|
| 6988 |
+
"step": 997
|
| 6989 |
+
},
|
| 6990 |
+
{
|
| 6991 |
+
"epoch": 0.38637243515292297,
|
| 6992 |
+
"grad_norm": 0.01769573800265789,
|
| 6993 |
+
"learning_rate": 2.435894364892005e-05,
|
| 6994 |
+
"loss": 0.0001,
|
| 6995 |
+
"step": 998
|
| 6996 |
+
},
|
| 6997 |
+
{
|
| 6998 |
+
"epoch": 0.3867595818815331,
|
| 6999 |
+
"grad_norm": 0.003815864212810993,
|
| 7000 |
+
"learning_rate": 2.42678236353525e-05,
|
| 7001 |
+
"loss": 0.0001,
|
| 7002 |
+
"step": 999
|
| 7003 |
+
},
|
| 7004 |
+
{
|
| 7005 |
+
"epoch": 0.38714672861014326,
|
| 7006 |
+
"grad_norm": 0.020971521735191345,
|
| 7007 |
+
"learning_rate": 2.4176819723798948e-05,
|
| 7008 |
+
"loss": 0.0001,
|
| 7009 |
+
"step": 1000
|
| 7010 |
+
},
|
| 7011 |
+
{
|
| 7012 |
+
"epoch": 0.3875338753387534,
|
| 7013 |
+
"grad_norm": 0.0321972593665123,
|
| 7014 |
+
"learning_rate": 2.408593232486351e-05,
|
| 7015 |
+
"loss": 0.003,
|
| 7016 |
+
"step": 1001
|
| 7017 |
+
},
|
| 7018 |
+
{
|
| 7019 |
+
"epoch": 0.3879210220673635,
|
| 7020 |
+
"grad_norm": 0.0008187716011889279,
|
| 7021 |
+
"learning_rate": 2.3995161848624597e-05,
|
| 7022 |
+
"loss": 0.0,
|
| 7023 |
+
"step": 1002
|
| 7024 |
+
},
|
| 7025 |
+
{
|
| 7026 |
+
"epoch": 0.3883081687959737,
|
| 7027 |
+
"grad_norm": 0.020097631961107254,
|
| 7028 |
+
"learning_rate": 2.390450870463307e-05,
|
| 7029 |
+
"loss": 0.0004,
|
| 7030 |
+
"step": 1003
|
| 7031 |
+
},
|
| 7032 |
+
{
|
| 7033 |
+
"epoch": 0.3886953155245838,
|
| 7034 |
+
"grad_norm": 0.0003399544220883399,
|
| 7035 |
+
"learning_rate": 2.381397330191043e-05,
|
| 7036 |
+
"loss": 0.0,
|
| 7037 |
+
"step": 1004
|
| 7038 |
+
},
|
| 7039 |
+
{
|
| 7040 |
+
"epoch": 0.389082462253194,
|
| 7041 |
+
"grad_norm": 0.048977453261613846,
|
| 7042 |
+
"learning_rate": 2.372355604894686e-05,
|
| 7043 |
+
"loss": 0.0025,
|
| 7044 |
+
"step": 1005
|
| 7045 |
+
},
|
| 7046 |
+
{
|
| 7047 |
+
"epoch": 0.3894696089818041,
|
| 7048 |
+
"grad_norm": 0.0018951277015730739,
|
| 7049 |
+
"learning_rate": 2.3633257353699524e-05,
|
| 7050 |
+
"loss": 0.0,
|
| 7051 |
+
"step": 1006
|
| 7052 |
+
},
|
| 7053 |
+
{
|
| 7054 |
+
"epoch": 0.3898567557104142,
|
| 7055 |
+
"grad_norm": 0.0007534907781518996,
|
| 7056 |
+
"learning_rate": 2.354307762359064e-05,
|
| 7057 |
+
"loss": 0.0,
|
| 7058 |
+
"step": 1007
|
| 7059 |
+
},
|
| 7060 |
+
{
|
| 7061 |
+
"epoch": 0.3902439024390244,
|
| 7062 |
+
"grad_norm": 0.00365415564738214,
|
| 7063 |
+
"learning_rate": 2.3453017265505673e-05,
|
| 7064 |
+
"loss": 0.0001,
|
| 7065 |
+
"step": 1008
|
| 7066 |
+
},
|
| 7067 |
+
{
|
| 7068 |
+
"epoch": 0.3906310491676345,
|
| 7069 |
+
"grad_norm": 0.07262806594371796,
|
| 7070 |
+
"learning_rate": 2.3363076685791435e-05,
|
| 7071 |
+
"loss": 0.0021,
|
| 7072 |
+
"step": 1009
|
| 7073 |
+
},
|
| 7074 |
+
{
|
| 7075 |
+
"epoch": 0.3910181958962447,
|
| 7076 |
+
"grad_norm": 0.0027157345321029425,
|
| 7077 |
+
"learning_rate": 2.3273256290254402e-05,
|
| 7078 |
+
"loss": 0.0,
|
| 7079 |
+
"step": 1010
|
| 7080 |
+
},
|
| 7081 |
+
{
|
| 7082 |
+
"epoch": 0.3914053426248548,
|
| 7083 |
+
"grad_norm": 0.002876731101423502,
|
| 7084 |
+
"learning_rate": 2.3183556484158736e-05,
|
| 7085 |
+
"loss": 0.0,
|
| 7086 |
+
"step": 1011
|
| 7087 |
+
},
|
| 7088 |
+
{
|
| 7089 |
+
"epoch": 0.39179248935346495,
|
| 7090 |
+
"grad_norm": 0.0003100191242992878,
|
| 7091 |
+
"learning_rate": 2.309397767222446e-05,
|
| 7092 |
+
"loss": 0.0,
|
| 7093 |
+
"step": 1012
|
| 7094 |
+
},
|
| 7095 |
+
{
|
| 7096 |
+
"epoch": 0.3921796360820751,
|
| 7097 |
+
"grad_norm": 0.0007354211411438882,
|
| 7098 |
+
"learning_rate": 2.300452025862574e-05,
|
| 7099 |
+
"loss": 0.0,
|
| 7100 |
+
"step": 1013
|
| 7101 |
+
},
|
| 7102 |
+
{
|
| 7103 |
+
"epoch": 0.39256678281068524,
|
| 7104 |
+
"grad_norm": 0.16168968379497528,
|
| 7105 |
+
"learning_rate": 2.291518464698899e-05,
|
| 7106 |
+
"loss": 0.0013,
|
| 7107 |
+
"step": 1014
|
| 7108 |
+
},
|
| 7109 |
+
{
|
| 7110 |
+
"epoch": 0.39295392953929537,
|
| 7111 |
+
"grad_norm": 0.0008131096255965531,
|
| 7112 |
+
"learning_rate": 2.282597124039107e-05,
|
| 7113 |
+
"loss": 0.0,
|
| 7114 |
+
"step": 1015
|
| 7115 |
+
},
|
| 7116 |
+
{
|
| 7117 |
+
"epoch": 0.39334107626790554,
|
| 7118 |
+
"grad_norm": 0.0007273323717527092,
|
| 7119 |
+
"learning_rate": 2.27368804413574e-05,
|
| 7120 |
+
"loss": 0.0,
|
| 7121 |
+
"step": 1016
|
| 7122 |
+
},
|
| 7123 |
+
{
|
| 7124 |
+
"epoch": 0.39372822299651566,
|
| 7125 |
+
"grad_norm": 0.0019192282343283296,
|
| 7126 |
+
"learning_rate": 2.2647912651860252e-05,
|
| 7127 |
+
"loss": 0.0,
|
| 7128 |
+
"step": 1017
|
| 7129 |
+
},
|
| 7130 |
+
{
|
| 7131 |
+
"epoch": 0.39411536972512584,
|
| 7132 |
+
"grad_norm": 0.0010810550302267075,
|
| 7133 |
+
"learning_rate": 2.255906827331693e-05,
|
| 7134 |
+
"loss": 0.0,
|
| 7135 |
+
"step": 1018
|
| 7136 |
+
},
|
| 7137 |
+
{
|
| 7138 |
+
"epoch": 0.39450251645373596,
|
| 7139 |
+
"grad_norm": 0.0015484696486964822,
|
| 7140 |
+
"learning_rate": 2.2470347706587813e-05,
|
| 7141 |
+
"loss": 0.0,
|
| 7142 |
+
"step": 1019
|
| 7143 |
+
},
|
| 7144 |
+
{
|
| 7145 |
+
"epoch": 0.3948896631823461,
|
| 7146 |
+
"grad_norm": 0.0008098200778476894,
|
| 7147 |
+
"learning_rate": 2.238175135197471e-05,
|
| 7148 |
+
"loss": 0.0,
|
| 7149 |
+
"step": 1020
|
| 7150 |
+
},
|
| 7151 |
+
{
|
| 7152 |
+
"epoch": 0.39527680991095626,
|
| 7153 |
+
"grad_norm": 0.0020570175256580114,
|
| 7154 |
+
"learning_rate": 2.2293279609219003e-05,
|
| 7155 |
+
"loss": 0.0,
|
| 7156 |
+
"step": 1021
|
| 7157 |
+
},
|
| 7158 |
+
{
|
| 7159 |
+
"epoch": 0.3956639566395664,
|
| 7160 |
+
"grad_norm": 0.0012723479885607958,
|
| 7161 |
+
"learning_rate": 2.220493287749978e-05,
|
| 7162 |
+
"loss": 0.0,
|
| 7163 |
+
"step": 1022
|
| 7164 |
+
},
|
| 7165 |
+
{
|
| 7166 |
+
"epoch": 0.39605110336817656,
|
| 7167 |
+
"grad_norm": 0.0003519760794006288,
|
| 7168 |
+
"learning_rate": 2.2116711555432136e-05,
|
| 7169 |
+
"loss": 0.0,
|
| 7170 |
+
"step": 1023
|
| 7171 |
+
},
|
| 7172 |
+
{
|
| 7173 |
+
"epoch": 0.3964382500967867,
|
| 7174 |
+
"grad_norm": 0.00034122299985028803,
|
| 7175 |
+
"learning_rate": 2.2028616041065304e-05,
|
| 7176 |
+
"loss": 0.0,
|
| 7177 |
+
"step": 1024
|
| 7178 |
+
},
|
| 7179 |
+
{
|
| 7180 |
+
"epoch": 0.3968253968253968,
|
| 7181 |
+
"grad_norm": 0.026362977921962738,
|
| 7182 |
+
"learning_rate": 2.194064673188089e-05,
|
| 7183 |
+
"loss": 0.0002,
|
| 7184 |
+
"step": 1025
|
| 7185 |
+
},
|
| 7186 |
+
{
|
| 7187 |
+
"epoch": 0.397212543554007,
|
| 7188 |
+
"grad_norm": 0.00036965496838092804,
|
| 7189 |
+
"learning_rate": 2.185280402479105e-05,
|
| 7190 |
+
"loss": 0.0,
|
| 7191 |
+
"step": 1026
|
| 7192 |
+
},
|
| 7193 |
+
{
|
| 7194 |
+
"epoch": 0.3975996902826171,
|
| 7195 |
+
"grad_norm": 0.11109102517366409,
|
| 7196 |
+
"learning_rate": 2.1765088316136755e-05,
|
| 7197 |
+
"loss": 0.001,
|
| 7198 |
+
"step": 1027
|
| 7199 |
+
},
|
| 7200 |
+
{
|
| 7201 |
+
"epoch": 0.3979868370112273,
|
| 7202 |
+
"grad_norm": 0.042729608714580536,
|
| 7203 |
+
"learning_rate": 2.1677500001685945e-05,
|
| 7204 |
+
"loss": 0.0004,
|
| 7205 |
+
"step": 1028
|
| 7206 |
+
},
|
| 7207 |
+
{
|
| 7208 |
+
"epoch": 0.3983739837398374,
|
| 7209 |
+
"grad_norm": 0.00787183828651905,
|
| 7210 |
+
"learning_rate": 2.159003947663174e-05,
|
| 7211 |
+
"loss": 0.0001,
|
| 7212 |
+
"step": 1029
|
| 7213 |
+
},
|
| 7214 |
+
{
|
| 7215 |
+
"epoch": 0.3987611304684475,
|
| 7216 |
+
"grad_norm": 0.003600550815463066,
|
| 7217 |
+
"learning_rate": 2.1502707135590723e-05,
|
| 7218 |
+
"loss": 0.0,
|
| 7219 |
+
"step": 1030
|
| 7220 |
+
},
|
| 7221 |
+
{
|
| 7222 |
+
"epoch": 0.3991482771970577,
|
| 7223 |
+
"grad_norm": 0.007428227458149195,
|
| 7224 |
+
"learning_rate": 2.1415503372601096e-05,
|
| 7225 |
+
"loss": 0.0001,
|
| 7226 |
+
"step": 1031
|
| 7227 |
+
},
|
| 7228 |
+
{
|
| 7229 |
+
"epoch": 0.3995354239256678,
|
| 7230 |
+
"grad_norm": 0.0009940440068021417,
|
| 7231 |
+
"learning_rate": 2.1328428581120964e-05,
|
| 7232 |
+
"loss": 0.0,
|
| 7233 |
+
"step": 1032
|
| 7234 |
+
},
|
| 7235 |
+
{
|
| 7236 |
+
"epoch": 0.399922570654278,
|
| 7237 |
+
"grad_norm": 0.00677646417170763,
|
| 7238 |
+
"learning_rate": 2.124148315402642e-05,
|
| 7239 |
+
"loss": 0.0001,
|
| 7240 |
+
"step": 1033
|
| 7241 |
+
},
|
| 7242 |
+
{
|
| 7243 |
+
"epoch": 0.4003097173828881,
|
| 7244 |
+
"grad_norm": 0.0008721842896193266,
|
| 7245 |
+
"learning_rate": 2.1154667483609998e-05,
|
| 7246 |
+
"loss": 0.0,
|
| 7247 |
+
"step": 1034
|
| 7248 |
+
},
|
| 7249 |
+
{
|
| 7250 |
+
"epoch": 0.40069686411149824,
|
| 7251 |
+
"grad_norm": 0.00037974066799506545,
|
| 7252 |
+
"learning_rate": 2.106798196157872e-05,
|
| 7253 |
+
"loss": 0.0,
|
| 7254 |
+
"step": 1035
|
| 7255 |
+
},
|
| 7256 |
+
{
|
| 7257 |
+
"epoch": 0.4010840108401084,
|
| 7258 |
+
"grad_norm": 0.0008370023570023477,
|
| 7259 |
+
"learning_rate": 2.098142697905236e-05,
|
| 7260 |
+
"loss": 0.0,
|
| 7261 |
+
"step": 1036
|
| 7262 |
+
},
|
| 7263 |
+
{
|
| 7264 |
+
"epoch": 0.40147115756871854,
|
| 7265 |
+
"grad_norm": 0.0004361419996712357,
|
| 7266 |
+
"learning_rate": 2.0895002926561736e-05,
|
| 7267 |
+
"loss": 0.0,
|
| 7268 |
+
"step": 1037
|
| 7269 |
+
},
|
| 7270 |
+
{
|
| 7271 |
+
"epoch": 0.40185830429732866,
|
| 7272 |
+
"grad_norm": 0.0016112083103507757,
|
| 7273 |
+
"learning_rate": 2.080871019404692e-05,
|
| 7274 |
+
"loss": 0.0,
|
| 7275 |
+
"step": 1038
|
| 7276 |
+
},
|
| 7277 |
+
{
|
| 7278 |
+
"epoch": 0.40224545102593884,
|
| 7279 |
+
"grad_norm": 0.0007014021975919604,
|
| 7280 |
+
"learning_rate": 2.0722549170855494e-05,
|
| 7281 |
+
"loss": 0.0,
|
| 7282 |
+
"step": 1039
|
| 7283 |
+
},
|
| 7284 |
+
{
|
| 7285 |
+
"epoch": 0.40263259775454896,
|
| 7286 |
+
"grad_norm": 0.00015222003275994211,
|
| 7287 |
+
"learning_rate": 2.063652024574071e-05,
|
| 7288 |
+
"loss": 0.0,
|
| 7289 |
+
"step": 1040
|
| 7290 |
+
},
|
| 7291 |
+
{
|
| 7292 |
+
"epoch": 0.40301974448315914,
|
| 7293 |
+
"grad_norm": 0.00020387036784086376,
|
| 7294 |
+
"learning_rate": 2.0550623806859844e-05,
|
| 7295 |
+
"loss": 0.0,
|
| 7296 |
+
"step": 1041
|
| 7297 |
+
},
|
| 7298 |
+
{
|
| 7299 |
+
"epoch": 0.40340689121176926,
|
| 7300 |
+
"grad_norm": 0.004351683426648378,
|
| 7301 |
+
"learning_rate": 2.0464860241772455e-05,
|
| 7302 |
+
"loss": 0.0,
|
| 7303 |
+
"step": 1042
|
| 7304 |
+
},
|
| 7305 |
+
{
|
| 7306 |
+
"epoch": 0.4037940379403794,
|
| 7307 |
+
"grad_norm": 0.00455818697810173,
|
| 7308 |
+
"learning_rate": 2.0379229937438477e-05,
|
| 7309 |
+
"loss": 0.0001,
|
| 7310 |
+
"step": 1043
|
| 7311 |
+
},
|
| 7312 |
+
{
|
| 7313 |
+
"epoch": 0.40418118466898956,
|
| 7314 |
+
"grad_norm": 0.000623556668870151,
|
| 7315 |
+
"learning_rate": 2.0293733280216648e-05,
|
| 7316 |
+
"loss": 0.0,
|
| 7317 |
+
"step": 1044
|
| 7318 |
+
},
|
| 7319 |
+
{
|
| 7320 |
+
"epoch": 0.4045683313975997,
|
| 7321 |
+
"grad_norm": 0.0013906165258958936,
|
| 7322 |
+
"learning_rate": 2.020837065586269e-05,
|
| 7323 |
+
"loss": 0.0,
|
| 7324 |
+
"step": 1045
|
| 7325 |
+
},
|
| 7326 |
+
{
|
| 7327 |
+
"epoch": 0.40495547812620986,
|
| 7328 |
+
"grad_norm": 0.003972301259636879,
|
| 7329 |
+
"learning_rate": 2.0123142449527584e-05,
|
| 7330 |
+
"loss": 0.0001,
|
| 7331 |
+
"step": 1046
|
| 7332 |
+
},
|
| 7333 |
+
{
|
| 7334 |
+
"epoch": 0.40534262485482,
|
| 7335 |
+
"grad_norm": 0.00038339104503393173,
|
| 7336 |
+
"learning_rate": 2.0038049045755775e-05,
|
| 7337 |
+
"loss": 0.0,
|
| 7338 |
+
"step": 1047
|
| 7339 |
+
},
|
| 7340 |
+
{
|
| 7341 |
+
"epoch": 0.4057297715834301,
|
| 7342 |
+
"grad_norm": 0.00043085686047561467,
|
| 7343 |
+
"learning_rate": 1.995309082848354e-05,
|
| 7344 |
+
"loss": 0.0,
|
| 7345 |
+
"step": 1048
|
| 7346 |
+
},
|
| 7347 |
+
{
|
| 7348 |
+
"epoch": 0.4061169183120403,
|
| 7349 |
+
"grad_norm": 0.0012404817389324307,
|
| 7350 |
+
"learning_rate": 1.9868268181037185e-05,
|
| 7351 |
+
"loss": 0.0,
|
| 7352 |
+
"step": 1049
|
| 7353 |
+
},
|
| 7354 |
+
{
|
| 7355 |
+
"epoch": 0.4065040650406504,
|
| 7356 |
+
"grad_norm": 0.0011212611570954323,
|
| 7357 |
+
"learning_rate": 1.978358148613134e-05,
|
| 7358 |
+
"loss": 0.0,
|
| 7359 |
+
"step": 1050
|
| 7360 |
+
},
|
| 7361 |
+
{
|
| 7362 |
+
"epoch": 0.4068912117692606,
|
| 7363 |
+
"grad_norm": 0.039180099964141846,
|
| 7364 |
+
"learning_rate": 1.9699031125867213e-05,
|
| 7365 |
+
"loss": 0.0023,
|
| 7366 |
+
"step": 1051
|
| 7367 |
+
},
|
| 7368 |
+
{
|
| 7369 |
+
"epoch": 0.4072783584978707,
|
| 7370 |
+
"grad_norm": 0.02465960942208767,
|
| 7371 |
+
"learning_rate": 1.9614617481730883e-05,
|
| 7372 |
+
"loss": 0.0004,
|
| 7373 |
+
"step": 1052
|
| 7374 |
+
},
|
| 7375 |
+
{
|
| 7376 |
+
"epoch": 0.4076655052264808,
|
| 7377 |
+
"grad_norm": 0.013811935670673847,
|
| 7378 |
+
"learning_rate": 1.9530340934591608e-05,
|
| 7379 |
+
"loss": 0.0003,
|
| 7380 |
+
"step": 1053
|
| 7381 |
+
},
|
| 7382 |
+
{
|
| 7383 |
+
"epoch": 0.408052651955091,
|
| 7384 |
+
"grad_norm": 0.008351047523319721,
|
| 7385 |
+
"learning_rate": 1.9446201864699987e-05,
|
| 7386 |
+
"loss": 0.0002,
|
| 7387 |
+
"step": 1054
|
| 7388 |
+
},
|
| 7389 |
+
{
|
| 7390 |
+
"epoch": 0.4084397986837011,
|
| 7391 |
+
"grad_norm": 0.0017275417922064662,
|
| 7392 |
+
"learning_rate": 1.9362200651686407e-05,
|
| 7393 |
+
"loss": 0.0001,
|
| 7394 |
+
"step": 1055
|
| 7395 |
+
},
|
| 7396 |
+
{
|
| 7397 |
+
"epoch": 0.4088269454123113,
|
| 7398 |
+
"grad_norm": 0.031212275847792625,
|
| 7399 |
+
"learning_rate": 1.9278337674559237e-05,
|
| 7400 |
+
"loss": 0.0005,
|
| 7401 |
+
"step": 1056
|
| 7402 |
+
},
|
| 7403 |
+
{
|
| 7404 |
+
"epoch": 0.4092140921409214,
|
| 7405 |
+
"grad_norm": 0.001833677408285439,
|
| 7406 |
+
"learning_rate": 1.919461331170312e-05,
|
| 7407 |
+
"loss": 0.0001,
|
| 7408 |
+
"step": 1057
|
| 7409 |
+
},
|
| 7410 |
+
{
|
| 7411 |
+
"epoch": 0.40960123886953154,
|
| 7412 |
+
"grad_norm": 0.00048409271403215826,
|
| 7413 |
+
"learning_rate": 1.9111027940877284e-05,
|
| 7414 |
+
"loss": 0.0,
|
| 7415 |
+
"step": 1058
|
| 7416 |
+
},
|
| 7417 |
+
{
|
| 7418 |
+
"epoch": 0.4099883855981417,
|
| 7419 |
+
"grad_norm": 0.004958099219948053,
|
| 7420 |
+
"learning_rate": 1.902758193921385e-05,
|
| 7421 |
+
"loss": 0.0001,
|
| 7422 |
+
"step": 1059
|
| 7423 |
+
},
|
| 7424 |
+
{
|
| 7425 |
+
"epoch": 0.41037553232675184,
|
| 7426 |
+
"grad_norm": 0.0003586974926292896,
|
| 7427 |
+
"learning_rate": 1.89442756832161e-05,
|
| 7428 |
+
"loss": 0.0,
|
| 7429 |
+
"step": 1060
|
| 7430 |
+
},
|
| 7431 |
+
{
|
| 7432 |
+
"epoch": 0.41076267905536196,
|
| 7433 |
+
"grad_norm": 0.004265904892235994,
|
| 7434 |
+
"learning_rate": 1.8861109548756766e-05,
|
| 7435 |
+
"loss": 0.0,
|
| 7436 |
+
"step": 1061
|
| 7437 |
+
},
|
| 7438 |
+
{
|
| 7439 |
+
"epoch": 0.41114982578397213,
|
| 7440 |
+
"grad_norm": 0.0007765475893393159,
|
| 7441 |
+
"learning_rate": 1.877808391107641e-05,
|
| 7442 |
+
"loss": 0.0,
|
| 7443 |
+
"step": 1062
|
| 7444 |
+
},
|
| 7445 |
+
{
|
| 7446 |
+
"epoch": 0.41153697251258226,
|
| 7447 |
+
"grad_norm": 0.003014490008354187,
|
| 7448 |
+
"learning_rate": 1.8695199144781645e-05,
|
| 7449 |
+
"loss": 0.0001,
|
| 7450 |
+
"step": 1063
|
| 7451 |
+
},
|
| 7452 |
+
{
|
| 7453 |
+
"epoch": 0.41192411924119243,
|
| 7454 |
+
"grad_norm": 0.0007445579976774752,
|
| 7455 |
+
"learning_rate": 1.861245562384351e-05,
|
| 7456 |
+
"loss": 0.0,
|
| 7457 |
+
"step": 1064
|
| 7458 |
+
},
|
| 7459 |
+
{
|
| 7460 |
+
"epoch": 0.41231126596980255,
|
| 7461 |
+
"grad_norm": 0.0004886957467533648,
|
| 7462 |
+
"learning_rate": 1.8529853721595697e-05,
|
| 7463 |
+
"loss": 0.0,
|
| 7464 |
+
"step": 1065
|
| 7465 |
+
},
|
| 7466 |
+
{
|
| 7467 |
+
"epoch": 0.4126984126984127,
|
| 7468 |
+
"grad_norm": 0.0007616803050041199,
|
| 7469 |
+
"learning_rate": 1.844739381073301e-05,
|
| 7470 |
+
"loss": 0.0,
|
| 7471 |
+
"step": 1066
|
| 7472 |
+
},
|
| 7473 |
+
{
|
| 7474 |
+
"epoch": 0.41308555942702285,
|
| 7475 |
+
"grad_norm": 0.0005557533586397767,
|
| 7476 |
+
"learning_rate": 1.8365076263309545e-05,
|
| 7477 |
+
"loss": 0.0,
|
| 7478 |
+
"step": 1067
|
| 7479 |
+
},
|
| 7480 |
+
{
|
| 7481 |
+
"epoch": 0.413472706155633,
|
| 7482 |
+
"grad_norm": 0.00033164405613206327,
|
| 7483 |
+
"learning_rate": 1.8282901450737033e-05,
|
| 7484 |
+
"loss": 0.0,
|
| 7485 |
+
"step": 1068
|
| 7486 |
+
},
|
| 7487 |
+
{
|
| 7488 |
+
"epoch": 0.41385985288424315,
|
| 7489 |
+
"grad_norm": 0.000697499024681747,
|
| 7490 |
+
"learning_rate": 1.8200869743783245e-05,
|
| 7491 |
+
"loss": 0.0,
|
| 7492 |
+
"step": 1069
|
| 7493 |
+
},
|
| 7494 |
+
{
|
| 7495 |
+
"epoch": 0.4142469996128533,
|
| 7496 |
+
"grad_norm": 0.008421828970313072,
|
| 7497 |
+
"learning_rate": 1.8118981512570253e-05,
|
| 7498 |
+
"loss": 0.0001,
|
| 7499 |
+
"step": 1070
|
| 7500 |
+
},
|
| 7501 |
+
{
|
| 7502 |
+
"epoch": 0.4146341463414634,
|
| 7503 |
+
"grad_norm": 0.0010353690013289452,
|
| 7504 |
+
"learning_rate": 1.80372371265728e-05,
|
| 7505 |
+
"loss": 0.0,
|
| 7506 |
+
"step": 1071
|
| 7507 |
+
},
|
| 7508 |
+
{
|
| 7509 |
+
"epoch": 0.41502129307007357,
|
| 7510 |
+
"grad_norm": 0.0004474584711715579,
|
| 7511 |
+
"learning_rate": 1.795563695461653e-05,
|
| 7512 |
+
"loss": 0.0,
|
| 7513 |
+
"step": 1072
|
| 7514 |
+
},
|
| 7515 |
+
{
|
| 7516 |
+
"epoch": 0.4154084397986837,
|
| 7517 |
+
"grad_norm": 0.0019867317751049995,
|
| 7518 |
+
"learning_rate": 1.7874181364876464e-05,
|
| 7519 |
+
"loss": 0.0,
|
| 7520 |
+
"step": 1073
|
| 7521 |
+
},
|
| 7522 |
+
{
|
| 7523 |
+
"epoch": 0.41579558652729387,
|
| 7524 |
+
"grad_norm": 0.001235696254298091,
|
| 7525 |
+
"learning_rate": 1.7792870724875314e-05,
|
| 7526 |
+
"loss": 0.0,
|
| 7527 |
+
"step": 1074
|
| 7528 |
+
},
|
| 7529 |
+
{
|
| 7530 |
+
"epoch": 0.416182733255904,
|
| 7531 |
+
"grad_norm": 0.0007534808246418834,
|
| 7532 |
+
"learning_rate": 1.77117054014817e-05,
|
| 7533 |
+
"loss": 0.0,
|
| 7534 |
+
"step": 1075
|
| 7535 |
+
},
|
| 7536 |
+
{
|
| 7537 |
+
"epoch": 0.4165698799845141,
|
| 7538 |
+
"grad_norm": 0.0005520595004782081,
|
| 7539 |
+
"learning_rate": 1.7630685760908622e-05,
|
| 7540 |
+
"loss": 0.0,
|
| 7541 |
+
"step": 1076
|
| 7542 |
+
},
|
| 7543 |
+
{
|
| 7544 |
+
"epoch": 0.4169570267131243,
|
| 7545 |
+
"grad_norm": 0.0005588960484601557,
|
| 7546 |
+
"learning_rate": 1.7549812168711777e-05,
|
| 7547 |
+
"loss": 0.0,
|
| 7548 |
+
"step": 1077
|
| 7549 |
+
},
|
| 7550 |
+
{
|
| 7551 |
+
"epoch": 0.4173441734417344,
|
| 7552 |
+
"grad_norm": 0.0007228697068057954,
|
| 7553 |
+
"learning_rate": 1.746908498978791e-05,
|
| 7554 |
+
"loss": 0.0,
|
| 7555 |
+
"step": 1078
|
| 7556 |
+
},
|
| 7557 |
+
{
|
| 7558 |
+
"epoch": 0.41773132017034453,
|
| 7559 |
+
"grad_norm": 0.00042682827915996313,
|
| 7560 |
+
"learning_rate": 1.73885045883731e-05,
|
| 7561 |
+
"loss": 0.0,
|
| 7562 |
+
"step": 1079
|
| 7563 |
+
},
|
| 7564 |
+
{
|
| 7565 |
+
"epoch": 0.4181184668989547,
|
| 7566 |
+
"grad_norm": 0.0011390242725610733,
|
| 7567 |
+
"learning_rate": 1.7308071328041243e-05,
|
| 7568 |
+
"loss": 0.0,
|
| 7569 |
+
"step": 1080
|
| 7570 |
+
},
|
| 7571 |
+
{
|
| 7572 |
+
"epoch": 0.41850561362756483,
|
| 7573 |
+
"grad_norm": 0.00039792185998521745,
|
| 7574 |
+
"learning_rate": 1.7227785571702303e-05,
|
| 7575 |
+
"loss": 0.0,
|
| 7576 |
+
"step": 1081
|
| 7577 |
+
},
|
| 7578 |
+
{
|
| 7579 |
+
"epoch": 0.418892760356175,
|
| 7580 |
+
"grad_norm": 0.0003800159611273557,
|
| 7581 |
+
"learning_rate": 1.7147647681600738e-05,
|
| 7582 |
+
"loss": 0.0,
|
| 7583 |
+
"step": 1082
|
| 7584 |
+
},
|
| 7585 |
+
{
|
| 7586 |
+
"epoch": 0.41927990708478513,
|
| 7587 |
+
"grad_norm": 0.0011553197400644422,
|
| 7588 |
+
"learning_rate": 1.7067658019313826e-05,
|
| 7589 |
+
"loss": 0.0,
|
| 7590 |
+
"step": 1083
|
| 7591 |
+
},
|
| 7592 |
+
{
|
| 7593 |
+
"epoch": 0.41966705381339525,
|
| 7594 |
+
"grad_norm": 0.0001602371339686215,
|
| 7595 |
+
"learning_rate": 1.6987816945750057e-05,
|
| 7596 |
+
"loss": 0.0,
|
| 7597 |
+
"step": 1084
|
| 7598 |
+
},
|
| 7599 |
+
{
|
| 7600 |
+
"epoch": 0.42005420054200543,
|
| 7601 |
+
"grad_norm": 0.0007585774874314666,
|
| 7602 |
+
"learning_rate": 1.690812482114752e-05,
|
| 7603 |
+
"loss": 0.0,
|
| 7604 |
+
"step": 1085
|
| 7605 |
+
},
|
| 7606 |
+
{
|
| 7607 |
+
"epoch": 0.42044134727061555,
|
| 7608 |
+
"grad_norm": 0.0005152882658876479,
|
| 7609 |
+
"learning_rate": 1.6828582005072187e-05,
|
| 7610 |
+
"loss": 0.0,
|
| 7611 |
+
"step": 1086
|
| 7612 |
+
},
|
| 7613 |
+
{
|
| 7614 |
+
"epoch": 0.42082849399922573,
|
| 7615 |
+
"grad_norm": 0.00029434141470119357,
|
| 7616 |
+
"learning_rate": 1.6749188856416425e-05,
|
| 7617 |
+
"loss": 0.0,
|
| 7618 |
+
"step": 1087
|
| 7619 |
+
},
|
| 7620 |
+
{
|
| 7621 |
+
"epoch": 0.42121564072783585,
|
| 7622 |
+
"grad_norm": 0.0014791539870202541,
|
| 7623 |
+
"learning_rate": 1.666994573339729e-05,
|
| 7624 |
+
"loss": 0.0,
|
| 7625 |
+
"step": 1088
|
| 7626 |
+
},
|
| 7627 |
+
{
|
| 7628 |
+
"epoch": 0.42160278745644597,
|
| 7629 |
+
"grad_norm": 0.0017479541711509228,
|
| 7630 |
+
"learning_rate": 1.6590852993554935e-05,
|
| 7631 |
+
"loss": 0.0,
|
| 7632 |
+
"step": 1089
|
| 7633 |
+
},
|
| 7634 |
+
{
|
| 7635 |
+
"epoch": 0.42198993418505615,
|
| 7636 |
+
"grad_norm": 0.0161809790879488,
|
| 7637 |
+
"learning_rate": 1.651191099375099e-05,
|
| 7638 |
+
"loss": 0.0002,
|
| 7639 |
+
"step": 1090
|
| 7640 |
+
},
|
| 7641 |
+
{
|
| 7642 |
+
"epoch": 0.42237708091366627,
|
| 7643 |
+
"grad_norm": 0.0017758719623088837,
|
| 7644 |
+
"learning_rate": 1.6433120090166943e-05,
|
| 7645 |
+
"loss": 0.0,
|
| 7646 |
+
"step": 1091
|
| 7647 |
+
},
|
| 7648 |
+
{
|
| 7649 |
+
"epoch": 0.42276422764227645,
|
| 7650 |
+
"grad_norm": 0.0008196650887839496,
|
| 7651 |
+
"learning_rate": 1.6354480638302588e-05,
|
| 7652 |
+
"loss": 0.0,
|
| 7653 |
+
"step": 1092
|
| 7654 |
+
},
|
| 7655 |
+
{
|
| 7656 |
+
"epoch": 0.42315137437088657,
|
| 7657 |
+
"grad_norm": 0.0007899802876636386,
|
| 7658 |
+
"learning_rate": 1.6275992992974308e-05,
|
| 7659 |
+
"loss": 0.0,
|
| 7660 |
+
"step": 1093
|
| 7661 |
+
},
|
| 7662 |
+
{
|
| 7663 |
+
"epoch": 0.4235385210994967,
|
| 7664 |
+
"grad_norm": 0.0006489930092357099,
|
| 7665 |
+
"learning_rate": 1.6197657508313597e-05,
|
| 7666 |
+
"loss": 0.0,
|
| 7667 |
+
"step": 1094
|
| 7668 |
+
},
|
| 7669 |
+
{
|
| 7670 |
+
"epoch": 0.42392566782810687,
|
| 7671 |
+
"grad_norm": 0.0010093534365296364,
|
| 7672 |
+
"learning_rate": 1.6119474537765395e-05,
|
| 7673 |
+
"loss": 0.0,
|
| 7674 |
+
"step": 1095
|
| 7675 |
+
},
|
| 7676 |
+
{
|
| 7677 |
+
"epoch": 0.424312814556717,
|
| 7678 |
+
"grad_norm": 0.0457945354282856,
|
| 7679 |
+
"learning_rate": 1.604144443408653e-05,
|
| 7680 |
+
"loss": 0.0005,
|
| 7681 |
+
"step": 1096
|
| 7682 |
+
},
|
| 7683 |
+
{
|
| 7684 |
+
"epoch": 0.42469996128532717,
|
| 7685 |
+
"grad_norm": 0.009583662264049053,
|
| 7686 |
+
"learning_rate": 1.5963567549344028e-05,
|
| 7687 |
+
"loss": 0.0001,
|
| 7688 |
+
"step": 1097
|
| 7689 |
+
},
|
| 7690 |
+
{
|
| 7691 |
+
"epoch": 0.4250871080139373,
|
| 7692 |
+
"grad_norm": 0.0023053644690662622,
|
| 7693 |
+
"learning_rate": 1.58858442349137e-05,
|
| 7694 |
+
"loss": 0.0,
|
| 7695 |
+
"step": 1098
|
| 7696 |
+
},
|
| 7697 |
+
{
|
| 7698 |
+
"epoch": 0.4254742547425474,
|
| 7699 |
+
"grad_norm": 0.0012813667999580503,
|
| 7700 |
+
"learning_rate": 1.580827484147842e-05,
|
| 7701 |
+
"loss": 0.0,
|
| 7702 |
+
"step": 1099
|
| 7703 |
+
},
|
| 7704 |
+
{
|
| 7705 |
+
"epoch": 0.4258614014711576,
|
| 7706 |
+
"grad_norm": 0.001665553660131991,
|
| 7707 |
+
"learning_rate": 1.5730859719026536e-05,
|
| 7708 |
+
"loss": 0.0,
|
| 7709 |
+
"step": 1100
|
| 7710 |
+
},
|
| 7711 |
+
{
|
| 7712 |
+
"epoch": 0.4262485481997677,
|
| 7713 |
+
"grad_norm": 0.03397725149989128,
|
| 7714 |
+
"learning_rate": 1.5653599216850378e-05,
|
| 7715 |
+
"loss": 0.0026,
|
| 7716 |
+
"step": 1101
|
| 7717 |
+
},
|
| 7718 |
+
{
|
| 7719 |
+
"epoch": 0.42663569492837783,
|
| 7720 |
+
"grad_norm": 0.0028618781361728907,
|
| 7721 |
+
"learning_rate": 1.557649368354464e-05,
|
| 7722 |
+
"loss": 0.0001,
|
| 7723 |
+
"step": 1102
|
| 7724 |
+
},
|
| 7725 |
+
{
|
| 7726 |
+
"epoch": 0.427022841656988,
|
| 7727 |
+
"grad_norm": 0.0466277040541172,
|
| 7728 |
+
"learning_rate": 1.5499543467004813e-05,
|
| 7729 |
+
"loss": 0.0011,
|
| 7730 |
+
"step": 1103
|
| 7731 |
+
},
|
| 7732 |
+
{
|
| 7733 |
+
"epoch": 0.4274099883855981,
|
| 7734 |
+
"grad_norm": 0.0036304357927292585,
|
| 7735 |
+
"learning_rate": 1.5422748914425557e-05,
|
| 7736 |
+
"loss": 0.0001,
|
| 7737 |
+
"step": 1104
|
| 7738 |
+
},
|
| 7739 |
+
{
|
| 7740 |
+
"epoch": 0.4277971351142083,
|
| 7741 |
+
"grad_norm": 0.0021317729260772467,
|
| 7742 |
+
"learning_rate": 1.5346110372299216e-05,
|
| 7743 |
+
"loss": 0.0,
|
| 7744 |
+
"step": 1105
|
| 7745 |
+
},
|
| 7746 |
+
{
|
| 7747 |
+
"epoch": 0.4281842818428184,
|
| 7748 |
+
"grad_norm": 0.0051021394319832325,
|
| 7749 |
+
"learning_rate": 1.526962818641428e-05,
|
| 7750 |
+
"loss": 0.0001,
|
| 7751 |
+
"step": 1106
|
| 7752 |
+
},
|
| 7753 |
+
{
|
| 7754 |
+
"epoch": 0.42857142857142855,
|
| 7755 |
+
"grad_norm": 0.0004642395942937583,
|
| 7756 |
+
"learning_rate": 1.5193302701853673e-05,
|
| 7757 |
+
"loss": 0.0,
|
| 7758 |
+
"step": 1107
|
| 7759 |
+
},
|
| 7760 |
+
{
|
| 7761 |
+
"epoch": 0.4289585753000387,
|
| 7762 |
+
"grad_norm": 0.0023515617940574884,
|
| 7763 |
+
"learning_rate": 1.5117134262993348e-05,
|
| 7764 |
+
"loss": 0.0,
|
| 7765 |
+
"step": 1108
|
| 7766 |
+
},
|
| 7767 |
+
{
|
| 7768 |
+
"epoch": 0.42934572202864885,
|
| 7769 |
+
"grad_norm": 0.004486995283514261,
|
| 7770 |
+
"learning_rate": 1.5041123213500674e-05,
|
| 7771 |
+
"loss": 0.0001,
|
| 7772 |
+
"step": 1109
|
| 7773 |
+
},
|
| 7774 |
+
{
|
| 7775 |
+
"epoch": 0.429732868757259,
|
| 7776 |
+
"grad_norm": 0.03434832766652107,
|
| 7777 |
+
"learning_rate": 1.4965269896332885e-05,
|
| 7778 |
+
"loss": 0.0004,
|
| 7779 |
+
"step": 1110
|
| 7780 |
+
},
|
| 7781 |
+
{
|
| 7782 |
+
"epoch": 0.43012001548586914,
|
| 7783 |
+
"grad_norm": 0.06803593039512634,
|
| 7784 |
+
"learning_rate": 1.488957465373551e-05,
|
| 7785 |
+
"loss": 0.0015,
|
| 7786 |
+
"step": 1111
|
| 7787 |
+
},
|
| 7788 |
+
{
|
| 7789 |
+
"epoch": 0.43050716221447927,
|
| 7790 |
+
"grad_norm": 0.0014034640043973923,
|
| 7791 |
+
"learning_rate": 1.4814037827240895e-05,
|
| 7792 |
+
"loss": 0.0,
|
| 7793 |
+
"step": 1112
|
| 7794 |
+
},
|
| 7795 |
+
{
|
| 7796 |
+
"epoch": 0.43089430894308944,
|
| 7797 |
+
"grad_norm": 0.00042223307536914945,
|
| 7798 |
+
"learning_rate": 1.47386597576666e-05,
|
| 7799 |
+
"loss": 0.0,
|
| 7800 |
+
"step": 1113
|
| 7801 |
}
|
| 7802 |
],
|
| 7803 |
"logging_steps": 1,
|
|
|
|
| 7817 |
"attributes": {}
|
| 7818 |
}
|
| 7819 |
},
|
| 7820 |
+
"total_flos": 3.17298751184044e+17,
|
| 7821 |
"train_batch_size": 4,
|
| 7822 |
"trial_name": null,
|
| 7823 |
"trial_params": null
|