Commit
•
bd1d7ee
1
Parent(s):
9e8b8c9
Upload folder using huggingface_hub
Browse files- data/model.safetensors +1 -1
- data/optimizer.pt +1 -1
- data/rng_state.pth +1 -1
- data/scheduler.pt +1 -1
- data/trainer_state.json +2341 -3
data/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 576008736
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dd1709db18e90edc24059b64a578b586ae315622e2466a225c5605e3fa28c2d7
|
| 3 |
size 576008736
|
data/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1152256984
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c3e4e0e6b57e3e8d11e9e3c939267d213e6cc8c2683a951981bc53a1252a6976
|
| 3 |
size 1152256984
|
data/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:f24c4425e9d6c449b00fd9a1872bee61b2463d9ca33c8e91c1c5a45cf6c44127
|
| 3 |
size 14244
|
data/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:a4d72a7225743ea21366860ec06ef85504326088672296a72343c37eced8e566
|
| 3 |
size 1064
|
data/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,
|
|
@@ -11669,6 +11669,2344 @@
|
|
| 11669 |
"learning_rate": 2.9825041073997102e-05,
|
| 11670 |
"loss": 1.9942,
|
| 11671 |
"step": 24990
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 11672 |
}
|
| 11673 |
],
|
| 11674 |
"logging_steps": 15,
|
|
@@ -11688,7 +14026,7 @@
|
|
| 11688 |
"attributes": {}
|
| 11689 |
}
|
| 11690 |
},
|
| 11691 |
-
"total_flos":
|
| 11692 |
"train_batch_size": 4,
|
| 11693 |
"trial_name": null,
|
| 11694 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.05842054222052586,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 30000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 11669 |
"learning_rate": 2.9825041073997102e-05,
|
| 11670 |
"loss": 1.9942,
|
| 11671 |
"step": 24990
|
| 11672 |
+
},
|
| 11673 |
+
{
|
| 11674 |
+
"epoch": 0.04869352194080831,
|
| 11675 |
+
"grad_norm": 3.3070085048675537,
|
| 11676 |
+
"learning_rate": 2.9824831386111103e-05,
|
| 11677 |
+
"loss": 1.5052,
|
| 11678 |
+
"step": 25005
|
| 11679 |
+
},
|
| 11680 |
+
{
|
| 11681 |
+
"epoch": 0.04872273221191857,
|
| 11682 |
+
"grad_norm": 2.769148826599121,
|
| 11683 |
+
"learning_rate": 2.9824621573383107e-05,
|
| 11684 |
+
"loss": 1.8361,
|
| 11685 |
+
"step": 25020
|
| 11686 |
+
},
|
| 11687 |
+
{
|
| 11688 |
+
"epoch": 0.04875194248302883,
|
| 11689 |
+
"grad_norm": 2.3901126384735107,
|
| 11690 |
+
"learning_rate": 2.982441163581489e-05,
|
| 11691 |
+
"loss": 1.8346,
|
| 11692 |
+
"step": 25035
|
| 11693 |
+
},
|
| 11694 |
+
{
|
| 11695 |
+
"epoch": 0.04878115275413909,
|
| 11696 |
+
"grad_norm": 3.8897323608398438,
|
| 11697 |
+
"learning_rate": 2.9824201573408218e-05,
|
| 11698 |
+
"loss": 1.8186,
|
| 11699 |
+
"step": 25050
|
| 11700 |
+
},
|
| 11701 |
+
{
|
| 11702 |
+
"epoch": 0.04881036302524936,
|
| 11703 |
+
"grad_norm": 1.947713017463684,
|
| 11704 |
+
"learning_rate": 2.982399138616486e-05,
|
| 11705 |
+
"loss": 1.8812,
|
| 11706 |
+
"step": 25065
|
| 11707 |
+
},
|
| 11708 |
+
{
|
| 11709 |
+
"epoch": 0.04883957329635962,
|
| 11710 |
+
"grad_norm": 1.6136304140090942,
|
| 11711 |
+
"learning_rate": 2.9823781074086582e-05,
|
| 11712 |
+
"loss": 2.0161,
|
| 11713 |
+
"step": 25080
|
| 11714 |
+
},
|
| 11715 |
+
{
|
| 11716 |
+
"epoch": 0.04886878356746988,
|
| 11717 |
+
"grad_norm": 2.6880128383636475,
|
| 11718 |
+
"learning_rate": 2.9823570637175166e-05,
|
| 11719 |
+
"loss": 1.8861,
|
| 11720 |
+
"step": 25095
|
| 11721 |
+
},
|
| 11722 |
+
{
|
| 11723 |
+
"epoch": 0.04889799383858015,
|
| 11724 |
+
"grad_norm": 4.511691570281982,
|
| 11725 |
+
"learning_rate": 2.982336007543237e-05,
|
| 11726 |
+
"loss": 1.8334,
|
| 11727 |
+
"step": 25110
|
| 11728 |
+
},
|
| 11729 |
+
{
|
| 11730 |
+
"epoch": 0.04892720410969041,
|
| 11731 |
+
"grad_norm": 2.540619134902954,
|
| 11732 |
+
"learning_rate": 2.9823149388859975e-05,
|
| 11733 |
+
"loss": 1.8554,
|
| 11734 |
+
"step": 25125
|
| 11735 |
+
},
|
| 11736 |
+
{
|
| 11737 |
+
"epoch": 0.04895641438080067,
|
| 11738 |
+
"grad_norm": 2.650416135787964,
|
| 11739 |
+
"learning_rate": 2.982293857745976e-05,
|
| 11740 |
+
"loss": 2.0556,
|
| 11741 |
+
"step": 25140
|
| 11742 |
+
},
|
| 11743 |
+
{
|
| 11744 |
+
"epoch": 0.04898562465191093,
|
| 11745 |
+
"grad_norm": 4.075965881347656,
|
| 11746 |
+
"learning_rate": 2.9822727641233488e-05,
|
| 11747 |
+
"loss": 1.7369,
|
| 11748 |
+
"step": 25155
|
| 11749 |
+
},
|
| 11750 |
+
{
|
| 11751 |
+
"epoch": 0.0490148349230212,
|
| 11752 |
+
"grad_norm": 4.21481466293335,
|
| 11753 |
+
"learning_rate": 2.9822516580182944e-05,
|
| 11754 |
+
"loss": 1.9624,
|
| 11755 |
+
"step": 25170
|
| 11756 |
+
},
|
| 11757 |
+
{
|
| 11758 |
+
"epoch": 0.04904404519413146,
|
| 11759 |
+
"grad_norm": 4.617081642150879,
|
| 11760 |
+
"learning_rate": 2.98223053943099e-05,
|
| 11761 |
+
"loss": 1.7276,
|
| 11762 |
+
"step": 25185
|
| 11763 |
+
},
|
| 11764 |
+
{
|
| 11765 |
+
"epoch": 0.04907325546524172,
|
| 11766 |
+
"grad_norm": 3.5099408626556396,
|
| 11767 |
+
"learning_rate": 2.9822094083616145e-05,
|
| 11768 |
+
"loss": 1.8783,
|
| 11769 |
+
"step": 25200
|
| 11770 |
+
},
|
| 11771 |
+
{
|
| 11772 |
+
"epoch": 0.049102465736351984,
|
| 11773 |
+
"grad_norm": 4.118253231048584,
|
| 11774 |
+
"learning_rate": 2.9821882648103445e-05,
|
| 11775 |
+
"loss": 1.7848,
|
| 11776 |
+
"step": 25215
|
| 11777 |
+
},
|
| 11778 |
+
{
|
| 11779 |
+
"epoch": 0.04913167600746225,
|
| 11780 |
+
"grad_norm": 3.617659091949463,
|
| 11781 |
+
"learning_rate": 2.982167108777359e-05,
|
| 11782 |
+
"loss": 1.8307,
|
| 11783 |
+
"step": 25230
|
| 11784 |
+
},
|
| 11785 |
+
{
|
| 11786 |
+
"epoch": 0.04916088627857251,
|
| 11787 |
+
"grad_norm": 3.3717000484466553,
|
| 11788 |
+
"learning_rate": 2.9821459402628357e-05,
|
| 11789 |
+
"loss": 1.6414,
|
| 11790 |
+
"step": 25245
|
| 11791 |
+
},
|
| 11792 |
+
{
|
| 11793 |
+
"epoch": 0.049190096549682774,
|
| 11794 |
+
"grad_norm": 3.3341469764709473,
|
| 11795 |
+
"learning_rate": 2.9821247592669526e-05,
|
| 11796 |
+
"loss": 1.9641,
|
| 11797 |
+
"step": 25260
|
| 11798 |
+
},
|
| 11799 |
+
{
|
| 11800 |
+
"epoch": 0.04921930682079304,
|
| 11801 |
+
"grad_norm": 3.8818039894104004,
|
| 11802 |
+
"learning_rate": 2.9821035657898886e-05,
|
| 11803 |
+
"loss": 1.8227,
|
| 11804 |
+
"step": 25275
|
| 11805 |
+
},
|
| 11806 |
+
{
|
| 11807 |
+
"epoch": 0.0492485170919033,
|
| 11808 |
+
"grad_norm": 2.4647955894470215,
|
| 11809 |
+
"learning_rate": 2.9820823598318226e-05,
|
| 11810 |
+
"loss": 1.9441,
|
| 11811 |
+
"step": 25290
|
| 11812 |
+
},
|
| 11813 |
+
{
|
| 11814 |
+
"epoch": 0.049277727363013564,
|
| 11815 |
+
"grad_norm": 5.145657062530518,
|
| 11816 |
+
"learning_rate": 2.9820611413929318e-05,
|
| 11817 |
+
"loss": 1.7515,
|
| 11818 |
+
"step": 25305
|
| 11819 |
+
},
|
| 11820 |
+
{
|
| 11821 |
+
"epoch": 0.049306937634123825,
|
| 11822 |
+
"grad_norm": 2.595554828643799,
|
| 11823 |
+
"learning_rate": 2.9820399104733964e-05,
|
| 11824 |
+
"loss": 1.8185,
|
| 11825 |
+
"step": 25320
|
| 11826 |
+
},
|
| 11827 |
+
{
|
| 11828 |
+
"epoch": 0.04933614790523409,
|
| 11829 |
+
"grad_norm": 3.0854387283325195,
|
| 11830 |
+
"learning_rate": 2.9820186670733944e-05,
|
| 11831 |
+
"loss": 2.0462,
|
| 11832 |
+
"step": 25335
|
| 11833 |
+
},
|
| 11834 |
+
{
|
| 11835 |
+
"epoch": 0.049365358176344354,
|
| 11836 |
+
"grad_norm": 4.086148262023926,
|
| 11837 |
+
"learning_rate": 2.9819974111931045e-05,
|
| 11838 |
+
"loss": 1.923,
|
| 11839 |
+
"step": 25350
|
| 11840 |
+
},
|
| 11841 |
+
{
|
| 11842 |
+
"epoch": 0.049394568447454615,
|
| 11843 |
+
"grad_norm": 2.3999664783477783,
|
| 11844 |
+
"learning_rate": 2.9819761428327057e-05,
|
| 11845 |
+
"loss": 1.9243,
|
| 11846 |
+
"step": 25365
|
| 11847 |
+
},
|
| 11848 |
+
{
|
| 11849 |
+
"epoch": 0.04942377871856488,
|
| 11850 |
+
"grad_norm": 1.8367825746536255,
|
| 11851 |
+
"learning_rate": 2.981954861992378e-05,
|
| 11852 |
+
"loss": 1.9104,
|
| 11853 |
+
"step": 25380
|
| 11854 |
+
},
|
| 11855 |
+
{
|
| 11856 |
+
"epoch": 0.049452988989675144,
|
| 11857 |
+
"grad_norm": 2.3233165740966797,
|
| 11858 |
+
"learning_rate": 2.9819335686722997e-05,
|
| 11859 |
+
"loss": 1.7446,
|
| 11860 |
+
"step": 25395
|
| 11861 |
+
},
|
| 11862 |
+
{
|
| 11863 |
+
"epoch": 0.049482199260785405,
|
| 11864 |
+
"grad_norm": 3.7498881816864014,
|
| 11865 |
+
"learning_rate": 2.98191226287265e-05,
|
| 11866 |
+
"loss": 2.0172,
|
| 11867 |
+
"step": 25410
|
| 11868 |
+
},
|
| 11869 |
+
{
|
| 11870 |
+
"epoch": 0.049511409531895666,
|
| 11871 |
+
"grad_norm": 2.636087656021118,
|
| 11872 |
+
"learning_rate": 2.9818909445936092e-05,
|
| 11873 |
+
"loss": 1.9958,
|
| 11874 |
+
"step": 25425
|
| 11875 |
+
},
|
| 11876 |
+
{
|
| 11877 |
+
"epoch": 0.049540619803005934,
|
| 11878 |
+
"grad_norm": 2.6167049407958984,
|
| 11879 |
+
"learning_rate": 2.9818696138353564e-05,
|
| 11880 |
+
"loss": 1.9746,
|
| 11881 |
+
"step": 25440
|
| 11882 |
+
},
|
| 11883 |
+
{
|
| 11884 |
+
"epoch": 0.049569830074116195,
|
| 11885 |
+
"grad_norm": 2.699380874633789,
|
| 11886 |
+
"learning_rate": 2.9818482705980708e-05,
|
| 11887 |
+
"loss": 2.0158,
|
| 11888 |
+
"step": 25455
|
| 11889 |
+
},
|
| 11890 |
+
{
|
| 11891 |
+
"epoch": 0.049599040345226456,
|
| 11892 |
+
"grad_norm": 2.8639230728149414,
|
| 11893 |
+
"learning_rate": 2.9818269148819326e-05,
|
| 11894 |
+
"loss": 1.8795,
|
| 11895 |
+
"step": 25470
|
| 11896 |
+
},
|
| 11897 |
+
{
|
| 11898 |
+
"epoch": 0.04962825061633672,
|
| 11899 |
+
"grad_norm": 3.6716597080230713,
|
| 11900 |
+
"learning_rate": 2.9818055466871217e-05,
|
| 11901 |
+
"loss": 2.0066,
|
| 11902 |
+
"step": 25485
|
| 11903 |
+
},
|
| 11904 |
+
{
|
| 11905 |
+
"epoch": 0.049657460887446984,
|
| 11906 |
+
"grad_norm": 3.0852763652801514,
|
| 11907 |
+
"learning_rate": 2.981784166013818e-05,
|
| 11908 |
+
"loss": 1.8392,
|
| 11909 |
+
"step": 25500
|
| 11910 |
+
},
|
| 11911 |
+
{
|
| 11912 |
+
"epoch": 0.049686671158557245,
|
| 11913 |
+
"grad_norm": 2.0497000217437744,
|
| 11914 |
+
"learning_rate": 2.981762772862201e-05,
|
| 11915 |
+
"loss": 2.0587,
|
| 11916 |
+
"step": 25515
|
| 11917 |
+
},
|
| 11918 |
+
{
|
| 11919 |
+
"epoch": 0.049715881429667506,
|
| 11920 |
+
"grad_norm": 4.302377700805664,
|
| 11921 |
+
"learning_rate": 2.9817413672324517e-05,
|
| 11922 |
+
"loss": 1.9415,
|
| 11923 |
+
"step": 25530
|
| 11924 |
+
},
|
| 11925 |
+
{
|
| 11926 |
+
"epoch": 0.049745091700777774,
|
| 11927 |
+
"grad_norm": 2.478428840637207,
|
| 11928 |
+
"learning_rate": 2.9817199491247495e-05,
|
| 11929 |
+
"loss": 2.0082,
|
| 11930 |
+
"step": 25545
|
| 11931 |
+
},
|
| 11932 |
+
{
|
| 11933 |
+
"epoch": 0.049774301971888035,
|
| 11934 |
+
"grad_norm": 3.375516891479492,
|
| 11935 |
+
"learning_rate": 2.9816985185392752e-05,
|
| 11936 |
+
"loss": 2.061,
|
| 11937 |
+
"step": 25560
|
| 11938 |
+
},
|
| 11939 |
+
{
|
| 11940 |
+
"epoch": 0.049803512242998296,
|
| 11941 |
+
"grad_norm": 2.3733456134796143,
|
| 11942 |
+
"learning_rate": 2.9816770754762094e-05,
|
| 11943 |
+
"loss": 1.8752,
|
| 11944 |
+
"step": 25575
|
| 11945 |
+
},
|
| 11946 |
+
{
|
| 11947 |
+
"epoch": 0.049832722514108564,
|
| 11948 |
+
"grad_norm": 2.8646862506866455,
|
| 11949 |
+
"learning_rate": 2.9816556199357334e-05,
|
| 11950 |
+
"loss": 1.9048,
|
| 11951 |
+
"step": 25590
|
| 11952 |
+
},
|
| 11953 |
+
{
|
| 11954 |
+
"epoch": 0.049861932785218825,
|
| 11955 |
+
"grad_norm": 3.711494207382202,
|
| 11956 |
+
"learning_rate": 2.981634151918026e-05,
|
| 11957 |
+
"loss": 1.8555,
|
| 11958 |
+
"step": 25605
|
| 11959 |
+
},
|
| 11960 |
+
{
|
| 11961 |
+
"epoch": 0.049891143056329086,
|
| 11962 |
+
"grad_norm": 4.652657985687256,
|
| 11963 |
+
"learning_rate": 2.9816126714232694e-05,
|
| 11964 |
+
"loss": 1.9591,
|
| 11965 |
+
"step": 25620
|
| 11966 |
+
},
|
| 11967 |
+
{
|
| 11968 |
+
"epoch": 0.04992035332743935,
|
| 11969 |
+
"grad_norm": 3.5854547023773193,
|
| 11970 |
+
"learning_rate": 2.981591178451644e-05,
|
| 11971 |
+
"loss": 1.8043,
|
| 11972 |
+
"step": 25635
|
| 11973 |
+
},
|
| 11974 |
+
{
|
| 11975 |
+
"epoch": 0.049949563598549615,
|
| 11976 |
+
"grad_norm": 2.532128095626831,
|
| 11977 |
+
"learning_rate": 2.981569673003331e-05,
|
| 11978 |
+
"loss": 1.7456,
|
| 11979 |
+
"step": 25650
|
| 11980 |
+
},
|
| 11981 |
+
{
|
| 11982 |
+
"epoch": 0.049978773869659876,
|
| 11983 |
+
"grad_norm": 1.8968348503112793,
|
| 11984 |
+
"learning_rate": 2.9815481550785116e-05,
|
| 11985 |
+
"loss": 1.8971,
|
| 11986 |
+
"step": 25665
|
| 11987 |
+
},
|
| 11988 |
+
{
|
| 11989 |
+
"epoch": 0.05000798414077014,
|
| 11990 |
+
"grad_norm": 4.662414073944092,
|
| 11991 |
+
"learning_rate": 2.9815266246773663e-05,
|
| 11992 |
+
"loss": 2.1576,
|
| 11993 |
+
"step": 25680
|
| 11994 |
+
},
|
| 11995 |
+
{
|
| 11996 |
+
"epoch": 0.050037194411880405,
|
| 11997 |
+
"grad_norm": 2.1178083419799805,
|
| 11998 |
+
"learning_rate": 2.9815050818000773e-05,
|
| 11999 |
+
"loss": 1.9055,
|
| 12000 |
+
"step": 25695
|
| 12001 |
+
},
|
| 12002 |
+
{
|
| 12003 |
+
"epoch": 0.050066404682990666,
|
| 12004 |
+
"grad_norm": 3.8748810291290283,
|
| 12005 |
+
"learning_rate": 2.9814835264468254e-05,
|
| 12006 |
+
"loss": 1.8177,
|
| 12007 |
+
"step": 25710
|
| 12008 |
+
},
|
| 12009 |
+
{
|
| 12010 |
+
"epoch": 0.05009561495410093,
|
| 12011 |
+
"grad_norm": 3.424405097961426,
|
| 12012 |
+
"learning_rate": 2.9814619586177926e-05,
|
| 12013 |
+
"loss": 1.8804,
|
| 12014 |
+
"step": 25725
|
| 12015 |
+
},
|
| 12016 |
+
{
|
| 12017 |
+
"epoch": 0.05012482522521119,
|
| 12018 |
+
"grad_norm": 4.104612827301025,
|
| 12019 |
+
"learning_rate": 2.98144037831316e-05,
|
| 12020 |
+
"loss": 1.973,
|
| 12021 |
+
"step": 25740
|
| 12022 |
+
},
|
| 12023 |
+
{
|
| 12024 |
+
"epoch": 0.050154035496321456,
|
| 12025 |
+
"grad_norm": 2.1848320960998535,
|
| 12026 |
+
"learning_rate": 2.98141878553311e-05,
|
| 12027 |
+
"loss": 1.8529,
|
| 12028 |
+
"step": 25755
|
| 12029 |
+
},
|
| 12030 |
+
{
|
| 12031 |
+
"epoch": 0.05018324576743172,
|
| 12032 |
+
"grad_norm": 4.191700458526611,
|
| 12033 |
+
"learning_rate": 2.981397180277824e-05,
|
| 12034 |
+
"loss": 1.8094,
|
| 12035 |
+
"step": 25770
|
| 12036 |
+
},
|
| 12037 |
+
{
|
| 12038 |
+
"epoch": 0.05021245603854198,
|
| 12039 |
+
"grad_norm": 3.3454208374023438,
|
| 12040 |
+
"learning_rate": 2.981375562547484e-05,
|
| 12041 |
+
"loss": 1.9105,
|
| 12042 |
+
"step": 25785
|
| 12043 |
+
},
|
| 12044 |
+
{
|
| 12045 |
+
"epoch": 0.05024166630965224,
|
| 12046 |
+
"grad_norm": 2.848618507385254,
|
| 12047 |
+
"learning_rate": 2.9813539323422717e-05,
|
| 12048 |
+
"loss": 1.9054,
|
| 12049 |
+
"step": 25800
|
| 12050 |
+
},
|
| 12051 |
+
{
|
| 12052 |
+
"epoch": 0.05027087658076251,
|
| 12053 |
+
"grad_norm": 1.7352811098098755,
|
| 12054 |
+
"learning_rate": 2.98133228966237e-05,
|
| 12055 |
+
"loss": 1.91,
|
| 12056 |
+
"step": 25815
|
| 12057 |
+
},
|
| 12058 |
+
{
|
| 12059 |
+
"epoch": 0.05030008685187277,
|
| 12060 |
+
"grad_norm": 2.177297592163086,
|
| 12061 |
+
"learning_rate": 2.9813106345079604e-05,
|
| 12062 |
+
"loss": 1.8276,
|
| 12063 |
+
"step": 25830
|
| 12064 |
+
},
|
| 12065 |
+
{
|
| 12066 |
+
"epoch": 0.05032929712298303,
|
| 12067 |
+
"grad_norm": 2.1822316646575928,
|
| 12068 |
+
"learning_rate": 2.981288966879226e-05,
|
| 12069 |
+
"loss": 1.7961,
|
| 12070 |
+
"step": 25845
|
| 12071 |
+
},
|
| 12072 |
+
{
|
| 12073 |
+
"epoch": 0.050358507394093296,
|
| 12074 |
+
"grad_norm": 2.70729398727417,
|
| 12075 |
+
"learning_rate": 2.9812672867763482e-05,
|
| 12076 |
+
"loss": 1.7497,
|
| 12077 |
+
"step": 25860
|
| 12078 |
+
},
|
| 12079 |
+
{
|
| 12080 |
+
"epoch": 0.05038771766520356,
|
| 12081 |
+
"grad_norm": 2.869450807571411,
|
| 12082 |
+
"learning_rate": 2.981245594199511e-05,
|
| 12083 |
+
"loss": 1.8309,
|
| 12084 |
+
"step": 25875
|
| 12085 |
+
},
|
| 12086 |
+
{
|
| 12087 |
+
"epoch": 0.05041692793631382,
|
| 12088 |
+
"grad_norm": 2.233219623565674,
|
| 12089 |
+
"learning_rate": 2.981223889148896e-05,
|
| 12090 |
+
"loss": 1.8577,
|
| 12091 |
+
"step": 25890
|
| 12092 |
+
},
|
| 12093 |
+
{
|
| 12094 |
+
"epoch": 0.05044613820742408,
|
| 12095 |
+
"grad_norm": 5.402493000030518,
|
| 12096 |
+
"learning_rate": 2.981202171624686e-05,
|
| 12097 |
+
"loss": 1.9151,
|
| 12098 |
+
"step": 25905
|
| 12099 |
+
},
|
| 12100 |
+
{
|
| 12101 |
+
"epoch": 0.05047534847853435,
|
| 12102 |
+
"grad_norm": 4.127412796020508,
|
| 12103 |
+
"learning_rate": 2.9811804416270648e-05,
|
| 12104 |
+
"loss": 1.8644,
|
| 12105 |
+
"step": 25920
|
| 12106 |
+
},
|
| 12107 |
+
{
|
| 12108 |
+
"epoch": 0.05050455874964461,
|
| 12109 |
+
"grad_norm": 2.9654555320739746,
|
| 12110 |
+
"learning_rate": 2.9811586991562145e-05,
|
| 12111 |
+
"loss": 1.99,
|
| 12112 |
+
"step": 25935
|
| 12113 |
+
},
|
| 12114 |
+
{
|
| 12115 |
+
"epoch": 0.05053376902075487,
|
| 12116 |
+
"grad_norm": 2.6316134929656982,
|
| 12117 |
+
"learning_rate": 2.981136944212318e-05,
|
| 12118 |
+
"loss": 1.7434,
|
| 12119 |
+
"step": 25950
|
| 12120 |
+
},
|
| 12121 |
+
{
|
| 12122 |
+
"epoch": 0.05056297929186514,
|
| 12123 |
+
"grad_norm": 2.6688010692596436,
|
| 12124 |
+
"learning_rate": 2.9811151767955597e-05,
|
| 12125 |
+
"loss": 1.8075,
|
| 12126 |
+
"step": 25965
|
| 12127 |
+
},
|
| 12128 |
+
{
|
| 12129 |
+
"epoch": 0.0505921895629754,
|
| 12130 |
+
"grad_norm": 1.7729136943817139,
|
| 12131 |
+
"learning_rate": 2.981093396906122e-05,
|
| 12132 |
+
"loss": 1.7828,
|
| 12133 |
+
"step": 25980
|
| 12134 |
+
},
|
| 12135 |
+
{
|
| 12136 |
+
"epoch": 0.05062139983408566,
|
| 12137 |
+
"grad_norm": 3.017512321472168,
|
| 12138 |
+
"learning_rate": 2.9810716045441884e-05,
|
| 12139 |
+
"loss": 1.9602,
|
| 12140 |
+
"step": 25995
|
| 12141 |
+
},
|
| 12142 |
+
{
|
| 12143 |
+
"epoch": 0.05065061010519592,
|
| 12144 |
+
"grad_norm": 3.954099416732788,
|
| 12145 |
+
"learning_rate": 2.9810497997099427e-05,
|
| 12146 |
+
"loss": 1.9158,
|
| 12147 |
+
"step": 26010
|
| 12148 |
+
},
|
| 12149 |
+
{
|
| 12150 |
+
"epoch": 0.05067982037630619,
|
| 12151 |
+
"grad_norm": 4.384187698364258,
|
| 12152 |
+
"learning_rate": 2.981027982403568e-05,
|
| 12153 |
+
"loss": 2.0172,
|
| 12154 |
+
"step": 26025
|
| 12155 |
+
},
|
| 12156 |
+
{
|
| 12157 |
+
"epoch": 0.05070903064741645,
|
| 12158 |
+
"grad_norm": 2.7591800689697266,
|
| 12159 |
+
"learning_rate": 2.9810061526252488e-05,
|
| 12160 |
+
"loss": 1.8372,
|
| 12161 |
+
"step": 26040
|
| 12162 |
+
},
|
| 12163 |
+
{
|
| 12164 |
+
"epoch": 0.05073824091852671,
|
| 12165 |
+
"grad_norm": 2.6615748405456543,
|
| 12166 |
+
"learning_rate": 2.980984310375168e-05,
|
| 12167 |
+
"loss": 1.8127,
|
| 12168 |
+
"step": 26055
|
| 12169 |
+
},
|
| 12170 |
+
{
|
| 12171 |
+
"epoch": 0.05076745118963698,
|
| 12172 |
+
"grad_norm": 3.1193888187408447,
|
| 12173 |
+
"learning_rate": 2.9809624556535106e-05,
|
| 12174 |
+
"loss": 1.8826,
|
| 12175 |
+
"step": 26070
|
| 12176 |
+
},
|
| 12177 |
+
{
|
| 12178 |
+
"epoch": 0.05079666146074724,
|
| 12179 |
+
"grad_norm": 2.338146448135376,
|
| 12180 |
+
"learning_rate": 2.9809405884604594e-05,
|
| 12181 |
+
"loss": 1.9003,
|
| 12182 |
+
"step": 26085
|
| 12183 |
+
},
|
| 12184 |
+
{
|
| 12185 |
+
"epoch": 0.0508258717318575,
|
| 12186 |
+
"grad_norm": 2.339162588119507,
|
| 12187 |
+
"learning_rate": 2.9809187087961993e-05,
|
| 12188 |
+
"loss": 1.8444,
|
| 12189 |
+
"step": 26100
|
| 12190 |
+
},
|
| 12191 |
+
{
|
| 12192 |
+
"epoch": 0.05085508200296776,
|
| 12193 |
+
"grad_norm": 6.189070224761963,
|
| 12194 |
+
"learning_rate": 2.980896816660915e-05,
|
| 12195 |
+
"loss": 1.8682,
|
| 12196 |
+
"step": 26115
|
| 12197 |
+
},
|
| 12198 |
+
{
|
| 12199 |
+
"epoch": 0.05088429227407803,
|
| 12200 |
+
"grad_norm": 3.7556257247924805,
|
| 12201 |
+
"learning_rate": 2.9808749120547898e-05,
|
| 12202 |
+
"loss": 1.8692,
|
| 12203 |
+
"step": 26130
|
| 12204 |
+
},
|
| 12205 |
+
{
|
| 12206 |
+
"epoch": 0.05091350254518829,
|
| 12207 |
+
"grad_norm": 2.4181950092315674,
|
| 12208 |
+
"learning_rate": 2.980852994978009e-05,
|
| 12209 |
+
"loss": 1.8497,
|
| 12210 |
+
"step": 26145
|
| 12211 |
+
},
|
| 12212 |
+
{
|
| 12213 |
+
"epoch": 0.05094271281629855,
|
| 12214 |
+
"grad_norm": 2.1751251220703125,
|
| 12215 |
+
"learning_rate": 2.9808310654307566e-05,
|
| 12216 |
+
"loss": 1.7236,
|
| 12217 |
+
"step": 26160
|
| 12218 |
+
},
|
| 12219 |
+
{
|
| 12220 |
+
"epoch": 0.05097192308740882,
|
| 12221 |
+
"grad_norm": 2.304203748703003,
|
| 12222 |
+
"learning_rate": 2.9808091234132177e-05,
|
| 12223 |
+
"loss": 1.8394,
|
| 12224 |
+
"step": 26175
|
| 12225 |
+
},
|
| 12226 |
+
{
|
| 12227 |
+
"epoch": 0.05100113335851908,
|
| 12228 |
+
"grad_norm": 2.03320050239563,
|
| 12229 |
+
"learning_rate": 2.980787168925577e-05,
|
| 12230 |
+
"loss": 2.0082,
|
| 12231 |
+
"step": 26190
|
| 12232 |
+
},
|
| 12233 |
+
{
|
| 12234 |
+
"epoch": 0.05103034362962934,
|
| 12235 |
+
"grad_norm": 3.1527299880981445,
|
| 12236 |
+
"learning_rate": 2.9807652019680195e-05,
|
| 12237 |
+
"loss": 1.8783,
|
| 12238 |
+
"step": 26205
|
| 12239 |
+
},
|
| 12240 |
+
{
|
| 12241 |
+
"epoch": 0.0510595539007396,
|
| 12242 |
+
"grad_norm": 3.638120412826538,
|
| 12243 |
+
"learning_rate": 2.9807432225407295e-05,
|
| 12244 |
+
"loss": 1.8912,
|
| 12245 |
+
"step": 26220
|
| 12246 |
+
},
|
| 12247 |
+
{
|
| 12248 |
+
"epoch": 0.05108876417184987,
|
| 12249 |
+
"grad_norm": 1.9177019596099854,
|
| 12250 |
+
"learning_rate": 2.9807212306438927e-05,
|
| 12251 |
+
"loss": 1.7949,
|
| 12252 |
+
"step": 26235
|
| 12253 |
+
},
|
| 12254 |
+
{
|
| 12255 |
+
"epoch": 0.05111797444296013,
|
| 12256 |
+
"grad_norm": 4.763120174407959,
|
| 12257 |
+
"learning_rate": 2.9806992262776945e-05,
|
| 12258 |
+
"loss": 1.9464,
|
| 12259 |
+
"step": 26250
|
| 12260 |
+
},
|
| 12261 |
+
{
|
| 12262 |
+
"epoch": 0.05114718471407039,
|
| 12263 |
+
"grad_norm": 4.4592132568359375,
|
| 12264 |
+
"learning_rate": 2.980677209442319e-05,
|
| 12265 |
+
"loss": 1.834,
|
| 12266 |
+
"step": 26265
|
| 12267 |
+
},
|
| 12268 |
+
{
|
| 12269 |
+
"epoch": 0.05117639498518065,
|
| 12270 |
+
"grad_norm": 2.6289279460906982,
|
| 12271 |
+
"learning_rate": 2.980655180137953e-05,
|
| 12272 |
+
"loss": 1.7644,
|
| 12273 |
+
"step": 26280
|
| 12274 |
+
},
|
| 12275 |
+
{
|
| 12276 |
+
"epoch": 0.05120560525629092,
|
| 12277 |
+
"grad_norm": 2.2794735431671143,
|
| 12278 |
+
"learning_rate": 2.9806331383647816e-05,
|
| 12279 |
+
"loss": 1.9125,
|
| 12280 |
+
"step": 26295
|
| 12281 |
+
},
|
| 12282 |
+
{
|
| 12283 |
+
"epoch": 0.05123481552740118,
|
| 12284 |
+
"grad_norm": 3.5884125232696533,
|
| 12285 |
+
"learning_rate": 2.9806110841229904e-05,
|
| 12286 |
+
"loss": 1.925,
|
| 12287 |
+
"step": 26310
|
| 12288 |
+
},
|
| 12289 |
+
{
|
| 12290 |
+
"epoch": 0.05126402579851144,
|
| 12291 |
+
"grad_norm": 2.8750176429748535,
|
| 12292 |
+
"learning_rate": 2.9805890174127648e-05,
|
| 12293 |
+
"loss": 1.6463,
|
| 12294 |
+
"step": 26325
|
| 12295 |
+
},
|
| 12296 |
+
{
|
| 12297 |
+
"epoch": 0.05129323606962171,
|
| 12298 |
+
"grad_norm": 3.1167027950286865,
|
| 12299 |
+
"learning_rate": 2.980566938234291e-05,
|
| 12300 |
+
"loss": 1.8897,
|
| 12301 |
+
"step": 26340
|
| 12302 |
+
},
|
| 12303 |
+
{
|
| 12304 |
+
"epoch": 0.05132244634073197,
|
| 12305 |
+
"grad_norm": 2.8106181621551514,
|
| 12306 |
+
"learning_rate": 2.9805448465877546e-05,
|
| 12307 |
+
"loss": 1.855,
|
| 12308 |
+
"step": 26355
|
| 12309 |
+
},
|
| 12310 |
+
{
|
| 12311 |
+
"epoch": 0.05135165661184223,
|
| 12312 |
+
"grad_norm": 3.8905303478240967,
|
| 12313 |
+
"learning_rate": 2.980522742473342e-05,
|
| 12314 |
+
"loss": 1.9127,
|
| 12315 |
+
"step": 26370
|
| 12316 |
+
},
|
| 12317 |
+
{
|
| 12318 |
+
"epoch": 0.05138086688295249,
|
| 12319 |
+
"grad_norm": 3.031163215637207,
|
| 12320 |
+
"learning_rate": 2.980500625891239e-05,
|
| 12321 |
+
"loss": 1.9577,
|
| 12322 |
+
"step": 26385
|
| 12323 |
+
},
|
| 12324 |
+
{
|
| 12325 |
+
"epoch": 0.05141007715406276,
|
| 12326 |
+
"grad_norm": 1.991543173789978,
|
| 12327 |
+
"learning_rate": 2.980478496841632e-05,
|
| 12328 |
+
"loss": 1.6416,
|
| 12329 |
+
"step": 26400
|
| 12330 |
+
},
|
| 12331 |
+
{
|
| 12332 |
+
"epoch": 0.05143928742517302,
|
| 12333 |
+
"grad_norm": 3.90432071685791,
|
| 12334 |
+
"learning_rate": 2.9804563553247076e-05,
|
| 12335 |
+
"loss": 1.7318,
|
| 12336 |
+
"step": 26415
|
| 12337 |
+
},
|
| 12338 |
+
{
|
| 12339 |
+
"epoch": 0.05146849769628328,
|
| 12340 |
+
"grad_norm": 3.582280158996582,
|
| 12341 |
+
"learning_rate": 2.980434201340652e-05,
|
| 12342 |
+
"loss": 1.7349,
|
| 12343 |
+
"step": 26430
|
| 12344 |
+
},
|
| 12345 |
+
{
|
| 12346 |
+
"epoch": 0.05149770796739355,
|
| 12347 |
+
"grad_norm": 3.1633496284484863,
|
| 12348 |
+
"learning_rate": 2.980412034889651e-05,
|
| 12349 |
+
"loss": 1.8023,
|
| 12350 |
+
"step": 26445
|
| 12351 |
+
},
|
| 12352 |
+
{
|
| 12353 |
+
"epoch": 0.05152691823850381,
|
| 12354 |
+
"grad_norm": 3.7939155101776123,
|
| 12355 |
+
"learning_rate": 2.9803898559718927e-05,
|
| 12356 |
+
"loss": 1.6626,
|
| 12357 |
+
"step": 26460
|
| 12358 |
+
},
|
| 12359 |
+
{
|
| 12360 |
+
"epoch": 0.05155612850961407,
|
| 12361 |
+
"grad_norm": 3.093492031097412,
|
| 12362 |
+
"learning_rate": 2.9803676645875634e-05,
|
| 12363 |
+
"loss": 1.7914,
|
| 12364 |
+
"step": 26475
|
| 12365 |
+
},
|
| 12366 |
+
{
|
| 12367 |
+
"epoch": 0.051585338780724334,
|
| 12368 |
+
"grad_norm": 2.4469592571258545,
|
| 12369 |
+
"learning_rate": 2.9803454607368493e-05,
|
| 12370 |
+
"loss": 1.7484,
|
| 12371 |
+
"step": 26490
|
| 12372 |
+
},
|
| 12373 |
+
{
|
| 12374 |
+
"epoch": 0.0516145490518346,
|
| 12375 |
+
"grad_norm": 2.613276243209839,
|
| 12376 |
+
"learning_rate": 2.9803232444199382e-05,
|
| 12377 |
+
"loss": 1.7984,
|
| 12378 |
+
"step": 26505
|
| 12379 |
+
},
|
| 12380 |
+
{
|
| 12381 |
+
"epoch": 0.05164375932294486,
|
| 12382 |
+
"grad_norm": 3.1793646812438965,
|
| 12383 |
+
"learning_rate": 2.9803010156370166e-05,
|
| 12384 |
+
"loss": 1.7114,
|
| 12385 |
+
"step": 26520
|
| 12386 |
+
},
|
| 12387 |
+
{
|
| 12388 |
+
"epoch": 0.051672969594055124,
|
| 12389 |
+
"grad_norm": 1.9021662473678589,
|
| 12390 |
+
"learning_rate": 2.980278774388272e-05,
|
| 12391 |
+
"loss": 1.6837,
|
| 12392 |
+
"step": 26535
|
| 12393 |
+
},
|
| 12394 |
+
{
|
| 12395 |
+
"epoch": 0.05170217986516539,
|
| 12396 |
+
"grad_norm": 2.415710687637329,
|
| 12397 |
+
"learning_rate": 2.9802565206738922e-05,
|
| 12398 |
+
"loss": 1.9818,
|
| 12399 |
+
"step": 26550
|
| 12400 |
+
},
|
| 12401 |
+
{
|
| 12402 |
+
"epoch": 0.05173139013627565,
|
| 12403 |
+
"grad_norm": 3.8227622509002686,
|
| 12404 |
+
"learning_rate": 2.9802342544940635e-05,
|
| 12405 |
+
"loss": 1.9691,
|
| 12406 |
+
"step": 26565
|
| 12407 |
+
},
|
| 12408 |
+
{
|
| 12409 |
+
"epoch": 0.051760600407385914,
|
| 12410 |
+
"grad_norm": 1.6945210695266724,
|
| 12411 |
+
"learning_rate": 2.980211975848974e-05,
|
| 12412 |
+
"loss": 1.9788,
|
| 12413 |
+
"step": 26580
|
| 12414 |
+
},
|
| 12415 |
+
{
|
| 12416 |
+
"epoch": 0.051789810678496175,
|
| 12417 |
+
"grad_norm": 2.1171348094940186,
|
| 12418 |
+
"learning_rate": 2.980189684738811e-05,
|
| 12419 |
+
"loss": 1.8498,
|
| 12420 |
+
"step": 26595
|
| 12421 |
+
},
|
| 12422 |
+
{
|
| 12423 |
+
"epoch": 0.05181902094960644,
|
| 12424 |
+
"grad_norm": 3.8129332065582275,
|
| 12425 |
+
"learning_rate": 2.9801673811637628e-05,
|
| 12426 |
+
"loss": 1.8109,
|
| 12427 |
+
"step": 26610
|
| 12428 |
+
},
|
| 12429 |
+
{
|
| 12430 |
+
"epoch": 0.0518482312207167,
|
| 12431 |
+
"grad_norm": 3.1358275413513184,
|
| 12432 |
+
"learning_rate": 2.9801450651240173e-05,
|
| 12433 |
+
"loss": 1.7809,
|
| 12434 |
+
"step": 26625
|
| 12435 |
+
},
|
| 12436 |
+
{
|
| 12437 |
+
"epoch": 0.051877441491826964,
|
| 12438 |
+
"grad_norm": 3.9287755489349365,
|
| 12439 |
+
"learning_rate": 2.9801227366197614e-05,
|
| 12440 |
+
"loss": 1.7791,
|
| 12441 |
+
"step": 26640
|
| 12442 |
+
},
|
| 12443 |
+
{
|
| 12444 |
+
"epoch": 0.05190665176293723,
|
| 12445 |
+
"grad_norm": 3.269742488861084,
|
| 12446 |
+
"learning_rate": 2.980100395651184e-05,
|
| 12447 |
+
"loss": 2.0053,
|
| 12448 |
+
"step": 26655
|
| 12449 |
+
},
|
| 12450 |
+
{
|
| 12451 |
+
"epoch": 0.05193586203404749,
|
| 12452 |
+
"grad_norm": 3.23358154296875,
|
| 12453 |
+
"learning_rate": 2.980078042218473e-05,
|
| 12454 |
+
"loss": 1.7863,
|
| 12455 |
+
"step": 26670
|
| 12456 |
+
},
|
| 12457 |
+
{
|
| 12458 |
+
"epoch": 0.051965072305157754,
|
| 12459 |
+
"grad_norm": 3.6313607692718506,
|
| 12460 |
+
"learning_rate": 2.980055676321817e-05,
|
| 12461 |
+
"loss": 1.8125,
|
| 12462 |
+
"step": 26685
|
| 12463 |
+
},
|
| 12464 |
+
{
|
| 12465 |
+
"epoch": 0.051994282576268015,
|
| 12466 |
+
"grad_norm": 2.303229331970215,
|
| 12467 |
+
"learning_rate": 2.9800332979614035e-05,
|
| 12468 |
+
"loss": 1.8336,
|
| 12469 |
+
"step": 26700
|
| 12470 |
+
},
|
| 12471 |
+
{
|
| 12472 |
+
"epoch": 0.05202349284737828,
|
| 12473 |
+
"grad_norm": 4.923130035400391,
|
| 12474 |
+
"learning_rate": 2.9800109071374216e-05,
|
| 12475 |
+
"loss": 1.9392,
|
| 12476 |
+
"step": 26715
|
| 12477 |
+
},
|
| 12478 |
+
{
|
| 12479 |
+
"epoch": 0.052052703118488544,
|
| 12480 |
+
"grad_norm": 6.160820484161377,
|
| 12481 |
+
"learning_rate": 2.9799885038500597e-05,
|
| 12482 |
+
"loss": 1.8991,
|
| 12483 |
+
"step": 26730
|
| 12484 |
+
},
|
| 12485 |
+
{
|
| 12486 |
+
"epoch": 0.052081913389598805,
|
| 12487 |
+
"grad_norm": 1.9204509258270264,
|
| 12488 |
+
"learning_rate": 2.9799660880995065e-05,
|
| 12489 |
+
"loss": 1.8511,
|
| 12490 |
+
"step": 26745
|
| 12491 |
+
},
|
| 12492 |
+
{
|
| 12493 |
+
"epoch": 0.05211112366070907,
|
| 12494 |
+
"grad_norm": 4.1540656089782715,
|
| 12495 |
+
"learning_rate": 2.9799436598859507e-05,
|
| 12496 |
+
"loss": 1.7353,
|
| 12497 |
+
"step": 26760
|
| 12498 |
+
},
|
| 12499 |
+
{
|
| 12500 |
+
"epoch": 0.052140333931819334,
|
| 12501 |
+
"grad_norm": 2.5721335411071777,
|
| 12502 |
+
"learning_rate": 2.979921219209581e-05,
|
| 12503 |
+
"loss": 1.6949,
|
| 12504 |
+
"step": 26775
|
| 12505 |
+
},
|
| 12506 |
+
{
|
| 12507 |
+
"epoch": 0.052169544202929595,
|
| 12508 |
+
"grad_norm": 2.5524466037750244,
|
| 12509 |
+
"learning_rate": 2.9798987660705867e-05,
|
| 12510 |
+
"loss": 1.9033,
|
| 12511 |
+
"step": 26790
|
| 12512 |
+
},
|
| 12513 |
+
{
|
| 12514 |
+
"epoch": 0.052198754474039856,
|
| 12515 |
+
"grad_norm": 2.8870813846588135,
|
| 12516 |
+
"learning_rate": 2.979876300469157e-05,
|
| 12517 |
+
"loss": 1.6899,
|
| 12518 |
+
"step": 26805
|
| 12519 |
+
},
|
| 12520 |
+
{
|
| 12521 |
+
"epoch": 0.052227964745150124,
|
| 12522 |
+
"grad_norm": 2.5837631225585938,
|
| 12523 |
+
"learning_rate": 2.9798538224054804e-05,
|
| 12524 |
+
"loss": 1.7715,
|
| 12525 |
+
"step": 26820
|
| 12526 |
+
},
|
| 12527 |
+
{
|
| 12528 |
+
"epoch": 0.052257175016260385,
|
| 12529 |
+
"grad_norm": 2.5015158653259277,
|
| 12530 |
+
"learning_rate": 2.979831331879747e-05,
|
| 12531 |
+
"loss": 1.9121,
|
| 12532 |
+
"step": 26835
|
| 12533 |
+
},
|
| 12534 |
+
{
|
| 12535 |
+
"epoch": 0.052286385287370646,
|
| 12536 |
+
"grad_norm": 3.3133704662323,
|
| 12537 |
+
"learning_rate": 2.9798088288921457e-05,
|
| 12538 |
+
"loss": 1.87,
|
| 12539 |
+
"step": 26850
|
| 12540 |
+
},
|
| 12541 |
+
{
|
| 12542 |
+
"epoch": 0.05231559555848091,
|
| 12543 |
+
"grad_norm": 2.159090042114258,
|
| 12544 |
+
"learning_rate": 2.979786313442866e-05,
|
| 12545 |
+
"loss": 1.9665,
|
| 12546 |
+
"step": 26865
|
| 12547 |
+
},
|
| 12548 |
+
{
|
| 12549 |
+
"epoch": 0.052344805829591175,
|
| 12550 |
+
"grad_norm": 3.6745707988739014,
|
| 12551 |
+
"learning_rate": 2.9797637855320977e-05,
|
| 12552 |
+
"loss": 1.8956,
|
| 12553 |
+
"step": 26880
|
| 12554 |
+
},
|
| 12555 |
+
{
|
| 12556 |
+
"epoch": 0.052374016100701436,
|
| 12557 |
+
"grad_norm": 2.593538761138916,
|
| 12558 |
+
"learning_rate": 2.9797412451600305e-05,
|
| 12559 |
+
"loss": 1.9206,
|
| 12560 |
+
"step": 26895
|
| 12561 |
+
},
|
| 12562 |
+
{
|
| 12563 |
+
"epoch": 0.0524032263718117,
|
| 12564 |
+
"grad_norm": 2.1524672508239746,
|
| 12565 |
+
"learning_rate": 2.979718692326854e-05,
|
| 12566 |
+
"loss": 2.0598,
|
| 12567 |
+
"step": 26910
|
| 12568 |
+
},
|
| 12569 |
+
{
|
| 12570 |
+
"epoch": 0.052432436642921965,
|
| 12571 |
+
"grad_norm": 5.403210639953613,
|
| 12572 |
+
"learning_rate": 2.9796961270327583e-05,
|
| 12573 |
+
"loss": 1.9446,
|
| 12574 |
+
"step": 26925
|
| 12575 |
+
},
|
| 12576 |
+
{
|
| 12577 |
+
"epoch": 0.052461646914032226,
|
| 12578 |
+
"grad_norm": 2.859320878982544,
|
| 12579 |
+
"learning_rate": 2.9796735492779338e-05,
|
| 12580 |
+
"loss": 1.6969,
|
| 12581 |
+
"step": 26940
|
| 12582 |
+
},
|
| 12583 |
+
{
|
| 12584 |
+
"epoch": 0.05249085718514249,
|
| 12585 |
+
"grad_norm": 2.0825371742248535,
|
| 12586 |
+
"learning_rate": 2.9796509590625696e-05,
|
| 12587 |
+
"loss": 1.8951,
|
| 12588 |
+
"step": 26955
|
| 12589 |
+
},
|
| 12590 |
+
{
|
| 12591 |
+
"epoch": 0.05252006745625275,
|
| 12592 |
+
"grad_norm": 2.3604981899261475,
|
| 12593 |
+
"learning_rate": 2.979628356386857e-05,
|
| 12594 |
+
"loss": 1.8861,
|
| 12595 |
+
"step": 26970
|
| 12596 |
+
},
|
| 12597 |
+
{
|
| 12598 |
+
"epoch": 0.052549277727363015,
|
| 12599 |
+
"grad_norm": 3.510629415512085,
|
| 12600 |
+
"learning_rate": 2.9796057412509856e-05,
|
| 12601 |
+
"loss": 1.9885,
|
| 12602 |
+
"step": 26985
|
| 12603 |
+
},
|
| 12604 |
+
{
|
| 12605 |
+
"epoch": 0.052578487998473276,
|
| 12606 |
+
"grad_norm": 1.72383713722229,
|
| 12607 |
+
"learning_rate": 2.9795831136551467e-05,
|
| 12608 |
+
"loss": 1.832,
|
| 12609 |
+
"step": 27000
|
| 12610 |
+
},
|
| 12611 |
+
{
|
| 12612 |
+
"epoch": 0.05260769826958354,
|
| 12613 |
+
"grad_norm": 2.767523765563965,
|
| 12614 |
+
"learning_rate": 2.9795604735995297e-05,
|
| 12615 |
+
"loss": 1.8956,
|
| 12616 |
+
"step": 27015
|
| 12617 |
+
},
|
| 12618 |
+
{
|
| 12619 |
+
"epoch": 0.052636908540693805,
|
| 12620 |
+
"grad_norm": 2.731154203414917,
|
| 12621 |
+
"learning_rate": 2.979537821084326e-05,
|
| 12622 |
+
"loss": 1.7157,
|
| 12623 |
+
"step": 27030
|
| 12624 |
+
},
|
| 12625 |
+
{
|
| 12626 |
+
"epoch": 0.052666118811804066,
|
| 12627 |
+
"grad_norm": 3.17053484916687,
|
| 12628 |
+
"learning_rate": 2.9795151561097265e-05,
|
| 12629 |
+
"loss": 1.9691,
|
| 12630 |
+
"step": 27045
|
| 12631 |
+
},
|
| 12632 |
+
{
|
| 12633 |
+
"epoch": 0.05269532908291433,
|
| 12634 |
+
"grad_norm": 3.9647130966186523,
|
| 12635 |
+
"learning_rate": 2.979492478675922e-05,
|
| 12636 |
+
"loss": 1.9736,
|
| 12637 |
+
"step": 27060
|
| 12638 |
+
},
|
| 12639 |
+
{
|
| 12640 |
+
"epoch": 0.05272453935402459,
|
| 12641 |
+
"grad_norm": 2.4074623584747314,
|
| 12642 |
+
"learning_rate": 2.9794697887831027e-05,
|
| 12643 |
+
"loss": 1.8325,
|
| 12644 |
+
"step": 27075
|
| 12645 |
+
},
|
| 12646 |
+
{
|
| 12647 |
+
"epoch": 0.052753749625134856,
|
| 12648 |
+
"grad_norm": 4.785901069641113,
|
| 12649 |
+
"learning_rate": 2.9794470864314603e-05,
|
| 12650 |
+
"loss": 1.9717,
|
| 12651 |
+
"step": 27090
|
| 12652 |
+
},
|
| 12653 |
+
{
|
| 12654 |
+
"epoch": 0.05278295989624512,
|
| 12655 |
+
"grad_norm": 5.4298577308654785,
|
| 12656 |
+
"learning_rate": 2.979424371621186e-05,
|
| 12657 |
+
"loss": 1.8316,
|
| 12658 |
+
"step": 27105
|
| 12659 |
+
},
|
| 12660 |
+
{
|
| 12661 |
+
"epoch": 0.05281217016735538,
|
| 12662 |
+
"grad_norm": 2.509413003921509,
|
| 12663 |
+
"learning_rate": 2.9794016443524713e-05,
|
| 12664 |
+
"loss": 1.8792,
|
| 12665 |
+
"step": 27120
|
| 12666 |
+
},
|
| 12667 |
+
{
|
| 12668 |
+
"epoch": 0.052841380438465646,
|
| 12669 |
+
"grad_norm": 1.903182029724121,
|
| 12670 |
+
"learning_rate": 2.979378904625507e-05,
|
| 12671 |
+
"loss": 1.8049,
|
| 12672 |
+
"step": 27135
|
| 12673 |
+
},
|
| 12674 |
+
{
|
| 12675 |
+
"epoch": 0.05287059070957591,
|
| 12676 |
+
"grad_norm": 3.3434927463531494,
|
| 12677 |
+
"learning_rate": 2.9793561524404846e-05,
|
| 12678 |
+
"loss": 1.7794,
|
| 12679 |
+
"step": 27150
|
| 12680 |
+
},
|
| 12681 |
+
{
|
| 12682 |
+
"epoch": 0.05289980098068617,
|
| 12683 |
+
"grad_norm": 5.064967632293701,
|
| 12684 |
+
"learning_rate": 2.9793333877975964e-05,
|
| 12685 |
+
"loss": 1.8726,
|
| 12686 |
+
"step": 27165
|
| 12687 |
+
},
|
| 12688 |
+
{
|
| 12689 |
+
"epoch": 0.05292901125179643,
|
| 12690 |
+
"grad_norm": 1.8450191020965576,
|
| 12691 |
+
"learning_rate": 2.9793106106970335e-05,
|
| 12692 |
+
"loss": 1.8586,
|
| 12693 |
+
"step": 27180
|
| 12694 |
+
},
|
| 12695 |
+
{
|
| 12696 |
+
"epoch": 0.0529582215229067,
|
| 12697 |
+
"grad_norm": 2.540570020675659,
|
| 12698 |
+
"learning_rate": 2.979287821138988e-05,
|
| 12699 |
+
"loss": 1.8988,
|
| 12700 |
+
"step": 27195
|
| 12701 |
+
},
|
| 12702 |
+
{
|
| 12703 |
+
"epoch": 0.05298743179401696,
|
| 12704 |
+
"grad_norm": 2.0893425941467285,
|
| 12705 |
+
"learning_rate": 2.9792650191236516e-05,
|
| 12706 |
+
"loss": 1.7794,
|
| 12707 |
+
"step": 27210
|
| 12708 |
+
},
|
| 12709 |
+
{
|
| 12710 |
+
"epoch": 0.05301664206512722,
|
| 12711 |
+
"grad_norm": 2.7562851905822754,
|
| 12712 |
+
"learning_rate": 2.979242204651216e-05,
|
| 12713 |
+
"loss": 1.7025,
|
| 12714 |
+
"step": 27225
|
| 12715 |
+
},
|
| 12716 |
+
{
|
| 12717 |
+
"epoch": 0.05304585233623749,
|
| 12718 |
+
"grad_norm": 4.634995937347412,
|
| 12719 |
+
"learning_rate": 2.9792193777218743e-05,
|
| 12720 |
+
"loss": 1.8236,
|
| 12721 |
+
"step": 27240
|
| 12722 |
+
},
|
| 12723 |
+
{
|
| 12724 |
+
"epoch": 0.05307506260734775,
|
| 12725 |
+
"grad_norm": 3.1855075359344482,
|
| 12726 |
+
"learning_rate": 2.9791965383358184e-05,
|
| 12727 |
+
"loss": 1.8439,
|
| 12728 |
+
"step": 27255
|
| 12729 |
+
},
|
| 12730 |
+
{
|
| 12731 |
+
"epoch": 0.05310427287845801,
|
| 12732 |
+
"grad_norm": 3.3286306858062744,
|
| 12733 |
+
"learning_rate": 2.9791736864932403e-05,
|
| 12734 |
+
"loss": 1.9314,
|
| 12735 |
+
"step": 27270
|
| 12736 |
+
},
|
| 12737 |
+
{
|
| 12738 |
+
"epoch": 0.05313348314956827,
|
| 12739 |
+
"grad_norm": 3.1028332710266113,
|
| 12740 |
+
"learning_rate": 2.979150822194332e-05,
|
| 12741 |
+
"loss": 2.0257,
|
| 12742 |
+
"step": 27285
|
| 12743 |
+
},
|
| 12744 |
+
{
|
| 12745 |
+
"epoch": 0.05316269342067854,
|
| 12746 |
+
"grad_norm": 4.1933393478393555,
|
| 12747 |
+
"learning_rate": 2.979127945439287e-05,
|
| 12748 |
+
"loss": 1.9908,
|
| 12749 |
+
"step": 27300
|
| 12750 |
+
},
|
| 12751 |
+
{
|
| 12752 |
+
"epoch": 0.0531919036917888,
|
| 12753 |
+
"grad_norm": 4.206679344177246,
|
| 12754 |
+
"learning_rate": 2.9791050562282974e-05,
|
| 12755 |
+
"loss": 1.8144,
|
| 12756 |
+
"step": 27315
|
| 12757 |
+
},
|
| 12758 |
+
{
|
| 12759 |
+
"epoch": 0.05322111396289906,
|
| 12760 |
+
"grad_norm": 2.231621265411377,
|
| 12761 |
+
"learning_rate": 2.9790821545615562e-05,
|
| 12762 |
+
"loss": 1.824,
|
| 12763 |
+
"step": 27330
|
| 12764 |
+
},
|
| 12765 |
+
{
|
| 12766 |
+
"epoch": 0.05325032423400932,
|
| 12767 |
+
"grad_norm": 3.625483512878418,
|
| 12768 |
+
"learning_rate": 2.9790592404392557e-05,
|
| 12769 |
+
"loss": 2.0087,
|
| 12770 |
+
"step": 27345
|
| 12771 |
+
},
|
| 12772 |
+
{
|
| 12773 |
+
"epoch": 0.05327953450511959,
|
| 12774 |
+
"grad_norm": 4.063029766082764,
|
| 12775 |
+
"learning_rate": 2.9790363138615902e-05,
|
| 12776 |
+
"loss": 1.8927,
|
| 12777 |
+
"step": 27360
|
| 12778 |
+
},
|
| 12779 |
+
{
|
| 12780 |
+
"epoch": 0.05330874477622985,
|
| 12781 |
+
"grad_norm": 4.166107654571533,
|
| 12782 |
+
"learning_rate": 2.979013374828751e-05,
|
| 12783 |
+
"loss": 1.9004,
|
| 12784 |
+
"step": 27375
|
| 12785 |
+
},
|
| 12786 |
+
{
|
| 12787 |
+
"epoch": 0.05333795504734011,
|
| 12788 |
+
"grad_norm": 2.737416982650757,
|
| 12789 |
+
"learning_rate": 2.9789904233409326e-05,
|
| 12790 |
+
"loss": 1.9678,
|
| 12791 |
+
"step": 27390
|
| 12792 |
+
},
|
| 12793 |
+
{
|
| 12794 |
+
"epoch": 0.05336716531845038,
|
| 12795 |
+
"grad_norm": 2.131272315979004,
|
| 12796 |
+
"learning_rate": 2.9789674593983277e-05,
|
| 12797 |
+
"loss": 1.895,
|
| 12798 |
+
"step": 27405
|
| 12799 |
+
},
|
| 12800 |
+
{
|
| 12801 |
+
"epoch": 0.05339637558956064,
|
| 12802 |
+
"grad_norm": 2.983872890472412,
|
| 12803 |
+
"learning_rate": 2.9789444830011302e-05,
|
| 12804 |
+
"loss": 1.7807,
|
| 12805 |
+
"step": 27420
|
| 12806 |
+
},
|
| 12807 |
+
{
|
| 12808 |
+
"epoch": 0.0534255858606709,
|
| 12809 |
+
"grad_norm": 5.785390377044678,
|
| 12810 |
+
"learning_rate": 2.978921494149533e-05,
|
| 12811 |
+
"loss": 1.9559,
|
| 12812 |
+
"step": 27435
|
| 12813 |
+
},
|
| 12814 |
+
{
|
| 12815 |
+
"epoch": 0.05345479613178116,
|
| 12816 |
+
"grad_norm": 2.7100813388824463,
|
| 12817 |
+
"learning_rate": 2.9788984928437298e-05,
|
| 12818 |
+
"loss": 1.9147,
|
| 12819 |
+
"step": 27450
|
| 12820 |
+
},
|
| 12821 |
+
{
|
| 12822 |
+
"epoch": 0.05348400640289143,
|
| 12823 |
+
"grad_norm": 2.0759334564208984,
|
| 12824 |
+
"learning_rate": 2.978875479083914e-05,
|
| 12825 |
+
"loss": 1.7345,
|
| 12826 |
+
"step": 27465
|
| 12827 |
+
},
|
| 12828 |
+
{
|
| 12829 |
+
"epoch": 0.05351321667400169,
|
| 12830 |
+
"grad_norm": 4.645893573760986,
|
| 12831 |
+
"learning_rate": 2.9788524528702804e-05,
|
| 12832 |
+
"loss": 1.7246,
|
| 12833 |
+
"step": 27480
|
| 12834 |
+
},
|
| 12835 |
+
{
|
| 12836 |
+
"epoch": 0.05354242694511195,
|
| 12837 |
+
"grad_norm": 3.1474320888519287,
|
| 12838 |
+
"learning_rate": 2.9788294142030225e-05,
|
| 12839 |
+
"loss": 1.8679,
|
| 12840 |
+
"step": 27495
|
| 12841 |
+
},
|
| 12842 |
+
{
|
| 12843 |
+
"epoch": 0.05357163721622222,
|
| 12844 |
+
"grad_norm": 2.8445558547973633,
|
| 12845 |
+
"learning_rate": 2.9788063630823335e-05,
|
| 12846 |
+
"loss": 1.748,
|
| 12847 |
+
"step": 27510
|
| 12848 |
+
},
|
| 12849 |
+
{
|
| 12850 |
+
"epoch": 0.05360084748733248,
|
| 12851 |
+
"grad_norm": 3.176330089569092,
|
| 12852 |
+
"learning_rate": 2.978783299508408e-05,
|
| 12853 |
+
"loss": 1.9698,
|
| 12854 |
+
"step": 27525
|
| 12855 |
+
},
|
| 12856 |
+
{
|
| 12857 |
+
"epoch": 0.05363005775844274,
|
| 12858 |
+
"grad_norm": 2.875300884246826,
|
| 12859 |
+
"learning_rate": 2.9787602234814407e-05,
|
| 12860 |
+
"loss": 1.7617,
|
| 12861 |
+
"step": 27540
|
| 12862 |
+
},
|
| 12863 |
+
{
|
| 12864 |
+
"epoch": 0.053659268029553,
|
| 12865 |
+
"grad_norm": 4.004800796508789,
|
| 12866 |
+
"learning_rate": 2.978737135001626e-05,
|
| 12867 |
+
"loss": 1.8204,
|
| 12868 |
+
"step": 27555
|
| 12869 |
+
},
|
| 12870 |
+
{
|
| 12871 |
+
"epoch": 0.05368847830066327,
|
| 12872 |
+
"grad_norm": 2.3588531017303467,
|
| 12873 |
+
"learning_rate": 2.9787140340691574e-05,
|
| 12874 |
+
"loss": 1.8028,
|
| 12875 |
+
"step": 27570
|
| 12876 |
+
},
|
| 12877 |
+
{
|
| 12878 |
+
"epoch": 0.05371768857177353,
|
| 12879 |
+
"grad_norm": 2.293210744857788,
|
| 12880 |
+
"learning_rate": 2.9786909206842297e-05,
|
| 12881 |
+
"loss": 1.9525,
|
| 12882 |
+
"step": 27585
|
| 12883 |
+
},
|
| 12884 |
+
{
|
| 12885 |
+
"epoch": 0.05374689884288379,
|
| 12886 |
+
"grad_norm": 3.4703333377838135,
|
| 12887 |
+
"learning_rate": 2.9786677948470382e-05,
|
| 12888 |
+
"loss": 1.8144,
|
| 12889 |
+
"step": 27600
|
| 12890 |
+
},
|
| 12891 |
+
{
|
| 12892 |
+
"epoch": 0.05377610911399406,
|
| 12893 |
+
"grad_norm": 2.1387853622436523,
|
| 12894 |
+
"learning_rate": 2.9786446565577772e-05,
|
| 12895 |
+
"loss": 1.7853,
|
| 12896 |
+
"step": 27615
|
| 12897 |
+
},
|
| 12898 |
+
{
|
| 12899 |
+
"epoch": 0.05380531938510432,
|
| 12900 |
+
"grad_norm": 1.9415435791015625,
|
| 12901 |
+
"learning_rate": 2.9786215058166417e-05,
|
| 12902 |
+
"loss": 1.8912,
|
| 12903 |
+
"step": 27630
|
| 12904 |
+
},
|
| 12905 |
+
{
|
| 12906 |
+
"epoch": 0.05383452965621458,
|
| 12907 |
+
"grad_norm": 3.315534830093384,
|
| 12908 |
+
"learning_rate": 2.978598342623826e-05,
|
| 12909 |
+
"loss": 1.7991,
|
| 12910 |
+
"step": 27645
|
| 12911 |
+
},
|
| 12912 |
+
{
|
| 12913 |
+
"epoch": 0.05386373992732484,
|
| 12914 |
+
"grad_norm": 3.2139084339141846,
|
| 12915 |
+
"learning_rate": 2.9785751669795265e-05,
|
| 12916 |
+
"loss": 1.8284,
|
| 12917 |
+
"step": 27660
|
| 12918 |
+
},
|
| 12919 |
+
{
|
| 12920 |
+
"epoch": 0.05389295019843511,
|
| 12921 |
+
"grad_norm": 2.0407660007476807,
|
| 12922 |
+
"learning_rate": 2.9785519788839368e-05,
|
| 12923 |
+
"loss": 1.8263,
|
| 12924 |
+
"step": 27675
|
| 12925 |
+
},
|
| 12926 |
+
{
|
| 12927 |
+
"epoch": 0.05392216046954537,
|
| 12928 |
+
"grad_norm": 3.6719486713409424,
|
| 12929 |
+
"learning_rate": 2.9785287783372538e-05,
|
| 12930 |
+
"loss": 1.7552,
|
| 12931 |
+
"step": 27690
|
| 12932 |
+
},
|
| 12933 |
+
{
|
| 12934 |
+
"epoch": 0.05395137074065563,
|
| 12935 |
+
"grad_norm": 2.3214800357818604,
|
| 12936 |
+
"learning_rate": 2.978505565339671e-05,
|
| 12937 |
+
"loss": 1.7752,
|
| 12938 |
+
"step": 27705
|
| 12939 |
+
},
|
| 12940 |
+
{
|
| 12941 |
+
"epoch": 0.0539805810117659,
|
| 12942 |
+
"grad_norm": 3.5290887355804443,
|
| 12943 |
+
"learning_rate": 2.9784823398913856e-05,
|
| 12944 |
+
"loss": 1.6684,
|
| 12945 |
+
"step": 27720
|
| 12946 |
+
},
|
| 12947 |
+
{
|
| 12948 |
+
"epoch": 0.05400979128287616,
|
| 12949 |
+
"grad_norm": 2.344564437866211,
|
| 12950 |
+
"learning_rate": 2.978459101992592e-05,
|
| 12951 |
+
"loss": 1.7649,
|
| 12952 |
+
"step": 27735
|
| 12953 |
+
},
|
| 12954 |
+
{
|
| 12955 |
+
"epoch": 0.05403900155398642,
|
| 12956 |
+
"grad_norm": 3.5880258083343506,
|
| 12957 |
+
"learning_rate": 2.9784358516434867e-05,
|
| 12958 |
+
"loss": 1.8194,
|
| 12959 |
+
"step": 27750
|
| 12960 |
+
},
|
| 12961 |
+
{
|
| 12962 |
+
"epoch": 0.05406821182509668,
|
| 12963 |
+
"grad_norm": 4.46945858001709,
|
| 12964 |
+
"learning_rate": 2.978412588844265e-05,
|
| 12965 |
+
"loss": 1.9552,
|
| 12966 |
+
"step": 27765
|
| 12967 |
+
},
|
| 12968 |
+
{
|
| 12969 |
+
"epoch": 0.05409742209620695,
|
| 12970 |
+
"grad_norm": 2.9063570499420166,
|
| 12971 |
+
"learning_rate": 2.978389313595123e-05,
|
| 12972 |
+
"loss": 1.9078,
|
| 12973 |
+
"step": 27780
|
| 12974 |
+
},
|
| 12975 |
+
{
|
| 12976 |
+
"epoch": 0.05412663236731721,
|
| 12977 |
+
"grad_norm": 4.740022659301758,
|
| 12978 |
+
"learning_rate": 2.9783660258962568e-05,
|
| 12979 |
+
"loss": 1.8773,
|
| 12980 |
+
"step": 27795
|
| 12981 |
+
},
|
| 12982 |
+
{
|
| 12983 |
+
"epoch": 0.05415584263842747,
|
| 12984 |
+
"grad_norm": 5.9590654373168945,
|
| 12985 |
+
"learning_rate": 2.9783427257478623e-05,
|
| 12986 |
+
"loss": 2.0891,
|
| 12987 |
+
"step": 27810
|
| 12988 |
+
},
|
| 12989 |
+
{
|
| 12990 |
+
"epoch": 0.05418505290953774,
|
| 12991 |
+
"grad_norm": 3.4905142784118652,
|
| 12992 |
+
"learning_rate": 2.978319413150136e-05,
|
| 12993 |
+
"loss": 1.912,
|
| 12994 |
+
"step": 27825
|
| 12995 |
+
},
|
| 12996 |
+
{
|
| 12997 |
+
"epoch": 0.054214263180648,
|
| 12998 |
+
"grad_norm": 4.312283039093018,
|
| 12999 |
+
"learning_rate": 2.978296088103273e-05,
|
| 13000 |
+
"loss": 1.7969,
|
| 13001 |
+
"step": 27840
|
| 13002 |
+
},
|
| 13003 |
+
{
|
| 13004 |
+
"epoch": 0.05424347345175826,
|
| 13005 |
+
"grad_norm": 3.419679880142212,
|
| 13006 |
+
"learning_rate": 2.978272750607472e-05,
|
| 13007 |
+
"loss": 2.0149,
|
| 13008 |
+
"step": 27855
|
| 13009 |
+
},
|
| 13010 |
+
{
|
| 13011 |
+
"epoch": 0.054272683722868524,
|
| 13012 |
+
"grad_norm": 1.4746593236923218,
|
| 13013 |
+
"learning_rate": 2.9782494006629275e-05,
|
| 13014 |
+
"loss": 1.7991,
|
| 13015 |
+
"step": 27870
|
| 13016 |
+
},
|
| 13017 |
+
{
|
| 13018 |
+
"epoch": 0.05430189399397879,
|
| 13019 |
+
"grad_norm": 2.6810593605041504,
|
| 13020 |
+
"learning_rate": 2.9782260382698374e-05,
|
| 13021 |
+
"loss": 1.8557,
|
| 13022 |
+
"step": 27885
|
| 13023 |
+
},
|
| 13024 |
+
{
|
| 13025 |
+
"epoch": 0.05433110426508905,
|
| 13026 |
+
"grad_norm": 4.144235134124756,
|
| 13027 |
+
"learning_rate": 2.9782026634283975e-05,
|
| 13028 |
+
"loss": 2.0213,
|
| 13029 |
+
"step": 27900
|
| 13030 |
+
},
|
| 13031 |
+
{
|
| 13032 |
+
"epoch": 0.054360314536199314,
|
| 13033 |
+
"grad_norm": 3.5831942558288574,
|
| 13034 |
+
"learning_rate": 2.9781792761388055e-05,
|
| 13035 |
+
"loss": 1.7606,
|
| 13036 |
+
"step": 27915
|
| 13037 |
+
},
|
| 13038 |
+
{
|
| 13039 |
+
"epoch": 0.054389524807309575,
|
| 13040 |
+
"grad_norm": 3.201408863067627,
|
| 13041 |
+
"learning_rate": 2.9781558764012573e-05,
|
| 13042 |
+
"loss": 1.7619,
|
| 13043 |
+
"step": 27930
|
| 13044 |
+
},
|
| 13045 |
+
{
|
| 13046 |
+
"epoch": 0.05441873507841984,
|
| 13047 |
+
"grad_norm": 2.8114073276519775,
|
| 13048 |
+
"learning_rate": 2.978132464215951e-05,
|
| 13049 |
+
"loss": 1.9014,
|
| 13050 |
+
"step": 27945
|
| 13051 |
+
},
|
| 13052 |
+
{
|
| 13053 |
+
"epoch": 0.054447945349530104,
|
| 13054 |
+
"grad_norm": 2.400465726852417,
|
| 13055 |
+
"learning_rate": 2.9781090395830834e-05,
|
| 13056 |
+
"loss": 1.9041,
|
| 13057 |
+
"step": 27960
|
| 13058 |
+
},
|
| 13059 |
+
{
|
| 13060 |
+
"epoch": 0.054477155620640365,
|
| 13061 |
+
"grad_norm": 2.1496269702911377,
|
| 13062 |
+
"learning_rate": 2.9780856025028513e-05,
|
| 13063 |
+
"loss": 1.7252,
|
| 13064 |
+
"step": 27975
|
| 13065 |
+
},
|
| 13066 |
+
{
|
| 13067 |
+
"epoch": 0.05450636589175063,
|
| 13068 |
+
"grad_norm": 3.477867603302002,
|
| 13069 |
+
"learning_rate": 2.978062152975453e-05,
|
| 13070 |
+
"loss": 1.7466,
|
| 13071 |
+
"step": 27990
|
| 13072 |
+
},
|
| 13073 |
+
{
|
| 13074 |
+
"epoch": 0.054535576162860894,
|
| 13075 |
+
"grad_norm": 2.1709723472595215,
|
| 13076 |
+
"learning_rate": 2.978038691001085e-05,
|
| 13077 |
+
"loss": 1.6808,
|
| 13078 |
+
"step": 28005
|
| 13079 |
+
},
|
| 13080 |
+
{
|
| 13081 |
+
"epoch": 0.054564786433971155,
|
| 13082 |
+
"grad_norm": 2.13655686378479,
|
| 13083 |
+
"learning_rate": 2.978015216579945e-05,
|
| 13084 |
+
"loss": 1.8089,
|
| 13085 |
+
"step": 28020
|
| 13086 |
+
},
|
| 13087 |
+
{
|
| 13088 |
+
"epoch": 0.054593996705081416,
|
| 13089 |
+
"grad_norm": 3.563598394393921,
|
| 13090 |
+
"learning_rate": 2.9779917297122318e-05,
|
| 13091 |
+
"loss": 1.8368,
|
| 13092 |
+
"step": 28035
|
| 13093 |
+
},
|
| 13094 |
+
{
|
| 13095 |
+
"epoch": 0.054623206976191684,
|
| 13096 |
+
"grad_norm": 3.1836514472961426,
|
| 13097 |
+
"learning_rate": 2.977968230398142e-05,
|
| 13098 |
+
"loss": 1.8042,
|
| 13099 |
+
"step": 28050
|
| 13100 |
+
},
|
| 13101 |
+
{
|
| 13102 |
+
"epoch": 0.054652417247301945,
|
| 13103 |
+
"grad_norm": 2.4931259155273438,
|
| 13104 |
+
"learning_rate": 2.9779447186378738e-05,
|
| 13105 |
+
"loss": 1.8061,
|
| 13106 |
+
"step": 28065
|
| 13107 |
+
},
|
| 13108 |
+
{
|
| 13109 |
+
"epoch": 0.054681627518412206,
|
| 13110 |
+
"grad_norm": 3.2399775981903076,
|
| 13111 |
+
"learning_rate": 2.977921194431625e-05,
|
| 13112 |
+
"loss": 1.8723,
|
| 13113 |
+
"step": 28080
|
| 13114 |
+
},
|
| 13115 |
+
{
|
| 13116 |
+
"epoch": 0.05471083778952247,
|
| 13117 |
+
"grad_norm": 3.281590461730957,
|
| 13118 |
+
"learning_rate": 2.977897657779594e-05,
|
| 13119 |
+
"loss": 1.8882,
|
| 13120 |
+
"step": 28095
|
| 13121 |
+
},
|
| 13122 |
+
{
|
| 13123 |
+
"epoch": 0.054740048060632734,
|
| 13124 |
+
"grad_norm": 2.607039451599121,
|
| 13125 |
+
"learning_rate": 2.9778741086819795e-05,
|
| 13126 |
+
"loss": 1.8796,
|
| 13127 |
+
"step": 28110
|
| 13128 |
+
},
|
| 13129 |
+
{
|
| 13130 |
+
"epoch": 0.054769258331742995,
|
| 13131 |
+
"grad_norm": 3.4543254375457764,
|
| 13132 |
+
"learning_rate": 2.977850547138979e-05,
|
| 13133 |
+
"loss": 1.8826,
|
| 13134 |
+
"step": 28125
|
| 13135 |
+
},
|
| 13136 |
+
{
|
| 13137 |
+
"epoch": 0.054798468602853256,
|
| 13138 |
+
"grad_norm": 3.8168885707855225,
|
| 13139 |
+
"learning_rate": 2.9778269731507914e-05,
|
| 13140 |
+
"loss": 1.9827,
|
| 13141 |
+
"step": 28140
|
| 13142 |
+
},
|
| 13143 |
+
{
|
| 13144 |
+
"epoch": 0.054827678873963524,
|
| 13145 |
+
"grad_norm": 2.7308406829833984,
|
| 13146 |
+
"learning_rate": 2.977803386717615e-05,
|
| 13147 |
+
"loss": 1.768,
|
| 13148 |
+
"step": 28155
|
| 13149 |
+
},
|
| 13150 |
+
{
|
| 13151 |
+
"epoch": 0.054856889145073785,
|
| 13152 |
+
"grad_norm": 4.276648998260498,
|
| 13153 |
+
"learning_rate": 2.9777797878396477e-05,
|
| 13154 |
+
"loss": 1.8696,
|
| 13155 |
+
"step": 28170
|
| 13156 |
+
},
|
| 13157 |
+
{
|
| 13158 |
+
"epoch": 0.054886099416184046,
|
| 13159 |
+
"grad_norm": 3.1874961853027344,
|
| 13160 |
+
"learning_rate": 2.97775617651709e-05,
|
| 13161 |
+
"loss": 1.881,
|
| 13162 |
+
"step": 28185
|
| 13163 |
+
},
|
| 13164 |
+
{
|
| 13165 |
+
"epoch": 0.054915309687294314,
|
| 13166 |
+
"grad_norm": 2.4808239936828613,
|
| 13167 |
+
"learning_rate": 2.977732552750139e-05,
|
| 13168 |
+
"loss": 1.7825,
|
| 13169 |
+
"step": 28200
|
| 13170 |
+
},
|
| 13171 |
+
{
|
| 13172 |
+
"epoch": 0.054944519958404575,
|
| 13173 |
+
"grad_norm": 3.0970373153686523,
|
| 13174 |
+
"learning_rate": 2.9777089165389942e-05,
|
| 13175 |
+
"loss": 1.7825,
|
| 13176 |
+
"step": 28215
|
| 13177 |
+
},
|
| 13178 |
+
{
|
| 13179 |
+
"epoch": 0.054973730229514836,
|
| 13180 |
+
"grad_norm": 1.757534384727478,
|
| 13181 |
+
"learning_rate": 2.9776852678838555e-05,
|
| 13182 |
+
"loss": 2.0145,
|
| 13183 |
+
"step": 28230
|
| 13184 |
+
},
|
| 13185 |
+
{
|
| 13186 |
+
"epoch": 0.0550029405006251,
|
| 13187 |
+
"grad_norm": 2.351555347442627,
|
| 13188 |
+
"learning_rate": 2.977661606784921e-05,
|
| 13189 |
+
"loss": 2.0268,
|
| 13190 |
+
"step": 28245
|
| 13191 |
+
},
|
| 13192 |
+
{
|
| 13193 |
+
"epoch": 0.055032150771735365,
|
| 13194 |
+
"grad_norm": 2.4561007022857666,
|
| 13195 |
+
"learning_rate": 2.9776379332423902e-05,
|
| 13196 |
+
"loss": 1.7745,
|
| 13197 |
+
"step": 28260
|
| 13198 |
+
},
|
| 13199 |
+
{
|
| 13200 |
+
"epoch": 0.055061361042845626,
|
| 13201 |
+
"grad_norm": 3.8144009113311768,
|
| 13202 |
+
"learning_rate": 2.9776142472564624e-05,
|
| 13203 |
+
"loss": 1.8408,
|
| 13204 |
+
"step": 28275
|
| 13205 |
+
},
|
| 13206 |
+
{
|
| 13207 |
+
"epoch": 0.05509057131395589,
|
| 13208 |
+
"grad_norm": 4.969006061553955,
|
| 13209 |
+
"learning_rate": 2.9775905488273373e-05,
|
| 13210 |
+
"loss": 1.9223,
|
| 13211 |
+
"step": 28290
|
| 13212 |
+
},
|
| 13213 |
+
{
|
| 13214 |
+
"epoch": 0.055119781585066155,
|
| 13215 |
+
"grad_norm": 5.71866512298584,
|
| 13216 |
+
"learning_rate": 2.9775668379552146e-05,
|
| 13217 |
+
"loss": 1.842,
|
| 13218 |
+
"step": 28305
|
| 13219 |
+
},
|
| 13220 |
+
{
|
| 13221 |
+
"epoch": 0.055148991856176416,
|
| 13222 |
+
"grad_norm": 1.9934380054473877,
|
| 13223 |
+
"learning_rate": 2.9775431146402937e-05,
|
| 13224 |
+
"loss": 1.976,
|
| 13225 |
+
"step": 28320
|
| 13226 |
+
},
|
| 13227 |
+
{
|
| 13228 |
+
"epoch": 0.05517820212728668,
|
| 13229 |
+
"grad_norm": 1.7895939350128174,
|
| 13230 |
+
"learning_rate": 2.9775193788827743e-05,
|
| 13231 |
+
"loss": 2.0921,
|
| 13232 |
+
"step": 28335
|
| 13233 |
+
},
|
| 13234 |
+
{
|
| 13235 |
+
"epoch": 0.05520741239839694,
|
| 13236 |
+
"grad_norm": 4.20900821685791,
|
| 13237 |
+
"learning_rate": 2.9774956306828566e-05,
|
| 13238 |
+
"loss": 1.9333,
|
| 13239 |
+
"step": 28350
|
| 13240 |
+
},
|
| 13241 |
+
{
|
| 13242 |
+
"epoch": 0.055236622669507206,
|
| 13243 |
+
"grad_norm": 1.7822163105010986,
|
| 13244 |
+
"learning_rate": 2.97747187004074e-05,
|
| 13245 |
+
"loss": 2.0872,
|
| 13246 |
+
"step": 28365
|
| 13247 |
+
},
|
| 13248 |
+
{
|
| 13249 |
+
"epoch": 0.05526583294061747,
|
| 13250 |
+
"grad_norm": 1.7469080686569214,
|
| 13251 |
+
"learning_rate": 2.9774480969566254e-05,
|
| 13252 |
+
"loss": 1.8781,
|
| 13253 |
+
"step": 28380
|
| 13254 |
+
},
|
| 13255 |
+
{
|
| 13256 |
+
"epoch": 0.05529504321172773,
|
| 13257 |
+
"grad_norm": 3.0533454418182373,
|
| 13258 |
+
"learning_rate": 2.977424311430712e-05,
|
| 13259 |
+
"loss": 1.7184,
|
| 13260 |
+
"step": 28395
|
| 13261 |
+
},
|
| 13262 |
+
{
|
| 13263 |
+
"epoch": 0.05532425348283799,
|
| 13264 |
+
"grad_norm": 2.729780912399292,
|
| 13265 |
+
"learning_rate": 2.977400513463201e-05,
|
| 13266 |
+
"loss": 1.7794,
|
| 13267 |
+
"step": 28410
|
| 13268 |
+
},
|
| 13269 |
+
{
|
| 13270 |
+
"epoch": 0.05535346375394826,
|
| 13271 |
+
"grad_norm": 3.2360620498657227,
|
| 13272 |
+
"learning_rate": 2.9773767030542926e-05,
|
| 13273 |
+
"loss": 1.7711,
|
| 13274 |
+
"step": 28425
|
| 13275 |
+
},
|
| 13276 |
+
{
|
| 13277 |
+
"epoch": 0.05538267402505852,
|
| 13278 |
+
"grad_norm": 2.3599355220794678,
|
| 13279 |
+
"learning_rate": 2.9773528802041873e-05,
|
| 13280 |
+
"loss": 1.8758,
|
| 13281 |
+
"step": 28440
|
| 13282 |
+
},
|
| 13283 |
+
{
|
| 13284 |
+
"epoch": 0.05541188429616878,
|
| 13285 |
+
"grad_norm": 3.3827106952667236,
|
| 13286 |
+
"learning_rate": 2.9773290449130856e-05,
|
| 13287 |
+
"loss": 1.901,
|
| 13288 |
+
"step": 28455
|
| 13289 |
+
},
|
| 13290 |
+
{
|
| 13291 |
+
"epoch": 0.055441094567279046,
|
| 13292 |
+
"grad_norm": 2.9506921768188477,
|
| 13293 |
+
"learning_rate": 2.977305197181188e-05,
|
| 13294 |
+
"loss": 1.9286,
|
| 13295 |
+
"step": 28470
|
| 13296 |
+
},
|
| 13297 |
+
{
|
| 13298 |
+
"epoch": 0.05547030483838931,
|
| 13299 |
+
"grad_norm": 3.381622791290283,
|
| 13300 |
+
"learning_rate": 2.9772813370086956e-05,
|
| 13301 |
+
"loss": 2.0169,
|
| 13302 |
+
"step": 28485
|
| 13303 |
+
},
|
| 13304 |
+
{
|
| 13305 |
+
"epoch": 0.05549951510949957,
|
| 13306 |
+
"grad_norm": 3.6618142127990723,
|
| 13307 |
+
"learning_rate": 2.9772574643958095e-05,
|
| 13308 |
+
"loss": 1.8318,
|
| 13309 |
+
"step": 28500
|
| 13310 |
+
},
|
| 13311 |
+
{
|
| 13312 |
+
"epoch": 0.05552872538060983,
|
| 13313 |
+
"grad_norm": 2.906064033508301,
|
| 13314 |
+
"learning_rate": 2.9772335793427304e-05,
|
| 13315 |
+
"loss": 1.8778,
|
| 13316 |
+
"step": 28515
|
| 13317 |
+
},
|
| 13318 |
+
{
|
| 13319 |
+
"epoch": 0.0555579356517201,
|
| 13320 |
+
"grad_norm": 1.9280356168746948,
|
| 13321 |
+
"learning_rate": 2.9772096818496592e-05,
|
| 13322 |
+
"loss": 1.9095,
|
| 13323 |
+
"step": 28530
|
| 13324 |
+
},
|
| 13325 |
+
{
|
| 13326 |
+
"epoch": 0.05558714592283036,
|
| 13327 |
+
"grad_norm": 2.451441764831543,
|
| 13328 |
+
"learning_rate": 2.977185771916798e-05,
|
| 13329 |
+
"loss": 1.7975,
|
| 13330 |
+
"step": 28545
|
| 13331 |
+
},
|
| 13332 |
+
{
|
| 13333 |
+
"epoch": 0.05561635619394062,
|
| 13334 |
+
"grad_norm": 3.4363293647766113,
|
| 13335 |
+
"learning_rate": 2.9771618495443473e-05,
|
| 13336 |
+
"loss": 1.9995,
|
| 13337 |
+
"step": 28560
|
| 13338 |
+
},
|
| 13339 |
+
{
|
| 13340 |
+
"epoch": 0.05564556646505089,
|
| 13341 |
+
"grad_norm": 3.402430772781372,
|
| 13342 |
+
"learning_rate": 2.9771379147325095e-05,
|
| 13343 |
+
"loss": 1.879,
|
| 13344 |
+
"step": 28575
|
| 13345 |
+
},
|
| 13346 |
+
{
|
| 13347 |
+
"epoch": 0.05567477673616115,
|
| 13348 |
+
"grad_norm": 3.7891762256622314,
|
| 13349 |
+
"learning_rate": 2.977113967481485e-05,
|
| 13350 |
+
"loss": 1.8275,
|
| 13351 |
+
"step": 28590
|
| 13352 |
+
},
|
| 13353 |
+
{
|
| 13354 |
+
"epoch": 0.05570398700727141,
|
| 13355 |
+
"grad_norm": 2.45809268951416,
|
| 13356 |
+
"learning_rate": 2.977090007791476e-05,
|
| 13357 |
+
"loss": 1.8131,
|
| 13358 |
+
"step": 28605
|
| 13359 |
+
},
|
| 13360 |
+
{
|
| 13361 |
+
"epoch": 0.05573319727838167,
|
| 13362 |
+
"grad_norm": 3.5447278022766113,
|
| 13363 |
+
"learning_rate": 2.9770660356626848e-05,
|
| 13364 |
+
"loss": 1.8373,
|
| 13365 |
+
"step": 28620
|
| 13366 |
+
},
|
| 13367 |
+
{
|
| 13368 |
+
"epoch": 0.05576240754949194,
|
| 13369 |
+
"grad_norm": 3.397735595703125,
|
| 13370 |
+
"learning_rate": 2.9770420510953124e-05,
|
| 13371 |
+
"loss": 1.7907,
|
| 13372 |
+
"step": 28635
|
| 13373 |
+
},
|
| 13374 |
+
{
|
| 13375 |
+
"epoch": 0.0557916178206022,
|
| 13376 |
+
"grad_norm": 2.185011386871338,
|
| 13377 |
+
"learning_rate": 2.9770180540895613e-05,
|
| 13378 |
+
"loss": 1.8909,
|
| 13379 |
+
"step": 28650
|
| 13380 |
+
},
|
| 13381 |
+
{
|
| 13382 |
+
"epoch": 0.05582082809171246,
|
| 13383 |
+
"grad_norm": 3.66780161857605,
|
| 13384 |
+
"learning_rate": 2.9769940446456332e-05,
|
| 13385 |
+
"loss": 1.8898,
|
| 13386 |
+
"step": 28665
|
| 13387 |
+
},
|
| 13388 |
+
{
|
| 13389 |
+
"epoch": 0.05585003836282273,
|
| 13390 |
+
"grad_norm": 3.8541507720947266,
|
| 13391 |
+
"learning_rate": 2.9769700227637307e-05,
|
| 13392 |
+
"loss": 1.8156,
|
| 13393 |
+
"step": 28680
|
| 13394 |
+
},
|
| 13395 |
+
{
|
| 13396 |
+
"epoch": 0.05587924863393299,
|
| 13397 |
+
"grad_norm": 2.504997491836548,
|
| 13398 |
+
"learning_rate": 2.9769459884440563e-05,
|
| 13399 |
+
"loss": 1.8584,
|
| 13400 |
+
"step": 28695
|
| 13401 |
+
},
|
| 13402 |
+
{
|
| 13403 |
+
"epoch": 0.05590845890504325,
|
| 13404 |
+
"grad_norm": 2.811286687850952,
|
| 13405 |
+
"learning_rate": 2.9769219416868114e-05,
|
| 13406 |
+
"loss": 1.8762,
|
| 13407 |
+
"step": 28710
|
| 13408 |
+
},
|
| 13409 |
+
{
|
| 13410 |
+
"epoch": 0.05593766917615351,
|
| 13411 |
+
"grad_norm": 3.2774808406829834,
|
| 13412 |
+
"learning_rate": 2.976897882492199e-05,
|
| 13413 |
+
"loss": 1.8637,
|
| 13414 |
+
"step": 28725
|
| 13415 |
+
},
|
| 13416 |
+
{
|
| 13417 |
+
"epoch": 0.05596687944726378,
|
| 13418 |
+
"grad_norm": 3.6415061950683594,
|
| 13419 |
+
"learning_rate": 2.9768738108604222e-05,
|
| 13420 |
+
"loss": 2.033,
|
| 13421 |
+
"step": 28740
|
| 13422 |
+
},
|
| 13423 |
+
{
|
| 13424 |
+
"epoch": 0.05599608971837404,
|
| 13425 |
+
"grad_norm": 2.5112550258636475,
|
| 13426 |
+
"learning_rate": 2.9768497267916833e-05,
|
| 13427 |
+
"loss": 1.9643,
|
| 13428 |
+
"step": 28755
|
| 13429 |
+
},
|
| 13430 |
+
{
|
| 13431 |
+
"epoch": 0.0560252999894843,
|
| 13432 |
+
"grad_norm": 3.3538918495178223,
|
| 13433 |
+
"learning_rate": 2.9768256302861852e-05,
|
| 13434 |
+
"loss": 1.7964,
|
| 13435 |
+
"step": 28770
|
| 13436 |
+
},
|
| 13437 |
+
{
|
| 13438 |
+
"epoch": 0.05605451026059457,
|
| 13439 |
+
"grad_norm": 1.7511709928512573,
|
| 13440 |
+
"learning_rate": 2.9768015213441306e-05,
|
| 13441 |
+
"loss": 1.9891,
|
| 13442 |
+
"step": 28785
|
| 13443 |
+
},
|
| 13444 |
+
{
|
| 13445 |
+
"epoch": 0.05608372053170483,
|
| 13446 |
+
"grad_norm": 4.007176399230957,
|
| 13447 |
+
"learning_rate": 2.9767773999657225e-05,
|
| 13448 |
+
"loss": 2.0263,
|
| 13449 |
+
"step": 28800
|
| 13450 |
+
},
|
| 13451 |
+
{
|
| 13452 |
+
"epoch": 0.05611293080281509,
|
| 13453 |
+
"grad_norm": 2.968200206756592,
|
| 13454 |
+
"learning_rate": 2.9767532661511644e-05,
|
| 13455 |
+
"loss": 1.7171,
|
| 13456 |
+
"step": 28815
|
| 13457 |
+
},
|
| 13458 |
+
{
|
| 13459 |
+
"epoch": 0.05614214107392535,
|
| 13460 |
+
"grad_norm": 3.2194066047668457,
|
| 13461 |
+
"learning_rate": 2.9767291199006594e-05,
|
| 13462 |
+
"loss": 1.8825,
|
| 13463 |
+
"step": 28830
|
| 13464 |
+
},
|
| 13465 |
+
{
|
| 13466 |
+
"epoch": 0.05617135134503562,
|
| 13467 |
+
"grad_norm": 3.994147539138794,
|
| 13468 |
+
"learning_rate": 2.976704961214411e-05,
|
| 13469 |
+
"loss": 1.6859,
|
| 13470 |
+
"step": 28845
|
| 13471 |
+
},
|
| 13472 |
+
{
|
| 13473 |
+
"epoch": 0.05620056161614588,
|
| 13474 |
+
"grad_norm": 2.1359071731567383,
|
| 13475 |
+
"learning_rate": 2.976680790092622e-05,
|
| 13476 |
+
"loss": 1.7796,
|
| 13477 |
+
"step": 28860
|
| 13478 |
+
},
|
| 13479 |
+
{
|
| 13480 |
+
"epoch": 0.05622977188725614,
|
| 13481 |
+
"grad_norm": 3.074885606765747,
|
| 13482 |
+
"learning_rate": 2.976656606535497e-05,
|
| 13483 |
+
"loss": 1.9507,
|
| 13484 |
+
"step": 28875
|
| 13485 |
+
},
|
| 13486 |
+
{
|
| 13487 |
+
"epoch": 0.05625898215836641,
|
| 13488 |
+
"grad_norm": 3.184913396835327,
|
| 13489 |
+
"learning_rate": 2.9766324105432385e-05,
|
| 13490 |
+
"loss": 1.9536,
|
| 13491 |
+
"step": 28890
|
| 13492 |
+
},
|
| 13493 |
+
{
|
| 13494 |
+
"epoch": 0.05628819242947667,
|
| 13495 |
+
"grad_norm": 3.911243438720703,
|
| 13496 |
+
"learning_rate": 2.976608202116051e-05,
|
| 13497 |
+
"loss": 1.8609,
|
| 13498 |
+
"step": 28905
|
| 13499 |
+
},
|
| 13500 |
+
{
|
| 13501 |
+
"epoch": 0.05631740270058693,
|
| 13502 |
+
"grad_norm": 4.364305019378662,
|
| 13503 |
+
"learning_rate": 2.9765839812541378e-05,
|
| 13504 |
+
"loss": 1.8573,
|
| 13505 |
+
"step": 28920
|
| 13506 |
+
},
|
| 13507 |
+
{
|
| 13508 |
+
"epoch": 0.05634661297169719,
|
| 13509 |
+
"grad_norm": 3.0088484287261963,
|
| 13510 |
+
"learning_rate": 2.9765597479577034e-05,
|
| 13511 |
+
"loss": 2.0879,
|
| 13512 |
+
"step": 28935
|
| 13513 |
+
},
|
| 13514 |
+
{
|
| 13515 |
+
"epoch": 0.05637582324280746,
|
| 13516 |
+
"grad_norm": 1.9867300987243652,
|
| 13517 |
+
"learning_rate": 2.9765355022269518e-05,
|
| 13518 |
+
"loss": 1.7799,
|
| 13519 |
+
"step": 28950
|
| 13520 |
+
},
|
| 13521 |
+
{
|
| 13522 |
+
"epoch": 0.05640503351391772,
|
| 13523 |
+
"grad_norm": 2.1443519592285156,
|
| 13524 |
+
"learning_rate": 2.9765112440620874e-05,
|
| 13525 |
+
"loss": 1.7201,
|
| 13526 |
+
"step": 28965
|
| 13527 |
+
},
|
| 13528 |
+
{
|
| 13529 |
+
"epoch": 0.05643424378502798,
|
| 13530 |
+
"grad_norm": 2.226407051086426,
|
| 13531 |
+
"learning_rate": 2.9764869734633134e-05,
|
| 13532 |
+
"loss": 1.7974,
|
| 13533 |
+
"step": 28980
|
| 13534 |
+
},
|
| 13535 |
+
{
|
| 13536 |
+
"epoch": 0.05646345405613824,
|
| 13537 |
+
"grad_norm": 4.08579158782959,
|
| 13538 |
+
"learning_rate": 2.9764626904308354e-05,
|
| 13539 |
+
"loss": 1.8633,
|
| 13540 |
+
"step": 28995
|
| 13541 |
+
},
|
| 13542 |
+
{
|
| 13543 |
+
"epoch": 0.05649266432724851,
|
| 13544 |
+
"grad_norm": 4.314965724945068,
|
| 13545 |
+
"learning_rate": 2.9764383949648576e-05,
|
| 13546 |
+
"loss": 1.9242,
|
| 13547 |
+
"step": 29010
|
| 13548 |
+
},
|
| 13549 |
+
{
|
| 13550 |
+
"epoch": 0.05652187459835877,
|
| 13551 |
+
"grad_norm": 4.151242733001709,
|
| 13552 |
+
"learning_rate": 2.976414087065584e-05,
|
| 13553 |
+
"loss": 1.942,
|
| 13554 |
+
"step": 29025
|
| 13555 |
+
},
|
| 13556 |
+
{
|
| 13557 |
+
"epoch": 0.05655108486946903,
|
| 13558 |
+
"grad_norm": 1.9677321910858154,
|
| 13559 |
+
"learning_rate": 2.97638976673322e-05,
|
| 13560 |
+
"loss": 1.8371,
|
| 13561 |
+
"step": 29040
|
| 13562 |
+
},
|
| 13563 |
+
{
|
| 13564 |
+
"epoch": 0.0565802951405793,
|
| 13565 |
+
"grad_norm": 2.814545154571533,
|
| 13566 |
+
"learning_rate": 2.97636543396797e-05,
|
| 13567 |
+
"loss": 1.9929,
|
| 13568 |
+
"step": 29055
|
| 13569 |
+
},
|
| 13570 |
+
{
|
| 13571 |
+
"epoch": 0.05660950541168956,
|
| 13572 |
+
"grad_norm": 3.4638845920562744,
|
| 13573 |
+
"learning_rate": 2.976341088770039e-05,
|
| 13574 |
+
"loss": 1.8635,
|
| 13575 |
+
"step": 29070
|
| 13576 |
+
},
|
| 13577 |
+
{
|
| 13578 |
+
"epoch": 0.05663871568279982,
|
| 13579 |
+
"grad_norm": 2.0748291015625,
|
| 13580 |
+
"learning_rate": 2.976316731139632e-05,
|
| 13581 |
+
"loss": 1.6827,
|
| 13582 |
+
"step": 29085
|
| 13583 |
+
},
|
| 13584 |
+
{
|
| 13585 |
+
"epoch": 0.056667925953910084,
|
| 13586 |
+
"grad_norm": 2.4861667156219482,
|
| 13587 |
+
"learning_rate": 2.9762923610769545e-05,
|
| 13588 |
+
"loss": 1.8022,
|
| 13589 |
+
"step": 29100
|
| 13590 |
+
},
|
| 13591 |
+
{
|
| 13592 |
+
"epoch": 0.05669713622502035,
|
| 13593 |
+
"grad_norm": 2.12156343460083,
|
| 13594 |
+
"learning_rate": 2.9762679785822113e-05,
|
| 13595 |
+
"loss": 1.8912,
|
| 13596 |
+
"step": 29115
|
| 13597 |
+
},
|
| 13598 |
+
{
|
| 13599 |
+
"epoch": 0.05672634649613061,
|
| 13600 |
+
"grad_norm": 2.3051788806915283,
|
| 13601 |
+
"learning_rate": 2.9762435836556075e-05,
|
| 13602 |
+
"loss": 1.786,
|
| 13603 |
+
"step": 29130
|
| 13604 |
+
},
|
| 13605 |
+
{
|
| 13606 |
+
"epoch": 0.056755556767240874,
|
| 13607 |
+
"grad_norm": 3.0045323371887207,
|
| 13608 |
+
"learning_rate": 2.9762191762973492e-05,
|
| 13609 |
+
"loss": 1.9105,
|
| 13610 |
+
"step": 29145
|
| 13611 |
+
},
|
| 13612 |
+
{
|
| 13613 |
+
"epoch": 0.05678476703835114,
|
| 13614 |
+
"grad_norm": 2.9411308765411377,
|
| 13615 |
+
"learning_rate": 2.9761947565076413e-05,
|
| 13616 |
+
"loss": 1.7949,
|
| 13617 |
+
"step": 29160
|
| 13618 |
+
},
|
| 13619 |
+
{
|
| 13620 |
+
"epoch": 0.0568139773094614,
|
| 13621 |
+
"grad_norm": 3.0440449714660645,
|
| 13622 |
+
"learning_rate": 2.97617032428669e-05,
|
| 13623 |
+
"loss": 2.0024,
|
| 13624 |
+
"step": 29175
|
| 13625 |
+
},
|
| 13626 |
+
{
|
| 13627 |
+
"epoch": 0.056843187580571664,
|
| 13628 |
+
"grad_norm": 3.836582660675049,
|
| 13629 |
+
"learning_rate": 2.976145879634701e-05,
|
| 13630 |
+
"loss": 1.9192,
|
| 13631 |
+
"step": 29190
|
| 13632 |
+
},
|
| 13633 |
+
{
|
| 13634 |
+
"epoch": 0.056872397851681925,
|
| 13635 |
+
"grad_norm": 3.63226580619812,
|
| 13636 |
+
"learning_rate": 2.9761214225518792e-05,
|
| 13637 |
+
"loss": 1.6728,
|
| 13638 |
+
"step": 29205
|
| 13639 |
+
},
|
| 13640 |
+
{
|
| 13641 |
+
"epoch": 0.05690160812279219,
|
| 13642 |
+
"grad_norm": 2.588270902633667,
|
| 13643 |
+
"learning_rate": 2.9760969530384317e-05,
|
| 13644 |
+
"loss": 1.8828,
|
| 13645 |
+
"step": 29220
|
| 13646 |
+
},
|
| 13647 |
+
{
|
| 13648 |
+
"epoch": 0.05693081839390245,
|
| 13649 |
+
"grad_norm": 3.4929933547973633,
|
| 13650 |
+
"learning_rate": 2.9760724710945642e-05,
|
| 13651 |
+
"loss": 2.0755,
|
| 13652 |
+
"step": 29235
|
| 13653 |
+
},
|
| 13654 |
+
{
|
| 13655 |
+
"epoch": 0.056960028665012714,
|
| 13656 |
+
"grad_norm": 2.916142463684082,
|
| 13657 |
+
"learning_rate": 2.976047976720483e-05,
|
| 13658 |
+
"loss": 1.8955,
|
| 13659 |
+
"step": 29250
|
| 13660 |
+
},
|
| 13661 |
+
{
|
| 13662 |
+
"epoch": 0.05698923893612298,
|
| 13663 |
+
"grad_norm": 3.606926918029785,
|
| 13664 |
+
"learning_rate": 2.976023469916394e-05,
|
| 13665 |
+
"loss": 1.9196,
|
| 13666 |
+
"step": 29265
|
| 13667 |
+
},
|
| 13668 |
+
{
|
| 13669 |
+
"epoch": 0.05701844920723324,
|
| 13670 |
+
"grad_norm": 4.033385276794434,
|
| 13671 |
+
"learning_rate": 2.9759989506825033e-05,
|
| 13672 |
+
"loss": 1.9555,
|
| 13673 |
+
"step": 29280
|
| 13674 |
+
},
|
| 13675 |
+
{
|
| 13676 |
+
"epoch": 0.057047659478343504,
|
| 13677 |
+
"grad_norm": 2.1779022216796875,
|
| 13678 |
+
"learning_rate": 2.9759744190190185e-05,
|
| 13679 |
+
"loss": 1.7735,
|
| 13680 |
+
"step": 29295
|
| 13681 |
+
},
|
| 13682 |
+
{
|
| 13683 |
+
"epoch": 0.057076869749453765,
|
| 13684 |
+
"grad_norm": 4.592677116394043,
|
| 13685 |
+
"learning_rate": 2.9759498749261452e-05,
|
| 13686 |
+
"loss": 1.8971,
|
| 13687 |
+
"step": 29310
|
| 13688 |
+
},
|
| 13689 |
+
{
|
| 13690 |
+
"epoch": 0.05710608002056403,
|
| 13691 |
+
"grad_norm": 4.797886371612549,
|
| 13692 |
+
"learning_rate": 2.9759253184040906e-05,
|
| 13693 |
+
"loss": 1.8494,
|
| 13694 |
+
"step": 29325
|
| 13695 |
+
},
|
| 13696 |
+
{
|
| 13697 |
+
"epoch": 0.057135290291674294,
|
| 13698 |
+
"grad_norm": 2.316049337387085,
|
| 13699 |
+
"learning_rate": 2.9759007494530615e-05,
|
| 13700 |
+
"loss": 1.7288,
|
| 13701 |
+
"step": 29340
|
| 13702 |
+
},
|
| 13703 |
+
{
|
| 13704 |
+
"epoch": 0.057164500562784555,
|
| 13705 |
+
"grad_norm": 2.9396653175354004,
|
| 13706 |
+
"learning_rate": 2.975876168073264e-05,
|
| 13707 |
+
"loss": 1.8751,
|
| 13708 |
+
"step": 29355
|
| 13709 |
+
},
|
| 13710 |
+
{
|
| 13711 |
+
"epoch": 0.05719371083389482,
|
| 13712 |
+
"grad_norm": 4.224997043609619,
|
| 13713 |
+
"learning_rate": 2.9758515742649063e-05,
|
| 13714 |
+
"loss": 1.78,
|
| 13715 |
+
"step": 29370
|
| 13716 |
+
},
|
| 13717 |
+
{
|
| 13718 |
+
"epoch": 0.057222921105005084,
|
| 13719 |
+
"grad_norm": 2.5549795627593994,
|
| 13720 |
+
"learning_rate": 2.9758269680281946e-05,
|
| 13721 |
+
"loss": 1.7375,
|
| 13722 |
+
"step": 29385
|
| 13723 |
+
},
|
| 13724 |
+
{
|
| 13725 |
+
"epoch": 0.057252131376115345,
|
| 13726 |
+
"grad_norm": 3.9945852756500244,
|
| 13727 |
+
"learning_rate": 2.9758023493633365e-05,
|
| 13728 |
+
"loss": 1.7078,
|
| 13729 |
+
"step": 29400
|
| 13730 |
+
},
|
| 13731 |
+
{
|
| 13732 |
+
"epoch": 0.057281341647225606,
|
| 13733 |
+
"grad_norm": 3.5278730392456055,
|
| 13734 |
+
"learning_rate": 2.975777718270539e-05,
|
| 13735 |
+
"loss": 1.7572,
|
| 13736 |
+
"step": 29415
|
| 13737 |
+
},
|
| 13738 |
+
{
|
| 13739 |
+
"epoch": 0.057310551918335874,
|
| 13740 |
+
"grad_norm": 3.9674766063690186,
|
| 13741 |
+
"learning_rate": 2.9757530747500104e-05,
|
| 13742 |
+
"loss": 1.9125,
|
| 13743 |
+
"step": 29430
|
| 13744 |
+
},
|
| 13745 |
+
{
|
| 13746 |
+
"epoch": 0.057339762189446135,
|
| 13747 |
+
"grad_norm": 3.498309373855591,
|
| 13748 |
+
"learning_rate": 2.9757284188019573e-05,
|
| 13749 |
+
"loss": 1.8242,
|
| 13750 |
+
"step": 29445
|
| 13751 |
+
},
|
| 13752 |
+
{
|
| 13753 |
+
"epoch": 0.057368972460556396,
|
| 13754 |
+
"grad_norm": 3.2913856506347656,
|
| 13755 |
+
"learning_rate": 2.9757037504265874e-05,
|
| 13756 |
+
"loss": 1.8559,
|
| 13757 |
+
"step": 29460
|
| 13758 |
+
},
|
| 13759 |
+
{
|
| 13760 |
+
"epoch": 0.05739818273166666,
|
| 13761 |
+
"grad_norm": 2.72259521484375,
|
| 13762 |
+
"learning_rate": 2.9756790696241088e-05,
|
| 13763 |
+
"loss": 1.7855,
|
| 13764 |
+
"step": 29475
|
| 13765 |
+
},
|
| 13766 |
+
{
|
| 13767 |
+
"epoch": 0.057427393002776925,
|
| 13768 |
+
"grad_norm": 7.3963704109191895,
|
| 13769 |
+
"learning_rate": 2.9756543763947292e-05,
|
| 13770 |
+
"loss": 1.9272,
|
| 13771 |
+
"step": 29490
|
| 13772 |
+
},
|
| 13773 |
+
{
|
| 13774 |
+
"epoch": 0.057456603273887186,
|
| 13775 |
+
"grad_norm": 2.901502847671509,
|
| 13776 |
+
"learning_rate": 2.9756296707386566e-05,
|
| 13777 |
+
"loss": 1.7294,
|
| 13778 |
+
"step": 29505
|
| 13779 |
+
},
|
| 13780 |
+
{
|
| 13781 |
+
"epoch": 0.05748581354499745,
|
| 13782 |
+
"grad_norm": 1.9630255699157715,
|
| 13783 |
+
"learning_rate": 2.9756049526560995e-05,
|
| 13784 |
+
"loss": 1.8477,
|
| 13785 |
+
"step": 29520
|
| 13786 |
+
},
|
| 13787 |
+
{
|
| 13788 |
+
"epoch": 0.057515023816107715,
|
| 13789 |
+
"grad_norm": 2.449836492538452,
|
| 13790 |
+
"learning_rate": 2.975580222147265e-05,
|
| 13791 |
+
"loss": 1.9282,
|
| 13792 |
+
"step": 29535
|
| 13793 |
+
},
|
| 13794 |
+
{
|
| 13795 |
+
"epoch": 0.057544234087217976,
|
| 13796 |
+
"grad_norm": 2.7461347579956055,
|
| 13797 |
+
"learning_rate": 2.9755554792123617e-05,
|
| 13798 |
+
"loss": 1.8883,
|
| 13799 |
+
"step": 29550
|
| 13800 |
+
},
|
| 13801 |
+
{
|
| 13802 |
+
"epoch": 0.05757344435832824,
|
| 13803 |
+
"grad_norm": 3.3567450046539307,
|
| 13804 |
+
"learning_rate": 2.9755307238515986e-05,
|
| 13805 |
+
"loss": 1.9717,
|
| 13806 |
+
"step": 29565
|
| 13807 |
+
},
|
| 13808 |
+
{
|
| 13809 |
+
"epoch": 0.0576026546294385,
|
| 13810 |
+
"grad_norm": 3.6341872215270996,
|
| 13811 |
+
"learning_rate": 2.975505956065184e-05,
|
| 13812 |
+
"loss": 1.9378,
|
| 13813 |
+
"step": 29580
|
| 13814 |
+
},
|
| 13815 |
+
{
|
| 13816 |
+
"epoch": 0.057631864900548765,
|
| 13817 |
+
"grad_norm": 4.170873165130615,
|
| 13818 |
+
"learning_rate": 2.9754811758533253e-05,
|
| 13819 |
+
"loss": 1.9077,
|
| 13820 |
+
"step": 29595
|
| 13821 |
+
},
|
| 13822 |
+
{
|
| 13823 |
+
"epoch": 0.057661075171659026,
|
| 13824 |
+
"grad_norm": 1.9185329675674438,
|
| 13825 |
+
"learning_rate": 2.975456383216233e-05,
|
| 13826 |
+
"loss": 1.7987,
|
| 13827 |
+
"step": 29610
|
| 13828 |
+
},
|
| 13829 |
+
{
|
| 13830 |
+
"epoch": 0.05769028544276929,
|
| 13831 |
+
"grad_norm": 2.3139781951904297,
|
| 13832 |
+
"learning_rate": 2.9754315781541144e-05,
|
| 13833 |
+
"loss": 1.9518,
|
| 13834 |
+
"step": 29625
|
| 13835 |
+
},
|
| 13836 |
+
{
|
| 13837 |
+
"epoch": 0.057719495713879555,
|
| 13838 |
+
"grad_norm": 3.581907033920288,
|
| 13839 |
+
"learning_rate": 2.9754067606671794e-05,
|
| 13840 |
+
"loss": 1.8729,
|
| 13841 |
+
"step": 29640
|
| 13842 |
+
},
|
| 13843 |
+
{
|
| 13844 |
+
"epoch": 0.057748705984989816,
|
| 13845 |
+
"grad_norm": 2.490278720855713,
|
| 13846 |
+
"learning_rate": 2.975381930755636e-05,
|
| 13847 |
+
"loss": 1.7576,
|
| 13848 |
+
"step": 29655
|
| 13849 |
+
},
|
| 13850 |
+
{
|
| 13851 |
+
"epoch": 0.05777791625610008,
|
| 13852 |
+
"grad_norm": 3.5740153789520264,
|
| 13853 |
+
"learning_rate": 2.9753570884196942e-05,
|
| 13854 |
+
"loss": 1.9749,
|
| 13855 |
+
"step": 29670
|
| 13856 |
+
},
|
| 13857 |
+
{
|
| 13858 |
+
"epoch": 0.05780712652721034,
|
| 13859 |
+
"grad_norm": 3.012131929397583,
|
| 13860 |
+
"learning_rate": 2.975332233659563e-05,
|
| 13861 |
+
"loss": 1.8704,
|
| 13862 |
+
"step": 29685
|
| 13863 |
+
},
|
| 13864 |
+
{
|
| 13865 |
+
"epoch": 0.057836336798320606,
|
| 13866 |
+
"grad_norm": 4.225703239440918,
|
| 13867 |
+
"learning_rate": 2.9753073664754514e-05,
|
| 13868 |
+
"loss": 1.6134,
|
| 13869 |
+
"step": 29700
|
| 13870 |
+
},
|
| 13871 |
+
{
|
| 13872 |
+
"epoch": 0.05786554706943087,
|
| 13873 |
+
"grad_norm": 3.840487480163574,
|
| 13874 |
+
"learning_rate": 2.9752824868675693e-05,
|
| 13875 |
+
"loss": 1.822,
|
| 13876 |
+
"step": 29715
|
| 13877 |
+
},
|
| 13878 |
+
{
|
| 13879 |
+
"epoch": 0.05789475734054113,
|
| 13880 |
+
"grad_norm": 2.8694260120391846,
|
| 13881 |
+
"learning_rate": 2.975257594836125e-05,
|
| 13882 |
+
"loss": 1.939,
|
| 13883 |
+
"step": 29730
|
| 13884 |
+
},
|
| 13885 |
+
{
|
| 13886 |
+
"epoch": 0.057923967611651396,
|
| 13887 |
+
"grad_norm": 3.4660465717315674,
|
| 13888 |
+
"learning_rate": 2.97523269038133e-05,
|
| 13889 |
+
"loss": 1.9517,
|
| 13890 |
+
"step": 29745
|
| 13891 |
+
},
|
| 13892 |
+
{
|
| 13893 |
+
"epoch": 0.05795317788276166,
|
| 13894 |
+
"grad_norm": 3.125666379928589,
|
| 13895 |
+
"learning_rate": 2.9752077735033924e-05,
|
| 13896 |
+
"loss": 1.7034,
|
| 13897 |
+
"step": 29760
|
| 13898 |
+
},
|
| 13899 |
+
{
|
| 13900 |
+
"epoch": 0.05798238815387192,
|
| 13901 |
+
"grad_norm": 4.895527362823486,
|
| 13902 |
+
"learning_rate": 2.975182844202523e-05,
|
| 13903 |
+
"loss": 1.8021,
|
| 13904 |
+
"step": 29775
|
| 13905 |
+
},
|
| 13906 |
+
{
|
| 13907 |
+
"epoch": 0.05801159842498218,
|
| 13908 |
+
"grad_norm": 3.4364778995513916,
|
| 13909 |
+
"learning_rate": 2.9751579024789314e-05,
|
| 13910 |
+
"loss": 1.9357,
|
| 13911 |
+
"step": 29790
|
| 13912 |
+
},
|
| 13913 |
+
{
|
| 13914 |
+
"epoch": 0.05804080869609245,
|
| 13915 |
+
"grad_norm": 3.209791421890259,
|
| 13916 |
+
"learning_rate": 2.9751329483328276e-05,
|
| 13917 |
+
"loss": 1.8969,
|
| 13918 |
+
"step": 29805
|
| 13919 |
+
},
|
| 13920 |
+
{
|
| 13921 |
+
"epoch": 0.05807001896720271,
|
| 13922 |
+
"grad_norm": 2.851810932159424,
|
| 13923 |
+
"learning_rate": 2.9751079817644217e-05,
|
| 13924 |
+
"loss": 1.8274,
|
| 13925 |
+
"step": 29820
|
| 13926 |
+
},
|
| 13927 |
+
{
|
| 13928 |
+
"epoch": 0.05809922923831297,
|
| 13929 |
+
"grad_norm": 1.9028266668319702,
|
| 13930 |
+
"learning_rate": 2.975083002773924e-05,
|
| 13931 |
+
"loss": 1.7498,
|
| 13932 |
+
"step": 29835
|
| 13933 |
+
},
|
| 13934 |
+
{
|
| 13935 |
+
"epoch": 0.05812843950942324,
|
| 13936 |
+
"grad_norm": 4.442659378051758,
|
| 13937 |
+
"learning_rate": 2.9750580113615448e-05,
|
| 13938 |
+
"loss": 1.906,
|
| 13939 |
+
"step": 29850
|
| 13940 |
+
},
|
| 13941 |
+
{
|
| 13942 |
+
"epoch": 0.0581576497805335,
|
| 13943 |
+
"grad_norm": 4.513755798339844,
|
| 13944 |
+
"learning_rate": 2.9750330075274948e-05,
|
| 13945 |
+
"loss": 1.7855,
|
| 13946 |
+
"step": 29865
|
| 13947 |
+
},
|
| 13948 |
+
{
|
| 13949 |
+
"epoch": 0.05818686005164376,
|
| 13950 |
+
"grad_norm": 2.4421234130859375,
|
| 13951 |
+
"learning_rate": 2.975007991271984e-05,
|
| 13952 |
+
"loss": 1.8355,
|
| 13953 |
+
"step": 29880
|
| 13954 |
+
},
|
| 13955 |
+
{
|
| 13956 |
+
"epoch": 0.05821607032275402,
|
| 13957 |
+
"grad_norm": 2.0383033752441406,
|
| 13958 |
+
"learning_rate": 2.974982962595224e-05,
|
| 13959 |
+
"loss": 1.8456,
|
| 13960 |
+
"step": 29895
|
| 13961 |
+
},
|
| 13962 |
+
{
|
| 13963 |
+
"epoch": 0.05824528059386429,
|
| 13964 |
+
"grad_norm": 2.9908933639526367,
|
| 13965 |
+
"learning_rate": 2.9749579214974245e-05,
|
| 13966 |
+
"loss": 1.7223,
|
| 13967 |
+
"step": 29910
|
| 13968 |
+
},
|
| 13969 |
+
{
|
| 13970 |
+
"epoch": 0.05827449086497455,
|
| 13971 |
+
"grad_norm": 3.1746137142181396,
|
| 13972 |
+
"learning_rate": 2.9749328679787976e-05,
|
| 13973 |
+
"loss": 1.8949,
|
| 13974 |
+
"step": 29925
|
| 13975 |
+
},
|
| 13976 |
+
{
|
| 13977 |
+
"epoch": 0.05830370113608481,
|
| 13978 |
+
"grad_norm": 3.351712942123413,
|
| 13979 |
+
"learning_rate": 2.9749078020395526e-05,
|
| 13980 |
+
"loss": 1.8423,
|
| 13981 |
+
"step": 29940
|
| 13982 |
+
},
|
| 13983 |
+
{
|
| 13984 |
+
"epoch": 0.05833291140719508,
|
| 13985 |
+
"grad_norm": 4.363383769989014,
|
| 13986 |
+
"learning_rate": 2.9748827236799024e-05,
|
| 13987 |
+
"loss": 2.0718,
|
| 13988 |
+
"step": 29955
|
| 13989 |
+
},
|
| 13990 |
+
{
|
| 13991 |
+
"epoch": 0.05836212167830534,
|
| 13992 |
+
"grad_norm": 3.5311388969421387,
|
| 13993 |
+
"learning_rate": 2.974857632900057e-05,
|
| 13994 |
+
"loss": 1.8393,
|
| 13995 |
+
"step": 29970
|
| 13996 |
+
},
|
| 13997 |
+
{
|
| 13998 |
+
"epoch": 0.0583913319494156,
|
| 13999 |
+
"grad_norm": 2.739178419113159,
|
| 14000 |
+
"learning_rate": 2.974832529700228e-05,
|
| 14001 |
+
"loss": 1.901,
|
| 14002 |
+
"step": 29985
|
| 14003 |
+
},
|
| 14004 |
+
{
|
| 14005 |
+
"epoch": 0.05842054222052586,
|
| 14006 |
+
"grad_norm": 5.376321315765381,
|
| 14007 |
+
"learning_rate": 2.974807414080627e-05,
|
| 14008 |
+
"loss": 1.9742,
|
| 14009 |
+
"step": 30000
|
| 14010 |
}
|
| 14011 |
],
|
| 14012 |
"logging_steps": 15,
|
|
|
|
| 14026 |
"attributes": {}
|
| 14027 |
}
|
| 14028 |
},
|
| 14029 |
+
"total_flos": 4.741878296671027e+16,
|
| 14030 |
"train_batch_size": 4,
|
| 14031 |
"trial_name": null,
|
| 14032 |
"trial_params": null
|