Training in progress, step 1023, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 83945296
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7e4f80648c997705b666e664ffadcb1627b78f74639290ea3d039ff0c8669a9d
|
3 |
size 83945296
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 43123028
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ddb9b64b89f85cf8f1e2c8f87d6d17e7ae4c874d9f24dcb4bd9b2c64dd1462aa
|
3 |
size 43123028
|
last-checkpoint/rng_state.pth
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 14244
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:158da7aa6e82ada935c7a8526b9425b4c5bbf20ddc5f8c48a6eb554f20b379e1
|
3 |
size 14244
|
last-checkpoint/scheduler.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1064
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7171d4744726664567e10d94beb7206e0d8f92056cef040c63510146c590be2d
|
3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
-
"epoch": 0.
|
5 |
"eval_steps": 341,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -4805,6 +4805,2401 @@
|
|
4805 |
"eval_samples_per_second": 27.594,
|
4806 |
"eval_steps_per_second": 13.797,
|
4807 |
"step": 682
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
4808 |
}
|
4809 |
],
|
4810 |
"logging_steps": 1,
|
@@ -4824,7 +7219,7 @@
|
|
4824 |
"attributes": {}
|
4825 |
}
|
4826 |
},
|
4827 |
-
"total_flos":
|
4828 |
"train_batch_size": 2,
|
4829 |
"trial_name": null,
|
4830 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.2006079027355623,
|
5 |
"eval_steps": 341,
|
6 |
+
"global_step": 1023,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
4805 |
"eval_samples_per_second": 27.594,
|
4806 |
"eval_steps_per_second": 13.797,
|
4807 |
"step": 682
|
4808 |
+
},
|
4809 |
+
{
|
4810 |
+
"epoch": 0.1339346994803412,
|
4811 |
+
"grad_norm": 11.698119163513184,
|
4812 |
+
"learning_rate": 0.00010058134251626983,
|
4813 |
+
"loss": 2.1025,
|
4814 |
+
"step": 683
|
4815 |
+
},
|
4816 |
+
{
|
4817 |
+
"epoch": 0.1341307971369742,
|
4818 |
+
"grad_norm": 5.185816287994385,
|
4819 |
+
"learning_rate": 0.00010034880676718832,
|
4820 |
+
"loss": 1.0986,
|
4821 |
+
"step": 684
|
4822 |
+
},
|
4823 |
+
{
|
4824 |
+
"epoch": 0.1343268947936072,
|
4825 |
+
"grad_norm": 6.936627388000488,
|
4826 |
+
"learning_rate": 0.0001001162691319675,
|
4827 |
+
"loss": 1.3077,
|
4828 |
+
"step": 685
|
4829 |
+
},
|
4830 |
+
{
|
4831 |
+
"epoch": 0.13452299245024021,
|
4832 |
+
"grad_norm": 4.661152362823486,
|
4833 |
+
"learning_rate": 9.988373086803251e-05,
|
4834 |
+
"loss": 1.6312,
|
4835 |
+
"step": 686
|
4836 |
+
},
|
4837 |
+
{
|
4838 |
+
"epoch": 0.13471909010687322,
|
4839 |
+
"grad_norm": 4.227341651916504,
|
4840 |
+
"learning_rate": 9.965119323281174e-05,
|
4841 |
+
"loss": 0.6856,
|
4842 |
+
"step": 687
|
4843 |
+
},
|
4844 |
+
{
|
4845 |
+
"epoch": 0.13491518776350622,
|
4846 |
+
"grad_norm": 24.00637435913086,
|
4847 |
+
"learning_rate": 9.941865748373017e-05,
|
4848 |
+
"loss": 2.8267,
|
4849 |
+
"step": 688
|
4850 |
+
},
|
4851 |
+
{
|
4852 |
+
"epoch": 0.13511128542013923,
|
4853 |
+
"grad_norm": 4.701898574829102,
|
4854 |
+
"learning_rate": 9.918612487820273e-05,
|
4855 |
+
"loss": 1.7314,
|
4856 |
+
"step": 689
|
4857 |
+
},
|
4858 |
+
{
|
4859 |
+
"epoch": 0.13530738307677223,
|
4860 |
+
"grad_norm": 7.700753688812256,
|
4861 |
+
"learning_rate": 9.895359667362724e-05,
|
4862 |
+
"loss": 1.1831,
|
4863 |
+
"step": 690
|
4864 |
+
},
|
4865 |
+
{
|
4866 |
+
"epoch": 0.13550348073340523,
|
4867 |
+
"grad_norm": 5.045268535614014,
|
4868 |
+
"learning_rate": 9.87210741273777e-05,
|
4869 |
+
"loss": 2.6148,
|
4870 |
+
"step": 691
|
4871 |
+
},
|
4872 |
+
{
|
4873 |
+
"epoch": 0.13569957839003824,
|
4874 |
+
"grad_norm": 8.903031349182129,
|
4875 |
+
"learning_rate": 9.848855849679754e-05,
|
4876 |
+
"loss": 2.3442,
|
4877 |
+
"step": 692
|
4878 |
+
},
|
4879 |
+
{
|
4880 |
+
"epoch": 0.13589567604667124,
|
4881 |
+
"grad_norm": 6.869392395019531,
|
4882 |
+
"learning_rate": 9.82560510391929e-05,
|
4883 |
+
"loss": 1.7329,
|
4884 |
+
"step": 693
|
4885 |
+
},
|
4886 |
+
{
|
4887 |
+
"epoch": 0.13609177370330425,
|
4888 |
+
"grad_norm": 5.181933403015137,
|
4889 |
+
"learning_rate": 9.802355301182556e-05,
|
4890 |
+
"loss": 2.2152,
|
4891 |
+
"step": 694
|
4892 |
+
},
|
4893 |
+
{
|
4894 |
+
"epoch": 0.13628787135993725,
|
4895 |
+
"grad_norm": 9.427200317382812,
|
4896 |
+
"learning_rate": 9.779106567190642e-05,
|
4897 |
+
"loss": 0.9606,
|
4898 |
+
"step": 695
|
4899 |
+
},
|
4900 |
+
{
|
4901 |
+
"epoch": 0.13648396901657026,
|
4902 |
+
"grad_norm": 5.893560409545898,
|
4903 |
+
"learning_rate": 9.755859027658848e-05,
|
4904 |
+
"loss": 0.7759,
|
4905 |
+
"step": 696
|
4906 |
+
},
|
4907 |
+
{
|
4908 |
+
"epoch": 0.13668006667320326,
|
4909 |
+
"grad_norm": 3.864348888397217,
|
4910 |
+
"learning_rate": 9.73261280829603e-05,
|
4911 |
+
"loss": 0.714,
|
4912 |
+
"step": 697
|
4913 |
+
},
|
4914 |
+
{
|
4915 |
+
"epoch": 0.13687616432983626,
|
4916 |
+
"grad_norm": 7.077169418334961,
|
4917 |
+
"learning_rate": 9.709368034803892e-05,
|
4918 |
+
"loss": 1.9489,
|
4919 |
+
"step": 698
|
4920 |
+
},
|
4921 |
+
{
|
4922 |
+
"epoch": 0.13707226198646927,
|
4923 |
+
"grad_norm": 10.102705955505371,
|
4924 |
+
"learning_rate": 9.686124832876328e-05,
|
4925 |
+
"loss": 2.2884,
|
4926 |
+
"step": 699
|
4927 |
+
},
|
4928 |
+
{
|
4929 |
+
"epoch": 0.13726835964310227,
|
4930 |
+
"grad_norm": 6.154596328735352,
|
4931 |
+
"learning_rate": 9.662883328198731e-05,
|
4932 |
+
"loss": 1.1614,
|
4933 |
+
"step": 700
|
4934 |
+
},
|
4935 |
+
{
|
4936 |
+
"epoch": 0.13746445729973528,
|
4937 |
+
"grad_norm": 8.005276679992676,
|
4938 |
+
"learning_rate": 9.639643646447316e-05,
|
4939 |
+
"loss": 2.5097,
|
4940 |
+
"step": 701
|
4941 |
+
},
|
4942 |
+
{
|
4943 |
+
"epoch": 0.13766055495636828,
|
4944 |
+
"grad_norm": 12.45338249206543,
|
4945 |
+
"learning_rate": 9.616405913288438e-05,
|
4946 |
+
"loss": 0.8686,
|
4947 |
+
"step": 702
|
4948 |
+
},
|
4949 |
+
{
|
4950 |
+
"epoch": 0.13785665261300128,
|
4951 |
+
"grad_norm": 4.405421257019043,
|
4952 |
+
"learning_rate": 9.593170254377916e-05,
|
4953 |
+
"loss": 0.8499,
|
4954 |
+
"step": 703
|
4955 |
+
},
|
4956 |
+
{
|
4957 |
+
"epoch": 0.1380527502696343,
|
4958 |
+
"grad_norm": 9.178966522216797,
|
4959 |
+
"learning_rate": 9.569936795360363e-05,
|
4960 |
+
"loss": 1.7378,
|
4961 |
+
"step": 704
|
4962 |
+
},
|
4963 |
+
{
|
4964 |
+
"epoch": 0.1382488479262673,
|
4965 |
+
"grad_norm": 9.432720184326172,
|
4966 |
+
"learning_rate": 9.546705661868484e-05,
|
4967 |
+
"loss": 1.5712,
|
4968 |
+
"step": 705
|
4969 |
+
},
|
4970 |
+
{
|
4971 |
+
"epoch": 0.1384449455829003,
|
4972 |
+
"grad_norm": 5.602529048919678,
|
4973 |
+
"learning_rate": 9.523476979522411e-05,
|
4974 |
+
"loss": 1.8486,
|
4975 |
+
"step": 706
|
4976 |
+
},
|
4977 |
+
{
|
4978 |
+
"epoch": 0.1386410432395333,
|
4979 |
+
"grad_norm": 7.334858417510986,
|
4980 |
+
"learning_rate": 9.50025087392902e-05,
|
4981 |
+
"loss": 1.3093,
|
4982 |
+
"step": 707
|
4983 |
+
},
|
4984 |
+
{
|
4985 |
+
"epoch": 0.13883714089616628,
|
4986 |
+
"grad_norm": 16.70349884033203,
|
4987 |
+
"learning_rate": 9.477027470681269e-05,
|
4988 |
+
"loss": 1.8117,
|
4989 |
+
"step": 708
|
4990 |
+
},
|
4991 |
+
{
|
4992 |
+
"epoch": 0.13903323855279928,
|
4993 |
+
"grad_norm": 4.915411472320557,
|
4994 |
+
"learning_rate": 9.453806895357482e-05,
|
4995 |
+
"loss": 1.296,
|
4996 |
+
"step": 709
|
4997 |
+
},
|
4998 |
+
{
|
4999 |
+
"epoch": 0.1392293362094323,
|
5000 |
+
"grad_norm": 9.790595054626465,
|
5001 |
+
"learning_rate": 9.430589273520703e-05,
|
5002 |
+
"loss": 1.7763,
|
5003 |
+
"step": 710
|
5004 |
+
},
|
5005 |
+
{
|
5006 |
+
"epoch": 0.1394254338660653,
|
5007 |
+
"grad_norm": 5.754830837249756,
|
5008 |
+
"learning_rate": 9.407374730718e-05,
|
5009 |
+
"loss": 1.2865,
|
5010 |
+
"step": 711
|
5011 |
+
},
|
5012 |
+
{
|
5013 |
+
"epoch": 0.1396215315226983,
|
5014 |
+
"grad_norm": 6.5165534019470215,
|
5015 |
+
"learning_rate": 9.3841633924798e-05,
|
5016 |
+
"loss": 1.0498,
|
5017 |
+
"step": 712
|
5018 |
+
},
|
5019 |
+
{
|
5020 |
+
"epoch": 0.1398176291793313,
|
5021 |
+
"grad_norm": 4.821907043457031,
|
5022 |
+
"learning_rate": 9.360955384319195e-05,
|
5023 |
+
"loss": 1.1823,
|
5024 |
+
"step": 713
|
5025 |
+
},
|
5026 |
+
{
|
5027 |
+
"epoch": 0.1400137268359643,
|
5028 |
+
"grad_norm": 7.226667881011963,
|
5029 |
+
"learning_rate": 9.337750831731266e-05,
|
5030 |
+
"loss": 1.0176,
|
5031 |
+
"step": 714
|
5032 |
+
},
|
5033 |
+
{
|
5034 |
+
"epoch": 0.1402098244925973,
|
5035 |
+
"grad_norm": 5.560497760772705,
|
5036 |
+
"learning_rate": 9.314549860192417e-05,
|
5037 |
+
"loss": 1.1645,
|
5038 |
+
"step": 715
|
5039 |
+
},
|
5040 |
+
{
|
5041 |
+
"epoch": 0.1404059221492303,
|
5042 |
+
"grad_norm": 4.834329128265381,
|
5043 |
+
"learning_rate": 9.291352595159682e-05,
|
5044 |
+
"loss": 1.682,
|
5045 |
+
"step": 716
|
5046 |
+
},
|
5047 |
+
{
|
5048 |
+
"epoch": 0.14060201980586332,
|
5049 |
+
"grad_norm": 6.791970729827881,
|
5050 |
+
"learning_rate": 9.268159162070058e-05,
|
5051 |
+
"loss": 1.5408,
|
5052 |
+
"step": 717
|
5053 |
+
},
|
5054 |
+
{
|
5055 |
+
"epoch": 0.14079811746249632,
|
5056 |
+
"grad_norm": 5.209107875823975,
|
5057 |
+
"learning_rate": 9.244969686339813e-05,
|
5058 |
+
"loss": 0.6371,
|
5059 |
+
"step": 718
|
5060 |
+
},
|
5061 |
+
{
|
5062 |
+
"epoch": 0.14099421511912932,
|
5063 |
+
"grad_norm": 7.674073696136475,
|
5064 |
+
"learning_rate": 9.221784293363824e-05,
|
5065 |
+
"loss": 1.9635,
|
5066 |
+
"step": 719
|
5067 |
+
},
|
5068 |
+
{
|
5069 |
+
"epoch": 0.14119031277576233,
|
5070 |
+
"grad_norm": 13.054336547851562,
|
5071 |
+
"learning_rate": 9.198603108514888e-05,
|
5072 |
+
"loss": 1.2977,
|
5073 |
+
"step": 720
|
5074 |
+
},
|
5075 |
+
{
|
5076 |
+
"epoch": 0.14138641043239533,
|
5077 |
+
"grad_norm": 7.0277299880981445,
|
5078 |
+
"learning_rate": 9.175426257143045e-05,
|
5079 |
+
"loss": 1.3063,
|
5080 |
+
"step": 721
|
5081 |
+
},
|
5082 |
+
{
|
5083 |
+
"epoch": 0.14158250808902834,
|
5084 |
+
"grad_norm": 10.085116386413574,
|
5085 |
+
"learning_rate": 9.152253864574898e-05,
|
5086 |
+
"loss": 1.5895,
|
5087 |
+
"step": 722
|
5088 |
+
},
|
5089 |
+
{
|
5090 |
+
"epoch": 0.14177860574566134,
|
5091 |
+
"grad_norm": 11.588119506835938,
|
5092 |
+
"learning_rate": 9.129086056112955e-05,
|
5093 |
+
"loss": 1.6528,
|
5094 |
+
"step": 723
|
5095 |
+
},
|
5096 |
+
{
|
5097 |
+
"epoch": 0.14197470340229434,
|
5098 |
+
"grad_norm": 6.269704818725586,
|
5099 |
+
"learning_rate": 9.10592295703492e-05,
|
5100 |
+
"loss": 1.1952,
|
5101 |
+
"step": 724
|
5102 |
+
},
|
5103 |
+
{
|
5104 |
+
"epoch": 0.14217080105892735,
|
5105 |
+
"grad_norm": 5.680713176727295,
|
5106 |
+
"learning_rate": 9.082764692593039e-05,
|
5107 |
+
"loss": 2.322,
|
5108 |
+
"step": 725
|
5109 |
+
},
|
5110 |
+
{
|
5111 |
+
"epoch": 0.14236689871556035,
|
5112 |
+
"grad_norm": 6.14839506149292,
|
5113 |
+
"learning_rate": 9.059611388013408e-05,
|
5114 |
+
"loss": 2.6696,
|
5115 |
+
"step": 726
|
5116 |
+
},
|
5117 |
+
{
|
5118 |
+
"epoch": 0.14256299637219336,
|
5119 |
+
"grad_norm": 13.12667179107666,
|
5120 |
+
"learning_rate": 9.03646316849532e-05,
|
5121 |
+
"loss": 2.0335,
|
5122 |
+
"step": 727
|
5123 |
+
},
|
5124 |
+
{
|
5125 |
+
"epoch": 0.14275909402882636,
|
5126 |
+
"grad_norm": 7.358575820922852,
|
5127 |
+
"learning_rate": 9.01332015921055e-05,
|
5128 |
+
"loss": 0.8538,
|
5129 |
+
"step": 728
|
5130 |
+
},
|
5131 |
+
{
|
5132 |
+
"epoch": 0.14295519168545937,
|
5133 |
+
"grad_norm": 11.932861328125,
|
5134 |
+
"learning_rate": 8.990182485302713e-05,
|
5135 |
+
"loss": 1.4472,
|
5136 |
+
"step": 729
|
5137 |
+
},
|
5138 |
+
{
|
5139 |
+
"epoch": 0.14315128934209237,
|
5140 |
+
"grad_norm": 8.689105987548828,
|
5141 |
+
"learning_rate": 8.967050271886564e-05,
|
5142 |
+
"loss": 1.8314,
|
5143 |
+
"step": 730
|
5144 |
+
},
|
5145 |
+
{
|
5146 |
+
"epoch": 0.14334738699872537,
|
5147 |
+
"grad_norm": 5.939968585968018,
|
5148 |
+
"learning_rate": 8.943923644047342e-05,
|
5149 |
+
"loss": 0.8076,
|
5150 |
+
"step": 731
|
5151 |
+
},
|
5152 |
+
{
|
5153 |
+
"epoch": 0.14354348465535838,
|
5154 |
+
"grad_norm": 8.181587219238281,
|
5155 |
+
"learning_rate": 8.920802726840076e-05,
|
5156 |
+
"loss": 1.4664,
|
5157 |
+
"step": 732
|
5158 |
+
},
|
5159 |
+
{
|
5160 |
+
"epoch": 0.14373958231199138,
|
5161 |
+
"grad_norm": 7.449463844299316,
|
5162 |
+
"learning_rate": 8.897687645288914e-05,
|
5163 |
+
"loss": 1.0997,
|
5164 |
+
"step": 733
|
5165 |
+
},
|
5166 |
+
{
|
5167 |
+
"epoch": 0.14393567996862439,
|
5168 |
+
"grad_norm": 5.9800848960876465,
|
5169 |
+
"learning_rate": 8.874578524386447e-05,
|
5170 |
+
"loss": 0.9288,
|
5171 |
+
"step": 734
|
5172 |
+
},
|
5173 |
+
{
|
5174 |
+
"epoch": 0.1441317776252574,
|
5175 |
+
"grad_norm": 8.202332496643066,
|
5176 |
+
"learning_rate": 8.851475489093045e-05,
|
5177 |
+
"loss": 1.5658,
|
5178 |
+
"step": 735
|
5179 |
+
},
|
5180 |
+
{
|
5181 |
+
"epoch": 0.1443278752818904,
|
5182 |
+
"grad_norm": 4.058600425720215,
|
5183 |
+
"learning_rate": 8.828378664336158e-05,
|
5184 |
+
"loss": 1.1746,
|
5185 |
+
"step": 736
|
5186 |
+
},
|
5187 |
+
{
|
5188 |
+
"epoch": 0.1445239729385234,
|
5189 |
+
"grad_norm": 5.467014789581299,
|
5190 |
+
"learning_rate": 8.805288175009659e-05,
|
5191 |
+
"loss": 0.9683,
|
5192 |
+
"step": 737
|
5193 |
+
},
|
5194 |
+
{
|
5195 |
+
"epoch": 0.14472007059515637,
|
5196 |
+
"grad_norm": 8.572188377380371,
|
5197 |
+
"learning_rate": 8.782204145973162e-05,
|
5198 |
+
"loss": 1.3502,
|
5199 |
+
"step": 738
|
5200 |
+
},
|
5201 |
+
{
|
5202 |
+
"epoch": 0.14491616825178938,
|
5203 |
+
"grad_norm": 6.00791072845459,
|
5204 |
+
"learning_rate": 8.759126702051348e-05,
|
5205 |
+
"loss": 1.5011,
|
5206 |
+
"step": 739
|
5207 |
+
},
|
5208 |
+
{
|
5209 |
+
"epoch": 0.14511226590842238,
|
5210 |
+
"grad_norm": 7.9531989097595215,
|
5211 |
+
"learning_rate": 8.736055968033289e-05,
|
5212 |
+
"loss": 2.3835,
|
5213 |
+
"step": 740
|
5214 |
+
},
|
5215 |
+
{
|
5216 |
+
"epoch": 0.1453083635650554,
|
5217 |
+
"grad_norm": 7.390120029449463,
|
5218 |
+
"learning_rate": 8.712992068671774e-05,
|
5219 |
+
"loss": 1.1469,
|
5220 |
+
"step": 741
|
5221 |
+
},
|
5222 |
+
{
|
5223 |
+
"epoch": 0.1455044612216884,
|
5224 |
+
"grad_norm": 6.503965377807617,
|
5225 |
+
"learning_rate": 8.689935128682635e-05,
|
5226 |
+
"loss": 1.9587,
|
5227 |
+
"step": 742
|
5228 |
+
},
|
5229 |
+
{
|
5230 |
+
"epoch": 0.1457005588783214,
|
5231 |
+
"grad_norm": 7.921380996704102,
|
5232 |
+
"learning_rate": 8.666885272744073e-05,
|
5233 |
+
"loss": 1.9997,
|
5234 |
+
"step": 743
|
5235 |
+
},
|
5236 |
+
{
|
5237 |
+
"epoch": 0.1458966565349544,
|
5238 |
+
"grad_norm": 7.979074001312256,
|
5239 |
+
"learning_rate": 8.643842625495978e-05,
|
5240 |
+
"loss": 1.4087,
|
5241 |
+
"step": 744
|
5242 |
+
},
|
5243 |
+
{
|
5244 |
+
"epoch": 0.1460927541915874,
|
5245 |
+
"grad_norm": 5.653082847595215,
|
5246 |
+
"learning_rate": 8.620807311539259e-05,
|
5247 |
+
"loss": 1.3974,
|
5248 |
+
"step": 745
|
5249 |
+
},
|
5250 |
+
{
|
5251 |
+
"epoch": 0.1462888518482204,
|
5252 |
+
"grad_norm": 7.764763355255127,
|
5253 |
+
"learning_rate": 8.597779455435183e-05,
|
5254 |
+
"loss": 1.4269,
|
5255 |
+
"step": 746
|
5256 |
+
},
|
5257 |
+
{
|
5258 |
+
"epoch": 0.1464849495048534,
|
5259 |
+
"grad_norm": 11.947178840637207,
|
5260 |
+
"learning_rate": 8.574759181704676e-05,
|
5261 |
+
"loss": 2.4265,
|
5262 |
+
"step": 747
|
5263 |
+
},
|
5264 |
+
{
|
5265 |
+
"epoch": 0.14668104716148642,
|
5266 |
+
"grad_norm": 9.1488037109375,
|
5267 |
+
"learning_rate": 8.551746614827669e-05,
|
5268 |
+
"loss": 2.2206,
|
5269 |
+
"step": 748
|
5270 |
+
},
|
5271 |
+
{
|
5272 |
+
"epoch": 0.14687714481811942,
|
5273 |
+
"grad_norm": 5.008571147918701,
|
5274 |
+
"learning_rate": 8.528741879242415e-05,
|
5275 |
+
"loss": 2.1277,
|
5276 |
+
"step": 749
|
5277 |
+
},
|
5278 |
+
{
|
5279 |
+
"epoch": 0.14707324247475242,
|
5280 |
+
"grad_norm": 4.040747165679932,
|
5281 |
+
"learning_rate": 8.505745099344827e-05,
|
5282 |
+
"loss": 0.7625,
|
5283 |
+
"step": 750
|
5284 |
+
},
|
5285 |
+
{
|
5286 |
+
"epoch": 0.14726934013138543,
|
5287 |
+
"grad_norm": 8.826830863952637,
|
5288 |
+
"learning_rate": 8.482756399487793e-05,
|
5289 |
+
"loss": 1.1674,
|
5290 |
+
"step": 751
|
5291 |
+
},
|
5292 |
+
{
|
5293 |
+
"epoch": 0.14746543778801843,
|
5294 |
+
"grad_norm": 4.931051731109619,
|
5295 |
+
"learning_rate": 8.45977590398051e-05,
|
5296 |
+
"loss": 1.1831,
|
5297 |
+
"step": 752
|
5298 |
+
},
|
5299 |
+
{
|
5300 |
+
"epoch": 0.14766153544465144,
|
5301 |
+
"grad_norm": 5.232216835021973,
|
5302 |
+
"learning_rate": 8.436803737087806e-05,
|
5303 |
+
"loss": 1.0998,
|
5304 |
+
"step": 753
|
5305 |
+
},
|
5306 |
+
{
|
5307 |
+
"epoch": 0.14785763310128444,
|
5308 |
+
"grad_norm": 9.30781364440918,
|
5309 |
+
"learning_rate": 8.413840023029488e-05,
|
5310 |
+
"loss": 2.8714,
|
5311 |
+
"step": 754
|
5312 |
+
},
|
5313 |
+
{
|
5314 |
+
"epoch": 0.14805373075791745,
|
5315 |
+
"grad_norm": 6.524617671966553,
|
5316 |
+
"learning_rate": 8.390884885979639e-05,
|
5317 |
+
"loss": 1.3297,
|
5318 |
+
"step": 755
|
5319 |
+
},
|
5320 |
+
{
|
5321 |
+
"epoch": 0.14824982841455045,
|
5322 |
+
"grad_norm": 6.3461432456970215,
|
5323 |
+
"learning_rate": 8.367938450065967e-05,
|
5324 |
+
"loss": 1.6513,
|
5325 |
+
"step": 756
|
5326 |
+
},
|
5327 |
+
{
|
5328 |
+
"epoch": 0.14844592607118345,
|
5329 |
+
"grad_norm": 9.379528999328613,
|
5330 |
+
"learning_rate": 8.345000839369132e-05,
|
5331 |
+
"loss": 1.3968,
|
5332 |
+
"step": 757
|
5333 |
+
},
|
5334 |
+
{
|
5335 |
+
"epoch": 0.14864202372781646,
|
5336 |
+
"grad_norm": 9.503636360168457,
|
5337 |
+
"learning_rate": 8.322072177922073e-05,
|
5338 |
+
"loss": 1.0089,
|
5339 |
+
"step": 758
|
5340 |
+
},
|
5341 |
+
{
|
5342 |
+
"epoch": 0.14883812138444946,
|
5343 |
+
"grad_norm": 4.907662868499756,
|
5344 |
+
"learning_rate": 8.299152589709336e-05,
|
5345 |
+
"loss": 2.4323,
|
5346 |
+
"step": 759
|
5347 |
+
},
|
5348 |
+
{
|
5349 |
+
"epoch": 0.14903421904108247,
|
5350 |
+
"grad_norm": 7.7067084312438965,
|
5351 |
+
"learning_rate": 8.2762421986664e-05,
|
5352 |
+
"loss": 1.4224,
|
5353 |
+
"step": 760
|
5354 |
+
},
|
5355 |
+
{
|
5356 |
+
"epoch": 0.14923031669771547,
|
5357 |
+
"grad_norm": 5.226773738861084,
|
5358 |
+
"learning_rate": 8.253341128679018e-05,
|
5359 |
+
"loss": 2.0016,
|
5360 |
+
"step": 761
|
5361 |
+
},
|
5362 |
+
{
|
5363 |
+
"epoch": 0.14942641435434847,
|
5364 |
+
"grad_norm": 8.012574195861816,
|
5365 |
+
"learning_rate": 8.230449503582538e-05,
|
5366 |
+
"loss": 1.8039,
|
5367 |
+
"step": 762
|
5368 |
+
},
|
5369 |
+
{
|
5370 |
+
"epoch": 0.14962251201098148,
|
5371 |
+
"grad_norm": 6.705967426300049,
|
5372 |
+
"learning_rate": 8.207567447161231e-05,
|
5373 |
+
"loss": 2.5853,
|
5374 |
+
"step": 763
|
5375 |
+
},
|
5376 |
+
{
|
5377 |
+
"epoch": 0.14981860966761448,
|
5378 |
+
"grad_norm": 9.187888145446777,
|
5379 |
+
"learning_rate": 8.184695083147627e-05,
|
5380 |
+
"loss": 1.0529,
|
5381 |
+
"step": 764
|
5382 |
+
},
|
5383 |
+
{
|
5384 |
+
"epoch": 0.1500147073242475,
|
5385 |
+
"grad_norm": 4.542299270629883,
|
5386 |
+
"learning_rate": 8.16183253522186e-05,
|
5387 |
+
"loss": 1.2803,
|
5388 |
+
"step": 765
|
5389 |
+
},
|
5390 |
+
{
|
5391 |
+
"epoch": 0.1502108049808805,
|
5392 |
+
"grad_norm": 5.094648361206055,
|
5393 |
+
"learning_rate": 8.138979927010964e-05,
|
5394 |
+
"loss": 0.9061,
|
5395 |
+
"step": 766
|
5396 |
+
},
|
5397 |
+
{
|
5398 |
+
"epoch": 0.1504069026375135,
|
5399 |
+
"grad_norm": 13.718326568603516,
|
5400 |
+
"learning_rate": 8.116137382088233e-05,
|
5401 |
+
"loss": 2.0593,
|
5402 |
+
"step": 767
|
5403 |
+
},
|
5404 |
+
{
|
5405 |
+
"epoch": 0.15060300029414647,
|
5406 |
+
"grad_norm": 6.143137454986572,
|
5407 |
+
"learning_rate": 8.093305023972546e-05,
|
5408 |
+
"loss": 0.8494,
|
5409 |
+
"step": 768
|
5410 |
+
},
|
5411 |
+
{
|
5412 |
+
"epoch": 0.15079909795077948,
|
5413 |
+
"grad_norm": 6.3516154289245605,
|
5414 |
+
"learning_rate": 8.070482976127703e-05,
|
5415 |
+
"loss": 2.214,
|
5416 |
+
"step": 769
|
5417 |
+
},
|
5418 |
+
{
|
5419 |
+
"epoch": 0.15099519560741248,
|
5420 |
+
"grad_norm": 9.909046173095703,
|
5421 |
+
"learning_rate": 8.047671361961741e-05,
|
5422 |
+
"loss": 0.928,
|
5423 |
+
"step": 770
|
5424 |
+
},
|
5425 |
+
{
|
5426 |
+
"epoch": 0.15119129326404548,
|
5427 |
+
"grad_norm": 6.7013373374938965,
|
5428 |
+
"learning_rate": 8.024870304826286e-05,
|
5429 |
+
"loss": 1.2937,
|
5430 |
+
"step": 771
|
5431 |
+
},
|
5432 |
+
{
|
5433 |
+
"epoch": 0.1513873909206785,
|
5434 |
+
"grad_norm": 7.552654266357422,
|
5435 |
+
"learning_rate": 8.002079928015867e-05,
|
5436 |
+
"loss": 2.3988,
|
5437 |
+
"step": 772
|
5438 |
+
},
|
5439 |
+
{
|
5440 |
+
"epoch": 0.1515834885773115,
|
5441 |
+
"grad_norm": 4.806221961975098,
|
5442 |
+
"learning_rate": 7.97930035476728e-05,
|
5443 |
+
"loss": 1.9656,
|
5444 |
+
"step": 773
|
5445 |
+
},
|
5446 |
+
{
|
5447 |
+
"epoch": 0.1517795862339445,
|
5448 |
+
"grad_norm": 13.320466041564941,
|
5449 |
+
"learning_rate": 7.956531708258887e-05,
|
5450 |
+
"loss": 2.1333,
|
5451 |
+
"step": 774
|
5452 |
+
},
|
5453 |
+
{
|
5454 |
+
"epoch": 0.1519756838905775,
|
5455 |
+
"grad_norm": 7.802790641784668,
|
5456 |
+
"learning_rate": 7.933774111609964e-05,
|
5457 |
+
"loss": 0.766,
|
5458 |
+
"step": 775
|
5459 |
+
},
|
5460 |
+
{
|
5461 |
+
"epoch": 0.1521717815472105,
|
5462 |
+
"grad_norm": 9.146486282348633,
|
5463 |
+
"learning_rate": 7.911027687880037e-05,
|
5464 |
+
"loss": 1.0215,
|
5465 |
+
"step": 776
|
5466 |
+
},
|
5467 |
+
{
|
5468 |
+
"epoch": 0.1523678792038435,
|
5469 |
+
"grad_norm": 5.029911518096924,
|
5470 |
+
"learning_rate": 7.888292560068226e-05,
|
5471 |
+
"loss": 0.6491,
|
5472 |
+
"step": 777
|
5473 |
+
},
|
5474 |
+
{
|
5475 |
+
"epoch": 0.1525639768604765,
|
5476 |
+
"grad_norm": 16.61541175842285,
|
5477 |
+
"learning_rate": 7.865568851112559e-05,
|
5478 |
+
"loss": 2.3993,
|
5479 |
+
"step": 778
|
5480 |
+
},
|
5481 |
+
{
|
5482 |
+
"epoch": 0.15276007451710952,
|
5483 |
+
"grad_norm": 8.573732376098633,
|
5484 |
+
"learning_rate": 7.842856683889321e-05,
|
5485 |
+
"loss": 1.5773,
|
5486 |
+
"step": 779
|
5487 |
+
},
|
5488 |
+
{
|
5489 |
+
"epoch": 0.15295617217374252,
|
5490 |
+
"grad_norm": 7.390541076660156,
|
5491 |
+
"learning_rate": 7.820156181212379e-05,
|
5492 |
+
"loss": 0.6461,
|
5493 |
+
"step": 780
|
5494 |
+
},
|
5495 |
+
{
|
5496 |
+
"epoch": 0.15315226983037553,
|
5497 |
+
"grad_norm": 4.455470085144043,
|
5498 |
+
"learning_rate": 7.79746746583254e-05,
|
5499 |
+
"loss": 1.9557,
|
5500 |
+
"step": 781
|
5501 |
+
},
|
5502 |
+
{
|
5503 |
+
"epoch": 0.15334836748700853,
|
5504 |
+
"grad_norm": 9.000947952270508,
|
5505 |
+
"learning_rate": 7.774790660436858e-05,
|
5506 |
+
"loss": 2.5368,
|
5507 |
+
"step": 782
|
5508 |
+
},
|
5509 |
+
{
|
5510 |
+
"epoch": 0.15354446514364153,
|
5511 |
+
"grad_norm": 7.608001232147217,
|
5512 |
+
"learning_rate": 7.752125887647992e-05,
|
5513 |
+
"loss": 1.607,
|
5514 |
+
"step": 783
|
5515 |
+
},
|
5516 |
+
{
|
5517 |
+
"epoch": 0.15374056280027454,
|
5518 |
+
"grad_norm": 4.594658851623535,
|
5519 |
+
"learning_rate": 7.729473270023537e-05,
|
5520 |
+
"loss": 0.9085,
|
5521 |
+
"step": 784
|
5522 |
+
},
|
5523 |
+
{
|
5524 |
+
"epoch": 0.15393666045690754,
|
5525 |
+
"grad_norm": 4.750558853149414,
|
5526 |
+
"learning_rate": 7.706832930055357e-05,
|
5527 |
+
"loss": 1.8871,
|
5528 |
+
"step": 785
|
5529 |
+
},
|
5530 |
+
{
|
5531 |
+
"epoch": 0.15413275811354055,
|
5532 |
+
"grad_norm": 6.5569987297058105,
|
5533 |
+
"learning_rate": 7.684204990168925e-05,
|
5534 |
+
"loss": 1.4386,
|
5535 |
+
"step": 786
|
5536 |
+
},
|
5537 |
+
{
|
5538 |
+
"epoch": 0.15432885577017355,
|
5539 |
+
"grad_norm": 4.655035495758057,
|
5540 |
+
"learning_rate": 7.66158957272266e-05,
|
5541 |
+
"loss": 1.9621,
|
5542 |
+
"step": 787
|
5543 |
+
},
|
5544 |
+
{
|
5545 |
+
"epoch": 0.15452495342680655,
|
5546 |
+
"grad_norm": 9.556004524230957,
|
5547 |
+
"learning_rate": 7.638986800007277e-05,
|
5548 |
+
"loss": 0.9825,
|
5549 |
+
"step": 788
|
5550 |
+
},
|
5551 |
+
{
|
5552 |
+
"epoch": 0.15472105108343956,
|
5553 |
+
"grad_norm": 4.108066082000732,
|
5554 |
+
"learning_rate": 7.616396794245107e-05,
|
5555 |
+
"loss": 2.0709,
|
5556 |
+
"step": 789
|
5557 |
+
},
|
5558 |
+
{
|
5559 |
+
"epoch": 0.15491714874007256,
|
5560 |
+
"grad_norm": 7.94503116607666,
|
5561 |
+
"learning_rate": 7.593819677589445e-05,
|
5562 |
+
"loss": 1.1699,
|
5563 |
+
"step": 790
|
5564 |
+
},
|
5565 |
+
{
|
5566 |
+
"epoch": 0.15511324639670557,
|
5567 |
+
"grad_norm": 6.220666408538818,
|
5568 |
+
"learning_rate": 7.571255572123889e-05,
|
5569 |
+
"loss": 1.1849,
|
5570 |
+
"step": 791
|
5571 |
+
},
|
5572 |
+
{
|
5573 |
+
"epoch": 0.15530934405333857,
|
5574 |
+
"grad_norm": 6.629292964935303,
|
5575 |
+
"learning_rate": 7.54870459986169e-05,
|
5576 |
+
"loss": 1.3573,
|
5577 |
+
"step": 792
|
5578 |
+
},
|
5579 |
+
{
|
5580 |
+
"epoch": 0.15550544170997158,
|
5581 |
+
"grad_norm": 4.624551296234131,
|
5582 |
+
"learning_rate": 7.52616688274507e-05,
|
5583 |
+
"loss": 1.1043,
|
5584 |
+
"step": 793
|
5585 |
+
},
|
5586 |
+
{
|
5587 |
+
"epoch": 0.15570153936660458,
|
5588 |
+
"grad_norm": 8.356186866760254,
|
5589 |
+
"learning_rate": 7.503642542644581e-05,
|
5590 |
+
"loss": 1.3448,
|
5591 |
+
"step": 794
|
5592 |
+
},
|
5593 |
+
{
|
5594 |
+
"epoch": 0.15589763702323758,
|
5595 |
+
"grad_norm": 5.954531192779541,
|
5596 |
+
"learning_rate": 7.481131701358434e-05,
|
5597 |
+
"loss": 1.2289,
|
5598 |
+
"step": 795
|
5599 |
+
},
|
5600 |
+
{
|
5601 |
+
"epoch": 0.1560937346798706,
|
5602 |
+
"grad_norm": 8.886476516723633,
|
5603 |
+
"learning_rate": 7.458634480611864e-05,
|
5604 |
+
"loss": 2.0398,
|
5605 |
+
"step": 796
|
5606 |
+
},
|
5607 |
+
{
|
5608 |
+
"epoch": 0.1562898323365036,
|
5609 |
+
"grad_norm": 5.7918524742126465,
|
5610 |
+
"learning_rate": 7.436151002056433e-05,
|
5611 |
+
"loss": 1.4364,
|
5612 |
+
"step": 797
|
5613 |
+
},
|
5614 |
+
{
|
5615 |
+
"epoch": 0.15648592999313657,
|
5616 |
+
"grad_norm": 9.434075355529785,
|
5617 |
+
"learning_rate": 7.413681387269406e-05,
|
5618 |
+
"loss": 1.8601,
|
5619 |
+
"step": 798
|
5620 |
+
},
|
5621 |
+
{
|
5622 |
+
"epoch": 0.15668202764976957,
|
5623 |
+
"grad_norm": 5.067231178283691,
|
5624 |
+
"learning_rate": 7.391225757753068e-05,
|
5625 |
+
"loss": 1.414,
|
5626 |
+
"step": 799
|
5627 |
+
},
|
5628 |
+
{
|
5629 |
+
"epoch": 0.15687812530640258,
|
5630 |
+
"grad_norm": 6.89833927154541,
|
5631 |
+
"learning_rate": 7.368784234934104e-05,
|
5632 |
+
"loss": 1.6815,
|
5633 |
+
"step": 800
|
5634 |
+
},
|
5635 |
+
{
|
5636 |
+
"epoch": 0.15707422296303558,
|
5637 |
+
"grad_norm": 15.612439155578613,
|
5638 |
+
"learning_rate": 7.346356940162895e-05,
|
5639 |
+
"loss": 3.0339,
|
5640 |
+
"step": 801
|
5641 |
+
},
|
5642 |
+
{
|
5643 |
+
"epoch": 0.15727032061966859,
|
5644 |
+
"grad_norm": 5.933378219604492,
|
5645 |
+
"learning_rate": 7.323943994712895e-05,
|
5646 |
+
"loss": 1.6531,
|
5647 |
+
"step": 802
|
5648 |
+
},
|
5649 |
+
{
|
5650 |
+
"epoch": 0.1574664182763016,
|
5651 |
+
"grad_norm": 4.7962727546691895,
|
5652 |
+
"learning_rate": 7.301545519779964e-05,
|
5653 |
+
"loss": 1.8519,
|
5654 |
+
"step": 803
|
5655 |
+
},
|
5656 |
+
{
|
5657 |
+
"epoch": 0.1576625159329346,
|
5658 |
+
"grad_norm": 7.490151882171631,
|
5659 |
+
"learning_rate": 7.279161636481715e-05,
|
5660 |
+
"loss": 1.429,
|
5661 |
+
"step": 804
|
5662 |
+
},
|
5663 |
+
{
|
5664 |
+
"epoch": 0.1578586135895676,
|
5665 |
+
"grad_norm": 10.799139976501465,
|
5666 |
+
"learning_rate": 7.256792465856858e-05,
|
5667 |
+
"loss": 1.391,
|
5668 |
+
"step": 805
|
5669 |
+
},
|
5670 |
+
{
|
5671 |
+
"epoch": 0.1580547112462006,
|
5672 |
+
"grad_norm": 4.731945514678955,
|
5673 |
+
"learning_rate": 7.23443812886454e-05,
|
5674 |
+
"loss": 1.3097,
|
5675 |
+
"step": 806
|
5676 |
+
},
|
5677 |
+
{
|
5678 |
+
"epoch": 0.1582508089028336,
|
5679 |
+
"grad_norm": 6.669147491455078,
|
5680 |
+
"learning_rate": 7.21209874638371e-05,
|
5681 |
+
"loss": 2.5443,
|
5682 |
+
"step": 807
|
5683 |
+
},
|
5684 |
+
{
|
5685 |
+
"epoch": 0.1584469065594666,
|
5686 |
+
"grad_norm": 7.046877861022949,
|
5687 |
+
"learning_rate": 7.189774439212442e-05,
|
5688 |
+
"loss": 1.5809,
|
5689 |
+
"step": 808
|
5690 |
+
},
|
5691 |
+
{
|
5692 |
+
"epoch": 0.15864300421609961,
|
5693 |
+
"grad_norm": 5.848972320556641,
|
5694 |
+
"learning_rate": 7.167465328067289e-05,
|
5695 |
+
"loss": 1.6708,
|
5696 |
+
"step": 809
|
5697 |
+
},
|
5698 |
+
{
|
5699 |
+
"epoch": 0.15883910187273262,
|
5700 |
+
"grad_norm": 6.136227607727051,
|
5701 |
+
"learning_rate": 7.145171533582636e-05,
|
5702 |
+
"loss": 1.2122,
|
5703 |
+
"step": 810
|
5704 |
+
},
|
5705 |
+
{
|
5706 |
+
"epoch": 0.15903519952936562,
|
5707 |
+
"grad_norm": 4.7808709144592285,
|
5708 |
+
"learning_rate": 7.122893176310052e-05,
|
5709 |
+
"loss": 1.8965,
|
5710 |
+
"step": 811
|
5711 |
+
},
|
5712 |
+
{
|
5713 |
+
"epoch": 0.15923129718599863,
|
5714 |
+
"grad_norm": 4.156777381896973,
|
5715 |
+
"learning_rate": 7.100630376717621e-05,
|
5716 |
+
"loss": 1.2742,
|
5717 |
+
"step": 812
|
5718 |
+
},
|
5719 |
+
{
|
5720 |
+
"epoch": 0.15942739484263163,
|
5721 |
+
"grad_norm": 5.69835090637207,
|
5722 |
+
"learning_rate": 7.078383255189304e-05,
|
5723 |
+
"loss": 0.7664,
|
5724 |
+
"step": 813
|
5725 |
+
},
|
5726 |
+
{
|
5727 |
+
"epoch": 0.15962349249926464,
|
5728 |
+
"grad_norm": 9.707277297973633,
|
5729 |
+
"learning_rate": 7.056151932024282e-05,
|
5730 |
+
"loss": 1.3492,
|
5731 |
+
"step": 814
|
5732 |
+
},
|
5733 |
+
{
|
5734 |
+
"epoch": 0.15981959015589764,
|
5735 |
+
"grad_norm": 10.071001052856445,
|
5736 |
+
"learning_rate": 7.033936527436318e-05,
|
5737 |
+
"loss": 1.2432,
|
5738 |
+
"step": 815
|
5739 |
+
},
|
5740 |
+
{
|
5741 |
+
"epoch": 0.16001568781253064,
|
5742 |
+
"grad_norm": 6.86775541305542,
|
5743 |
+
"learning_rate": 7.011737161553081e-05,
|
5744 |
+
"loss": 1.7583,
|
5745 |
+
"step": 816
|
5746 |
+
},
|
5747 |
+
{
|
5748 |
+
"epoch": 0.16021178546916365,
|
5749 |
+
"grad_norm": 5.526086330413818,
|
5750 |
+
"learning_rate": 6.989553954415524e-05,
|
5751 |
+
"loss": 1.5891,
|
5752 |
+
"step": 817
|
5753 |
+
},
|
5754 |
+
{
|
5755 |
+
"epoch": 0.16040788312579665,
|
5756 |
+
"grad_norm": 26.293426513671875,
|
5757 |
+
"learning_rate": 6.967387025977215e-05,
|
5758 |
+
"loss": 1.7835,
|
5759 |
+
"step": 818
|
5760 |
+
},
|
5761 |
+
{
|
5762 |
+
"epoch": 0.16060398078242966,
|
5763 |
+
"grad_norm": 5.713183879852295,
|
5764 |
+
"learning_rate": 6.945236496103707e-05,
|
5765 |
+
"loss": 1.7831,
|
5766 |
+
"step": 819
|
5767 |
+
},
|
5768 |
+
{
|
5769 |
+
"epoch": 0.16080007843906266,
|
5770 |
+
"grad_norm": 7.669525146484375,
|
5771 |
+
"learning_rate": 6.92310248457187e-05,
|
5772 |
+
"loss": 1.8961,
|
5773 |
+
"step": 820
|
5774 |
+
},
|
5775 |
+
{
|
5776 |
+
"epoch": 0.16099617609569566,
|
5777 |
+
"grad_norm": 7.351653099060059,
|
5778 |
+
"learning_rate": 6.900985111069256e-05,
|
5779 |
+
"loss": 1.8533,
|
5780 |
+
"step": 821
|
5781 |
+
},
|
5782 |
+
{
|
5783 |
+
"epoch": 0.16119227375232867,
|
5784 |
+
"grad_norm": 6.009294509887695,
|
5785 |
+
"learning_rate": 6.878884495193448e-05,
|
5786 |
+
"loss": 1.7683,
|
5787 |
+
"step": 822
|
5788 |
+
},
|
5789 |
+
{
|
5790 |
+
"epoch": 0.16138837140896167,
|
5791 |
+
"grad_norm": 5.90309476852417,
|
5792 |
+
"learning_rate": 6.856800756451414e-05,
|
5793 |
+
"loss": 1.7174,
|
5794 |
+
"step": 823
|
5795 |
+
},
|
5796 |
+
{
|
5797 |
+
"epoch": 0.16158446906559468,
|
5798 |
+
"grad_norm": 6.903116703033447,
|
5799 |
+
"learning_rate": 6.83473401425886e-05,
|
5800 |
+
"loss": 1.6063,
|
5801 |
+
"step": 824
|
5802 |
+
},
|
5803 |
+
{
|
5804 |
+
"epoch": 0.16178056672222768,
|
5805 |
+
"grad_norm": 5.747986793518066,
|
5806 |
+
"learning_rate": 6.812684387939585e-05,
|
5807 |
+
"loss": 1.7352,
|
5808 |
+
"step": 825
|
5809 |
+
},
|
5810 |
+
{
|
5811 |
+
"epoch": 0.16197666437886069,
|
5812 |
+
"grad_norm": 4.83687686920166,
|
5813 |
+
"learning_rate": 6.79065199672484e-05,
|
5814 |
+
"loss": 1.0662,
|
5815 |
+
"step": 826
|
5816 |
+
},
|
5817 |
+
{
|
5818 |
+
"epoch": 0.1621727620354937,
|
5819 |
+
"grad_norm": 4.761837482452393,
|
5820 |
+
"learning_rate": 6.768636959752669e-05,
|
5821 |
+
"loss": 1.4324,
|
5822 |
+
"step": 827
|
5823 |
+
},
|
5824 |
+
{
|
5825 |
+
"epoch": 0.16236885969212667,
|
5826 |
+
"grad_norm": 4.991310119628906,
|
5827 |
+
"learning_rate": 6.746639396067283e-05,
|
5828 |
+
"loss": 0.6861,
|
5829 |
+
"step": 828
|
5830 |
+
},
|
5831 |
+
{
|
5832 |
+
"epoch": 0.16256495734875967,
|
5833 |
+
"grad_norm": 9.824933052062988,
|
5834 |
+
"learning_rate": 6.724659424618401e-05,
|
5835 |
+
"loss": 1.2527,
|
5836 |
+
"step": 829
|
5837 |
+
},
|
5838 |
+
{
|
5839 |
+
"epoch": 0.16276105500539267,
|
5840 |
+
"grad_norm": 9.043631553649902,
|
5841 |
+
"learning_rate": 6.702697164260626e-05,
|
5842 |
+
"loss": 1.2587,
|
5843 |
+
"step": 830
|
5844 |
+
},
|
5845 |
+
{
|
5846 |
+
"epoch": 0.16295715266202568,
|
5847 |
+
"grad_norm": 6.001374244689941,
|
5848 |
+
"learning_rate": 6.680752733752777e-05,
|
5849 |
+
"loss": 1.0727,
|
5850 |
+
"step": 831
|
5851 |
+
},
|
5852 |
+
{
|
5853 |
+
"epoch": 0.16315325031865868,
|
5854 |
+
"grad_norm": 7.015263557434082,
|
5855 |
+
"learning_rate": 6.658826251757264e-05,
|
5856 |
+
"loss": 0.963,
|
5857 |
+
"step": 832
|
5858 |
+
},
|
5859 |
+
{
|
5860 |
+
"epoch": 0.1633493479752917,
|
5861 |
+
"grad_norm": 16.830978393554688,
|
5862 |
+
"learning_rate": 6.63691783683944e-05,
|
5863 |
+
"loss": 1.3392,
|
5864 |
+
"step": 833
|
5865 |
+
},
|
5866 |
+
{
|
5867 |
+
"epoch": 0.1635454456319247,
|
5868 |
+
"grad_norm": 12.247193336486816,
|
5869 |
+
"learning_rate": 6.61502760746697e-05,
|
5870 |
+
"loss": 2.2754,
|
5871 |
+
"step": 834
|
5872 |
+
},
|
5873 |
+
{
|
5874 |
+
"epoch": 0.1637415432885577,
|
5875 |
+
"grad_norm": 3.836216449737549,
|
5876 |
+
"learning_rate": 6.593155682009176e-05,
|
5877 |
+
"loss": 2.4028,
|
5878 |
+
"step": 835
|
5879 |
+
},
|
5880 |
+
{
|
5881 |
+
"epoch": 0.1639376409451907,
|
5882 |
+
"grad_norm": 5.886529445648193,
|
5883 |
+
"learning_rate": 6.571302178736404e-05,
|
5884 |
+
"loss": 1.0402,
|
5885 |
+
"step": 836
|
5886 |
+
},
|
5887 |
+
{
|
5888 |
+
"epoch": 0.1641337386018237,
|
5889 |
+
"grad_norm": 5.377973556518555,
|
5890 |
+
"learning_rate": 6.549467215819378e-05,
|
5891 |
+
"loss": 1.4784,
|
5892 |
+
"step": 837
|
5893 |
+
},
|
5894 |
+
{
|
5895 |
+
"epoch": 0.1643298362584567,
|
5896 |
+
"grad_norm": 4.867081165313721,
|
5897 |
+
"learning_rate": 6.527650911328585e-05,
|
5898 |
+
"loss": 1.9024,
|
5899 |
+
"step": 838
|
5900 |
+
},
|
5901 |
+
{
|
5902 |
+
"epoch": 0.1645259339150897,
|
5903 |
+
"grad_norm": 8.56846809387207,
|
5904 |
+
"learning_rate": 6.505853383233596e-05,
|
5905 |
+
"loss": 1.2514,
|
5906 |
+
"step": 839
|
5907 |
+
},
|
5908 |
+
{
|
5909 |
+
"epoch": 0.16472203157172272,
|
5910 |
+
"grad_norm": 14.424854278564453,
|
5911 |
+
"learning_rate": 6.484074749402467e-05,
|
5912 |
+
"loss": 1.2373,
|
5913 |
+
"step": 840
|
5914 |
+
},
|
5915 |
+
{
|
5916 |
+
"epoch": 0.16491812922835572,
|
5917 |
+
"grad_norm": 5.4463067054748535,
|
5918 |
+
"learning_rate": 6.462315127601071e-05,
|
5919 |
+
"loss": 1.5188,
|
5920 |
+
"step": 841
|
5921 |
+
},
|
5922 |
+
{
|
5923 |
+
"epoch": 0.16511422688498872,
|
5924 |
+
"grad_norm": 5.752686500549316,
|
5925 |
+
"learning_rate": 6.440574635492493e-05,
|
5926 |
+
"loss": 1.0012,
|
5927 |
+
"step": 842
|
5928 |
+
},
|
5929 |
+
{
|
5930 |
+
"epoch": 0.16531032454162173,
|
5931 |
+
"grad_norm": 5.5835442543029785,
|
5932 |
+
"learning_rate": 6.418853390636364e-05,
|
5933 |
+
"loss": 1.8797,
|
5934 |
+
"step": 843
|
5935 |
+
},
|
5936 |
+
{
|
5937 |
+
"epoch": 0.16550642219825473,
|
5938 |
+
"grad_norm": 8.692926406860352,
|
5939 |
+
"learning_rate": 6.39715151048823e-05,
|
5940 |
+
"loss": 1.2265,
|
5941 |
+
"step": 844
|
5942 |
+
},
|
5943 |
+
{
|
5944 |
+
"epoch": 0.16570251985488774,
|
5945 |
+
"grad_norm": 5.524386882781982,
|
5946 |
+
"learning_rate": 6.375469112398946e-05,
|
5947 |
+
"loss": 0.7208,
|
5948 |
+
"step": 845
|
5949 |
+
},
|
5950 |
+
{
|
5951 |
+
"epoch": 0.16589861751152074,
|
5952 |
+
"grad_norm": 5.7656145095825195,
|
5953 |
+
"learning_rate": 6.353806313614001e-05,
|
5954 |
+
"loss": 1.2716,
|
5955 |
+
"step": 846
|
5956 |
+
},
|
5957 |
+
{
|
5958 |
+
"epoch": 0.16609471516815374,
|
5959 |
+
"grad_norm": 12.103293418884277,
|
5960 |
+
"learning_rate": 6.332163231272911e-05,
|
5961 |
+
"loss": 3.2642,
|
5962 |
+
"step": 847
|
5963 |
+
},
|
5964 |
+
{
|
5965 |
+
"epoch": 0.16629081282478675,
|
5966 |
+
"grad_norm": 22.349411010742188,
|
5967 |
+
"learning_rate": 6.310539982408568e-05,
|
5968 |
+
"loss": 2.3537,
|
5969 |
+
"step": 848
|
5970 |
+
},
|
5971 |
+
{
|
5972 |
+
"epoch": 0.16648691048141975,
|
5973 |
+
"grad_norm": 7.782928466796875,
|
5974 |
+
"learning_rate": 6.288936683946633e-05,
|
5975 |
+
"loss": 1.1905,
|
5976 |
+
"step": 849
|
5977 |
+
},
|
5978 |
+
{
|
5979 |
+
"epoch": 0.16668300813805276,
|
5980 |
+
"grad_norm": 6.026412487030029,
|
5981 |
+
"learning_rate": 6.267353452704876e-05,
|
5982 |
+
"loss": 1.1583,
|
5983 |
+
"step": 850
|
5984 |
+
},
|
5985 |
+
{
|
5986 |
+
"epoch": 0.16687910579468576,
|
5987 |
+
"grad_norm": 10.18653392791748,
|
5988 |
+
"learning_rate": 6.245790405392553e-05,
|
5989 |
+
"loss": 2.5888,
|
5990 |
+
"step": 851
|
5991 |
+
},
|
5992 |
+
{
|
5993 |
+
"epoch": 0.16707520345131877,
|
5994 |
+
"grad_norm": 6.149682998657227,
|
5995 |
+
"learning_rate": 6.224247658609781e-05,
|
5996 |
+
"loss": 1.0514,
|
5997 |
+
"step": 852
|
5998 |
+
},
|
5999 |
+
{
|
6000 |
+
"epoch": 0.16727130110795177,
|
6001 |
+
"grad_norm": 5.528243541717529,
|
6002 |
+
"learning_rate": 6.202725328846909e-05,
|
6003 |
+
"loss": 0.8576,
|
6004 |
+
"step": 853
|
6005 |
+
},
|
6006 |
+
{
|
6007 |
+
"epoch": 0.16746739876458477,
|
6008 |
+
"grad_norm": 7.682480812072754,
|
6009 |
+
"learning_rate": 6.181223532483875e-05,
|
6010 |
+
"loss": 1.5959,
|
6011 |
+
"step": 854
|
6012 |
+
},
|
6013 |
+
{
|
6014 |
+
"epoch": 0.16766349642121778,
|
6015 |
+
"grad_norm": 4.653623580932617,
|
6016 |
+
"learning_rate": 6.159742385789588e-05,
|
6017 |
+
"loss": 0.7513,
|
6018 |
+
"step": 855
|
6019 |
+
},
|
6020 |
+
{
|
6021 |
+
"epoch": 0.16785959407785078,
|
6022 |
+
"grad_norm": 18.469745635986328,
|
6023 |
+
"learning_rate": 6.138282004921295e-05,
|
6024 |
+
"loss": 3.1673,
|
6025 |
+
"step": 856
|
6026 |
+
},
|
6027 |
+
{
|
6028 |
+
"epoch": 0.1680556917344838,
|
6029 |
+
"grad_norm": 4.403334140777588,
|
6030 |
+
"learning_rate": 6.116842505923955e-05,
|
6031 |
+
"loss": 0.9741,
|
6032 |
+
"step": 857
|
6033 |
+
},
|
6034 |
+
{
|
6035 |
+
"epoch": 0.16825178939111676,
|
6036 |
+
"grad_norm": 5.073941230773926,
|
6037 |
+
"learning_rate": 6.095424004729611e-05,
|
6038 |
+
"loss": 0.8638,
|
6039 |
+
"step": 858
|
6040 |
+
},
|
6041 |
+
{
|
6042 |
+
"epoch": 0.16844788704774977,
|
6043 |
+
"grad_norm": 6.09031867980957,
|
6044 |
+
"learning_rate": 6.0740266171567606e-05,
|
6045 |
+
"loss": 0.7445,
|
6046 |
+
"step": 859
|
6047 |
+
},
|
6048 |
+
{
|
6049 |
+
"epoch": 0.16864398470438277,
|
6050 |
+
"grad_norm": 5.235655784606934,
|
6051 |
+
"learning_rate": 6.0526504589097254e-05,
|
6052 |
+
"loss": 1.79,
|
6053 |
+
"step": 860
|
6054 |
+
},
|
6055 |
+
{
|
6056 |
+
"epoch": 0.16884008236101578,
|
6057 |
+
"grad_norm": 5.974957466125488,
|
6058 |
+
"learning_rate": 6.031295645578049e-05,
|
6059 |
+
"loss": 2.638,
|
6060 |
+
"step": 861
|
6061 |
+
},
|
6062 |
+
{
|
6063 |
+
"epoch": 0.16903618001764878,
|
6064 |
+
"grad_norm": 4.034956932067871,
|
6065 |
+
"learning_rate": 6.009962292635838e-05,
|
6066 |
+
"loss": 1.2214,
|
6067 |
+
"step": 862
|
6068 |
+
},
|
6069 |
+
{
|
6070 |
+
"epoch": 0.16923227767428178,
|
6071 |
+
"grad_norm": 5.100616931915283,
|
6072 |
+
"learning_rate": 5.988650515441156e-05,
|
6073 |
+
"loss": 1.7694,
|
6074 |
+
"step": 863
|
6075 |
+
},
|
6076 |
+
{
|
6077 |
+
"epoch": 0.1694283753309148,
|
6078 |
+
"grad_norm": 6.878417491912842,
|
6079 |
+
"learning_rate": 5.967360429235407e-05,
|
6080 |
+
"loss": 1.2678,
|
6081 |
+
"step": 864
|
6082 |
+
},
|
6083 |
+
{
|
6084 |
+
"epoch": 0.1696244729875478,
|
6085 |
+
"grad_norm": 6.435969352722168,
|
6086 |
+
"learning_rate": 5.946092149142692e-05,
|
6087 |
+
"loss": 1.449,
|
6088 |
+
"step": 865
|
6089 |
+
},
|
6090 |
+
{
|
6091 |
+
"epoch": 0.1698205706441808,
|
6092 |
+
"grad_norm": 5.022701740264893,
|
6093 |
+
"learning_rate": 5.924845790169209e-05,
|
6094 |
+
"loss": 0.8444,
|
6095 |
+
"step": 866
|
6096 |
+
},
|
6097 |
+
{
|
6098 |
+
"epoch": 0.1700166683008138,
|
6099 |
+
"grad_norm": 4.173498153686523,
|
6100 |
+
"learning_rate": 5.903621467202608e-05,
|
6101 |
+
"loss": 1.089,
|
6102 |
+
"step": 867
|
6103 |
+
},
|
6104 |
+
{
|
6105 |
+
"epoch": 0.1702127659574468,
|
6106 |
+
"grad_norm": 6.396774768829346,
|
6107 |
+
"learning_rate": 5.88241929501139e-05,
|
6108 |
+
"loss": 1.1751,
|
6109 |
+
"step": 868
|
6110 |
+
},
|
6111 |
+
{
|
6112 |
+
"epoch": 0.1704088636140798,
|
6113 |
+
"grad_norm": 4.628440856933594,
|
6114 |
+
"learning_rate": 5.861239388244273e-05,
|
6115 |
+
"loss": 1.4271,
|
6116 |
+
"step": 869
|
6117 |
+
},
|
6118 |
+
{
|
6119 |
+
"epoch": 0.1706049612707128,
|
6120 |
+
"grad_norm": 13.596205711364746,
|
6121 |
+
"learning_rate": 5.8400818614295806e-05,
|
6122 |
+
"loss": 0.7325,
|
6123 |
+
"step": 870
|
6124 |
+
},
|
6125 |
+
{
|
6126 |
+
"epoch": 0.17080105892734582,
|
6127 |
+
"grad_norm": 8.975894927978516,
|
6128 |
+
"learning_rate": 5.818946828974607e-05,
|
6129 |
+
"loss": 1.8689,
|
6130 |
+
"step": 871
|
6131 |
+
},
|
6132 |
+
{
|
6133 |
+
"epoch": 0.17099715658397882,
|
6134 |
+
"grad_norm": 5.900057315826416,
|
6135 |
+
"learning_rate": 5.797834405165035e-05,
|
6136 |
+
"loss": 0.8738,
|
6137 |
+
"step": 872
|
6138 |
+
},
|
6139 |
+
{
|
6140 |
+
"epoch": 0.17119325424061183,
|
6141 |
+
"grad_norm": 6.28643798828125,
|
6142 |
+
"learning_rate": 5.776744704164272e-05,
|
6143 |
+
"loss": 2.4116,
|
6144 |
+
"step": 873
|
6145 |
+
},
|
6146 |
+
{
|
6147 |
+
"epoch": 0.17138935189724483,
|
6148 |
+
"grad_norm": 6.7851033210754395,
|
6149 |
+
"learning_rate": 5.7556778400128583e-05,
|
6150 |
+
"loss": 1.5964,
|
6151 |
+
"step": 874
|
6152 |
+
},
|
6153 |
+
{
|
6154 |
+
"epoch": 0.17158544955387783,
|
6155 |
+
"grad_norm": 12.063843727111816,
|
6156 |
+
"learning_rate": 5.7346339266278505e-05,
|
6157 |
+
"loss": 1.9227,
|
6158 |
+
"step": 875
|
6159 |
+
},
|
6160 |
+
{
|
6161 |
+
"epoch": 0.17178154721051084,
|
6162 |
+
"grad_norm": 6.9710469245910645,
|
6163 |
+
"learning_rate": 5.713613077802199e-05,
|
6164 |
+
"loss": 1.3143,
|
6165 |
+
"step": 876
|
6166 |
+
},
|
6167 |
+
{
|
6168 |
+
"epoch": 0.17197764486714384,
|
6169 |
+
"grad_norm": 10.355106353759766,
|
6170 |
+
"learning_rate": 5.6926154072041295e-05,
|
6171 |
+
"loss": 2.005,
|
6172 |
+
"step": 877
|
6173 |
+
},
|
6174 |
+
{
|
6175 |
+
"epoch": 0.17217374252377685,
|
6176 |
+
"grad_norm": 8.473671913146973,
|
6177 |
+
"learning_rate": 5.671641028376546e-05,
|
6178 |
+
"loss": 1.4054,
|
6179 |
+
"step": 878
|
6180 |
+
},
|
6181 |
+
{
|
6182 |
+
"epoch": 0.17236984018040985,
|
6183 |
+
"grad_norm": 6.026031494140625,
|
6184 |
+
"learning_rate": 5.650690054736396e-05,
|
6185 |
+
"loss": 0.9981,
|
6186 |
+
"step": 879
|
6187 |
+
},
|
6188 |
+
{
|
6189 |
+
"epoch": 0.17256593783704285,
|
6190 |
+
"grad_norm": 7.832981109619141,
|
6191 |
+
"learning_rate": 5.629762599574064e-05,
|
6192 |
+
"loss": 1.318,
|
6193 |
+
"step": 880
|
6194 |
+
},
|
6195 |
+
{
|
6196 |
+
"epoch": 0.17276203549367586,
|
6197 |
+
"grad_norm": 8.796941757202148,
|
6198 |
+
"learning_rate": 5.6088587760527656e-05,
|
6199 |
+
"loss": 1.7696,
|
6200 |
+
"step": 881
|
6201 |
+
},
|
6202 |
+
{
|
6203 |
+
"epoch": 0.17295813315030886,
|
6204 |
+
"grad_norm": 5.9997124671936035,
|
6205 |
+
"learning_rate": 5.587978697207927e-05,
|
6206 |
+
"loss": 1.5515,
|
6207 |
+
"step": 882
|
6208 |
+
},
|
6209 |
+
{
|
6210 |
+
"epoch": 0.17315423080694187,
|
6211 |
+
"grad_norm": 10.373340606689453,
|
6212 |
+
"learning_rate": 5.5671224759465775e-05,
|
6213 |
+
"loss": 1.735,
|
6214 |
+
"step": 883
|
6215 |
+
},
|
6216 |
+
{
|
6217 |
+
"epoch": 0.17335032846357487,
|
6218 |
+
"grad_norm": 6.619256973266602,
|
6219 |
+
"learning_rate": 5.546290225046744e-05,
|
6220 |
+
"loss": 1.3402,
|
6221 |
+
"step": 884
|
6222 |
+
},
|
6223 |
+
{
|
6224 |
+
"epoch": 0.17354642612020788,
|
6225 |
+
"grad_norm": 7.469152927398682,
|
6226 |
+
"learning_rate": 5.5254820571568325e-05,
|
6227 |
+
"loss": 1.0792,
|
6228 |
+
"step": 885
|
6229 |
+
},
|
6230 |
+
{
|
6231 |
+
"epoch": 0.17374252377684088,
|
6232 |
+
"grad_norm": 4.927844524383545,
|
6233 |
+
"learning_rate": 5.504698084795014e-05,
|
6234 |
+
"loss": 1.4485,
|
6235 |
+
"step": 886
|
6236 |
+
},
|
6237 |
+
{
|
6238 |
+
"epoch": 0.17393862143347388,
|
6239 |
+
"grad_norm": 4.025864601135254,
|
6240 |
+
"learning_rate": 5.483938420348642e-05,
|
6241 |
+
"loss": 0.9515,
|
6242 |
+
"step": 887
|
6243 |
+
},
|
6244 |
+
{
|
6245 |
+
"epoch": 0.17413471909010686,
|
6246 |
+
"grad_norm": 14.059304237365723,
|
6247 |
+
"learning_rate": 5.4632031760736146e-05,
|
6248 |
+
"loss": 1.9276,
|
6249 |
+
"step": 888
|
6250 |
+
},
|
6251 |
+
{
|
6252 |
+
"epoch": 0.17433081674673986,
|
6253 |
+
"grad_norm": 6.336188793182373,
|
6254 |
+
"learning_rate": 5.4424924640937824e-05,
|
6255 |
+
"loss": 1.7752,
|
6256 |
+
"step": 889
|
6257 |
+
},
|
6258 |
+
{
|
6259 |
+
"epoch": 0.17452691440337287,
|
6260 |
+
"grad_norm": 7.820261478424072,
|
6261 |
+
"learning_rate": 5.421806396400339e-05,
|
6262 |
+
"loss": 1.4182,
|
6263 |
+
"step": 890
|
6264 |
+
},
|
6265 |
+
{
|
6266 |
+
"epoch": 0.17472301206000587,
|
6267 |
+
"grad_norm": 3.8585610389709473,
|
6268 |
+
"learning_rate": 5.4011450848512225e-05,
|
6269 |
+
"loss": 1.1074,
|
6270 |
+
"step": 891
|
6271 |
+
},
|
6272 |
+
{
|
6273 |
+
"epoch": 0.17491910971663888,
|
6274 |
+
"grad_norm": 5.7243123054504395,
|
6275 |
+
"learning_rate": 5.3805086411704985e-05,
|
6276 |
+
"loss": 2.2836,
|
6277 |
+
"step": 892
|
6278 |
+
},
|
6279 |
+
{
|
6280 |
+
"epoch": 0.17511520737327188,
|
6281 |
+
"grad_norm": 3.6346094608306885,
|
6282 |
+
"learning_rate": 5.3598971769477655e-05,
|
6283 |
+
"loss": 1.2185,
|
6284 |
+
"step": 893
|
6285 |
+
},
|
6286 |
+
{
|
6287 |
+
"epoch": 0.17531130502990488,
|
6288 |
+
"grad_norm": 5.283199310302734,
|
6289 |
+
"learning_rate": 5.339310803637538e-05,
|
6290 |
+
"loss": 1.0693,
|
6291 |
+
"step": 894
|
6292 |
+
},
|
6293 |
+
{
|
6294 |
+
"epoch": 0.1755074026865379,
|
6295 |
+
"grad_norm": 5.19635009765625,
|
6296 |
+
"learning_rate": 5.318749632558674e-05,
|
6297 |
+
"loss": 1.3408,
|
6298 |
+
"step": 895
|
6299 |
+
},
|
6300 |
+
{
|
6301 |
+
"epoch": 0.1757035003431709,
|
6302 |
+
"grad_norm": 7.379054546356201,
|
6303 |
+
"learning_rate": 5.298213774893735e-05,
|
6304 |
+
"loss": 1.6441,
|
6305 |
+
"step": 896
|
6306 |
+
},
|
6307 |
+
{
|
6308 |
+
"epoch": 0.1758995979998039,
|
6309 |
+
"grad_norm": 10.650435447692871,
|
6310 |
+
"learning_rate": 5.277703341688407e-05,
|
6311 |
+
"loss": 1.1186,
|
6312 |
+
"step": 897
|
6313 |
+
},
|
6314 |
+
{
|
6315 |
+
"epoch": 0.1760956956564369,
|
6316 |
+
"grad_norm": 4.219560623168945,
|
6317 |
+
"learning_rate": 5.257218443850895e-05,
|
6318 |
+
"loss": 1.1465,
|
6319 |
+
"step": 898
|
6320 |
+
},
|
6321 |
+
{
|
6322 |
+
"epoch": 0.1762917933130699,
|
6323 |
+
"grad_norm": 9.091025352478027,
|
6324 |
+
"learning_rate": 5.236759192151336e-05,
|
6325 |
+
"loss": 2.0643,
|
6326 |
+
"step": 899
|
6327 |
+
},
|
6328 |
+
{
|
6329 |
+
"epoch": 0.1764878909697029,
|
6330 |
+
"grad_norm": 6.650880336761475,
|
6331 |
+
"learning_rate": 5.2163256972211714e-05,
|
6332 |
+
"loss": 1.6255,
|
6333 |
+
"step": 900
|
6334 |
+
},
|
6335 |
+
{
|
6336 |
+
"epoch": 0.1766839886263359,
|
6337 |
+
"grad_norm": 5.008049011230469,
|
6338 |
+
"learning_rate": 5.195918069552572e-05,
|
6339 |
+
"loss": 1.0065,
|
6340 |
+
"step": 901
|
6341 |
+
},
|
6342 |
+
{
|
6343 |
+
"epoch": 0.17688008628296892,
|
6344 |
+
"grad_norm": 6.407417297363281,
|
6345 |
+
"learning_rate": 5.175536419497831e-05,
|
6346 |
+
"loss": 0.7189,
|
6347 |
+
"step": 902
|
6348 |
+
},
|
6349 |
+
{
|
6350 |
+
"epoch": 0.17707618393960192,
|
6351 |
+
"grad_norm": 4.843297004699707,
|
6352 |
+
"learning_rate": 5.155180857268779e-05,
|
6353 |
+
"loss": 1.5639,
|
6354 |
+
"step": 903
|
6355 |
+
},
|
6356 |
+
{
|
6357 |
+
"epoch": 0.17727228159623493,
|
6358 |
+
"grad_norm": 3.7099854946136475,
|
6359 |
+
"learning_rate": 5.1348514929361714e-05,
|
6360 |
+
"loss": 1.0337,
|
6361 |
+
"step": 904
|
6362 |
+
},
|
6363 |
+
{
|
6364 |
+
"epoch": 0.17746837925286793,
|
6365 |
+
"grad_norm": 7.940112590789795,
|
6366 |
+
"learning_rate": 5.114548436429098e-05,
|
6367 |
+
"loss": 1.9378,
|
6368 |
+
"step": 905
|
6369 |
+
},
|
6370 |
+
{
|
6371 |
+
"epoch": 0.17766447690950093,
|
6372 |
+
"grad_norm": 6.462250232696533,
|
6373 |
+
"learning_rate": 5.0942717975344035e-05,
|
6374 |
+
"loss": 1.2462,
|
6375 |
+
"step": 906
|
6376 |
+
},
|
6377 |
+
{
|
6378 |
+
"epoch": 0.17786057456613394,
|
6379 |
+
"grad_norm": 3.7596821784973145,
|
6380 |
+
"learning_rate": 5.074021685896072e-05,
|
6381 |
+
"loss": 0.5025,
|
6382 |
+
"step": 907
|
6383 |
+
},
|
6384 |
+
{
|
6385 |
+
"epoch": 0.17805667222276694,
|
6386 |
+
"grad_norm": 5.532608509063721,
|
6387 |
+
"learning_rate": 5.0537982110146464e-05,
|
6388 |
+
"loss": 1.2263,
|
6389 |
+
"step": 908
|
6390 |
+
},
|
6391 |
+
{
|
6392 |
+
"epoch": 0.17825276987939995,
|
6393 |
+
"grad_norm": 6.286501884460449,
|
6394 |
+
"learning_rate": 5.033601482246628e-05,
|
6395 |
+
"loss": 0.9899,
|
6396 |
+
"step": 909
|
6397 |
+
},
|
6398 |
+
{
|
6399 |
+
"epoch": 0.17844886753603295,
|
6400 |
+
"grad_norm": 8.714296340942383,
|
6401 |
+
"learning_rate": 5.013431608803909e-05,
|
6402 |
+
"loss": 1.0626,
|
6403 |
+
"step": 910
|
6404 |
+
},
|
6405 |
+
{
|
6406 |
+
"epoch": 0.17864496519266596,
|
6407 |
+
"grad_norm": 6.827104568481445,
|
6408 |
+
"learning_rate": 4.993288699753146e-05,
|
6409 |
+
"loss": 1.1741,
|
6410 |
+
"step": 911
|
6411 |
+
},
|
6412 |
+
{
|
6413 |
+
"epoch": 0.17884106284929896,
|
6414 |
+
"grad_norm": 7.155247688293457,
|
6415 |
+
"learning_rate": 4.973172864015193e-05,
|
6416 |
+
"loss": 1.9644,
|
6417 |
+
"step": 912
|
6418 |
+
},
|
6419 |
+
{
|
6420 |
+
"epoch": 0.17903716050593196,
|
6421 |
+
"grad_norm": 4.7466139793396,
|
6422 |
+
"learning_rate": 4.953084210364508e-05,
|
6423 |
+
"loss": 1.2649,
|
6424 |
+
"step": 913
|
6425 |
+
},
|
6426 |
+
{
|
6427 |
+
"epoch": 0.17923325816256497,
|
6428 |
+
"grad_norm": 6.292973041534424,
|
6429 |
+
"learning_rate": 4.933022847428571e-05,
|
6430 |
+
"loss": 2.3925,
|
6431 |
+
"step": 914
|
6432 |
+
},
|
6433 |
+
{
|
6434 |
+
"epoch": 0.17942935581919797,
|
6435 |
+
"grad_norm": 6.9081950187683105,
|
6436 |
+
"learning_rate": 4.912988883687283e-05,
|
6437 |
+
"loss": 2.7031,
|
6438 |
+
"step": 915
|
6439 |
+
},
|
6440 |
+
{
|
6441 |
+
"epoch": 0.17962545347583098,
|
6442 |
+
"grad_norm": 10.206450462341309,
|
6443 |
+
"learning_rate": 4.892982427472387e-05,
|
6444 |
+
"loss": 1.2479,
|
6445 |
+
"step": 916
|
6446 |
+
},
|
6447 |
+
{
|
6448 |
+
"epoch": 0.17982155113246398,
|
6449 |
+
"grad_norm": 9.753491401672363,
|
6450 |
+
"learning_rate": 4.873003586966881e-05,
|
6451 |
+
"loss": 1.1683,
|
6452 |
+
"step": 917
|
6453 |
+
},
|
6454 |
+
{
|
6455 |
+
"epoch": 0.18001764878909696,
|
6456 |
+
"grad_norm": 5.422173023223877,
|
6457 |
+
"learning_rate": 4.853052470204438e-05,
|
6458 |
+
"loss": 1.0117,
|
6459 |
+
"step": 918
|
6460 |
+
},
|
6461 |
+
{
|
6462 |
+
"epoch": 0.18021374644572996,
|
6463 |
+
"grad_norm": 4.505789756774902,
|
6464 |
+
"learning_rate": 4.8331291850688096e-05,
|
6465 |
+
"loss": 0.7031,
|
6466 |
+
"step": 919
|
6467 |
+
},
|
6468 |
+
{
|
6469 |
+
"epoch": 0.18040984410236297,
|
6470 |
+
"grad_norm": 6.756499290466309,
|
6471 |
+
"learning_rate": 4.813233839293265e-05,
|
6472 |
+
"loss": 1.5295,
|
6473 |
+
"step": 920
|
6474 |
+
},
|
6475 |
+
{
|
6476 |
+
"epoch": 0.18060594175899597,
|
6477 |
+
"grad_norm": 7.535793781280518,
|
6478 |
+
"learning_rate": 4.7933665404599804e-05,
|
6479 |
+
"loss": 1.2856,
|
6480 |
+
"step": 921
|
6481 |
+
},
|
6482 |
+
{
|
6483 |
+
"epoch": 0.18080203941562897,
|
6484 |
+
"grad_norm": 8.04592227935791,
|
6485 |
+
"learning_rate": 4.773527395999474e-05,
|
6486 |
+
"loss": 0.9891,
|
6487 |
+
"step": 922
|
6488 |
+
},
|
6489 |
+
{
|
6490 |
+
"epoch": 0.18099813707226198,
|
6491 |
+
"grad_norm": 5.000694274902344,
|
6492 |
+
"learning_rate": 4.753716513190024e-05,
|
6493 |
+
"loss": 1.3128,
|
6494 |
+
"step": 923
|
6495 |
+
},
|
6496 |
+
{
|
6497 |
+
"epoch": 0.18119423472889498,
|
6498 |
+
"grad_norm": 8.065174102783203,
|
6499 |
+
"learning_rate": 4.7339339991570876e-05,
|
6500 |
+
"loss": 1.9393,
|
6501 |
+
"step": 924
|
6502 |
+
},
|
6503 |
+
{
|
6504 |
+
"epoch": 0.18139033238552799,
|
6505 |
+
"grad_norm": 6.1632256507873535,
|
6506 |
+
"learning_rate": 4.714179960872712e-05,
|
6507 |
+
"loss": 1.7183,
|
6508 |
+
"step": 925
|
6509 |
+
},
|
6510 |
+
{
|
6511 |
+
"epoch": 0.181586430042161,
|
6512 |
+
"grad_norm": 6.445769786834717,
|
6513 |
+
"learning_rate": 4.69445450515498e-05,
|
6514 |
+
"loss": 2.6642,
|
6515 |
+
"step": 926
|
6516 |
+
},
|
6517 |
+
{
|
6518 |
+
"epoch": 0.181782527698794,
|
6519 |
+
"grad_norm": 8.51395320892334,
|
6520 |
+
"learning_rate": 4.674757738667405e-05,
|
6521 |
+
"loss": 1.2191,
|
6522 |
+
"step": 927
|
6523 |
+
},
|
6524 |
+
{
|
6525 |
+
"epoch": 0.181978625355427,
|
6526 |
+
"grad_norm": 9.159540176391602,
|
6527 |
+
"learning_rate": 4.6550897679183625e-05,
|
6528 |
+
"loss": 2.1541,
|
6529 |
+
"step": 928
|
6530 |
+
},
|
6531 |
+
{
|
6532 |
+
"epoch": 0.18217472301206,
|
6533 |
+
"grad_norm": 7.939080715179443,
|
6534 |
+
"learning_rate": 4.635450699260535e-05,
|
6535 |
+
"loss": 1.4875,
|
6536 |
+
"step": 929
|
6537 |
+
},
|
6538 |
+
{
|
6539 |
+
"epoch": 0.182370820668693,
|
6540 |
+
"grad_norm": 11.180910110473633,
|
6541 |
+
"learning_rate": 4.615840638890305e-05,
|
6542 |
+
"loss": 1.9817,
|
6543 |
+
"step": 930
|
6544 |
+
},
|
6545 |
+
{
|
6546 |
+
"epoch": 0.182566918325326,
|
6547 |
+
"grad_norm": 7.014312744140625,
|
6548 |
+
"learning_rate": 4.596259692847198e-05,
|
6549 |
+
"loss": 1.2057,
|
6550 |
+
"step": 931
|
6551 |
+
},
|
6552 |
+
{
|
6553 |
+
"epoch": 0.18276301598195901,
|
6554 |
+
"grad_norm": 3.89388108253479,
|
6555 |
+
"learning_rate": 4.5767079670133064e-05,
|
6556 |
+
"loss": 1.2204,
|
6557 |
+
"step": 932
|
6558 |
+
},
|
6559 |
+
{
|
6560 |
+
"epoch": 0.18295911363859202,
|
6561 |
+
"grad_norm": 7.551772594451904,
|
6562 |
+
"learning_rate": 4.557185567112725e-05,
|
6563 |
+
"loss": 1.271,
|
6564 |
+
"step": 933
|
6565 |
+
},
|
6566 |
+
{
|
6567 |
+
"epoch": 0.18315521129522502,
|
6568 |
+
"grad_norm": 5.167237758636475,
|
6569 |
+
"learning_rate": 4.537692598710962e-05,
|
6570 |
+
"loss": 2.0885,
|
6571 |
+
"step": 934
|
6572 |
+
},
|
6573 |
+
{
|
6574 |
+
"epoch": 0.18335130895185803,
|
6575 |
+
"grad_norm": 7.082968235015869,
|
6576 |
+
"learning_rate": 4.5182291672143796e-05,
|
6577 |
+
"loss": 2.0684,
|
6578 |
+
"step": 935
|
6579 |
+
},
|
6580 |
+
{
|
6581 |
+
"epoch": 0.18354740660849103,
|
6582 |
+
"grad_norm": 5.6844801902771,
|
6583 |
+
"learning_rate": 4.498795377869622e-05,
|
6584 |
+
"loss": 1.2413,
|
6585 |
+
"step": 936
|
6586 |
+
},
|
6587 |
+
{
|
6588 |
+
"epoch": 0.18374350426512404,
|
6589 |
+
"grad_norm": 8.226604461669922,
|
6590 |
+
"learning_rate": 4.479391335763054e-05,
|
6591 |
+
"loss": 1.7761,
|
6592 |
+
"step": 937
|
6593 |
+
},
|
6594 |
+
{
|
6595 |
+
"epoch": 0.18393960192175704,
|
6596 |
+
"grad_norm": 6.878251075744629,
|
6597 |
+
"learning_rate": 4.4600171458201735e-05,
|
6598 |
+
"loss": 1.4162,
|
6599 |
+
"step": 938
|
6600 |
+
},
|
6601 |
+
{
|
6602 |
+
"epoch": 0.18413569957839004,
|
6603 |
+
"grad_norm": 6.791055202484131,
|
6604 |
+
"learning_rate": 4.440672912805063e-05,
|
6605 |
+
"loss": 1.3623,
|
6606 |
+
"step": 939
|
6607 |
+
},
|
6608 |
+
{
|
6609 |
+
"epoch": 0.18433179723502305,
|
6610 |
+
"grad_norm": 4.67439079284668,
|
6611 |
+
"learning_rate": 4.4213587413198085e-05,
|
6612 |
+
"loss": 0.9896,
|
6613 |
+
"step": 940
|
6614 |
+
},
|
6615 |
+
{
|
6616 |
+
"epoch": 0.18452789489165605,
|
6617 |
+
"grad_norm": 6.860576629638672,
|
6618 |
+
"learning_rate": 4.402074735803955e-05,
|
6619 |
+
"loss": 1.7911,
|
6620 |
+
"step": 941
|
6621 |
+
},
|
6622 |
+
{
|
6623 |
+
"epoch": 0.18472399254828906,
|
6624 |
+
"grad_norm": 5.80997896194458,
|
6625 |
+
"learning_rate": 4.382821000533913e-05,
|
6626 |
+
"loss": 2.8923,
|
6627 |
+
"step": 942
|
6628 |
+
},
|
6629 |
+
{
|
6630 |
+
"epoch": 0.18492009020492206,
|
6631 |
+
"grad_norm": 5.034128189086914,
|
6632 |
+
"learning_rate": 4.3635976396224186e-05,
|
6633 |
+
"loss": 1.7938,
|
6634 |
+
"step": 943
|
6635 |
+
},
|
6636 |
+
{
|
6637 |
+
"epoch": 0.18511618786155506,
|
6638 |
+
"grad_norm": 4.669545650482178,
|
6639 |
+
"learning_rate": 4.3444047570179525e-05,
|
6640 |
+
"loss": 0.8484,
|
6641 |
+
"step": 944
|
6642 |
+
},
|
6643 |
+
{
|
6644 |
+
"epoch": 0.18531228551818807,
|
6645 |
+
"grad_norm": 6.147574424743652,
|
6646 |
+
"learning_rate": 4.3252424565042017e-05,
|
6647 |
+
"loss": 1.1467,
|
6648 |
+
"step": 945
|
6649 |
+
},
|
6650 |
+
{
|
6651 |
+
"epoch": 0.18550838317482107,
|
6652 |
+
"grad_norm": 21.214567184448242,
|
6653 |
+
"learning_rate": 4.306110841699471e-05,
|
6654 |
+
"loss": 3.0427,
|
6655 |
+
"step": 946
|
6656 |
+
},
|
6657 |
+
{
|
6658 |
+
"epoch": 0.18570448083145408,
|
6659 |
+
"grad_norm": 5.828713893890381,
|
6660 |
+
"learning_rate": 4.2870100160561344e-05,
|
6661 |
+
"loss": 0.9022,
|
6662 |
+
"step": 947
|
6663 |
+
},
|
6664 |
+
{
|
6665 |
+
"epoch": 0.18590057848808705,
|
6666 |
+
"grad_norm": 8.350537300109863,
|
6667 |
+
"learning_rate": 4.267940082860088e-05,
|
6668 |
+
"loss": 1.5416,
|
6669 |
+
"step": 948
|
6670 |
+
},
|
6671 |
+
{
|
6672 |
+
"epoch": 0.18609667614472006,
|
6673 |
+
"grad_norm": 3.421091318130493,
|
6674 |
+
"learning_rate": 4.2489011452301705e-05,
|
6675 |
+
"loss": 0.7294,
|
6676 |
+
"step": 949
|
6677 |
+
},
|
6678 |
+
{
|
6679 |
+
"epoch": 0.18629277380135306,
|
6680 |
+
"grad_norm": 5.414788246154785,
|
6681 |
+
"learning_rate": 4.229893306117616e-05,
|
6682 |
+
"loss": 0.9339,
|
6683 |
+
"step": 950
|
6684 |
+
},
|
6685 |
+
{
|
6686 |
+
"epoch": 0.18648887145798607,
|
6687 |
+
"grad_norm": 4.0585761070251465,
|
6688 |
+
"learning_rate": 4.210916668305496e-05,
|
6689 |
+
"loss": 0.9102,
|
6690 |
+
"step": 951
|
6691 |
+
},
|
6692 |
+
{
|
6693 |
+
"epoch": 0.18668496911461907,
|
6694 |
+
"grad_norm": 13.428049087524414,
|
6695 |
+
"learning_rate": 4.1919713344081704e-05,
|
6696 |
+
"loss": 2.287,
|
6697 |
+
"step": 952
|
6698 |
+
},
|
6699 |
+
{
|
6700 |
+
"epoch": 0.18688106677125207,
|
6701 |
+
"grad_norm": 7.367785453796387,
|
6702 |
+
"learning_rate": 4.17305740687072e-05,
|
6703 |
+
"loss": 1.5482,
|
6704 |
+
"step": 953
|
6705 |
+
},
|
6706 |
+
{
|
6707 |
+
"epoch": 0.18707716442788508,
|
6708 |
+
"grad_norm": 4.9446635246276855,
|
6709 |
+
"learning_rate": 4.154174987968398e-05,
|
6710 |
+
"loss": 1.5513,
|
6711 |
+
"step": 954
|
6712 |
+
},
|
6713 |
+
{
|
6714 |
+
"epoch": 0.18727326208451808,
|
6715 |
+
"grad_norm": 6.660252094268799,
|
6716 |
+
"learning_rate": 4.135324179806078e-05,
|
6717 |
+
"loss": 1.6624,
|
6718 |
+
"step": 955
|
6719 |
+
},
|
6720 |
+
{
|
6721 |
+
"epoch": 0.1874693597411511,
|
6722 |
+
"grad_norm": 8.531050682067871,
|
6723 |
+
"learning_rate": 4.11650508431771e-05,
|
6724 |
+
"loss": 1.58,
|
6725 |
+
"step": 956
|
6726 |
+
},
|
6727 |
+
{
|
6728 |
+
"epoch": 0.1876654573977841,
|
6729 |
+
"grad_norm": 5.770985126495361,
|
6730 |
+
"learning_rate": 4.09771780326575e-05,
|
6731 |
+
"loss": 1.2329,
|
6732 |
+
"step": 957
|
6733 |
+
},
|
6734 |
+
{
|
6735 |
+
"epoch": 0.1878615550544171,
|
6736 |
+
"grad_norm": 6.347046375274658,
|
6737 |
+
"learning_rate": 4.0789624382406264e-05,
|
6738 |
+
"loss": 1.3816,
|
6739 |
+
"step": 958
|
6740 |
+
},
|
6741 |
+
{
|
6742 |
+
"epoch": 0.1880576527110501,
|
6743 |
+
"grad_norm": 10.09997272491455,
|
6744 |
+
"learning_rate": 4.060239090660181e-05,
|
6745 |
+
"loss": 1.6532,
|
6746 |
+
"step": 959
|
6747 |
+
},
|
6748 |
+
{
|
6749 |
+
"epoch": 0.1882537503676831,
|
6750 |
+
"grad_norm": 5.894500255584717,
|
6751 |
+
"learning_rate": 4.041547861769126e-05,
|
6752 |
+
"loss": 2.0952,
|
6753 |
+
"step": 960
|
6754 |
+
},
|
6755 |
+
{
|
6756 |
+
"epoch": 0.1884498480243161,
|
6757 |
+
"grad_norm": 3.906419038772583,
|
6758 |
+
"learning_rate": 4.0228888526384936e-05,
|
6759 |
+
"loss": 1.5555,
|
6760 |
+
"step": 961
|
6761 |
+
},
|
6762 |
+
{
|
6763 |
+
"epoch": 0.1886459456809491,
|
6764 |
+
"grad_norm": 6.6089630126953125,
|
6765 |
+
"learning_rate": 4.004262164165098e-05,
|
6766 |
+
"loss": 2.0901,
|
6767 |
+
"step": 962
|
6768 |
+
},
|
6769 |
+
{
|
6770 |
+
"epoch": 0.18884204333758212,
|
6771 |
+
"grad_norm": 5.344242095947266,
|
6772 |
+
"learning_rate": 3.985667897070974e-05,
|
6773 |
+
"loss": 2.2121,
|
6774 |
+
"step": 963
|
6775 |
+
},
|
6776 |
+
{
|
6777 |
+
"epoch": 0.18903814099421512,
|
6778 |
+
"grad_norm": 3.924384117126465,
|
6779 |
+
"learning_rate": 3.967106151902842e-05,
|
6780 |
+
"loss": 2.3173,
|
6781 |
+
"step": 964
|
6782 |
+
},
|
6783 |
+
{
|
6784 |
+
"epoch": 0.18923423865084812,
|
6785 |
+
"grad_norm": 5.392980098724365,
|
6786 |
+
"learning_rate": 3.948577029031566e-05,
|
6787 |
+
"loss": 1.3514,
|
6788 |
+
"step": 965
|
6789 |
+
},
|
6790 |
+
{
|
6791 |
+
"epoch": 0.18943033630748113,
|
6792 |
+
"grad_norm": 5.208516597747803,
|
6793 |
+
"learning_rate": 3.930080628651607e-05,
|
6794 |
+
"loss": 1.1333,
|
6795 |
+
"step": 966
|
6796 |
+
},
|
6797 |
+
{
|
6798 |
+
"epoch": 0.18962643396411413,
|
6799 |
+
"grad_norm": 8.4103364944458,
|
6800 |
+
"learning_rate": 3.9116170507804784e-05,
|
6801 |
+
"loss": 3.0624,
|
6802 |
+
"step": 967
|
6803 |
+
},
|
6804 |
+
{
|
6805 |
+
"epoch": 0.18982253162074714,
|
6806 |
+
"grad_norm": 4.458950996398926,
|
6807 |
+
"learning_rate": 3.89318639525822e-05,
|
6808 |
+
"loss": 1.974,
|
6809 |
+
"step": 968
|
6810 |
+
},
|
6811 |
+
{
|
6812 |
+
"epoch": 0.19001862927738014,
|
6813 |
+
"grad_norm": 4.55155086517334,
|
6814 |
+
"learning_rate": 3.874788761746836e-05,
|
6815 |
+
"loss": 1.2587,
|
6816 |
+
"step": 969
|
6817 |
+
},
|
6818 |
+
{
|
6819 |
+
"epoch": 0.19021472693401315,
|
6820 |
+
"grad_norm": 3.517223834991455,
|
6821 |
+
"learning_rate": 3.8564242497297686e-05,
|
6822 |
+
"loss": 0.9581,
|
6823 |
+
"step": 970
|
6824 |
+
},
|
6825 |
+
{
|
6826 |
+
"epoch": 0.19041082459064615,
|
6827 |
+
"grad_norm": 6.368021011352539,
|
6828 |
+
"learning_rate": 3.8380929585113666e-05,
|
6829 |
+
"loss": 1.6365,
|
6830 |
+
"step": 971
|
6831 |
+
},
|
6832 |
+
{
|
6833 |
+
"epoch": 0.19060692224727915,
|
6834 |
+
"grad_norm": 5.153208255767822,
|
6835 |
+
"learning_rate": 3.819794987216333e-05,
|
6836 |
+
"loss": 0.9107,
|
6837 |
+
"step": 972
|
6838 |
+
},
|
6839 |
+
{
|
6840 |
+
"epoch": 0.19080301990391216,
|
6841 |
+
"grad_norm": 9.487665176391602,
|
6842 |
+
"learning_rate": 3.801530434789199e-05,
|
6843 |
+
"loss": 2.9295,
|
6844 |
+
"step": 973
|
6845 |
+
},
|
6846 |
+
{
|
6847 |
+
"epoch": 0.19099911756054516,
|
6848 |
+
"grad_norm": 10.716341018676758,
|
6849 |
+
"learning_rate": 3.783299399993781e-05,
|
6850 |
+
"loss": 1.9429,
|
6851 |
+
"step": 974
|
6852 |
+
},
|
6853 |
+
{
|
6854 |
+
"epoch": 0.19119521521717817,
|
6855 |
+
"grad_norm": 4.834690093994141,
|
6856 |
+
"learning_rate": 3.7651019814126654e-05,
|
6857 |
+
"loss": 1.7383,
|
6858 |
+
"step": 975
|
6859 |
+
},
|
6860 |
+
{
|
6861 |
+
"epoch": 0.19139131287381117,
|
6862 |
+
"grad_norm": 4.203547954559326,
|
6863 |
+
"learning_rate": 3.74693827744665e-05,
|
6864 |
+
"loss": 0.7084,
|
6865 |
+
"step": 976
|
6866 |
+
},
|
6867 |
+
{
|
6868 |
+
"epoch": 0.19158741053044417,
|
6869 |
+
"grad_norm": 8.38505744934082,
|
6870 |
+
"learning_rate": 3.7288083863142274e-05,
|
6871 |
+
"loss": 2.7231,
|
6872 |
+
"step": 977
|
6873 |
+
},
|
6874 |
+
{
|
6875 |
+
"epoch": 0.19178350818707715,
|
6876 |
+
"grad_norm": 6.748301029205322,
|
6877 |
+
"learning_rate": 3.710712406051049e-05,
|
6878 |
+
"loss": 1.3249,
|
6879 |
+
"step": 978
|
6880 |
+
},
|
6881 |
+
{
|
6882 |
+
"epoch": 0.19197960584371015,
|
6883 |
+
"grad_norm": 5.358412265777588,
|
6884 |
+
"learning_rate": 3.692650434509404e-05,
|
6885 |
+
"loss": 0.7995,
|
6886 |
+
"step": 979
|
6887 |
+
},
|
6888 |
+
{
|
6889 |
+
"epoch": 0.19217570350034316,
|
6890 |
+
"grad_norm": 3.909745693206787,
|
6891 |
+
"learning_rate": 3.674622569357673e-05,
|
6892 |
+
"loss": 2.3175,
|
6893 |
+
"step": 980
|
6894 |
+
},
|
6895 |
+
{
|
6896 |
+
"epoch": 0.19237180115697616,
|
6897 |
+
"grad_norm": 11.555536270141602,
|
6898 |
+
"learning_rate": 3.6566289080798153e-05,
|
6899 |
+
"loss": 1.7112,
|
6900 |
+
"step": 981
|
6901 |
+
},
|
6902 |
+
{
|
6903 |
+
"epoch": 0.19256789881360917,
|
6904 |
+
"grad_norm": 10.718825340270996,
|
6905 |
+
"learning_rate": 3.638669547974831e-05,
|
6906 |
+
"loss": 1.1124,
|
6907 |
+
"step": 982
|
6908 |
+
},
|
6909 |
+
{
|
6910 |
+
"epoch": 0.19276399647024217,
|
6911 |
+
"grad_norm": 4.483185291290283,
|
6912 |
+
"learning_rate": 3.6207445861562494e-05,
|
6913 |
+
"loss": 1.85,
|
6914 |
+
"step": 983
|
6915 |
+
},
|
6916 |
+
{
|
6917 |
+
"epoch": 0.19296009412687518,
|
6918 |
+
"grad_norm": 4.881377696990967,
|
6919 |
+
"learning_rate": 3.602854119551585e-05,
|
6920 |
+
"loss": 0.8783,
|
6921 |
+
"step": 984
|
6922 |
+
},
|
6923 |
+
{
|
6924 |
+
"epoch": 0.19315619178350818,
|
6925 |
+
"grad_norm": 5.172372817993164,
|
6926 |
+
"learning_rate": 3.584998244901826e-05,
|
6927 |
+
"loss": 1.1851,
|
6928 |
+
"step": 985
|
6929 |
+
},
|
6930 |
+
{
|
6931 |
+
"epoch": 0.19335228944014118,
|
6932 |
+
"grad_norm": 8.540918350219727,
|
6933 |
+
"learning_rate": 3.567177058760905e-05,
|
6934 |
+
"loss": 0.8138,
|
6935 |
+
"step": 986
|
6936 |
+
},
|
6937 |
+
{
|
6938 |
+
"epoch": 0.1935483870967742,
|
6939 |
+
"grad_norm": 10.905487060546875,
|
6940 |
+
"learning_rate": 3.549390657495188e-05,
|
6941 |
+
"loss": 2.8036,
|
6942 |
+
"step": 987
|
6943 |
+
},
|
6944 |
+
{
|
6945 |
+
"epoch": 0.1937444847534072,
|
6946 |
+
"grad_norm": 6.448824405670166,
|
6947 |
+
"learning_rate": 3.531639137282938e-05,
|
6948 |
+
"loss": 2.4083,
|
6949 |
+
"step": 988
|
6950 |
+
},
|
6951 |
+
{
|
6952 |
+
"epoch": 0.1939405824100402,
|
6953 |
+
"grad_norm": 4.409211158752441,
|
6954 |
+
"learning_rate": 3.5139225941138e-05,
|
6955 |
+
"loss": 1.2686,
|
6956 |
+
"step": 989
|
6957 |
+
},
|
6958 |
+
{
|
6959 |
+
"epoch": 0.1941366800666732,
|
6960 |
+
"grad_norm": 4.925105094909668,
|
6961 |
+
"learning_rate": 3.496241123788294e-05,
|
6962 |
+
"loss": 3.3955,
|
6963 |
+
"step": 990
|
6964 |
+
},
|
6965 |
+
{
|
6966 |
+
"epoch": 0.1943327777233062,
|
6967 |
+
"grad_norm": 6.541469097137451,
|
6968 |
+
"learning_rate": 3.478594821917279e-05,
|
6969 |
+
"loss": 1.4906,
|
6970 |
+
"step": 991
|
6971 |
+
},
|
6972 |
+
{
|
6973 |
+
"epoch": 0.1945288753799392,
|
6974 |
+
"grad_norm": 7.749048233032227,
|
6975 |
+
"learning_rate": 3.460983783921443e-05,
|
6976 |
+
"loss": 1.7907,
|
6977 |
+
"step": 992
|
6978 |
+
},
|
6979 |
+
{
|
6980 |
+
"epoch": 0.1947249730365722,
|
6981 |
+
"grad_norm": 4.833098888397217,
|
6982 |
+
"learning_rate": 3.44340810503079e-05,
|
6983 |
+
"loss": 1.2777,
|
6984 |
+
"step": 993
|
6985 |
+
},
|
6986 |
+
{
|
6987 |
+
"epoch": 0.19492107069320522,
|
6988 |
+
"grad_norm": 5.138740062713623,
|
6989 |
+
"learning_rate": 3.425867880284129e-05,
|
6990 |
+
"loss": 1.7267,
|
6991 |
+
"step": 994
|
6992 |
+
},
|
6993 |
+
{
|
6994 |
+
"epoch": 0.19511716834983822,
|
6995 |
+
"grad_norm": 8.098197937011719,
|
6996 |
+
"learning_rate": 3.4083632045285466e-05,
|
6997 |
+
"loss": 1.7961,
|
6998 |
+
"step": 995
|
6999 |
+
},
|
7000 |
+
{
|
7001 |
+
"epoch": 0.19531326600647123,
|
7002 |
+
"grad_norm": 4.534963130950928,
|
7003 |
+
"learning_rate": 3.390894172418903e-05,
|
7004 |
+
"loss": 1.773,
|
7005 |
+
"step": 996
|
7006 |
+
},
|
7007 |
+
{
|
7008 |
+
"epoch": 0.19550936366310423,
|
7009 |
+
"grad_norm": 9.311367988586426,
|
7010 |
+
"learning_rate": 3.373460878417315e-05,
|
7011 |
+
"loss": 1.2173,
|
7012 |
+
"step": 997
|
7013 |
+
},
|
7014 |
+
{
|
7015 |
+
"epoch": 0.19570546131973723,
|
7016 |
+
"grad_norm": 4.8790388107299805,
|
7017 |
+
"learning_rate": 3.356063416792662e-05,
|
7018 |
+
"loss": 1.3003,
|
7019 |
+
"step": 998
|
7020 |
+
},
|
7021 |
+
{
|
7022 |
+
"epoch": 0.19590155897637024,
|
7023 |
+
"grad_norm": 4.137528419494629,
|
7024 |
+
"learning_rate": 3.338701881620052e-05,
|
7025 |
+
"loss": 1.8794,
|
7026 |
+
"step": 999
|
7027 |
+
},
|
7028 |
+
{
|
7029 |
+
"epoch": 0.19609765663300324,
|
7030 |
+
"grad_norm": 8.082904815673828,
|
7031 |
+
"learning_rate": 3.321376366780326e-05,
|
7032 |
+
"loss": 1.3442,
|
7033 |
+
"step": 1000
|
7034 |
+
},
|
7035 |
+
{
|
7036 |
+
"epoch": 0.19629375428963625,
|
7037 |
+
"grad_norm": 7.494518280029297,
|
7038 |
+
"learning_rate": 3.3040869659595485e-05,
|
7039 |
+
"loss": 2.3671,
|
7040 |
+
"step": 1001
|
7041 |
+
},
|
7042 |
+
{
|
7043 |
+
"epoch": 0.19648985194626925,
|
7044 |
+
"grad_norm": 4.482415676116943,
|
7045 |
+
"learning_rate": 3.286833772648504e-05,
|
7046 |
+
"loss": 0.7332,
|
7047 |
+
"step": 1002
|
7048 |
+
},
|
7049 |
+
{
|
7050 |
+
"epoch": 0.19668594960290225,
|
7051 |
+
"grad_norm": 6.464245796203613,
|
7052 |
+
"learning_rate": 3.269616880142182e-05,
|
7053 |
+
"loss": 1.6453,
|
7054 |
+
"step": 1003
|
7055 |
+
},
|
7056 |
+
{
|
7057 |
+
"epoch": 0.19688204725953526,
|
7058 |
+
"grad_norm": 6.342423915863037,
|
7059 |
+
"learning_rate": 3.252436381539291e-05,
|
7060 |
+
"loss": 1.5742,
|
7061 |
+
"step": 1004
|
7062 |
+
},
|
7063 |
+
{
|
7064 |
+
"epoch": 0.19707814491616826,
|
7065 |
+
"grad_norm": 2.4557156562805176,
|
7066 |
+
"learning_rate": 3.235292369741729e-05,
|
7067 |
+
"loss": 0.3941,
|
7068 |
+
"step": 1005
|
7069 |
+
},
|
7070 |
+
{
|
7071 |
+
"epoch": 0.19727424257280127,
|
7072 |
+
"grad_norm": 6.024993419647217,
|
7073 |
+
"learning_rate": 3.218184937454103e-05,
|
7074 |
+
"loss": 1.7341,
|
7075 |
+
"step": 1006
|
7076 |
+
},
|
7077 |
+
{
|
7078 |
+
"epoch": 0.19747034022943427,
|
7079 |
+
"grad_norm": 4.793758392333984,
|
7080 |
+
"learning_rate": 3.2011141771832184e-05,
|
7081 |
+
"loss": 0.8558,
|
7082 |
+
"step": 1007
|
7083 |
+
},
|
7084 |
+
{
|
7085 |
+
"epoch": 0.19766643788606725,
|
7086 |
+
"grad_norm": 8.792879104614258,
|
7087 |
+
"learning_rate": 3.184080181237575e-05,
|
7088 |
+
"loss": 0.6646,
|
7089 |
+
"step": 1008
|
7090 |
+
},
|
7091 |
+
{
|
7092 |
+
"epoch": 0.19786253554270025,
|
7093 |
+
"grad_norm": 8.334936141967773,
|
7094 |
+
"learning_rate": 3.167083041726886e-05,
|
7095 |
+
"loss": 1.1602,
|
7096 |
+
"step": 1009
|
7097 |
+
},
|
7098 |
+
{
|
7099 |
+
"epoch": 0.19805863319933326,
|
7100 |
+
"grad_norm": 5.780764579772949,
|
7101 |
+
"learning_rate": 3.1501228505615534e-05,
|
7102 |
+
"loss": 2.342,
|
7103 |
+
"step": 1010
|
7104 |
+
},
|
7105 |
+
{
|
7106 |
+
"epoch": 0.19825473085596626,
|
7107 |
+
"grad_norm": 9.708879470825195,
|
7108 |
+
"learning_rate": 3.1331996994521915e-05,
|
7109 |
+
"loss": 0.981,
|
7110 |
+
"step": 1011
|
7111 |
+
},
|
7112 |
+
{
|
7113 |
+
"epoch": 0.19845082851259926,
|
7114 |
+
"grad_norm": 5.935221195220947,
|
7115 |
+
"learning_rate": 3.1163136799091186e-05,
|
7116 |
+
"loss": 0.8119,
|
7117 |
+
"step": 1012
|
7118 |
+
},
|
7119 |
+
{
|
7120 |
+
"epoch": 0.19864692616923227,
|
7121 |
+
"grad_norm": 9.303512573242188,
|
7122 |
+
"learning_rate": 3.0994648832418784e-05,
|
7123 |
+
"loss": 1.5852,
|
7124 |
+
"step": 1013
|
7125 |
+
},
|
7126 |
+
{
|
7127 |
+
"epoch": 0.19884302382586527,
|
7128 |
+
"grad_norm": 7.7904052734375,
|
7129 |
+
"learning_rate": 3.082653400558724e-05,
|
7130 |
+
"loss": 1.4118,
|
7131 |
+
"step": 1014
|
7132 |
+
},
|
7133 |
+
{
|
7134 |
+
"epoch": 0.19903912148249828,
|
7135 |
+
"grad_norm": 4.866517066955566,
|
7136 |
+
"learning_rate": 3.065879322766142e-05,
|
7137 |
+
"loss": 0.7066,
|
7138 |
+
"step": 1015
|
7139 |
+
},
|
7140 |
+
{
|
7141 |
+
"epoch": 0.19923521913913128,
|
7142 |
+
"grad_norm": 4.0074896812438965,
|
7143 |
+
"learning_rate": 3.0491427405683514e-05,
|
7144 |
+
"loss": 1.0785,
|
7145 |
+
"step": 1016
|
7146 |
+
},
|
7147 |
+
{
|
7148 |
+
"epoch": 0.19943131679576429,
|
7149 |
+
"grad_norm": 5.893292427062988,
|
7150 |
+
"learning_rate": 3.0324437444668274e-05,
|
7151 |
+
"loss": 2.6953,
|
7152 |
+
"step": 1017
|
7153 |
+
},
|
7154 |
+
{
|
7155 |
+
"epoch": 0.1996274144523973,
|
7156 |
+
"grad_norm": 10.129006385803223,
|
7157 |
+
"learning_rate": 3.0157824247597922e-05,
|
7158 |
+
"loss": 1.4669,
|
7159 |
+
"step": 1018
|
7160 |
+
},
|
7161 |
+
{
|
7162 |
+
"epoch": 0.1998235121090303,
|
7163 |
+
"grad_norm": 6.336552143096924,
|
7164 |
+
"learning_rate": 2.9991588715417383e-05,
|
7165 |
+
"loss": 1.9093,
|
7166 |
+
"step": 1019
|
7167 |
+
},
|
7168 |
+
{
|
7169 |
+
"epoch": 0.2000196097656633,
|
7170 |
+
"grad_norm": 7.2662482261657715,
|
7171 |
+
"learning_rate": 2.9825731747029394e-05,
|
7172 |
+
"loss": 1.3408,
|
7173 |
+
"step": 1020
|
7174 |
+
},
|
7175 |
+
{
|
7176 |
+
"epoch": 0.2002157074222963,
|
7177 |
+
"grad_norm": 4.900631427764893,
|
7178 |
+
"learning_rate": 2.966025423928972e-05,
|
7179 |
+
"loss": 1.2219,
|
7180 |
+
"step": 1021
|
7181 |
+
},
|
7182 |
+
{
|
7183 |
+
"epoch": 0.2004118050789293,
|
7184 |
+
"grad_norm": 10.661945343017578,
|
7185 |
+
"learning_rate": 2.9495157087002123e-05,
|
7186 |
+
"loss": 2.2735,
|
7187 |
+
"step": 1022
|
7188 |
+
},
|
7189 |
+
{
|
7190 |
+
"epoch": 0.2006079027355623,
|
7191 |
+
"grad_norm": 6.423792362213135,
|
7192 |
+
"learning_rate": 2.9330441182913694e-05,
|
7193 |
+
"loss": 1.362,
|
7194 |
+
"step": 1023
|
7195 |
+
},
|
7196 |
+
{
|
7197 |
+
"epoch": 0.2006079027355623,
|
7198 |
+
"eval_loss": 0.3885071277618408,
|
7199 |
+
"eval_runtime": 77.8128,
|
7200 |
+
"eval_samples_per_second": 27.605,
|
7201 |
+
"eval_steps_per_second": 13.802,
|
7202 |
+
"step": 1023
|
7203 |
}
|
7204 |
],
|
7205 |
"logging_steps": 1,
|
|
|
7219 |
"attributes": {}
|
7220 |
}
|
7221 |
},
|
7222 |
+
"total_flos": 2.1978504362144563e+17,
|
7223 |
"train_batch_size": 2,
|
7224 |
"trial_name": null,
|
7225 |
"trial_params": null
|