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 +2334 -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 sha256:3f7fc3027701fc162cacfe1f164f9286c19e2c312c8caa0ac162e5c2466bc3fa
|
| 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:fb8bef0278b83e5db495bff5d5d0002f9f7c872c67c02a1d8ad135e9b6ab6ca5
|
| 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:bdbddfb9ed7da6ee0fba930abf6b54364e831087db6cbd0b392f9d8e545a0fb2
|
| 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:e543c4939980849c5a85545ea803fa092b2cc6c74d0eae335c9796c38c00f741
|
| 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,
|
|
@@ -9338,6 +9338,2337 @@
|
|
| 9338 |
"learning_rate": 2.9887914128987465e-05,
|
| 9339 |
"loss": 1.852,
|
| 9340 |
"step": 19995
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9341 |
}
|
| 9342 |
],
|
| 9343 |
"logging_steps": 15,
|
|
@@ -9357,7 +11688,7 @@
|
|
| 9357 |
"attributes": {}
|
| 9358 |
}
|
| 9359 |
},
|
| 9360 |
-
"total_flos": 3.
|
| 9361 |
"train_batch_size": 4,
|
| 9362 |
"trial_name": null,
|
| 9363 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.048683785183771554,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 25000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 9338 |
"learning_rate": 2.9887914128987465e-05,
|
| 9339 |
"loss": 1.852,
|
| 9340 |
"step": 19995
|
| 9341 |
+
},
|
| 9342 |
+
{
|
| 9343 |
+
"epoch": 0.03896650166109075,
|
| 9344 |
+
"grad_norm": 4.136464595794678,
|
| 9345 |
+
"learning_rate": 2.9887746105205264e-05,
|
| 9346 |
+
"loss": 1.8378,
|
| 9347 |
+
"step": 20010
|
| 9348 |
+
},
|
| 9349 |
+
{
|
| 9350 |
+
"epoch": 0.03899571193220101,
|
| 9351 |
+
"grad_norm": 3.3061976432800293,
|
| 9352 |
+
"learning_rate": 2.9887577956051263e-05,
|
| 9353 |
+
"loss": 1.8702,
|
| 9354 |
+
"step": 20025
|
| 9355 |
+
},
|
| 9356 |
+
{
|
| 9357 |
+
"epoch": 0.039024922203311274,
|
| 9358 |
+
"grad_norm": 3.5822839736938477,
|
| 9359 |
+
"learning_rate": 2.9887409681526876e-05,
|
| 9360 |
+
"loss": 1.87,
|
| 9361 |
+
"step": 20040
|
| 9362 |
+
},
|
| 9363 |
+
{
|
| 9364 |
+
"epoch": 0.03905413247442154,
|
| 9365 |
+
"grad_norm": 4.132532119750977,
|
| 9366 |
+
"learning_rate": 2.9887241281633518e-05,
|
| 9367 |
+
"loss": 2.051,
|
| 9368 |
+
"step": 20055
|
| 9369 |
+
},
|
| 9370 |
+
{
|
| 9371 |
+
"epoch": 0.0390833427455318,
|
| 9372 |
+
"grad_norm": 3.9095962047576904,
|
| 9373 |
+
"learning_rate": 2.9887072756372606e-05,
|
| 9374 |
+
"loss": 1.8271,
|
| 9375 |
+
"step": 20070
|
| 9376 |
+
},
|
| 9377 |
+
{
|
| 9378 |
+
"epoch": 0.039112553016642064,
|
| 9379 |
+
"grad_norm": 3.028393507003784,
|
| 9380 |
+
"learning_rate": 2.988690410574556e-05,
|
| 9381 |
+
"loss": 1.825,
|
| 9382 |
+
"step": 20085
|
| 9383 |
+
},
|
| 9384 |
+
{
|
| 9385 |
+
"epoch": 0.039141763287752325,
|
| 9386 |
+
"grad_norm": 1.7903603315353394,
|
| 9387 |
+
"learning_rate": 2.98867353297538e-05,
|
| 9388 |
+
"loss": 1.9565,
|
| 9389 |
+
"step": 20100
|
| 9390 |
+
},
|
| 9391 |
+
{
|
| 9392 |
+
"epoch": 0.03917097355886259,
|
| 9393 |
+
"grad_norm": 2.168895959854126,
|
| 9394 |
+
"learning_rate": 2.988656642839875e-05,
|
| 9395 |
+
"loss": 1.8483,
|
| 9396 |
+
"step": 20115
|
| 9397 |
+
},
|
| 9398 |
+
{
|
| 9399 |
+
"epoch": 0.039200183829972854,
|
| 9400 |
+
"grad_norm": 7.789179801940918,
|
| 9401 |
+
"learning_rate": 2.988639740168183e-05,
|
| 9402 |
+
"loss": 2.0627,
|
| 9403 |
+
"step": 20130
|
| 9404 |
+
},
|
| 9405 |
+
{
|
| 9406 |
+
"epoch": 0.039229394101083115,
|
| 9407 |
+
"grad_norm": 2.033658981323242,
|
| 9408 |
+
"learning_rate": 2.9886228249604464e-05,
|
| 9409 |
+
"loss": 1.8905,
|
| 9410 |
+
"step": 20145
|
| 9411 |
+
},
|
| 9412 |
+
{
|
| 9413 |
+
"epoch": 0.03925860437219338,
|
| 9414 |
+
"grad_norm": 3.6176931858062744,
|
| 9415 |
+
"learning_rate": 2.9886058972168076e-05,
|
| 9416 |
+
"loss": 1.8981,
|
| 9417 |
+
"step": 20160
|
| 9418 |
+
},
|
| 9419 |
+
{
|
| 9420 |
+
"epoch": 0.039287814643303644,
|
| 9421 |
+
"grad_norm": 3.776540517807007,
|
| 9422 |
+
"learning_rate": 2.9885889569374088e-05,
|
| 9423 |
+
"loss": 1.7417,
|
| 9424 |
+
"step": 20175
|
| 9425 |
+
},
|
| 9426 |
+
{
|
| 9427 |
+
"epoch": 0.039317024914413905,
|
| 9428 |
+
"grad_norm": 2.138796329498291,
|
| 9429 |
+
"learning_rate": 2.9885720041223934e-05,
|
| 9430 |
+
"loss": 1.8845,
|
| 9431 |
+
"step": 20190
|
| 9432 |
+
},
|
| 9433 |
+
{
|
| 9434 |
+
"epoch": 0.039346235185524166,
|
| 9435 |
+
"grad_norm": 5.0401482582092285,
|
| 9436 |
+
"learning_rate": 2.988555038771904e-05,
|
| 9437 |
+
"loss": 1.8966,
|
| 9438 |
+
"step": 20205
|
| 9439 |
+
},
|
| 9440 |
+
{
|
| 9441 |
+
"epoch": 0.039375445456634434,
|
| 9442 |
+
"grad_norm": 2.2886552810668945,
|
| 9443 |
+
"learning_rate": 2.9885380608860827e-05,
|
| 9444 |
+
"loss": 1.8898,
|
| 9445 |
+
"step": 20220
|
| 9446 |
+
},
|
| 9447 |
+
{
|
| 9448 |
+
"epoch": 0.039404655727744695,
|
| 9449 |
+
"grad_norm": 2.286959648132324,
|
| 9450 |
+
"learning_rate": 2.9885210704650734e-05,
|
| 9451 |
+
"loss": 1.9597,
|
| 9452 |
+
"step": 20235
|
| 9453 |
+
},
|
| 9454 |
+
{
|
| 9455 |
+
"epoch": 0.039433865998854956,
|
| 9456 |
+
"grad_norm": 2.538752794265747,
|
| 9457 |
+
"learning_rate": 2.988504067509019e-05,
|
| 9458 |
+
"loss": 1.8451,
|
| 9459 |
+
"step": 20250
|
| 9460 |
+
},
|
| 9461 |
+
{
|
| 9462 |
+
"epoch": 0.039463076269965223,
|
| 9463 |
+
"grad_norm": 3.4922502040863037,
|
| 9464 |
+
"learning_rate": 2.988487052018062e-05,
|
| 9465 |
+
"loss": 1.8717,
|
| 9466 |
+
"step": 20265
|
| 9467 |
+
},
|
| 9468 |
+
{
|
| 9469 |
+
"epoch": 0.039492286541075484,
|
| 9470 |
+
"grad_norm": 3.5919086933135986,
|
| 9471 |
+
"learning_rate": 2.9884700239923467e-05,
|
| 9472 |
+
"loss": 1.8182,
|
| 9473 |
+
"step": 20280
|
| 9474 |
+
},
|
| 9475 |
+
{
|
| 9476 |
+
"epoch": 0.039521496812185745,
|
| 9477 |
+
"grad_norm": 1.9253063201904297,
|
| 9478 |
+
"learning_rate": 2.988452983432016e-05,
|
| 9479 |
+
"loss": 1.8067,
|
| 9480 |
+
"step": 20295
|
| 9481 |
+
},
|
| 9482 |
+
{
|
| 9483 |
+
"epoch": 0.039550707083296006,
|
| 9484 |
+
"grad_norm": 3.168278217315674,
|
| 9485 |
+
"learning_rate": 2.9884359303372127e-05,
|
| 9486 |
+
"loss": 1.9619,
|
| 9487 |
+
"step": 20310
|
| 9488 |
+
},
|
| 9489 |
+
{
|
| 9490 |
+
"epoch": 0.039579917354406274,
|
| 9491 |
+
"grad_norm": 2.574296236038208,
|
| 9492 |
+
"learning_rate": 2.9884188647080816e-05,
|
| 9493 |
+
"loss": 1.9037,
|
| 9494 |
+
"step": 20325
|
| 9495 |
+
},
|
| 9496 |
+
{
|
| 9497 |
+
"epoch": 0.039609127625516535,
|
| 9498 |
+
"grad_norm": 3.87908673286438,
|
| 9499 |
+
"learning_rate": 2.9884017865447657e-05,
|
| 9500 |
+
"loss": 1.8106,
|
| 9501 |
+
"step": 20340
|
| 9502 |
+
},
|
| 9503 |
+
{
|
| 9504 |
+
"epoch": 0.039638337896626796,
|
| 9505 |
+
"grad_norm": 3.060088872909546,
|
| 9506 |
+
"learning_rate": 2.9883846958474093e-05,
|
| 9507 |
+
"loss": 1.8878,
|
| 9508 |
+
"step": 20355
|
| 9509 |
+
},
|
| 9510 |
+
{
|
| 9511 |
+
"epoch": 0.039667548167737064,
|
| 9512 |
+
"grad_norm": 2.6092071533203125,
|
| 9513 |
+
"learning_rate": 2.988367592616156e-05,
|
| 9514 |
+
"loss": 2.0189,
|
| 9515 |
+
"step": 20370
|
| 9516 |
+
},
|
| 9517 |
+
{
|
| 9518 |
+
"epoch": 0.039696758438847325,
|
| 9519 |
+
"grad_norm": 2.688831329345703,
|
| 9520 |
+
"learning_rate": 2.9883504768511496e-05,
|
| 9521 |
+
"loss": 1.9439,
|
| 9522 |
+
"step": 20385
|
| 9523 |
+
},
|
| 9524 |
+
{
|
| 9525 |
+
"epoch": 0.039725968709957586,
|
| 9526 |
+
"grad_norm": 4.126415252685547,
|
| 9527 |
+
"learning_rate": 2.988333348552535e-05,
|
| 9528 |
+
"loss": 1.7334,
|
| 9529 |
+
"step": 20400
|
| 9530 |
+
},
|
| 9531 |
+
{
|
| 9532 |
+
"epoch": 0.03975517898106785,
|
| 9533 |
+
"grad_norm": 2.5536105632781982,
|
| 9534 |
+
"learning_rate": 2.988316207720455e-05,
|
| 9535 |
+
"loss": 1.8741,
|
| 9536 |
+
"step": 20415
|
| 9537 |
+
},
|
| 9538 |
+
{
|
| 9539 |
+
"epoch": 0.039784389252178115,
|
| 9540 |
+
"grad_norm": 4.146584510803223,
|
| 9541 |
+
"learning_rate": 2.9882990543550557e-05,
|
| 9542 |
+
"loss": 1.839,
|
| 9543 |
+
"step": 20430
|
| 9544 |
+
},
|
| 9545 |
+
{
|
| 9546 |
+
"epoch": 0.039813599523288376,
|
| 9547 |
+
"grad_norm": 4.531203746795654,
|
| 9548 |
+
"learning_rate": 2.9882818884564805e-05,
|
| 9549 |
+
"loss": 1.801,
|
| 9550 |
+
"step": 20445
|
| 9551 |
+
},
|
| 9552 |
+
{
|
| 9553 |
+
"epoch": 0.03984280979439864,
|
| 9554 |
+
"grad_norm": 4.3206562995910645,
|
| 9555 |
+
"learning_rate": 2.988264710024874e-05,
|
| 9556 |
+
"loss": 1.9127,
|
| 9557 |
+
"step": 20460
|
| 9558 |
+
},
|
| 9559 |
+
{
|
| 9560 |
+
"epoch": 0.0398720200655089,
|
| 9561 |
+
"grad_norm": 2.258627414703369,
|
| 9562 |
+
"learning_rate": 2.9882475190603815e-05,
|
| 9563 |
+
"loss": 1.9329,
|
| 9564 |
+
"step": 20475
|
| 9565 |
+
},
|
| 9566 |
+
{
|
| 9567 |
+
"epoch": 0.039901230336619166,
|
| 9568 |
+
"grad_norm": 2.874171018600464,
|
| 9569 |
+
"learning_rate": 2.988230315563147e-05,
|
| 9570 |
+
"loss": 1.8632,
|
| 9571 |
+
"step": 20490
|
| 9572 |
+
},
|
| 9573 |
+
{
|
| 9574 |
+
"epoch": 0.03993044060772943,
|
| 9575 |
+
"grad_norm": 2.2248058319091797,
|
| 9576 |
+
"learning_rate": 2.988213099533316e-05,
|
| 9577 |
+
"loss": 1.7653,
|
| 9578 |
+
"step": 20505
|
| 9579 |
+
},
|
| 9580 |
+
{
|
| 9581 |
+
"epoch": 0.03995965087883969,
|
| 9582 |
+
"grad_norm": 2.2211756706237793,
|
| 9583 |
+
"learning_rate": 2.988195870971033e-05,
|
| 9584 |
+
"loss": 2.0817,
|
| 9585 |
+
"step": 20520
|
| 9586 |
+
},
|
| 9587 |
+
{
|
| 9588 |
+
"epoch": 0.039988861149949956,
|
| 9589 |
+
"grad_norm": 2.266099214553833,
|
| 9590 |
+
"learning_rate": 2.9881786298764432e-05,
|
| 9591 |
+
"loss": 2.0993,
|
| 9592 |
+
"step": 20535
|
| 9593 |
+
},
|
| 9594 |
+
{
|
| 9595 |
+
"epoch": 0.04001807142106022,
|
| 9596 |
+
"grad_norm": 3.2420692443847656,
|
| 9597 |
+
"learning_rate": 2.988161376249692e-05,
|
| 9598 |
+
"loss": 1.8448,
|
| 9599 |
+
"step": 20550
|
| 9600 |
+
},
|
| 9601 |
+
{
|
| 9602 |
+
"epoch": 0.04004728169217048,
|
| 9603 |
+
"grad_norm": 2.1219029426574707,
|
| 9604 |
+
"learning_rate": 2.9881441100909244e-05,
|
| 9605 |
+
"loss": 1.9547,
|
| 9606 |
+
"step": 20565
|
| 9607 |
+
},
|
| 9608 |
+
{
|
| 9609 |
+
"epoch": 0.04007649196328074,
|
| 9610 |
+
"grad_norm": 4.585850238800049,
|
| 9611 |
+
"learning_rate": 2.988126831400286e-05,
|
| 9612 |
+
"loss": 1.9619,
|
| 9613 |
+
"step": 20580
|
| 9614 |
+
},
|
| 9615 |
+
{
|
| 9616 |
+
"epoch": 0.04010570223439101,
|
| 9617 |
+
"grad_norm": 3.4670188426971436,
|
| 9618 |
+
"learning_rate": 2.9881095401779224e-05,
|
| 9619 |
+
"loss": 2.0,
|
| 9620 |
+
"step": 20595
|
| 9621 |
+
},
|
| 9622 |
+
{
|
| 9623 |
+
"epoch": 0.04013491250550127,
|
| 9624 |
+
"grad_norm": 2.0285186767578125,
|
| 9625 |
+
"learning_rate": 2.9880922364239787e-05,
|
| 9626 |
+
"loss": 1.9434,
|
| 9627 |
+
"step": 20610
|
| 9628 |
+
},
|
| 9629 |
+
{
|
| 9630 |
+
"epoch": 0.04016412277661153,
|
| 9631 |
+
"grad_norm": 1.873279094696045,
|
| 9632 |
+
"learning_rate": 2.9880749201386014e-05,
|
| 9633 |
+
"loss": 1.8546,
|
| 9634 |
+
"step": 20625
|
| 9635 |
+
},
|
| 9636 |
+
{
|
| 9637 |
+
"epoch": 0.040193333047721796,
|
| 9638 |
+
"grad_norm": 2.1878137588500977,
|
| 9639 |
+
"learning_rate": 2.9880575913219354e-05,
|
| 9640 |
+
"loss": 1.8624,
|
| 9641 |
+
"step": 20640
|
| 9642 |
+
},
|
| 9643 |
+
{
|
| 9644 |
+
"epoch": 0.04022254331883206,
|
| 9645 |
+
"grad_norm": 3.9366180896759033,
|
| 9646 |
+
"learning_rate": 2.988040249974128e-05,
|
| 9647 |
+
"loss": 1.9143,
|
| 9648 |
+
"step": 20655
|
| 9649 |
+
},
|
| 9650 |
+
{
|
| 9651 |
+
"epoch": 0.04025175358994232,
|
| 9652 |
+
"grad_norm": 2.3610141277313232,
|
| 9653 |
+
"learning_rate": 2.9880228960953236e-05,
|
| 9654 |
+
"loss": 1.9616,
|
| 9655 |
+
"step": 20670
|
| 9656 |
+
},
|
| 9657 |
+
{
|
| 9658 |
+
"epoch": 0.04028096386105258,
|
| 9659 |
+
"grad_norm": 2.8658604621887207,
|
| 9660 |
+
"learning_rate": 2.9880055296856695e-05,
|
| 9661 |
+
"loss": 1.7601,
|
| 9662 |
+
"step": 20685
|
| 9663 |
+
},
|
| 9664 |
+
{
|
| 9665 |
+
"epoch": 0.04031017413216285,
|
| 9666 |
+
"grad_norm": 4.454057216644287,
|
| 9667 |
+
"learning_rate": 2.9879881507453112e-05,
|
| 9668 |
+
"loss": 1.7578,
|
| 9669 |
+
"step": 20700
|
| 9670 |
+
},
|
| 9671 |
+
{
|
| 9672 |
+
"epoch": 0.04033938440327311,
|
| 9673 |
+
"grad_norm": 2.244455575942993,
|
| 9674 |
+
"learning_rate": 2.9879707592743957e-05,
|
| 9675 |
+
"loss": 1.8334,
|
| 9676 |
+
"step": 20715
|
| 9677 |
+
},
|
| 9678 |
+
{
|
| 9679 |
+
"epoch": 0.04036859467438337,
|
| 9680 |
+
"grad_norm": 1.61302649974823,
|
| 9681 |
+
"learning_rate": 2.987953355273069e-05,
|
| 9682 |
+
"loss": 1.9746,
|
| 9683 |
+
"step": 20730
|
| 9684 |
+
},
|
| 9685 |
+
{
|
| 9686 |
+
"epoch": 0.04039780494549364,
|
| 9687 |
+
"grad_norm": 2.8323678970336914,
|
| 9688 |
+
"learning_rate": 2.987935938741478e-05,
|
| 9689 |
+
"loss": 2.1251,
|
| 9690 |
+
"step": 20745
|
| 9691 |
+
},
|
| 9692 |
+
{
|
| 9693 |
+
"epoch": 0.0404270152166039,
|
| 9694 |
+
"grad_norm": 2.418241262435913,
|
| 9695 |
+
"learning_rate": 2.987918509679769e-05,
|
| 9696 |
+
"loss": 2.0798,
|
| 9697 |
+
"step": 20760
|
| 9698 |
+
},
|
| 9699 |
+
{
|
| 9700 |
+
"epoch": 0.04045622548771416,
|
| 9701 |
+
"grad_norm": 3.465550661087036,
|
| 9702 |
+
"learning_rate": 2.987901068088089e-05,
|
| 9703 |
+
"loss": 1.8706,
|
| 9704 |
+
"step": 20775
|
| 9705 |
+
},
|
| 9706 |
+
{
|
| 9707 |
+
"epoch": 0.04048543575882442,
|
| 9708 |
+
"grad_norm": 4.148263931274414,
|
| 9709 |
+
"learning_rate": 2.987883613966585e-05,
|
| 9710 |
+
"loss": 1.7925,
|
| 9711 |
+
"step": 20790
|
| 9712 |
+
},
|
| 9713 |
+
{
|
| 9714 |
+
"epoch": 0.04051464602993469,
|
| 9715 |
+
"grad_norm": 2.4039666652679443,
|
| 9716 |
+
"learning_rate": 2.9878661473154037e-05,
|
| 9717 |
+
"loss": 1.8672,
|
| 9718 |
+
"step": 20805
|
| 9719 |
+
},
|
| 9720 |
+
{
|
| 9721 |
+
"epoch": 0.04054385630104495,
|
| 9722 |
+
"grad_norm": 2.12880277633667,
|
| 9723 |
+
"learning_rate": 2.9878486681346923e-05,
|
| 9724 |
+
"loss": 1.8978,
|
| 9725 |
+
"step": 20820
|
| 9726 |
+
},
|
| 9727 |
+
{
|
| 9728 |
+
"epoch": 0.04057306657215521,
|
| 9729 |
+
"grad_norm": 1.9612979888916016,
|
| 9730 |
+
"learning_rate": 2.987831176424598e-05,
|
| 9731 |
+
"loss": 1.9175,
|
| 9732 |
+
"step": 20835
|
| 9733 |
+
},
|
| 9734 |
+
{
|
| 9735 |
+
"epoch": 0.04060227684326548,
|
| 9736 |
+
"grad_norm": 2.161982297897339,
|
| 9737 |
+
"learning_rate": 2.9878136721852682e-05,
|
| 9738 |
+
"loss": 1.7752,
|
| 9739 |
+
"step": 20850
|
| 9740 |
+
},
|
| 9741 |
+
{
|
| 9742 |
+
"epoch": 0.04063148711437574,
|
| 9743 |
+
"grad_norm": 4.4879961013793945,
|
| 9744 |
+
"learning_rate": 2.9877961554168498e-05,
|
| 9745 |
+
"loss": 2.0857,
|
| 9746 |
+
"step": 20865
|
| 9747 |
+
},
|
| 9748 |
+
{
|
| 9749 |
+
"epoch": 0.040660697385486,
|
| 9750 |
+
"grad_norm": 4.1571364402771,
|
| 9751 |
+
"learning_rate": 2.9877786261194914e-05,
|
| 9752 |
+
"loss": 1.949,
|
| 9753 |
+
"step": 20880
|
| 9754 |
+
},
|
| 9755 |
+
{
|
| 9756 |
+
"epoch": 0.04068990765659626,
|
| 9757 |
+
"grad_norm": 3.3120033740997314,
|
| 9758 |
+
"learning_rate": 2.9877610842933397e-05,
|
| 9759 |
+
"loss": 1.8585,
|
| 9760 |
+
"step": 20895
|
| 9761 |
+
},
|
| 9762 |
+
{
|
| 9763 |
+
"epoch": 0.04071911792770653,
|
| 9764 |
+
"grad_norm": 3.193117618560791,
|
| 9765 |
+
"learning_rate": 2.9877435299385424e-05,
|
| 9766 |
+
"loss": 1.731,
|
| 9767 |
+
"step": 20910
|
| 9768 |
+
},
|
| 9769 |
+
{
|
| 9770 |
+
"epoch": 0.04074832819881679,
|
| 9771 |
+
"grad_norm": 2.375343084335327,
|
| 9772 |
+
"learning_rate": 2.987725963055248e-05,
|
| 9773 |
+
"loss": 1.8269,
|
| 9774 |
+
"step": 20925
|
| 9775 |
+
},
|
| 9776 |
+
{
|
| 9777 |
+
"epoch": 0.04077753846992705,
|
| 9778 |
+
"grad_norm": 2.3607242107391357,
|
| 9779 |
+
"learning_rate": 2.9877083836436036e-05,
|
| 9780 |
+
"loss": 1.8305,
|
| 9781 |
+
"step": 20940
|
| 9782 |
+
},
|
| 9783 |
+
{
|
| 9784 |
+
"epoch": 0.04080674874103731,
|
| 9785 |
+
"grad_norm": 3.0205342769622803,
|
| 9786 |
+
"learning_rate": 2.987690791703758e-05,
|
| 9787 |
+
"loss": 1.8631,
|
| 9788 |
+
"step": 20955
|
| 9789 |
+
},
|
| 9790 |
+
{
|
| 9791 |
+
"epoch": 0.04083595901214758,
|
| 9792 |
+
"grad_norm": 3.530947685241699,
|
| 9793 |
+
"learning_rate": 2.9876731872358585e-05,
|
| 9794 |
+
"loss": 1.8431,
|
| 9795 |
+
"step": 20970
|
| 9796 |
+
},
|
| 9797 |
+
{
|
| 9798 |
+
"epoch": 0.04086516928325784,
|
| 9799 |
+
"grad_norm": 2.8419220447540283,
|
| 9800 |
+
"learning_rate": 2.987655570240054e-05,
|
| 9801 |
+
"loss": 1.8519,
|
| 9802 |
+
"step": 20985
|
| 9803 |
+
},
|
| 9804 |
+
{
|
| 9805 |
+
"epoch": 0.0408943795543681,
|
| 9806 |
+
"grad_norm": 2.253532886505127,
|
| 9807 |
+
"learning_rate": 2.9876379407164933e-05,
|
| 9808 |
+
"loss": 1.8688,
|
| 9809 |
+
"step": 21000
|
| 9810 |
+
},
|
| 9811 |
+
{
|
| 9812 |
+
"epoch": 0.04092358982547837,
|
| 9813 |
+
"grad_norm": 1.9279251098632812,
|
| 9814 |
+
"learning_rate": 2.987620298665324e-05,
|
| 9815 |
+
"loss": 1.9111,
|
| 9816 |
+
"step": 21015
|
| 9817 |
+
},
|
| 9818 |
+
{
|
| 9819 |
+
"epoch": 0.04095280009658863,
|
| 9820 |
+
"grad_norm": 3.0645790100097656,
|
| 9821 |
+
"learning_rate": 2.987602644086695e-05,
|
| 9822 |
+
"loss": 1.8359,
|
| 9823 |
+
"step": 21030
|
| 9824 |
+
},
|
| 9825 |
+
{
|
| 9826 |
+
"epoch": 0.04098201036769889,
|
| 9827 |
+
"grad_norm": 2.9384896755218506,
|
| 9828 |
+
"learning_rate": 2.9875849769807544e-05,
|
| 9829 |
+
"loss": 1.747,
|
| 9830 |
+
"step": 21045
|
| 9831 |
+
},
|
| 9832 |
+
{
|
| 9833 |
+
"epoch": 0.04101122063880915,
|
| 9834 |
+
"grad_norm": 3.7138500213623047,
|
| 9835 |
+
"learning_rate": 2.987567297347652e-05,
|
| 9836 |
+
"loss": 1.8528,
|
| 9837 |
+
"step": 21060
|
| 9838 |
+
},
|
| 9839 |
+
{
|
| 9840 |
+
"epoch": 0.04104043090991942,
|
| 9841 |
+
"grad_norm": 4.440821170806885,
|
| 9842 |
+
"learning_rate": 2.987549605187536e-05,
|
| 9843 |
+
"loss": 1.9611,
|
| 9844 |
+
"step": 21075
|
| 9845 |
+
},
|
| 9846 |
+
{
|
| 9847 |
+
"epoch": 0.04106964118102968,
|
| 9848 |
+
"grad_norm": 3.0378854274749756,
|
| 9849 |
+
"learning_rate": 2.9875319005005552e-05,
|
| 9850 |
+
"loss": 2.012,
|
| 9851 |
+
"step": 21090
|
| 9852 |
+
},
|
| 9853 |
+
{
|
| 9854 |
+
"epoch": 0.04109885145213994,
|
| 9855 |
+
"grad_norm": 4.378201961517334,
|
| 9856 |
+
"learning_rate": 2.9875141832868598e-05,
|
| 9857 |
+
"loss": 1.7366,
|
| 9858 |
+
"step": 21105
|
| 9859 |
+
},
|
| 9860 |
+
{
|
| 9861 |
+
"epoch": 0.04112806172325021,
|
| 9862 |
+
"grad_norm": 4.794327735900879,
|
| 9863 |
+
"learning_rate": 2.9874964535465978e-05,
|
| 9864 |
+
"loss": 1.8728,
|
| 9865 |
+
"step": 21120
|
| 9866 |
+
},
|
| 9867 |
+
{
|
| 9868 |
+
"epoch": 0.04115727199436047,
|
| 9869 |
+
"grad_norm": 2.5137903690338135,
|
| 9870 |
+
"learning_rate": 2.987478711279919e-05,
|
| 9871 |
+
"loss": 1.9813,
|
| 9872 |
+
"step": 21135
|
| 9873 |
+
},
|
| 9874 |
+
{
|
| 9875 |
+
"epoch": 0.04118648226547073,
|
| 9876 |
+
"grad_norm": 2.2024412155151367,
|
| 9877 |
+
"learning_rate": 2.987460956486973e-05,
|
| 9878 |
+
"loss": 1.7721,
|
| 9879 |
+
"step": 21150
|
| 9880 |
+
},
|
| 9881 |
+
{
|
| 9882 |
+
"epoch": 0.04121569253658099,
|
| 9883 |
+
"grad_norm": 3.0043609142303467,
|
| 9884 |
+
"learning_rate": 2.987443189167909e-05,
|
| 9885 |
+
"loss": 1.9015,
|
| 9886 |
+
"step": 21165
|
| 9887 |
+
},
|
| 9888 |
+
{
|
| 9889 |
+
"epoch": 0.04124490280769126,
|
| 9890 |
+
"grad_norm": 4.627270698547363,
|
| 9891 |
+
"learning_rate": 2.9874254093228763e-05,
|
| 9892 |
+
"loss": 1.9487,
|
| 9893 |
+
"step": 21180
|
| 9894 |
+
},
|
| 9895 |
+
{
|
| 9896 |
+
"epoch": 0.04127411307880152,
|
| 9897 |
+
"grad_norm": 3.208395481109619,
|
| 9898 |
+
"learning_rate": 2.987407616952025e-05,
|
| 9899 |
+
"loss": 1.6989,
|
| 9900 |
+
"step": 21195
|
| 9901 |
+
},
|
| 9902 |
+
{
|
| 9903 |
+
"epoch": 0.04130332334991178,
|
| 9904 |
+
"grad_norm": 4.102930545806885,
|
| 9905 |
+
"learning_rate": 2.9873898120555055e-05,
|
| 9906 |
+
"loss": 2.0639,
|
| 9907 |
+
"step": 21210
|
| 9908 |
+
},
|
| 9909 |
+
{
|
| 9910 |
+
"epoch": 0.04133253362102205,
|
| 9911 |
+
"grad_norm": 3.846593141555786,
|
| 9912 |
+
"learning_rate": 2.987371994633467e-05,
|
| 9913 |
+
"loss": 1.7067,
|
| 9914 |
+
"step": 21225
|
| 9915 |
+
},
|
| 9916 |
+
{
|
| 9917 |
+
"epoch": 0.04136174389213231,
|
| 9918 |
+
"grad_norm": 3.6651105880737305,
|
| 9919 |
+
"learning_rate": 2.9873541646860597e-05,
|
| 9920 |
+
"loss": 1.8983,
|
| 9921 |
+
"step": 21240
|
| 9922 |
+
},
|
| 9923 |
+
{
|
| 9924 |
+
"epoch": 0.04139095416324257,
|
| 9925 |
+
"grad_norm": 3.715604543685913,
|
| 9926 |
+
"learning_rate": 2.987336322213434e-05,
|
| 9927 |
+
"loss": 1.6676,
|
| 9928 |
+
"step": 21255
|
| 9929 |
+
},
|
| 9930 |
+
{
|
| 9931 |
+
"epoch": 0.041420164434352834,
|
| 9932 |
+
"grad_norm": 3.0780601501464844,
|
| 9933 |
+
"learning_rate": 2.9873184672157395e-05,
|
| 9934 |
+
"loss": 1.9342,
|
| 9935 |
+
"step": 21270
|
| 9936 |
+
},
|
| 9937 |
+
{
|
| 9938 |
+
"epoch": 0.0414493747054631,
|
| 9939 |
+
"grad_norm": 3.385103225708008,
|
| 9940 |
+
"learning_rate": 2.9873005996931274e-05,
|
| 9941 |
+
"loss": 1.9494,
|
| 9942 |
+
"step": 21285
|
| 9943 |
+
},
|
| 9944 |
+
{
|
| 9945 |
+
"epoch": 0.04147858497657336,
|
| 9946 |
+
"grad_norm": 2.9652836322784424,
|
| 9947 |
+
"learning_rate": 2.9872827196457475e-05,
|
| 9948 |
+
"loss": 1.8491,
|
| 9949 |
+
"step": 21300
|
| 9950 |
+
},
|
| 9951 |
+
{
|
| 9952 |
+
"epoch": 0.041507795247683624,
|
| 9953 |
+
"grad_norm": 2.346210479736328,
|
| 9954 |
+
"learning_rate": 2.9872648270737507e-05,
|
| 9955 |
+
"loss": 1.7948,
|
| 9956 |
+
"step": 21315
|
| 9957 |
+
},
|
| 9958 |
+
{
|
| 9959 |
+
"epoch": 0.04153700551879389,
|
| 9960 |
+
"grad_norm": 2.0421793460845947,
|
| 9961 |
+
"learning_rate": 2.9872469219772877e-05,
|
| 9962 |
+
"loss": 1.7642,
|
| 9963 |
+
"step": 21330
|
| 9964 |
+
},
|
| 9965 |
+
{
|
| 9966 |
+
"epoch": 0.04156621578990415,
|
| 9967 |
+
"grad_norm": 3.2347426414489746,
|
| 9968 |
+
"learning_rate": 2.9872290043565094e-05,
|
| 9969 |
+
"loss": 1.9741,
|
| 9970 |
+
"step": 21345
|
| 9971 |
+
},
|
| 9972 |
+
{
|
| 9973 |
+
"epoch": 0.041595426061014414,
|
| 9974 |
+
"grad_norm": 2.8749160766601562,
|
| 9975 |
+
"learning_rate": 2.987211074211566e-05,
|
| 9976 |
+
"loss": 1.8364,
|
| 9977 |
+
"step": 21360
|
| 9978 |
+
},
|
| 9979 |
+
{
|
| 9980 |
+
"epoch": 0.041624636332124675,
|
| 9981 |
+
"grad_norm": 3.484539031982422,
|
| 9982 |
+
"learning_rate": 2.9871931315426094e-05,
|
| 9983 |
+
"loss": 1.7097,
|
| 9984 |
+
"step": 21375
|
| 9985 |
+
},
|
| 9986 |
+
{
|
| 9987 |
+
"epoch": 0.04165384660323494,
|
| 9988 |
+
"grad_norm": 3.105286121368408,
|
| 9989 |
+
"learning_rate": 2.98717517634979e-05,
|
| 9990 |
+
"loss": 1.8239,
|
| 9991 |
+
"step": 21390
|
| 9992 |
+
},
|
| 9993 |
+
{
|
| 9994 |
+
"epoch": 0.0416830568743452,
|
| 9995 |
+
"grad_norm": 3.804901361465454,
|
| 9996 |
+
"learning_rate": 2.9871572086332594e-05,
|
| 9997 |
+
"loss": 1.7356,
|
| 9998 |
+
"step": 21405
|
| 9999 |
+
},
|
| 10000 |
+
{
|
| 10001 |
+
"epoch": 0.041712267145455464,
|
| 10002 |
+
"grad_norm": 2.4536283016204834,
|
| 10003 |
+
"learning_rate": 2.9871392283931686e-05,
|
| 10004 |
+
"loss": 1.8601,
|
| 10005 |
+
"step": 21420
|
| 10006 |
+
},
|
| 10007 |
+
{
|
| 10008 |
+
"epoch": 0.04174147741656573,
|
| 10009 |
+
"grad_norm": 2.8864688873291016,
|
| 10010 |
+
"learning_rate": 2.9871212356296697e-05,
|
| 10011 |
+
"loss": 1.802,
|
| 10012 |
+
"step": 21435
|
| 10013 |
+
},
|
| 10014 |
+
{
|
| 10015 |
+
"epoch": 0.04177068768767599,
|
| 10016 |
+
"grad_norm": 3.4735238552093506,
|
| 10017 |
+
"learning_rate": 2.9871032303429133e-05,
|
| 10018 |
+
"loss": 1.9557,
|
| 10019 |
+
"step": 21450
|
| 10020 |
+
},
|
| 10021 |
+
{
|
| 10022 |
+
"epoch": 0.041799897958786254,
|
| 10023 |
+
"grad_norm": 4.21823263168335,
|
| 10024 |
+
"learning_rate": 2.9870852125330513e-05,
|
| 10025 |
+
"loss": 1.8513,
|
| 10026 |
+
"step": 21465
|
| 10027 |
+
},
|
| 10028 |
+
{
|
| 10029 |
+
"epoch": 0.041829108229896515,
|
| 10030 |
+
"grad_norm": 1.6568303108215332,
|
| 10031 |
+
"learning_rate": 2.9870671822002357e-05,
|
| 10032 |
+
"loss": 1.8443,
|
| 10033 |
+
"step": 21480
|
| 10034 |
+
},
|
| 10035 |
+
{
|
| 10036 |
+
"epoch": 0.04185831850100678,
|
| 10037 |
+
"grad_norm": 3.8471972942352295,
|
| 10038 |
+
"learning_rate": 2.9870491393446184e-05,
|
| 10039 |
+
"loss": 1.8831,
|
| 10040 |
+
"step": 21495
|
| 10041 |
+
},
|
| 10042 |
+
{
|
| 10043 |
+
"epoch": 0.041887528772117044,
|
| 10044 |
+
"grad_norm": 4.5181803703308105,
|
| 10045 |
+
"learning_rate": 2.987031083966351e-05,
|
| 10046 |
+
"loss": 2.0743,
|
| 10047 |
+
"step": 21510
|
| 10048 |
+
},
|
| 10049 |
+
{
|
| 10050 |
+
"epoch": 0.041916739043227305,
|
| 10051 |
+
"grad_norm": 2.7604262828826904,
|
| 10052 |
+
"learning_rate": 2.987013016065586e-05,
|
| 10053 |
+
"loss": 1.7647,
|
| 10054 |
+
"step": 21525
|
| 10055 |
+
},
|
| 10056 |
+
{
|
| 10057 |
+
"epoch": 0.041945949314337566,
|
| 10058 |
+
"grad_norm": 1.6995900869369507,
|
| 10059 |
+
"learning_rate": 2.986994935642475e-05,
|
| 10060 |
+
"loss": 1.9221,
|
| 10061 |
+
"step": 21540
|
| 10062 |
+
},
|
| 10063 |
+
{
|
| 10064 |
+
"epoch": 0.041975159585447834,
|
| 10065 |
+
"grad_norm": 3.400073766708374,
|
| 10066 |
+
"learning_rate": 2.9869768426971706e-05,
|
| 10067 |
+
"loss": 1.7217,
|
| 10068 |
+
"step": 21555
|
| 10069 |
+
},
|
| 10070 |
+
{
|
| 10071 |
+
"epoch": 0.042004369856558095,
|
| 10072 |
+
"grad_norm": 2.7783281803131104,
|
| 10073 |
+
"learning_rate": 2.986958737229825e-05,
|
| 10074 |
+
"loss": 1.8149,
|
| 10075 |
+
"step": 21570
|
| 10076 |
+
},
|
| 10077 |
+
{
|
| 10078 |
+
"epoch": 0.042033580127668356,
|
| 10079 |
+
"grad_norm": 1.9334073066711426,
|
| 10080 |
+
"learning_rate": 2.9869406192405904e-05,
|
| 10081 |
+
"loss": 1.7657,
|
| 10082 |
+
"step": 21585
|
| 10083 |
+
},
|
| 10084 |
+
{
|
| 10085 |
+
"epoch": 0.042062790398778624,
|
| 10086 |
+
"grad_norm": 3.244271993637085,
|
| 10087 |
+
"learning_rate": 2.9869224887296205e-05,
|
| 10088 |
+
"loss": 1.8122,
|
| 10089 |
+
"step": 21600
|
| 10090 |
+
},
|
| 10091 |
+
{
|
| 10092 |
+
"epoch": 0.042092000669888885,
|
| 10093 |
+
"grad_norm": 3.9582252502441406,
|
| 10094 |
+
"learning_rate": 2.9869043456970662e-05,
|
| 10095 |
+
"loss": 1.8296,
|
| 10096 |
+
"step": 21615
|
| 10097 |
+
},
|
| 10098 |
+
{
|
| 10099 |
+
"epoch": 0.042121210940999146,
|
| 10100 |
+
"grad_norm": 3.9118635654449463,
|
| 10101 |
+
"learning_rate": 2.986886190143082e-05,
|
| 10102 |
+
"loss": 1.9162,
|
| 10103 |
+
"step": 21630
|
| 10104 |
+
},
|
| 10105 |
+
{
|
| 10106 |
+
"epoch": 0.04215042121210941,
|
| 10107 |
+
"grad_norm": 3.3247733116149902,
|
| 10108 |
+
"learning_rate": 2.9868680220678198e-05,
|
| 10109 |
+
"loss": 1.9345,
|
| 10110 |
+
"step": 21645
|
| 10111 |
+
},
|
| 10112 |
+
{
|
| 10113 |
+
"epoch": 0.042179631483219675,
|
| 10114 |
+
"grad_norm": 4.189952373504639,
|
| 10115 |
+
"learning_rate": 2.9868498414714332e-05,
|
| 10116 |
+
"loss": 1.7394,
|
| 10117 |
+
"step": 21660
|
| 10118 |
+
},
|
| 10119 |
+
{
|
| 10120 |
+
"epoch": 0.042208841754329936,
|
| 10121 |
+
"grad_norm": 4.380448818206787,
|
| 10122 |
+
"learning_rate": 2.986831648354075e-05,
|
| 10123 |
+
"loss": 1.8411,
|
| 10124 |
+
"step": 21675
|
| 10125 |
+
},
|
| 10126 |
+
{
|
| 10127 |
+
"epoch": 0.0422380520254402,
|
| 10128 |
+
"grad_norm": 3.3280911445617676,
|
| 10129 |
+
"learning_rate": 2.986813442715898e-05,
|
| 10130 |
+
"loss": 1.7338,
|
| 10131 |
+
"step": 21690
|
| 10132 |
+
},
|
| 10133 |
+
{
|
| 10134 |
+
"epoch": 0.042267262296550465,
|
| 10135 |
+
"grad_norm": 2.147976875305176,
|
| 10136 |
+
"learning_rate": 2.9867952245570557e-05,
|
| 10137 |
+
"loss": 1.7418,
|
| 10138 |
+
"step": 21705
|
| 10139 |
+
},
|
| 10140 |
+
{
|
| 10141 |
+
"epoch": 0.042296472567660726,
|
| 10142 |
+
"grad_norm": 4.927828788757324,
|
| 10143 |
+
"learning_rate": 2.9867769938777025e-05,
|
| 10144 |
+
"loss": 1.9729,
|
| 10145 |
+
"step": 21720
|
| 10146 |
+
},
|
| 10147 |
+
{
|
| 10148 |
+
"epoch": 0.04232568283877099,
|
| 10149 |
+
"grad_norm": 2.1555793285369873,
|
| 10150 |
+
"learning_rate": 2.9867587506779903e-05,
|
| 10151 |
+
"loss": 1.9733,
|
| 10152 |
+
"step": 21735
|
| 10153 |
+
},
|
| 10154 |
+
{
|
| 10155 |
+
"epoch": 0.04235489310988125,
|
| 10156 |
+
"grad_norm": 4.826751232147217,
|
| 10157 |
+
"learning_rate": 2.986740494958074e-05,
|
| 10158 |
+
"loss": 1.9141,
|
| 10159 |
+
"step": 21750
|
| 10160 |
+
},
|
| 10161 |
+
{
|
| 10162 |
+
"epoch": 0.042384103380991515,
|
| 10163 |
+
"grad_norm": 2.62713360786438,
|
| 10164 |
+
"learning_rate": 2.986722226718107e-05,
|
| 10165 |
+
"loss": 1.7623,
|
| 10166 |
+
"step": 21765
|
| 10167 |
+
},
|
| 10168 |
+
{
|
| 10169 |
+
"epoch": 0.042413313652101776,
|
| 10170 |
+
"grad_norm": 2.9087536334991455,
|
| 10171 |
+
"learning_rate": 2.9867039459582422e-05,
|
| 10172 |
+
"loss": 2.0298,
|
| 10173 |
+
"step": 21780
|
| 10174 |
+
},
|
| 10175 |
+
{
|
| 10176 |
+
"epoch": 0.04244252392321204,
|
| 10177 |
+
"grad_norm": 3.2233824729919434,
|
| 10178 |
+
"learning_rate": 2.986685652678635e-05,
|
| 10179 |
+
"loss": 1.8063,
|
| 10180 |
+
"step": 21795
|
| 10181 |
+
},
|
| 10182 |
+
{
|
| 10183 |
+
"epoch": 0.042471734194322305,
|
| 10184 |
+
"grad_norm": 3.8226282596588135,
|
| 10185 |
+
"learning_rate": 2.9866673468794392e-05,
|
| 10186 |
+
"loss": 1.8173,
|
| 10187 |
+
"step": 21810
|
| 10188 |
+
},
|
| 10189 |
+
{
|
| 10190 |
+
"epoch": 0.042500944465432566,
|
| 10191 |
+
"grad_norm": 2.362210273742676,
|
| 10192 |
+
"learning_rate": 2.986649028560808e-05,
|
| 10193 |
+
"loss": 2.0107,
|
| 10194 |
+
"step": 21825
|
| 10195 |
+
},
|
| 10196 |
+
{
|
| 10197 |
+
"epoch": 0.04253015473654283,
|
| 10198 |
+
"grad_norm": 1.4686235189437866,
|
| 10199 |
+
"learning_rate": 2.9866306977228964e-05,
|
| 10200 |
+
"loss": 1.8339,
|
| 10201 |
+
"step": 21840
|
| 10202 |
+
},
|
| 10203 |
+
{
|
| 10204 |
+
"epoch": 0.04255936500765309,
|
| 10205 |
+
"grad_norm": 3.2174501419067383,
|
| 10206 |
+
"learning_rate": 2.9866123543658585e-05,
|
| 10207 |
+
"loss": 1.9372,
|
| 10208 |
+
"step": 21855
|
| 10209 |
+
},
|
| 10210 |
+
{
|
| 10211 |
+
"epoch": 0.042588575278763356,
|
| 10212 |
+
"grad_norm": 4.215010643005371,
|
| 10213 |
+
"learning_rate": 2.9865939984898494e-05,
|
| 10214 |
+
"loss": 1.7492,
|
| 10215 |
+
"step": 21870
|
| 10216 |
+
},
|
| 10217 |
+
{
|
| 10218 |
+
"epoch": 0.04261778554987362,
|
| 10219 |
+
"grad_norm": 6.015155792236328,
|
| 10220 |
+
"learning_rate": 2.9865756300950224e-05,
|
| 10221 |
+
"loss": 1.8146,
|
| 10222 |
+
"step": 21885
|
| 10223 |
+
},
|
| 10224 |
+
{
|
| 10225 |
+
"epoch": 0.04264699582098388,
|
| 10226 |
+
"grad_norm": 3.102923631668091,
|
| 10227 |
+
"learning_rate": 2.9865572491815336e-05,
|
| 10228 |
+
"loss": 1.9206,
|
| 10229 |
+
"step": 21900
|
| 10230 |
+
},
|
| 10231 |
+
{
|
| 10232 |
+
"epoch": 0.042676206092094146,
|
| 10233 |
+
"grad_norm": 3.5606796741485596,
|
| 10234 |
+
"learning_rate": 2.986538855749537e-05,
|
| 10235 |
+
"loss": 1.7442,
|
| 10236 |
+
"step": 21915
|
| 10237 |
+
},
|
| 10238 |
+
{
|
| 10239 |
+
"epoch": 0.04270541636320441,
|
| 10240 |
+
"grad_norm": 3.877696990966797,
|
| 10241 |
+
"learning_rate": 2.9865204497991874e-05,
|
| 10242 |
+
"loss": 1.8356,
|
| 10243 |
+
"step": 21930
|
| 10244 |
+
},
|
| 10245 |
+
{
|
| 10246 |
+
"epoch": 0.04273462663431467,
|
| 10247 |
+
"grad_norm": 2.7707462310791016,
|
| 10248 |
+
"learning_rate": 2.98650203133064e-05,
|
| 10249 |
+
"loss": 2.1114,
|
| 10250 |
+
"step": 21945
|
| 10251 |
+
},
|
| 10252 |
+
{
|
| 10253 |
+
"epoch": 0.04276383690542493,
|
| 10254 |
+
"grad_norm": 3.8471269607543945,
|
| 10255 |
+
"learning_rate": 2.9864836003440496e-05,
|
| 10256 |
+
"loss": 1.974,
|
| 10257 |
+
"step": 21960
|
| 10258 |
+
},
|
| 10259 |
+
{
|
| 10260 |
+
"epoch": 0.0427930471765352,
|
| 10261 |
+
"grad_norm": 4.265331268310547,
|
| 10262 |
+
"learning_rate": 2.9864651568395728e-05,
|
| 10263 |
+
"loss": 1.7851,
|
| 10264 |
+
"step": 21975
|
| 10265 |
+
},
|
| 10266 |
+
{
|
| 10267 |
+
"epoch": 0.04282225744764546,
|
| 10268 |
+
"grad_norm": 3.578641414642334,
|
| 10269 |
+
"learning_rate": 2.986446700817363e-05,
|
| 10270 |
+
"loss": 1.7996,
|
| 10271 |
+
"step": 21990
|
| 10272 |
+
},
|
| 10273 |
+
{
|
| 10274 |
+
"epoch": 0.04285146771875572,
|
| 10275 |
+
"grad_norm": 3.289726495742798,
|
| 10276 |
+
"learning_rate": 2.9864282322775768e-05,
|
| 10277 |
+
"loss": 1.7165,
|
| 10278 |
+
"step": 22005
|
| 10279 |
+
},
|
| 10280 |
+
{
|
| 10281 |
+
"epoch": 0.04288067798986598,
|
| 10282 |
+
"grad_norm": 1.883353590965271,
|
| 10283 |
+
"learning_rate": 2.986409751220369e-05,
|
| 10284 |
+
"loss": 1.8633,
|
| 10285 |
+
"step": 22020
|
| 10286 |
+
},
|
| 10287 |
+
{
|
| 10288 |
+
"epoch": 0.04290988826097625,
|
| 10289 |
+
"grad_norm": 1.7199418544769287,
|
| 10290 |
+
"learning_rate": 2.9863912576458955e-05,
|
| 10291 |
+
"loss": 1.7696,
|
| 10292 |
+
"step": 22035
|
| 10293 |
+
},
|
| 10294 |
+
{
|
| 10295 |
+
"epoch": 0.04293909853208651,
|
| 10296 |
+
"grad_norm": 3.445349931716919,
|
| 10297 |
+
"learning_rate": 2.986372751554313e-05,
|
| 10298 |
+
"loss": 1.8349,
|
| 10299 |
+
"step": 22050
|
| 10300 |
+
},
|
| 10301 |
+
{
|
| 10302 |
+
"epoch": 0.04296830880319677,
|
| 10303 |
+
"grad_norm": 2.827427387237549,
|
| 10304 |
+
"learning_rate": 2.986354232945776e-05,
|
| 10305 |
+
"loss": 1.8647,
|
| 10306 |
+
"step": 22065
|
| 10307 |
+
},
|
| 10308 |
+
{
|
| 10309 |
+
"epoch": 0.04299751907430704,
|
| 10310 |
+
"grad_norm": 2.4225332736968994,
|
| 10311 |
+
"learning_rate": 2.986335701820441e-05,
|
| 10312 |
+
"loss": 2.0951,
|
| 10313 |
+
"step": 22080
|
| 10314 |
+
},
|
| 10315 |
+
{
|
| 10316 |
+
"epoch": 0.0430267293454173,
|
| 10317 |
+
"grad_norm": 2.9156410694122314,
|
| 10318 |
+
"learning_rate": 2.986317158178464e-05,
|
| 10319 |
+
"loss": 1.7989,
|
| 10320 |
+
"step": 22095
|
| 10321 |
+
},
|
| 10322 |
+
{
|
| 10323 |
+
"epoch": 0.04305593961652756,
|
| 10324 |
+
"grad_norm": 2.3038766384124756,
|
| 10325 |
+
"learning_rate": 2.986298602020001e-05,
|
| 10326 |
+
"loss": 1.8867,
|
| 10327 |
+
"step": 22110
|
| 10328 |
+
},
|
| 10329 |
+
{
|
| 10330 |
+
"epoch": 0.04308514988763782,
|
| 10331 |
+
"grad_norm": 3.085129976272583,
|
| 10332 |
+
"learning_rate": 2.986280033345209e-05,
|
| 10333 |
+
"loss": 1.6969,
|
| 10334 |
+
"step": 22125
|
| 10335 |
+
},
|
| 10336 |
+
{
|
| 10337 |
+
"epoch": 0.04311436015874809,
|
| 10338 |
+
"grad_norm": 4.071490287780762,
|
| 10339 |
+
"learning_rate": 2.986261452154243e-05,
|
| 10340 |
+
"loss": 1.7865,
|
| 10341 |
+
"step": 22140
|
| 10342 |
+
},
|
| 10343 |
+
{
|
| 10344 |
+
"epoch": 0.04314357042985835,
|
| 10345 |
+
"grad_norm": 2.2542295455932617,
|
| 10346 |
+
"learning_rate": 2.986242858447261e-05,
|
| 10347 |
+
"loss": 1.8687,
|
| 10348 |
+
"step": 22155
|
| 10349 |
+
},
|
| 10350 |
+
{
|
| 10351 |
+
"epoch": 0.04317278070096861,
|
| 10352 |
+
"grad_norm": 3.3869330883026123,
|
| 10353 |
+
"learning_rate": 2.9862242522244183e-05,
|
| 10354 |
+
"loss": 1.8037,
|
| 10355 |
+
"step": 22170
|
| 10356 |
+
},
|
| 10357 |
+
{
|
| 10358 |
+
"epoch": 0.04320199097207888,
|
| 10359 |
+
"grad_norm": 3.7993576526641846,
|
| 10360 |
+
"learning_rate": 2.9862056334858727e-05,
|
| 10361 |
+
"loss": 1.8819,
|
| 10362 |
+
"step": 22185
|
| 10363 |
+
},
|
| 10364 |
+
{
|
| 10365 |
+
"epoch": 0.04323120124318914,
|
| 10366 |
+
"grad_norm": 3.019289255142212,
|
| 10367 |
+
"learning_rate": 2.9861870022317798e-05,
|
| 10368 |
+
"loss": 1.8932,
|
| 10369 |
+
"step": 22200
|
| 10370 |
+
},
|
| 10371 |
+
{
|
| 10372 |
+
"epoch": 0.0432604115142994,
|
| 10373 |
+
"grad_norm": 2.831664562225342,
|
| 10374 |
+
"learning_rate": 2.9861683584622976e-05,
|
| 10375 |
+
"loss": 1.8813,
|
| 10376 |
+
"step": 22215
|
| 10377 |
+
},
|
| 10378 |
+
{
|
| 10379 |
+
"epoch": 0.04328962178540966,
|
| 10380 |
+
"grad_norm": 4.9506754875183105,
|
| 10381 |
+
"learning_rate": 2.9861497021775825e-05,
|
| 10382 |
+
"loss": 1.7917,
|
| 10383 |
+
"step": 22230
|
| 10384 |
+
},
|
| 10385 |
+
{
|
| 10386 |
+
"epoch": 0.04331883205651993,
|
| 10387 |
+
"grad_norm": 2.384033203125,
|
| 10388 |
+
"learning_rate": 2.986131033377792e-05,
|
| 10389 |
+
"loss": 2.0226,
|
| 10390 |
+
"step": 22245
|
| 10391 |
+
},
|
| 10392 |
+
{
|
| 10393 |
+
"epoch": 0.04334804232763019,
|
| 10394 |
+
"grad_norm": 1.8661621809005737,
|
| 10395 |
+
"learning_rate": 2.9861123520630828e-05,
|
| 10396 |
+
"loss": 1.817,
|
| 10397 |
+
"step": 22260
|
| 10398 |
+
},
|
| 10399 |
+
{
|
| 10400 |
+
"epoch": 0.04337725259874045,
|
| 10401 |
+
"grad_norm": 3.5283803939819336,
|
| 10402 |
+
"learning_rate": 2.9860936582336123e-05,
|
| 10403 |
+
"loss": 1.806,
|
| 10404 |
+
"step": 22275
|
| 10405 |
+
},
|
| 10406 |
+
{
|
| 10407 |
+
"epoch": 0.04340646286985072,
|
| 10408 |
+
"grad_norm": 4.240048408508301,
|
| 10409 |
+
"learning_rate": 2.9860749518895386e-05,
|
| 10410 |
+
"loss": 1.7773,
|
| 10411 |
+
"step": 22290
|
| 10412 |
+
},
|
| 10413 |
+
{
|
| 10414 |
+
"epoch": 0.04343567314096098,
|
| 10415 |
+
"grad_norm": 3.9326212406158447,
|
| 10416 |
+
"learning_rate": 2.986056233031018e-05,
|
| 10417 |
+
"loss": 1.8494,
|
| 10418 |
+
"step": 22305
|
| 10419 |
+
},
|
| 10420 |
+
{
|
| 10421 |
+
"epoch": 0.04346488341207124,
|
| 10422 |
+
"grad_norm": 4.390437602996826,
|
| 10423 |
+
"learning_rate": 2.986037501658209e-05,
|
| 10424 |
+
"loss": 1.7608,
|
| 10425 |
+
"step": 22320
|
| 10426 |
+
},
|
| 10427 |
+
{
|
| 10428 |
+
"epoch": 0.0434940936831815,
|
| 10429 |
+
"grad_norm": 2.5305089950561523,
|
| 10430 |
+
"learning_rate": 2.98601875777127e-05,
|
| 10431 |
+
"loss": 1.7295,
|
| 10432 |
+
"step": 22335
|
| 10433 |
+
},
|
| 10434 |
+
{
|
| 10435 |
+
"epoch": 0.04352330395429177,
|
| 10436 |
+
"grad_norm": 2.4276509284973145,
|
| 10437 |
+
"learning_rate": 2.9860000013703576e-05,
|
| 10438 |
+
"loss": 2.0082,
|
| 10439 |
+
"step": 22350
|
| 10440 |
+
},
|
| 10441 |
+
{
|
| 10442 |
+
"epoch": 0.04355251422540203,
|
| 10443 |
+
"grad_norm": 3.4239320755004883,
|
| 10444 |
+
"learning_rate": 2.9859812324556298e-05,
|
| 10445 |
+
"loss": 1.6979,
|
| 10446 |
+
"step": 22365
|
| 10447 |
+
},
|
| 10448 |
+
{
|
| 10449 |
+
"epoch": 0.04358172449651229,
|
| 10450 |
+
"grad_norm": 2.544837474822998,
|
| 10451 |
+
"learning_rate": 2.985962451027245e-05,
|
| 10452 |
+
"loss": 1.7844,
|
| 10453 |
+
"step": 22380
|
| 10454 |
+
},
|
| 10455 |
+
{
|
| 10456 |
+
"epoch": 0.04361093476762256,
|
| 10457 |
+
"grad_norm": 5.094317436218262,
|
| 10458 |
+
"learning_rate": 2.985943657085362e-05,
|
| 10459 |
+
"loss": 1.8792,
|
| 10460 |
+
"step": 22395
|
| 10461 |
+
},
|
| 10462 |
+
{
|
| 10463 |
+
"epoch": 0.04364014503873282,
|
| 10464 |
+
"grad_norm": 3.235121250152588,
|
| 10465 |
+
"learning_rate": 2.985924850630138e-05,
|
| 10466 |
+
"loss": 1.7795,
|
| 10467 |
+
"step": 22410
|
| 10468 |
+
},
|
| 10469 |
+
{
|
| 10470 |
+
"epoch": 0.04366935530984308,
|
| 10471 |
+
"grad_norm": 4.88664436340332,
|
| 10472 |
+
"learning_rate": 2.9859060316617325e-05,
|
| 10473 |
+
"loss": 1.9035,
|
| 10474 |
+
"step": 22425
|
| 10475 |
+
},
|
| 10476 |
+
{
|
| 10477 |
+
"epoch": 0.04369856558095334,
|
| 10478 |
+
"grad_norm": 2.401301622390747,
|
| 10479 |
+
"learning_rate": 2.9858872001803025e-05,
|
| 10480 |
+
"loss": 1.9182,
|
| 10481 |
+
"step": 22440
|
| 10482 |
+
},
|
| 10483 |
+
{
|
| 10484 |
+
"epoch": 0.04372777585206361,
|
| 10485 |
+
"grad_norm": 2.24088978767395,
|
| 10486 |
+
"learning_rate": 2.9858683561860077e-05,
|
| 10487 |
+
"loss": 1.688,
|
| 10488 |
+
"step": 22455
|
| 10489 |
+
},
|
| 10490 |
+
{
|
| 10491 |
+
"epoch": 0.04375698612317387,
|
| 10492 |
+
"grad_norm": 2.521176338195801,
|
| 10493 |
+
"learning_rate": 2.9858494996790065e-05,
|
| 10494 |
+
"loss": 1.962,
|
| 10495 |
+
"step": 22470
|
| 10496 |
+
},
|
| 10497 |
+
{
|
| 10498 |
+
"epoch": 0.04378619639428413,
|
| 10499 |
+
"grad_norm": 2.897905111312866,
|
| 10500 |
+
"learning_rate": 2.9858306306594578e-05,
|
| 10501 |
+
"loss": 1.7388,
|
| 10502 |
+
"step": 22485
|
| 10503 |
+
},
|
| 10504 |
+
{
|
| 10505 |
+
"epoch": 0.0438154066653944,
|
| 10506 |
+
"grad_norm": 2.3301055431365967,
|
| 10507 |
+
"learning_rate": 2.9858117491275204e-05,
|
| 10508 |
+
"loss": 1.8834,
|
| 10509 |
+
"step": 22500
|
| 10510 |
+
},
|
| 10511 |
+
{
|
| 10512 |
+
"epoch": 0.04384461693650466,
|
| 10513 |
+
"grad_norm": 2.078205108642578,
|
| 10514 |
+
"learning_rate": 2.9857928550833533e-05,
|
| 10515 |
+
"loss": 1.8959,
|
| 10516 |
+
"step": 22515
|
| 10517 |
+
},
|
| 10518 |
+
{
|
| 10519 |
+
"epoch": 0.04387382720761492,
|
| 10520 |
+
"grad_norm": 2.8938355445861816,
|
| 10521 |
+
"learning_rate": 2.9857739485271153e-05,
|
| 10522 |
+
"loss": 1.8529,
|
| 10523 |
+
"step": 22530
|
| 10524 |
+
},
|
| 10525 |
+
{
|
| 10526 |
+
"epoch": 0.04390303747872518,
|
| 10527 |
+
"grad_norm": 2.0363752841949463,
|
| 10528 |
+
"learning_rate": 2.9857550294589663e-05,
|
| 10529 |
+
"loss": 1.9974,
|
| 10530 |
+
"step": 22545
|
| 10531 |
+
},
|
| 10532 |
+
{
|
| 10533 |
+
"epoch": 0.04393224774983545,
|
| 10534 |
+
"grad_norm": 3.58372163772583,
|
| 10535 |
+
"learning_rate": 2.9857360978790647e-05,
|
| 10536 |
+
"loss": 1.7921,
|
| 10537 |
+
"step": 22560
|
| 10538 |
+
},
|
| 10539 |
+
{
|
| 10540 |
+
"epoch": 0.04396145802094571,
|
| 10541 |
+
"grad_norm": 3.4086432456970215,
|
| 10542 |
+
"learning_rate": 2.985717153787571e-05,
|
| 10543 |
+
"loss": 1.8401,
|
| 10544 |
+
"step": 22575
|
| 10545 |
+
},
|
| 10546 |
+
{
|
| 10547 |
+
"epoch": 0.04399066829205597,
|
| 10548 |
+
"grad_norm": 2.3264150619506836,
|
| 10549 |
+
"learning_rate": 2.985698197184644e-05,
|
| 10550 |
+
"loss": 1.8789,
|
| 10551 |
+
"step": 22590
|
| 10552 |
+
},
|
| 10553 |
+
{
|
| 10554 |
+
"epoch": 0.044019878563166234,
|
| 10555 |
+
"grad_norm": 2.727571964263916,
|
| 10556 |
+
"learning_rate": 2.9856792280704435e-05,
|
| 10557 |
+
"loss": 1.7859,
|
| 10558 |
+
"step": 22605
|
| 10559 |
+
},
|
| 10560 |
+
{
|
| 10561 |
+
"epoch": 0.0440490888342765,
|
| 10562 |
+
"grad_norm": 3.833146095275879,
|
| 10563 |
+
"learning_rate": 2.9856602464451293e-05,
|
| 10564 |
+
"loss": 1.9976,
|
| 10565 |
+
"step": 22620
|
| 10566 |
+
},
|
| 10567 |
+
{
|
| 10568 |
+
"epoch": 0.04407829910538676,
|
| 10569 |
+
"grad_norm": 1.8543720245361328,
|
| 10570 |
+
"learning_rate": 2.9856412523088612e-05,
|
| 10571 |
+
"loss": 1.8875,
|
| 10572 |
+
"step": 22635
|
| 10573 |
+
},
|
| 10574 |
+
{
|
| 10575 |
+
"epoch": 0.044107509376497024,
|
| 10576 |
+
"grad_norm": 2.6003103256225586,
|
| 10577 |
+
"learning_rate": 2.9856222456617993e-05,
|
| 10578 |
+
"loss": 2.0006,
|
| 10579 |
+
"step": 22650
|
| 10580 |
+
},
|
| 10581 |
+
{
|
| 10582 |
+
"epoch": 0.04413671964760729,
|
| 10583 |
+
"grad_norm": 3.0136611461639404,
|
| 10584 |
+
"learning_rate": 2.9856032265041035e-05,
|
| 10585 |
+
"loss": 1.8309,
|
| 10586 |
+
"step": 22665
|
| 10587 |
+
},
|
| 10588 |
+
{
|
| 10589 |
+
"epoch": 0.04416592991871755,
|
| 10590 |
+
"grad_norm": 1.948042631149292,
|
| 10591 |
+
"learning_rate": 2.9855841948359337e-05,
|
| 10592 |
+
"loss": 1.8721,
|
| 10593 |
+
"step": 22680
|
| 10594 |
+
},
|
| 10595 |
+
{
|
| 10596 |
+
"epoch": 0.044195140189827814,
|
| 10597 |
+
"grad_norm": 3.1908979415893555,
|
| 10598 |
+
"learning_rate": 2.9855651506574507e-05,
|
| 10599 |
+
"loss": 1.8373,
|
| 10600 |
+
"step": 22695
|
| 10601 |
+
},
|
| 10602 |
+
{
|
| 10603 |
+
"epoch": 0.044224350460938075,
|
| 10604 |
+
"grad_norm": 3.519826889038086,
|
| 10605 |
+
"learning_rate": 2.985546093968815e-05,
|
| 10606 |
+
"loss": 1.8992,
|
| 10607 |
+
"step": 22710
|
| 10608 |
+
},
|
| 10609 |
+
{
|
| 10610 |
+
"epoch": 0.04425356073204834,
|
| 10611 |
+
"grad_norm": 4.259973526000977,
|
| 10612 |
+
"learning_rate": 2.985527024770186e-05,
|
| 10613 |
+
"loss": 1.9466,
|
| 10614 |
+
"step": 22725
|
| 10615 |
+
},
|
| 10616 |
+
{
|
| 10617 |
+
"epoch": 0.044282771003158604,
|
| 10618 |
+
"grad_norm": 2.69942569732666,
|
| 10619 |
+
"learning_rate": 2.9855079430617253e-05,
|
| 10620 |
+
"loss": 1.6805,
|
| 10621 |
+
"step": 22740
|
| 10622 |
+
},
|
| 10623 |
+
{
|
| 10624 |
+
"epoch": 0.044311981274268865,
|
| 10625 |
+
"grad_norm": 3.0210318565368652,
|
| 10626 |
+
"learning_rate": 2.9854888488435933e-05,
|
| 10627 |
+
"loss": 1.8744,
|
| 10628 |
+
"step": 22755
|
| 10629 |
+
},
|
| 10630 |
+
{
|
| 10631 |
+
"epoch": 0.04434119154537913,
|
| 10632 |
+
"grad_norm": 2.448091745376587,
|
| 10633 |
+
"learning_rate": 2.9854697421159505e-05,
|
| 10634 |
+
"loss": 1.8449,
|
| 10635 |
+
"step": 22770
|
| 10636 |
+
},
|
| 10637 |
+
{
|
| 10638 |
+
"epoch": 0.044370401816489394,
|
| 10639 |
+
"grad_norm": 3.8297231197357178,
|
| 10640 |
+
"learning_rate": 2.9854506228789586e-05,
|
| 10641 |
+
"loss": 1.8071,
|
| 10642 |
+
"step": 22785
|
| 10643 |
+
},
|
| 10644 |
+
{
|
| 10645 |
+
"epoch": 0.044399612087599655,
|
| 10646 |
+
"grad_norm": 4.065975189208984,
|
| 10647 |
+
"learning_rate": 2.9854314911327777e-05,
|
| 10648 |
+
"loss": 1.9335,
|
| 10649 |
+
"step": 22800
|
| 10650 |
+
},
|
| 10651 |
+
{
|
| 10652 |
+
"epoch": 0.044428822358709916,
|
| 10653 |
+
"grad_norm": 2.0296108722686768,
|
| 10654 |
+
"learning_rate": 2.9854123468775693e-05,
|
| 10655 |
+
"loss": 1.748,
|
| 10656 |
+
"step": 22815
|
| 10657 |
+
},
|
| 10658 |
+
{
|
| 10659 |
+
"epoch": 0.044458032629820184,
|
| 10660 |
+
"grad_norm": 1.7504347562789917,
|
| 10661 |
+
"learning_rate": 2.985393190113495e-05,
|
| 10662 |
+
"loss": 1.9084,
|
| 10663 |
+
"step": 22830
|
| 10664 |
+
},
|
| 10665 |
+
{
|
| 10666 |
+
"epoch": 0.044487242900930445,
|
| 10667 |
+
"grad_norm": 2.8802502155303955,
|
| 10668 |
+
"learning_rate": 2.9853740208407152e-05,
|
| 10669 |
+
"loss": 1.8649,
|
| 10670 |
+
"step": 22845
|
| 10671 |
+
},
|
| 10672 |
+
{
|
| 10673 |
+
"epoch": 0.044516453172040706,
|
| 10674 |
+
"grad_norm": 3.5376250743865967,
|
| 10675 |
+
"learning_rate": 2.985354839059392e-05,
|
| 10676 |
+
"loss": 1.7617,
|
| 10677 |
+
"step": 22860
|
| 10678 |
+
},
|
| 10679 |
+
{
|
| 10680 |
+
"epoch": 0.044545663443150973,
|
| 10681 |
+
"grad_norm": 4.846216201782227,
|
| 10682 |
+
"learning_rate": 2.985335644769687e-05,
|
| 10683 |
+
"loss": 1.7531,
|
| 10684 |
+
"step": 22875
|
| 10685 |
+
},
|
| 10686 |
+
{
|
| 10687 |
+
"epoch": 0.044574873714261234,
|
| 10688 |
+
"grad_norm": 1.9456549882888794,
|
| 10689 |
+
"learning_rate": 2.9853164379717615e-05,
|
| 10690 |
+
"loss": 1.9782,
|
| 10691 |
+
"step": 22890
|
| 10692 |
+
},
|
| 10693 |
+
{
|
| 10694 |
+
"epoch": 0.044604083985371495,
|
| 10695 |
+
"grad_norm": 3.7463254928588867,
|
| 10696 |
+
"learning_rate": 2.9852972186657774e-05,
|
| 10697 |
+
"loss": 1.9347,
|
| 10698 |
+
"step": 22905
|
| 10699 |
+
},
|
| 10700 |
+
{
|
| 10701 |
+
"epoch": 0.044633294256481756,
|
| 10702 |
+
"grad_norm": 1.9641201496124268,
|
| 10703 |
+
"learning_rate": 2.9852779868518967e-05,
|
| 10704 |
+
"loss": 1.8366,
|
| 10705 |
+
"step": 22920
|
| 10706 |
+
},
|
| 10707 |
+
{
|
| 10708 |
+
"epoch": 0.044662504527592024,
|
| 10709 |
+
"grad_norm": 3.9136605262756348,
|
| 10710 |
+
"learning_rate": 2.9852587425302812e-05,
|
| 10711 |
+
"loss": 2.0271,
|
| 10712 |
+
"step": 22935
|
| 10713 |
+
},
|
| 10714 |
+
{
|
| 10715 |
+
"epoch": 0.044691714798702285,
|
| 10716 |
+
"grad_norm": 2.161766767501831,
|
| 10717 |
+
"learning_rate": 2.9852394857010923e-05,
|
| 10718 |
+
"loss": 1.8681,
|
| 10719 |
+
"step": 22950
|
| 10720 |
+
},
|
| 10721 |
+
{
|
| 10722 |
+
"epoch": 0.044720925069812546,
|
| 10723 |
+
"grad_norm": 3.1569862365722656,
|
| 10724 |
+
"learning_rate": 2.9852202163644937e-05,
|
| 10725 |
+
"loss": 1.7996,
|
| 10726 |
+
"step": 22965
|
| 10727 |
+
},
|
| 10728 |
+
{
|
| 10729 |
+
"epoch": 0.044750135340922814,
|
| 10730 |
+
"grad_norm": 2.0586421489715576,
|
| 10731 |
+
"learning_rate": 2.9852009345206458e-05,
|
| 10732 |
+
"loss": 1.7727,
|
| 10733 |
+
"step": 22980
|
| 10734 |
+
},
|
| 10735 |
+
{
|
| 10736 |
+
"epoch": 0.044779345612033075,
|
| 10737 |
+
"grad_norm": 5.425686359405518,
|
| 10738 |
+
"learning_rate": 2.9851816401697127e-05,
|
| 10739 |
+
"loss": 1.9209,
|
| 10740 |
+
"step": 22995
|
| 10741 |
+
},
|
| 10742 |
+
{
|
| 10743 |
+
"epoch": 0.044808555883143336,
|
| 10744 |
+
"grad_norm": 5.385043621063232,
|
| 10745 |
+
"learning_rate": 2.985162333311856e-05,
|
| 10746 |
+
"loss": 1.8473,
|
| 10747 |
+
"step": 23010
|
| 10748 |
+
},
|
| 10749 |
+
{
|
| 10750 |
+
"epoch": 0.0448377661542536,
|
| 10751 |
+
"grad_norm": 3.4577736854553223,
|
| 10752 |
+
"learning_rate": 2.985143013947238e-05,
|
| 10753 |
+
"loss": 1.848,
|
| 10754 |
+
"step": 23025
|
| 10755 |
+
},
|
| 10756 |
+
{
|
| 10757 |
+
"epoch": 0.044866976425363865,
|
| 10758 |
+
"grad_norm": 3.0630152225494385,
|
| 10759 |
+
"learning_rate": 2.985123682076022e-05,
|
| 10760 |
+
"loss": 1.7272,
|
| 10761 |
+
"step": 23040
|
| 10762 |
+
},
|
| 10763 |
+
{
|
| 10764 |
+
"epoch": 0.044896186696474126,
|
| 10765 |
+
"grad_norm": 3.8338427543640137,
|
| 10766 |
+
"learning_rate": 2.985104337698371e-05,
|
| 10767 |
+
"loss": 1.9042,
|
| 10768 |
+
"step": 23055
|
| 10769 |
+
},
|
| 10770 |
+
{
|
| 10771 |
+
"epoch": 0.04492539696758439,
|
| 10772 |
+
"grad_norm": 3.1702969074249268,
|
| 10773 |
+
"learning_rate": 2.985084980814447e-05,
|
| 10774 |
+
"loss": 1.9389,
|
| 10775 |
+
"step": 23070
|
| 10776 |
+
},
|
| 10777 |
+
{
|
| 10778 |
+
"epoch": 0.04495460723869465,
|
| 10779 |
+
"grad_norm": 2.758162021636963,
|
| 10780 |
+
"learning_rate": 2.985065611424414e-05,
|
| 10781 |
+
"loss": 1.8825,
|
| 10782 |
+
"step": 23085
|
| 10783 |
+
},
|
| 10784 |
+
{
|
| 10785 |
+
"epoch": 0.044983817509804916,
|
| 10786 |
+
"grad_norm": 2.1373486518859863,
|
| 10787 |
+
"learning_rate": 2.985046229528434e-05,
|
| 10788 |
+
"loss": 1.9451,
|
| 10789 |
+
"step": 23100
|
| 10790 |
+
},
|
| 10791 |
+
{
|
| 10792 |
+
"epoch": 0.04501302778091518,
|
| 10793 |
+
"grad_norm": 2.1302478313446045,
|
| 10794 |
+
"learning_rate": 2.985026835126671e-05,
|
| 10795 |
+
"loss": 2.0208,
|
| 10796 |
+
"step": 23115
|
| 10797 |
+
},
|
| 10798 |
+
{
|
| 10799 |
+
"epoch": 0.04504223805202544,
|
| 10800 |
+
"grad_norm": 3.8777012825012207,
|
| 10801 |
+
"learning_rate": 2.985007428219289e-05,
|
| 10802 |
+
"loss": 2.0138,
|
| 10803 |
+
"step": 23130
|
| 10804 |
+
},
|
| 10805 |
+
{
|
| 10806 |
+
"epoch": 0.045071448323135706,
|
| 10807 |
+
"grad_norm": 3.428769111633301,
|
| 10808 |
+
"learning_rate": 2.9849880088064497e-05,
|
| 10809 |
+
"loss": 1.9868,
|
| 10810 |
+
"step": 23145
|
| 10811 |
+
},
|
| 10812 |
+
{
|
| 10813 |
+
"epoch": 0.04510065859424597,
|
| 10814 |
+
"grad_norm": 5.036014556884766,
|
| 10815 |
+
"learning_rate": 2.9849685768883172e-05,
|
| 10816 |
+
"loss": 1.7348,
|
| 10817 |
+
"step": 23160
|
| 10818 |
+
},
|
| 10819 |
+
{
|
| 10820 |
+
"epoch": 0.04512986886535623,
|
| 10821 |
+
"grad_norm": 3.206932306289673,
|
| 10822 |
+
"learning_rate": 2.9849491324650563e-05,
|
| 10823 |
+
"loss": 1.9991,
|
| 10824 |
+
"step": 23175
|
| 10825 |
+
},
|
| 10826 |
+
{
|
| 10827 |
+
"epoch": 0.04515907913646649,
|
| 10828 |
+
"grad_norm": 2.5311954021453857,
|
| 10829 |
+
"learning_rate": 2.9849296755368297e-05,
|
| 10830 |
+
"loss": 1.7501,
|
| 10831 |
+
"step": 23190
|
| 10832 |
+
},
|
| 10833 |
+
{
|
| 10834 |
+
"epoch": 0.04518828940757676,
|
| 10835 |
+
"grad_norm": 2.1587252616882324,
|
| 10836 |
+
"learning_rate": 2.984910206103801e-05,
|
| 10837 |
+
"loss": 1.7981,
|
| 10838 |
+
"step": 23205
|
| 10839 |
+
},
|
| 10840 |
+
{
|
| 10841 |
+
"epoch": 0.04521749967868702,
|
| 10842 |
+
"grad_norm": 2.198859453201294,
|
| 10843 |
+
"learning_rate": 2.984890724166135e-05,
|
| 10844 |
+
"loss": 1.9349,
|
| 10845 |
+
"step": 23220
|
| 10846 |
+
},
|
| 10847 |
+
{
|
| 10848 |
+
"epoch": 0.04524670994979728,
|
| 10849 |
+
"grad_norm": 3.513099193572998,
|
| 10850 |
+
"learning_rate": 2.9848712297239955e-05,
|
| 10851 |
+
"loss": 1.7639,
|
| 10852 |
+
"step": 23235
|
| 10853 |
+
},
|
| 10854 |
+
{
|
| 10855 |
+
"epoch": 0.045275920220907546,
|
| 10856 |
+
"grad_norm": 4.66519021987915,
|
| 10857 |
+
"learning_rate": 2.984851722777546e-05,
|
| 10858 |
+
"loss": 1.6865,
|
| 10859 |
+
"step": 23250
|
| 10860 |
+
},
|
| 10861 |
+
{
|
| 10862 |
+
"epoch": 0.04530513049201781,
|
| 10863 |
+
"grad_norm": 2.0970771312713623,
|
| 10864 |
+
"learning_rate": 2.9848322033269523e-05,
|
| 10865 |
+
"loss": 1.8861,
|
| 10866 |
+
"step": 23265
|
| 10867 |
+
},
|
| 10868 |
+
{
|
| 10869 |
+
"epoch": 0.04533434076312807,
|
| 10870 |
+
"grad_norm": 4.110560417175293,
|
| 10871 |
+
"learning_rate": 2.984812671372377e-05,
|
| 10872 |
+
"loss": 1.8482,
|
| 10873 |
+
"step": 23280
|
| 10874 |
+
},
|
| 10875 |
+
{
|
| 10876 |
+
"epoch": 0.04536355103423833,
|
| 10877 |
+
"grad_norm": 3.2896182537078857,
|
| 10878 |
+
"learning_rate": 2.9847931269139854e-05,
|
| 10879 |
+
"loss": 1.9021,
|
| 10880 |
+
"step": 23295
|
| 10881 |
+
},
|
| 10882 |
+
{
|
| 10883 |
+
"epoch": 0.0453927613053486,
|
| 10884 |
+
"grad_norm": 1.929777979850769,
|
| 10885 |
+
"learning_rate": 2.9847735699519423e-05,
|
| 10886 |
+
"loss": 1.9236,
|
| 10887 |
+
"step": 23310
|
| 10888 |
+
},
|
| 10889 |
+
{
|
| 10890 |
+
"epoch": 0.04542197157645886,
|
| 10891 |
+
"grad_norm": 2.989177703857422,
|
| 10892 |
+
"learning_rate": 2.9847540004864115e-05,
|
| 10893 |
+
"loss": 1.8256,
|
| 10894 |
+
"step": 23325
|
| 10895 |
+
},
|
| 10896 |
+
{
|
| 10897 |
+
"epoch": 0.04545118184756912,
|
| 10898 |
+
"grad_norm": 2.735318899154663,
|
| 10899 |
+
"learning_rate": 2.984734418517559e-05,
|
| 10900 |
+
"loss": 1.8279,
|
| 10901 |
+
"step": 23340
|
| 10902 |
+
},
|
| 10903 |
+
{
|
| 10904 |
+
"epoch": 0.04548039211867939,
|
| 10905 |
+
"grad_norm": 2.597491502761841,
|
| 10906 |
+
"learning_rate": 2.9847148240455495e-05,
|
| 10907 |
+
"loss": 1.8952,
|
| 10908 |
+
"step": 23355
|
| 10909 |
+
},
|
| 10910 |
+
{
|
| 10911 |
+
"epoch": 0.04550960238978965,
|
| 10912 |
+
"grad_norm": 3.5635156631469727,
|
| 10913 |
+
"learning_rate": 2.9846952170705473e-05,
|
| 10914 |
+
"loss": 1.7302,
|
| 10915 |
+
"step": 23370
|
| 10916 |
+
},
|
| 10917 |
+
{
|
| 10918 |
+
"epoch": 0.04553881266089991,
|
| 10919 |
+
"grad_norm": 6.853522777557373,
|
| 10920 |
+
"learning_rate": 2.9846755975927174e-05,
|
| 10921 |
+
"loss": 2.0572,
|
| 10922 |
+
"step": 23385
|
| 10923 |
+
},
|
| 10924 |
+
{
|
| 10925 |
+
"epoch": 0.04556802293201017,
|
| 10926 |
+
"grad_norm": 2.006582498550415,
|
| 10927 |
+
"learning_rate": 2.984655965612226e-05,
|
| 10928 |
+
"loss": 1.939,
|
| 10929 |
+
"step": 23400
|
| 10930 |
+
},
|
| 10931 |
+
{
|
| 10932 |
+
"epoch": 0.04559723320312044,
|
| 10933 |
+
"grad_norm": 3.6737654209136963,
|
| 10934 |
+
"learning_rate": 2.984636321129238e-05,
|
| 10935 |
+
"loss": 1.8714,
|
| 10936 |
+
"step": 23415
|
| 10937 |
+
},
|
| 10938 |
+
{
|
| 10939 |
+
"epoch": 0.0456264434742307,
|
| 10940 |
+
"grad_norm": 2.5802881717681885,
|
| 10941 |
+
"learning_rate": 2.9846166641439185e-05,
|
| 10942 |
+
"loss": 1.8687,
|
| 10943 |
+
"step": 23430
|
| 10944 |
+
},
|
| 10945 |
+
{
|
| 10946 |
+
"epoch": 0.04565565374534096,
|
| 10947 |
+
"grad_norm": 3.748178005218506,
|
| 10948 |
+
"learning_rate": 2.984596994656433e-05,
|
| 10949 |
+
"loss": 1.9146,
|
| 10950 |
+
"step": 23445
|
| 10951 |
+
},
|
| 10952 |
+
{
|
| 10953 |
+
"epoch": 0.04568486401645123,
|
| 10954 |
+
"grad_norm": 2.5193021297454834,
|
| 10955 |
+
"learning_rate": 2.9845773126669475e-05,
|
| 10956 |
+
"loss": 1.8845,
|
| 10957 |
+
"step": 23460
|
| 10958 |
+
},
|
| 10959 |
+
{
|
| 10960 |
+
"epoch": 0.04571407428756149,
|
| 10961 |
+
"grad_norm": 2.3217403888702393,
|
| 10962 |
+
"learning_rate": 2.984557618175628e-05,
|
| 10963 |
+
"loss": 1.9149,
|
| 10964 |
+
"step": 23475
|
| 10965 |
+
},
|
| 10966 |
+
{
|
| 10967 |
+
"epoch": 0.04574328455867175,
|
| 10968 |
+
"grad_norm": 3.9094045162200928,
|
| 10969 |
+
"learning_rate": 2.98453791118264e-05,
|
| 10970 |
+
"loss": 1.8615,
|
| 10971 |
+
"step": 23490
|
| 10972 |
+
},
|
| 10973 |
+
{
|
| 10974 |
+
"epoch": 0.04577249482978201,
|
| 10975 |
+
"grad_norm": 2.190436840057373,
|
| 10976 |
+
"learning_rate": 2.9845181916881495e-05,
|
| 10977 |
+
"loss": 1.8181,
|
| 10978 |
+
"step": 23505
|
| 10979 |
+
},
|
| 10980 |
+
{
|
| 10981 |
+
"epoch": 0.04580170510089228,
|
| 10982 |
+
"grad_norm": 3.33785343170166,
|
| 10983 |
+
"learning_rate": 2.984498459692322e-05,
|
| 10984 |
+
"loss": 1.9345,
|
| 10985 |
+
"step": 23520
|
| 10986 |
+
},
|
| 10987 |
+
{
|
| 10988 |
+
"epoch": 0.04583091537200254,
|
| 10989 |
+
"grad_norm": 4.065570831298828,
|
| 10990 |
+
"learning_rate": 2.9844787151953242e-05,
|
| 10991 |
+
"loss": 1.7801,
|
| 10992 |
+
"step": 23535
|
| 10993 |
+
},
|
| 10994 |
+
{
|
| 10995 |
+
"epoch": 0.0458601256431128,
|
| 10996 |
+
"grad_norm": 3.4585118293762207,
|
| 10997 |
+
"learning_rate": 2.984458958197323e-05,
|
| 10998 |
+
"loss": 1.8181,
|
| 10999 |
+
"step": 23550
|
| 11000 |
+
},
|
| 11001 |
+
{
|
| 11002 |
+
"epoch": 0.04588933591422307,
|
| 11003 |
+
"grad_norm": 2.1615355014801025,
|
| 11004 |
+
"learning_rate": 2.984439188698484e-05,
|
| 11005 |
+
"loss": 1.8265,
|
| 11006 |
+
"step": 23565
|
| 11007 |
+
},
|
| 11008 |
+
{
|
| 11009 |
+
"epoch": 0.04591854618533333,
|
| 11010 |
+
"grad_norm": 3.9168038368225098,
|
| 11011 |
+
"learning_rate": 2.9844194066989737e-05,
|
| 11012 |
+
"loss": 1.8508,
|
| 11013 |
+
"step": 23580
|
| 11014 |
+
},
|
| 11015 |
+
{
|
| 11016 |
+
"epoch": 0.04594775645644359,
|
| 11017 |
+
"grad_norm": 2.436638593673706,
|
| 11018 |
+
"learning_rate": 2.9843996121989587e-05,
|
| 11019 |
+
"loss": 1.9247,
|
| 11020 |
+
"step": 23595
|
| 11021 |
+
},
|
| 11022 |
+
{
|
| 11023 |
+
"epoch": 0.04597696672755385,
|
| 11024 |
+
"grad_norm": 3.637157678604126,
|
| 11025 |
+
"learning_rate": 2.9843798051986053e-05,
|
| 11026 |
+
"loss": 1.9791,
|
| 11027 |
+
"step": 23610
|
| 11028 |
+
},
|
| 11029 |
+
{
|
| 11030 |
+
"epoch": 0.04600617699866412,
|
| 11031 |
+
"grad_norm": 2.9777755737304688,
|
| 11032 |
+
"learning_rate": 2.9843599856980815e-05,
|
| 11033 |
+
"loss": 1.8227,
|
| 11034 |
+
"step": 23625
|
| 11035 |
+
},
|
| 11036 |
+
{
|
| 11037 |
+
"epoch": 0.04603538726977438,
|
| 11038 |
+
"grad_norm": 5.862057209014893,
|
| 11039 |
+
"learning_rate": 2.9843401536975533e-05,
|
| 11040 |
+
"loss": 1.9745,
|
| 11041 |
+
"step": 23640
|
| 11042 |
+
},
|
| 11043 |
+
{
|
| 11044 |
+
"epoch": 0.04606459754088464,
|
| 11045 |
+
"grad_norm": 2.937347173690796,
|
| 11046 |
+
"learning_rate": 2.9843203091971878e-05,
|
| 11047 |
+
"loss": 1.7575,
|
| 11048 |
+
"step": 23655
|
| 11049 |
+
},
|
| 11050 |
+
{
|
| 11051 |
+
"epoch": 0.0460938078119949,
|
| 11052 |
+
"grad_norm": 3.7695438861846924,
|
| 11053 |
+
"learning_rate": 2.984300452197152e-05,
|
| 11054 |
+
"loss": 1.9655,
|
| 11055 |
+
"step": 23670
|
| 11056 |
+
},
|
| 11057 |
+
{
|
| 11058 |
+
"epoch": 0.04612301808310517,
|
| 11059 |
+
"grad_norm": 3.1122677326202393,
|
| 11060 |
+
"learning_rate": 2.9842805826976137e-05,
|
| 11061 |
+
"loss": 2.0459,
|
| 11062 |
+
"step": 23685
|
| 11063 |
+
},
|
| 11064 |
+
{
|
| 11065 |
+
"epoch": 0.04615222835421543,
|
| 11066 |
+
"grad_norm": 4.503921031951904,
|
| 11067 |
+
"learning_rate": 2.98426070069874e-05,
|
| 11068 |
+
"loss": 1.8319,
|
| 11069 |
+
"step": 23700
|
| 11070 |
+
},
|
| 11071 |
+
{
|
| 11072 |
+
"epoch": 0.04618143862532569,
|
| 11073 |
+
"grad_norm": 2.247570514678955,
|
| 11074 |
+
"learning_rate": 2.9842408062006982e-05,
|
| 11075 |
+
"loss": 1.765,
|
| 11076 |
+
"step": 23715
|
| 11077 |
+
},
|
| 11078 |
+
{
|
| 11079 |
+
"epoch": 0.04621064889643596,
|
| 11080 |
+
"grad_norm": 3.512427806854248,
|
| 11081 |
+
"learning_rate": 2.9842208992036554e-05,
|
| 11082 |
+
"loss": 1.8118,
|
| 11083 |
+
"step": 23730
|
| 11084 |
+
},
|
| 11085 |
+
{
|
| 11086 |
+
"epoch": 0.04623985916754622,
|
| 11087 |
+
"grad_norm": 3.174893379211426,
|
| 11088 |
+
"learning_rate": 2.98420097970778e-05,
|
| 11089 |
+
"loss": 1.8791,
|
| 11090 |
+
"step": 23745
|
| 11091 |
+
},
|
| 11092 |
+
{
|
| 11093 |
+
"epoch": 0.04626906943865648,
|
| 11094 |
+
"grad_norm": 5.151320934295654,
|
| 11095 |
+
"learning_rate": 2.9841810477132392e-05,
|
| 11096 |
+
"loss": 1.817,
|
| 11097 |
+
"step": 23760
|
| 11098 |
+
},
|
| 11099 |
+
{
|
| 11100 |
+
"epoch": 0.04629827970976674,
|
| 11101 |
+
"grad_norm": 2.6584534645080566,
|
| 11102 |
+
"learning_rate": 2.984161103220201e-05,
|
| 11103 |
+
"loss": 1.8441,
|
| 11104 |
+
"step": 23775
|
| 11105 |
+
},
|
| 11106 |
+
{
|
| 11107 |
+
"epoch": 0.04632748998087701,
|
| 11108 |
+
"grad_norm": 1.967596173286438,
|
| 11109 |
+
"learning_rate": 2.9841411462288335e-05,
|
| 11110 |
+
"loss": 2.0349,
|
| 11111 |
+
"step": 23790
|
| 11112 |
+
},
|
| 11113 |
+
{
|
| 11114 |
+
"epoch": 0.04635670025198727,
|
| 11115 |
+
"grad_norm": 3.7256250381469727,
|
| 11116 |
+
"learning_rate": 2.9841211767393048e-05,
|
| 11117 |
+
"loss": 1.8131,
|
| 11118 |
+
"step": 23805
|
| 11119 |
+
},
|
| 11120 |
+
{
|
| 11121 |
+
"epoch": 0.04638591052309753,
|
| 11122 |
+
"grad_norm": 4.081748962402344,
|
| 11123 |
+
"learning_rate": 2.9841011947517826e-05,
|
| 11124 |
+
"loss": 1.8664,
|
| 11125 |
+
"step": 23820
|
| 11126 |
+
},
|
| 11127 |
+
{
|
| 11128 |
+
"epoch": 0.0464151207942078,
|
| 11129 |
+
"grad_norm": 2.527451276779175,
|
| 11130 |
+
"learning_rate": 2.984081200266436e-05,
|
| 11131 |
+
"loss": 1.8916,
|
| 11132 |
+
"step": 23835
|
| 11133 |
+
},
|
| 11134 |
+
{
|
| 11135 |
+
"epoch": 0.04644433106531806,
|
| 11136 |
+
"grad_norm": 2.159346103668213,
|
| 11137 |
+
"learning_rate": 2.9840611932834326e-05,
|
| 11138 |
+
"loss": 1.9254,
|
| 11139 |
+
"step": 23850
|
| 11140 |
+
},
|
| 11141 |
+
{
|
| 11142 |
+
"epoch": 0.04647354133642832,
|
| 11143 |
+
"grad_norm": 2.636519432067871,
|
| 11144 |
+
"learning_rate": 2.9840411738029412e-05,
|
| 11145 |
+
"loss": 1.8062,
|
| 11146 |
+
"step": 23865
|
| 11147 |
+
},
|
| 11148 |
+
{
|
| 11149 |
+
"epoch": 0.046502751607538584,
|
| 11150 |
+
"grad_norm": 2.3750388622283936,
|
| 11151 |
+
"learning_rate": 2.9840211418251303e-05,
|
| 11152 |
+
"loss": 1.7347,
|
| 11153 |
+
"step": 23880
|
| 11154 |
+
},
|
| 11155 |
+
{
|
| 11156 |
+
"epoch": 0.04653196187864885,
|
| 11157 |
+
"grad_norm": 2.181845188140869,
|
| 11158 |
+
"learning_rate": 2.9840010973501685e-05,
|
| 11159 |
+
"loss": 1.9207,
|
| 11160 |
+
"step": 23895
|
| 11161 |
+
},
|
| 11162 |
+
{
|
| 11163 |
+
"epoch": 0.04656117214975911,
|
| 11164 |
+
"grad_norm": 3.098555564880371,
|
| 11165 |
+
"learning_rate": 2.9839810403782252e-05,
|
| 11166 |
+
"loss": 1.9566,
|
| 11167 |
+
"step": 23910
|
| 11168 |
+
},
|
| 11169 |
+
{
|
| 11170 |
+
"epoch": 0.046590382420869374,
|
| 11171 |
+
"grad_norm": 5.142632484436035,
|
| 11172 |
+
"learning_rate": 2.9839609709094685e-05,
|
| 11173 |
+
"loss": 1.8871,
|
| 11174 |
+
"step": 23925
|
| 11175 |
+
},
|
| 11176 |
+
{
|
| 11177 |
+
"epoch": 0.04661959269197964,
|
| 11178 |
+
"grad_norm": 1.7162665128707886,
|
| 11179 |
+
"learning_rate": 2.9839408889440678e-05,
|
| 11180 |
+
"loss": 1.8784,
|
| 11181 |
+
"step": 23940
|
| 11182 |
+
},
|
| 11183 |
+
{
|
| 11184 |
+
"epoch": 0.0466488029630899,
|
| 11185 |
+
"grad_norm": 3.743401050567627,
|
| 11186 |
+
"learning_rate": 2.9839207944821925e-05,
|
| 11187 |
+
"loss": 1.9179,
|
| 11188 |
+
"step": 23955
|
| 11189 |
+
},
|
| 11190 |
+
{
|
| 11191 |
+
"epoch": 0.046678013234200164,
|
| 11192 |
+
"grad_norm": 2.952817440032959,
|
| 11193 |
+
"learning_rate": 2.983900687524011e-05,
|
| 11194 |
+
"loss": 1.9182,
|
| 11195 |
+
"step": 23970
|
| 11196 |
+
},
|
| 11197 |
+
{
|
| 11198 |
+
"epoch": 0.046707223505310425,
|
| 11199 |
+
"grad_norm": 2.3484249114990234,
|
| 11200 |
+
"learning_rate": 2.983880568069693e-05,
|
| 11201 |
+
"loss": 1.8336,
|
| 11202 |
+
"step": 23985
|
| 11203 |
+
},
|
| 11204 |
+
{
|
| 11205 |
+
"epoch": 0.04673643377642069,
|
| 11206 |
+
"grad_norm": 1.931921362876892,
|
| 11207 |
+
"learning_rate": 2.9838604361194087e-05,
|
| 11208 |
+
"loss": 1.771,
|
| 11209 |
+
"step": 24000
|
| 11210 |
+
},
|
| 11211 |
+
{
|
| 11212 |
+
"epoch": 0.04676564404753095,
|
| 11213 |
+
"grad_norm": 5.197054386138916,
|
| 11214 |
+
"learning_rate": 2.9838402916733263e-05,
|
| 11215 |
+
"loss": 1.8558,
|
| 11216 |
+
"step": 24015
|
| 11217 |
+
},
|
| 11218 |
+
{
|
| 11219 |
+
"epoch": 0.046794854318641214,
|
| 11220 |
+
"grad_norm": 2.862427234649658,
|
| 11221 |
+
"learning_rate": 2.9838201347316164e-05,
|
| 11222 |
+
"loss": 1.6956,
|
| 11223 |
+
"step": 24030
|
| 11224 |
+
},
|
| 11225 |
+
{
|
| 11226 |
+
"epoch": 0.04682406458975148,
|
| 11227 |
+
"grad_norm": 2.6003293991088867,
|
| 11228 |
+
"learning_rate": 2.9837999652944487e-05,
|
| 11229 |
+
"loss": 2.0473,
|
| 11230 |
+
"step": 24045
|
| 11231 |
+
},
|
| 11232 |
+
{
|
| 11233 |
+
"epoch": 0.04685327486086174,
|
| 11234 |
+
"grad_norm": 2.068455934524536,
|
| 11235 |
+
"learning_rate": 2.9837797833619926e-05,
|
| 11236 |
+
"loss": 1.9233,
|
| 11237 |
+
"step": 24060
|
| 11238 |
+
},
|
| 11239 |
+
{
|
| 11240 |
+
"epoch": 0.046882485131972004,
|
| 11241 |
+
"grad_norm": 3.4722557067871094,
|
| 11242 |
+
"learning_rate": 2.983759588934418e-05,
|
| 11243 |
+
"loss": 2.0052,
|
| 11244 |
+
"step": 24075
|
| 11245 |
+
},
|
| 11246 |
+
{
|
| 11247 |
+
"epoch": 0.046911695403082265,
|
| 11248 |
+
"grad_norm": 3.198732852935791,
|
| 11249 |
+
"learning_rate": 2.9837393820118954e-05,
|
| 11250 |
+
"loss": 1.906,
|
| 11251 |
+
"step": 24090
|
| 11252 |
+
},
|
| 11253 |
+
{
|
| 11254 |
+
"epoch": 0.04694090567419253,
|
| 11255 |
+
"grad_norm": 2.5050504207611084,
|
| 11256 |
+
"learning_rate": 2.983719162594595e-05,
|
| 11257 |
+
"loss": 1.833,
|
| 11258 |
+
"step": 24105
|
| 11259 |
+
},
|
| 11260 |
+
{
|
| 11261 |
+
"epoch": 0.046970115945302794,
|
| 11262 |
+
"grad_norm": 2.1438305377960205,
|
| 11263 |
+
"learning_rate": 2.9836989306826866e-05,
|
| 11264 |
+
"loss": 1.7439,
|
| 11265 |
+
"step": 24120
|
| 11266 |
+
},
|
| 11267 |
+
{
|
| 11268 |
+
"epoch": 0.046999326216413055,
|
| 11269 |
+
"grad_norm": 1.5980552434921265,
|
| 11270 |
+
"learning_rate": 2.983678686276341e-05,
|
| 11271 |
+
"loss": 1.8018,
|
| 11272 |
+
"step": 24135
|
| 11273 |
+
},
|
| 11274 |
+
{
|
| 11275 |
+
"epoch": 0.047028536487523316,
|
| 11276 |
+
"grad_norm": 2.566882610321045,
|
| 11277 |
+
"learning_rate": 2.9836584293757282e-05,
|
| 11278 |
+
"loss": 1.8807,
|
| 11279 |
+
"step": 24150
|
| 11280 |
+
},
|
| 11281 |
+
{
|
| 11282 |
+
"epoch": 0.047057746758633584,
|
| 11283 |
+
"grad_norm": 4.125818252563477,
|
| 11284 |
+
"learning_rate": 2.9836381599810196e-05,
|
| 11285 |
+
"loss": 1.8198,
|
| 11286 |
+
"step": 24165
|
| 11287 |
+
},
|
| 11288 |
+
{
|
| 11289 |
+
"epoch": 0.047086957029743845,
|
| 11290 |
+
"grad_norm": 3.1217780113220215,
|
| 11291 |
+
"learning_rate": 2.983617878092385e-05,
|
| 11292 |
+
"loss": 2.0202,
|
| 11293 |
+
"step": 24180
|
| 11294 |
+
},
|
| 11295 |
+
{
|
| 11296 |
+
"epoch": 0.047116167300854106,
|
| 11297 |
+
"grad_norm": 3.001835823059082,
|
| 11298 |
+
"learning_rate": 2.9835975837099956e-05,
|
| 11299 |
+
"loss": 1.851,
|
| 11300 |
+
"step": 24195
|
| 11301 |
+
},
|
| 11302 |
+
{
|
| 11303 |
+
"epoch": 0.047145377571964374,
|
| 11304 |
+
"grad_norm": 4.060051918029785,
|
| 11305 |
+
"learning_rate": 2.9835772768340225e-05,
|
| 11306 |
+
"loss": 1.8918,
|
| 11307 |
+
"step": 24210
|
| 11308 |
+
},
|
| 11309 |
+
{
|
| 11310 |
+
"epoch": 0.047174587843074635,
|
| 11311 |
+
"grad_norm": 4.199372291564941,
|
| 11312 |
+
"learning_rate": 2.9835569574646363e-05,
|
| 11313 |
+
"loss": 1.848,
|
| 11314 |
+
"step": 24225
|
| 11315 |
+
},
|
| 11316 |
+
{
|
| 11317 |
+
"epoch": 0.047203798114184896,
|
| 11318 |
+
"grad_norm": 3.1068427562713623,
|
| 11319 |
+
"learning_rate": 2.9835366256020085e-05,
|
| 11320 |
+
"loss": 1.793,
|
| 11321 |
+
"step": 24240
|
| 11322 |
+
},
|
| 11323 |
+
{
|
| 11324 |
+
"epoch": 0.04723300838529516,
|
| 11325 |
+
"grad_norm": 3.30556321144104,
|
| 11326 |
+
"learning_rate": 2.9835162812463098e-05,
|
| 11327 |
+
"loss": 1.8914,
|
| 11328 |
+
"step": 24255
|
| 11329 |
+
},
|
| 11330 |
+
{
|
| 11331 |
+
"epoch": 0.047262218656405425,
|
| 11332 |
+
"grad_norm": 3.898911476135254,
|
| 11333 |
+
"learning_rate": 2.9834959243977123e-05,
|
| 11334 |
+
"loss": 1.929,
|
| 11335 |
+
"step": 24270
|
| 11336 |
+
},
|
| 11337 |
+
{
|
| 11338 |
+
"epoch": 0.047291428927515686,
|
| 11339 |
+
"grad_norm": 2.1169817447662354,
|
| 11340 |
+
"learning_rate": 2.9834755550563865e-05,
|
| 11341 |
+
"loss": 1.9513,
|
| 11342 |
+
"step": 24285
|
| 11343 |
+
},
|
| 11344 |
+
{
|
| 11345 |
+
"epoch": 0.04732063919862595,
|
| 11346 |
+
"grad_norm": 3.0540499687194824,
|
| 11347 |
+
"learning_rate": 2.9834551732225044e-05,
|
| 11348 |
+
"loss": 1.8283,
|
| 11349 |
+
"step": 24300
|
| 11350 |
+
},
|
| 11351 |
+
{
|
| 11352 |
+
"epoch": 0.047349849469736215,
|
| 11353 |
+
"grad_norm": 2.495234727859497,
|
| 11354 |
+
"learning_rate": 2.9834347788962383e-05,
|
| 11355 |
+
"loss": 1.8268,
|
| 11356 |
+
"step": 24315
|
| 11357 |
+
},
|
| 11358 |
+
{
|
| 11359 |
+
"epoch": 0.047379059740846476,
|
| 11360 |
+
"grad_norm": 4.254831790924072,
|
| 11361 |
+
"learning_rate": 2.9834143720777588e-05,
|
| 11362 |
+
"loss": 1.8933,
|
| 11363 |
+
"step": 24330
|
| 11364 |
+
},
|
| 11365 |
+
{
|
| 11366 |
+
"epoch": 0.04740827001195674,
|
| 11367 |
+
"grad_norm": 5.430483818054199,
|
| 11368 |
+
"learning_rate": 2.9833939527672384e-05,
|
| 11369 |
+
"loss": 1.8128,
|
| 11370 |
+
"step": 24345
|
| 11371 |
+
},
|
| 11372 |
+
{
|
| 11373 |
+
"epoch": 0.047437480283067,
|
| 11374 |
+
"grad_norm": 3.332515239715576,
|
| 11375 |
+
"learning_rate": 2.983373520964849e-05,
|
| 11376 |
+
"loss": 1.9089,
|
| 11377 |
+
"step": 24360
|
| 11378 |
+
},
|
| 11379 |
+
{
|
| 11380 |
+
"epoch": 0.047466690554177265,
|
| 11381 |
+
"grad_norm": 2.1588125228881836,
|
| 11382 |
+
"learning_rate": 2.983353076670762e-05,
|
| 11383 |
+
"loss": 1.7578,
|
| 11384 |
+
"step": 24375
|
| 11385 |
+
},
|
| 11386 |
+
{
|
| 11387 |
+
"epoch": 0.047495900825287526,
|
| 11388 |
+
"grad_norm": 2.6773619651794434,
|
| 11389 |
+
"learning_rate": 2.9833326198851503e-05,
|
| 11390 |
+
"loss": 1.9486,
|
| 11391 |
+
"step": 24390
|
| 11392 |
+
},
|
| 11393 |
+
{
|
| 11394 |
+
"epoch": 0.04752511109639779,
|
| 11395 |
+
"grad_norm": 2.855078935623169,
|
| 11396 |
+
"learning_rate": 2.9833121506081862e-05,
|
| 11397 |
+
"loss": 1.9339,
|
| 11398 |
+
"step": 24405
|
| 11399 |
+
},
|
| 11400 |
+
{
|
| 11401 |
+
"epoch": 0.047554321367508055,
|
| 11402 |
+
"grad_norm": 2.539210557937622,
|
| 11403 |
+
"learning_rate": 2.983291668840042e-05,
|
| 11404 |
+
"loss": 1.8479,
|
| 11405 |
+
"step": 24420
|
| 11406 |
+
},
|
| 11407 |
+
{
|
| 11408 |
+
"epoch": 0.047583531638618316,
|
| 11409 |
+
"grad_norm": 3.0605132579803467,
|
| 11410 |
+
"learning_rate": 2.9832711745808895e-05,
|
| 11411 |
+
"loss": 1.8924,
|
| 11412 |
+
"step": 24435
|
| 11413 |
+
},
|
| 11414 |
+
{
|
| 11415 |
+
"epoch": 0.04761274190972858,
|
| 11416 |
+
"grad_norm": 2.142998218536377,
|
| 11417 |
+
"learning_rate": 2.9832506678309025e-05,
|
| 11418 |
+
"loss": 2.0331,
|
| 11419 |
+
"step": 24450
|
| 11420 |
+
},
|
| 11421 |
+
{
|
| 11422 |
+
"epoch": 0.04764195218083884,
|
| 11423 |
+
"grad_norm": 3.082610845565796,
|
| 11424 |
+
"learning_rate": 2.983230148590253e-05,
|
| 11425 |
+
"loss": 1.8299,
|
| 11426 |
+
"step": 24465
|
| 11427 |
+
},
|
| 11428 |
+
{
|
| 11429 |
+
"epoch": 0.047671162451949106,
|
| 11430 |
+
"grad_norm": 4.011190414428711,
|
| 11431 |
+
"learning_rate": 2.9832096168591128e-05,
|
| 11432 |
+
"loss": 1.8591,
|
| 11433 |
+
"step": 24480
|
| 11434 |
+
},
|
| 11435 |
+
{
|
| 11436 |
+
"epoch": 0.04770037272305937,
|
| 11437 |
+
"grad_norm": 2.326756238937378,
|
| 11438 |
+
"learning_rate": 2.983189072637657e-05,
|
| 11439 |
+
"loss": 1.8687,
|
| 11440 |
+
"step": 24495
|
| 11441 |
+
},
|
| 11442 |
+
{
|
| 11443 |
+
"epoch": 0.04772958299416963,
|
| 11444 |
+
"grad_norm": 2.1641993522644043,
|
| 11445 |
+
"learning_rate": 2.9831685159260568e-05,
|
| 11446 |
+
"loss": 1.9402,
|
| 11447 |
+
"step": 24510
|
| 11448 |
+
},
|
| 11449 |
+
{
|
| 11450 |
+
"epoch": 0.047758793265279896,
|
| 11451 |
+
"grad_norm": 2.614429473876953,
|
| 11452 |
+
"learning_rate": 2.983147946724486e-05,
|
| 11453 |
+
"loss": 1.8911,
|
| 11454 |
+
"step": 24525
|
| 11455 |
+
},
|
| 11456 |
+
{
|
| 11457 |
+
"epoch": 0.04778800353639016,
|
| 11458 |
+
"grad_norm": 2.163760185241699,
|
| 11459 |
+
"learning_rate": 2.983127365033118e-05,
|
| 11460 |
+
"loss": 1.7623,
|
| 11461 |
+
"step": 24540
|
| 11462 |
+
},
|
| 11463 |
+
{
|
| 11464 |
+
"epoch": 0.04781721380750042,
|
| 11465 |
+
"grad_norm": 1.837815761566162,
|
| 11466 |
+
"learning_rate": 2.9831067708521257e-05,
|
| 11467 |
+
"loss": 2.0103,
|
| 11468 |
+
"step": 24555
|
| 11469 |
+
},
|
| 11470 |
+
{
|
| 11471 |
+
"epoch": 0.04784642407861068,
|
| 11472 |
+
"grad_norm": 4.629458904266357,
|
| 11473 |
+
"learning_rate": 2.9830861641816826e-05,
|
| 11474 |
+
"loss": 1.8473,
|
| 11475 |
+
"step": 24570
|
| 11476 |
+
},
|
| 11477 |
+
{
|
| 11478 |
+
"epoch": 0.04787563434972095,
|
| 11479 |
+
"grad_norm": 2.741942882537842,
|
| 11480 |
+
"learning_rate": 2.9830655450219623e-05,
|
| 11481 |
+
"loss": 1.8219,
|
| 11482 |
+
"step": 24585
|
| 11483 |
+
},
|
| 11484 |
+
{
|
| 11485 |
+
"epoch": 0.04790484462083121,
|
| 11486 |
+
"grad_norm": 2.6273906230926514,
|
| 11487 |
+
"learning_rate": 2.9830449133731387e-05,
|
| 11488 |
+
"loss": 1.7688,
|
| 11489 |
+
"step": 24600
|
| 11490 |
+
},
|
| 11491 |
+
{
|
| 11492 |
+
"epoch": 0.04793405489194147,
|
| 11493 |
+
"grad_norm": 2.2749216556549072,
|
| 11494 |
+
"learning_rate": 2.983024269235385e-05,
|
| 11495 |
+
"loss": 1.9809,
|
| 11496 |
+
"step": 24615
|
| 11497 |
+
},
|
| 11498 |
+
{
|
| 11499 |
+
"epoch": 0.04796326516305174,
|
| 11500 |
+
"grad_norm": 4.1970367431640625,
|
| 11501 |
+
"learning_rate": 2.9830036126088754e-05,
|
| 11502 |
+
"loss": 1.7998,
|
| 11503 |
+
"step": 24630
|
| 11504 |
+
},
|
| 11505 |
+
{
|
| 11506 |
+
"epoch": 0.047992475434162,
|
| 11507 |
+
"grad_norm": 2.87906813621521,
|
| 11508 |
+
"learning_rate": 2.982982943493784e-05,
|
| 11509 |
+
"loss": 1.8496,
|
| 11510 |
+
"step": 24645
|
| 11511 |
+
},
|
| 11512 |
+
{
|
| 11513 |
+
"epoch": 0.04802168570527226,
|
| 11514 |
+
"grad_norm": 1.9954204559326172,
|
| 11515 |
+
"learning_rate": 2.9829622618902848e-05,
|
| 11516 |
+
"loss": 1.8782,
|
| 11517 |
+
"step": 24660
|
| 11518 |
+
},
|
| 11519 |
+
{
|
| 11520 |
+
"epoch": 0.04805089597638252,
|
| 11521 |
+
"grad_norm": 3.3936777114868164,
|
| 11522 |
+
"learning_rate": 2.982941567798551e-05,
|
| 11523 |
+
"loss": 1.7,
|
| 11524 |
+
"step": 24675
|
| 11525 |
+
},
|
| 11526 |
+
{
|
| 11527 |
+
"epoch": 0.04808010624749279,
|
| 11528 |
+
"grad_norm": 5.30580997467041,
|
| 11529 |
+
"learning_rate": 2.9829208612187585e-05,
|
| 11530 |
+
"loss": 1.8712,
|
| 11531 |
+
"step": 24690
|
| 11532 |
+
},
|
| 11533 |
+
{
|
| 11534 |
+
"epoch": 0.04810931651860305,
|
| 11535 |
+
"grad_norm": 1.797080636024475,
|
| 11536 |
+
"learning_rate": 2.982900142151081e-05,
|
| 11537 |
+
"loss": 1.8711,
|
| 11538 |
+
"step": 24705
|
| 11539 |
+
},
|
| 11540 |
+
{
|
| 11541 |
+
"epoch": 0.04813852678971331,
|
| 11542 |
+
"grad_norm": 1.7338413000106812,
|
| 11543 |
+
"learning_rate": 2.9828794105956922e-05,
|
| 11544 |
+
"loss": 1.9125,
|
| 11545 |
+
"step": 24720
|
| 11546 |
+
},
|
| 11547 |
+
{
|
| 11548 |
+
"epoch": 0.04816773706082357,
|
| 11549 |
+
"grad_norm": 6.833633899688721,
|
| 11550 |
+
"learning_rate": 2.9828586665527677e-05,
|
| 11551 |
+
"loss": 1.8622,
|
| 11552 |
+
"step": 24735
|
| 11553 |
+
},
|
| 11554 |
+
{
|
| 11555 |
+
"epoch": 0.04819694733193384,
|
| 11556 |
+
"grad_norm": 2.8855645656585693,
|
| 11557 |
+
"learning_rate": 2.9828379100224814e-05,
|
| 11558 |
+
"loss": 1.7596,
|
| 11559 |
+
"step": 24750
|
| 11560 |
+
},
|
| 11561 |
+
{
|
| 11562 |
+
"epoch": 0.0482261576030441,
|
| 11563 |
+
"grad_norm": 3.7425286769866943,
|
| 11564 |
+
"learning_rate": 2.982817141005009e-05,
|
| 11565 |
+
"loss": 1.8452,
|
| 11566 |
+
"step": 24765
|
| 11567 |
+
},
|
| 11568 |
+
{
|
| 11569 |
+
"epoch": 0.04825536787415436,
|
| 11570 |
+
"grad_norm": 4.187405109405518,
|
| 11571 |
+
"learning_rate": 2.9827963595005248e-05,
|
| 11572 |
+
"loss": 1.8669,
|
| 11573 |
+
"step": 24780
|
| 11574 |
+
},
|
| 11575 |
+
{
|
| 11576 |
+
"epoch": 0.04828457814526463,
|
| 11577 |
+
"grad_norm": 4.476016998291016,
|
| 11578 |
+
"learning_rate": 2.982775565509204e-05,
|
| 11579 |
+
"loss": 1.9221,
|
| 11580 |
+
"step": 24795
|
| 11581 |
+
},
|
| 11582 |
+
{
|
| 11583 |
+
"epoch": 0.04831378841637489,
|
| 11584 |
+
"grad_norm": 3.6991500854492188,
|
| 11585 |
+
"learning_rate": 2.9827547590312213e-05,
|
| 11586 |
+
"loss": 1.5434,
|
| 11587 |
+
"step": 24810
|
| 11588 |
+
},
|
| 11589 |
+
{
|
| 11590 |
+
"epoch": 0.04834299868748515,
|
| 11591 |
+
"grad_norm": 3.0707576274871826,
|
| 11592 |
+
"learning_rate": 2.9827339400667524e-05,
|
| 11593 |
+
"loss": 1.88,
|
| 11594 |
+
"step": 24825
|
| 11595 |
+
},
|
| 11596 |
+
{
|
| 11597 |
+
"epoch": 0.04837220895859541,
|
| 11598 |
+
"grad_norm": 2.604163646697998,
|
| 11599 |
+
"learning_rate": 2.9827131086159723e-05,
|
| 11600 |
+
"loss": 1.9946,
|
| 11601 |
+
"step": 24840
|
| 11602 |
+
},
|
| 11603 |
+
{
|
| 11604 |
+
"epoch": 0.04840141922970568,
|
| 11605 |
+
"grad_norm": 4.7181549072265625,
|
| 11606 |
+
"learning_rate": 2.9826922646790568e-05,
|
| 11607 |
+
"loss": 1.8807,
|
| 11608 |
+
"step": 24855
|
| 11609 |
+
},
|
| 11610 |
+
{
|
| 11611 |
+
"epoch": 0.04843062950081594,
|
| 11612 |
+
"grad_norm": 3.983146905899048,
|
| 11613 |
+
"learning_rate": 2.9826714082561808e-05,
|
| 11614 |
+
"loss": 1.8593,
|
| 11615 |
+
"step": 24870
|
| 11616 |
+
},
|
| 11617 |
+
{
|
| 11618 |
+
"epoch": 0.0484598397719262,
|
| 11619 |
+
"grad_norm": 4.0342912673950195,
|
| 11620 |
+
"learning_rate": 2.982650539347521e-05,
|
| 11621 |
+
"loss": 1.9208,
|
| 11622 |
+
"step": 24885
|
| 11623 |
+
},
|
| 11624 |
+
{
|
| 11625 |
+
"epoch": 0.04848905004303647,
|
| 11626 |
+
"grad_norm": 3.092644691467285,
|
| 11627 |
+
"learning_rate": 2.982629657953252e-05,
|
| 11628 |
+
"loss": 1.9611,
|
| 11629 |
+
"step": 24900
|
| 11630 |
+
},
|
| 11631 |
+
{
|
| 11632 |
+
"epoch": 0.04851826031414673,
|
| 11633 |
+
"grad_norm": 2.244311809539795,
|
| 11634 |
+
"learning_rate": 2.98260876407355e-05,
|
| 11635 |
+
"loss": 1.7965,
|
| 11636 |
+
"step": 24915
|
| 11637 |
+
},
|
| 11638 |
+
{
|
| 11639 |
+
"epoch": 0.04854747058525699,
|
| 11640 |
+
"grad_norm": 4.740140914916992,
|
| 11641 |
+
"learning_rate": 2.9825878577085917e-05,
|
| 11642 |
+
"loss": 2.0418,
|
| 11643 |
+
"step": 24930
|
| 11644 |
+
},
|
| 11645 |
+
{
|
| 11646 |
+
"epoch": 0.04857668085636725,
|
| 11647 |
+
"grad_norm": 1.6131473779678345,
|
| 11648 |
+
"learning_rate": 2.9825669388585523e-05,
|
| 11649 |
+
"loss": 1.7781,
|
| 11650 |
+
"step": 24945
|
| 11651 |
+
},
|
| 11652 |
+
{
|
| 11653 |
+
"epoch": 0.04860589112747752,
|
| 11654 |
+
"grad_norm": 3.3024797439575195,
|
| 11655 |
+
"learning_rate": 2.9825460075236077e-05,
|
| 11656 |
+
"loss": 1.9176,
|
| 11657 |
+
"step": 24960
|
| 11658 |
+
},
|
| 11659 |
+
{
|
| 11660 |
+
"epoch": 0.04863510139858778,
|
| 11661 |
+
"grad_norm": 3.6958138942718506,
|
| 11662 |
+
"learning_rate": 2.9825250637039348e-05,
|
| 11663 |
+
"loss": 1.8613,
|
| 11664 |
+
"step": 24975
|
| 11665 |
+
},
|
| 11666 |
+
{
|
| 11667 |
+
"epoch": 0.04866431166969804,
|
| 11668 |
+
"grad_norm": 2.277597427368164,
|
| 11669 |
+
"learning_rate": 2.9825041073997102e-05,
|
| 11670 |
+
"loss": 1.9942,
|
| 11671 |
+
"step": 24990
|
| 11672 |
}
|
| 11673 |
],
|
| 11674 |
"logging_steps": 15,
|
|
|
|
| 11688 |
"attributes": {}
|
| 11689 |
}
|
| 11690 |
},
|
| 11691 |
+
"total_flos": 3.955653771590246e+16,
|
| 11692 |
"train_batch_size": 4,
|
| 11693 |
"trial_name": null,
|
| 11694 |
"trial_params": null
|