Training in progress, step 1373, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 27024
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:32d9dd50f89199a8d088db40af270ad72d1cb575a22e64c4d3ba386618485784
|
| 3 |
size 27024
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 64038
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b9405347f8e33cdbd1bf3081e2b7af9b4ca30ef3c19a49a55f53f52421aa35b0
|
| 3 |
size 64038
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:45a94c03e52871c883dee04b6aa48bf80d33bd04e8df6ff3892a707dcf722af8
|
| 3 |
size 14244
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1f589134abe94927f45df4ff22b731a3fe2caf6169bbe770d2041f613d81dcbc
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 344,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -7263,6 +7263,2393 @@
|
|
| 7263 |
"eval_samples_per_second": 266.35,
|
| 7264 |
"eval_steps_per_second": 133.175,
|
| 7265 |
"step": 1032
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7266 |
}
|
| 7267 |
],
|
| 7268 |
"logging_steps": 1,
|
|
@@ -7277,12 +9664,12 @@
|
|
| 7277 |
"should_evaluate": false,
|
| 7278 |
"should_log": false,
|
| 7279 |
"should_save": true,
|
| 7280 |
-
"should_training_stop":
|
| 7281 |
},
|
| 7282 |
"attributes": {}
|
| 7283 |
}
|
| 7284 |
},
|
| 7285 |
-
"total_flos":
|
| 7286 |
"train_batch_size": 2,
|
| 7287 |
"trial_name": null,
|
| 7288 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.1158527581478747,
|
| 5 |
"eval_steps": 344,
|
| 6 |
+
"global_step": 1373,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 7263 |
"eval_samples_per_second": 266.35,
|
| 7264 |
"eval_steps_per_second": 133.175,
|
| 7265 |
"step": 1032
|
| 7266 |
+
},
|
| 7267 |
+
{
|
| 7268 |
+
"epoch": 0.0871638012867841,
|
| 7269 |
+
"grad_norm": 0.051210373640060425,
|
| 7270 |
+
"learning_rate": 2.916719127169053e-05,
|
| 7271 |
+
"loss": 11.7361,
|
| 7272 |
+
"step": 1033
|
| 7273 |
+
},
|
| 7274 |
+
{
|
| 7275 |
+
"epoch": 0.08724818057166965,
|
| 7276 |
+
"grad_norm": 0.05944953113794327,
|
| 7277 |
+
"learning_rate": 2.900467979050192e-05,
|
| 7278 |
+
"loss": 11.7394,
|
| 7279 |
+
"step": 1034
|
| 7280 |
+
},
|
| 7281 |
+
{
|
| 7282 |
+
"epoch": 0.08733255985655522,
|
| 7283 |
+
"grad_norm": 0.05057380720973015,
|
| 7284 |
+
"learning_rate": 2.8842545479679793e-05,
|
| 7285 |
+
"loss": 11.7424,
|
| 7286 |
+
"step": 1035
|
| 7287 |
+
},
|
| 7288 |
+
{
|
| 7289 |
+
"epoch": 0.08741693914144077,
|
| 7290 |
+
"grad_norm": 0.04771542549133301,
|
| 7291 |
+
"learning_rate": 2.868078920058037e-05,
|
| 7292 |
+
"loss": 11.7351,
|
| 7293 |
+
"step": 1036
|
| 7294 |
+
},
|
| 7295 |
+
{
|
| 7296 |
+
"epoch": 0.08750131842632633,
|
| 7297 |
+
"grad_norm": 0.050783440470695496,
|
| 7298 |
+
"learning_rate": 2.8519411812551388e-05,
|
| 7299 |
+
"loss": 11.7382,
|
| 7300 |
+
"step": 1037
|
| 7301 |
+
},
|
| 7302 |
+
{
|
| 7303 |
+
"epoch": 0.0875856977112119,
|
| 7304 |
+
"grad_norm": 0.04163803160190582,
|
| 7305 |
+
"learning_rate": 2.8358414172927883e-05,
|
| 7306 |
+
"loss": 11.7386,
|
| 7307 |
+
"step": 1038
|
| 7308 |
+
},
|
| 7309 |
+
{
|
| 7310 |
+
"epoch": 0.08767007699609745,
|
| 7311 |
+
"grad_norm": 0.04715771973133087,
|
| 7312 |
+
"learning_rate": 2.8197797137027337e-05,
|
| 7313 |
+
"loss": 11.7355,
|
| 7314 |
+
"step": 1039
|
| 7315 |
+
},
|
| 7316 |
+
{
|
| 7317 |
+
"epoch": 0.08775445628098302,
|
| 7318 |
+
"grad_norm": 0.044405248016119,
|
| 7319 |
+
"learning_rate": 2.8037561558145155e-05,
|
| 7320 |
+
"loss": 11.7307,
|
| 7321 |
+
"step": 1040
|
| 7322 |
+
},
|
| 7323 |
+
{
|
| 7324 |
+
"epoch": 0.08783883556586858,
|
| 7325 |
+
"grad_norm": 0.03523321449756622,
|
| 7326 |
+
"learning_rate": 2.7877708287550363e-05,
|
| 7327 |
+
"loss": 11.7391,
|
| 7328 |
+
"step": 1041
|
| 7329 |
+
},
|
| 7330 |
+
{
|
| 7331 |
+
"epoch": 0.08792321485075413,
|
| 7332 |
+
"grad_norm": 0.04324277117848396,
|
| 7333 |
+
"learning_rate": 2.7718238174480848e-05,
|
| 7334 |
+
"loss": 11.7402,
|
| 7335 |
+
"step": 1042
|
| 7336 |
+
},
|
| 7337 |
+
{
|
| 7338 |
+
"epoch": 0.0880075941356397,
|
| 7339 |
+
"grad_norm": 0.047308653593063354,
|
| 7340 |
+
"learning_rate": 2.755915206613895e-05,
|
| 7341 |
+
"loss": 11.7358,
|
| 7342 |
+
"step": 1043
|
| 7343 |
+
},
|
| 7344 |
+
{
|
| 7345 |
+
"epoch": 0.08809197342052526,
|
| 7346 |
+
"grad_norm": 0.03767863288521767,
|
| 7347 |
+
"learning_rate": 2.7400450807686938e-05,
|
| 7348 |
+
"loss": 11.7307,
|
| 7349 |
+
"step": 1044
|
| 7350 |
+
},
|
| 7351 |
+
{
|
| 7352 |
+
"epoch": 0.08817635270541083,
|
| 7353 |
+
"grad_norm": 0.045568328350782394,
|
| 7354 |
+
"learning_rate": 2.724213524224246e-05,
|
| 7355 |
+
"loss": 11.7356,
|
| 7356 |
+
"step": 1045
|
| 7357 |
+
},
|
| 7358 |
+
{
|
| 7359 |
+
"epoch": 0.08826073199029638,
|
| 7360 |
+
"grad_norm": 0.06720924377441406,
|
| 7361 |
+
"learning_rate": 2.7084206210874274e-05,
|
| 7362 |
+
"loss": 11.7429,
|
| 7363 |
+
"step": 1046
|
| 7364 |
+
},
|
| 7365 |
+
{
|
| 7366 |
+
"epoch": 0.08834511127518194,
|
| 7367 |
+
"grad_norm": 0.045691728591918945,
|
| 7368 |
+
"learning_rate": 2.6926664552597535e-05,
|
| 7369 |
+
"loss": 11.7409,
|
| 7370 |
+
"step": 1047
|
| 7371 |
+
},
|
| 7372 |
+
{
|
| 7373 |
+
"epoch": 0.0884294905600675,
|
| 7374 |
+
"grad_norm": 0.04011031985282898,
|
| 7375 |
+
"learning_rate": 2.6769511104369382e-05,
|
| 7376 |
+
"loss": 11.7419,
|
| 7377 |
+
"step": 1048
|
| 7378 |
+
},
|
| 7379 |
+
{
|
| 7380 |
+
"epoch": 0.08851386984495306,
|
| 7381 |
+
"grad_norm": 0.04756263270974159,
|
| 7382 |
+
"learning_rate": 2.661274670108469e-05,
|
| 7383 |
+
"loss": 11.7421,
|
| 7384 |
+
"step": 1049
|
| 7385 |
+
},
|
| 7386 |
+
{
|
| 7387 |
+
"epoch": 0.08859824912983863,
|
| 7388 |
+
"grad_norm": 0.048668891191482544,
|
| 7389 |
+
"learning_rate": 2.6456372175571386e-05,
|
| 7390 |
+
"loss": 11.7406,
|
| 7391 |
+
"step": 1050
|
| 7392 |
+
},
|
| 7393 |
+
{
|
| 7394 |
+
"epoch": 0.08868262841472419,
|
| 7395 |
+
"grad_norm": 0.04588236287236214,
|
| 7396 |
+
"learning_rate": 2.630038835858617e-05,
|
| 7397 |
+
"loss": 11.7316,
|
| 7398 |
+
"step": 1051
|
| 7399 |
+
},
|
| 7400 |
+
{
|
| 7401 |
+
"epoch": 0.08876700769960974,
|
| 7402 |
+
"grad_norm": 0.05487845838069916,
|
| 7403 |
+
"learning_rate": 2.6144796078810063e-05,
|
| 7404 |
+
"loss": 11.7308,
|
| 7405 |
+
"step": 1052
|
| 7406 |
+
},
|
| 7407 |
+
{
|
| 7408 |
+
"epoch": 0.08885138698449531,
|
| 7409 |
+
"grad_norm": 0.04577934741973877,
|
| 7410 |
+
"learning_rate": 2.598959616284391e-05,
|
| 7411 |
+
"loss": 11.7363,
|
| 7412 |
+
"step": 1053
|
| 7413 |
+
},
|
| 7414 |
+
{
|
| 7415 |
+
"epoch": 0.08893576626938086,
|
| 7416 |
+
"grad_norm": 0.03855929523706436,
|
| 7417 |
+
"learning_rate": 2.5834789435204243e-05,
|
| 7418 |
+
"loss": 11.7328,
|
| 7419 |
+
"step": 1054
|
| 7420 |
+
},
|
| 7421 |
+
{
|
| 7422 |
+
"epoch": 0.08902014555426643,
|
| 7423 |
+
"grad_norm": 0.050304148346185684,
|
| 7424 |
+
"learning_rate": 2.5680376718318654e-05,
|
| 7425 |
+
"loss": 11.7383,
|
| 7426 |
+
"step": 1055
|
| 7427 |
+
},
|
| 7428 |
+
{
|
| 7429 |
+
"epoch": 0.08910452483915199,
|
| 7430 |
+
"grad_norm": 0.053080540150403976,
|
| 7431 |
+
"learning_rate": 2.5526358832521425e-05,
|
| 7432 |
+
"loss": 11.7442,
|
| 7433 |
+
"step": 1056
|
| 7434 |
+
},
|
| 7435 |
+
{
|
| 7436 |
+
"epoch": 0.08918890412403754,
|
| 7437 |
+
"grad_norm": 0.04133858159184456,
|
| 7438 |
+
"learning_rate": 2.5372736596049418e-05,
|
| 7439 |
+
"loss": 11.7321,
|
| 7440 |
+
"step": 1057
|
| 7441 |
+
},
|
| 7442 |
+
{
|
| 7443 |
+
"epoch": 0.08927328340892311,
|
| 7444 |
+
"grad_norm": 0.043758735060691833,
|
| 7445 |
+
"learning_rate": 2.5219510825037462e-05,
|
| 7446 |
+
"loss": 11.7319,
|
| 7447 |
+
"step": 1058
|
| 7448 |
+
},
|
| 7449 |
+
{
|
| 7450 |
+
"epoch": 0.08935766269380867,
|
| 7451 |
+
"grad_norm": 0.057619646191596985,
|
| 7452 |
+
"learning_rate": 2.5066682333514137e-05,
|
| 7453 |
+
"loss": 11.7434,
|
| 7454 |
+
"step": 1059
|
| 7455 |
+
},
|
| 7456 |
+
{
|
| 7457 |
+
"epoch": 0.08944204197869424,
|
| 7458 |
+
"grad_norm": 0.051560867577791214,
|
| 7459 |
+
"learning_rate": 2.4914251933397483e-05,
|
| 7460 |
+
"loss": 11.7345,
|
| 7461 |
+
"step": 1060
|
| 7462 |
+
},
|
| 7463 |
+
{
|
| 7464 |
+
"epoch": 0.08952642126357979,
|
| 7465 |
+
"grad_norm": 0.05797383561730385,
|
| 7466 |
+
"learning_rate": 2.4762220434490503e-05,
|
| 7467 |
+
"loss": 11.7331,
|
| 7468 |
+
"step": 1061
|
| 7469 |
+
},
|
| 7470 |
+
{
|
| 7471 |
+
"epoch": 0.08961080054846535,
|
| 7472 |
+
"grad_norm": 0.038218650966882706,
|
| 7473 |
+
"learning_rate": 2.4610588644477163e-05,
|
| 7474 |
+
"loss": 11.7396,
|
| 7475 |
+
"step": 1062
|
| 7476 |
+
},
|
| 7477 |
+
{
|
| 7478 |
+
"epoch": 0.08969517983335092,
|
| 7479 |
+
"grad_norm": 0.052235107868909836,
|
| 7480 |
+
"learning_rate": 2.4459357368917845e-05,
|
| 7481 |
+
"loss": 11.7346,
|
| 7482 |
+
"step": 1063
|
| 7483 |
+
},
|
| 7484 |
+
{
|
| 7485 |
+
"epoch": 0.08977955911823647,
|
| 7486 |
+
"grad_norm": 0.04374772310256958,
|
| 7487 |
+
"learning_rate": 2.4308527411245097e-05,
|
| 7488 |
+
"loss": 11.7321,
|
| 7489 |
+
"step": 1064
|
| 7490 |
+
},
|
| 7491 |
+
{
|
| 7492 |
+
"epoch": 0.08986393840312204,
|
| 7493 |
+
"grad_norm": 0.04937068372964859,
|
| 7494 |
+
"learning_rate": 2.415809957275956e-05,
|
| 7495 |
+
"loss": 11.7335,
|
| 7496 |
+
"step": 1065
|
| 7497 |
+
},
|
| 7498 |
+
{
|
| 7499 |
+
"epoch": 0.0899483176880076,
|
| 7500 |
+
"grad_norm": 0.058082547038793564,
|
| 7501 |
+
"learning_rate": 2.40080746526255e-05,
|
| 7502 |
+
"loss": 11.7296,
|
| 7503 |
+
"step": 1066
|
| 7504 |
+
},
|
| 7505 |
+
{
|
| 7506 |
+
"epoch": 0.09003269697289315,
|
| 7507 |
+
"grad_norm": 0.054038409143686295,
|
| 7508 |
+
"learning_rate": 2.3858453447866557e-05,
|
| 7509 |
+
"loss": 11.7343,
|
| 7510 |
+
"step": 1067
|
| 7511 |
+
},
|
| 7512 |
+
{
|
| 7513 |
+
"epoch": 0.09011707625777872,
|
| 7514 |
+
"grad_norm": 0.058869801461696625,
|
| 7515 |
+
"learning_rate": 2.3709236753361774e-05,
|
| 7516 |
+
"loss": 11.7355,
|
| 7517 |
+
"step": 1068
|
| 7518 |
+
},
|
| 7519 |
+
{
|
| 7520 |
+
"epoch": 0.09020145554266427,
|
| 7521 |
+
"grad_norm": 0.04889821633696556,
|
| 7522 |
+
"learning_rate": 2.3560425361840976e-05,
|
| 7523 |
+
"loss": 11.7325,
|
| 7524 |
+
"step": 1069
|
| 7525 |
+
},
|
| 7526 |
+
{
|
| 7527 |
+
"epoch": 0.09028583482754983,
|
| 7528 |
+
"grad_norm": 0.049528591334819794,
|
| 7529 |
+
"learning_rate": 2.341202006388096e-05,
|
| 7530 |
+
"loss": 11.7385,
|
| 7531 |
+
"step": 1070
|
| 7532 |
+
},
|
| 7533 |
+
{
|
| 7534 |
+
"epoch": 0.0903702141124354,
|
| 7535 |
+
"grad_norm": 0.048903752118349075,
|
| 7536 |
+
"learning_rate": 2.3264021647901014e-05,
|
| 7537 |
+
"loss": 11.7305,
|
| 7538 |
+
"step": 1071
|
| 7539 |
+
},
|
| 7540 |
+
{
|
| 7541 |
+
"epoch": 0.09045459339732095,
|
| 7542 |
+
"grad_norm": 0.05058593675494194,
|
| 7543 |
+
"learning_rate": 2.311643090015877e-05,
|
| 7544 |
+
"loss": 11.7379,
|
| 7545 |
+
"step": 1072
|
| 7546 |
+
},
|
| 7547 |
+
{
|
| 7548 |
+
"epoch": 0.09053897268220652,
|
| 7549 |
+
"grad_norm": 0.048489585518836975,
|
| 7550 |
+
"learning_rate": 2.2969248604746208e-05,
|
| 7551 |
+
"loss": 11.7313,
|
| 7552 |
+
"step": 1073
|
| 7553 |
+
},
|
| 7554 |
+
{
|
| 7555 |
+
"epoch": 0.09062335196709208,
|
| 7556 |
+
"grad_norm": 0.04381921514868736,
|
| 7557 |
+
"learning_rate": 2.282247554358531e-05,
|
| 7558 |
+
"loss": 11.7402,
|
| 7559 |
+
"step": 1074
|
| 7560 |
+
},
|
| 7561 |
+
{
|
| 7562 |
+
"epoch": 0.09070773125197763,
|
| 7563 |
+
"grad_norm": 0.07846806198358536,
|
| 7564 |
+
"learning_rate": 2.267611249642383e-05,
|
| 7565 |
+
"loss": 11.7376,
|
| 7566 |
+
"step": 1075
|
| 7567 |
+
},
|
| 7568 |
+
{
|
| 7569 |
+
"epoch": 0.0907921105368632,
|
| 7570 |
+
"grad_norm": 0.050075385719537735,
|
| 7571 |
+
"learning_rate": 2.2530160240831498e-05,
|
| 7572 |
+
"loss": 11.7374,
|
| 7573 |
+
"step": 1076
|
| 7574 |
+
},
|
| 7575 |
+
{
|
| 7576 |
+
"epoch": 0.09087648982174876,
|
| 7577 |
+
"grad_norm": 0.0697568878531456,
|
| 7578 |
+
"learning_rate": 2.2384619552195518e-05,
|
| 7579 |
+
"loss": 11.7338,
|
| 7580 |
+
"step": 1077
|
| 7581 |
+
},
|
| 7582 |
+
{
|
| 7583 |
+
"epoch": 0.09096086910663433,
|
| 7584 |
+
"grad_norm": 0.04641594737768173,
|
| 7585 |
+
"learning_rate": 2.2239491203716646e-05,
|
| 7586 |
+
"loss": 11.7423,
|
| 7587 |
+
"step": 1078
|
| 7588 |
+
},
|
| 7589 |
+
{
|
| 7590 |
+
"epoch": 0.09104524839151988,
|
| 7591 |
+
"grad_norm": 0.038387954235076904,
|
| 7592 |
+
"learning_rate": 2.2094775966405045e-05,
|
| 7593 |
+
"loss": 11.7405,
|
| 7594 |
+
"step": 1079
|
| 7595 |
+
},
|
| 7596 |
+
{
|
| 7597 |
+
"epoch": 0.09112962767640544,
|
| 7598 |
+
"grad_norm": 0.050520624965429306,
|
| 7599 |
+
"learning_rate": 2.1950474609076098e-05,
|
| 7600 |
+
"loss": 11.7344,
|
| 7601 |
+
"step": 1080
|
| 7602 |
+
},
|
| 7603 |
+
{
|
| 7604 |
+
"epoch": 0.091214006961291,
|
| 7605 |
+
"grad_norm": 0.03694351762533188,
|
| 7606 |
+
"learning_rate": 2.1806587898346552e-05,
|
| 7607 |
+
"loss": 11.7326,
|
| 7608 |
+
"step": 1081
|
| 7609 |
+
},
|
| 7610 |
+
{
|
| 7611 |
+
"epoch": 0.09129838624617656,
|
| 7612 |
+
"grad_norm": 0.059930723160505295,
|
| 7613 |
+
"learning_rate": 2.1663116598630205e-05,
|
| 7614 |
+
"loss": 11.7346,
|
| 7615 |
+
"step": 1082
|
| 7616 |
+
},
|
| 7617 |
+
{
|
| 7618 |
+
"epoch": 0.09138276553106213,
|
| 7619 |
+
"grad_norm": 0.033798351883888245,
|
| 7620 |
+
"learning_rate": 2.1520061472133902e-05,
|
| 7621 |
+
"loss": 11.7386,
|
| 7622 |
+
"step": 1083
|
| 7623 |
+
},
|
| 7624 |
+
{
|
| 7625 |
+
"epoch": 0.09146714481594768,
|
| 7626 |
+
"grad_norm": 0.036779649555683136,
|
| 7627 |
+
"learning_rate": 2.1377423278853627e-05,
|
| 7628 |
+
"loss": 11.738,
|
| 7629 |
+
"step": 1084
|
| 7630 |
+
},
|
| 7631 |
+
{
|
| 7632 |
+
"epoch": 0.09155152410083324,
|
| 7633 |
+
"grad_norm": 0.037904247641563416,
|
| 7634 |
+
"learning_rate": 2.1235202776570294e-05,
|
| 7635 |
+
"loss": 11.743,
|
| 7636 |
+
"step": 1085
|
| 7637 |
+
},
|
| 7638 |
+
{
|
| 7639 |
+
"epoch": 0.09163590338571881,
|
| 7640 |
+
"grad_norm": 0.028789686039090157,
|
| 7641 |
+
"learning_rate": 2.1093400720845814e-05,
|
| 7642 |
+
"loss": 11.7359,
|
| 7643 |
+
"step": 1086
|
| 7644 |
+
},
|
| 7645 |
+
{
|
| 7646 |
+
"epoch": 0.09172028267060436,
|
| 7647 |
+
"grad_norm": 0.05105876550078392,
|
| 7648 |
+
"learning_rate": 2.0952017865019036e-05,
|
| 7649 |
+
"loss": 11.7333,
|
| 7650 |
+
"step": 1087
|
| 7651 |
+
},
|
| 7652 |
+
{
|
| 7653 |
+
"epoch": 0.09180466195548993,
|
| 7654 |
+
"grad_norm": 0.051357559859752655,
|
| 7655 |
+
"learning_rate": 2.0811054960201725e-05,
|
| 7656 |
+
"loss": 11.7391,
|
| 7657 |
+
"step": 1088
|
| 7658 |
+
},
|
| 7659 |
+
{
|
| 7660 |
+
"epoch": 0.09188904124037549,
|
| 7661 |
+
"grad_norm": 0.05480816960334778,
|
| 7662 |
+
"learning_rate": 2.067051275527472e-05,
|
| 7663 |
+
"loss": 11.7396,
|
| 7664 |
+
"step": 1089
|
| 7665 |
+
},
|
| 7666 |
+
{
|
| 7667 |
+
"epoch": 0.09197342052526104,
|
| 7668 |
+
"grad_norm": 0.06252170354127884,
|
| 7669 |
+
"learning_rate": 2.053039199688378e-05,
|
| 7670 |
+
"loss": 11.7303,
|
| 7671 |
+
"step": 1090
|
| 7672 |
+
},
|
| 7673 |
+
{
|
| 7674 |
+
"epoch": 0.09205779981014661,
|
| 7675 |
+
"grad_norm": 0.04678226262331009,
|
| 7676 |
+
"learning_rate": 2.0390693429435627e-05,
|
| 7677 |
+
"loss": 11.7393,
|
| 7678 |
+
"step": 1091
|
| 7679 |
+
},
|
| 7680 |
+
{
|
| 7681 |
+
"epoch": 0.09214217909503217,
|
| 7682 |
+
"grad_norm": 0.04634862765669823,
|
| 7683 |
+
"learning_rate": 2.0251417795094165e-05,
|
| 7684 |
+
"loss": 11.7432,
|
| 7685 |
+
"step": 1092
|
| 7686 |
+
},
|
| 7687 |
+
{
|
| 7688 |
+
"epoch": 0.09222655837991774,
|
| 7689 |
+
"grad_norm": 0.05414288491010666,
|
| 7690 |
+
"learning_rate": 2.0112565833776365e-05,
|
| 7691 |
+
"loss": 11.7431,
|
| 7692 |
+
"step": 1093
|
| 7693 |
+
},
|
| 7694 |
+
{
|
| 7695 |
+
"epoch": 0.09231093766480329,
|
| 7696 |
+
"grad_norm": 0.04650377482175827,
|
| 7697 |
+
"learning_rate": 1.99741382831483e-05,
|
| 7698 |
+
"loss": 11.7344,
|
| 7699 |
+
"step": 1094
|
| 7700 |
+
},
|
| 7701 |
+
{
|
| 7702 |
+
"epoch": 0.09239531694968885,
|
| 7703 |
+
"grad_norm": 0.039519403129816055,
|
| 7704 |
+
"learning_rate": 1.983613587862153e-05,
|
| 7705 |
+
"loss": 11.7402,
|
| 7706 |
+
"step": 1095
|
| 7707 |
+
},
|
| 7708 |
+
{
|
| 7709 |
+
"epoch": 0.09247969623457442,
|
| 7710 |
+
"grad_norm": 0.054441023617982864,
|
| 7711 |
+
"learning_rate": 1.9698559353348734e-05,
|
| 7712 |
+
"loss": 11.7331,
|
| 7713 |
+
"step": 1096
|
| 7714 |
+
},
|
| 7715 |
+
{
|
| 7716 |
+
"epoch": 0.09256407551945997,
|
| 7717 |
+
"grad_norm": 0.06373140960931778,
|
| 7718 |
+
"learning_rate": 1.9561409438220247e-05,
|
| 7719 |
+
"loss": 11.7349,
|
| 7720 |
+
"step": 1097
|
| 7721 |
+
},
|
| 7722 |
+
{
|
| 7723 |
+
"epoch": 0.09264845480434554,
|
| 7724 |
+
"grad_norm": 0.04323618486523628,
|
| 7725 |
+
"learning_rate": 1.942468686185993e-05,
|
| 7726 |
+
"loss": 11.7335,
|
| 7727 |
+
"step": 1098
|
| 7728 |
+
},
|
| 7729 |
+
{
|
| 7730 |
+
"epoch": 0.0927328340892311,
|
| 7731 |
+
"grad_norm": 0.0476679727435112,
|
| 7732 |
+
"learning_rate": 1.9288392350621277e-05,
|
| 7733 |
+
"loss": 11.732,
|
| 7734 |
+
"step": 1099
|
| 7735 |
+
},
|
| 7736 |
+
{
|
| 7737 |
+
"epoch": 0.09281721337411665,
|
| 7738 |
+
"grad_norm": 0.060586925595998764,
|
| 7739 |
+
"learning_rate": 1.9152526628583778e-05,
|
| 7740 |
+
"loss": 11.7313,
|
| 7741 |
+
"step": 1100
|
| 7742 |
+
},
|
| 7743 |
+
{
|
| 7744 |
+
"epoch": 0.09290159265900222,
|
| 7745 |
+
"grad_norm": 0.07433529198169708,
|
| 7746 |
+
"learning_rate": 1.901709041754889e-05,
|
| 7747 |
+
"loss": 11.7321,
|
| 7748 |
+
"step": 1101
|
| 7749 |
+
},
|
| 7750 |
+
{
|
| 7751 |
+
"epoch": 0.09298597194388777,
|
| 7752 |
+
"grad_norm": 0.04580956697463989,
|
| 7753 |
+
"learning_rate": 1.888208443703614e-05,
|
| 7754 |
+
"loss": 11.7349,
|
| 7755 |
+
"step": 1102
|
| 7756 |
+
},
|
| 7757 |
+
{
|
| 7758 |
+
"epoch": 0.09307035122877333,
|
| 7759 |
+
"grad_norm": 0.05141078308224678,
|
| 7760 |
+
"learning_rate": 1.8747509404279594e-05,
|
| 7761 |
+
"loss": 11.7406,
|
| 7762 |
+
"step": 1103
|
| 7763 |
+
},
|
| 7764 |
+
{
|
| 7765 |
+
"epoch": 0.0931547305136589,
|
| 7766 |
+
"grad_norm": 0.04042494297027588,
|
| 7767 |
+
"learning_rate": 1.8613366034223666e-05,
|
| 7768 |
+
"loss": 11.7414,
|
| 7769 |
+
"step": 1104
|
| 7770 |
+
},
|
| 7771 |
+
{
|
| 7772 |
+
"epoch": 0.09323910979854445,
|
| 7773 |
+
"grad_norm": 0.043851088732481,
|
| 7774 |
+
"learning_rate": 1.8479655039519683e-05,
|
| 7775 |
+
"loss": 11.7351,
|
| 7776 |
+
"step": 1105
|
| 7777 |
+
},
|
| 7778 |
+
{
|
| 7779 |
+
"epoch": 0.09332348908343002,
|
| 7780 |
+
"grad_norm": 0.06234266236424446,
|
| 7781 |
+
"learning_rate": 1.8346377130521863e-05,
|
| 7782 |
+
"loss": 11.7343,
|
| 7783 |
+
"step": 1106
|
| 7784 |
+
},
|
| 7785 |
+
{
|
| 7786 |
+
"epoch": 0.09340786836831558,
|
| 7787 |
+
"grad_norm": 0.05132495239377022,
|
| 7788 |
+
"learning_rate": 1.8213533015283525e-05,
|
| 7789 |
+
"loss": 11.7299,
|
| 7790 |
+
"step": 1107
|
| 7791 |
+
},
|
| 7792 |
+
{
|
| 7793 |
+
"epoch": 0.09349224765320113,
|
| 7794 |
+
"grad_norm": 0.04665987193584442,
|
| 7795 |
+
"learning_rate": 1.8081123399553568e-05,
|
| 7796 |
+
"loss": 11.7322,
|
| 7797 |
+
"step": 1108
|
| 7798 |
+
},
|
| 7799 |
+
{
|
| 7800 |
+
"epoch": 0.0935766269380867,
|
| 7801 |
+
"grad_norm": 0.053191788494586945,
|
| 7802 |
+
"learning_rate": 1.794914898677249e-05,
|
| 7803 |
+
"loss": 11.7332,
|
| 7804 |
+
"step": 1109
|
| 7805 |
+
},
|
| 7806 |
+
{
|
| 7807 |
+
"epoch": 0.09366100622297226,
|
| 7808 |
+
"grad_norm": 0.043996527791023254,
|
| 7809 |
+
"learning_rate": 1.781761047806866e-05,
|
| 7810 |
+
"loss": 11.7411,
|
| 7811 |
+
"step": 1110
|
| 7812 |
+
},
|
| 7813 |
+
{
|
| 7814 |
+
"epoch": 0.09374538550785783,
|
| 7815 |
+
"grad_norm": 0.06540839374065399,
|
| 7816 |
+
"learning_rate": 1.76865085722548e-05,
|
| 7817 |
+
"loss": 11.7387,
|
| 7818 |
+
"step": 1111
|
| 7819 |
+
},
|
| 7820 |
+
{
|
| 7821 |
+
"epoch": 0.09382976479274338,
|
| 7822 |
+
"grad_norm": 0.0516204908490181,
|
| 7823 |
+
"learning_rate": 1.7555843965823992e-05,
|
| 7824 |
+
"loss": 11.7424,
|
| 7825 |
+
"step": 1112
|
| 7826 |
+
},
|
| 7827 |
+
{
|
| 7828 |
+
"epoch": 0.09391414407762894,
|
| 7829 |
+
"grad_norm": 0.04256313294172287,
|
| 7830 |
+
"learning_rate": 1.742561735294618e-05,
|
| 7831 |
+
"loss": 11.7374,
|
| 7832 |
+
"step": 1113
|
| 7833 |
+
},
|
| 7834 |
+
{
|
| 7835 |
+
"epoch": 0.0939985233625145,
|
| 7836 |
+
"grad_norm": 0.047496214509010315,
|
| 7837 |
+
"learning_rate": 1.7295829425464494e-05,
|
| 7838 |
+
"loss": 11.7408,
|
| 7839 |
+
"step": 1114
|
| 7840 |
+
},
|
| 7841 |
+
{
|
| 7842 |
+
"epoch": 0.09408290264740006,
|
| 7843 |
+
"grad_norm": 0.050380922853946686,
|
| 7844 |
+
"learning_rate": 1.7166480872891333e-05,
|
| 7845 |
+
"loss": 11.7318,
|
| 7846 |
+
"step": 1115
|
| 7847 |
+
},
|
| 7848 |
+
{
|
| 7849 |
+
"epoch": 0.09416728193228563,
|
| 7850 |
+
"grad_norm": 0.04991568624973297,
|
| 7851 |
+
"learning_rate": 1.7037572382405033e-05,
|
| 7852 |
+
"loss": 11.7353,
|
| 7853 |
+
"step": 1116
|
| 7854 |
+
},
|
| 7855 |
+
{
|
| 7856 |
+
"epoch": 0.09425166121717118,
|
| 7857 |
+
"grad_norm": 0.053983237594366074,
|
| 7858 |
+
"learning_rate": 1.6909104638845986e-05,
|
| 7859 |
+
"loss": 11.7369,
|
| 7860 |
+
"step": 1117
|
| 7861 |
+
},
|
| 7862 |
+
{
|
| 7863 |
+
"epoch": 0.09433604050205674,
|
| 7864 |
+
"grad_norm": 0.04841995984315872,
|
| 7865 |
+
"learning_rate": 1.6781078324712975e-05,
|
| 7866 |
+
"loss": 11.7386,
|
| 7867 |
+
"step": 1118
|
| 7868 |
+
},
|
| 7869 |
+
{
|
| 7870 |
+
"epoch": 0.09442041978694231,
|
| 7871 |
+
"grad_norm": 0.04017956554889679,
|
| 7872 |
+
"learning_rate": 1.665349412015984e-05,
|
| 7873 |
+
"loss": 11.7403,
|
| 7874 |
+
"step": 1119
|
| 7875 |
+
},
|
| 7876 |
+
{
|
| 7877 |
+
"epoch": 0.09450479907182786,
|
| 7878 |
+
"grad_norm": 0.04265524446964264,
|
| 7879 |
+
"learning_rate": 1.6526352702991478e-05,
|
| 7880 |
+
"loss": 11.7406,
|
| 7881 |
+
"step": 1120
|
| 7882 |
+
},
|
| 7883 |
+
{
|
| 7884 |
+
"epoch": 0.09458917835671343,
|
| 7885 |
+
"grad_norm": 0.04521246254444122,
|
| 7886 |
+
"learning_rate": 1.6399654748660498e-05,
|
| 7887 |
+
"loss": 11.7349,
|
| 7888 |
+
"step": 1121
|
| 7889 |
+
},
|
| 7890 |
+
{
|
| 7891 |
+
"epoch": 0.09467355764159899,
|
| 7892 |
+
"grad_norm": 0.0627727061510086,
|
| 7893 |
+
"learning_rate": 1.627340093026367e-05,
|
| 7894 |
+
"loss": 11.7341,
|
| 7895 |
+
"step": 1122
|
| 7896 |
+
},
|
| 7897 |
+
{
|
| 7898 |
+
"epoch": 0.09475793692648454,
|
| 7899 |
+
"grad_norm": 0.04507485404610634,
|
| 7900 |
+
"learning_rate": 1.614759191853803e-05,
|
| 7901 |
+
"loss": 11.7403,
|
| 7902 |
+
"step": 1123
|
| 7903 |
+
},
|
| 7904 |
+
{
|
| 7905 |
+
"epoch": 0.09484231621137011,
|
| 7906 |
+
"grad_norm": 0.05335237458348274,
|
| 7907 |
+
"learning_rate": 1.602222838185773e-05,
|
| 7908 |
+
"loss": 11.7322,
|
| 7909 |
+
"step": 1124
|
| 7910 |
+
},
|
| 7911 |
+
{
|
| 7912 |
+
"epoch": 0.09492669549625567,
|
| 7913 |
+
"grad_norm": 0.053999755531549454,
|
| 7914 |
+
"learning_rate": 1.5897310986230197e-05,
|
| 7915 |
+
"loss": 11.7342,
|
| 7916 |
+
"step": 1125
|
| 7917 |
+
},
|
| 7918 |
+
{
|
| 7919 |
+
"epoch": 0.09501107478114124,
|
| 7920 |
+
"grad_norm": 0.04442942142486572,
|
| 7921 |
+
"learning_rate": 1.5772840395292633e-05,
|
| 7922 |
+
"loss": 11.7362,
|
| 7923 |
+
"step": 1126
|
| 7924 |
+
},
|
| 7925 |
+
{
|
| 7926 |
+
"epoch": 0.09509545406602679,
|
| 7927 |
+
"grad_norm": 0.05467358976602554,
|
| 7928 |
+
"learning_rate": 1.5648817270308645e-05,
|
| 7929 |
+
"loss": 11.7332,
|
| 7930 |
+
"step": 1127
|
| 7931 |
+
},
|
| 7932 |
+
{
|
| 7933 |
+
"epoch": 0.09517983335091235,
|
| 7934 |
+
"grad_norm": 0.06570015102624893,
|
| 7935 |
+
"learning_rate": 1.5525242270164618e-05,
|
| 7936 |
+
"loss": 11.7381,
|
| 7937 |
+
"step": 1128
|
| 7938 |
+
},
|
| 7939 |
+
{
|
| 7940 |
+
"epoch": 0.09526421263579792,
|
| 7941 |
+
"grad_norm": 0.08328992873430252,
|
| 7942 |
+
"learning_rate": 1.540211605136611e-05,
|
| 7943 |
+
"loss": 11.7376,
|
| 7944 |
+
"step": 1129
|
| 7945 |
+
},
|
| 7946 |
+
{
|
| 7947 |
+
"epoch": 0.09534859192068347,
|
| 7948 |
+
"grad_norm": 0.05460549518465996,
|
| 7949 |
+
"learning_rate": 1.5279439268034633e-05,
|
| 7950 |
+
"loss": 11.7361,
|
| 7951 |
+
"step": 1130
|
| 7952 |
+
},
|
| 7953 |
+
{
|
| 7954 |
+
"epoch": 0.09543297120556904,
|
| 7955 |
+
"grad_norm": 0.03204495459794998,
|
| 7956 |
+
"learning_rate": 1.5157212571903889e-05,
|
| 7957 |
+
"loss": 11.7371,
|
| 7958 |
+
"step": 1131
|
| 7959 |
+
},
|
| 7960 |
+
{
|
| 7961 |
+
"epoch": 0.0955173504904546,
|
| 7962 |
+
"grad_norm": 0.047763891518116,
|
| 7963 |
+
"learning_rate": 1.5035436612316567e-05,
|
| 7964 |
+
"loss": 11.7334,
|
| 7965 |
+
"step": 1132
|
| 7966 |
+
},
|
| 7967 |
+
{
|
| 7968 |
+
"epoch": 0.09560172977534015,
|
| 7969 |
+
"grad_norm": 0.05876670405268669,
|
| 7970 |
+
"learning_rate": 1.4914112036220695e-05,
|
| 7971 |
+
"loss": 11.744,
|
| 7972 |
+
"step": 1133
|
| 7973 |
+
},
|
| 7974 |
+
{
|
| 7975 |
+
"epoch": 0.09568610906022572,
|
| 7976 |
+
"grad_norm": 0.053105805069208145,
|
| 7977 |
+
"learning_rate": 1.4793239488166222e-05,
|
| 7978 |
+
"loss": 11.7336,
|
| 7979 |
+
"step": 1134
|
| 7980 |
+
},
|
| 7981 |
+
{
|
| 7982 |
+
"epoch": 0.09577048834511127,
|
| 7983 |
+
"grad_norm": 0.04326646029949188,
|
| 7984 |
+
"learning_rate": 1.4672819610301802e-05,
|
| 7985 |
+
"loss": 11.7365,
|
| 7986 |
+
"step": 1135
|
| 7987 |
+
},
|
| 7988 |
+
{
|
| 7989 |
+
"epoch": 0.09585486762999683,
|
| 7990 |
+
"grad_norm": 0.04853566363453865,
|
| 7991 |
+
"learning_rate": 1.455285304237114e-05,
|
| 7992 |
+
"loss": 11.739,
|
| 7993 |
+
"step": 1136
|
| 7994 |
+
},
|
| 7995 |
+
{
|
| 7996 |
+
"epoch": 0.0959392469148824,
|
| 7997 |
+
"grad_norm": 0.03289419412612915,
|
| 7998 |
+
"learning_rate": 1.4433340421709595e-05,
|
| 7999 |
+
"loss": 11.7348,
|
| 8000 |
+
"step": 1137
|
| 8001 |
+
},
|
| 8002 |
+
{
|
| 8003 |
+
"epoch": 0.09602362619976795,
|
| 8004 |
+
"grad_norm": 0.05929587036371231,
|
| 8005 |
+
"learning_rate": 1.4314282383241096e-05,
|
| 8006 |
+
"loss": 11.7395,
|
| 8007 |
+
"step": 1138
|
| 8008 |
+
},
|
| 8009 |
+
{
|
| 8010 |
+
"epoch": 0.09610800548465352,
|
| 8011 |
+
"grad_norm": 0.052573513239622116,
|
| 8012 |
+
"learning_rate": 1.419567955947434e-05,
|
| 8013 |
+
"loss": 11.7392,
|
| 8014 |
+
"step": 1139
|
| 8015 |
+
},
|
| 8016 |
+
{
|
| 8017 |
+
"epoch": 0.09619238476953908,
|
| 8018 |
+
"grad_norm": 0.04547189921140671,
|
| 8019 |
+
"learning_rate": 1.4077532580499752e-05,
|
| 8020 |
+
"loss": 11.7403,
|
| 8021 |
+
"step": 1140
|
| 8022 |
+
},
|
| 8023 |
+
{
|
| 8024 |
+
"epoch": 0.09627676405442463,
|
| 8025 |
+
"grad_norm": 0.05477345362305641,
|
| 8026 |
+
"learning_rate": 1.3959842073986085e-05,
|
| 8027 |
+
"loss": 11.7385,
|
| 8028 |
+
"step": 1141
|
| 8029 |
+
},
|
| 8030 |
+
{
|
| 8031 |
+
"epoch": 0.0963611433393102,
|
| 8032 |
+
"grad_norm": 0.0408143512904644,
|
| 8033 |
+
"learning_rate": 1.384260866517686e-05,
|
| 8034 |
+
"loss": 11.7403,
|
| 8035 |
+
"step": 1142
|
| 8036 |
+
},
|
| 8037 |
+
{
|
| 8038 |
+
"epoch": 0.09644552262419576,
|
| 8039 |
+
"grad_norm": 0.04860100522637367,
|
| 8040 |
+
"learning_rate": 1.3725832976887409e-05,
|
| 8041 |
+
"loss": 11.7447,
|
| 8042 |
+
"step": 1143
|
| 8043 |
+
},
|
| 8044 |
+
{
|
| 8045 |
+
"epoch": 0.09652990190908133,
|
| 8046 |
+
"grad_norm": 0.045273557305336,
|
| 8047 |
+
"learning_rate": 1.360951562950128e-05,
|
| 8048 |
+
"loss": 11.7387,
|
| 8049 |
+
"step": 1144
|
| 8050 |
+
},
|
| 8051 |
+
{
|
| 8052 |
+
"epoch": 0.09661428119396688,
|
| 8053 |
+
"grad_norm": 0.03969378396868706,
|
| 8054 |
+
"learning_rate": 1.3493657240966972e-05,
|
| 8055 |
+
"loss": 11.7301,
|
| 8056 |
+
"step": 1145
|
| 8057 |
+
},
|
| 8058 |
+
{
|
| 8059 |
+
"epoch": 0.09669866047885244,
|
| 8060 |
+
"grad_norm": 0.04256885126233101,
|
| 8061 |
+
"learning_rate": 1.3378258426794888e-05,
|
| 8062 |
+
"loss": 11.738,
|
| 8063 |
+
"step": 1146
|
| 8064 |
+
},
|
| 8065 |
+
{
|
| 8066 |
+
"epoch": 0.096783039763738,
|
| 8067 |
+
"grad_norm": 0.05963636934757233,
|
| 8068 |
+
"learning_rate": 1.3263319800053697e-05,
|
| 8069 |
+
"loss": 11.737,
|
| 8070 |
+
"step": 1147
|
| 8071 |
+
},
|
| 8072 |
+
{
|
| 8073 |
+
"epoch": 0.09686741904862356,
|
| 8074 |
+
"grad_norm": 0.05352253094315529,
|
| 8075 |
+
"learning_rate": 1.3148841971367388e-05,
|
| 8076 |
+
"loss": 11.733,
|
| 8077 |
+
"step": 1148
|
| 8078 |
+
},
|
| 8079 |
+
{
|
| 8080 |
+
"epoch": 0.09695179833350913,
|
| 8081 |
+
"grad_norm": 0.03994332253932953,
|
| 8082 |
+
"learning_rate": 1.3034825548911944e-05,
|
| 8083 |
+
"loss": 11.737,
|
| 8084 |
+
"step": 1149
|
| 8085 |
+
},
|
| 8086 |
+
{
|
| 8087 |
+
"epoch": 0.09703617761839468,
|
| 8088 |
+
"grad_norm": 0.052082277834415436,
|
| 8089 |
+
"learning_rate": 1.2921271138411928e-05,
|
| 8090 |
+
"loss": 11.7396,
|
| 8091 |
+
"step": 1150
|
| 8092 |
+
},
|
| 8093 |
+
{
|
| 8094 |
+
"epoch": 0.09712055690328024,
|
| 8095 |
+
"grad_norm": 0.03420352563261986,
|
| 8096 |
+
"learning_rate": 1.2808179343137583e-05,
|
| 8097 |
+
"loss": 11.7341,
|
| 8098 |
+
"step": 1151
|
| 8099 |
+
},
|
| 8100 |
+
{
|
| 8101 |
+
"epoch": 0.09720493618816581,
|
| 8102 |
+
"grad_norm": 0.048619840294122696,
|
| 8103 |
+
"learning_rate": 1.2695550763901375e-05,
|
| 8104 |
+
"loss": 11.7382,
|
| 8105 |
+
"step": 1152
|
| 8106 |
+
},
|
| 8107 |
+
{
|
| 8108 |
+
"epoch": 0.09728931547305136,
|
| 8109 |
+
"grad_norm": 0.05240659415721893,
|
| 8110 |
+
"learning_rate": 1.258338599905482e-05,
|
| 8111 |
+
"loss": 11.7405,
|
| 8112 |
+
"step": 1153
|
| 8113 |
+
},
|
| 8114 |
+
{
|
| 8115 |
+
"epoch": 0.09737369475793693,
|
| 8116 |
+
"grad_norm": 0.03384154662489891,
|
| 8117 |
+
"learning_rate": 1.2471685644485542e-05,
|
| 8118 |
+
"loss": 11.7397,
|
| 8119 |
+
"step": 1154
|
| 8120 |
+
},
|
| 8121 |
+
{
|
| 8122 |
+
"epoch": 0.09745807404282249,
|
| 8123 |
+
"grad_norm": 0.03211098536849022,
|
| 8124 |
+
"learning_rate": 1.2360450293613757e-05,
|
| 8125 |
+
"loss": 11.7327,
|
| 8126 |
+
"step": 1155
|
| 8127 |
+
},
|
| 8128 |
+
{
|
| 8129 |
+
"epoch": 0.09754245332770804,
|
| 8130 |
+
"grad_norm": 0.06345736980438232,
|
| 8131 |
+
"learning_rate": 1.2249680537389375e-05,
|
| 8132 |
+
"loss": 11.7408,
|
| 8133 |
+
"step": 1156
|
| 8134 |
+
},
|
| 8135 |
+
{
|
| 8136 |
+
"epoch": 0.09762683261259361,
|
| 8137 |
+
"grad_norm": 0.056650202721357346,
|
| 8138 |
+
"learning_rate": 1.2139376964288852e-05,
|
| 8139 |
+
"loss": 11.7378,
|
| 8140 |
+
"step": 1157
|
| 8141 |
+
},
|
| 8142 |
+
{
|
| 8143 |
+
"epoch": 0.09771121189747917,
|
| 8144 |
+
"grad_norm": 0.04024751856923103,
|
| 8145 |
+
"learning_rate": 1.2029540160311858e-05,
|
| 8146 |
+
"loss": 11.7431,
|
| 8147 |
+
"step": 1158
|
| 8148 |
+
},
|
| 8149 |
+
{
|
| 8150 |
+
"epoch": 0.09779559118236474,
|
| 8151 |
+
"grad_norm": 0.04658965766429901,
|
| 8152 |
+
"learning_rate": 1.1920170708978374e-05,
|
| 8153 |
+
"loss": 11.7371,
|
| 8154 |
+
"step": 1159
|
| 8155 |
+
},
|
| 8156 |
+
{
|
| 8157 |
+
"epoch": 0.09787997046725029,
|
| 8158 |
+
"grad_norm": 0.04927515611052513,
|
| 8159 |
+
"learning_rate": 1.181126919132557e-05,
|
| 8160 |
+
"loss": 11.74,
|
| 8161 |
+
"step": 1160
|
| 8162 |
+
},
|
| 8163 |
+
{
|
| 8164 |
+
"epoch": 0.09796434975213585,
|
| 8165 |
+
"grad_norm": 0.03423786163330078,
|
| 8166 |
+
"learning_rate": 1.1702836185904543e-05,
|
| 8167 |
+
"loss": 11.738,
|
| 8168 |
+
"step": 1161
|
| 8169 |
+
},
|
| 8170 |
+
{
|
| 8171 |
+
"epoch": 0.09804872903702141,
|
| 8172 |
+
"grad_norm": 0.03046330064535141,
|
| 8173 |
+
"learning_rate": 1.1594872268777513e-05,
|
| 8174 |
+
"loss": 11.742,
|
| 8175 |
+
"step": 1162
|
| 8176 |
+
},
|
| 8177 |
+
{
|
| 8178 |
+
"epoch": 0.09813310832190697,
|
| 8179 |
+
"grad_norm": 0.04873904958367348,
|
| 8180 |
+
"learning_rate": 1.1487378013514483e-05,
|
| 8181 |
+
"loss": 11.7425,
|
| 8182 |
+
"step": 1163
|
| 8183 |
+
},
|
| 8184 |
+
{
|
| 8185 |
+
"epoch": 0.09821748760679254,
|
| 8186 |
+
"grad_norm": 0.040266990661621094,
|
| 8187 |
+
"learning_rate": 1.1380353991190374e-05,
|
| 8188 |
+
"loss": 11.7373,
|
| 8189 |
+
"step": 1164
|
| 8190 |
+
},
|
| 8191 |
+
{
|
| 8192 |
+
"epoch": 0.0983018668916781,
|
| 8193 |
+
"grad_norm": 0.05132141709327698,
|
| 8194 |
+
"learning_rate": 1.1273800770382026e-05,
|
| 8195 |
+
"loss": 11.7348,
|
| 8196 |
+
"step": 1165
|
| 8197 |
+
},
|
| 8198 |
+
{
|
| 8199 |
+
"epoch": 0.09838624617656365,
|
| 8200 |
+
"grad_norm": 0.05011948198080063,
|
| 8201 |
+
"learning_rate": 1.116771891716496e-05,
|
| 8202 |
+
"loss": 11.7358,
|
| 8203 |
+
"step": 1166
|
| 8204 |
+
},
|
| 8205 |
+
{
|
| 8206 |
+
"epoch": 0.09847062546144922,
|
| 8207 |
+
"grad_norm": 0.05387256667017937,
|
| 8208 |
+
"learning_rate": 1.1062108995110565e-05,
|
| 8209 |
+
"loss": 11.7396,
|
| 8210 |
+
"step": 1167
|
| 8211 |
+
},
|
| 8212 |
+
{
|
| 8213 |
+
"epoch": 0.09855500474633477,
|
| 8214 |
+
"grad_norm": 0.049708906561136246,
|
| 8215 |
+
"learning_rate": 1.0956971565283114e-05,
|
| 8216 |
+
"loss": 11.7366,
|
| 8217 |
+
"step": 1168
|
| 8218 |
+
},
|
| 8219 |
+
{
|
| 8220 |
+
"epoch": 0.09863938403122033,
|
| 8221 |
+
"grad_norm": 0.048757653683423996,
|
| 8222 |
+
"learning_rate": 1.0852307186236555e-05,
|
| 8223 |
+
"loss": 11.7302,
|
| 8224 |
+
"step": 1169
|
| 8225 |
+
},
|
| 8226 |
+
{
|
| 8227 |
+
"epoch": 0.0987237633161059,
|
| 8228 |
+
"grad_norm": 0.039620526134967804,
|
| 8229 |
+
"learning_rate": 1.0748116414011888e-05,
|
| 8230 |
+
"loss": 11.7354,
|
| 8231 |
+
"step": 1170
|
| 8232 |
+
},
|
| 8233 |
+
{
|
| 8234 |
+
"epoch": 0.09880814260099145,
|
| 8235 |
+
"grad_norm": 0.037166379392147064,
|
| 8236 |
+
"learning_rate": 1.0644399802133875e-05,
|
| 8237 |
+
"loss": 11.7381,
|
| 8238 |
+
"step": 1171
|
| 8239 |
+
},
|
| 8240 |
+
{
|
| 8241 |
+
"epoch": 0.09889252188587702,
|
| 8242 |
+
"grad_norm": 0.05658088997006416,
|
| 8243 |
+
"learning_rate": 1.0541157901608289e-05,
|
| 8244 |
+
"loss": 11.7269,
|
| 8245 |
+
"step": 1172
|
| 8246 |
+
},
|
| 8247 |
+
{
|
| 8248 |
+
"epoch": 0.09897690117076258,
|
| 8249 |
+
"grad_norm": 0.04697306454181671,
|
| 8250 |
+
"learning_rate": 1.0438391260919033e-05,
|
| 8251 |
+
"loss": 11.7282,
|
| 8252 |
+
"step": 1173
|
| 8253 |
+
},
|
| 8254 |
+
{
|
| 8255 |
+
"epoch": 0.09906128045564813,
|
| 8256 |
+
"grad_norm": 0.0378534160554409,
|
| 8257 |
+
"learning_rate": 1.0336100426025008e-05,
|
| 8258 |
+
"loss": 11.7364,
|
| 8259 |
+
"step": 1174
|
| 8260 |
+
},
|
| 8261 |
+
{
|
| 8262 |
+
"epoch": 0.0991456597405337,
|
| 8263 |
+
"grad_norm": 0.04863833636045456,
|
| 8264 |
+
"learning_rate": 1.0234285940357425e-05,
|
| 8265 |
+
"loss": 11.7378,
|
| 8266 |
+
"step": 1175
|
| 8267 |
+
},
|
| 8268 |
+
{
|
| 8269 |
+
"epoch": 0.09923003902541926,
|
| 8270 |
+
"grad_norm": 0.0404275618493557,
|
| 8271 |
+
"learning_rate": 1.013294834481686e-05,
|
| 8272 |
+
"loss": 11.7357,
|
| 8273 |
+
"step": 1176
|
| 8274 |
+
},
|
| 8275 |
+
{
|
| 8276 |
+
"epoch": 0.09931441831030483,
|
| 8277 |
+
"grad_norm": 0.03083287924528122,
|
| 8278 |
+
"learning_rate": 1.003208817777025e-05,
|
| 8279 |
+
"loss": 11.7349,
|
| 8280 |
+
"step": 1177
|
| 8281 |
+
},
|
| 8282 |
+
{
|
| 8283 |
+
"epoch": 0.09939879759519038,
|
| 8284 |
+
"grad_norm": 0.04483843222260475,
|
| 8285 |
+
"learning_rate": 9.931705975048278e-06,
|
| 8286 |
+
"loss": 11.7366,
|
| 8287 |
+
"step": 1178
|
| 8288 |
+
},
|
| 8289 |
+
{
|
| 8290 |
+
"epoch": 0.09948317688007594,
|
| 8291 |
+
"grad_norm": 0.041158974170684814,
|
| 8292 |
+
"learning_rate": 9.831802269942302e-06,
|
| 8293 |
+
"loss": 11.7334,
|
| 8294 |
+
"step": 1179
|
| 8295 |
+
},
|
| 8296 |
+
{
|
| 8297 |
+
"epoch": 0.0995675561649615,
|
| 8298 |
+
"grad_norm": 0.04370011016726494,
|
| 8299 |
+
"learning_rate": 9.732377593201592e-06,
|
| 8300 |
+
"loss": 11.7369,
|
| 8301 |
+
"step": 1180
|
| 8302 |
+
},
|
| 8303 |
+
{
|
| 8304 |
+
"epoch": 0.09965193544984706,
|
| 8305 |
+
"grad_norm": 0.04717973247170448,
|
| 8306 |
+
"learning_rate": 9.633432473030634e-06,
|
| 8307 |
+
"loss": 11.7374,
|
| 8308 |
+
"step": 1181
|
| 8309 |
+
},
|
| 8310 |
+
{
|
| 8311 |
+
"epoch": 0.09973631473473263,
|
| 8312 |
+
"grad_norm": 0.04466686770319939,
|
| 8313 |
+
"learning_rate": 9.534967435086096e-06,
|
| 8314 |
+
"loss": 11.745,
|
| 8315 |
+
"step": 1182
|
| 8316 |
+
},
|
| 8317 |
+
{
|
| 8318 |
+
"epoch": 0.09982069401961818,
|
| 8319 |
+
"grad_norm": 0.035717520862817764,
|
| 8320 |
+
"learning_rate": 9.436983002474208e-06,
|
| 8321 |
+
"loss": 11.737,
|
| 8322 |
+
"step": 1183
|
| 8323 |
+
},
|
| 8324 |
+
{
|
| 8325 |
+
"epoch": 0.09990507330450374,
|
| 8326 |
+
"grad_norm": 0.0347609668970108,
|
| 8327 |
+
"learning_rate": 9.339479695747988e-06,
|
| 8328 |
+
"loss": 11.7352,
|
| 8329 |
+
"step": 1184
|
| 8330 |
+
},
|
| 8331 |
+
{
|
| 8332 |
+
"epoch": 0.09998945258938931,
|
| 8333 |
+
"grad_norm": 0.0553274005651474,
|
| 8334 |
+
"learning_rate": 9.242458032904311e-06,
|
| 8335 |
+
"loss": 11.739,
|
| 8336 |
+
"step": 1185
|
| 8337 |
+
},
|
| 8338 |
+
{
|
| 8339 |
+
"epoch": 0.10007383187427486,
|
| 8340 |
+
"grad_norm": 0.03988463431596756,
|
| 8341 |
+
"learning_rate": 9.145918529381314e-06,
|
| 8342 |
+
"loss": 11.7381,
|
| 8343 |
+
"step": 1186
|
| 8344 |
+
},
|
| 8345 |
+
{
|
| 8346 |
+
"epoch": 0.10015821115916043,
|
| 8347 |
+
"grad_norm": 0.05592465028166771,
|
| 8348 |
+
"learning_rate": 9.049861698055694e-06,
|
| 8349 |
+
"loss": 11.7429,
|
| 8350 |
+
"step": 1187
|
| 8351 |
+
},
|
| 8352 |
+
{
|
| 8353 |
+
"epoch": 0.10024259044404599,
|
| 8354 |
+
"grad_norm": 0.0538526326417923,
|
| 8355 |
+
"learning_rate": 8.954288049239733e-06,
|
| 8356 |
+
"loss": 11.7365,
|
| 8357 |
+
"step": 1188
|
| 8358 |
+
},
|
| 8359 |
+
{
|
| 8360 |
+
"epoch": 0.10032696972893154,
|
| 8361 |
+
"grad_norm": 0.044326748698949814,
|
| 8362 |
+
"learning_rate": 8.859198090678922e-06,
|
| 8363 |
+
"loss": 11.7408,
|
| 8364 |
+
"step": 1189
|
| 8365 |
+
},
|
| 8366 |
+
{
|
| 8367 |
+
"epoch": 0.10041134901381711,
|
| 8368 |
+
"grad_norm": 0.038209687918424606,
|
| 8369 |
+
"learning_rate": 8.764592327548949e-06,
|
| 8370 |
+
"loss": 11.7339,
|
| 8371 |
+
"step": 1190
|
| 8372 |
+
},
|
| 8373 |
+
{
|
| 8374 |
+
"epoch": 0.10049572829870267,
|
| 8375 |
+
"grad_norm": 0.03617362305521965,
|
| 8376 |
+
"learning_rate": 8.67047126245325e-06,
|
| 8377 |
+
"loss": 11.7308,
|
| 8378 |
+
"step": 1191
|
| 8379 |
+
},
|
| 8380 |
+
{
|
| 8381 |
+
"epoch": 0.10058010758358824,
|
| 8382 |
+
"grad_norm": 0.05320918932557106,
|
| 8383 |
+
"learning_rate": 8.57683539542028e-06,
|
| 8384 |
+
"loss": 11.7386,
|
| 8385 |
+
"step": 1192
|
| 8386 |
+
},
|
| 8387 |
+
{
|
| 8388 |
+
"epoch": 0.10066448686847379,
|
| 8389 |
+
"grad_norm": 0.05507320538163185,
|
| 8390 |
+
"learning_rate": 8.483685223900705e-06,
|
| 8391 |
+
"loss": 11.7356,
|
| 8392 |
+
"step": 1193
|
| 8393 |
+
},
|
| 8394 |
+
{
|
| 8395 |
+
"epoch": 0.10074886615335935,
|
| 8396 |
+
"grad_norm": 0.03567732498049736,
|
| 8397 |
+
"learning_rate": 8.391021242764963e-06,
|
| 8398 |
+
"loss": 11.7399,
|
| 8399 |
+
"step": 1194
|
| 8400 |
+
},
|
| 8401 |
+
{
|
| 8402 |
+
"epoch": 0.10083324543824491,
|
| 8403 |
+
"grad_norm": 0.0483282096683979,
|
| 8404 |
+
"learning_rate": 8.298843944300583e-06,
|
| 8405 |
+
"loss": 11.7409,
|
| 8406 |
+
"step": 1195
|
| 8407 |
+
},
|
| 8408 |
+
{
|
| 8409 |
+
"epoch": 0.10091762472313047,
|
| 8410 |
+
"grad_norm": 0.05166078358888626,
|
| 8411 |
+
"learning_rate": 8.207153818209446e-06,
|
| 8412 |
+
"loss": 11.7345,
|
| 8413 |
+
"step": 1196
|
| 8414 |
+
},
|
| 8415 |
+
{
|
| 8416 |
+
"epoch": 0.10100200400801604,
|
| 8417 |
+
"grad_norm": 0.04812915250658989,
|
| 8418 |
+
"learning_rate": 8.115951351605378e-06,
|
| 8419 |
+
"loss": 11.7316,
|
| 8420 |
+
"step": 1197
|
| 8421 |
+
},
|
| 8422 |
+
{
|
| 8423 |
+
"epoch": 0.1010863832929016,
|
| 8424 |
+
"grad_norm": 0.05382709950208664,
|
| 8425 |
+
"learning_rate": 8.025237029011368e-06,
|
| 8426 |
+
"loss": 11.7298,
|
| 8427 |
+
"step": 1198
|
| 8428 |
+
},
|
| 8429 |
+
{
|
| 8430 |
+
"epoch": 0.10117076257778715,
|
| 8431 |
+
"grad_norm": 0.04726829007267952,
|
| 8432 |
+
"learning_rate": 7.935011332357112e-06,
|
| 8433 |
+
"loss": 11.739,
|
| 8434 |
+
"step": 1199
|
| 8435 |
+
},
|
| 8436 |
+
{
|
| 8437 |
+
"epoch": 0.10125514186267272,
|
| 8438 |
+
"grad_norm": 0.04048336297273636,
|
| 8439 |
+
"learning_rate": 7.845274740976528e-06,
|
| 8440 |
+
"loss": 11.7374,
|
| 8441 |
+
"step": 1200
|
| 8442 |
+
},
|
| 8443 |
+
{
|
| 8444 |
+
"epoch": 0.10133952114755827,
|
| 8445 |
+
"grad_norm": 0.06581093370914459,
|
| 8446 |
+
"learning_rate": 7.756027731604942e-06,
|
| 8447 |
+
"loss": 11.7342,
|
| 8448 |
+
"step": 1201
|
| 8449 |
+
},
|
| 8450 |
+
{
|
| 8451 |
+
"epoch": 0.10142390043244383,
|
| 8452 |
+
"grad_norm": 0.05036183446645737,
|
| 8453 |
+
"learning_rate": 7.667270778376834e-06,
|
| 8454 |
+
"loss": 11.7373,
|
| 8455 |
+
"step": 1202
|
| 8456 |
+
},
|
| 8457 |
+
{
|
| 8458 |
+
"epoch": 0.1015082797173294,
|
| 8459 |
+
"grad_norm": 0.04229575768113136,
|
| 8460 |
+
"learning_rate": 7.5790043528232045e-06,
|
| 8461 |
+
"loss": 11.7406,
|
| 8462 |
+
"step": 1203
|
| 8463 |
+
},
|
| 8464 |
+
{
|
| 8465 |
+
"epoch": 0.10159265900221495,
|
| 8466 |
+
"grad_norm": 0.0635019838809967,
|
| 8467 |
+
"learning_rate": 7.491228923868998e-06,
|
| 8468 |
+
"loss": 11.7338,
|
| 8469 |
+
"step": 1204
|
| 8470 |
+
},
|
| 8471 |
+
{
|
| 8472 |
+
"epoch": 0.10167703828710052,
|
| 8473 |
+
"grad_norm": 0.0493345782160759,
|
| 8474 |
+
"learning_rate": 7.4039449578307994e-06,
|
| 8475 |
+
"loss": 11.728,
|
| 8476 |
+
"step": 1205
|
| 8477 |
+
},
|
| 8478 |
+
{
|
| 8479 |
+
"epoch": 0.10176141757198608,
|
| 8480 |
+
"grad_norm": 0.04717019572854042,
|
| 8481 |
+
"learning_rate": 7.317152918414117e-06,
|
| 8482 |
+
"loss": 11.7432,
|
| 8483 |
+
"step": 1206
|
| 8484 |
+
},
|
| 8485 |
+
{
|
| 8486 |
+
"epoch": 0.10184579685687163,
|
| 8487 |
+
"grad_norm": 0.05204099789261818,
|
| 8488 |
+
"learning_rate": 7.230853266711124e-06,
|
| 8489 |
+
"loss": 11.7355,
|
| 8490 |
+
"step": 1207
|
| 8491 |
+
},
|
| 8492 |
+
{
|
| 8493 |
+
"epoch": 0.1019301761417572,
|
| 8494 |
+
"grad_norm": 0.038699548691511154,
|
| 8495 |
+
"learning_rate": 7.145046461198146e-06,
|
| 8496 |
+
"loss": 11.7382,
|
| 8497 |
+
"step": 1208
|
| 8498 |
+
},
|
| 8499 |
+
{
|
| 8500 |
+
"epoch": 0.10201455542664276,
|
| 8501 |
+
"grad_norm": 0.04183535650372505,
|
| 8502 |
+
"learning_rate": 7.059732957733145e-06,
|
| 8503 |
+
"loss": 11.7325,
|
| 8504 |
+
"step": 1209
|
| 8505 |
+
},
|
| 8506 |
+
{
|
| 8507 |
+
"epoch": 0.10209893471152832,
|
| 8508 |
+
"grad_norm": 0.046674247831106186,
|
| 8509 |
+
"learning_rate": 6.97491320955338e-06,
|
| 8510 |
+
"loss": 11.7354,
|
| 8511 |
+
"step": 1210
|
| 8512 |
+
},
|
| 8513 |
+
{
|
| 8514 |
+
"epoch": 0.10218331399641388,
|
| 8515 |
+
"grad_norm": 0.04590325802564621,
|
| 8516 |
+
"learning_rate": 6.890587667273063e-06,
|
| 8517 |
+
"loss": 11.7363,
|
| 8518 |
+
"step": 1211
|
| 8519 |
+
},
|
| 8520 |
+
{
|
| 8521 |
+
"epoch": 0.10226769328129943,
|
| 8522 |
+
"grad_norm": 0.0393228605389595,
|
| 8523 |
+
"learning_rate": 6.806756778880752e-06,
|
| 8524 |
+
"loss": 11.7346,
|
| 8525 |
+
"step": 1212
|
| 8526 |
+
},
|
| 8527 |
+
{
|
| 8528 |
+
"epoch": 0.102352072566185,
|
| 8529 |
+
"grad_norm": 0.05303521826863289,
|
| 8530 |
+
"learning_rate": 6.7234209897371835e-06,
|
| 8531 |
+
"loss": 11.7317,
|
| 8532 |
+
"step": 1213
|
| 8533 |
+
},
|
| 8534 |
+
{
|
| 8535 |
+
"epoch": 0.10243645185107056,
|
| 8536 |
+
"grad_norm": 0.05112370848655701,
|
| 8537 |
+
"learning_rate": 6.6405807425728105e-06,
|
| 8538 |
+
"loss": 11.7312,
|
| 8539 |
+
"step": 1214
|
| 8540 |
+
},
|
| 8541 |
+
{
|
| 8542 |
+
"epoch": 0.10252083113595613,
|
| 8543 |
+
"grad_norm": 0.05076449736952782,
|
| 8544 |
+
"learning_rate": 6.558236477485458e-06,
|
| 8545 |
+
"loss": 11.7398,
|
| 8546 |
+
"step": 1215
|
| 8547 |
+
},
|
| 8548 |
+
{
|
| 8549 |
+
"epoch": 0.10260521042084168,
|
| 8550 |
+
"grad_norm": 0.04005897417664528,
|
| 8551 |
+
"learning_rate": 6.476388631938024e-06,
|
| 8552 |
+
"loss": 11.7392,
|
| 8553 |
+
"step": 1216
|
| 8554 |
+
},
|
| 8555 |
+
{
|
| 8556 |
+
"epoch": 0.10268958970572724,
|
| 8557 |
+
"grad_norm": 0.041421402245759964,
|
| 8558 |
+
"learning_rate": 6.395037640756074e-06,
|
| 8559 |
+
"loss": 11.7274,
|
| 8560 |
+
"step": 1217
|
| 8561 |
+
},
|
| 8562 |
+
{
|
| 8563 |
+
"epoch": 0.10277396899061281,
|
| 8564 |
+
"grad_norm": 0.04897172749042511,
|
| 8565 |
+
"learning_rate": 6.314183936125584e-06,
|
| 8566 |
+
"loss": 11.7319,
|
| 8567 |
+
"step": 1218
|
| 8568 |
+
},
|
| 8569 |
+
{
|
| 8570 |
+
"epoch": 0.10285834827549836,
|
| 8571 |
+
"grad_norm": 0.06160645931959152,
|
| 8572 |
+
"learning_rate": 6.233827947590709e-06,
|
| 8573 |
+
"loss": 11.7425,
|
| 8574 |
+
"step": 1219
|
| 8575 |
+
},
|
| 8576 |
+
{
|
| 8577 |
+
"epoch": 0.10294272756038393,
|
| 8578 |
+
"grad_norm": 0.052552200853824615,
|
| 8579 |
+
"learning_rate": 6.153970102051354e-06,
|
| 8580 |
+
"loss": 11.7322,
|
| 8581 |
+
"step": 1220
|
| 8582 |
+
},
|
| 8583 |
+
{
|
| 8584 |
+
"epoch": 0.10302710684526949,
|
| 8585 |
+
"grad_norm": 0.04836604744195938,
|
| 8586 |
+
"learning_rate": 6.074610823761029e-06,
|
| 8587 |
+
"loss": 11.7397,
|
| 8588 |
+
"step": 1221
|
| 8589 |
+
},
|
| 8590 |
+
{
|
| 8591 |
+
"epoch": 0.10311148613015504,
|
| 8592 |
+
"grad_norm": 0.034704506397247314,
|
| 8593 |
+
"learning_rate": 5.9957505343245176e-06,
|
| 8594 |
+
"loss": 11.7391,
|
| 8595 |
+
"step": 1222
|
| 8596 |
+
},
|
| 8597 |
+
{
|
| 8598 |
+
"epoch": 0.10319586541504061,
|
| 8599 |
+
"grad_norm": 0.047300923615694046,
|
| 8600 |
+
"learning_rate": 5.917389652695693e-06,
|
| 8601 |
+
"loss": 11.7378,
|
| 8602 |
+
"step": 1223
|
| 8603 |
+
},
|
| 8604 |
+
{
|
| 8605 |
+
"epoch": 0.10328024469992617,
|
| 8606 |
+
"grad_norm": 0.04731455072760582,
|
| 8607 |
+
"learning_rate": 5.839528595175314e-06,
|
| 8608 |
+
"loss": 11.744,
|
| 8609 |
+
"step": 1224
|
| 8610 |
+
},
|
| 8611 |
+
{
|
| 8612 |
+
"epoch": 0.10336462398481173,
|
| 8613 |
+
"grad_norm": 0.05412326753139496,
|
| 8614 |
+
"learning_rate": 5.7621677754086775e-06,
|
| 8615 |
+
"loss": 11.7367,
|
| 8616 |
+
"step": 1225
|
| 8617 |
+
},
|
| 8618 |
+
{
|
| 8619 |
+
"epoch": 0.10344900326969729,
|
| 8620 |
+
"grad_norm": 0.03682037815451622,
|
| 8621 |
+
"learning_rate": 5.685307604383561e-06,
|
| 8622 |
+
"loss": 11.7389,
|
| 8623 |
+
"step": 1226
|
| 8624 |
+
},
|
| 8625 |
+
{
|
| 8626 |
+
"epoch": 0.10353338255458285,
|
| 8627 |
+
"grad_norm": 0.054178204387426376,
|
| 8628 |
+
"learning_rate": 5.608948490428023e-06,
|
| 8629 |
+
"loss": 11.7367,
|
| 8630 |
+
"step": 1227
|
| 8631 |
+
},
|
| 8632 |
+
{
|
| 8633 |
+
"epoch": 0.10361776183946841,
|
| 8634 |
+
"grad_norm": 0.050201497972011566,
|
| 8635 |
+
"learning_rate": 5.533090839208133e-06,
|
| 8636 |
+
"loss": 11.7403,
|
| 8637 |
+
"step": 1228
|
| 8638 |
+
},
|
| 8639 |
+
{
|
| 8640 |
+
"epoch": 0.10370214112435397,
|
| 8641 |
+
"grad_norm": 0.0559217743575573,
|
| 8642 |
+
"learning_rate": 5.45773505372591e-06,
|
| 8643 |
+
"loss": 11.7386,
|
| 8644 |
+
"step": 1229
|
| 8645 |
+
},
|
| 8646 |
+
{
|
| 8647 |
+
"epoch": 0.10378652040923954,
|
| 8648 |
+
"grad_norm": 0.05557003617286682,
|
| 8649 |
+
"learning_rate": 5.38288153431723e-06,
|
| 8650 |
+
"loss": 11.74,
|
| 8651 |
+
"step": 1230
|
| 8652 |
+
},
|
| 8653 |
+
{
|
| 8654 |
+
"epoch": 0.1038708996941251,
|
| 8655 |
+
"grad_norm": 0.03945248946547508,
|
| 8656 |
+
"learning_rate": 5.308530678649504e-06,
|
| 8657 |
+
"loss": 11.739,
|
| 8658 |
+
"step": 1231
|
| 8659 |
+
},
|
| 8660 |
+
{
|
| 8661 |
+
"epoch": 0.10395527897901065,
|
| 8662 |
+
"grad_norm": 0.04193837195634842,
|
| 8663 |
+
"learning_rate": 5.2346828817197655e-06,
|
| 8664 |
+
"loss": 11.7392,
|
| 8665 |
+
"step": 1232
|
| 8666 |
+
},
|
| 8667 |
+
{
|
| 8668 |
+
"epoch": 0.10403965826389622,
|
| 8669 |
+
"grad_norm": 0.04187193512916565,
|
| 8670 |
+
"learning_rate": 5.161338535852467e-06,
|
| 8671 |
+
"loss": 11.7353,
|
| 8672 |
+
"step": 1233
|
| 8673 |
+
},
|
| 8674 |
+
{
|
| 8675 |
+
"epoch": 0.10412403754878177,
|
| 8676 |
+
"grad_norm": 0.045496273785829544,
|
| 8677 |
+
"learning_rate": 5.08849803069743e-06,
|
| 8678 |
+
"loss": 11.7391,
|
| 8679 |
+
"step": 1234
|
| 8680 |
+
},
|
| 8681 |
+
{
|
| 8682 |
+
"epoch": 0.10420841683366733,
|
| 8683 |
+
"grad_norm": 0.043216168880462646,
|
| 8684 |
+
"learning_rate": 5.016161753227799e-06,
|
| 8685 |
+
"loss": 11.7402,
|
| 8686 |
+
"step": 1235
|
| 8687 |
+
},
|
| 8688 |
+
{
|
| 8689 |
+
"epoch": 0.1042927961185529,
|
| 8690 |
+
"grad_norm": 0.08071540296077728,
|
| 8691 |
+
"learning_rate": 4.944330087737881e-06,
|
| 8692 |
+
"loss": 11.742,
|
| 8693 |
+
"step": 1236
|
| 8694 |
+
},
|
| 8695 |
+
{
|
| 8696 |
+
"epoch": 0.10437717540343845,
|
| 8697 |
+
"grad_norm": 0.03432680293917656,
|
| 8698 |
+
"learning_rate": 4.8730034158412155e-06,
|
| 8699 |
+
"loss": 11.7371,
|
| 8700 |
+
"step": 1237
|
| 8701 |
+
},
|
| 8702 |
+
{
|
| 8703 |
+
"epoch": 0.10446155468832402,
|
| 8704 |
+
"grad_norm": 0.054856572300195694,
|
| 8705 |
+
"learning_rate": 4.802182116468557e-06,
|
| 8706 |
+
"loss": 11.7376,
|
| 8707 |
+
"step": 1238
|
| 8708 |
+
},
|
| 8709 |
+
{
|
| 8710 |
+
"epoch": 0.10454593397320958,
|
| 8711 |
+
"grad_norm": 0.05000922828912735,
|
| 8712 |
+
"learning_rate": 4.731866565865717e-06,
|
| 8713 |
+
"loss": 11.7358,
|
| 8714 |
+
"step": 1239
|
| 8715 |
+
},
|
| 8716 |
+
{
|
| 8717 |
+
"epoch": 0.10463031325809513,
|
| 8718 |
+
"grad_norm": 0.046056319028139114,
|
| 8719 |
+
"learning_rate": 4.662057137591735e-06,
|
| 8720 |
+
"loss": 11.7372,
|
| 8721 |
+
"step": 1240
|
| 8722 |
+
},
|
| 8723 |
+
{
|
| 8724 |
+
"epoch": 0.1047146925429807,
|
| 8725 |
+
"grad_norm": 0.048261385411024094,
|
| 8726 |
+
"learning_rate": 4.592754202516803e-06,
|
| 8727 |
+
"loss": 11.7406,
|
| 8728 |
+
"step": 1241
|
| 8729 |
+
},
|
| 8730 |
+
{
|
| 8731 |
+
"epoch": 0.10479907182786626,
|
| 8732 |
+
"grad_norm": 0.027361886575818062,
|
| 8733 |
+
"learning_rate": 4.523958128820283e-06,
|
| 8734 |
+
"loss": 11.732,
|
| 8735 |
+
"step": 1242
|
| 8736 |
+
},
|
| 8737 |
+
{
|
| 8738 |
+
"epoch": 0.10488345111275182,
|
| 8739 |
+
"grad_norm": 0.04859941080212593,
|
| 8740 |
+
"learning_rate": 4.45566928198885e-06,
|
| 8741 |
+
"loss": 11.7345,
|
| 8742 |
+
"step": 1243
|
| 8743 |
+
},
|
| 8744 |
+
{
|
| 8745 |
+
"epoch": 0.10496783039763738,
|
| 8746 |
+
"grad_norm": 0.04576332867145538,
|
| 8747 |
+
"learning_rate": 4.3878880248143905e-06,
|
| 8748 |
+
"loss": 11.7374,
|
| 8749 |
+
"step": 1244
|
| 8750 |
+
},
|
| 8751 |
+
{
|
| 8752 |
+
"epoch": 0.10505220968252293,
|
| 8753 |
+
"grad_norm": 0.040460649877786636,
|
| 8754 |
+
"learning_rate": 4.320614717392213e-06,
|
| 8755 |
+
"loss": 11.742,
|
| 8756 |
+
"step": 1245
|
| 8757 |
+
},
|
| 8758 |
+
{
|
| 8759 |
+
"epoch": 0.1051365889674085,
|
| 8760 |
+
"grad_norm": 0.03785835951566696,
|
| 8761 |
+
"learning_rate": 4.25384971711913e-06,
|
| 8762 |
+
"loss": 11.7336,
|
| 8763 |
+
"step": 1246
|
| 8764 |
+
},
|
| 8765 |
+
{
|
| 8766 |
+
"epoch": 0.10522096825229406,
|
| 8767 |
+
"grad_norm": 0.07303830236196518,
|
| 8768 |
+
"learning_rate": 4.1875933786914345e-06,
|
| 8769 |
+
"loss": 11.741,
|
| 8770 |
+
"step": 1247
|
| 8771 |
+
},
|
| 8772 |
+
{
|
| 8773 |
+
"epoch": 0.10530534753717963,
|
| 8774 |
+
"grad_norm": 0.03281836956739426,
|
| 8775 |
+
"learning_rate": 4.12184605410314e-06,
|
| 8776 |
+
"loss": 11.7397,
|
| 8777 |
+
"step": 1248
|
| 8778 |
+
},
|
| 8779 |
+
{
|
| 8780 |
+
"epoch": 0.10538972682206518,
|
| 8781 |
+
"grad_norm": 0.03648677468299866,
|
| 8782 |
+
"learning_rate": 4.056608092644077e-06,
|
| 8783 |
+
"loss": 11.7403,
|
| 8784 |
+
"step": 1249
|
| 8785 |
+
},
|
| 8786 |
+
{
|
| 8787 |
+
"epoch": 0.10547410610695074,
|
| 8788 |
+
"grad_norm": 0.03989635780453682,
|
| 8789 |
+
"learning_rate": 3.991879840897994e-06,
|
| 8790 |
+
"loss": 11.7391,
|
| 8791 |
+
"step": 1250
|
| 8792 |
+
},
|
| 8793 |
+
{
|
| 8794 |
+
"epoch": 0.10555848539183631,
|
| 8795 |
+
"grad_norm": 0.03752193599939346,
|
| 8796 |
+
"learning_rate": 3.927661642740821e-06,
|
| 8797 |
+
"loss": 11.7387,
|
| 8798 |
+
"step": 1251
|
| 8799 |
+
},
|
| 8800 |
+
{
|
| 8801 |
+
"epoch": 0.10564286467672186,
|
| 8802 |
+
"grad_norm": 0.06209597364068031,
|
| 8803 |
+
"learning_rate": 3.863953839338685e-06,
|
| 8804 |
+
"loss": 11.7418,
|
| 8805 |
+
"step": 1252
|
| 8806 |
+
},
|
| 8807 |
+
{
|
| 8808 |
+
"epoch": 0.10572724396160743,
|
| 8809 |
+
"grad_norm": 0.056839268654584885,
|
| 8810 |
+
"learning_rate": 3.8007567691462185e-06,
|
| 8811 |
+
"loss": 11.7308,
|
| 8812 |
+
"step": 1253
|
| 8813 |
+
},
|
| 8814 |
+
{
|
| 8815 |
+
"epoch": 0.10581162324649299,
|
| 8816 |
+
"grad_norm": 0.048007652163505554,
|
| 8817 |
+
"learning_rate": 3.738070767904778e-06,
|
| 8818 |
+
"loss": 11.7374,
|
| 8819 |
+
"step": 1254
|
| 8820 |
+
},
|
| 8821 |
+
{
|
| 8822 |
+
"epoch": 0.10589600253137854,
|
| 8823 |
+
"grad_norm": 0.03774699568748474,
|
| 8824 |
+
"learning_rate": 3.675896168640536e-06,
|
| 8825 |
+
"loss": 11.738,
|
| 8826 |
+
"step": 1255
|
| 8827 |
+
},
|
| 8828 |
+
{
|
| 8829 |
+
"epoch": 0.10598038181626411,
|
| 8830 |
+
"grad_norm": 0.02863268554210663,
|
| 8831 |
+
"learning_rate": 3.6142333016628283e-06,
|
| 8832 |
+
"loss": 11.7302,
|
| 8833 |
+
"step": 1256
|
| 8834 |
+
},
|
| 8835 |
+
{
|
| 8836 |
+
"epoch": 0.10606476110114967,
|
| 8837 |
+
"grad_norm": 0.03822990879416466,
|
| 8838 |
+
"learning_rate": 3.5530824945623542e-06,
|
| 8839 |
+
"loss": 11.7308,
|
| 8840 |
+
"step": 1257
|
| 8841 |
+
},
|
| 8842 |
+
{
|
| 8843 |
+
"epoch": 0.10614914038603523,
|
| 8844 |
+
"grad_norm": 0.04675639420747757,
|
| 8845 |
+
"learning_rate": 3.492444072209411e-06,
|
| 8846 |
+
"loss": 11.7328,
|
| 8847 |
+
"step": 1258
|
| 8848 |
+
},
|
| 8849 |
+
{
|
| 8850 |
+
"epoch": 0.10623351967092079,
|
| 8851 |
+
"grad_norm": 0.030831044539809227,
|
| 8852 |
+
"learning_rate": 3.43231835675224e-06,
|
| 8853 |
+
"loss": 11.7346,
|
| 8854 |
+
"step": 1259
|
| 8855 |
+
},
|
| 8856 |
+
{
|
| 8857 |
+
"epoch": 0.10631789895580634,
|
| 8858 |
+
"grad_norm": 0.05338575690984726,
|
| 8859 |
+
"learning_rate": 3.3727056676152503e-06,
|
| 8860 |
+
"loss": 11.7364,
|
| 8861 |
+
"step": 1260
|
| 8862 |
+
},
|
| 8863 |
+
{
|
| 8864 |
+
"epoch": 0.10640227824069191,
|
| 8865 |
+
"grad_norm": 0.04862627759575844,
|
| 8866 |
+
"learning_rate": 3.313606321497309e-06,
|
| 8867 |
+
"loss": 11.7405,
|
| 8868 |
+
"step": 1261
|
| 8869 |
+
},
|
| 8870 |
+
{
|
| 8871 |
+
"epoch": 0.10648665752557747,
|
| 8872 |
+
"grad_norm": 0.036035746335983276,
|
| 8873 |
+
"learning_rate": 3.255020632370176e-06,
|
| 8874 |
+
"loss": 11.7424,
|
| 8875 |
+
"step": 1262
|
| 8876 |
+
},
|
| 8877 |
+
{
|
| 8878 |
+
"epoch": 0.10657103681046304,
|
| 8879 |
+
"grad_norm": 0.05689811706542969,
|
| 8880 |
+
"learning_rate": 3.1969489114766715e-06,
|
| 8881 |
+
"loss": 11.7322,
|
| 8882 |
+
"step": 1263
|
| 8883 |
+
},
|
| 8884 |
+
{
|
| 8885 |
+
"epoch": 0.10665541609534859,
|
| 8886 |
+
"grad_norm": 0.05613115802407265,
|
| 8887 |
+
"learning_rate": 3.1393914673291337e-06,
|
| 8888 |
+
"loss": 11.7383,
|
| 8889 |
+
"step": 1264
|
| 8890 |
+
},
|
| 8891 |
+
{
|
| 8892 |
+
"epoch": 0.10673979538023415,
|
| 8893 |
+
"grad_norm": 0.04067595675587654,
|
| 8894 |
+
"learning_rate": 3.082348605707752e-06,
|
| 8895 |
+
"loss": 11.7393,
|
| 8896 |
+
"step": 1265
|
| 8897 |
+
},
|
| 8898 |
+
{
|
| 8899 |
+
"epoch": 0.10682417466511972,
|
| 8900 |
+
"grad_norm": 0.05096471682190895,
|
| 8901 |
+
"learning_rate": 3.0258206296589485e-06,
|
| 8902 |
+
"loss": 11.7451,
|
| 8903 |
+
"step": 1266
|
| 8904 |
+
},
|
| 8905 |
+
{
|
| 8906 |
+
"epoch": 0.10690855395000527,
|
| 8907 |
+
"grad_norm": 0.05544022098183632,
|
| 8908 |
+
"learning_rate": 2.969807839493732e-06,
|
| 8909 |
+
"loss": 11.7435,
|
| 8910 |
+
"step": 1267
|
| 8911 |
+
},
|
| 8912 |
+
{
|
| 8913 |
+
"epoch": 0.10699293323489083,
|
| 8914 |
+
"grad_norm": 0.05149846151471138,
|
| 8915 |
+
"learning_rate": 2.9143105327861574e-06,
|
| 8916 |
+
"loss": 11.7374,
|
| 8917 |
+
"step": 1268
|
| 8918 |
+
},
|
| 8919 |
+
{
|
| 8920 |
+
"epoch": 0.1070773125197764,
|
| 8921 |
+
"grad_norm": 0.04375915229320526,
|
| 8922 |
+
"learning_rate": 2.8593290043717027e-06,
|
| 8923 |
+
"loss": 11.7425,
|
| 8924 |
+
"step": 1269
|
| 8925 |
+
},
|
| 8926 |
+
{
|
| 8927 |
+
"epoch": 0.10716169180466195,
|
| 8928 |
+
"grad_norm": 0.05639617145061493,
|
| 8929 |
+
"learning_rate": 2.8048635463457484e-06,
|
| 8930 |
+
"loss": 11.7412,
|
| 8931 |
+
"step": 1270
|
| 8932 |
+
},
|
| 8933 |
+
{
|
| 8934 |
+
"epoch": 0.10724607108954752,
|
| 8935 |
+
"grad_norm": 0.044010311365127563,
|
| 8936 |
+
"learning_rate": 2.7509144480619254e-06,
|
| 8937 |
+
"loss": 11.7319,
|
| 8938 |
+
"step": 1271
|
| 8939 |
+
},
|
| 8940 |
+
{
|
| 8941 |
+
"epoch": 0.10733045037443308,
|
| 8942 |
+
"grad_norm": 0.057349592447280884,
|
| 8943 |
+
"learning_rate": 2.6974819961307128e-06,
|
| 8944 |
+
"loss": 11.7398,
|
| 8945 |
+
"step": 1272
|
| 8946 |
+
},
|
| 8947 |
+
{
|
| 8948 |
+
"epoch": 0.10741482965931863,
|
| 8949 |
+
"grad_norm": 0.05181263014674187,
|
| 8950 |
+
"learning_rate": 2.644566474417831e-06,
|
| 8951 |
+
"loss": 11.7455,
|
| 8952 |
+
"step": 1273
|
| 8953 |
+
},
|
| 8954 |
+
{
|
| 8955 |
+
"epoch": 0.1074992089442042,
|
| 8956 |
+
"grad_norm": 0.05245477333664894,
|
| 8957 |
+
"learning_rate": 2.592168164042741e-06,
|
| 8958 |
+
"loss": 11.7344,
|
| 8959 |
+
"step": 1274
|
| 8960 |
+
},
|
| 8961 |
+
{
|
| 8962 |
+
"epoch": 0.10758358822908975,
|
| 8963 |
+
"grad_norm": 0.04753034934401512,
|
| 8964 |
+
"learning_rate": 2.540287343377179e-06,
|
| 8965 |
+
"loss": 11.7392,
|
| 8966 |
+
"step": 1275
|
| 8967 |
+
},
|
| 8968 |
+
{
|
| 8969 |
+
"epoch": 0.10766796751397532,
|
| 8970 |
+
"grad_norm": 0.04903138801455498,
|
| 8971 |
+
"learning_rate": 2.488924288043648e-06,
|
| 8972 |
+
"loss": 11.7393,
|
| 8973 |
+
"step": 1276
|
| 8974 |
+
},
|
| 8975 |
+
{
|
| 8976 |
+
"epoch": 0.10775234679886088,
|
| 8977 |
+
"grad_norm": 0.03462986648082733,
|
| 8978 |
+
"learning_rate": 2.438079270913951e-06,
|
| 8979 |
+
"loss": 11.7373,
|
| 8980 |
+
"step": 1277
|
| 8981 |
+
},
|
| 8982 |
+
{
|
| 8983 |
+
"epoch": 0.10783672608374643,
|
| 8984 |
+
"grad_norm": 0.03600209951400757,
|
| 8985 |
+
"learning_rate": 2.3877525621078255e-06,
|
| 8986 |
+
"loss": 11.735,
|
| 8987 |
+
"step": 1278
|
| 8988 |
+
},
|
| 8989 |
+
{
|
| 8990 |
+
"epoch": 0.107921105368632,
|
| 8991 |
+
"grad_norm": 0.05916762724518776,
|
| 8992 |
+
"learning_rate": 2.3379444289913342e-06,
|
| 8993 |
+
"loss": 11.7404,
|
| 8994 |
+
"step": 1279
|
| 8995 |
+
},
|
| 8996 |
+
{
|
| 8997 |
+
"epoch": 0.10800548465351756,
|
| 8998 |
+
"grad_norm": 0.04649950563907623,
|
| 8999 |
+
"learning_rate": 2.2886551361756324e-06,
|
| 9000 |
+
"loss": 11.728,
|
| 9001 |
+
"step": 1280
|
| 9002 |
+
},
|
| 9003 |
+
{
|
| 9004 |
+
"epoch": 0.10808986393840313,
|
| 9005 |
+
"grad_norm": 0.04231094568967819,
|
| 9006 |
+
"learning_rate": 2.239884945515469e-06,
|
| 9007 |
+
"loss": 11.734,
|
| 9008 |
+
"step": 1281
|
| 9009 |
+
},
|
| 9010 |
+
{
|
| 9011 |
+
"epoch": 0.10817424322328868,
|
| 9012 |
+
"grad_norm": 0.033634379506111145,
|
| 9013 |
+
"learning_rate": 2.191634116107755e-06,
|
| 9014 |
+
"loss": 11.7337,
|
| 9015 |
+
"step": 1282
|
| 9016 |
+
},
|
| 9017 |
+
{
|
| 9018 |
+
"epoch": 0.10825862250817424,
|
| 9019 |
+
"grad_norm": 0.040879081934690475,
|
| 9020 |
+
"learning_rate": 2.143902904290296e-06,
|
| 9021 |
+
"loss": 11.7407,
|
| 9022 |
+
"step": 1283
|
| 9023 |
+
},
|
| 9024 |
+
{
|
| 9025 |
+
"epoch": 0.1083430017930598,
|
| 9026 |
+
"grad_norm": 0.04845665395259857,
|
| 9027 |
+
"learning_rate": 2.0966915636403516e-06,
|
| 9028 |
+
"loss": 11.74,
|
| 9029 |
+
"step": 1284
|
| 9030 |
+
},
|
| 9031 |
+
{
|
| 9032 |
+
"epoch": 0.10842738107794536,
|
| 9033 |
+
"grad_norm": 0.05107761174440384,
|
| 9034 |
+
"learning_rate": 2.0500003449733017e-06,
|
| 9035 |
+
"loss": 11.7431,
|
| 9036 |
+
"step": 1285
|
| 9037 |
+
},
|
| 9038 |
+
{
|
| 9039 |
+
"epoch": 0.10851176036283093,
|
| 9040 |
+
"grad_norm": 0.04913756251335144,
|
| 9041 |
+
"learning_rate": 2.003829496341325e-06,
|
| 9042 |
+
"loss": 11.7349,
|
| 9043 |
+
"step": 1286
|
| 9044 |
+
},
|
| 9045 |
+
{
|
| 9046 |
+
"epoch": 0.10859613964771649,
|
| 9047 |
+
"grad_norm": 0.05221975967288017,
|
| 9048 |
+
"learning_rate": 1.9581792630320784e-06,
|
| 9049 |
+
"loss": 11.7328,
|
| 9050 |
+
"step": 1287
|
| 9051 |
+
},
|
| 9052 |
+
{
|
| 9053 |
+
"epoch": 0.10868051893260204,
|
| 9054 |
+
"grad_norm": 0.05367531627416611,
|
| 9055 |
+
"learning_rate": 1.9130498875673973e-06,
|
| 9056 |
+
"loss": 11.7395,
|
| 9057 |
+
"step": 1288
|
| 9058 |
+
},
|
| 9059 |
+
{
|
| 9060 |
+
"epoch": 0.10876489821748761,
|
| 9061 |
+
"grad_norm": 0.04138621315360069,
|
| 9062 |
+
"learning_rate": 1.868441609702032e-06,
|
| 9063 |
+
"loss": 11.7352,
|
| 9064 |
+
"step": 1289
|
| 9065 |
+
},
|
| 9066 |
+
{
|
| 9067 |
+
"epoch": 0.10884927750237317,
|
| 9068 |
+
"grad_norm": 0.04803365841507912,
|
| 9069 |
+
"learning_rate": 1.8243546664222677e-06,
|
| 9070 |
+
"loss": 11.7341,
|
| 9071 |
+
"step": 1290
|
| 9072 |
+
},
|
| 9073 |
+
{
|
| 9074 |
+
"epoch": 0.10893365678725873,
|
| 9075 |
+
"grad_norm": 0.04994744807481766,
|
| 9076 |
+
"learning_rate": 1.7807892919448178e-06,
|
| 9077 |
+
"loss": 11.7382,
|
| 9078 |
+
"step": 1291
|
| 9079 |
+
},
|
| 9080 |
+
{
|
| 9081 |
+
"epoch": 0.10901803607214429,
|
| 9082 |
+
"grad_norm": 0.06141884997487068,
|
| 9083 |
+
"learning_rate": 1.7377457177154555e-06,
|
| 9084 |
+
"loss": 11.7384,
|
| 9085 |
+
"step": 1292
|
| 9086 |
+
},
|
| 9087 |
+
{
|
| 9088 |
+
"epoch": 0.10910241535702984,
|
| 9089 |
+
"grad_norm": 0.05196886137127876,
|
| 9090 |
+
"learning_rate": 1.695224172407872e-06,
|
| 9091 |
+
"loss": 11.7427,
|
| 9092 |
+
"step": 1293
|
| 9093 |
+
},
|
| 9094 |
+
{
|
| 9095 |
+
"epoch": 0.10918679464191541,
|
| 9096 |
+
"grad_norm": 0.046019818633794785,
|
| 9097 |
+
"learning_rate": 1.6532248819223995e-06,
|
| 9098 |
+
"loss": 11.7394,
|
| 9099 |
+
"step": 1294
|
| 9100 |
+
},
|
| 9101 |
+
{
|
| 9102 |
+
"epoch": 0.10927117392680097,
|
| 9103 |
+
"grad_norm": 0.03441344574093819,
|
| 9104 |
+
"learning_rate": 1.611748069384844e-06,
|
| 9105 |
+
"loss": 11.7376,
|
| 9106 |
+
"step": 1295
|
| 9107 |
+
},
|
| 9108 |
+
{
|
| 9109 |
+
"epoch": 0.10935555321168654,
|
| 9110 |
+
"grad_norm": 0.03796188905835152,
|
| 9111 |
+
"learning_rate": 1.5707939551452776e-06,
|
| 9112 |
+
"loss": 11.7397,
|
| 9113 |
+
"step": 1296
|
| 9114 |
+
},
|
| 9115 |
+
{
|
| 9116 |
+
"epoch": 0.10943993249657209,
|
| 9117 |
+
"grad_norm": 0.037003178149461746,
|
| 9118 |
+
"learning_rate": 1.5303627567769041e-06,
|
| 9119 |
+
"loss": 11.7388,
|
| 9120 |
+
"step": 1297
|
| 9121 |
+
},
|
| 9122 |
+
{
|
| 9123 |
+
"epoch": 0.10952431178145765,
|
| 9124 |
+
"grad_norm": 0.06004711613059044,
|
| 9125 |
+
"learning_rate": 1.4904546890748606e-06,
|
| 9126 |
+
"loss": 11.7377,
|
| 9127 |
+
"step": 1298
|
| 9128 |
+
},
|
| 9129 |
+
{
|
| 9130 |
+
"epoch": 0.10960869106634322,
|
| 9131 |
+
"grad_norm": 0.03814685344696045,
|
| 9132 |
+
"learning_rate": 1.4510699640550852e-06,
|
| 9133 |
+
"loss": 11.7324,
|
| 9134 |
+
"step": 1299
|
| 9135 |
+
},
|
| 9136 |
+
{
|
| 9137 |
+
"epoch": 0.10969307035122877,
|
| 9138 |
+
"grad_norm": 0.054289188235998154,
|
| 9139 |
+
"learning_rate": 1.412208790953229e-06,
|
| 9140 |
+
"loss": 11.7418,
|
| 9141 |
+
"step": 1300
|
| 9142 |
+
},
|
| 9143 |
+
{
|
| 9144 |
+
"epoch": 0.10977744963611433,
|
| 9145 |
+
"grad_norm": 0.04306815192103386,
|
| 9146 |
+
"learning_rate": 1.3738713762235123e-06,
|
| 9147 |
+
"loss": 11.7396,
|
| 9148 |
+
"step": 1301
|
| 9149 |
+
},
|
| 9150 |
+
{
|
| 9151 |
+
"epoch": 0.1098618289209999,
|
| 9152 |
+
"grad_norm": 0.053123876452445984,
|
| 9153 |
+
"learning_rate": 1.3360579235376147e-06,
|
| 9154 |
+
"loss": 11.7362,
|
| 9155 |
+
"step": 1302
|
| 9156 |
+
},
|
| 9157 |
+
{
|
| 9158 |
+
"epoch": 0.10994620820588545,
|
| 9159 |
+
"grad_norm": 0.06724987924098969,
|
| 9160 |
+
"learning_rate": 1.2987686337836202e-06,
|
| 9161 |
+
"loss": 11.7369,
|
| 9162 |
+
"step": 1303
|
| 9163 |
+
},
|
| 9164 |
+
{
|
| 9165 |
+
"epoch": 0.11003058749077102,
|
| 9166 |
+
"grad_norm": 0.04288502410054207,
|
| 9167 |
+
"learning_rate": 1.2620037050649403e-06,
|
| 9168 |
+
"loss": 11.7417,
|
| 9169 |
+
"step": 1304
|
| 9170 |
+
},
|
| 9171 |
+
{
|
| 9172 |
+
"epoch": 0.11011496677565658,
|
| 9173 |
+
"grad_norm": 0.041843898594379425,
|
| 9174 |
+
"learning_rate": 1.225763332699259e-06,
|
| 9175 |
+
"loss": 11.7349,
|
| 9176 |
+
"step": 1305
|
| 9177 |
+
},
|
| 9178 |
+
{
|
| 9179 |
+
"epoch": 0.11019934606054213,
|
| 9180 |
+
"grad_norm": 0.052616916596889496,
|
| 9181 |
+
"learning_rate": 1.1900477092175011e-06,
|
| 9182 |
+
"loss": 11.7324,
|
| 9183 |
+
"step": 1306
|
| 9184 |
+
},
|
| 9185 |
+
{
|
| 9186 |
+
"epoch": 0.1102837253454277,
|
| 9187 |
+
"grad_norm": 0.055366210639476776,
|
| 9188 |
+
"learning_rate": 1.1548570243627987e-06,
|
| 9189 |
+
"loss": 11.7341,
|
| 9190 |
+
"step": 1307
|
| 9191 |
+
},
|
| 9192 |
+
{
|
| 9193 |
+
"epoch": 0.11036810463031325,
|
| 9194 |
+
"grad_norm": 0.05027055740356445,
|
| 9195 |
+
"learning_rate": 1.120191465089515e-06,
|
| 9196 |
+
"loss": 11.7393,
|
| 9197 |
+
"step": 1308
|
| 9198 |
+
},
|
| 9199 |
+
{
|
| 9200 |
+
"epoch": 0.11045248391519882,
|
| 9201 |
+
"grad_norm": 0.038022324442863464,
|
| 9202 |
+
"learning_rate": 1.0860512155621783e-06,
|
| 9203 |
+
"loss": 11.7377,
|
| 9204 |
+
"step": 1309
|
| 9205 |
+
},
|
| 9206 |
+
{
|
| 9207 |
+
"epoch": 0.11053686320008438,
|
| 9208 |
+
"grad_norm": 0.05578352138400078,
|
| 9209 |
+
"learning_rate": 1.0524364571546041e-06,
|
| 9210 |
+
"loss": 11.7355,
|
| 9211 |
+
"step": 1310
|
| 9212 |
+
},
|
| 9213 |
+
{
|
| 9214 |
+
"epoch": 0.11062124248496993,
|
| 9215 |
+
"grad_norm": 0.04797317460179329,
|
| 9216 |
+
"learning_rate": 1.019347368448842e-06,
|
| 9217 |
+
"loss": 11.7412,
|
| 9218 |
+
"step": 1311
|
| 9219 |
+
},
|
| 9220 |
+
{
|
| 9221 |
+
"epoch": 0.1107056217698555,
|
| 9222 |
+
"grad_norm": 0.04522743076086044,
|
| 9223 |
+
"learning_rate": 9.867841252342746e-07,
|
| 9224 |
+
"loss": 11.7355,
|
| 9225 |
+
"step": 1312
|
| 9226 |
+
},
|
| 9227 |
+
{
|
| 9228 |
+
"epoch": 0.11079000105474106,
|
| 9229 |
+
"grad_norm": 0.06885165721178055,
|
| 9230 |
+
"learning_rate": 9.547469005066979e-07,
|
| 9231 |
+
"loss": 11.7399,
|
| 9232 |
+
"step": 1313
|
| 9233 |
+
},
|
| 9234 |
+
{
|
| 9235 |
+
"epoch": 0.11087438033962663,
|
| 9236 |
+
"grad_norm": 0.05798881873488426,
|
| 9237 |
+
"learning_rate": 9.232358644673311e-07,
|
| 9238 |
+
"loss": 11.7363,
|
| 9239 |
+
"step": 1314
|
| 9240 |
+
},
|
| 9241 |
+
{
|
| 9242 |
+
"epoch": 0.11095875962451218,
|
| 9243 |
+
"grad_norm": 0.04205925762653351,
|
| 9244 |
+
"learning_rate": 8.922511845219971e-07,
|
| 9245 |
+
"loss": 11.7349,
|
| 9246 |
+
"step": 1315
|
| 9247 |
+
},
|
| 9248 |
+
{
|
| 9249 |
+
"epoch": 0.11104313890939774,
|
| 9250 |
+
"grad_norm": 0.05150824412703514,
|
| 9251 |
+
"learning_rate": 8.617930252801665e-07,
|
| 9252 |
+
"loss": 11.7446,
|
| 9253 |
+
"step": 1316
|
| 9254 |
+
},
|
| 9255 |
+
{
|
| 9256 |
+
"epoch": 0.1111275181942833,
|
| 9257 |
+
"grad_norm": 0.032801125198602676,
|
| 9258 |
+
"learning_rate": 8.31861548554147e-07,
|
| 9259 |
+
"loss": 11.7419,
|
| 9260 |
+
"step": 1317
|
| 9261 |
+
},
|
| 9262 |
+
{
|
| 9263 |
+
"epoch": 0.11121189747916886,
|
| 9264 |
+
"grad_norm": 0.03714403510093689,
|
| 9265 |
+
"learning_rate": 8.024569133581516e-07,
|
| 9266 |
+
"loss": 11.7379,
|
| 9267 |
+
"step": 1318
|
| 9268 |
+
},
|
| 9269 |
+
{
|
| 9270 |
+
"epoch": 0.11129627676405443,
|
| 9271 |
+
"grad_norm": 0.04341082647442818,
|
| 9272 |
+
"learning_rate": 7.7357927590751e-07,
|
| 9273 |
+
"loss": 11.7353,
|
| 9274 |
+
"step": 1319
|
| 9275 |
+
},
|
| 9276 |
+
{
|
| 9277 |
+
"epoch": 0.11138065604893999,
|
| 9278 |
+
"grad_norm": 0.04713427275419235,
|
| 9279 |
+
"learning_rate": 7.452287896178133e-07,
|
| 9280 |
+
"loss": 11.7349,
|
| 9281 |
+
"step": 1320
|
| 9282 |
+
},
|
| 9283 |
+
{
|
| 9284 |
+
"epoch": 0.11146503533382554,
|
| 9285 |
+
"grad_norm": 0.05022663623094559,
|
| 9286 |
+
"learning_rate": 7.174056051041045e-07,
|
| 9287 |
+
"loss": 11.7325,
|
| 9288 |
+
"step": 1321
|
| 9289 |
+
},
|
| 9290 |
+
{
|
| 9291 |
+
"epoch": 0.11154941461871111,
|
| 9292 |
+
"grad_norm": 0.04256540164351463,
|
| 9293 |
+
"learning_rate": 6.901098701800779e-07,
|
| 9294 |
+
"loss": 11.7368,
|
| 9295 |
+
"step": 1322
|
| 9296 |
+
},
|
| 9297 |
+
{
|
| 9298 |
+
"epoch": 0.11163379390359666,
|
| 9299 |
+
"grad_norm": 0.04342963546514511,
|
| 9300 |
+
"learning_rate": 6.633417298573142e-07,
|
| 9301 |
+
"loss": 11.7346,
|
| 9302 |
+
"step": 1323
|
| 9303 |
+
},
|
| 9304 |
+
{
|
| 9305 |
+
"epoch": 0.11171817318848223,
|
| 9306 |
+
"grad_norm": 0.046290263533592224,
|
| 9307 |
+
"learning_rate": 6.37101326344447e-07,
|
| 9308 |
+
"loss": 11.7324,
|
| 9309 |
+
"step": 1324
|
| 9310 |
+
},
|
| 9311 |
+
{
|
| 9312 |
+
"epoch": 0.11180255247336779,
|
| 9313 |
+
"grad_norm": 0.0498012937605381,
|
| 9314 |
+
"learning_rate": 6.113887990464861e-07,
|
| 9315 |
+
"loss": 11.7368,
|
| 9316 |
+
"step": 1325
|
| 9317 |
+
},
|
| 9318 |
+
{
|
| 9319 |
+
"epoch": 0.11188693175825334,
|
| 9320 |
+
"grad_norm": 0.04222957789897919,
|
| 9321 |
+
"learning_rate": 5.862042845640403e-07,
|
| 9322 |
+
"loss": 11.7331,
|
| 9323 |
+
"step": 1326
|
| 9324 |
+
},
|
| 9325 |
+
{
|
| 9326 |
+
"epoch": 0.11197131104313891,
|
| 9327 |
+
"grad_norm": 0.04639262706041336,
|
| 9328 |
+
"learning_rate": 5.61547916692573e-07,
|
| 9329 |
+
"loss": 11.7359,
|
| 9330 |
+
"step": 1327
|
| 9331 |
+
},
|
| 9332 |
+
{
|
| 9333 |
+
"epoch": 0.11205569032802447,
|
| 9334 |
+
"grad_norm": 0.04550090432167053,
|
| 9335 |
+
"learning_rate": 5.374198264217367e-07,
|
| 9336 |
+
"loss": 11.7349,
|
| 9337 |
+
"step": 1328
|
| 9338 |
+
},
|
| 9339 |
+
{
|
| 9340 |
+
"epoch": 0.11214006961291004,
|
| 9341 |
+
"grad_norm": 0.03915134817361832,
|
| 9342 |
+
"learning_rate": 5.138201419346178e-07,
|
| 9343 |
+
"loss": 11.7315,
|
| 9344 |
+
"step": 1329
|
| 9345 |
+
},
|
| 9346 |
+
{
|
| 9347 |
+
"epoch": 0.11222444889779559,
|
| 9348 |
+
"grad_norm": 0.05578768625855446,
|
| 9349 |
+
"learning_rate": 4.907489886071148e-07,
|
| 9350 |
+
"loss": 11.7411,
|
| 9351 |
+
"step": 1330
|
| 9352 |
+
},
|
| 9353 |
+
{
|
| 9354 |
+
"epoch": 0.11230882818268115,
|
| 9355 |
+
"grad_norm": 0.05181988701224327,
|
| 9356 |
+
"learning_rate": 4.682064890072502e-07,
|
| 9357 |
+
"loss": 11.7337,
|
| 9358 |
+
"step": 1331
|
| 9359 |
+
},
|
| 9360 |
+
{
|
| 9361 |
+
"epoch": 0.11239320746756672,
|
| 9362 |
+
"grad_norm": 0.06327461451292038,
|
| 9363 |
+
"learning_rate": 4.4619276289450397e-07,
|
| 9364 |
+
"loss": 11.7257,
|
| 9365 |
+
"step": 1332
|
| 9366 |
+
},
|
| 9367 |
+
{
|
| 9368 |
+
"epoch": 0.11247758675245227,
|
| 9369 |
+
"grad_norm": 0.060607727617025375,
|
| 9370 |
+
"learning_rate": 4.247079272191812e-07,
|
| 9371 |
+
"loss": 11.7296,
|
| 9372 |
+
"step": 1333
|
| 9373 |
+
},
|
| 9374 |
+
{
|
| 9375 |
+
"epoch": 0.11256196603733783,
|
| 9376 |
+
"grad_norm": 0.05216168984770775,
|
| 9377 |
+
"learning_rate": 4.037520961218233e-07,
|
| 9378 |
+
"loss": 11.7352,
|
| 9379 |
+
"step": 1334
|
| 9380 |
+
},
|
| 9381 |
+
{
|
| 9382 |
+
"epoch": 0.1126463453222234,
|
| 9383 |
+
"grad_norm": 0.049930181354284286,
|
| 9384 |
+
"learning_rate": 3.833253809325643e-07,
|
| 9385 |
+
"loss": 11.73,
|
| 9386 |
+
"step": 1335
|
| 9387 |
+
},
|
| 9388 |
+
{
|
| 9389 |
+
"epoch": 0.11273072460710895,
|
| 9390 |
+
"grad_norm": 0.04583895578980446,
|
| 9391 |
+
"learning_rate": 3.6342789017054234e-07,
|
| 9392 |
+
"loss": 11.7404,
|
| 9393 |
+
"step": 1336
|
| 9394 |
+
},
|
| 9395 |
+
{
|
| 9396 |
+
"epoch": 0.11281510389199452,
|
| 9397 |
+
"grad_norm": 0.0481545627117157,
|
| 9398 |
+
"learning_rate": 3.440597295433445e-07,
|
| 9399 |
+
"loss": 11.7377,
|
| 9400 |
+
"step": 1337
|
| 9401 |
+
},
|
| 9402 |
+
{
|
| 9403 |
+
"epoch": 0.11289948317688007,
|
| 9404 |
+
"grad_norm": 0.036132924258708954,
|
| 9405 |
+
"learning_rate": 3.2522100194644076e-07,
|
| 9406 |
+
"loss": 11.7386,
|
| 9407 |
+
"step": 1338
|
| 9408 |
+
},
|
| 9409 |
+
{
|
| 9410 |
+
"epoch": 0.11298386246176563,
|
| 9411 |
+
"grad_norm": 0.036397356539964676,
|
| 9412 |
+
"learning_rate": 3.069118074626176e-07,
|
| 9413 |
+
"loss": 11.7341,
|
| 9414 |
+
"step": 1339
|
| 9415 |
+
},
|
| 9416 |
+
{
|
| 9417 |
+
"epoch": 0.1130682417466512,
|
| 9418 |
+
"grad_norm": 0.04515663906931877,
|
| 9419 |
+
"learning_rate": 2.8913224336145625e-07,
|
| 9420 |
+
"loss": 11.7406,
|
| 9421 |
+
"step": 1340
|
| 9422 |
+
},
|
| 9423 |
+
{
|
| 9424 |
+
"epoch": 0.11315262103153675,
|
| 9425 |
+
"grad_norm": 0.04933186247944832,
|
| 9426 |
+
"learning_rate": 2.7188240409883324e-07,
|
| 9427 |
+
"loss": 11.7403,
|
| 9428 |
+
"step": 1341
|
| 9429 |
+
},
|
| 9430 |
+
{
|
| 9431 |
+
"epoch": 0.11323700031642232,
|
| 9432 |
+
"grad_norm": 0.035385578870773315,
|
| 9433 |
+
"learning_rate": 2.5516238131640945e-07,
|
| 9434 |
+
"loss": 11.7395,
|
| 9435 |
+
"step": 1342
|
| 9436 |
+
},
|
| 9437 |
+
{
|
| 9438 |
+
"epoch": 0.11332137960130788,
|
| 9439 |
+
"grad_norm": 0.03938493877649307,
|
| 9440 |
+
"learning_rate": 2.389722638410974e-07,
|
| 9441 |
+
"loss": 11.7361,
|
| 9442 |
+
"step": 1343
|
| 9443 |
+
},
|
| 9444 |
+
{
|
| 9445 |
+
"epoch": 0.11340575888619343,
|
| 9446 |
+
"grad_norm": 0.0395924337208271,
|
| 9447 |
+
"learning_rate": 2.2331213768468363e-07,
|
| 9448 |
+
"loss": 11.7371,
|
| 9449 |
+
"step": 1344
|
| 9450 |
+
},
|
| 9451 |
+
{
|
| 9452 |
+
"epoch": 0.113490138171079,
|
| 9453 |
+
"grad_norm": 0.045525308698415756,
|
| 9454 |
+
"learning_rate": 2.081820860432848e-07,
|
| 9455 |
+
"loss": 11.7387,
|
| 9456 |
+
"step": 1345
|
| 9457 |
+
},
|
| 9458 |
+
{
|
| 9459 |
+
"epoch": 0.11357451745596456,
|
| 9460 |
+
"grad_norm": 0.06034231185913086,
|
| 9461 |
+
"learning_rate": 1.9358218929693695e-07,
|
| 9462 |
+
"loss": 11.7373,
|
| 9463 |
+
"step": 1346
|
| 9464 |
+
},
|
| 9465 |
+
{
|
| 9466 |
+
"epoch": 0.11365889674085013,
|
| 9467 |
+
"grad_norm": 0.04225464537739754,
|
| 9468 |
+
"learning_rate": 1.795125250092067e-07,
|
| 9469 |
+
"loss": 11.7395,
|
| 9470 |
+
"step": 1347
|
| 9471 |
+
},
|
| 9472 |
+
{
|
| 9473 |
+
"epoch": 0.11374327602573568,
|
| 9474 |
+
"grad_norm": 0.04094812273979187,
|
| 9475 |
+
"learning_rate": 1.6597316792668072e-07,
|
| 9476 |
+
"loss": 11.7392,
|
| 9477 |
+
"step": 1348
|
| 9478 |
+
},
|
| 9479 |
+
{
|
| 9480 |
+
"epoch": 0.11382765531062124,
|
| 9481 |
+
"grad_norm": 0.03248784691095352,
|
| 9482 |
+
"learning_rate": 1.529641899786993e-07,
|
| 9483 |
+
"loss": 11.7423,
|
| 9484 |
+
"step": 1349
|
| 9485 |
+
},
|
| 9486 |
+
{
|
| 9487 |
+
"epoch": 0.1139120345955068,
|
| 9488 |
+
"grad_norm": 0.049137309193611145,
|
| 9489 |
+
"learning_rate": 1.4048566027685673e-07,
|
| 9490 |
+
"loss": 11.7341,
|
| 9491 |
+
"step": 1350
|
| 9492 |
+
},
|
| 9493 |
+
{
|
| 9494 |
+
"epoch": 0.11399641388039236,
|
| 9495 |
+
"grad_norm": 0.048104751855134964,
|
| 9496 |
+
"learning_rate": 1.2853764511471245e-07,
|
| 9497 |
+
"loss": 11.7328,
|
| 9498 |
+
"step": 1351
|
| 9499 |
+
},
|
| 9500 |
+
{
|
| 9501 |
+
"epoch": 0.11408079316527793,
|
| 9502 |
+
"grad_norm": 0.055573612451553345,
|
| 9503 |
+
"learning_rate": 1.1712020796738054e-07,
|
| 9504 |
+
"loss": 11.7343,
|
| 9505 |
+
"step": 1352
|
| 9506 |
+
},
|
| 9507 |
+
{
|
| 9508 |
+
"epoch": 0.11416517245016349,
|
| 9509 |
+
"grad_norm": 0.04170134663581848,
|
| 9510 |
+
"learning_rate": 1.0623340949125204e-07,
|
| 9511 |
+
"loss": 11.741,
|
| 9512 |
+
"step": 1353
|
| 9513 |
+
},
|
| 9514 |
+
{
|
| 9515 |
+
"epoch": 0.11424955173504904,
|
| 9516 |
+
"grad_norm": 0.04001680761575699,
|
| 9517 |
+
"learning_rate": 9.587730752362856e-08,
|
| 9518 |
+
"loss": 11.7284,
|
| 9519 |
+
"step": 1354
|
| 9520 |
+
},
|
| 9521 |
+
{
|
| 9522 |
+
"epoch": 0.11433393101993461,
|
| 9523 |
+
"grad_norm": 0.03968476876616478,
|
| 9524 |
+
"learning_rate": 8.605195708242252e-08,
|
| 9525 |
+
"loss": 11.7372,
|
| 9526 |
+
"step": 1355
|
| 9527 |
+
},
|
| 9528 |
+
{
|
| 9529 |
+
"epoch": 0.11441831030482016,
|
| 9530 |
+
"grad_norm": 0.034107908606529236,
|
| 9531 |
+
"learning_rate": 7.675741036587969e-08,
|
| 9532 |
+
"loss": 11.7374,
|
| 9533 |
+
"step": 1356
|
| 9534 |
+
},
|
| 9535 |
+
{
|
| 9536 |
+
"epoch": 0.11450268958970573,
|
| 9537 |
+
"grad_norm": 0.05979205295443535,
|
| 9538 |
+
"learning_rate": 6.799371675230148e-08,
|
| 9539 |
+
"loss": 11.7372,
|
| 9540 |
+
"step": 1357
|
| 9541 |
+
},
|
| 9542 |
+
{
|
| 9543 |
+
"epoch": 0.11458706887459129,
|
| 9544 |
+
"grad_norm": 0.06530775129795074,
|
| 9545 |
+
"learning_rate": 5.976092279974532e-08,
|
| 9546 |
+
"loss": 11.7462,
|
| 9547 |
+
"step": 1358
|
| 9548 |
+
},
|
| 9549 |
+
{
|
| 9550 |
+
"epoch": 0.11467144815947684,
|
| 9551 |
+
"grad_norm": 0.04594489932060242,
|
| 9552 |
+
"learning_rate": 5.205907224583584e-08,
|
| 9553 |
+
"loss": 11.7336,
|
| 9554 |
+
"step": 1359
|
| 9555 |
+
},
|
| 9556 |
+
{
|
| 9557 |
+
"epoch": 0.11475582744436241,
|
| 9558 |
+
"grad_norm": 0.05702807754278183,
|
| 9559 |
+
"learning_rate": 4.4888206007498436e-08,
|
| 9560 |
+
"loss": 11.7473,
|
| 9561 |
+
"step": 1360
|
| 9562 |
+
},
|
| 9563 |
+
{
|
| 9564 |
+
"epoch": 0.11484020672924797,
|
| 9565 |
+
"grad_norm": 0.029665730893611908,
|
| 9566 |
+
"learning_rate": 3.824836218072614e-08,
|
| 9567 |
+
"loss": 11.7388,
|
| 9568 |
+
"step": 1361
|
| 9569 |
+
},
|
| 9570 |
+
{
|
| 9571 |
+
"epoch": 0.11492458601413354,
|
| 9572 |
+
"grad_norm": 0.04100637510418892,
|
| 9573 |
+
"learning_rate": 3.2139576040413066e-08,
|
| 9574 |
+
"loss": 11.7351,
|
| 9575 |
+
"step": 1362
|
| 9576 |
+
},
|
| 9577 |
+
{
|
| 9578 |
+
"epoch": 0.11500896529901909,
|
| 9579 |
+
"grad_norm": 0.04064370319247246,
|
| 9580 |
+
"learning_rate": 2.6561880040165688e-08,
|
| 9581 |
+
"loss": 11.7427,
|
| 9582 |
+
"step": 1363
|
| 9583 |
+
},
|
| 9584 |
+
{
|
| 9585 |
+
"epoch": 0.11509334458390465,
|
| 9586 |
+
"grad_norm": 0.04216818884015083,
|
| 9587 |
+
"learning_rate": 2.151530381209188e-08,
|
| 9588 |
+
"loss": 11.7385,
|
| 9589 |
+
"step": 1364
|
| 9590 |
+
},
|
| 9591 |
+
{
|
| 9592 |
+
"epoch": 0.11517772386879022,
|
| 9593 |
+
"grad_norm": 0.05198872834444046,
|
| 9594 |
+
"learning_rate": 1.699987416667881e-08,
|
| 9595 |
+
"loss": 11.7315,
|
| 9596 |
+
"step": 1365
|
| 9597 |
+
},
|
| 9598 |
+
{
|
| 9599 |
+
"epoch": 0.11526210315367577,
|
| 9600 |
+
"grad_norm": 0.04517610743641853,
|
| 9601 |
+
"learning_rate": 1.3015615092637489e-08,
|
| 9602 |
+
"loss": 11.7394,
|
| 9603 |
+
"step": 1366
|
| 9604 |
+
},
|
| 9605 |
+
{
|
| 9606 |
+
"epoch": 0.11534648243856133,
|
| 9607 |
+
"grad_norm": 0.06256133317947388,
|
| 9608 |
+
"learning_rate": 9.56254775678067e-09,
|
| 9609 |
+
"loss": 11.7426,
|
| 9610 |
+
"step": 1367
|
| 9611 |
+
},
|
| 9612 |
+
{
|
| 9613 |
+
"epoch": 0.1154308617234469,
|
| 9614 |
+
"grad_norm": 0.049372196197509766,
|
| 9615 |
+
"learning_rate": 6.640690503889602e-09,
|
| 9616 |
+
"loss": 11.741,
|
| 9617 |
+
"step": 1368
|
| 9618 |
+
},
|
| 9619 |
+
{
|
| 9620 |
+
"epoch": 0.11551524100833245,
|
| 9621 |
+
"grad_norm": 0.03914564475417137,
|
| 9622 |
+
"learning_rate": 4.250058856669626e-09,
|
| 9623 |
+
"loss": 11.7337,
|
| 9624 |
+
"step": 1369
|
| 9625 |
+
},
|
| 9626 |
+
{
|
| 9627 |
+
"epoch": 0.11559962029321802,
|
| 9628 |
+
"grad_norm": 0.04299304261803627,
|
| 9629 |
+
"learning_rate": 2.3906655156058498e-09,
|
| 9630 |
+
"loss": 11.7355,
|
| 9631 |
+
"step": 1370
|
| 9632 |
+
},
|
| 9633 |
+
{
|
| 9634 |
+
"epoch": 0.11568399957810357,
|
| 9635 |
+
"grad_norm": 0.03753099963068962,
|
| 9636 |
+
"learning_rate": 1.0625203589187394e-09,
|
| 9637 |
+
"loss": 11.7316,
|
| 9638 |
+
"step": 1371
|
| 9639 |
+
},
|
| 9640 |
+
{
|
| 9641 |
+
"epoch": 0.11576837886298913,
|
| 9642 |
+
"grad_norm": 0.03077964298427105,
|
| 9643 |
+
"learning_rate": 2.656304425308065e-10,
|
| 9644 |
+
"loss": 11.7441,
|
| 9645 |
+
"step": 1372
|
| 9646 |
+
},
|
| 9647 |
+
{
|
| 9648 |
+
"epoch": 0.1158527581478747,
|
| 9649 |
+
"grad_norm": 0.04818045720458031,
|
| 9650 |
+
"learning_rate": 0.0,
|
| 9651 |
+
"loss": 11.7332,
|
| 9652 |
+
"step": 1373
|
| 9653 |
}
|
| 9654 |
],
|
| 9655 |
"logging_steps": 1,
|
|
|
|
| 9664 |
"should_evaluate": false,
|
| 9665 |
"should_log": false,
|
| 9666 |
"should_save": true,
|
| 9667 |
+
"should_training_stop": true
|
| 9668 |
},
|
| 9669 |
"attributes": {}
|
| 9670 |
}
|
| 9671 |
},
|
| 9672 |
+
"total_flos": 69721362137088.0,
|
| 9673 |
"train_batch_size": 2,
|
| 9674 |
"trial_name": null,
|
| 9675 |
"trial_params": null
|