Training in progress, step 758, 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 60010048
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:510a52ac805c222aef3cf677e392c568b0e01bfafa8b84d5fe860020a4affe93
|
| 3 |
size 60010048
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 30428180
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cd03054be855d638dc8730b0926abbe4afba52392c7c1ae7bffc7808f73ee7c2
|
| 3 |
size 30428180
|
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 sha256:2eaf8013c84ba7b2a92613bbe740d03f2043c8adf3497981ed60c3e17ec19ae9
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -2912,6 +2912,2407 @@
|
|
| 2912 |
"learning_rate": 1.2231812577065353e-05,
|
| 2913 |
"loss": 1.2451,
|
| 2914 |
"step": 415
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2915 |
}
|
| 2916 |
],
|
| 2917 |
"logging_steps": 1,
|
|
@@ -2931,7 +5332,7 @@
|
|
| 2931 |
"attributes": {}
|
| 2932 |
}
|
| 2933 |
},
|
| 2934 |
-
"total_flos":
|
| 2935 |
"train_batch_size": 4,
|
| 2936 |
"trial_name": null,
|
| 2937 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.8318244170096022,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 758,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 2912 |
"learning_rate": 1.2231812577065353e-05,
|
| 2913 |
"loss": 1.2451,
|
| 2914 |
"step": 415
|
| 2915 |
+
},
|
| 2916 |
+
{
|
| 2917 |
+
"epoch": 0.4565157750342936,
|
| 2918 |
+
"grad_norm": 0.2396046668291092,
|
| 2919 |
+
"learning_rate": 1.2207151664611591e-05,
|
| 2920 |
+
"loss": 1.2292,
|
| 2921 |
+
"step": 416
|
| 2922 |
+
},
|
| 2923 |
+
{
|
| 2924 |
+
"epoch": 0.4576131687242798,
|
| 2925 |
+
"grad_norm": 0.25232142210006714,
|
| 2926 |
+
"learning_rate": 1.2182490752157832e-05,
|
| 2927 |
+
"loss": 1.1666,
|
| 2928 |
+
"step": 417
|
| 2929 |
+
},
|
| 2930 |
+
{
|
| 2931 |
+
"epoch": 0.4587105624142661,
|
| 2932 |
+
"grad_norm": 0.2253977507352829,
|
| 2933 |
+
"learning_rate": 1.215782983970407e-05,
|
| 2934 |
+
"loss": 1.2109,
|
| 2935 |
+
"step": 418
|
| 2936 |
+
},
|
| 2937 |
+
{
|
| 2938 |
+
"epoch": 0.4598079561042524,
|
| 2939 |
+
"grad_norm": 0.22141477465629578,
|
| 2940 |
+
"learning_rate": 1.213316892725031e-05,
|
| 2941 |
+
"loss": 1.3102,
|
| 2942 |
+
"step": 419
|
| 2943 |
+
},
|
| 2944 |
+
{
|
| 2945 |
+
"epoch": 0.4609053497942387,
|
| 2946 |
+
"grad_norm": 0.24601267278194427,
|
| 2947 |
+
"learning_rate": 1.2108508014796549e-05,
|
| 2948 |
+
"loss": 1.2358,
|
| 2949 |
+
"step": 420
|
| 2950 |
+
},
|
| 2951 |
+
{
|
| 2952 |
+
"epoch": 0.462002743484225,
|
| 2953 |
+
"grad_norm": 0.25011298060417175,
|
| 2954 |
+
"learning_rate": 1.2083847102342788e-05,
|
| 2955 |
+
"loss": 1.2069,
|
| 2956 |
+
"step": 421
|
| 2957 |
+
},
|
| 2958 |
+
{
|
| 2959 |
+
"epoch": 0.4631001371742112,
|
| 2960 |
+
"grad_norm": 0.24054169654846191,
|
| 2961 |
+
"learning_rate": 1.2059186189889027e-05,
|
| 2962 |
+
"loss": 1.3289,
|
| 2963 |
+
"step": 422
|
| 2964 |
+
},
|
| 2965 |
+
{
|
| 2966 |
+
"epoch": 0.4641975308641975,
|
| 2967 |
+
"grad_norm": 0.236949160695076,
|
| 2968 |
+
"learning_rate": 1.2034525277435265e-05,
|
| 2969 |
+
"loss": 1.237,
|
| 2970 |
+
"step": 423
|
| 2971 |
+
},
|
| 2972 |
+
{
|
| 2973 |
+
"epoch": 0.4652949245541838,
|
| 2974 |
+
"grad_norm": 0.24198143184185028,
|
| 2975 |
+
"learning_rate": 1.2009864364981504e-05,
|
| 2976 |
+
"loss": 1.244,
|
| 2977 |
+
"step": 424
|
| 2978 |
+
},
|
| 2979 |
+
{
|
| 2980 |
+
"epoch": 0.4663923182441701,
|
| 2981 |
+
"grad_norm": 0.21737278997898102,
|
| 2982 |
+
"learning_rate": 1.1985203452527745e-05,
|
| 2983 |
+
"loss": 1.2327,
|
| 2984 |
+
"step": 425
|
| 2985 |
+
},
|
| 2986 |
+
{
|
| 2987 |
+
"epoch": 0.4674897119341564,
|
| 2988 |
+
"grad_norm": 0.25471195578575134,
|
| 2989 |
+
"learning_rate": 1.1960542540073982e-05,
|
| 2990 |
+
"loss": 1.2097,
|
| 2991 |
+
"step": 426
|
| 2992 |
+
},
|
| 2993 |
+
{
|
| 2994 |
+
"epoch": 0.4685871056241427,
|
| 2995 |
+
"grad_norm": 0.2115827053785324,
|
| 2996 |
+
"learning_rate": 1.1935881627620223e-05,
|
| 2997 |
+
"loss": 1.3403,
|
| 2998 |
+
"step": 427
|
| 2999 |
+
},
|
| 3000 |
+
{
|
| 3001 |
+
"epoch": 0.4696844993141289,
|
| 3002 |
+
"grad_norm": 0.2685728669166565,
|
| 3003 |
+
"learning_rate": 1.1911220715166463e-05,
|
| 3004 |
+
"loss": 1.213,
|
| 3005 |
+
"step": 428
|
| 3006 |
+
},
|
| 3007 |
+
{
|
| 3008 |
+
"epoch": 0.4707818930041152,
|
| 3009 |
+
"grad_norm": 0.21602900326251984,
|
| 3010 |
+
"learning_rate": 1.18865598027127e-05,
|
| 3011 |
+
"loss": 1.246,
|
| 3012 |
+
"step": 429
|
| 3013 |
+
},
|
| 3014 |
+
{
|
| 3015 |
+
"epoch": 0.4718792866941015,
|
| 3016 |
+
"grad_norm": 0.21242216229438782,
|
| 3017 |
+
"learning_rate": 1.1861898890258941e-05,
|
| 3018 |
+
"loss": 1.2576,
|
| 3019 |
+
"step": 430
|
| 3020 |
+
},
|
| 3021 |
+
{
|
| 3022 |
+
"epoch": 0.4729766803840878,
|
| 3023 |
+
"grad_norm": 0.24854035675525665,
|
| 3024 |
+
"learning_rate": 1.183723797780518e-05,
|
| 3025 |
+
"loss": 1.3029,
|
| 3026 |
+
"step": 431
|
| 3027 |
+
},
|
| 3028 |
+
{
|
| 3029 |
+
"epoch": 0.4740740740740741,
|
| 3030 |
+
"grad_norm": 0.21696555614471436,
|
| 3031 |
+
"learning_rate": 1.1812577065351419e-05,
|
| 3032 |
+
"loss": 1.2895,
|
| 3033 |
+
"step": 432
|
| 3034 |
+
},
|
| 3035 |
+
{
|
| 3036 |
+
"epoch": 0.47517146776406033,
|
| 3037 |
+
"grad_norm": 0.23158885538578033,
|
| 3038 |
+
"learning_rate": 1.1787916152897658e-05,
|
| 3039 |
+
"loss": 1.1982,
|
| 3040 |
+
"step": 433
|
| 3041 |
+
},
|
| 3042 |
+
{
|
| 3043 |
+
"epoch": 0.4762688614540466,
|
| 3044 |
+
"grad_norm": 0.198283389210701,
|
| 3045 |
+
"learning_rate": 1.1763255240443899e-05,
|
| 3046 |
+
"loss": 1.2691,
|
| 3047 |
+
"step": 434
|
| 3048 |
+
},
|
| 3049 |
+
{
|
| 3050 |
+
"epoch": 0.4773662551440329,
|
| 3051 |
+
"grad_norm": 0.23988409340381622,
|
| 3052 |
+
"learning_rate": 1.1738594327990136e-05,
|
| 3053 |
+
"loss": 1.2835,
|
| 3054 |
+
"step": 435
|
| 3055 |
+
},
|
| 3056 |
+
{
|
| 3057 |
+
"epoch": 0.4784636488340192,
|
| 3058 |
+
"grad_norm": 0.25393354892730713,
|
| 3059 |
+
"learning_rate": 1.1713933415536376e-05,
|
| 3060 |
+
"loss": 1.208,
|
| 3061 |
+
"step": 436
|
| 3062 |
+
},
|
| 3063 |
+
{
|
| 3064 |
+
"epoch": 0.4795610425240055,
|
| 3065 |
+
"grad_norm": 0.2547854483127594,
|
| 3066 |
+
"learning_rate": 1.1689272503082615e-05,
|
| 3067 |
+
"loss": 1.2615,
|
| 3068 |
+
"step": 437
|
| 3069 |
+
},
|
| 3070 |
+
{
|
| 3071 |
+
"epoch": 0.4806584362139918,
|
| 3072 |
+
"grad_norm": 0.25701621174812317,
|
| 3073 |
+
"learning_rate": 1.1664611590628854e-05,
|
| 3074 |
+
"loss": 1.1242,
|
| 3075 |
+
"step": 438
|
| 3076 |
+
},
|
| 3077 |
+
{
|
| 3078 |
+
"epoch": 0.48175582990397803,
|
| 3079 |
+
"grad_norm": 0.20124734938144684,
|
| 3080 |
+
"learning_rate": 1.1639950678175093e-05,
|
| 3081 |
+
"loss": 1.2604,
|
| 3082 |
+
"step": 439
|
| 3083 |
+
},
|
| 3084 |
+
{
|
| 3085 |
+
"epoch": 0.4828532235939643,
|
| 3086 |
+
"grad_norm": 0.19933612644672394,
|
| 3087 |
+
"learning_rate": 1.1615289765721334e-05,
|
| 3088 |
+
"loss": 1.2232,
|
| 3089 |
+
"step": 440
|
| 3090 |
+
},
|
| 3091 |
+
{
|
| 3092 |
+
"epoch": 0.4839506172839506,
|
| 3093 |
+
"grad_norm": 0.21274983882904053,
|
| 3094 |
+
"learning_rate": 1.1590628853267571e-05,
|
| 3095 |
+
"loss": 1.2456,
|
| 3096 |
+
"step": 441
|
| 3097 |
+
},
|
| 3098 |
+
{
|
| 3099 |
+
"epoch": 0.4850480109739369,
|
| 3100 |
+
"grad_norm": 0.2665160894393921,
|
| 3101 |
+
"learning_rate": 1.1565967940813812e-05,
|
| 3102 |
+
"loss": 1.2256,
|
| 3103 |
+
"step": 442
|
| 3104 |
+
},
|
| 3105 |
+
{
|
| 3106 |
+
"epoch": 0.4861454046639232,
|
| 3107 |
+
"grad_norm": 0.23317915201187134,
|
| 3108 |
+
"learning_rate": 1.1541307028360049e-05,
|
| 3109 |
+
"loss": 1.2468,
|
| 3110 |
+
"step": 443
|
| 3111 |
+
},
|
| 3112 |
+
{
|
| 3113 |
+
"epoch": 0.4872427983539095,
|
| 3114 |
+
"grad_norm": 0.23791036009788513,
|
| 3115 |
+
"learning_rate": 1.151664611590629e-05,
|
| 3116 |
+
"loss": 1.1422,
|
| 3117 |
+
"step": 444
|
| 3118 |
+
},
|
| 3119 |
+
{
|
| 3120 |
+
"epoch": 0.4883401920438957,
|
| 3121 |
+
"grad_norm": 0.22260887920856476,
|
| 3122 |
+
"learning_rate": 1.149198520345253e-05,
|
| 3123 |
+
"loss": 1.2848,
|
| 3124 |
+
"step": 445
|
| 3125 |
+
},
|
| 3126 |
+
{
|
| 3127 |
+
"epoch": 0.489437585733882,
|
| 3128 |
+
"grad_norm": 0.2535932660102844,
|
| 3129 |
+
"learning_rate": 1.1467324290998767e-05,
|
| 3130 |
+
"loss": 1.2961,
|
| 3131 |
+
"step": 446
|
| 3132 |
+
},
|
| 3133 |
+
{
|
| 3134 |
+
"epoch": 0.4905349794238683,
|
| 3135 |
+
"grad_norm": 0.23415204882621765,
|
| 3136 |
+
"learning_rate": 1.1442663378545008e-05,
|
| 3137 |
+
"loss": 1.189,
|
| 3138 |
+
"step": 447
|
| 3139 |
+
},
|
| 3140 |
+
{
|
| 3141 |
+
"epoch": 0.4916323731138546,
|
| 3142 |
+
"grad_norm": 0.2729712128639221,
|
| 3143 |
+
"learning_rate": 1.1418002466091247e-05,
|
| 3144 |
+
"loss": 1.1392,
|
| 3145 |
+
"step": 448
|
| 3146 |
+
},
|
| 3147 |
+
{
|
| 3148 |
+
"epoch": 0.4927297668038409,
|
| 3149 |
+
"grad_norm": 0.22939635813236237,
|
| 3150 |
+
"learning_rate": 1.1393341553637486e-05,
|
| 3151 |
+
"loss": 1.2495,
|
| 3152 |
+
"step": 449
|
| 3153 |
+
},
|
| 3154 |
+
{
|
| 3155 |
+
"epoch": 0.49382716049382713,
|
| 3156 |
+
"grad_norm": 0.2639816403388977,
|
| 3157 |
+
"learning_rate": 1.1368680641183724e-05,
|
| 3158 |
+
"loss": 1.2232,
|
| 3159 |
+
"step": 450
|
| 3160 |
+
},
|
| 3161 |
+
{
|
| 3162 |
+
"epoch": 0.4949245541838134,
|
| 3163 |
+
"grad_norm": 0.2258535474538803,
|
| 3164 |
+
"learning_rate": 1.1344019728729965e-05,
|
| 3165 |
+
"loss": 1.3115,
|
| 3166 |
+
"step": 451
|
| 3167 |
+
},
|
| 3168 |
+
{
|
| 3169 |
+
"epoch": 0.4960219478737997,
|
| 3170 |
+
"grad_norm": 0.23552873730659485,
|
| 3171 |
+
"learning_rate": 1.1319358816276202e-05,
|
| 3172 |
+
"loss": 1.3037,
|
| 3173 |
+
"step": 452
|
| 3174 |
+
},
|
| 3175 |
+
{
|
| 3176 |
+
"epoch": 0.497119341563786,
|
| 3177 |
+
"grad_norm": 0.22219283878803253,
|
| 3178 |
+
"learning_rate": 1.1294697903822443e-05,
|
| 3179 |
+
"loss": 1.3038,
|
| 3180 |
+
"step": 453
|
| 3181 |
+
},
|
| 3182 |
+
{
|
| 3183 |
+
"epoch": 0.4982167352537723,
|
| 3184 |
+
"grad_norm": 0.2428772896528244,
|
| 3185 |
+
"learning_rate": 1.1270036991368682e-05,
|
| 3186 |
+
"loss": 1.1719,
|
| 3187 |
+
"step": 454
|
| 3188 |
+
},
|
| 3189 |
+
{
|
| 3190 |
+
"epoch": 0.4993141289437586,
|
| 3191 |
+
"grad_norm": 0.2560546100139618,
|
| 3192 |
+
"learning_rate": 1.124537607891492e-05,
|
| 3193 |
+
"loss": 1.1719,
|
| 3194 |
+
"step": 455
|
| 3195 |
+
},
|
| 3196 |
+
{
|
| 3197 |
+
"epoch": 0.5004115226337449,
|
| 3198 |
+
"grad_norm": 0.25714465975761414,
|
| 3199 |
+
"learning_rate": 1.122071516646116e-05,
|
| 3200 |
+
"loss": 1.31,
|
| 3201 |
+
"step": 456
|
| 3202 |
+
},
|
| 3203 |
+
{
|
| 3204 |
+
"epoch": 0.5015089163237312,
|
| 3205 |
+
"grad_norm": 0.2870176136493683,
|
| 3206 |
+
"learning_rate": 1.11960542540074e-05,
|
| 3207 |
+
"loss": 1.3136,
|
| 3208 |
+
"step": 457
|
| 3209 |
+
},
|
| 3210 |
+
{
|
| 3211 |
+
"epoch": 0.5026063100137175,
|
| 3212 |
+
"grad_norm": 0.22786717116832733,
|
| 3213 |
+
"learning_rate": 1.1171393341553637e-05,
|
| 3214 |
+
"loss": 1.2645,
|
| 3215 |
+
"step": 458
|
| 3216 |
+
},
|
| 3217 |
+
{
|
| 3218 |
+
"epoch": 0.5037037037037037,
|
| 3219 |
+
"grad_norm": 0.22393208742141724,
|
| 3220 |
+
"learning_rate": 1.1146732429099878e-05,
|
| 3221 |
+
"loss": 1.2478,
|
| 3222 |
+
"step": 459
|
| 3223 |
+
},
|
| 3224 |
+
{
|
| 3225 |
+
"epoch": 0.50480109739369,
|
| 3226 |
+
"grad_norm": 0.2480694204568863,
|
| 3227 |
+
"learning_rate": 1.1122071516646115e-05,
|
| 3228 |
+
"loss": 1.2668,
|
| 3229 |
+
"step": 460
|
| 3230 |
+
},
|
| 3231 |
+
{
|
| 3232 |
+
"epoch": 0.5058984910836762,
|
| 3233 |
+
"grad_norm": 0.2237052023410797,
|
| 3234 |
+
"learning_rate": 1.1097410604192356e-05,
|
| 3235 |
+
"loss": 1.2082,
|
| 3236 |
+
"step": 461
|
| 3237 |
+
},
|
| 3238 |
+
{
|
| 3239 |
+
"epoch": 0.5069958847736625,
|
| 3240 |
+
"grad_norm": 0.2555733323097229,
|
| 3241 |
+
"learning_rate": 1.1072749691738596e-05,
|
| 3242 |
+
"loss": 1.2893,
|
| 3243 |
+
"step": 462
|
| 3244 |
+
},
|
| 3245 |
+
{
|
| 3246 |
+
"epoch": 0.5080932784636488,
|
| 3247 |
+
"grad_norm": 0.22584785521030426,
|
| 3248 |
+
"learning_rate": 1.1048088779284834e-05,
|
| 3249 |
+
"loss": 1.2195,
|
| 3250 |
+
"step": 463
|
| 3251 |
+
},
|
| 3252 |
+
{
|
| 3253 |
+
"epoch": 0.5091906721536351,
|
| 3254 |
+
"grad_norm": 0.26389771699905396,
|
| 3255 |
+
"learning_rate": 1.1023427866831074e-05,
|
| 3256 |
+
"loss": 1.3014,
|
| 3257 |
+
"step": 464
|
| 3258 |
+
},
|
| 3259 |
+
{
|
| 3260 |
+
"epoch": 0.5102880658436214,
|
| 3261 |
+
"grad_norm": 0.25652697682380676,
|
| 3262 |
+
"learning_rate": 1.0998766954377313e-05,
|
| 3263 |
+
"loss": 1.2505,
|
| 3264 |
+
"step": 465
|
| 3265 |
+
},
|
| 3266 |
+
{
|
| 3267 |
+
"epoch": 0.5113854595336077,
|
| 3268 |
+
"grad_norm": 0.2295251190662384,
|
| 3269 |
+
"learning_rate": 1.097410604192355e-05,
|
| 3270 |
+
"loss": 1.3409,
|
| 3271 |
+
"step": 466
|
| 3272 |
+
},
|
| 3273 |
+
{
|
| 3274 |
+
"epoch": 0.512482853223594,
|
| 3275 |
+
"grad_norm": 0.26109209656715393,
|
| 3276 |
+
"learning_rate": 1.0949445129469791e-05,
|
| 3277 |
+
"loss": 1.2654,
|
| 3278 |
+
"step": 467
|
| 3279 |
+
},
|
| 3280 |
+
{
|
| 3281 |
+
"epoch": 0.5135802469135803,
|
| 3282 |
+
"grad_norm": 0.2687307596206665,
|
| 3283 |
+
"learning_rate": 1.0924784217016032e-05,
|
| 3284 |
+
"loss": 1.2555,
|
| 3285 |
+
"step": 468
|
| 3286 |
+
},
|
| 3287 |
+
{
|
| 3288 |
+
"epoch": 0.5146776406035666,
|
| 3289 |
+
"grad_norm": 0.25440263748168945,
|
| 3290 |
+
"learning_rate": 1.0900123304562269e-05,
|
| 3291 |
+
"loss": 1.3099,
|
| 3292 |
+
"step": 469
|
| 3293 |
+
},
|
| 3294 |
+
{
|
| 3295 |
+
"epoch": 0.5157750342935528,
|
| 3296 |
+
"grad_norm": 0.2510749101638794,
|
| 3297 |
+
"learning_rate": 1.087546239210851e-05,
|
| 3298 |
+
"loss": 1.2036,
|
| 3299 |
+
"step": 470
|
| 3300 |
+
},
|
| 3301 |
+
{
|
| 3302 |
+
"epoch": 0.516872427983539,
|
| 3303 |
+
"grad_norm": 0.26294615864753723,
|
| 3304 |
+
"learning_rate": 1.0850801479654748e-05,
|
| 3305 |
+
"loss": 1.2649,
|
| 3306 |
+
"step": 471
|
| 3307 |
+
},
|
| 3308 |
+
{
|
| 3309 |
+
"epoch": 0.5179698216735253,
|
| 3310 |
+
"grad_norm": 0.23490417003631592,
|
| 3311 |
+
"learning_rate": 1.0826140567200987e-05,
|
| 3312 |
+
"loss": 1.2192,
|
| 3313 |
+
"step": 472
|
| 3314 |
+
},
|
| 3315 |
+
{
|
| 3316 |
+
"epoch": 0.5190672153635116,
|
| 3317 |
+
"grad_norm": 0.22545063495635986,
|
| 3318 |
+
"learning_rate": 1.0801479654747226e-05,
|
| 3319 |
+
"loss": 1.2995,
|
| 3320 |
+
"step": 473
|
| 3321 |
+
},
|
| 3322 |
+
{
|
| 3323 |
+
"epoch": 0.5201646090534979,
|
| 3324 |
+
"grad_norm": 0.23707161843776703,
|
| 3325 |
+
"learning_rate": 1.0776818742293467e-05,
|
| 3326 |
+
"loss": 1.2689,
|
| 3327 |
+
"step": 474
|
| 3328 |
+
},
|
| 3329 |
+
{
|
| 3330 |
+
"epoch": 0.5212620027434842,
|
| 3331 |
+
"grad_norm": 0.2702939212322235,
|
| 3332 |
+
"learning_rate": 1.0752157829839704e-05,
|
| 3333 |
+
"loss": 1.1514,
|
| 3334 |
+
"step": 475
|
| 3335 |
+
},
|
| 3336 |
+
{
|
| 3337 |
+
"epoch": 0.5223593964334705,
|
| 3338 |
+
"grad_norm": 0.23080959916114807,
|
| 3339 |
+
"learning_rate": 1.0727496917385945e-05,
|
| 3340 |
+
"loss": 1.2791,
|
| 3341 |
+
"step": 476
|
| 3342 |
+
},
|
| 3343 |
+
{
|
| 3344 |
+
"epoch": 0.5234567901234568,
|
| 3345 |
+
"grad_norm": 0.22967612743377686,
|
| 3346 |
+
"learning_rate": 1.0702836004932185e-05,
|
| 3347 |
+
"loss": 1.2839,
|
| 3348 |
+
"step": 477
|
| 3349 |
+
},
|
| 3350 |
+
{
|
| 3351 |
+
"epoch": 0.5245541838134431,
|
| 3352 |
+
"grad_norm": 0.24124501645565033,
|
| 3353 |
+
"learning_rate": 1.0678175092478422e-05,
|
| 3354 |
+
"loss": 1.2594,
|
| 3355 |
+
"step": 478
|
| 3356 |
+
},
|
| 3357 |
+
{
|
| 3358 |
+
"epoch": 0.5256515775034294,
|
| 3359 |
+
"grad_norm": 0.21366974711418152,
|
| 3360 |
+
"learning_rate": 1.0653514180024663e-05,
|
| 3361 |
+
"loss": 1.2265,
|
| 3362 |
+
"step": 479
|
| 3363 |
+
},
|
| 3364 |
+
{
|
| 3365 |
+
"epoch": 0.5267489711934157,
|
| 3366 |
+
"grad_norm": 0.26275762915611267,
|
| 3367 |
+
"learning_rate": 1.06288532675709e-05,
|
| 3368 |
+
"loss": 1.1795,
|
| 3369 |
+
"step": 480
|
| 3370 |
+
},
|
| 3371 |
+
{
|
| 3372 |
+
"epoch": 0.527846364883402,
|
| 3373 |
+
"grad_norm": 0.2695659101009369,
|
| 3374 |
+
"learning_rate": 1.060419235511714e-05,
|
| 3375 |
+
"loss": 1.1478,
|
| 3376 |
+
"step": 481
|
| 3377 |
+
},
|
| 3378 |
+
{
|
| 3379 |
+
"epoch": 0.5289437585733882,
|
| 3380 |
+
"grad_norm": 0.25220680236816406,
|
| 3381 |
+
"learning_rate": 1.057953144266338e-05,
|
| 3382 |
+
"loss": 1.2368,
|
| 3383 |
+
"step": 482
|
| 3384 |
+
},
|
| 3385 |
+
{
|
| 3386 |
+
"epoch": 0.5300411522633744,
|
| 3387 |
+
"grad_norm": 0.26438310742378235,
|
| 3388 |
+
"learning_rate": 1.0554870530209617e-05,
|
| 3389 |
+
"loss": 1.2154,
|
| 3390 |
+
"step": 483
|
| 3391 |
+
},
|
| 3392 |
+
{
|
| 3393 |
+
"epoch": 0.5311385459533607,
|
| 3394 |
+
"grad_norm": 0.2283749133348465,
|
| 3395 |
+
"learning_rate": 1.0530209617755858e-05,
|
| 3396 |
+
"loss": 1.2494,
|
| 3397 |
+
"step": 484
|
| 3398 |
+
},
|
| 3399 |
+
{
|
| 3400 |
+
"epoch": 0.532235939643347,
|
| 3401 |
+
"grad_norm": 0.2409619837999344,
|
| 3402 |
+
"learning_rate": 1.0505548705302098e-05,
|
| 3403 |
+
"loss": 1.2972,
|
| 3404 |
+
"step": 485
|
| 3405 |
+
},
|
| 3406 |
+
{
|
| 3407 |
+
"epoch": 0.5333333333333333,
|
| 3408 |
+
"grad_norm": 0.26836997270584106,
|
| 3409 |
+
"learning_rate": 1.0480887792848335e-05,
|
| 3410 |
+
"loss": 1.2379,
|
| 3411 |
+
"step": 486
|
| 3412 |
+
},
|
| 3413 |
+
{
|
| 3414 |
+
"epoch": 0.5344307270233196,
|
| 3415 |
+
"grad_norm": 0.2486780285835266,
|
| 3416 |
+
"learning_rate": 1.0456226880394576e-05,
|
| 3417 |
+
"loss": 1.2301,
|
| 3418 |
+
"step": 487
|
| 3419 |
+
},
|
| 3420 |
+
{
|
| 3421 |
+
"epoch": 0.5355281207133059,
|
| 3422 |
+
"grad_norm": 0.22093676030635834,
|
| 3423 |
+
"learning_rate": 1.0431565967940815e-05,
|
| 3424 |
+
"loss": 1.3068,
|
| 3425 |
+
"step": 488
|
| 3426 |
+
},
|
| 3427 |
+
{
|
| 3428 |
+
"epoch": 0.5366255144032922,
|
| 3429 |
+
"grad_norm": 0.2776540219783783,
|
| 3430 |
+
"learning_rate": 1.0406905055487054e-05,
|
| 3431 |
+
"loss": 1.1694,
|
| 3432 |
+
"step": 489
|
| 3433 |
+
},
|
| 3434 |
+
{
|
| 3435 |
+
"epoch": 0.5377229080932785,
|
| 3436 |
+
"grad_norm": 0.2561147212982178,
|
| 3437 |
+
"learning_rate": 1.0382244143033293e-05,
|
| 3438 |
+
"loss": 1.2451,
|
| 3439 |
+
"step": 490
|
| 3440 |
+
},
|
| 3441 |
+
{
|
| 3442 |
+
"epoch": 0.5388203017832648,
|
| 3443 |
+
"grad_norm": 0.2775615453720093,
|
| 3444 |
+
"learning_rate": 1.0357583230579533e-05,
|
| 3445 |
+
"loss": 1.1917,
|
| 3446 |
+
"step": 491
|
| 3447 |
+
},
|
| 3448 |
+
{
|
| 3449 |
+
"epoch": 0.5399176954732511,
|
| 3450 |
+
"grad_norm": 0.23608152568340302,
|
| 3451 |
+
"learning_rate": 1.033292231812577e-05,
|
| 3452 |
+
"loss": 1.2625,
|
| 3453 |
+
"step": 492
|
| 3454 |
+
},
|
| 3455 |
+
{
|
| 3456 |
+
"epoch": 0.5410150891632373,
|
| 3457 |
+
"grad_norm": 0.23603923618793488,
|
| 3458 |
+
"learning_rate": 1.0308261405672011e-05,
|
| 3459 |
+
"loss": 1.3082,
|
| 3460 |
+
"step": 493
|
| 3461 |
+
},
|
| 3462 |
+
{
|
| 3463 |
+
"epoch": 0.5421124828532236,
|
| 3464 |
+
"grad_norm": 0.24645058810710907,
|
| 3465 |
+
"learning_rate": 1.0283600493218252e-05,
|
| 3466 |
+
"loss": 1.2167,
|
| 3467 |
+
"step": 494
|
| 3468 |
+
},
|
| 3469 |
+
{
|
| 3470 |
+
"epoch": 0.5432098765432098,
|
| 3471 |
+
"grad_norm": 0.2328295260667801,
|
| 3472 |
+
"learning_rate": 1.0258939580764489e-05,
|
| 3473 |
+
"loss": 1.2517,
|
| 3474 |
+
"step": 495
|
| 3475 |
+
},
|
| 3476 |
+
{
|
| 3477 |
+
"epoch": 0.5443072702331961,
|
| 3478 |
+
"grad_norm": 0.25134629011154175,
|
| 3479 |
+
"learning_rate": 1.023427866831073e-05,
|
| 3480 |
+
"loss": 1.2552,
|
| 3481 |
+
"step": 496
|
| 3482 |
+
},
|
| 3483 |
+
{
|
| 3484 |
+
"epoch": 0.5454046639231824,
|
| 3485 |
+
"grad_norm": 0.23348143696784973,
|
| 3486 |
+
"learning_rate": 1.0209617755856967e-05,
|
| 3487 |
+
"loss": 1.1874,
|
| 3488 |
+
"step": 497
|
| 3489 |
+
},
|
| 3490 |
+
{
|
| 3491 |
+
"epoch": 0.5465020576131687,
|
| 3492 |
+
"grad_norm": 0.22104202210903168,
|
| 3493 |
+
"learning_rate": 1.0184956843403206e-05,
|
| 3494 |
+
"loss": 1.3255,
|
| 3495 |
+
"step": 498
|
| 3496 |
+
},
|
| 3497 |
+
{
|
| 3498 |
+
"epoch": 0.547599451303155,
|
| 3499 |
+
"grad_norm": 0.3350814878940582,
|
| 3500 |
+
"learning_rate": 1.0160295930949446e-05,
|
| 3501 |
+
"loss": 1.0423,
|
| 3502 |
+
"step": 499
|
| 3503 |
+
},
|
| 3504 |
+
{
|
| 3505 |
+
"epoch": 0.5486968449931413,
|
| 3506 |
+
"grad_norm": 0.26277315616607666,
|
| 3507 |
+
"learning_rate": 1.0135635018495683e-05,
|
| 3508 |
+
"loss": 1.1521,
|
| 3509 |
+
"step": 500
|
| 3510 |
+
},
|
| 3511 |
+
{
|
| 3512 |
+
"epoch": 0.5497942386831276,
|
| 3513 |
+
"grad_norm": 0.2433277815580368,
|
| 3514 |
+
"learning_rate": 1.0110974106041924e-05,
|
| 3515 |
+
"loss": 1.2761,
|
| 3516 |
+
"step": 501
|
| 3517 |
+
},
|
| 3518 |
+
{
|
| 3519 |
+
"epoch": 0.5508916323731139,
|
| 3520 |
+
"grad_norm": 0.21314479410648346,
|
| 3521 |
+
"learning_rate": 1.0086313193588165e-05,
|
| 3522 |
+
"loss": 1.2759,
|
| 3523 |
+
"step": 502
|
| 3524 |
+
},
|
| 3525 |
+
{
|
| 3526 |
+
"epoch": 0.5519890260631002,
|
| 3527 |
+
"grad_norm": 0.2336982637643814,
|
| 3528 |
+
"learning_rate": 1.0061652281134402e-05,
|
| 3529 |
+
"loss": 1.2843,
|
| 3530 |
+
"step": 503
|
| 3531 |
+
},
|
| 3532 |
+
{
|
| 3533 |
+
"epoch": 0.5530864197530864,
|
| 3534 |
+
"grad_norm": 0.24953609704971313,
|
| 3535 |
+
"learning_rate": 1.0036991368680642e-05,
|
| 3536 |
+
"loss": 1.2541,
|
| 3537 |
+
"step": 504
|
| 3538 |
+
},
|
| 3539 |
+
{
|
| 3540 |
+
"epoch": 0.5541838134430727,
|
| 3541 |
+
"grad_norm": 0.233303040266037,
|
| 3542 |
+
"learning_rate": 1.0012330456226881e-05,
|
| 3543 |
+
"loss": 1.2984,
|
| 3544 |
+
"step": 505
|
| 3545 |
+
},
|
| 3546 |
+
{
|
| 3547 |
+
"epoch": 0.555281207133059,
|
| 3548 |
+
"grad_norm": 0.25009945034980774,
|
| 3549 |
+
"learning_rate": 9.98766954377312e-06,
|
| 3550 |
+
"loss": 1.1962,
|
| 3551 |
+
"step": 506
|
| 3552 |
+
},
|
| 3553 |
+
{
|
| 3554 |
+
"epoch": 0.5563786008230452,
|
| 3555 |
+
"grad_norm": 0.277473121881485,
|
| 3556 |
+
"learning_rate": 9.96300863131936e-06,
|
| 3557 |
+
"loss": 1.2319,
|
| 3558 |
+
"step": 507
|
| 3559 |
+
},
|
| 3560 |
+
{
|
| 3561 |
+
"epoch": 0.5574759945130315,
|
| 3562 |
+
"grad_norm": 0.2756619155406952,
|
| 3563 |
+
"learning_rate": 9.938347718865598e-06,
|
| 3564 |
+
"loss": 1.2274,
|
| 3565 |
+
"step": 508
|
| 3566 |
+
},
|
| 3567 |
+
{
|
| 3568 |
+
"epoch": 0.5585733882030178,
|
| 3569 |
+
"grad_norm": 0.2504933178424835,
|
| 3570 |
+
"learning_rate": 9.913686806411839e-06,
|
| 3571 |
+
"loss": 1.2285,
|
| 3572 |
+
"step": 509
|
| 3573 |
+
},
|
| 3574 |
+
{
|
| 3575 |
+
"epoch": 0.5596707818930041,
|
| 3576 |
+
"grad_norm": 0.2296246439218521,
|
| 3577 |
+
"learning_rate": 9.889025893958078e-06,
|
| 3578 |
+
"loss": 1.2451,
|
| 3579 |
+
"step": 510
|
| 3580 |
+
},
|
| 3581 |
+
{
|
| 3582 |
+
"epoch": 0.5607681755829904,
|
| 3583 |
+
"grad_norm": 0.2273918092250824,
|
| 3584 |
+
"learning_rate": 9.864364981504317e-06,
|
| 3585 |
+
"loss": 1.3187,
|
| 3586 |
+
"step": 511
|
| 3587 |
+
},
|
| 3588 |
+
{
|
| 3589 |
+
"epoch": 0.5618655692729767,
|
| 3590 |
+
"grad_norm": 0.25282397866249084,
|
| 3591 |
+
"learning_rate": 9.839704069050555e-06,
|
| 3592 |
+
"loss": 1.2307,
|
| 3593 |
+
"step": 512
|
| 3594 |
+
},
|
| 3595 |
+
{
|
| 3596 |
+
"epoch": 0.562962962962963,
|
| 3597 |
+
"grad_norm": 0.2543501555919647,
|
| 3598 |
+
"learning_rate": 9.815043156596796e-06,
|
| 3599 |
+
"loss": 1.1582,
|
| 3600 |
+
"step": 513
|
| 3601 |
+
},
|
| 3602 |
+
{
|
| 3603 |
+
"epoch": 0.5640603566529493,
|
| 3604 |
+
"grad_norm": 0.2439233809709549,
|
| 3605 |
+
"learning_rate": 9.790382244143033e-06,
|
| 3606 |
+
"loss": 1.2101,
|
| 3607 |
+
"step": 514
|
| 3608 |
+
},
|
| 3609 |
+
{
|
| 3610 |
+
"epoch": 0.5651577503429356,
|
| 3611 |
+
"grad_norm": 0.2775726318359375,
|
| 3612 |
+
"learning_rate": 9.765721331689272e-06,
|
| 3613 |
+
"loss": 1.1793,
|
| 3614 |
+
"step": 515
|
| 3615 |
+
},
|
| 3616 |
+
{
|
| 3617 |
+
"epoch": 0.5662551440329218,
|
| 3618 |
+
"grad_norm": 0.24230070412158966,
|
| 3619 |
+
"learning_rate": 9.741060419235513e-06,
|
| 3620 |
+
"loss": 1.2533,
|
| 3621 |
+
"step": 516
|
| 3622 |
+
},
|
| 3623 |
+
{
|
| 3624 |
+
"epoch": 0.5673525377229081,
|
| 3625 |
+
"grad_norm": 0.2427159547805786,
|
| 3626 |
+
"learning_rate": 9.716399506781752e-06,
|
| 3627 |
+
"loss": 1.2027,
|
| 3628 |
+
"step": 517
|
| 3629 |
+
},
|
| 3630 |
+
{
|
| 3631 |
+
"epoch": 0.5684499314128943,
|
| 3632 |
+
"grad_norm": 0.23517057299613953,
|
| 3633 |
+
"learning_rate": 9.69173859432799e-06,
|
| 3634 |
+
"loss": 1.2124,
|
| 3635 |
+
"step": 518
|
| 3636 |
+
},
|
| 3637 |
+
{
|
| 3638 |
+
"epoch": 0.5695473251028806,
|
| 3639 |
+
"grad_norm": 0.2562308609485626,
|
| 3640 |
+
"learning_rate": 9.667077681874231e-06,
|
| 3641 |
+
"loss": 1.2731,
|
| 3642 |
+
"step": 519
|
| 3643 |
+
},
|
| 3644 |
+
{
|
| 3645 |
+
"epoch": 0.5706447187928669,
|
| 3646 |
+
"grad_norm": 0.22711578011512756,
|
| 3647 |
+
"learning_rate": 9.64241676942047e-06,
|
| 3648 |
+
"loss": 1.2964,
|
| 3649 |
+
"step": 520
|
| 3650 |
+
},
|
| 3651 |
+
{
|
| 3652 |
+
"epoch": 0.5717421124828532,
|
| 3653 |
+
"grad_norm": 0.24960756301879883,
|
| 3654 |
+
"learning_rate": 9.617755856966709e-06,
|
| 3655 |
+
"loss": 1.2228,
|
| 3656 |
+
"step": 521
|
| 3657 |
+
},
|
| 3658 |
+
{
|
| 3659 |
+
"epoch": 0.5728395061728395,
|
| 3660 |
+
"grad_norm": 0.23647604882717133,
|
| 3661 |
+
"learning_rate": 9.593094944512948e-06,
|
| 3662 |
+
"loss": 1.2372,
|
| 3663 |
+
"step": 522
|
| 3664 |
+
},
|
| 3665 |
+
{
|
| 3666 |
+
"epoch": 0.5739368998628258,
|
| 3667 |
+
"grad_norm": 0.2726448178291321,
|
| 3668 |
+
"learning_rate": 9.568434032059187e-06,
|
| 3669 |
+
"loss": 1.1503,
|
| 3670 |
+
"step": 523
|
| 3671 |
+
},
|
| 3672 |
+
{
|
| 3673 |
+
"epoch": 0.5750342935528121,
|
| 3674 |
+
"grad_norm": 0.23898166418075562,
|
| 3675 |
+
"learning_rate": 9.543773119605426e-06,
|
| 3676 |
+
"loss": 1.2561,
|
| 3677 |
+
"step": 524
|
| 3678 |
+
},
|
| 3679 |
+
{
|
| 3680 |
+
"epoch": 0.5761316872427984,
|
| 3681 |
+
"grad_norm": 0.2247282862663269,
|
| 3682 |
+
"learning_rate": 9.519112207151665e-06,
|
| 3683 |
+
"loss": 1.3139,
|
| 3684 |
+
"step": 525
|
| 3685 |
+
},
|
| 3686 |
+
{
|
| 3687 |
+
"epoch": 0.5772290809327847,
|
| 3688 |
+
"grad_norm": 0.2438066303730011,
|
| 3689 |
+
"learning_rate": 9.494451294697905e-06,
|
| 3690 |
+
"loss": 1.2585,
|
| 3691 |
+
"step": 526
|
| 3692 |
+
},
|
| 3693 |
+
{
|
| 3694 |
+
"epoch": 0.5783264746227709,
|
| 3695 |
+
"grad_norm": 0.26008307933807373,
|
| 3696 |
+
"learning_rate": 9.469790382244144e-06,
|
| 3697 |
+
"loss": 1.2154,
|
| 3698 |
+
"step": 527
|
| 3699 |
+
},
|
| 3700 |
+
{
|
| 3701 |
+
"epoch": 0.5794238683127572,
|
| 3702 |
+
"grad_norm": 0.24809014797210693,
|
| 3703 |
+
"learning_rate": 9.445129469790383e-06,
|
| 3704 |
+
"loss": 1.2685,
|
| 3705 |
+
"step": 528
|
| 3706 |
+
},
|
| 3707 |
+
{
|
| 3708 |
+
"epoch": 0.5805212620027435,
|
| 3709 |
+
"grad_norm": 0.2819349765777588,
|
| 3710 |
+
"learning_rate": 9.420468557336622e-06,
|
| 3711 |
+
"loss": 1.2329,
|
| 3712 |
+
"step": 529
|
| 3713 |
+
},
|
| 3714 |
+
{
|
| 3715 |
+
"epoch": 0.5816186556927297,
|
| 3716 |
+
"grad_norm": 0.2312726080417633,
|
| 3717 |
+
"learning_rate": 9.39580764488286e-06,
|
| 3718 |
+
"loss": 1.2034,
|
| 3719 |
+
"step": 530
|
| 3720 |
+
},
|
| 3721 |
+
{
|
| 3722 |
+
"epoch": 0.582716049382716,
|
| 3723 |
+
"grad_norm": 0.2416796237230301,
|
| 3724 |
+
"learning_rate": 9.3711467324291e-06,
|
| 3725 |
+
"loss": 1.2599,
|
| 3726 |
+
"step": 531
|
| 3727 |
+
},
|
| 3728 |
+
{
|
| 3729 |
+
"epoch": 0.5838134430727023,
|
| 3730 |
+
"grad_norm": 0.2733534276485443,
|
| 3731 |
+
"learning_rate": 9.346485819975339e-06,
|
| 3732 |
+
"loss": 1.1968,
|
| 3733 |
+
"step": 532
|
| 3734 |
+
},
|
| 3735 |
+
{
|
| 3736 |
+
"epoch": 0.5849108367626886,
|
| 3737 |
+
"grad_norm": 0.27570605278015137,
|
| 3738 |
+
"learning_rate": 9.32182490752158e-06,
|
| 3739 |
+
"loss": 1.2419,
|
| 3740 |
+
"step": 533
|
| 3741 |
+
},
|
| 3742 |
+
{
|
| 3743 |
+
"epoch": 0.5860082304526749,
|
| 3744 |
+
"grad_norm": 0.25366488099098206,
|
| 3745 |
+
"learning_rate": 9.297163995067818e-06,
|
| 3746 |
+
"loss": 1.2086,
|
| 3747 |
+
"step": 534
|
| 3748 |
+
},
|
| 3749 |
+
{
|
| 3750 |
+
"epoch": 0.5871056241426612,
|
| 3751 |
+
"grad_norm": 0.290729820728302,
|
| 3752 |
+
"learning_rate": 9.272503082614057e-06,
|
| 3753 |
+
"loss": 1.2264,
|
| 3754 |
+
"step": 535
|
| 3755 |
+
},
|
| 3756 |
+
{
|
| 3757 |
+
"epoch": 0.5882030178326475,
|
| 3758 |
+
"grad_norm": 0.2413792908191681,
|
| 3759 |
+
"learning_rate": 9.247842170160298e-06,
|
| 3760 |
+
"loss": 1.2113,
|
| 3761 |
+
"step": 536
|
| 3762 |
+
},
|
| 3763 |
+
{
|
| 3764 |
+
"epoch": 0.5893004115226338,
|
| 3765 |
+
"grad_norm": 0.2590895891189575,
|
| 3766 |
+
"learning_rate": 9.223181257706537e-06,
|
| 3767 |
+
"loss": 1.2003,
|
| 3768 |
+
"step": 537
|
| 3769 |
+
},
|
| 3770 |
+
{
|
| 3771 |
+
"epoch": 0.59039780521262,
|
| 3772 |
+
"grad_norm": 0.27695104479789734,
|
| 3773 |
+
"learning_rate": 9.198520345252775e-06,
|
| 3774 |
+
"loss": 1.1252,
|
| 3775 |
+
"step": 538
|
| 3776 |
+
},
|
| 3777 |
+
{
|
| 3778 |
+
"epoch": 0.5914951989026063,
|
| 3779 |
+
"grad_norm": 0.29316338896751404,
|
| 3780 |
+
"learning_rate": 9.173859432799014e-06,
|
| 3781 |
+
"loss": 1.1854,
|
| 3782 |
+
"step": 539
|
| 3783 |
+
},
|
| 3784 |
+
{
|
| 3785 |
+
"epoch": 0.5925925925925926,
|
| 3786 |
+
"grad_norm": 0.24757929146289825,
|
| 3787 |
+
"learning_rate": 9.149198520345253e-06,
|
| 3788 |
+
"loss": 1.2679,
|
| 3789 |
+
"step": 540
|
| 3790 |
+
},
|
| 3791 |
+
{
|
| 3792 |
+
"epoch": 0.5936899862825789,
|
| 3793 |
+
"grad_norm": 0.2792201042175293,
|
| 3794 |
+
"learning_rate": 9.124537607891492e-06,
|
| 3795 |
+
"loss": 1.2385,
|
| 3796 |
+
"step": 541
|
| 3797 |
+
},
|
| 3798 |
+
{
|
| 3799 |
+
"epoch": 0.5947873799725651,
|
| 3800 |
+
"grad_norm": 0.250360906124115,
|
| 3801 |
+
"learning_rate": 9.099876695437731e-06,
|
| 3802 |
+
"loss": 1.2251,
|
| 3803 |
+
"step": 542
|
| 3804 |
+
},
|
| 3805 |
+
{
|
| 3806 |
+
"epoch": 0.5958847736625514,
|
| 3807 |
+
"grad_norm": 0.24334873259067535,
|
| 3808 |
+
"learning_rate": 9.075215782983972e-06,
|
| 3809 |
+
"loss": 1.2531,
|
| 3810 |
+
"step": 543
|
| 3811 |
+
},
|
| 3812 |
+
{
|
| 3813 |
+
"epoch": 0.5969821673525377,
|
| 3814 |
+
"grad_norm": 0.25853219628334045,
|
| 3815 |
+
"learning_rate": 9.05055487053021e-06,
|
| 3816 |
+
"loss": 1.2103,
|
| 3817 |
+
"step": 544
|
| 3818 |
+
},
|
| 3819 |
+
{
|
| 3820 |
+
"epoch": 0.598079561042524,
|
| 3821 |
+
"grad_norm": 0.24322326481342316,
|
| 3822 |
+
"learning_rate": 9.02589395807645e-06,
|
| 3823 |
+
"loss": 1.2026,
|
| 3824 |
+
"step": 545
|
| 3825 |
+
},
|
| 3826 |
+
{
|
| 3827 |
+
"epoch": 0.5991769547325103,
|
| 3828 |
+
"grad_norm": 0.2686692774295807,
|
| 3829 |
+
"learning_rate": 9.001233045622688e-06,
|
| 3830 |
+
"loss": 1.2199,
|
| 3831 |
+
"step": 546
|
| 3832 |
+
},
|
| 3833 |
+
{
|
| 3834 |
+
"epoch": 0.6002743484224966,
|
| 3835 |
+
"grad_norm": 0.23173104226589203,
|
| 3836 |
+
"learning_rate": 8.976572133168927e-06,
|
| 3837 |
+
"loss": 1.2873,
|
| 3838 |
+
"step": 547
|
| 3839 |
+
},
|
| 3840 |
+
{
|
| 3841 |
+
"epoch": 0.6013717421124829,
|
| 3842 |
+
"grad_norm": 0.3123144507408142,
|
| 3843 |
+
"learning_rate": 8.951911220715166e-06,
|
| 3844 |
+
"loss": 1.2072,
|
| 3845 |
+
"step": 548
|
| 3846 |
+
},
|
| 3847 |
+
{
|
| 3848 |
+
"epoch": 0.6024691358024692,
|
| 3849 |
+
"grad_norm": 0.2637168765068054,
|
| 3850 |
+
"learning_rate": 8.927250308261405e-06,
|
| 3851 |
+
"loss": 1.3029,
|
| 3852 |
+
"step": 549
|
| 3853 |
+
},
|
| 3854 |
+
{
|
| 3855 |
+
"epoch": 0.6035665294924554,
|
| 3856 |
+
"grad_norm": 0.2372378557920456,
|
| 3857 |
+
"learning_rate": 8.902589395807646e-06,
|
| 3858 |
+
"loss": 1.2345,
|
| 3859 |
+
"step": 550
|
| 3860 |
+
},
|
| 3861 |
+
{
|
| 3862 |
+
"epoch": 0.6046639231824417,
|
| 3863 |
+
"grad_norm": 0.2722484767436981,
|
| 3864 |
+
"learning_rate": 8.877928483353885e-06,
|
| 3865 |
+
"loss": 1.1733,
|
| 3866 |
+
"step": 551
|
| 3867 |
+
},
|
| 3868 |
+
{
|
| 3869 |
+
"epoch": 0.605761316872428,
|
| 3870 |
+
"grad_norm": 0.25364038348197937,
|
| 3871 |
+
"learning_rate": 8.853267570900124e-06,
|
| 3872 |
+
"loss": 1.2077,
|
| 3873 |
+
"step": 552
|
| 3874 |
+
},
|
| 3875 |
+
{
|
| 3876 |
+
"epoch": 0.6068587105624142,
|
| 3877 |
+
"grad_norm": 0.23805555701255798,
|
| 3878 |
+
"learning_rate": 8.828606658446364e-06,
|
| 3879 |
+
"loss": 1.311,
|
| 3880 |
+
"step": 553
|
| 3881 |
+
},
|
| 3882 |
+
{
|
| 3883 |
+
"epoch": 0.6079561042524005,
|
| 3884 |
+
"grad_norm": 0.22436513006687164,
|
| 3885 |
+
"learning_rate": 8.803945745992603e-06,
|
| 3886 |
+
"loss": 1.2665,
|
| 3887 |
+
"step": 554
|
| 3888 |
+
},
|
| 3889 |
+
{
|
| 3890 |
+
"epoch": 0.6090534979423868,
|
| 3891 |
+
"grad_norm": 0.251506507396698,
|
| 3892 |
+
"learning_rate": 8.779284833538842e-06,
|
| 3893 |
+
"loss": 1.2218,
|
| 3894 |
+
"step": 555
|
| 3895 |
+
},
|
| 3896 |
+
{
|
| 3897 |
+
"epoch": 0.6101508916323731,
|
| 3898 |
+
"grad_norm": 0.2255803495645523,
|
| 3899 |
+
"learning_rate": 8.754623921085081e-06,
|
| 3900 |
+
"loss": 1.262,
|
| 3901 |
+
"step": 556
|
| 3902 |
+
},
|
| 3903 |
+
{
|
| 3904 |
+
"epoch": 0.6112482853223594,
|
| 3905 |
+
"grad_norm": 0.2577172517776489,
|
| 3906 |
+
"learning_rate": 8.72996300863132e-06,
|
| 3907 |
+
"loss": 1.2658,
|
| 3908 |
+
"step": 557
|
| 3909 |
+
},
|
| 3910 |
+
{
|
| 3911 |
+
"epoch": 0.6123456790123457,
|
| 3912 |
+
"grad_norm": 0.234437495470047,
|
| 3913 |
+
"learning_rate": 8.705302096177559e-06,
|
| 3914 |
+
"loss": 1.2065,
|
| 3915 |
+
"step": 558
|
| 3916 |
+
},
|
| 3917 |
+
{
|
| 3918 |
+
"epoch": 0.613443072702332,
|
| 3919 |
+
"grad_norm": 0.2587389647960663,
|
| 3920 |
+
"learning_rate": 8.680641183723798e-06,
|
| 3921 |
+
"loss": 1.232,
|
| 3922 |
+
"step": 559
|
| 3923 |
+
},
|
| 3924 |
+
{
|
| 3925 |
+
"epoch": 0.6145404663923183,
|
| 3926 |
+
"grad_norm": 0.2382604330778122,
|
| 3927 |
+
"learning_rate": 8.655980271270038e-06,
|
| 3928 |
+
"loss": 1.2491,
|
| 3929 |
+
"step": 560
|
| 3930 |
+
},
|
| 3931 |
+
{
|
| 3932 |
+
"epoch": 0.6156378600823045,
|
| 3933 |
+
"grad_norm": 0.2326079159975052,
|
| 3934 |
+
"learning_rate": 8.631319358816277e-06,
|
| 3935 |
+
"loss": 1.2397,
|
| 3936 |
+
"step": 561
|
| 3937 |
+
},
|
| 3938 |
+
{
|
| 3939 |
+
"epoch": 0.6167352537722908,
|
| 3940 |
+
"grad_norm": 0.24208717048168182,
|
| 3941 |
+
"learning_rate": 8.606658446362516e-06,
|
| 3942 |
+
"loss": 1.2494,
|
| 3943 |
+
"step": 562
|
| 3944 |
+
},
|
| 3945 |
+
{
|
| 3946 |
+
"epoch": 0.6178326474622771,
|
| 3947 |
+
"grad_norm": 0.2493620663881302,
|
| 3948 |
+
"learning_rate": 8.581997533908755e-06,
|
| 3949 |
+
"loss": 1.1917,
|
| 3950 |
+
"step": 563
|
| 3951 |
+
},
|
| 3952 |
+
{
|
| 3953 |
+
"epoch": 0.6189300411522634,
|
| 3954 |
+
"grad_norm": 0.24771496653556824,
|
| 3955 |
+
"learning_rate": 8.557336621454994e-06,
|
| 3956 |
+
"loss": 1.2364,
|
| 3957 |
+
"step": 564
|
| 3958 |
+
},
|
| 3959 |
+
{
|
| 3960 |
+
"epoch": 0.6200274348422496,
|
| 3961 |
+
"grad_norm": 0.2429998517036438,
|
| 3962 |
+
"learning_rate": 8.532675709001233e-06,
|
| 3963 |
+
"loss": 1.2242,
|
| 3964 |
+
"step": 565
|
| 3965 |
+
},
|
| 3966 |
+
{
|
| 3967 |
+
"epoch": 0.6211248285322359,
|
| 3968 |
+
"grad_norm": 0.25622135400772095,
|
| 3969 |
+
"learning_rate": 8.508014796547472e-06,
|
| 3970 |
+
"loss": 1.1917,
|
| 3971 |
+
"step": 566
|
| 3972 |
+
},
|
| 3973 |
+
{
|
| 3974 |
+
"epoch": 0.6222222222222222,
|
| 3975 |
+
"grad_norm": 0.23954106867313385,
|
| 3976 |
+
"learning_rate": 8.483353884093712e-06,
|
| 3977 |
+
"loss": 1.3033,
|
| 3978 |
+
"step": 567
|
| 3979 |
+
},
|
| 3980 |
+
{
|
| 3981 |
+
"epoch": 0.6233196159122085,
|
| 3982 |
+
"grad_norm": 0.2756274938583374,
|
| 3983 |
+
"learning_rate": 8.458692971639951e-06,
|
| 3984 |
+
"loss": 1.2447,
|
| 3985 |
+
"step": 568
|
| 3986 |
+
},
|
| 3987 |
+
{
|
| 3988 |
+
"epoch": 0.6244170096021948,
|
| 3989 |
+
"grad_norm": 0.26642876863479614,
|
| 3990 |
+
"learning_rate": 8.43403205918619e-06,
|
| 3991 |
+
"loss": 1.2941,
|
| 3992 |
+
"step": 569
|
| 3993 |
+
},
|
| 3994 |
+
{
|
| 3995 |
+
"epoch": 0.6255144032921811,
|
| 3996 |
+
"grad_norm": 0.2476617991924286,
|
| 3997 |
+
"learning_rate": 8.40937114673243e-06,
|
| 3998 |
+
"loss": 1.2448,
|
| 3999 |
+
"step": 570
|
| 4000 |
+
},
|
| 4001 |
+
{
|
| 4002 |
+
"epoch": 0.6266117969821674,
|
| 4003 |
+
"grad_norm": 0.2281130701303482,
|
| 4004 |
+
"learning_rate": 8.38471023427867e-06,
|
| 4005 |
+
"loss": 1.3206,
|
| 4006 |
+
"step": 571
|
| 4007 |
+
},
|
| 4008 |
+
{
|
| 4009 |
+
"epoch": 0.6277091906721536,
|
| 4010 |
+
"grad_norm": 0.2541149854660034,
|
| 4011 |
+
"learning_rate": 8.360049321824909e-06,
|
| 4012 |
+
"loss": 1.1509,
|
| 4013 |
+
"step": 572
|
| 4014 |
+
},
|
| 4015 |
+
{
|
| 4016 |
+
"epoch": 0.6288065843621399,
|
| 4017 |
+
"grad_norm": 0.23095186054706573,
|
| 4018 |
+
"learning_rate": 8.335388409371147e-06,
|
| 4019 |
+
"loss": 1.2499,
|
| 4020 |
+
"step": 573
|
| 4021 |
+
},
|
| 4022 |
+
{
|
| 4023 |
+
"epoch": 0.6299039780521262,
|
| 4024 |
+
"grad_norm": 0.25589171051979065,
|
| 4025 |
+
"learning_rate": 8.310727496917386e-06,
|
| 4026 |
+
"loss": 1.2262,
|
| 4027 |
+
"step": 574
|
| 4028 |
+
},
|
| 4029 |
+
{
|
| 4030 |
+
"epoch": 0.6310013717421125,
|
| 4031 |
+
"grad_norm": 0.2345532923936844,
|
| 4032 |
+
"learning_rate": 8.286066584463625e-06,
|
| 4033 |
+
"loss": 1.2438,
|
| 4034 |
+
"step": 575
|
| 4035 |
+
},
|
| 4036 |
+
{
|
| 4037 |
+
"epoch": 0.6320987654320988,
|
| 4038 |
+
"grad_norm": 0.24153324961662292,
|
| 4039 |
+
"learning_rate": 8.261405672009864e-06,
|
| 4040 |
+
"loss": 1.2189,
|
| 4041 |
+
"step": 576
|
| 4042 |
+
},
|
| 4043 |
+
{
|
| 4044 |
+
"epoch": 0.633196159122085,
|
| 4045 |
+
"grad_norm": 0.2964321970939636,
|
| 4046 |
+
"learning_rate": 8.236744759556105e-06,
|
| 4047 |
+
"loss": 1.242,
|
| 4048 |
+
"step": 577
|
| 4049 |
+
},
|
| 4050 |
+
{
|
| 4051 |
+
"epoch": 0.6342935528120713,
|
| 4052 |
+
"grad_norm": 0.26954197883605957,
|
| 4053 |
+
"learning_rate": 8.212083847102344e-06,
|
| 4054 |
+
"loss": 1.1554,
|
| 4055 |
+
"step": 578
|
| 4056 |
+
},
|
| 4057 |
+
{
|
| 4058 |
+
"epoch": 0.6353909465020576,
|
| 4059 |
+
"grad_norm": 0.2815469801425934,
|
| 4060 |
+
"learning_rate": 8.187422934648583e-06,
|
| 4061 |
+
"loss": 1.2922,
|
| 4062 |
+
"step": 579
|
| 4063 |
+
},
|
| 4064 |
+
{
|
| 4065 |
+
"epoch": 0.6364883401920439,
|
| 4066 |
+
"grad_norm": 0.2681712806224823,
|
| 4067 |
+
"learning_rate": 8.162762022194821e-06,
|
| 4068 |
+
"loss": 1.165,
|
| 4069 |
+
"step": 580
|
| 4070 |
+
},
|
| 4071 |
+
{
|
| 4072 |
+
"epoch": 0.6375857338820302,
|
| 4073 |
+
"grad_norm": 0.23276597261428833,
|
| 4074 |
+
"learning_rate": 8.13810110974106e-06,
|
| 4075 |
+
"loss": 1.2156,
|
| 4076 |
+
"step": 581
|
| 4077 |
+
},
|
| 4078 |
+
{
|
| 4079 |
+
"epoch": 0.6386831275720165,
|
| 4080 |
+
"grad_norm": 0.22940269112586975,
|
| 4081 |
+
"learning_rate": 8.1134401972873e-06,
|
| 4082 |
+
"loss": 1.2471,
|
| 4083 |
+
"step": 582
|
| 4084 |
+
},
|
| 4085 |
+
{
|
| 4086 |
+
"epoch": 0.6397805212620027,
|
| 4087 |
+
"grad_norm": 0.24708326160907745,
|
| 4088 |
+
"learning_rate": 8.08877928483354e-06,
|
| 4089 |
+
"loss": 1.2742,
|
| 4090 |
+
"step": 583
|
| 4091 |
+
},
|
| 4092 |
+
{
|
| 4093 |
+
"epoch": 0.640877914951989,
|
| 4094 |
+
"grad_norm": 0.24238231778144836,
|
| 4095 |
+
"learning_rate": 8.064118372379779e-06,
|
| 4096 |
+
"loss": 1.2485,
|
| 4097 |
+
"step": 584
|
| 4098 |
+
},
|
| 4099 |
+
{
|
| 4100 |
+
"epoch": 0.6419753086419753,
|
| 4101 |
+
"grad_norm": 0.25867024064064026,
|
| 4102 |
+
"learning_rate": 8.039457459926018e-06,
|
| 4103 |
+
"loss": 1.1782,
|
| 4104 |
+
"step": 585
|
| 4105 |
+
},
|
| 4106 |
+
{
|
| 4107 |
+
"epoch": 0.6430727023319616,
|
| 4108 |
+
"grad_norm": 0.24773919582366943,
|
| 4109 |
+
"learning_rate": 8.014796547472257e-06,
|
| 4110 |
+
"loss": 1.3088,
|
| 4111 |
+
"step": 586
|
| 4112 |
+
},
|
| 4113 |
+
{
|
| 4114 |
+
"epoch": 0.6441700960219479,
|
| 4115 |
+
"grad_norm": 0.24750138819217682,
|
| 4116 |
+
"learning_rate": 7.990135635018497e-06,
|
| 4117 |
+
"loss": 1.2712,
|
| 4118 |
+
"step": 587
|
| 4119 |
+
},
|
| 4120 |
+
{
|
| 4121 |
+
"epoch": 0.6452674897119342,
|
| 4122 |
+
"grad_norm": 0.25289419293403625,
|
| 4123 |
+
"learning_rate": 7.965474722564736e-06,
|
| 4124 |
+
"loss": 1.2251,
|
| 4125 |
+
"step": 588
|
| 4126 |
+
},
|
| 4127 |
+
{
|
| 4128 |
+
"epoch": 0.6463648834019204,
|
| 4129 |
+
"grad_norm": 0.2552140951156616,
|
| 4130 |
+
"learning_rate": 7.940813810110975e-06,
|
| 4131 |
+
"loss": 1.2337,
|
| 4132 |
+
"step": 589
|
| 4133 |
+
},
|
| 4134 |
+
{
|
| 4135 |
+
"epoch": 0.6474622770919067,
|
| 4136 |
+
"grad_norm": 0.2686692774295807,
|
| 4137 |
+
"learning_rate": 7.916152897657214e-06,
|
| 4138 |
+
"loss": 1.2644,
|
| 4139 |
+
"step": 590
|
| 4140 |
+
},
|
| 4141 |
+
{
|
| 4142 |
+
"epoch": 0.648559670781893,
|
| 4143 |
+
"grad_norm": 0.2389717698097229,
|
| 4144 |
+
"learning_rate": 7.891491985203453e-06,
|
| 4145 |
+
"loss": 1.2691,
|
| 4146 |
+
"step": 591
|
| 4147 |
+
},
|
| 4148 |
+
{
|
| 4149 |
+
"epoch": 0.6496570644718793,
|
| 4150 |
+
"grad_norm": 0.279037207365036,
|
| 4151 |
+
"learning_rate": 7.866831072749692e-06,
|
| 4152 |
+
"loss": 1.1433,
|
| 4153 |
+
"step": 592
|
| 4154 |
+
},
|
| 4155 |
+
{
|
| 4156 |
+
"epoch": 0.6507544581618656,
|
| 4157 |
+
"grad_norm": 0.23143412172794342,
|
| 4158 |
+
"learning_rate": 7.842170160295932e-06,
|
| 4159 |
+
"loss": 1.208,
|
| 4160 |
+
"step": 593
|
| 4161 |
+
},
|
| 4162 |
+
{
|
| 4163 |
+
"epoch": 0.6518518518518519,
|
| 4164 |
+
"grad_norm": 0.24017734825611115,
|
| 4165 |
+
"learning_rate": 7.817509247842171e-06,
|
| 4166 |
+
"loss": 1.2327,
|
| 4167 |
+
"step": 594
|
| 4168 |
+
},
|
| 4169 |
+
{
|
| 4170 |
+
"epoch": 0.6529492455418381,
|
| 4171 |
+
"grad_norm": 0.21986091136932373,
|
| 4172 |
+
"learning_rate": 7.79284833538841e-06,
|
| 4173 |
+
"loss": 1.2135,
|
| 4174 |
+
"step": 595
|
| 4175 |
+
},
|
| 4176 |
+
{
|
| 4177 |
+
"epoch": 0.6540466392318244,
|
| 4178 |
+
"grad_norm": 0.2538710832595825,
|
| 4179 |
+
"learning_rate": 7.768187422934649e-06,
|
| 4180 |
+
"loss": 1.2321,
|
| 4181 |
+
"step": 596
|
| 4182 |
+
},
|
| 4183 |
+
{
|
| 4184 |
+
"epoch": 0.6551440329218107,
|
| 4185 |
+
"grad_norm": 0.2757505178451538,
|
| 4186 |
+
"learning_rate": 7.743526510480888e-06,
|
| 4187 |
+
"loss": 1.2606,
|
| 4188 |
+
"step": 597
|
| 4189 |
+
},
|
| 4190 |
+
{
|
| 4191 |
+
"epoch": 0.656241426611797,
|
| 4192 |
+
"grad_norm": 0.25676995515823364,
|
| 4193 |
+
"learning_rate": 7.718865598027127e-06,
|
| 4194 |
+
"loss": 1.1645,
|
| 4195 |
+
"step": 598
|
| 4196 |
+
},
|
| 4197 |
+
{
|
| 4198 |
+
"epoch": 0.6573388203017833,
|
| 4199 |
+
"grad_norm": 0.21630260348320007,
|
| 4200 |
+
"learning_rate": 7.694204685573366e-06,
|
| 4201 |
+
"loss": 1.2465,
|
| 4202 |
+
"step": 599
|
| 4203 |
+
},
|
| 4204 |
+
{
|
| 4205 |
+
"epoch": 0.6584362139917695,
|
| 4206 |
+
"grad_norm": 0.25539395213127136,
|
| 4207 |
+
"learning_rate": 7.669543773119606e-06,
|
| 4208 |
+
"loss": 1.2697,
|
| 4209 |
+
"step": 600
|
| 4210 |
+
},
|
| 4211 |
+
{
|
| 4212 |
+
"epoch": 0.6595336076817558,
|
| 4213 |
+
"grad_norm": 0.252244770526886,
|
| 4214 |
+
"learning_rate": 7.644882860665845e-06,
|
| 4215 |
+
"loss": 1.247,
|
| 4216 |
+
"step": 601
|
| 4217 |
+
},
|
| 4218 |
+
{
|
| 4219 |
+
"epoch": 0.6606310013717421,
|
| 4220 |
+
"grad_norm": 0.23825043439865112,
|
| 4221 |
+
"learning_rate": 7.620221948212084e-06,
|
| 4222 |
+
"loss": 1.2331,
|
| 4223 |
+
"step": 602
|
| 4224 |
+
},
|
| 4225 |
+
{
|
| 4226 |
+
"epoch": 0.6617283950617284,
|
| 4227 |
+
"grad_norm": 0.25092631578445435,
|
| 4228 |
+
"learning_rate": 7.595561035758323e-06,
|
| 4229 |
+
"loss": 1.223,
|
| 4230 |
+
"step": 603
|
| 4231 |
+
},
|
| 4232 |
+
{
|
| 4233 |
+
"epoch": 0.6628257887517147,
|
| 4234 |
+
"grad_norm": 0.2373005449771881,
|
| 4235 |
+
"learning_rate": 7.570900123304563e-06,
|
| 4236 |
+
"loss": 1.2458,
|
| 4237 |
+
"step": 604
|
| 4238 |
+
},
|
| 4239 |
+
{
|
| 4240 |
+
"epoch": 0.663923182441701,
|
| 4241 |
+
"grad_norm": 0.2368205338716507,
|
| 4242 |
+
"learning_rate": 7.546239210850802e-06,
|
| 4243 |
+
"loss": 1.1767,
|
| 4244 |
+
"step": 605
|
| 4245 |
+
},
|
| 4246 |
+
{
|
| 4247 |
+
"epoch": 0.6650205761316872,
|
| 4248 |
+
"grad_norm": 0.26858776807785034,
|
| 4249 |
+
"learning_rate": 7.521578298397041e-06,
|
| 4250 |
+
"loss": 1.1943,
|
| 4251 |
+
"step": 606
|
| 4252 |
+
},
|
| 4253 |
+
{
|
| 4254 |
+
"epoch": 0.6661179698216735,
|
| 4255 |
+
"grad_norm": 0.23163066804409027,
|
| 4256 |
+
"learning_rate": 7.4969173859432805e-06,
|
| 4257 |
+
"loss": 1.2312,
|
| 4258 |
+
"step": 607
|
| 4259 |
+
},
|
| 4260 |
+
{
|
| 4261 |
+
"epoch": 0.6672153635116598,
|
| 4262 |
+
"grad_norm": 0.2653629183769226,
|
| 4263 |
+
"learning_rate": 7.472256473489519e-06,
|
| 4264 |
+
"loss": 1.2662,
|
| 4265 |
+
"step": 608
|
| 4266 |
+
},
|
| 4267 |
+
{
|
| 4268 |
+
"epoch": 0.6683127572016461,
|
| 4269 |
+
"grad_norm": 0.2933473587036133,
|
| 4270 |
+
"learning_rate": 7.447595561035758e-06,
|
| 4271 |
+
"loss": 1.1553,
|
| 4272 |
+
"step": 609
|
| 4273 |
+
},
|
| 4274 |
+
{
|
| 4275 |
+
"epoch": 0.6694101508916324,
|
| 4276 |
+
"grad_norm": 0.2857976257801056,
|
| 4277 |
+
"learning_rate": 7.422934648581999e-06,
|
| 4278 |
+
"loss": 1.245,
|
| 4279 |
+
"step": 610
|
| 4280 |
+
},
|
| 4281 |
+
{
|
| 4282 |
+
"epoch": 0.6705075445816187,
|
| 4283 |
+
"grad_norm": 0.247391477227211,
|
| 4284 |
+
"learning_rate": 7.398273736128238e-06,
|
| 4285 |
+
"loss": 1.2444,
|
| 4286 |
+
"step": 611
|
| 4287 |
+
},
|
| 4288 |
+
{
|
| 4289 |
+
"epoch": 0.671604938271605,
|
| 4290 |
+
"grad_norm": 0.2513018250465393,
|
| 4291 |
+
"learning_rate": 7.373612823674477e-06,
|
| 4292 |
+
"loss": 1.2437,
|
| 4293 |
+
"step": 612
|
| 4294 |
+
},
|
| 4295 |
+
{
|
| 4296 |
+
"epoch": 0.6727023319615912,
|
| 4297 |
+
"grad_norm": 0.23699401319026947,
|
| 4298 |
+
"learning_rate": 7.348951911220716e-06,
|
| 4299 |
+
"loss": 1.2296,
|
| 4300 |
+
"step": 613
|
| 4301 |
+
},
|
| 4302 |
+
{
|
| 4303 |
+
"epoch": 0.6737997256515775,
|
| 4304 |
+
"grad_norm": 0.2637063264846802,
|
| 4305 |
+
"learning_rate": 7.324290998766955e-06,
|
| 4306 |
+
"loss": 1.1789,
|
| 4307 |
+
"step": 614
|
| 4308 |
+
},
|
| 4309 |
+
{
|
| 4310 |
+
"epoch": 0.6748971193415638,
|
| 4311 |
+
"grad_norm": 0.26773321628570557,
|
| 4312 |
+
"learning_rate": 7.299630086313194e-06,
|
| 4313 |
+
"loss": 1.2081,
|
| 4314 |
+
"step": 615
|
| 4315 |
+
},
|
| 4316 |
+
{
|
| 4317 |
+
"epoch": 0.6759945130315501,
|
| 4318 |
+
"grad_norm": 0.2662067711353302,
|
| 4319 |
+
"learning_rate": 7.274969173859433e-06,
|
| 4320 |
+
"loss": 1.2615,
|
| 4321 |
+
"step": 616
|
| 4322 |
+
},
|
| 4323 |
+
{
|
| 4324 |
+
"epoch": 0.6770919067215363,
|
| 4325 |
+
"grad_norm": 0.2632356286048889,
|
| 4326 |
+
"learning_rate": 7.250308261405673e-06,
|
| 4327 |
+
"loss": 1.2003,
|
| 4328 |
+
"step": 617
|
| 4329 |
+
},
|
| 4330 |
+
{
|
| 4331 |
+
"epoch": 0.6781893004115226,
|
| 4332 |
+
"grad_norm": 0.24281790852546692,
|
| 4333 |
+
"learning_rate": 7.225647348951912e-06,
|
| 4334 |
+
"loss": 1.1169,
|
| 4335 |
+
"step": 618
|
| 4336 |
+
},
|
| 4337 |
+
{
|
| 4338 |
+
"epoch": 0.6792866941015089,
|
| 4339 |
+
"grad_norm": 0.24905385076999664,
|
| 4340 |
+
"learning_rate": 7.200986436498151e-06,
|
| 4341 |
+
"loss": 1.2919,
|
| 4342 |
+
"step": 619
|
| 4343 |
+
},
|
| 4344 |
+
{
|
| 4345 |
+
"epoch": 0.6803840877914952,
|
| 4346 |
+
"grad_norm": 0.24835814535617828,
|
| 4347 |
+
"learning_rate": 7.1763255240443905e-06,
|
| 4348 |
+
"loss": 1.2515,
|
| 4349 |
+
"step": 620
|
| 4350 |
+
},
|
| 4351 |
+
{
|
| 4352 |
+
"epoch": 0.6814814814814815,
|
| 4353 |
+
"grad_norm": 0.261608362197876,
|
| 4354 |
+
"learning_rate": 7.1516646115906294e-06,
|
| 4355 |
+
"loss": 1.1304,
|
| 4356 |
+
"step": 621
|
| 4357 |
+
},
|
| 4358 |
+
{
|
| 4359 |
+
"epoch": 0.6825788751714678,
|
| 4360 |
+
"grad_norm": 0.2825202941894531,
|
| 4361 |
+
"learning_rate": 7.127003699136868e-06,
|
| 4362 |
+
"loss": 1.1905,
|
| 4363 |
+
"step": 622
|
| 4364 |
+
},
|
| 4365 |
+
{
|
| 4366 |
+
"epoch": 0.683676268861454,
|
| 4367 |
+
"grad_norm": 0.25553426146507263,
|
| 4368 |
+
"learning_rate": 7.102342786683107e-06,
|
| 4369 |
+
"loss": 1.2196,
|
| 4370 |
+
"step": 623
|
| 4371 |
+
},
|
| 4372 |
+
{
|
| 4373 |
+
"epoch": 0.6847736625514403,
|
| 4374 |
+
"grad_norm": 0.23880372941493988,
|
| 4375 |
+
"learning_rate": 7.077681874229347e-06,
|
| 4376 |
+
"loss": 1.1902,
|
| 4377 |
+
"step": 624
|
| 4378 |
+
},
|
| 4379 |
+
{
|
| 4380 |
+
"epoch": 0.6858710562414266,
|
| 4381 |
+
"grad_norm": 0.24629957973957062,
|
| 4382 |
+
"learning_rate": 7.053020961775586e-06,
|
| 4383 |
+
"loss": 1.224,
|
| 4384 |
+
"step": 625
|
| 4385 |
+
},
|
| 4386 |
+
{
|
| 4387 |
+
"epoch": 0.6869684499314129,
|
| 4388 |
+
"grad_norm": 0.2556987702846527,
|
| 4389 |
+
"learning_rate": 7.028360049321825e-06,
|
| 4390 |
+
"loss": 1.2526,
|
| 4391 |
+
"step": 626
|
| 4392 |
+
},
|
| 4393 |
+
{
|
| 4394 |
+
"epoch": 0.6880658436213992,
|
| 4395 |
+
"grad_norm": 0.24384275078773499,
|
| 4396 |
+
"learning_rate": 7.003699136868065e-06,
|
| 4397 |
+
"loss": 1.2501,
|
| 4398 |
+
"step": 627
|
| 4399 |
+
},
|
| 4400 |
+
{
|
| 4401 |
+
"epoch": 0.6891632373113855,
|
| 4402 |
+
"grad_norm": 0.24641302227973938,
|
| 4403 |
+
"learning_rate": 6.979038224414304e-06,
|
| 4404 |
+
"loss": 1.2761,
|
| 4405 |
+
"step": 628
|
| 4406 |
+
},
|
| 4407 |
+
{
|
| 4408 |
+
"epoch": 0.6902606310013717,
|
| 4409 |
+
"grad_norm": 0.22154922783374786,
|
| 4410 |
+
"learning_rate": 6.954377311960543e-06,
|
| 4411 |
+
"loss": 1.3006,
|
| 4412 |
+
"step": 629
|
| 4413 |
+
},
|
| 4414 |
+
{
|
| 4415 |
+
"epoch": 0.691358024691358,
|
| 4416 |
+
"grad_norm": 0.22881589829921722,
|
| 4417 |
+
"learning_rate": 6.929716399506782e-06,
|
| 4418 |
+
"loss": 1.2458,
|
| 4419 |
+
"step": 630
|
| 4420 |
+
},
|
| 4421 |
+
{
|
| 4422 |
+
"epoch": 0.6924554183813443,
|
| 4423 |
+
"grad_norm": 0.25358015298843384,
|
| 4424 |
+
"learning_rate": 6.905055487053022e-06,
|
| 4425 |
+
"loss": 1.1781,
|
| 4426 |
+
"step": 631
|
| 4427 |
+
},
|
| 4428 |
+
{
|
| 4429 |
+
"epoch": 0.6935528120713306,
|
| 4430 |
+
"grad_norm": 0.2565036118030548,
|
| 4431 |
+
"learning_rate": 6.880394574599261e-06,
|
| 4432 |
+
"loss": 1.1634,
|
| 4433 |
+
"step": 632
|
| 4434 |
+
},
|
| 4435 |
+
{
|
| 4436 |
+
"epoch": 0.6946502057613169,
|
| 4437 |
+
"grad_norm": 0.2973492443561554,
|
| 4438 |
+
"learning_rate": 6.8557336621455e-06,
|
| 4439 |
+
"loss": 1.2238,
|
| 4440 |
+
"step": 633
|
| 4441 |
+
},
|
| 4442 |
+
{
|
| 4443 |
+
"epoch": 0.6957475994513032,
|
| 4444 |
+
"grad_norm": 0.26048916578292847,
|
| 4445 |
+
"learning_rate": 6.8310727496917395e-06,
|
| 4446 |
+
"loss": 1.3067,
|
| 4447 |
+
"step": 634
|
| 4448 |
+
},
|
| 4449 |
+
{
|
| 4450 |
+
"epoch": 0.6968449931412894,
|
| 4451 |
+
"grad_norm": 0.2569065988063812,
|
| 4452 |
+
"learning_rate": 6.806411837237978e-06,
|
| 4453 |
+
"loss": 1.2878,
|
| 4454 |
+
"step": 635
|
| 4455 |
+
},
|
| 4456 |
+
{
|
| 4457 |
+
"epoch": 0.6979423868312757,
|
| 4458 |
+
"grad_norm": 0.3082578480243683,
|
| 4459 |
+
"learning_rate": 6.781750924784217e-06,
|
| 4460 |
+
"loss": 1.1396,
|
| 4461 |
+
"step": 636
|
| 4462 |
+
},
|
| 4463 |
+
{
|
| 4464 |
+
"epoch": 0.699039780521262,
|
| 4465 |
+
"grad_norm": 0.23459649085998535,
|
| 4466 |
+
"learning_rate": 6.757090012330457e-06,
|
| 4467 |
+
"loss": 1.3344,
|
| 4468 |
+
"step": 637
|
| 4469 |
+
},
|
| 4470 |
+
{
|
| 4471 |
+
"epoch": 0.7001371742112483,
|
| 4472 |
+
"grad_norm": 0.27508512139320374,
|
| 4473 |
+
"learning_rate": 6.732429099876696e-06,
|
| 4474 |
+
"loss": 1.1655,
|
| 4475 |
+
"step": 638
|
| 4476 |
+
},
|
| 4477 |
+
{
|
| 4478 |
+
"epoch": 0.7012345679012346,
|
| 4479 |
+
"grad_norm": 0.26993826031684875,
|
| 4480 |
+
"learning_rate": 6.707768187422935e-06,
|
| 4481 |
+
"loss": 1.2815,
|
| 4482 |
+
"step": 639
|
| 4483 |
+
},
|
| 4484 |
+
{
|
| 4485 |
+
"epoch": 0.7023319615912208,
|
| 4486 |
+
"grad_norm": 0.24884283542633057,
|
| 4487 |
+
"learning_rate": 6.683107274969174e-06,
|
| 4488 |
+
"loss": 1.1821,
|
| 4489 |
+
"step": 640
|
| 4490 |
+
},
|
| 4491 |
+
{
|
| 4492 |
+
"epoch": 0.7034293552812071,
|
| 4493 |
+
"grad_norm": 0.24485498666763306,
|
| 4494 |
+
"learning_rate": 6.6584463625154135e-06,
|
| 4495 |
+
"loss": 1.2706,
|
| 4496 |
+
"step": 641
|
| 4497 |
+
},
|
| 4498 |
+
{
|
| 4499 |
+
"epoch": 0.7045267489711934,
|
| 4500 |
+
"grad_norm": 0.23114155232906342,
|
| 4501 |
+
"learning_rate": 6.633785450061652e-06,
|
| 4502 |
+
"loss": 1.2459,
|
| 4503 |
+
"step": 642
|
| 4504 |
+
},
|
| 4505 |
+
{
|
| 4506 |
+
"epoch": 0.7056241426611797,
|
| 4507 |
+
"grad_norm": 0.2295503467321396,
|
| 4508 |
+
"learning_rate": 6.609124537607891e-06,
|
| 4509 |
+
"loss": 1.2172,
|
| 4510 |
+
"step": 643
|
| 4511 |
+
},
|
| 4512 |
+
{
|
| 4513 |
+
"epoch": 0.706721536351166,
|
| 4514 |
+
"grad_norm": 0.23798534274101257,
|
| 4515 |
+
"learning_rate": 6.584463625154132e-06,
|
| 4516 |
+
"loss": 1.2519,
|
| 4517 |
+
"step": 644
|
| 4518 |
+
},
|
| 4519 |
+
{
|
| 4520 |
+
"epoch": 0.7078189300411523,
|
| 4521 |
+
"grad_norm": 0.23953770101070404,
|
| 4522 |
+
"learning_rate": 6.559802712700371e-06,
|
| 4523 |
+
"loss": 1.2205,
|
| 4524 |
+
"step": 645
|
| 4525 |
+
},
|
| 4526 |
+
{
|
| 4527 |
+
"epoch": 0.7089163237311386,
|
| 4528 |
+
"grad_norm": 0.2792452871799469,
|
| 4529 |
+
"learning_rate": 6.53514180024661e-06,
|
| 4530 |
+
"loss": 1.1207,
|
| 4531 |
+
"step": 646
|
| 4532 |
+
},
|
| 4533 |
+
{
|
| 4534 |
+
"epoch": 0.7100137174211248,
|
| 4535 |
+
"grad_norm": 0.23514851927757263,
|
| 4536 |
+
"learning_rate": 6.5104808877928495e-06,
|
| 4537 |
+
"loss": 1.2907,
|
| 4538 |
+
"step": 647
|
| 4539 |
+
},
|
| 4540 |
+
{
|
| 4541 |
+
"epoch": 0.7111111111111111,
|
| 4542 |
+
"grad_norm": 0.2464553862810135,
|
| 4543 |
+
"learning_rate": 6.485819975339088e-06,
|
| 4544 |
+
"loss": 1.3023,
|
| 4545 |
+
"step": 648
|
| 4546 |
+
},
|
| 4547 |
+
{
|
| 4548 |
+
"epoch": 0.7122085048010974,
|
| 4549 |
+
"grad_norm": 0.32537391781806946,
|
| 4550 |
+
"learning_rate": 6.461159062885327e-06,
|
| 4551 |
+
"loss": 1.152,
|
| 4552 |
+
"step": 649
|
| 4553 |
+
},
|
| 4554 |
+
{
|
| 4555 |
+
"epoch": 0.7133058984910837,
|
| 4556 |
+
"grad_norm": 0.22448231279850006,
|
| 4557 |
+
"learning_rate": 6.436498150431566e-06,
|
| 4558 |
+
"loss": 1.2991,
|
| 4559 |
+
"step": 650
|
| 4560 |
+
},
|
| 4561 |
+
{
|
| 4562 |
+
"epoch": 0.7144032921810699,
|
| 4563 |
+
"grad_norm": 0.24453403055667877,
|
| 4564 |
+
"learning_rate": 6.411837237977806e-06,
|
| 4565 |
+
"loss": 1.2375,
|
| 4566 |
+
"step": 651
|
| 4567 |
+
},
|
| 4568 |
+
{
|
| 4569 |
+
"epoch": 0.7155006858710562,
|
| 4570 |
+
"grad_norm": 0.2662949860095978,
|
| 4571 |
+
"learning_rate": 6.387176325524045e-06,
|
| 4572 |
+
"loss": 1.2031,
|
| 4573 |
+
"step": 652
|
| 4574 |
+
},
|
| 4575 |
+
{
|
| 4576 |
+
"epoch": 0.7165980795610425,
|
| 4577 |
+
"grad_norm": 0.27420365810394287,
|
| 4578 |
+
"learning_rate": 6.362515413070284e-06,
|
| 4579 |
+
"loss": 1.1687,
|
| 4580 |
+
"step": 653
|
| 4581 |
+
},
|
| 4582 |
+
{
|
| 4583 |
+
"epoch": 0.7176954732510288,
|
| 4584 |
+
"grad_norm": 0.2973918616771698,
|
| 4585 |
+
"learning_rate": 6.3378545006165236e-06,
|
| 4586 |
+
"loss": 1.1109,
|
| 4587 |
+
"step": 654
|
| 4588 |
+
},
|
| 4589 |
+
{
|
| 4590 |
+
"epoch": 0.7187928669410151,
|
| 4591 |
+
"grad_norm": 0.23176413774490356,
|
| 4592 |
+
"learning_rate": 6.3131935881627625e-06,
|
| 4593 |
+
"loss": 1.2921,
|
| 4594 |
+
"step": 655
|
| 4595 |
+
},
|
| 4596 |
+
{
|
| 4597 |
+
"epoch": 0.7198902606310014,
|
| 4598 |
+
"grad_norm": 0.2920747399330139,
|
| 4599 |
+
"learning_rate": 6.288532675709001e-06,
|
| 4600 |
+
"loss": 1.2242,
|
| 4601 |
+
"step": 656
|
| 4602 |
+
},
|
| 4603 |
+
{
|
| 4604 |
+
"epoch": 0.7209876543209877,
|
| 4605 |
+
"grad_norm": 0.3083314597606659,
|
| 4606 |
+
"learning_rate": 6.263871763255241e-06,
|
| 4607 |
+
"loss": 1.1806,
|
| 4608 |
+
"step": 657
|
| 4609 |
+
},
|
| 4610 |
+
{
|
| 4611 |
+
"epoch": 0.722085048010974,
|
| 4612 |
+
"grad_norm": 0.261935293674469,
|
| 4613 |
+
"learning_rate": 6.23921085080148e-06,
|
| 4614 |
+
"loss": 1.2915,
|
| 4615 |
+
"step": 658
|
| 4616 |
+
},
|
| 4617 |
+
{
|
| 4618 |
+
"epoch": 0.7231824417009602,
|
| 4619 |
+
"grad_norm": 0.2600385248661041,
|
| 4620 |
+
"learning_rate": 6.214549938347719e-06,
|
| 4621 |
+
"loss": 1.2212,
|
| 4622 |
+
"step": 659
|
| 4623 |
+
},
|
| 4624 |
+
{
|
| 4625 |
+
"epoch": 0.7242798353909465,
|
| 4626 |
+
"grad_norm": 0.2797953486442566,
|
| 4627 |
+
"learning_rate": 6.189889025893958e-06,
|
| 4628 |
+
"loss": 1.2557,
|
| 4629 |
+
"step": 660
|
| 4630 |
+
},
|
| 4631 |
+
{
|
| 4632 |
+
"epoch": 0.7253772290809328,
|
| 4633 |
+
"grad_norm": 0.25546994805336,
|
| 4634 |
+
"learning_rate": 6.1652281134401985e-06,
|
| 4635 |
+
"loss": 1.1503,
|
| 4636 |
+
"step": 661
|
| 4637 |
+
},
|
| 4638 |
+
{
|
| 4639 |
+
"epoch": 0.7264746227709191,
|
| 4640 |
+
"grad_norm": 0.26473572850227356,
|
| 4641 |
+
"learning_rate": 6.140567200986437e-06,
|
| 4642 |
+
"loss": 1.3061,
|
| 4643 |
+
"step": 662
|
| 4644 |
+
},
|
| 4645 |
+
{
|
| 4646 |
+
"epoch": 0.7275720164609053,
|
| 4647 |
+
"grad_norm": 0.28287649154663086,
|
| 4648 |
+
"learning_rate": 6.115906288532676e-06,
|
| 4649 |
+
"loss": 1.2115,
|
| 4650 |
+
"step": 663
|
| 4651 |
+
},
|
| 4652 |
+
{
|
| 4653 |
+
"epoch": 0.7286694101508916,
|
| 4654 |
+
"grad_norm": 0.27184855937957764,
|
| 4655 |
+
"learning_rate": 6.091245376078916e-06,
|
| 4656 |
+
"loss": 1.1986,
|
| 4657 |
+
"step": 664
|
| 4658 |
+
},
|
| 4659 |
+
{
|
| 4660 |
+
"epoch": 0.7297668038408779,
|
| 4661 |
+
"grad_norm": 0.260458767414093,
|
| 4662 |
+
"learning_rate": 6.066584463625155e-06,
|
| 4663 |
+
"loss": 1.2452,
|
| 4664 |
+
"step": 665
|
| 4665 |
+
},
|
| 4666 |
+
{
|
| 4667 |
+
"epoch": 0.7308641975308642,
|
| 4668 |
+
"grad_norm": 0.25575020909309387,
|
| 4669 |
+
"learning_rate": 6.041923551171394e-06,
|
| 4670 |
+
"loss": 1.3161,
|
| 4671 |
+
"step": 666
|
| 4672 |
+
},
|
| 4673 |
+
{
|
| 4674 |
+
"epoch": 0.7319615912208505,
|
| 4675 |
+
"grad_norm": 0.26954489946365356,
|
| 4676 |
+
"learning_rate": 6.017262638717633e-06,
|
| 4677 |
+
"loss": 1.3298,
|
| 4678 |
+
"step": 667
|
| 4679 |
+
},
|
| 4680 |
+
{
|
| 4681 |
+
"epoch": 0.7330589849108368,
|
| 4682 |
+
"grad_norm": 0.2576771080493927,
|
| 4683 |
+
"learning_rate": 5.9926017262638725e-06,
|
| 4684 |
+
"loss": 1.2151,
|
| 4685 |
+
"step": 668
|
| 4686 |
+
},
|
| 4687 |
+
{
|
| 4688 |
+
"epoch": 0.7341563786008231,
|
| 4689 |
+
"grad_norm": 0.24782928824424744,
|
| 4690 |
+
"learning_rate": 5.967940813810111e-06,
|
| 4691 |
+
"loss": 1.2997,
|
| 4692 |
+
"step": 669
|
| 4693 |
+
},
|
| 4694 |
+
{
|
| 4695 |
+
"epoch": 0.7352537722908093,
|
| 4696 |
+
"grad_norm": 0.24412375688552856,
|
| 4697 |
+
"learning_rate": 5.94327990135635e-06,
|
| 4698 |
+
"loss": 1.2091,
|
| 4699 |
+
"step": 670
|
| 4700 |
+
},
|
| 4701 |
+
{
|
| 4702 |
+
"epoch": 0.7363511659807956,
|
| 4703 |
+
"grad_norm": 0.35788780450820923,
|
| 4704 |
+
"learning_rate": 5.91861898890259e-06,
|
| 4705 |
+
"loss": 1.1713,
|
| 4706 |
+
"step": 671
|
| 4707 |
+
},
|
| 4708 |
+
{
|
| 4709 |
+
"epoch": 0.7374485596707819,
|
| 4710 |
+
"grad_norm": 0.265381783246994,
|
| 4711 |
+
"learning_rate": 5.893958076448829e-06,
|
| 4712 |
+
"loss": 1.2459,
|
| 4713 |
+
"step": 672
|
| 4714 |
+
},
|
| 4715 |
+
{
|
| 4716 |
+
"epoch": 0.7385459533607682,
|
| 4717 |
+
"grad_norm": 0.2551698088645935,
|
| 4718 |
+
"learning_rate": 5.869297163995068e-06,
|
| 4719 |
+
"loss": 1.2756,
|
| 4720 |
+
"step": 673
|
| 4721 |
+
},
|
| 4722 |
+
{
|
| 4723 |
+
"epoch": 0.7396433470507544,
|
| 4724 |
+
"grad_norm": 0.2247520238161087,
|
| 4725 |
+
"learning_rate": 5.844636251541308e-06,
|
| 4726 |
+
"loss": 1.2513,
|
| 4727 |
+
"step": 674
|
| 4728 |
+
},
|
| 4729 |
+
{
|
| 4730 |
+
"epoch": 0.7407407407407407,
|
| 4731 |
+
"grad_norm": 0.2614815831184387,
|
| 4732 |
+
"learning_rate": 5.8199753390875466e-06,
|
| 4733 |
+
"loss": 1.2349,
|
| 4734 |
+
"step": 675
|
| 4735 |
+
},
|
| 4736 |
+
{
|
| 4737 |
+
"epoch": 0.741838134430727,
|
| 4738 |
+
"grad_norm": 0.2318747192621231,
|
| 4739 |
+
"learning_rate": 5.7953144266337855e-06,
|
| 4740 |
+
"loss": 1.2905,
|
| 4741 |
+
"step": 676
|
| 4742 |
+
},
|
| 4743 |
+
{
|
| 4744 |
+
"epoch": 0.7429355281207133,
|
| 4745 |
+
"grad_norm": 0.25398141145706177,
|
| 4746 |
+
"learning_rate": 5.770653514180024e-06,
|
| 4747 |
+
"loss": 1.2403,
|
| 4748 |
+
"step": 677
|
| 4749 |
+
},
|
| 4750 |
+
{
|
| 4751 |
+
"epoch": 0.7440329218106996,
|
| 4752 |
+
"grad_norm": 0.2757530212402344,
|
| 4753 |
+
"learning_rate": 5.745992601726265e-06,
|
| 4754 |
+
"loss": 1.3219,
|
| 4755 |
+
"step": 678
|
| 4756 |
+
},
|
| 4757 |
+
{
|
| 4758 |
+
"epoch": 0.7451303155006859,
|
| 4759 |
+
"grad_norm": 0.2670435309410095,
|
| 4760 |
+
"learning_rate": 5.721331689272504e-06,
|
| 4761 |
+
"loss": 1.2375,
|
| 4762 |
+
"step": 679
|
| 4763 |
+
},
|
| 4764 |
+
{
|
| 4765 |
+
"epoch": 0.7462277091906722,
|
| 4766 |
+
"grad_norm": 0.238840714097023,
|
| 4767 |
+
"learning_rate": 5.696670776818743e-06,
|
| 4768 |
+
"loss": 1.299,
|
| 4769 |
+
"step": 680
|
| 4770 |
+
},
|
| 4771 |
+
{
|
| 4772 |
+
"epoch": 0.7473251028806585,
|
| 4773 |
+
"grad_norm": 0.24049274623394012,
|
| 4774 |
+
"learning_rate": 5.6720098643649825e-06,
|
| 4775 |
+
"loss": 1.2417,
|
| 4776 |
+
"step": 681
|
| 4777 |
+
},
|
| 4778 |
+
{
|
| 4779 |
+
"epoch": 0.7484224965706447,
|
| 4780 |
+
"grad_norm": 0.2531554102897644,
|
| 4781 |
+
"learning_rate": 5.6473489519112214e-06,
|
| 4782 |
+
"loss": 1.2263,
|
| 4783 |
+
"step": 682
|
| 4784 |
+
},
|
| 4785 |
+
{
|
| 4786 |
+
"epoch": 0.749519890260631,
|
| 4787 |
+
"grad_norm": 0.2929145097732544,
|
| 4788 |
+
"learning_rate": 5.62268803945746e-06,
|
| 4789 |
+
"loss": 1.1856,
|
| 4790 |
+
"step": 683
|
| 4791 |
+
},
|
| 4792 |
+
{
|
| 4793 |
+
"epoch": 0.7506172839506173,
|
| 4794 |
+
"grad_norm": 0.2513459622859955,
|
| 4795 |
+
"learning_rate": 5.5980271270037e-06,
|
| 4796 |
+
"loss": 1.2045,
|
| 4797 |
+
"step": 684
|
| 4798 |
+
},
|
| 4799 |
+
{
|
| 4800 |
+
"epoch": 0.7517146776406035,
|
| 4801 |
+
"grad_norm": 0.24417419731616974,
|
| 4802 |
+
"learning_rate": 5.573366214549939e-06,
|
| 4803 |
+
"loss": 1.2951,
|
| 4804 |
+
"step": 685
|
| 4805 |
+
},
|
| 4806 |
+
{
|
| 4807 |
+
"epoch": 0.7528120713305898,
|
| 4808 |
+
"grad_norm": 0.25240176916122437,
|
| 4809 |
+
"learning_rate": 5.548705302096178e-06,
|
| 4810 |
+
"loss": 1.1853,
|
| 4811 |
+
"step": 686
|
| 4812 |
+
},
|
| 4813 |
+
{
|
| 4814 |
+
"epoch": 0.7539094650205761,
|
| 4815 |
+
"grad_norm": 0.23649027943611145,
|
| 4816 |
+
"learning_rate": 5.524044389642417e-06,
|
| 4817 |
+
"loss": 1.2184,
|
| 4818 |
+
"step": 687
|
| 4819 |
+
},
|
| 4820 |
+
{
|
| 4821 |
+
"epoch": 0.7550068587105624,
|
| 4822 |
+
"grad_norm": 0.25390321016311646,
|
| 4823 |
+
"learning_rate": 5.499383477188657e-06,
|
| 4824 |
+
"loss": 1.2343,
|
| 4825 |
+
"step": 688
|
| 4826 |
+
},
|
| 4827 |
+
{
|
| 4828 |
+
"epoch": 0.7561042524005487,
|
| 4829 |
+
"grad_norm": 0.28937819600105286,
|
| 4830 |
+
"learning_rate": 5.4747225647348955e-06,
|
| 4831 |
+
"loss": 1.1617,
|
| 4832 |
+
"step": 689
|
| 4833 |
+
},
|
| 4834 |
+
{
|
| 4835 |
+
"epoch": 0.757201646090535,
|
| 4836 |
+
"grad_norm": 0.23086318373680115,
|
| 4837 |
+
"learning_rate": 5.450061652281134e-06,
|
| 4838 |
+
"loss": 1.2191,
|
| 4839 |
+
"step": 690
|
| 4840 |
+
},
|
| 4841 |
+
{
|
| 4842 |
+
"epoch": 0.7582990397805213,
|
| 4843 |
+
"grad_norm": 0.33580687642097473,
|
| 4844 |
+
"learning_rate": 5.425400739827374e-06,
|
| 4845 |
+
"loss": 1.3355,
|
| 4846 |
+
"step": 691
|
| 4847 |
+
},
|
| 4848 |
+
{
|
| 4849 |
+
"epoch": 0.7593964334705076,
|
| 4850 |
+
"grad_norm": 0.2412438690662384,
|
| 4851 |
+
"learning_rate": 5.400739827373613e-06,
|
| 4852 |
+
"loss": 1.2707,
|
| 4853 |
+
"step": 692
|
| 4854 |
+
},
|
| 4855 |
+
{
|
| 4856 |
+
"epoch": 0.7604938271604939,
|
| 4857 |
+
"grad_norm": 0.23220427334308624,
|
| 4858 |
+
"learning_rate": 5.376078914919852e-06,
|
| 4859 |
+
"loss": 1.1805,
|
| 4860 |
+
"step": 693
|
| 4861 |
+
},
|
| 4862 |
+
{
|
| 4863 |
+
"epoch": 0.7615912208504801,
|
| 4864 |
+
"grad_norm": 0.24792106449604034,
|
| 4865 |
+
"learning_rate": 5.351418002466093e-06,
|
| 4866 |
+
"loss": 1.1673,
|
| 4867 |
+
"step": 694
|
| 4868 |
+
},
|
| 4869 |
+
{
|
| 4870 |
+
"epoch": 0.7626886145404664,
|
| 4871 |
+
"grad_norm": 0.25883495807647705,
|
| 4872 |
+
"learning_rate": 5.3267570900123315e-06,
|
| 4873 |
+
"loss": 1.2681,
|
| 4874 |
+
"step": 695
|
| 4875 |
+
},
|
| 4876 |
+
{
|
| 4877 |
+
"epoch": 0.7637860082304527,
|
| 4878 |
+
"grad_norm": 0.24207496643066406,
|
| 4879 |
+
"learning_rate": 5.30209617755857e-06,
|
| 4880 |
+
"loss": 1.2634,
|
| 4881 |
+
"step": 696
|
| 4882 |
+
},
|
| 4883 |
+
{
|
| 4884 |
+
"epoch": 0.7648834019204389,
|
| 4885 |
+
"grad_norm": 0.3035838305950165,
|
| 4886 |
+
"learning_rate": 5.2774352651048085e-06,
|
| 4887 |
+
"loss": 1.2719,
|
| 4888 |
+
"step": 697
|
| 4889 |
+
},
|
| 4890 |
+
{
|
| 4891 |
+
"epoch": 0.7659807956104252,
|
| 4892 |
+
"grad_norm": 0.25187456607818604,
|
| 4893 |
+
"learning_rate": 5.252774352651049e-06,
|
| 4894 |
+
"loss": 1.2197,
|
| 4895 |
+
"step": 698
|
| 4896 |
+
},
|
| 4897 |
+
{
|
| 4898 |
+
"epoch": 0.7670781893004115,
|
| 4899 |
+
"grad_norm": 0.2787846624851227,
|
| 4900 |
+
"learning_rate": 5.228113440197288e-06,
|
| 4901 |
+
"loss": 1.2146,
|
| 4902 |
+
"step": 699
|
| 4903 |
+
},
|
| 4904 |
+
{
|
| 4905 |
+
"epoch": 0.7681755829903978,
|
| 4906 |
+
"grad_norm": 0.29621052742004395,
|
| 4907 |
+
"learning_rate": 5.203452527743527e-06,
|
| 4908 |
+
"loss": 1.2072,
|
| 4909 |
+
"step": 700
|
| 4910 |
+
},
|
| 4911 |
+
{
|
| 4912 |
+
"epoch": 0.7692729766803841,
|
| 4913 |
+
"grad_norm": 0.2812381088733673,
|
| 4914 |
+
"learning_rate": 5.178791615289767e-06,
|
| 4915 |
+
"loss": 1.1203,
|
| 4916 |
+
"step": 701
|
| 4917 |
+
},
|
| 4918 |
+
{
|
| 4919 |
+
"epoch": 0.7703703703703704,
|
| 4920 |
+
"grad_norm": 0.25507667660713196,
|
| 4921 |
+
"learning_rate": 5.1541307028360055e-06,
|
| 4922 |
+
"loss": 1.1712,
|
| 4923 |
+
"step": 702
|
| 4924 |
+
},
|
| 4925 |
+
{
|
| 4926 |
+
"epoch": 0.7714677640603567,
|
| 4927 |
+
"grad_norm": 0.23398463428020477,
|
| 4928 |
+
"learning_rate": 5.1294697903822444e-06,
|
| 4929 |
+
"loss": 1.2786,
|
| 4930 |
+
"step": 703
|
| 4931 |
+
},
|
| 4932 |
+
{
|
| 4933 |
+
"epoch": 0.772565157750343,
|
| 4934 |
+
"grad_norm": 0.27228790521621704,
|
| 4935 |
+
"learning_rate": 5.104808877928483e-06,
|
| 4936 |
+
"loss": 1.1069,
|
| 4937 |
+
"step": 704
|
| 4938 |
+
},
|
| 4939 |
+
{
|
| 4940 |
+
"epoch": 0.7736625514403292,
|
| 4941 |
+
"grad_norm": 0.2365400493144989,
|
| 4942 |
+
"learning_rate": 5.080147965474723e-06,
|
| 4943 |
+
"loss": 1.2509,
|
| 4944 |
+
"step": 705
|
| 4945 |
+
},
|
| 4946 |
+
{
|
| 4947 |
+
"epoch": 0.7747599451303155,
|
| 4948 |
+
"grad_norm": 0.2600681781768799,
|
| 4949 |
+
"learning_rate": 5.055487053020962e-06,
|
| 4950 |
+
"loss": 1.2548,
|
| 4951 |
+
"step": 706
|
| 4952 |
+
},
|
| 4953 |
+
{
|
| 4954 |
+
"epoch": 0.7758573388203018,
|
| 4955 |
+
"grad_norm": 0.23477765917778015,
|
| 4956 |
+
"learning_rate": 5.030826140567201e-06,
|
| 4957 |
+
"loss": 1.2885,
|
| 4958 |
+
"step": 707
|
| 4959 |
+
},
|
| 4960 |
+
{
|
| 4961 |
+
"epoch": 0.776954732510288,
|
| 4962 |
+
"grad_norm": 0.23022587597370148,
|
| 4963 |
+
"learning_rate": 5.006165228113441e-06,
|
| 4964 |
+
"loss": 1.312,
|
| 4965 |
+
"step": 708
|
| 4966 |
+
},
|
| 4967 |
+
{
|
| 4968 |
+
"epoch": 0.7780521262002743,
|
| 4969 |
+
"grad_norm": 0.255176842212677,
|
| 4970 |
+
"learning_rate": 4.98150431565968e-06,
|
| 4971 |
+
"loss": 1.172,
|
| 4972 |
+
"step": 709
|
| 4973 |
+
},
|
| 4974 |
+
{
|
| 4975 |
+
"epoch": 0.7791495198902606,
|
| 4976 |
+
"grad_norm": 0.31377795338630676,
|
| 4977 |
+
"learning_rate": 4.956843403205919e-06,
|
| 4978 |
+
"loss": 1.1991,
|
| 4979 |
+
"step": 710
|
| 4980 |
+
},
|
| 4981 |
+
{
|
| 4982 |
+
"epoch": 0.7802469135802469,
|
| 4983 |
+
"grad_norm": 0.23953334987163544,
|
| 4984 |
+
"learning_rate": 4.932182490752158e-06,
|
| 4985 |
+
"loss": 1.2951,
|
| 4986 |
+
"step": 711
|
| 4987 |
+
},
|
| 4988 |
+
{
|
| 4989 |
+
"epoch": 0.7813443072702332,
|
| 4990 |
+
"grad_norm": 0.2673497200012207,
|
| 4991 |
+
"learning_rate": 4.907521578298398e-06,
|
| 4992 |
+
"loss": 1.2645,
|
| 4993 |
+
"step": 712
|
| 4994 |
+
},
|
| 4995 |
+
{
|
| 4996 |
+
"epoch": 0.7824417009602195,
|
| 4997 |
+
"grad_norm": 0.2460653930902481,
|
| 4998 |
+
"learning_rate": 4.882860665844636e-06,
|
| 4999 |
+
"loss": 1.2481,
|
| 5000 |
+
"step": 713
|
| 5001 |
+
},
|
| 5002 |
+
{
|
| 5003 |
+
"epoch": 0.7835390946502058,
|
| 5004 |
+
"grad_norm": 0.2654884457588196,
|
| 5005 |
+
"learning_rate": 4.858199753390876e-06,
|
| 5006 |
+
"loss": 1.2675,
|
| 5007 |
+
"step": 714
|
| 5008 |
+
},
|
| 5009 |
+
{
|
| 5010 |
+
"epoch": 0.7846364883401921,
|
| 5011 |
+
"grad_norm": 0.2592717409133911,
|
| 5012 |
+
"learning_rate": 4.833538840937116e-06,
|
| 5013 |
+
"loss": 1.2298,
|
| 5014 |
+
"step": 715
|
| 5015 |
+
},
|
| 5016 |
+
{
|
| 5017 |
+
"epoch": 0.7857338820301784,
|
| 5018 |
+
"grad_norm": 0.24062027037143707,
|
| 5019 |
+
"learning_rate": 4.8088779284833545e-06,
|
| 5020 |
+
"loss": 1.2382,
|
| 5021 |
+
"step": 716
|
| 5022 |
+
},
|
| 5023 |
+
{
|
| 5024 |
+
"epoch": 0.7868312757201646,
|
| 5025 |
+
"grad_norm": 0.24740296602249146,
|
| 5026 |
+
"learning_rate": 4.784217016029593e-06,
|
| 5027 |
+
"loss": 1.2934,
|
| 5028 |
+
"step": 717
|
| 5029 |
+
},
|
| 5030 |
+
{
|
| 5031 |
+
"epoch": 0.7879286694101509,
|
| 5032 |
+
"grad_norm": 0.2442992925643921,
|
| 5033 |
+
"learning_rate": 4.759556103575832e-06,
|
| 5034 |
+
"loss": 1.208,
|
| 5035 |
+
"step": 718
|
| 5036 |
+
},
|
| 5037 |
+
{
|
| 5038 |
+
"epoch": 0.7890260631001371,
|
| 5039 |
+
"grad_norm": 0.23618517816066742,
|
| 5040 |
+
"learning_rate": 4.734895191122072e-06,
|
| 5041 |
+
"loss": 1.175,
|
| 5042 |
+
"step": 719
|
| 5043 |
+
},
|
| 5044 |
+
{
|
| 5045 |
+
"epoch": 0.7901234567901234,
|
| 5046 |
+
"grad_norm": 0.2576107978820801,
|
| 5047 |
+
"learning_rate": 4.710234278668311e-06,
|
| 5048 |
+
"loss": 1.2199,
|
| 5049 |
+
"step": 720
|
| 5050 |
+
},
|
| 5051 |
+
{
|
| 5052 |
+
"epoch": 0.7912208504801097,
|
| 5053 |
+
"grad_norm": 0.25478753447532654,
|
| 5054 |
+
"learning_rate": 4.68557336621455e-06,
|
| 5055 |
+
"loss": 1.1987,
|
| 5056 |
+
"step": 721
|
| 5057 |
+
},
|
| 5058 |
+
{
|
| 5059 |
+
"epoch": 0.792318244170096,
|
| 5060 |
+
"grad_norm": 0.26377934217453003,
|
| 5061 |
+
"learning_rate": 4.66091245376079e-06,
|
| 5062 |
+
"loss": 1.2576,
|
| 5063 |
+
"step": 722
|
| 5064 |
+
},
|
| 5065 |
+
{
|
| 5066 |
+
"epoch": 0.7934156378600823,
|
| 5067 |
+
"grad_norm": 0.2350923717021942,
|
| 5068 |
+
"learning_rate": 4.6362515413070285e-06,
|
| 5069 |
+
"loss": 1.2626,
|
| 5070 |
+
"step": 723
|
| 5071 |
+
},
|
| 5072 |
+
{
|
| 5073 |
+
"epoch": 0.7945130315500686,
|
| 5074 |
+
"grad_norm": 0.26834535598754883,
|
| 5075 |
+
"learning_rate": 4.611590628853268e-06,
|
| 5076 |
+
"loss": 1.285,
|
| 5077 |
+
"step": 724
|
| 5078 |
+
},
|
| 5079 |
+
{
|
| 5080 |
+
"epoch": 0.7956104252400549,
|
| 5081 |
+
"grad_norm": 0.2490326464176178,
|
| 5082 |
+
"learning_rate": 4.586929716399507e-06,
|
| 5083 |
+
"loss": 1.2008,
|
| 5084 |
+
"step": 725
|
| 5085 |
+
},
|
| 5086 |
+
{
|
| 5087 |
+
"epoch": 0.7967078189300412,
|
| 5088 |
+
"grad_norm": 0.2497320920228958,
|
| 5089 |
+
"learning_rate": 4.562268803945746e-06,
|
| 5090 |
+
"loss": 1.2807,
|
| 5091 |
+
"step": 726
|
| 5092 |
+
},
|
| 5093 |
+
{
|
| 5094 |
+
"epoch": 0.7978052126200275,
|
| 5095 |
+
"grad_norm": 0.2902284562587738,
|
| 5096 |
+
"learning_rate": 4.537607891491986e-06,
|
| 5097 |
+
"loss": 1.1737,
|
| 5098 |
+
"step": 727
|
| 5099 |
+
},
|
| 5100 |
+
{
|
| 5101 |
+
"epoch": 0.7989026063100138,
|
| 5102 |
+
"grad_norm": 0.26056787371635437,
|
| 5103 |
+
"learning_rate": 4.512946979038225e-06,
|
| 5104 |
+
"loss": 1.2053,
|
| 5105 |
+
"step": 728
|
| 5106 |
+
},
|
| 5107 |
+
{
|
| 5108 |
+
"epoch": 0.8,
|
| 5109 |
+
"grad_norm": 0.25973862409591675,
|
| 5110 |
+
"learning_rate": 4.488286066584464e-06,
|
| 5111 |
+
"loss": 1.2349,
|
| 5112 |
+
"step": 729
|
| 5113 |
+
},
|
| 5114 |
+
{
|
| 5115 |
+
"epoch": 0.8010973936899863,
|
| 5116 |
+
"grad_norm": 0.2638239860534668,
|
| 5117 |
+
"learning_rate": 4.463625154130703e-06,
|
| 5118 |
+
"loss": 1.2361,
|
| 5119 |
+
"step": 730
|
| 5120 |
+
},
|
| 5121 |
+
{
|
| 5122 |
+
"epoch": 0.8021947873799725,
|
| 5123 |
+
"grad_norm": 0.2783791124820709,
|
| 5124 |
+
"learning_rate": 4.438964241676942e-06,
|
| 5125 |
+
"loss": 1.2097,
|
| 5126 |
+
"step": 731
|
| 5127 |
+
},
|
| 5128 |
+
{
|
| 5129 |
+
"epoch": 0.8032921810699588,
|
| 5130 |
+
"grad_norm": 0.23621784150600433,
|
| 5131 |
+
"learning_rate": 4.414303329223182e-06,
|
| 5132 |
+
"loss": 1.3246,
|
| 5133 |
+
"step": 732
|
| 5134 |
+
},
|
| 5135 |
+
{
|
| 5136 |
+
"epoch": 0.8043895747599451,
|
| 5137 |
+
"grad_norm": 0.26742273569107056,
|
| 5138 |
+
"learning_rate": 4.389642416769421e-06,
|
| 5139 |
+
"loss": 1.2619,
|
| 5140 |
+
"step": 733
|
| 5141 |
+
},
|
| 5142 |
+
{
|
| 5143 |
+
"epoch": 0.8054869684499314,
|
| 5144 |
+
"grad_norm": 0.2638753354549408,
|
| 5145 |
+
"learning_rate": 4.36498150431566e-06,
|
| 5146 |
+
"loss": 1.2389,
|
| 5147 |
+
"step": 734
|
| 5148 |
+
},
|
| 5149 |
+
{
|
| 5150 |
+
"epoch": 0.8065843621399177,
|
| 5151 |
+
"grad_norm": 0.2494557499885559,
|
| 5152 |
+
"learning_rate": 4.340320591861899e-06,
|
| 5153 |
+
"loss": 1.2475,
|
| 5154 |
+
"step": 735
|
| 5155 |
+
},
|
| 5156 |
+
{
|
| 5157 |
+
"epoch": 0.807681755829904,
|
| 5158 |
+
"grad_norm": 0.23512186110019684,
|
| 5159 |
+
"learning_rate": 4.315659679408139e-06,
|
| 5160 |
+
"loss": 1.2393,
|
| 5161 |
+
"step": 736
|
| 5162 |
+
},
|
| 5163 |
+
{
|
| 5164 |
+
"epoch": 0.8087791495198903,
|
| 5165 |
+
"grad_norm": 0.25906118750572205,
|
| 5166 |
+
"learning_rate": 4.2909987669543775e-06,
|
| 5167 |
+
"loss": 1.205,
|
| 5168 |
+
"step": 737
|
| 5169 |
+
},
|
| 5170 |
+
{
|
| 5171 |
+
"epoch": 0.8098765432098766,
|
| 5172 |
+
"grad_norm": 0.33473852276802063,
|
| 5173 |
+
"learning_rate": 4.266337854500616e-06,
|
| 5174 |
+
"loss": 1.1913,
|
| 5175 |
+
"step": 738
|
| 5176 |
+
},
|
| 5177 |
+
{
|
| 5178 |
+
"epoch": 0.8109739368998629,
|
| 5179 |
+
"grad_norm": 0.2536832094192505,
|
| 5180 |
+
"learning_rate": 4.241676942046856e-06,
|
| 5181 |
+
"loss": 1.2255,
|
| 5182 |
+
"step": 739
|
| 5183 |
+
},
|
| 5184 |
+
{
|
| 5185 |
+
"epoch": 0.8120713305898491,
|
| 5186 |
+
"grad_norm": 0.2659871578216553,
|
| 5187 |
+
"learning_rate": 4.217016029593095e-06,
|
| 5188 |
+
"loss": 1.1904,
|
| 5189 |
+
"step": 740
|
| 5190 |
+
},
|
| 5191 |
+
{
|
| 5192 |
+
"epoch": 0.8131687242798354,
|
| 5193 |
+
"grad_norm": 0.26739829778671265,
|
| 5194 |
+
"learning_rate": 4.192355117139335e-06,
|
| 5195 |
+
"loss": 1.2491,
|
| 5196 |
+
"step": 741
|
| 5197 |
+
},
|
| 5198 |
+
{
|
| 5199 |
+
"epoch": 0.8142661179698216,
|
| 5200 |
+
"grad_norm": 0.2865275740623474,
|
| 5201 |
+
"learning_rate": 4.167694204685574e-06,
|
| 5202 |
+
"loss": 1.1686,
|
| 5203 |
+
"step": 742
|
| 5204 |
+
},
|
| 5205 |
+
{
|
| 5206 |
+
"epoch": 0.8153635116598079,
|
| 5207 |
+
"grad_norm": 0.27131780982017517,
|
| 5208 |
+
"learning_rate": 4.143033292231813e-06,
|
| 5209 |
+
"loss": 1.2858,
|
| 5210 |
+
"step": 743
|
| 5211 |
+
},
|
| 5212 |
+
{
|
| 5213 |
+
"epoch": 0.8164609053497942,
|
| 5214 |
+
"grad_norm": 0.23862424492835999,
|
| 5215 |
+
"learning_rate": 4.118372379778052e-06,
|
| 5216 |
+
"loss": 1.1968,
|
| 5217 |
+
"step": 744
|
| 5218 |
+
},
|
| 5219 |
+
{
|
| 5220 |
+
"epoch": 0.8175582990397805,
|
| 5221 |
+
"grad_norm": 0.26961636543273926,
|
| 5222 |
+
"learning_rate": 4.093711467324291e-06,
|
| 5223 |
+
"loss": 1.1896,
|
| 5224 |
+
"step": 745
|
| 5225 |
+
},
|
| 5226 |
+
{
|
| 5227 |
+
"epoch": 0.8186556927297668,
|
| 5228 |
+
"grad_norm": 0.2831185460090637,
|
| 5229 |
+
"learning_rate": 4.06905055487053e-06,
|
| 5230 |
+
"loss": 1.1653,
|
| 5231 |
+
"step": 746
|
| 5232 |
+
},
|
| 5233 |
+
{
|
| 5234 |
+
"epoch": 0.8197530864197531,
|
| 5235 |
+
"grad_norm": 0.2668111324310303,
|
| 5236 |
+
"learning_rate": 4.04438964241677e-06,
|
| 5237 |
+
"loss": 1.232,
|
| 5238 |
+
"step": 747
|
| 5239 |
+
},
|
| 5240 |
+
{
|
| 5241 |
+
"epoch": 0.8208504801097394,
|
| 5242 |
+
"grad_norm": 0.274718314409256,
|
| 5243 |
+
"learning_rate": 4.019728729963009e-06,
|
| 5244 |
+
"loss": 1.1473,
|
| 5245 |
+
"step": 748
|
| 5246 |
+
},
|
| 5247 |
+
{
|
| 5248 |
+
"epoch": 0.8219478737997257,
|
| 5249 |
+
"grad_norm": 0.24919269979000092,
|
| 5250 |
+
"learning_rate": 3.995067817509249e-06,
|
| 5251 |
+
"loss": 1.3063,
|
| 5252 |
+
"step": 749
|
| 5253 |
+
},
|
| 5254 |
+
{
|
| 5255 |
+
"epoch": 0.823045267489712,
|
| 5256 |
+
"grad_norm": 0.26878875494003296,
|
| 5257 |
+
"learning_rate": 3.9704069050554875e-06,
|
| 5258 |
+
"loss": 1.1749,
|
| 5259 |
+
"step": 750
|
| 5260 |
+
},
|
| 5261 |
+
{
|
| 5262 |
+
"epoch": 0.8241426611796983,
|
| 5263 |
+
"grad_norm": 0.2315559983253479,
|
| 5264 |
+
"learning_rate": 3.9457459926017264e-06,
|
| 5265 |
+
"loss": 1.2932,
|
| 5266 |
+
"step": 751
|
| 5267 |
+
},
|
| 5268 |
+
{
|
| 5269 |
+
"epoch": 0.8252400548696845,
|
| 5270 |
+
"grad_norm": 0.2781643569469452,
|
| 5271 |
+
"learning_rate": 3.921085080147966e-06,
|
| 5272 |
+
"loss": 1.2611,
|
| 5273 |
+
"step": 752
|
| 5274 |
+
},
|
| 5275 |
+
{
|
| 5276 |
+
"epoch": 0.8263374485596707,
|
| 5277 |
+
"grad_norm": 0.26966774463653564,
|
| 5278 |
+
"learning_rate": 3.896424167694205e-06,
|
| 5279 |
+
"loss": 1.2085,
|
| 5280 |
+
"step": 753
|
| 5281 |
+
},
|
| 5282 |
+
{
|
| 5283 |
+
"epoch": 0.827434842249657,
|
| 5284 |
+
"grad_norm": 0.2580644488334656,
|
| 5285 |
+
"learning_rate": 3.871763255240444e-06,
|
| 5286 |
+
"loss": 1.3157,
|
| 5287 |
+
"step": 754
|
| 5288 |
+
},
|
| 5289 |
+
{
|
| 5290 |
+
"epoch": 0.8285322359396433,
|
| 5291 |
+
"grad_norm": 0.241165891289711,
|
| 5292 |
+
"learning_rate": 3.847102342786683e-06,
|
| 5293 |
+
"loss": 1.237,
|
| 5294 |
+
"step": 755
|
| 5295 |
+
},
|
| 5296 |
+
{
|
| 5297 |
+
"epoch": 0.8296296296296296,
|
| 5298 |
+
"grad_norm": 0.24749965965747833,
|
| 5299 |
+
"learning_rate": 3.822441430332923e-06,
|
| 5300 |
+
"loss": 1.2067,
|
| 5301 |
+
"step": 756
|
| 5302 |
+
},
|
| 5303 |
+
{
|
| 5304 |
+
"epoch": 0.8307270233196159,
|
| 5305 |
+
"grad_norm": 0.26655301451683044,
|
| 5306 |
+
"learning_rate": 3.7977805178791616e-06,
|
| 5307 |
+
"loss": 1.206,
|
| 5308 |
+
"step": 757
|
| 5309 |
+
},
|
| 5310 |
+
{
|
| 5311 |
+
"epoch": 0.8318244170096022,
|
| 5312 |
+
"grad_norm": 0.24877706170082092,
|
| 5313 |
+
"learning_rate": 3.773119605425401e-06,
|
| 5314 |
+
"loss": 1.2571,
|
| 5315 |
+
"step": 758
|
| 5316 |
}
|
| 5317 |
],
|
| 5318 |
"logging_steps": 1,
|
|
|
|
| 5332 |
"attributes": {}
|
| 5333 |
}
|
| 5334 |
},
|
| 5335 |
+
"total_flos": 4.6050522467731046e+17,
|
| 5336 |
"train_batch_size": 4,
|
| 5337 |
"trial_name": null,
|
| 5338 |
"trial_params": null
|