Training in progress, step 1316, 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 125248064
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:dd41fc97630170141dc508ec07570c754fab8d996df8653022e6bf199c96a846
|
3 |
size 125248064
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 64220436
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5ee389341dc923054dc48c9489b23e0003f34ebfa04e4f787e56453d6f5b1e60
|
3 |
size 64220436
|
last-checkpoint/rng_state.pth
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 14244
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:83146f5398af193bd20f19e815a6a80c30ac74ff9e56871d64fc318266a6a143
|
3 |
size 14244
|
last-checkpoint/scheduler.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1064
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a1a6ae50029ebc12ca97684c200ec4fe6ed551d6b557bd2a588d20bd09d7084e
|
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,
|
@@ -6916,6 +6916,2309 @@
|
|
6916 |
"learning_rate": 1.4750723355605823e-05,
|
6917 |
"loss": 0.4075,
|
6918 |
"step": 987
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
6919 |
}
|
6920 |
],
|
6921 |
"logging_steps": 1,
|
@@ -6930,12 +9233,12 @@
|
|
6930 |
"should_evaluate": false,
|
6931 |
"should_log": false,
|
6932 |
"should_save": true,
|
6933 |
-
"should_training_stop":
|
6934 |
},
|
6935 |
"attributes": {}
|
6936 |
}
|
6937 |
},
|
6938 |
-
"total_flos": 1.
|
6939 |
"train_batch_size": 4,
|
6940 |
"trial_name": null,
|
6941 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.9937700585236926,
|
5 |
"eval_steps": 500,
|
6 |
+
"global_step": 1316,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
6916 |
"learning_rate": 1.4750723355605823e-05,
|
6917 |
"loss": 0.4075,
|
6918 |
"step": 987
|
6919 |
+
},
|
6920 |
+
{
|
6921 |
+
"epoch": 0.74608268831414,
|
6922 |
+
"grad_norm": 0.28927087783813477,
|
6923 |
+
"learning_rate": 1.4665848075973787e-05,
|
6924 |
+
"loss": 0.4336,
|
6925 |
+
"step": 988
|
6926 |
+
},
|
6927 |
+
{
|
6928 |
+
"epoch": 0.7468378327355106,
|
6929 |
+
"grad_norm": 0.3212716281414032,
|
6930 |
+
"learning_rate": 1.4581175699508982e-05,
|
6931 |
+
"loss": 0.4231,
|
6932 |
+
"step": 989
|
6933 |
+
},
|
6934 |
+
{
|
6935 |
+
"epoch": 0.7475929771568812,
|
6936 |
+
"grad_norm": 0.31144043803215027,
|
6937 |
+
"learning_rate": 1.4496706712434827e-05,
|
6938 |
+
"loss": 0.4016,
|
6939 |
+
"step": 990
|
6940 |
+
},
|
6941 |
+
{
|
6942 |
+
"epoch": 0.7483481215782518,
|
6943 |
+
"grad_norm": 0.2838822305202484,
|
6944 |
+
"learning_rate": 1.4412441599806864e-05,
|
6945 |
+
"loss": 0.3471,
|
6946 |
+
"step": 991
|
6947 |
+
},
|
6948 |
+
{
|
6949 |
+
"epoch": 0.7491032659996224,
|
6950 |
+
"grad_norm": 0.32325074076652527,
|
6951 |
+
"learning_rate": 1.4328380845509837e-05,
|
6952 |
+
"loss": 0.3989,
|
6953 |
+
"step": 992
|
6954 |
+
},
|
6955 |
+
{
|
6956 |
+
"epoch": 0.749858410420993,
|
6957 |
+
"grad_norm": 0.32019558548927307,
|
6958 |
+
"learning_rate": 1.4244524932255027e-05,
|
6959 |
+
"loss": 0.3604,
|
6960 |
+
"step": 993
|
6961 |
+
},
|
6962 |
+
{
|
6963 |
+
"epoch": 0.7506135548423636,
|
6964 |
+
"grad_norm": 0.3578037619590759,
|
6965 |
+
"learning_rate": 1.4160874341577446e-05,
|
6966 |
+
"loss": 0.4305,
|
6967 |
+
"step": 994
|
6968 |
+
},
|
6969 |
+
{
|
6970 |
+
"epoch": 0.7513686992637342,
|
6971 |
+
"grad_norm": 0.32821208238601685,
|
6972 |
+
"learning_rate": 1.4077429553832993e-05,
|
6973 |
+
"loss": 0.3459,
|
6974 |
+
"step": 995
|
6975 |
+
},
|
6976 |
+
{
|
6977 |
+
"epoch": 0.7521238436851048,
|
6978 |
+
"grad_norm": 0.32228267192840576,
|
6979 |
+
"learning_rate": 1.3994191048195827e-05,
|
6980 |
+
"loss": 0.3113,
|
6981 |
+
"step": 996
|
6982 |
+
},
|
6983 |
+
{
|
6984 |
+
"epoch": 0.7528789881064754,
|
6985 |
+
"grad_norm": 0.4038134217262268,
|
6986 |
+
"learning_rate": 1.391115930265554e-05,
|
6987 |
+
"loss": 0.4053,
|
6988 |
+
"step": 997
|
6989 |
+
},
|
6990 |
+
{
|
6991 |
+
"epoch": 0.753634132527846,
|
6992 |
+
"grad_norm": 0.36193010210990906,
|
6993 |
+
"learning_rate": 1.3828334794014381e-05,
|
6994 |
+
"loss": 0.3359,
|
6995 |
+
"step": 998
|
6996 |
+
},
|
6997 |
+
{
|
6998 |
+
"epoch": 0.7543892769492165,
|
6999 |
+
"grad_norm": 0.38052475452423096,
|
7000 |
+
"learning_rate": 1.37457179978846e-05,
|
7001 |
+
"loss": 0.3603,
|
7002 |
+
"step": 999
|
7003 |
+
},
|
7004 |
+
{
|
7005 |
+
"epoch": 0.7551444213705871,
|
7006 |
+
"grad_norm": 0.4941888153553009,
|
7007 |
+
"learning_rate": 1.366330938868569e-05,
|
7008 |
+
"loss": 0.4244,
|
7009 |
+
"step": 1000
|
7010 |
+
},
|
7011 |
+
{
|
7012 |
+
"epoch": 0.7558995657919577,
|
7013 |
+
"grad_norm": 0.15037553012371063,
|
7014 |
+
"learning_rate": 1.3581109439641588e-05,
|
7015 |
+
"loss": 0.3263,
|
7016 |
+
"step": 1001
|
7017 |
+
},
|
7018 |
+
{
|
7019 |
+
"epoch": 0.7566547102133283,
|
7020 |
+
"grad_norm": 0.14941640198230743,
|
7021 |
+
"learning_rate": 1.349911862277809e-05,
|
7022 |
+
"loss": 0.3179,
|
7023 |
+
"step": 1002
|
7024 |
+
},
|
7025 |
+
{
|
7026 |
+
"epoch": 0.7574098546346989,
|
7027 |
+
"grad_norm": 0.1701534390449524,
|
7028 |
+
"learning_rate": 1.3417337408920006e-05,
|
7029 |
+
"loss": 0.3566,
|
7030 |
+
"step": 1003
|
7031 |
+
},
|
7032 |
+
{
|
7033 |
+
"epoch": 0.7581649990560695,
|
7034 |
+
"grad_norm": 0.18360120058059692,
|
7035 |
+
"learning_rate": 1.3335766267688565e-05,
|
7036 |
+
"loss": 0.3882,
|
7037 |
+
"step": 1004
|
7038 |
+
},
|
7039 |
+
{
|
7040 |
+
"epoch": 0.7589201434774401,
|
7041 |
+
"grad_norm": 0.19040322303771973,
|
7042 |
+
"learning_rate": 1.3254405667498671e-05,
|
7043 |
+
"loss": 0.3905,
|
7044 |
+
"step": 1005
|
7045 |
+
},
|
7046 |
+
{
|
7047 |
+
"epoch": 0.7596752878988107,
|
7048 |
+
"grad_norm": 0.19472241401672363,
|
7049 |
+
"learning_rate": 1.3173256075556167e-05,
|
7050 |
+
"loss": 0.4387,
|
7051 |
+
"step": 1006
|
7052 |
+
},
|
7053 |
+
{
|
7054 |
+
"epoch": 0.7604304323201813,
|
7055 |
+
"grad_norm": 0.1870664656162262,
|
7056 |
+
"learning_rate": 1.309231795785526e-05,
|
7057 |
+
"loss": 0.4015,
|
7058 |
+
"step": 1007
|
7059 |
+
},
|
7060 |
+
{
|
7061 |
+
"epoch": 0.7611855767415519,
|
7062 |
+
"grad_norm": 0.18746329843997955,
|
7063 |
+
"learning_rate": 1.3011591779175775e-05,
|
7064 |
+
"loss": 0.4036,
|
7065 |
+
"step": 1008
|
7066 |
+
},
|
7067 |
+
{
|
7068 |
+
"epoch": 0.7619407211629224,
|
7069 |
+
"grad_norm": 0.1918107122182846,
|
7070 |
+
"learning_rate": 1.2931078003080437e-05,
|
7071 |
+
"loss": 0.3803,
|
7072 |
+
"step": 1009
|
7073 |
+
},
|
7074 |
+
{
|
7075 |
+
"epoch": 0.762695865584293,
|
7076 |
+
"grad_norm": 0.20022304356098175,
|
7077 |
+
"learning_rate": 1.2850777091912364e-05,
|
7078 |
+
"loss": 0.3884,
|
7079 |
+
"step": 1010
|
7080 |
+
},
|
7081 |
+
{
|
7082 |
+
"epoch": 0.7634510100056636,
|
7083 |
+
"grad_norm": 0.19047093391418457,
|
7084 |
+
"learning_rate": 1.2770689506792216e-05,
|
7085 |
+
"loss": 0.3729,
|
7086 |
+
"step": 1011
|
7087 |
+
},
|
7088 |
+
{
|
7089 |
+
"epoch": 0.7642061544270342,
|
7090 |
+
"grad_norm": 0.2117670625448227,
|
7091 |
+
"learning_rate": 1.2690815707615727e-05,
|
7092 |
+
"loss": 0.436,
|
7093 |
+
"step": 1012
|
7094 |
+
},
|
7095 |
+
{
|
7096 |
+
"epoch": 0.7649612988484048,
|
7097 |
+
"grad_norm": 0.20901797711849213,
|
7098 |
+
"learning_rate": 1.2611156153050963e-05,
|
7099 |
+
"loss": 0.3903,
|
7100 |
+
"step": 1013
|
7101 |
+
},
|
7102 |
+
{
|
7103 |
+
"epoch": 0.7657164432697754,
|
7104 |
+
"grad_norm": 0.2227080911397934,
|
7105 |
+
"learning_rate": 1.253171130053567e-05,
|
7106 |
+
"loss": 0.4772,
|
7107 |
+
"step": 1014
|
7108 |
+
},
|
7109 |
+
{
|
7110 |
+
"epoch": 0.766471587691146,
|
7111 |
+
"grad_norm": 0.21282415091991425,
|
7112 |
+
"learning_rate": 1.2452481606274734e-05,
|
7113 |
+
"loss": 0.4207,
|
7114 |
+
"step": 1015
|
7115 |
+
},
|
7116 |
+
{
|
7117 |
+
"epoch": 0.7672267321125166,
|
7118 |
+
"grad_norm": 0.21110861003398895,
|
7119 |
+
"learning_rate": 1.237346752523752e-05,
|
7120 |
+
"loss": 0.3875,
|
7121 |
+
"step": 1016
|
7122 |
+
},
|
7123 |
+
{
|
7124 |
+
"epoch": 0.767981876533887,
|
7125 |
+
"grad_norm": 0.23210763931274414,
|
7126 |
+
"learning_rate": 1.2294669511155193e-05,
|
7127 |
+
"loss": 0.439,
|
7128 |
+
"step": 1017
|
7129 |
+
},
|
7130 |
+
{
|
7131 |
+
"epoch": 0.7687370209552576,
|
7132 |
+
"grad_norm": 0.2336515635251999,
|
7133 |
+
"learning_rate": 1.2216088016518234e-05,
|
7134 |
+
"loss": 0.4468,
|
7135 |
+
"step": 1018
|
7136 |
+
},
|
7137 |
+
{
|
7138 |
+
"epoch": 0.7694921653766282,
|
7139 |
+
"grad_norm": 0.2305041402578354,
|
7140 |
+
"learning_rate": 1.2137723492573766e-05,
|
7141 |
+
"loss": 0.4231,
|
7142 |
+
"step": 1019
|
7143 |
+
},
|
7144 |
+
{
|
7145 |
+
"epoch": 0.7702473097979988,
|
7146 |
+
"grad_norm": 0.22647303342819214,
|
7147 |
+
"learning_rate": 1.2059576389322946e-05,
|
7148 |
+
"loss": 0.406,
|
7149 |
+
"step": 1020
|
7150 |
+
},
|
7151 |
+
{
|
7152 |
+
"epoch": 0.7710024542193694,
|
7153 |
+
"grad_norm": 0.2186606526374817,
|
7154 |
+
"learning_rate": 1.1981647155518472e-05,
|
7155 |
+
"loss": 0.3418,
|
7156 |
+
"step": 1021
|
7157 |
+
},
|
7158 |
+
{
|
7159 |
+
"epoch": 0.77175759864074,
|
7160 |
+
"grad_norm": 0.21919424831867218,
|
7161 |
+
"learning_rate": 1.1903936238661866e-05,
|
7162 |
+
"loss": 0.3603,
|
7163 |
+
"step": 1022
|
7164 |
+
},
|
7165 |
+
{
|
7166 |
+
"epoch": 0.7725127430621106,
|
7167 |
+
"grad_norm": 0.22741521894931793,
|
7168 |
+
"learning_rate": 1.1826444085001048e-05,
|
7169 |
+
"loss": 0.3957,
|
7170 |
+
"step": 1023
|
7171 |
+
},
|
7172 |
+
{
|
7173 |
+
"epoch": 0.7732678874834812,
|
7174 |
+
"grad_norm": 0.23761725425720215,
|
7175 |
+
"learning_rate": 1.174917113952772e-05,
|
7176 |
+
"loss": 0.3747,
|
7177 |
+
"step": 1024
|
7178 |
+
},
|
7179 |
+
{
|
7180 |
+
"epoch": 0.7740230319048518,
|
7181 |
+
"grad_norm": 0.23279255628585815,
|
7182 |
+
"learning_rate": 1.167211784597474e-05,
|
7183 |
+
"loss": 0.3978,
|
7184 |
+
"step": 1025
|
7185 |
+
},
|
7186 |
+
{
|
7187 |
+
"epoch": 0.7747781763262224,
|
7188 |
+
"grad_norm": 0.2562216520309448,
|
7189 |
+
"learning_rate": 1.1595284646813686e-05,
|
7190 |
+
"loss": 0.4132,
|
7191 |
+
"step": 1026
|
7192 |
+
},
|
7193 |
+
{
|
7194 |
+
"epoch": 0.775533320747593,
|
7195 |
+
"grad_norm": 0.23929475247859955,
|
7196 |
+
"learning_rate": 1.1518671983252228e-05,
|
7197 |
+
"loss": 0.3736,
|
7198 |
+
"step": 1027
|
7199 |
+
},
|
7200 |
+
{
|
7201 |
+
"epoch": 0.7762884651689635,
|
7202 |
+
"grad_norm": 0.2509770691394806,
|
7203 |
+
"learning_rate": 1.1442280295231656e-05,
|
7204 |
+
"loss": 0.4275,
|
7205 |
+
"step": 1028
|
7206 |
+
},
|
7207 |
+
{
|
7208 |
+
"epoch": 0.7770436095903341,
|
7209 |
+
"grad_norm": 0.2668601870536804,
|
7210 |
+
"learning_rate": 1.1366110021424337e-05,
|
7211 |
+
"loss": 0.406,
|
7212 |
+
"step": 1029
|
7213 |
+
},
|
7214 |
+
{
|
7215 |
+
"epoch": 0.7777987540117047,
|
7216 |
+
"grad_norm": 0.24542871117591858,
|
7217 |
+
"learning_rate": 1.1290161599231137e-05,
|
7218 |
+
"loss": 0.4133,
|
7219 |
+
"step": 1030
|
7220 |
+
},
|
7221 |
+
{
|
7222 |
+
"epoch": 0.7785538984330753,
|
7223 |
+
"grad_norm": 0.2547529935836792,
|
7224 |
+
"learning_rate": 1.1214435464779006e-05,
|
7225 |
+
"loss": 0.4081,
|
7226 |
+
"step": 1031
|
7227 |
+
},
|
7228 |
+
{
|
7229 |
+
"epoch": 0.7793090428544459,
|
7230 |
+
"grad_norm": 0.2759033441543579,
|
7231 |
+
"learning_rate": 1.1138932052918428e-05,
|
7232 |
+
"loss": 0.3879,
|
7233 |
+
"step": 1032
|
7234 |
+
},
|
7235 |
+
{
|
7236 |
+
"epoch": 0.7800641872758165,
|
7237 |
+
"grad_norm": 0.2765044867992401,
|
7238 |
+
"learning_rate": 1.106365179722088e-05,
|
7239 |
+
"loss": 0.4076,
|
7240 |
+
"step": 1033
|
7241 |
+
},
|
7242 |
+
{
|
7243 |
+
"epoch": 0.7808193316971871,
|
7244 |
+
"grad_norm": 0.27078738808631897,
|
7245 |
+
"learning_rate": 1.0988595129976442e-05,
|
7246 |
+
"loss": 0.3987,
|
7247 |
+
"step": 1034
|
7248 |
+
},
|
7249 |
+
{
|
7250 |
+
"epoch": 0.7815744761185577,
|
7251 |
+
"grad_norm": 0.27419719099998474,
|
7252 |
+
"learning_rate": 1.0913762482191198e-05,
|
7253 |
+
"loss": 0.3885,
|
7254 |
+
"step": 1035
|
7255 |
+
},
|
7256 |
+
{
|
7257 |
+
"epoch": 0.7823296205399283,
|
7258 |
+
"grad_norm": 0.26837533712387085,
|
7259 |
+
"learning_rate": 1.0839154283584868e-05,
|
7260 |
+
"loss": 0.3646,
|
7261 |
+
"step": 1036
|
7262 |
+
},
|
7263 |
+
{
|
7264 |
+
"epoch": 0.7830847649612989,
|
7265 |
+
"grad_norm": 0.30757594108581543,
|
7266 |
+
"learning_rate": 1.0764770962588277e-05,
|
7267 |
+
"loss": 0.4398,
|
7268 |
+
"step": 1037
|
7269 |
+
},
|
7270 |
+
{
|
7271 |
+
"epoch": 0.7838399093826695,
|
7272 |
+
"grad_norm": 0.28836846351623535,
|
7273 |
+
"learning_rate": 1.0690612946340894e-05,
|
7274 |
+
"loss": 0.4293,
|
7275 |
+
"step": 1038
|
7276 |
+
},
|
7277 |
+
{
|
7278 |
+
"epoch": 0.78459505380404,
|
7279 |
+
"grad_norm": 0.3034648597240448,
|
7280 |
+
"learning_rate": 1.0616680660688421e-05,
|
7281 |
+
"loss": 0.3929,
|
7282 |
+
"step": 1039
|
7283 |
+
},
|
7284 |
+
{
|
7285 |
+
"epoch": 0.7853501982254106,
|
7286 |
+
"grad_norm": 0.3083866834640503,
|
7287 |
+
"learning_rate": 1.0542974530180327e-05,
|
7288 |
+
"loss": 0.3692,
|
7289 |
+
"step": 1040
|
7290 |
+
},
|
7291 |
+
{
|
7292 |
+
"epoch": 0.7861053426467812,
|
7293 |
+
"grad_norm": 0.318313866853714,
|
7294 |
+
"learning_rate": 1.0469494978067357e-05,
|
7295 |
+
"loss": 0.381,
|
7296 |
+
"step": 1041
|
7297 |
+
},
|
7298 |
+
{
|
7299 |
+
"epoch": 0.7868604870681518,
|
7300 |
+
"grad_norm": 0.3297711908817291,
|
7301 |
+
"learning_rate": 1.0396242426299208e-05,
|
7302 |
+
"loss": 0.4136,
|
7303 |
+
"step": 1042
|
7304 |
+
},
|
7305 |
+
{
|
7306 |
+
"epoch": 0.7876156314895224,
|
7307 |
+
"grad_norm": 0.33657291531562805,
|
7308 |
+
"learning_rate": 1.0323217295522026e-05,
|
7309 |
+
"loss": 0.3967,
|
7310 |
+
"step": 1043
|
7311 |
+
},
|
7312 |
+
{
|
7313 |
+
"epoch": 0.788370775910893,
|
7314 |
+
"grad_norm": 0.3322165608406067,
|
7315 |
+
"learning_rate": 1.0250420005075995e-05,
|
7316 |
+
"loss": 0.362,
|
7317 |
+
"step": 1044
|
7318 |
+
},
|
7319 |
+
{
|
7320 |
+
"epoch": 0.7891259203322636,
|
7321 |
+
"grad_norm": 0.37014830112457275,
|
7322 |
+
"learning_rate": 1.017785097299298e-05,
|
7323 |
+
"loss": 0.4297,
|
7324 |
+
"step": 1045
|
7325 |
+
},
|
7326 |
+
{
|
7327 |
+
"epoch": 0.7898810647536342,
|
7328 |
+
"grad_norm": 0.33871424198150635,
|
7329 |
+
"learning_rate": 1.0105510615994051e-05,
|
7330 |
+
"loss": 0.3868,
|
7331 |
+
"step": 1046
|
7332 |
+
},
|
7333 |
+
{
|
7334 |
+
"epoch": 0.7906362091750048,
|
7335 |
+
"grad_norm": 0.4414469003677368,
|
7336 |
+
"learning_rate": 1.0033399349487171e-05,
|
7337 |
+
"loss": 0.2905,
|
7338 |
+
"step": 1047
|
7339 |
+
},
|
7340 |
+
{
|
7341 |
+
"epoch": 0.7913913535963754,
|
7342 |
+
"grad_norm": 0.42578282952308655,
|
7343 |
+
"learning_rate": 9.961517587564773e-06,
|
7344 |
+
"loss": 0.3776,
|
7345 |
+
"step": 1048
|
7346 |
+
},
|
7347 |
+
{
|
7348 |
+
"epoch": 0.792146498017746,
|
7349 |
+
"grad_norm": 0.45136335492134094,
|
7350 |
+
"learning_rate": 9.88986574300133e-06,
|
7351 |
+
"loss": 0.3735,
|
7352 |
+
"step": 1049
|
7353 |
+
},
|
7354 |
+
{
|
7355 |
+
"epoch": 0.7929016424391164,
|
7356 |
+
"grad_norm": 0.4850107729434967,
|
7357 |
+
"learning_rate": 9.818444227251089e-06,
|
7358 |
+
"loss": 0.3717,
|
7359 |
+
"step": 1050
|
7360 |
+
},
|
7361 |
+
{
|
7362 |
+
"epoch": 0.793656786860487,
|
7363 |
+
"grad_norm": 0.15895645320415497,
|
7364 |
+
"learning_rate": 9.747253450445649e-06,
|
7365 |
+
"loss": 0.3262,
|
7366 |
+
"step": 1051
|
7367 |
+
},
|
7368 |
+
{
|
7369 |
+
"epoch": 0.7944119312818576,
|
7370 |
+
"grad_norm": 0.15929242968559265,
|
7371 |
+
"learning_rate": 9.67629382139157e-06,
|
7372 |
+
"loss": 0.3419,
|
7373 |
+
"step": 1052
|
7374 |
+
},
|
7375 |
+
{
|
7376 |
+
"epoch": 0.7951670757032282,
|
7377 |
+
"grad_norm": 0.18021705746650696,
|
7378 |
+
"learning_rate": 9.605565747568107e-06,
|
7379 |
+
"loss": 0.3528,
|
7380 |
+
"step": 1053
|
7381 |
+
},
|
7382 |
+
{
|
7383 |
+
"epoch": 0.7959222201245988,
|
7384 |
+
"grad_norm": 0.16313131153583527,
|
7385 |
+
"learning_rate": 9.535069635124838e-06,
|
7386 |
+
"loss": 0.3082,
|
7387 |
+
"step": 1054
|
7388 |
+
},
|
7389 |
+
{
|
7390 |
+
"epoch": 0.7966773645459694,
|
7391 |
+
"grad_norm": 0.18114276230335236,
|
7392 |
+
"learning_rate": 9.464805888879263e-06,
|
7393 |
+
"loss": 0.3534,
|
7394 |
+
"step": 1055
|
7395 |
+
},
|
7396 |
+
{
|
7397 |
+
"epoch": 0.79743250896734,
|
7398 |
+
"grad_norm": 0.17138569056987762,
|
7399 |
+
"learning_rate": 9.394774912314624e-06,
|
7400 |
+
"loss": 0.3362,
|
7401 |
+
"step": 1056
|
7402 |
+
},
|
7403 |
+
{
|
7404 |
+
"epoch": 0.7981876533887106,
|
7405 |
+
"grad_norm": 0.18607540428638458,
|
7406 |
+
"learning_rate": 9.324977107577426e-06,
|
7407 |
+
"loss": 0.354,
|
7408 |
+
"step": 1057
|
7409 |
+
},
|
7410 |
+
{
|
7411 |
+
"epoch": 0.7989427978100812,
|
7412 |
+
"grad_norm": 0.19752830266952515,
|
7413 |
+
"learning_rate": 9.255412875475256e-06,
|
7414 |
+
"loss": 0.389,
|
7415 |
+
"step": 1058
|
7416 |
+
},
|
7417 |
+
{
|
7418 |
+
"epoch": 0.7996979422314517,
|
7419 |
+
"grad_norm": 0.20828400552272797,
|
7420 |
+
"learning_rate": 9.186082615474428e-06,
|
7421 |
+
"loss": 0.4721,
|
7422 |
+
"step": 1059
|
7423 |
+
},
|
7424 |
+
{
|
7425 |
+
"epoch": 0.8004530866528223,
|
7426 |
+
"grad_norm": 0.22494232654571533,
|
7427 |
+
"learning_rate": 9.116986725697657e-06,
|
7428 |
+
"loss": 0.4824,
|
7429 |
+
"step": 1060
|
7430 |
+
},
|
7431 |
+
{
|
7432 |
+
"epoch": 0.8012082310741929,
|
7433 |
+
"grad_norm": 0.19258789718151093,
|
7434 |
+
"learning_rate": 9.048125602921842e-06,
|
7435 |
+
"loss": 0.3916,
|
7436 |
+
"step": 1061
|
7437 |
+
},
|
7438 |
+
{
|
7439 |
+
"epoch": 0.8019633754955635,
|
7440 |
+
"grad_norm": 0.19615770876407623,
|
7441 |
+
"learning_rate": 8.979499642575762e-06,
|
7442 |
+
"loss": 0.3399,
|
7443 |
+
"step": 1062
|
7444 |
+
},
|
7445 |
+
{
|
7446 |
+
"epoch": 0.8027185199169341,
|
7447 |
+
"grad_norm": 0.19866204261779785,
|
7448 |
+
"learning_rate": 8.911109238737747e-06,
|
7449 |
+
"loss": 0.3476,
|
7450 |
+
"step": 1063
|
7451 |
+
},
|
7452 |
+
{
|
7453 |
+
"epoch": 0.8034736643383047,
|
7454 |
+
"grad_norm": 0.23535782098770142,
|
7455 |
+
"learning_rate": 8.842954784133516e-06,
|
7456 |
+
"loss": 0.469,
|
7457 |
+
"step": 1064
|
7458 |
+
},
|
7459 |
+
{
|
7460 |
+
"epoch": 0.8042288087596753,
|
7461 |
+
"grad_norm": 0.20158474147319794,
|
7462 |
+
"learning_rate": 8.775036670133857e-06,
|
7463 |
+
"loss": 0.3352,
|
7464 |
+
"step": 1065
|
7465 |
+
},
|
7466 |
+
{
|
7467 |
+
"epoch": 0.8049839531810459,
|
7468 |
+
"grad_norm": 0.1965356171131134,
|
7469 |
+
"learning_rate": 8.707355286752366e-06,
|
7470 |
+
"loss": 0.3574,
|
7471 |
+
"step": 1066
|
7472 |
+
},
|
7473 |
+
{
|
7474 |
+
"epoch": 0.8057390976024165,
|
7475 |
+
"grad_norm": 0.2335047870874405,
|
7476 |
+
"learning_rate": 8.639911022643288e-06,
|
7477 |
+
"loss": 0.4357,
|
7478 |
+
"step": 1067
|
7479 |
+
},
|
7480 |
+
{
|
7481 |
+
"epoch": 0.8064942420237871,
|
7482 |
+
"grad_norm": 0.22290942072868347,
|
7483 |
+
"learning_rate": 8.572704265099157e-06,
|
7484 |
+
"loss": 0.4266,
|
7485 |
+
"step": 1068
|
7486 |
+
},
|
7487 |
+
{
|
7488 |
+
"epoch": 0.8072493864451576,
|
7489 |
+
"grad_norm": 0.23795855045318604,
|
7490 |
+
"learning_rate": 8.505735400048748e-06,
|
7491 |
+
"loss": 0.4148,
|
7492 |
+
"step": 1069
|
7493 |
+
},
|
7494 |
+
{
|
7495 |
+
"epoch": 0.8080045308665282,
|
7496 |
+
"grad_norm": 0.228322833776474,
|
7497 |
+
"learning_rate": 8.439004812054657e-06,
|
7498 |
+
"loss": 0.3988,
|
7499 |
+
"step": 1070
|
7500 |
+
},
|
7501 |
+
{
|
7502 |
+
"epoch": 0.8087596752878988,
|
7503 |
+
"grad_norm": 0.23256799578666687,
|
7504 |
+
"learning_rate": 8.372512884311256e-06,
|
7505 |
+
"loss": 0.4268,
|
7506 |
+
"step": 1071
|
7507 |
+
},
|
7508 |
+
{
|
7509 |
+
"epoch": 0.8095148197092694,
|
7510 |
+
"grad_norm": 0.21915295720100403,
|
7511 |
+
"learning_rate": 8.306259998642441e-06,
|
7512 |
+
"loss": 0.3493,
|
7513 |
+
"step": 1072
|
7514 |
+
},
|
7515 |
+
{
|
7516 |
+
"epoch": 0.81026996413064,
|
7517 |
+
"grad_norm": 0.2517222464084625,
|
7518 |
+
"learning_rate": 8.24024653549937e-06,
|
7519 |
+
"loss": 0.4839,
|
7520 |
+
"step": 1073
|
7521 |
+
},
|
7522 |
+
{
|
7523 |
+
"epoch": 0.8110251085520106,
|
7524 |
+
"grad_norm": 0.24079398810863495,
|
7525 |
+
"learning_rate": 8.17447287395839e-06,
|
7526 |
+
"loss": 0.3799,
|
7527 |
+
"step": 1074
|
7528 |
+
},
|
7529 |
+
{
|
7530 |
+
"epoch": 0.8117802529733812,
|
7531 |
+
"grad_norm": 0.2544167935848236,
|
7532 |
+
"learning_rate": 8.108939391718812e-06,
|
7533 |
+
"loss": 0.3979,
|
7534 |
+
"step": 1075
|
7535 |
+
},
|
7536 |
+
{
|
7537 |
+
"epoch": 0.8125353973947518,
|
7538 |
+
"grad_norm": 0.2537594735622406,
|
7539 |
+
"learning_rate": 8.043646465100697e-06,
|
7540 |
+
"loss": 0.425,
|
7541 |
+
"step": 1076
|
7542 |
+
},
|
7543 |
+
{
|
7544 |
+
"epoch": 0.8132905418161224,
|
7545 |
+
"grad_norm": 0.2539171278476715,
|
7546 |
+
"learning_rate": 7.978594469042772e-06,
|
7547 |
+
"loss": 0.426,
|
7548 |
+
"step": 1077
|
7549 |
+
},
|
7550 |
+
{
|
7551 |
+
"epoch": 0.814045686237493,
|
7552 |
+
"grad_norm": 0.2699001729488373,
|
7553 |
+
"learning_rate": 7.913783777100253e-06,
|
7554 |
+
"loss": 0.414,
|
7555 |
+
"step": 1078
|
7556 |
+
},
|
7557 |
+
{
|
7558 |
+
"epoch": 0.8148008306588636,
|
7559 |
+
"grad_norm": 0.2574417293071747,
|
7560 |
+
"learning_rate": 7.849214761442636e-06,
|
7561 |
+
"loss": 0.4067,
|
7562 |
+
"step": 1079
|
7563 |
+
},
|
7564 |
+
{
|
7565 |
+
"epoch": 0.8155559750802341,
|
7566 |
+
"grad_norm": 0.2690460979938507,
|
7567 |
+
"learning_rate": 7.78488779285167e-06,
|
7568 |
+
"loss": 0.4546,
|
7569 |
+
"step": 1080
|
7570 |
+
},
|
7571 |
+
{
|
7572 |
+
"epoch": 0.8163111195016047,
|
7573 |
+
"grad_norm": 0.2663021683692932,
|
7574 |
+
"learning_rate": 7.72080324071912e-06,
|
7575 |
+
"loss": 0.3783,
|
7576 |
+
"step": 1081
|
7577 |
+
},
|
7578 |
+
{
|
7579 |
+
"epoch": 0.8170662639229753,
|
7580 |
+
"grad_norm": 0.24762804806232452,
|
7581 |
+
"learning_rate": 7.656961473044743e-06,
|
7582 |
+
"loss": 0.3468,
|
7583 |
+
"step": 1082
|
7584 |
+
},
|
7585 |
+
{
|
7586 |
+
"epoch": 0.8178214083443458,
|
7587 |
+
"grad_norm": 0.26688170433044434,
|
7588 |
+
"learning_rate": 7.593362856434116e-06,
|
7589 |
+
"loss": 0.3368,
|
7590 |
+
"step": 1083
|
7591 |
+
},
|
7592 |
+
{
|
7593 |
+
"epoch": 0.8185765527657164,
|
7594 |
+
"grad_norm": 0.27109459042549133,
|
7595 |
+
"learning_rate": 7.530007756096513e-06,
|
7596 |
+
"loss": 0.3406,
|
7597 |
+
"step": 1084
|
7598 |
+
},
|
7599 |
+
{
|
7600 |
+
"epoch": 0.819331697187087,
|
7601 |
+
"grad_norm": 0.28974488377571106,
|
7602 |
+
"learning_rate": 7.466896535842866e-06,
|
7603 |
+
"loss": 0.3889,
|
7604 |
+
"step": 1085
|
7605 |
+
},
|
7606 |
+
{
|
7607 |
+
"epoch": 0.8200868416084576,
|
7608 |
+
"grad_norm": 0.30808672308921814,
|
7609 |
+
"learning_rate": 7.404029558083653e-06,
|
7610 |
+
"loss": 0.418,
|
7611 |
+
"step": 1086
|
7612 |
+
},
|
7613 |
+
{
|
7614 |
+
"epoch": 0.8208419860298282,
|
7615 |
+
"grad_norm": 0.2833620309829712,
|
7616 |
+
"learning_rate": 7.341407183826776e-06,
|
7617 |
+
"loss": 0.3779,
|
7618 |
+
"step": 1087
|
7619 |
+
},
|
7620 |
+
{
|
7621 |
+
"epoch": 0.8215971304511988,
|
7622 |
+
"grad_norm": 0.2893945872783661,
|
7623 |
+
"learning_rate": 7.2790297726755716e-06,
|
7624 |
+
"loss": 0.3697,
|
7625 |
+
"step": 1088
|
7626 |
+
},
|
7627 |
+
{
|
7628 |
+
"epoch": 0.8223522748725693,
|
7629 |
+
"grad_norm": 0.3081832230091095,
|
7630 |
+
"learning_rate": 7.216897682826668e-06,
|
7631 |
+
"loss": 0.4526,
|
7632 |
+
"step": 1089
|
7633 |
+
},
|
7634 |
+
{
|
7635 |
+
"epoch": 0.8231074192939399,
|
7636 |
+
"grad_norm": 0.3169601261615753,
|
7637 |
+
"learning_rate": 7.1550112710679515e-06,
|
7638 |
+
"loss": 0.3671,
|
7639 |
+
"step": 1090
|
7640 |
+
},
|
7641 |
+
{
|
7642 |
+
"epoch": 0.8238625637153105,
|
7643 |
+
"grad_norm": 0.29276758432388306,
|
7644 |
+
"learning_rate": 7.093370892776558e-06,
|
7645 |
+
"loss": 0.2938,
|
7646 |
+
"step": 1091
|
7647 |
+
},
|
7648 |
+
{
|
7649 |
+
"epoch": 0.8246177081366811,
|
7650 |
+
"grad_norm": 0.31486350297927856,
|
7651 |
+
"learning_rate": 7.031976901916754e-06,
|
7652 |
+
"loss": 0.3838,
|
7653 |
+
"step": 1092
|
7654 |
+
},
|
7655 |
+
{
|
7656 |
+
"epoch": 0.8253728525580517,
|
7657 |
+
"grad_norm": 0.3386092782020569,
|
7658 |
+
"learning_rate": 6.9708296510379804e-06,
|
7659 |
+
"loss": 0.3715,
|
7660 |
+
"step": 1093
|
7661 |
+
},
|
7662 |
+
{
|
7663 |
+
"epoch": 0.8261279969794223,
|
7664 |
+
"grad_norm": 0.3289138376712799,
|
7665 |
+
"learning_rate": 6.909929491272799e-06,
|
7666 |
+
"loss": 0.3238,
|
7667 |
+
"step": 1094
|
7668 |
+
},
|
7669 |
+
{
|
7670 |
+
"epoch": 0.8268831414007929,
|
7671 |
+
"grad_norm": 0.3308430016040802,
|
7672 |
+
"learning_rate": 6.8492767723348396e-06,
|
7673 |
+
"loss": 0.3407,
|
7674 |
+
"step": 1095
|
7675 |
+
},
|
7676 |
+
{
|
7677 |
+
"epoch": 0.8276382858221635,
|
7678 |
+
"grad_norm": 0.32838174700737,
|
7679 |
+
"learning_rate": 6.788871842516864e-06,
|
7680 |
+
"loss": 0.3113,
|
7681 |
+
"step": 1096
|
7682 |
+
},
|
7683 |
+
{
|
7684 |
+
"epoch": 0.8283934302435341,
|
7685 |
+
"grad_norm": 0.37880030274391174,
|
7686 |
+
"learning_rate": 6.728715048688711e-06,
|
7687 |
+
"loss": 0.3602,
|
7688 |
+
"step": 1097
|
7689 |
+
},
|
7690 |
+
{
|
7691 |
+
"epoch": 0.8291485746649047,
|
7692 |
+
"grad_norm": 0.4009070098400116,
|
7693 |
+
"learning_rate": 6.668806736295307e-06,
|
7694 |
+
"loss": 0.42,
|
7695 |
+
"step": 1098
|
7696 |
+
},
|
7697 |
+
{
|
7698 |
+
"epoch": 0.8299037190862752,
|
7699 |
+
"grad_norm": 0.4191438853740692,
|
7700 |
+
"learning_rate": 6.6091472493547354e-06,
|
7701 |
+
"loss": 0.2983,
|
7702 |
+
"step": 1099
|
7703 |
+
},
|
7704 |
+
{
|
7705 |
+
"epoch": 0.8306588635076458,
|
7706 |
+
"grad_norm": 0.47517555952072144,
|
7707 |
+
"learning_rate": 6.549736930456163e-06,
|
7708 |
+
"loss": 0.3691,
|
7709 |
+
"step": 1100
|
7710 |
+
},
|
7711 |
+
{
|
7712 |
+
"epoch": 0.8314140079290164,
|
7713 |
+
"grad_norm": 0.1712975651025772,
|
7714 |
+
"learning_rate": 6.490576120757985e-06,
|
7715 |
+
"loss": 0.3235,
|
7716 |
+
"step": 1101
|
7717 |
+
},
|
7718 |
+
{
|
7719 |
+
"epoch": 0.832169152350387,
|
7720 |
+
"grad_norm": 0.16649426519870758,
|
7721 |
+
"learning_rate": 6.4316651599858e-06,
|
7722 |
+
"loss": 0.3349,
|
7723 |
+
"step": 1102
|
7724 |
+
},
|
7725 |
+
{
|
7726 |
+
"epoch": 0.8329242967717576,
|
7727 |
+
"grad_norm": 0.17389081418514252,
|
7728 |
+
"learning_rate": 6.373004386430442e-06,
|
7729 |
+
"loss": 0.3611,
|
7730 |
+
"step": 1103
|
7731 |
+
},
|
7732 |
+
{
|
7733 |
+
"epoch": 0.8336794411931282,
|
7734 |
+
"grad_norm": 0.16818854212760925,
|
7735 |
+
"learning_rate": 6.314594136946106e-06,
|
7736 |
+
"loss": 0.3444,
|
7737 |
+
"step": 1104
|
7738 |
+
},
|
7739 |
+
{
|
7740 |
+
"epoch": 0.8344345856144988,
|
7741 |
+
"grad_norm": 0.1820095032453537,
|
7742 |
+
"learning_rate": 6.25643474694837e-06,
|
7743 |
+
"loss": 0.3664,
|
7744 |
+
"step": 1105
|
7745 |
+
},
|
7746 |
+
{
|
7747 |
+
"epoch": 0.8351897300358694,
|
7748 |
+
"grad_norm": 0.18415705859661102,
|
7749 |
+
"learning_rate": 6.1985265504122314e-06,
|
7750 |
+
"loss": 0.3634,
|
7751 |
+
"step": 1106
|
7752 |
+
},
|
7753 |
+
{
|
7754 |
+
"epoch": 0.83594487445724,
|
7755 |
+
"grad_norm": 0.18334518373012543,
|
7756 |
+
"learning_rate": 6.140869879870287e-06,
|
7757 |
+
"loss": 0.3764,
|
7758 |
+
"step": 1107
|
7759 |
+
},
|
7760 |
+
{
|
7761 |
+
"epoch": 0.8367000188786106,
|
7762 |
+
"grad_norm": 0.19977064430713654,
|
7763 |
+
"learning_rate": 6.083465066410754e-06,
|
7764 |
+
"loss": 0.382,
|
7765 |
+
"step": 1108
|
7766 |
+
},
|
7767 |
+
{
|
7768 |
+
"epoch": 0.8374551632999812,
|
7769 |
+
"grad_norm": 0.20824985206127167,
|
7770 |
+
"learning_rate": 6.026312439675552e-06,
|
7771 |
+
"loss": 0.42,
|
7772 |
+
"step": 1109
|
7773 |
+
},
|
7774 |
+
{
|
7775 |
+
"epoch": 0.8382103077213517,
|
7776 |
+
"grad_norm": 0.20618076622486115,
|
7777 |
+
"learning_rate": 5.969412327858481e-06,
|
7778 |
+
"loss": 0.4066,
|
7779 |
+
"step": 1110
|
7780 |
+
},
|
7781 |
+
{
|
7782 |
+
"epoch": 0.8389654521427223,
|
7783 |
+
"grad_norm": 0.22045300900936127,
|
7784 |
+
"learning_rate": 5.912765057703262e-06,
|
7785 |
+
"loss": 0.4721,
|
7786 |
+
"step": 1111
|
7787 |
+
},
|
7788 |
+
{
|
7789 |
+
"epoch": 0.8397205965640929,
|
7790 |
+
"grad_norm": 0.20817910134792328,
|
7791 |
+
"learning_rate": 5.856370954501722e-06,
|
7792 |
+
"loss": 0.3477,
|
7793 |
+
"step": 1112
|
7794 |
+
},
|
7795 |
+
{
|
7796 |
+
"epoch": 0.8404757409854635,
|
7797 |
+
"grad_norm": 0.22744186222553253,
|
7798 |
+
"learning_rate": 5.8002303420919e-06,
|
7799 |
+
"loss": 0.409,
|
7800 |
+
"step": 1113
|
7801 |
+
},
|
7802 |
+
{
|
7803 |
+
"epoch": 0.8412308854068341,
|
7804 |
+
"grad_norm": 0.22021333873271942,
|
7805 |
+
"learning_rate": 5.744343542856162e-06,
|
7806 |
+
"loss": 0.422,
|
7807 |
+
"step": 1114
|
7808 |
+
},
|
7809 |
+
{
|
7810 |
+
"epoch": 0.8419860298282047,
|
7811 |
+
"grad_norm": 0.20742209255695343,
|
7812 |
+
"learning_rate": 5.688710877719416e-06,
|
7813 |
+
"loss": 0.4019,
|
7814 |
+
"step": 1115
|
7815 |
+
},
|
7816 |
+
{
|
7817 |
+
"epoch": 0.8427411742495753,
|
7818 |
+
"grad_norm": 0.21647979319095612,
|
7819 |
+
"learning_rate": 5.633332666147173e-06,
|
7820 |
+
"loss": 0.3796,
|
7821 |
+
"step": 1116
|
7822 |
+
},
|
7823 |
+
{
|
7824 |
+
"epoch": 0.8434963186709458,
|
7825 |
+
"grad_norm": 0.2143869549036026,
|
7826 |
+
"learning_rate": 5.5782092261438105e-06,
|
7827 |
+
"loss": 0.335,
|
7828 |
+
"step": 1117
|
7829 |
+
},
|
7830 |
+
{
|
7831 |
+
"epoch": 0.8442514630923164,
|
7832 |
+
"grad_norm": 0.2301177829504013,
|
7833 |
+
"learning_rate": 5.523340874250704e-06,
|
7834 |
+
"loss": 0.4501,
|
7835 |
+
"step": 1118
|
7836 |
+
},
|
7837 |
+
{
|
7838 |
+
"epoch": 0.845006607513687,
|
7839 |
+
"grad_norm": 0.23003332316875458,
|
7840 |
+
"learning_rate": 5.4687279255443705e-06,
|
7841 |
+
"loss": 0.4533,
|
7842 |
+
"step": 1119
|
7843 |
+
},
|
7844 |
+
{
|
7845 |
+
"epoch": 0.8457617519350575,
|
7846 |
+
"grad_norm": 0.22590281069278717,
|
7847 |
+
"learning_rate": 5.41437069363473e-06,
|
7848 |
+
"loss": 0.4036,
|
7849 |
+
"step": 1120
|
7850 |
+
},
|
7851 |
+
{
|
7852 |
+
"epoch": 0.8465168963564281,
|
7853 |
+
"grad_norm": 0.226993128657341,
|
7854 |
+
"learning_rate": 5.360269490663278e-06,
|
7855 |
+
"loss": 0.4468,
|
7856 |
+
"step": 1121
|
7857 |
+
},
|
7858 |
+
{
|
7859 |
+
"epoch": 0.8472720407777987,
|
7860 |
+
"grad_norm": 0.23445524275302887,
|
7861 |
+
"learning_rate": 5.306424627301248e-06,
|
7862 |
+
"loss": 0.384,
|
7863 |
+
"step": 1122
|
7864 |
+
},
|
7865 |
+
{
|
7866 |
+
"epoch": 0.8480271851991693,
|
7867 |
+
"grad_norm": 0.23912809789180756,
|
7868 |
+
"learning_rate": 5.252836412747919e-06,
|
7869 |
+
"loss": 0.4073,
|
7870 |
+
"step": 1123
|
7871 |
+
},
|
7872 |
+
{
|
7873 |
+
"epoch": 0.8487823296205399,
|
7874 |
+
"grad_norm": 0.24784360826015472,
|
7875 |
+
"learning_rate": 5.1995051547287285e-06,
|
7876 |
+
"loss": 0.407,
|
7877 |
+
"step": 1124
|
7878 |
+
},
|
7879 |
+
{
|
7880 |
+
"epoch": 0.8495374740419105,
|
7881 |
+
"grad_norm": 0.24575859308242798,
|
7882 |
+
"learning_rate": 5.14643115949362e-06,
|
7883 |
+
"loss": 0.3848,
|
7884 |
+
"step": 1125
|
7885 |
+
},
|
7886 |
+
{
|
7887 |
+
"epoch": 0.8502926184632811,
|
7888 |
+
"grad_norm": 0.25169506669044495,
|
7889 |
+
"learning_rate": 5.0936147318152e-06,
|
7890 |
+
"loss": 0.3523,
|
7891 |
+
"step": 1126
|
7892 |
+
},
|
7893 |
+
{
|
7894 |
+
"epoch": 0.8510477628846517,
|
7895 |
+
"grad_norm": 0.2474677860736847,
|
7896 |
+
"learning_rate": 5.041056174987008e-06,
|
7897 |
+
"loss": 0.4185,
|
7898 |
+
"step": 1127
|
7899 |
+
},
|
7900 |
+
{
|
7901 |
+
"epoch": 0.8518029073060223,
|
7902 |
+
"grad_norm": 0.2616688907146454,
|
7903 |
+
"learning_rate": 4.988755790821803e-06,
|
7904 |
+
"loss": 0.4394,
|
7905 |
+
"step": 1128
|
7906 |
+
},
|
7907 |
+
{
|
7908 |
+
"epoch": 0.8525580517273929,
|
7909 |
+
"grad_norm": 0.2649298310279846,
|
7910 |
+
"learning_rate": 4.936713879649818e-06,
|
7911 |
+
"loss": 0.4319,
|
7912 |
+
"step": 1129
|
7913 |
+
},
|
7914 |
+
{
|
7915 |
+
"epoch": 0.8533131961487634,
|
7916 |
+
"grad_norm": 0.2544507086277008,
|
7917 |
+
"learning_rate": 4.88493074031699e-06,
|
7918 |
+
"loss": 0.3631,
|
7919 |
+
"step": 1130
|
7920 |
+
},
|
7921 |
+
{
|
7922 |
+
"epoch": 0.854068340570134,
|
7923 |
+
"grad_norm": 0.24985194206237793,
|
7924 |
+
"learning_rate": 4.833406670183327e-06,
|
7925 |
+
"loss": 0.3581,
|
7926 |
+
"step": 1131
|
7927 |
+
},
|
7928 |
+
{
|
7929 |
+
"epoch": 0.8548234849915046,
|
7930 |
+
"grad_norm": 0.2733616828918457,
|
7931 |
+
"learning_rate": 4.782141965121128e-06,
|
7932 |
+
"loss": 0.4217,
|
7933 |
+
"step": 1132
|
7934 |
+
},
|
7935 |
+
{
|
7936 |
+
"epoch": 0.8555786294128752,
|
7937 |
+
"grad_norm": 0.2666637599468231,
|
7938 |
+
"learning_rate": 4.731136919513312e-06,
|
7939 |
+
"loss": 0.3794,
|
7940 |
+
"step": 1133
|
7941 |
+
},
|
7942 |
+
{
|
7943 |
+
"epoch": 0.8563337738342458,
|
7944 |
+
"grad_norm": 0.2879069149494171,
|
7945 |
+
"learning_rate": 4.680391826251751e-06,
|
7946 |
+
"loss": 0.35,
|
7947 |
+
"step": 1134
|
7948 |
+
},
|
7949 |
+
{
|
7950 |
+
"epoch": 0.8570889182556164,
|
7951 |
+
"grad_norm": 0.30357176065444946,
|
7952 |
+
"learning_rate": 4.629906976735526e-06,
|
7953 |
+
"loss": 0.4039,
|
7954 |
+
"step": 1135
|
7955 |
+
},
|
7956 |
+
{
|
7957 |
+
"epoch": 0.857844062676987,
|
7958 |
+
"grad_norm": 0.30282896757125854,
|
7959 |
+
"learning_rate": 4.5796826608693274e-06,
|
7960 |
+
"loss": 0.4026,
|
7961 |
+
"step": 1136
|
7962 |
+
},
|
7963 |
+
{
|
7964 |
+
"epoch": 0.8585992070983576,
|
7965 |
+
"grad_norm": 0.27673810720443726,
|
7966 |
+
"learning_rate": 4.529719167061741e-06,
|
7967 |
+
"loss": 0.3365,
|
7968 |
+
"step": 1137
|
7969 |
+
},
|
7970 |
+
{
|
7971 |
+
"epoch": 0.8593543515197282,
|
7972 |
+
"grad_norm": 0.28368133306503296,
|
7973 |
+
"learning_rate": 4.480016782223589e-06,
|
7974 |
+
"loss": 0.3572,
|
7975 |
+
"step": 1138
|
7976 |
+
},
|
7977 |
+
{
|
7978 |
+
"epoch": 0.8601094959410988,
|
7979 |
+
"grad_norm": 0.2899417281150818,
|
7980 |
+
"learning_rate": 4.430575791766328e-06,
|
7981 |
+
"loss": 0.3424,
|
7982 |
+
"step": 1139
|
7983 |
+
},
|
7984 |
+
{
|
7985 |
+
"epoch": 0.8608646403624693,
|
7986 |
+
"grad_norm": 0.328281432390213,
|
7987 |
+
"learning_rate": 4.381396479600375e-06,
|
7988 |
+
"loss": 0.3891,
|
7989 |
+
"step": 1140
|
7990 |
+
},
|
7991 |
+
{
|
7992 |
+
"epoch": 0.8616197847838399,
|
7993 |
+
"grad_norm": 0.3324677348136902,
|
7994 |
+
"learning_rate": 4.3324791281334595e-06,
|
7995 |
+
"loss": 0.4318,
|
7996 |
+
"step": 1141
|
7997 |
+
},
|
7998 |
+
{
|
7999 |
+
"epoch": 0.8623749292052105,
|
8000 |
+
"grad_norm": 0.30871015787124634,
|
8001 |
+
"learning_rate": 4.283824018269045e-06,
|
8002 |
+
"loss": 0.3513,
|
8003 |
+
"step": 1142
|
8004 |
+
},
|
8005 |
+
{
|
8006 |
+
"epoch": 0.8631300736265811,
|
8007 |
+
"grad_norm": 0.3389221131801605,
|
8008 |
+
"learning_rate": 4.235431429404707e-06,
|
8009 |
+
"loss": 0.4268,
|
8010 |
+
"step": 1143
|
8011 |
+
},
|
8012 |
+
{
|
8013 |
+
"epoch": 0.8638852180479517,
|
8014 |
+
"grad_norm": 0.30663684010505676,
|
8015 |
+
"learning_rate": 4.187301639430485e-06,
|
8016 |
+
"loss": 0.3358,
|
8017 |
+
"step": 1144
|
8018 |
+
},
|
8019 |
+
{
|
8020 |
+
"epoch": 0.8646403624693223,
|
8021 |
+
"grad_norm": 0.3269205093383789,
|
8022 |
+
"learning_rate": 4.139434924727359e-06,
|
8023 |
+
"loss": 0.3542,
|
8024 |
+
"step": 1145
|
8025 |
+
},
|
8026 |
+
{
|
8027 |
+
"epoch": 0.8653955068906929,
|
8028 |
+
"grad_norm": 0.3500070571899414,
|
8029 |
+
"learning_rate": 4.091831560165577e-06,
|
8030 |
+
"loss": 0.3445,
|
8031 |
+
"step": 1146
|
8032 |
+
},
|
8033 |
+
{
|
8034 |
+
"epoch": 0.8661506513120635,
|
8035 |
+
"grad_norm": 0.3916202783584595,
|
8036 |
+
"learning_rate": 4.044491819103169e-06,
|
8037 |
+
"loss": 0.4275,
|
8038 |
+
"step": 1147
|
8039 |
+
},
|
8040 |
+
{
|
8041 |
+
"epoch": 0.8669057957334341,
|
8042 |
+
"grad_norm": 0.41067856550216675,
|
8043 |
+
"learning_rate": 3.997415973384311e-06,
|
8044 |
+
"loss": 0.4219,
|
8045 |
+
"step": 1148
|
8046 |
+
},
|
8047 |
+
{
|
8048 |
+
"epoch": 0.8676609401548047,
|
8049 |
+
"grad_norm": 0.418266624212265,
|
8050 |
+
"learning_rate": 3.950604293337773e-06,
|
8051 |
+
"loss": 0.4184,
|
8052 |
+
"step": 1149
|
8053 |
+
},
|
8054 |
+
{
|
8055 |
+
"epoch": 0.8684160845761751,
|
8056 |
+
"grad_norm": 0.4629250168800354,
|
8057 |
+
"learning_rate": 3.904057047775389e-06,
|
8058 |
+
"loss": 0.3859,
|
8059 |
+
"step": 1150
|
8060 |
+
},
|
8061 |
+
{
|
8062 |
+
"epoch": 0.8691712289975457,
|
8063 |
+
"grad_norm": 0.1396147459745407,
|
8064 |
+
"learning_rate": 3.857774503990514e-06,
|
8065 |
+
"loss": 0.2758,
|
8066 |
+
"step": 1151
|
8067 |
+
},
|
8068 |
+
{
|
8069 |
+
"epoch": 0.8699263734189163,
|
8070 |
+
"grad_norm": 0.16692477464675903,
|
8071 |
+
"learning_rate": 3.8117569277564434e-06,
|
8072 |
+
"loss": 0.3403,
|
8073 |
+
"step": 1152
|
8074 |
+
},
|
8075 |
+
{
|
8076 |
+
"epoch": 0.8706815178402869,
|
8077 |
+
"grad_norm": 0.17089448869228363,
|
8078 |
+
"learning_rate": 3.766004583324956e-06,
|
8079 |
+
"loss": 0.3199,
|
8080 |
+
"step": 1153
|
8081 |
+
},
|
8082 |
+
{
|
8083 |
+
"epoch": 0.8714366622616575,
|
8084 |
+
"grad_norm": 0.1756073534488678,
|
8085 |
+
"learning_rate": 3.7205177334247443e-06,
|
8086 |
+
"loss": 0.349,
|
8087 |
+
"step": 1154
|
8088 |
+
},
|
8089 |
+
{
|
8090 |
+
"epoch": 0.8721918066830281,
|
8091 |
+
"grad_norm": 0.18730361759662628,
|
8092 |
+
"learning_rate": 3.675296639259912e-06,
|
8093 |
+
"loss": 0.3708,
|
8094 |
+
"step": 1155
|
8095 |
+
},
|
8096 |
+
{
|
8097 |
+
"epoch": 0.8729469511043987,
|
8098 |
+
"grad_norm": 0.1944558471441269,
|
8099 |
+
"learning_rate": 3.630341560508521e-06,
|
8100 |
+
"loss": 0.4133,
|
8101 |
+
"step": 1156
|
8102 |
+
},
|
8103 |
+
{
|
8104 |
+
"epoch": 0.8737020955257693,
|
8105 |
+
"grad_norm": 0.19459518790245056,
|
8106 |
+
"learning_rate": 3.5856527553210116e-06,
|
8107 |
+
"loss": 0.355,
|
8108 |
+
"step": 1157
|
8109 |
+
},
|
8110 |
+
{
|
8111 |
+
"epoch": 0.8744572399471399,
|
8112 |
+
"grad_norm": 0.1979731023311615,
|
8113 |
+
"learning_rate": 3.5412304803188136e-06,
|
8114 |
+
"loss": 0.3999,
|
8115 |
+
"step": 1158
|
8116 |
+
},
|
8117 |
+
{
|
8118 |
+
"epoch": 0.8752123843685105,
|
8119 |
+
"grad_norm": 0.1909814327955246,
|
8120 |
+
"learning_rate": 3.497074990592819e-06,
|
8121 |
+
"loss": 0.3762,
|
8122 |
+
"step": 1159
|
8123 |
+
},
|
8124 |
+
{
|
8125 |
+
"epoch": 0.875967528789881,
|
8126 |
+
"grad_norm": 0.20063309371471405,
|
8127 |
+
"learning_rate": 3.4531865397019246e-06,
|
8128 |
+
"loss": 0.4081,
|
8129 |
+
"step": 1160
|
8130 |
+
},
|
8131 |
+
{
|
8132 |
+
"epoch": 0.8767226732112516,
|
8133 |
+
"grad_norm": 0.19899629056453705,
|
8134 |
+
"learning_rate": 3.409565379671592e-06,
|
8135 |
+
"loss": 0.3843,
|
8136 |
+
"step": 1161
|
8137 |
+
},
|
8138 |
+
{
|
8139 |
+
"epoch": 0.8774778176326222,
|
8140 |
+
"grad_norm": 0.2057962417602539,
|
8141 |
+
"learning_rate": 3.3662117609923715e-06,
|
8142 |
+
"loss": 0.3534,
|
8143 |
+
"step": 1162
|
8144 |
+
},
|
8145 |
+
{
|
8146 |
+
"epoch": 0.8782329620539928,
|
8147 |
+
"grad_norm": 0.20951855182647705,
|
8148 |
+
"learning_rate": 3.3231259326184984e-06,
|
8149 |
+
"loss": 0.4041,
|
8150 |
+
"step": 1163
|
8151 |
+
},
|
8152 |
+
{
|
8153 |
+
"epoch": 0.8789881064753634,
|
8154 |
+
"grad_norm": 0.21981364488601685,
|
8155 |
+
"learning_rate": 3.2803081419664484e-06,
|
8156 |
+
"loss": 0.4235,
|
8157 |
+
"step": 1164
|
8158 |
+
},
|
8159 |
+
{
|
8160 |
+
"epoch": 0.879743250896734,
|
8161 |
+
"grad_norm": 0.22572548687458038,
|
8162 |
+
"learning_rate": 3.2377586349134846e-06,
|
8163 |
+
"loss": 0.4469,
|
8164 |
+
"step": 1165
|
8165 |
+
},
|
8166 |
+
{
|
8167 |
+
"epoch": 0.8804983953181046,
|
8168 |
+
"grad_norm": 0.22527821362018585,
|
8169 |
+
"learning_rate": 3.1954776557963085e-06,
|
8170 |
+
"loss": 0.3998,
|
8171 |
+
"step": 1166
|
8172 |
+
},
|
8173 |
+
{
|
8174 |
+
"epoch": 0.8812535397394752,
|
8175 |
+
"grad_norm": 0.2290899008512497,
|
8176 |
+
"learning_rate": 3.1534654474096206e-06,
|
8177 |
+
"loss": 0.4271,
|
8178 |
+
"step": 1167
|
8179 |
+
},
|
8180 |
+
{
|
8181 |
+
"epoch": 0.8820086841608458,
|
8182 |
+
"grad_norm": 0.23082055151462555,
|
8183 |
+
"learning_rate": 3.1117222510047118e-06,
|
8184 |
+
"loss": 0.4626,
|
8185 |
+
"step": 1168
|
8186 |
+
},
|
8187 |
+
{
|
8188 |
+
"epoch": 0.8827638285822164,
|
8189 |
+
"grad_norm": 0.2215840071439743,
|
8190 |
+
"learning_rate": 3.0702483062881203e-06,
|
8191 |
+
"loss": 0.366,
|
8192 |
+
"step": 1169
|
8193 |
+
},
|
8194 |
+
{
|
8195 |
+
"epoch": 0.883518973003587,
|
8196 |
+
"grad_norm": 0.2405186891555786,
|
8197 |
+
"learning_rate": 3.0290438514201936e-06,
|
8198 |
+
"loss": 0.4314,
|
8199 |
+
"step": 1170
|
8200 |
+
},
|
8201 |
+
{
|
8202 |
+
"epoch": 0.8842741174249575,
|
8203 |
+
"grad_norm": 0.23838458955287933,
|
8204 |
+
"learning_rate": 2.988109123013805e-06,
|
8205 |
+
"loss": 0.4155,
|
8206 |
+
"step": 1171
|
8207 |
+
},
|
8208 |
+
{
|
8209 |
+
"epoch": 0.8850292618463281,
|
8210 |
+
"grad_norm": 0.2328627109527588,
|
8211 |
+
"learning_rate": 2.9474443561329167e-06,
|
8212 |
+
"loss": 0.3708,
|
8213 |
+
"step": 1172
|
8214 |
+
},
|
8215 |
+
{
|
8216 |
+
"epoch": 0.8857844062676987,
|
8217 |
+
"grad_norm": 0.23276405036449432,
|
8218 |
+
"learning_rate": 2.907049784291266e-06,
|
8219 |
+
"loss": 0.3825,
|
8220 |
+
"step": 1173
|
8221 |
+
},
|
8222 |
+
{
|
8223 |
+
"epoch": 0.8865395506890693,
|
8224 |
+
"grad_norm": 0.22206051647663116,
|
8225 |
+
"learning_rate": 2.8669256394510234e-06,
|
8226 |
+
"loss": 0.3593,
|
8227 |
+
"step": 1174
|
8228 |
+
},
|
8229 |
+
{
|
8230 |
+
"epoch": 0.8872946951104399,
|
8231 |
+
"grad_norm": 0.24987028539180756,
|
8232 |
+
"learning_rate": 2.827072152021465e-06,
|
8233 |
+
"loss": 0.4427,
|
8234 |
+
"step": 1175
|
8235 |
+
},
|
8236 |
+
{
|
8237 |
+
"epoch": 0.8880498395318105,
|
8238 |
+
"grad_norm": 0.2531154155731201,
|
8239 |
+
"learning_rate": 2.787489550857619e-06,
|
8240 |
+
"loss": 0.3854,
|
8241 |
+
"step": 1176
|
8242 |
+
},
|
8243 |
+
{
|
8244 |
+
"epoch": 0.8888049839531811,
|
8245 |
+
"grad_norm": 0.24906668066978455,
|
8246 |
+
"learning_rate": 2.7481780632589993e-06,
|
8247 |
+
"loss": 0.3828,
|
8248 |
+
"step": 1177
|
8249 |
+
},
|
8250 |
+
{
|
8251 |
+
"epoch": 0.8895601283745517,
|
8252 |
+
"grad_norm": 0.26225507259368896,
|
8253 |
+
"learning_rate": 2.7091379149682685e-06,
|
8254 |
+
"loss": 0.3999,
|
8255 |
+
"step": 1178
|
8256 |
+
},
|
8257 |
+
{
|
8258 |
+
"epoch": 0.8903152727959223,
|
8259 |
+
"grad_norm": 0.2617410719394684,
|
8260 |
+
"learning_rate": 2.670369330169925e-06,
|
8261 |
+
"loss": 0.4004,
|
8262 |
+
"step": 1179
|
8263 |
+
},
|
8264 |
+
{
|
8265 |
+
"epoch": 0.8910704172172929,
|
8266 |
+
"grad_norm": 0.26333704590797424,
|
8267 |
+
"learning_rate": 2.6318725314890715e-06,
|
8268 |
+
"loss": 0.4155,
|
8269 |
+
"step": 1180
|
8270 |
+
},
|
8271 |
+
{
|
8272 |
+
"epoch": 0.8918255616386634,
|
8273 |
+
"grad_norm": 0.24778012931346893,
|
8274 |
+
"learning_rate": 2.593647739990068e-06,
|
8275 |
+
"loss": 0.3652,
|
8276 |
+
"step": 1181
|
8277 |
+
},
|
8278 |
+
{
|
8279 |
+
"epoch": 0.892580706060034,
|
8280 |
+
"grad_norm": 0.27770712971687317,
|
8281 |
+
"learning_rate": 2.555695175175338e-06,
|
8282 |
+
"loss": 0.435,
|
8283 |
+
"step": 1182
|
8284 |
+
},
|
8285 |
+
{
|
8286 |
+
"epoch": 0.8933358504814045,
|
8287 |
+
"grad_norm": 0.26981207728385925,
|
8288 |
+
"learning_rate": 2.518015054984041e-06,
|
8289 |
+
"loss": 0.3979,
|
8290 |
+
"step": 1183
|
8291 |
+
},
|
8292 |
+
{
|
8293 |
+
"epoch": 0.8940909949027751,
|
8294 |
+
"grad_norm": 0.28787314891815186,
|
8295 |
+
"learning_rate": 2.480607595790846e-06,
|
8296 |
+
"loss": 0.3894,
|
8297 |
+
"step": 1184
|
8298 |
+
},
|
8299 |
+
{
|
8300 |
+
"epoch": 0.8948461393241457,
|
8301 |
+
"grad_norm": 0.27000874280929565,
|
8302 |
+
"learning_rate": 2.4434730124047068e-06,
|
8303 |
+
"loss": 0.3587,
|
8304 |
+
"step": 1185
|
8305 |
+
},
|
8306 |
+
{
|
8307 |
+
"epoch": 0.8956012837455163,
|
8308 |
+
"grad_norm": 0.27463385462760925,
|
8309 |
+
"learning_rate": 2.406611518067614e-06,
|
8310 |
+
"loss": 0.3957,
|
8311 |
+
"step": 1186
|
8312 |
+
},
|
8313 |
+
{
|
8314 |
+
"epoch": 0.8963564281668869,
|
8315 |
+
"grad_norm": 0.28618261218070984,
|
8316 |
+
"learning_rate": 2.370023324453341e-06,
|
8317 |
+
"loss": 0.4097,
|
8318 |
+
"step": 1187
|
8319 |
+
},
|
8320 |
+
{
|
8321 |
+
"epoch": 0.8971115725882575,
|
8322 |
+
"grad_norm": 0.32782208919525146,
|
8323 |
+
"learning_rate": 2.333708641666299e-06,
|
8324 |
+
"loss": 0.4427,
|
8325 |
+
"step": 1188
|
8326 |
+
},
|
8327 |
+
{
|
8328 |
+
"epoch": 0.8978667170096281,
|
8329 |
+
"grad_norm": 0.30927029252052307,
|
8330 |
+
"learning_rate": 2.2976676782402396e-06,
|
8331 |
+
"loss": 0.3857,
|
8332 |
+
"step": 1189
|
8333 |
+
},
|
8334 |
+
{
|
8335 |
+
"epoch": 0.8986218614309986,
|
8336 |
+
"grad_norm": 0.2805861830711365,
|
8337 |
+
"learning_rate": 2.2619006411371435e-06,
|
8338 |
+
"loss": 0.3028,
|
8339 |
+
"step": 1190
|
8340 |
+
},
|
8341 |
+
{
|
8342 |
+
"epoch": 0.8993770058523692,
|
8343 |
+
"grad_norm": 0.3102644979953766,
|
8344 |
+
"learning_rate": 2.2264077357459944e-06,
|
8345 |
+
"loss": 0.4104,
|
8346 |
+
"step": 1191
|
8347 |
+
},
|
8348 |
+
{
|
8349 |
+
"epoch": 0.9001321502737398,
|
8350 |
+
"grad_norm": 0.3404208719730377,
|
8351 |
+
"learning_rate": 2.191189165881563e-06,
|
8352 |
+
"loss": 0.4215,
|
8353 |
+
"step": 1192
|
8354 |
+
},
|
8355 |
+
{
|
8356 |
+
"epoch": 0.9008872946951104,
|
8357 |
+
"grad_norm": 0.34622740745544434,
|
8358 |
+
"learning_rate": 2.156245133783308e-06,
|
8359 |
+
"loss": 0.3945,
|
8360 |
+
"step": 1193
|
8361 |
+
},
|
8362 |
+
{
|
8363 |
+
"epoch": 0.901642439116481,
|
8364 |
+
"grad_norm": 0.3446052670478821,
|
8365 |
+
"learning_rate": 2.121575840114176e-06,
|
8366 |
+
"loss": 0.3668,
|
8367 |
+
"step": 1194
|
8368 |
+
},
|
8369 |
+
{
|
8370 |
+
"epoch": 0.9023975835378516,
|
8371 |
+
"grad_norm": 0.3375178277492523,
|
8372 |
+
"learning_rate": 2.087181483959433e-06,
|
8373 |
+
"loss": 0.3543,
|
8374 |
+
"step": 1195
|
8375 |
+
},
|
8376 |
+
{
|
8377 |
+
"epoch": 0.9031527279592222,
|
8378 |
+
"grad_norm": 0.3674304187297821,
|
8379 |
+
"learning_rate": 2.0530622628255615e-06,
|
8380 |
+
"loss": 0.3769,
|
8381 |
+
"step": 1196
|
8382 |
+
},
|
8383 |
+
{
|
8384 |
+
"epoch": 0.9039078723805928,
|
8385 |
+
"grad_norm": 0.37681370973587036,
|
8386 |
+
"learning_rate": 2.019218372639098e-06,
|
8387 |
+
"loss": 0.4049,
|
8388 |
+
"step": 1197
|
8389 |
+
},
|
8390 |
+
{
|
8391 |
+
"epoch": 0.9046630168019634,
|
8392 |
+
"grad_norm": 0.3854749798774719,
|
8393 |
+
"learning_rate": 1.9856500077455155e-06,
|
8394 |
+
"loss": 0.4216,
|
8395 |
+
"step": 1198
|
8396 |
+
},
|
8397 |
+
{
|
8398 |
+
"epoch": 0.905418161223334,
|
8399 |
+
"grad_norm": 0.47074857354164124,
|
8400 |
+
"learning_rate": 1.952357360908114e-06,
|
8401 |
+
"loss": 0.4559,
|
8402 |
+
"step": 1199
|
8403 |
+
},
|
8404 |
+
{
|
8405 |
+
"epoch": 0.9061733056447046,
|
8406 |
+
"grad_norm": 0.6017633080482483,
|
8407 |
+
"learning_rate": 1.9193406233068924e-06,
|
8408 |
+
"loss": 0.4365,
|
8409 |
+
"step": 1200
|
8410 |
+
},
|
8411 |
+
{
|
8412 |
+
"epoch": 0.9069284500660751,
|
8413 |
+
"grad_norm": 0.14071007072925568,
|
8414 |
+
"learning_rate": 1.8865999845374793e-06,
|
8415 |
+
"loss": 0.2597,
|
8416 |
+
"step": 1201
|
8417 |
+
},
|
8418 |
+
{
|
8419 |
+
"epoch": 0.9076835944874457,
|
8420 |
+
"grad_norm": 0.15932346880435944,
|
8421 |
+
"learning_rate": 1.8541356326100433e-06,
|
8422 |
+
"loss": 0.3266,
|
8423 |
+
"step": 1202
|
8424 |
+
},
|
8425 |
+
{
|
8426 |
+
"epoch": 0.9084387389088163,
|
8427 |
+
"grad_norm": 0.1692121922969818,
|
8428 |
+
"learning_rate": 1.8219477539481555e-06,
|
8429 |
+
"loss": 0.3298,
|
8430 |
+
"step": 1203
|
8431 |
+
},
|
8432 |
+
{
|
8433 |
+
"epoch": 0.9091938833301869,
|
8434 |
+
"grad_norm": 0.16915957629680634,
|
8435 |
+
"learning_rate": 1.7900365333878177e-06,
|
8436 |
+
"loss": 0.3622,
|
8437 |
+
"step": 1204
|
8438 |
+
},
|
8439 |
+
{
|
8440 |
+
"epoch": 0.9099490277515575,
|
8441 |
+
"grad_norm": 0.18877923488616943,
|
8442 |
+
"learning_rate": 1.758402154176314e-06,
|
8443 |
+
"loss": 0.3674,
|
8444 |
+
"step": 1205
|
8445 |
+
},
|
8446 |
+
{
|
8447 |
+
"epoch": 0.9107041721729281,
|
8448 |
+
"grad_norm": 0.18287624418735504,
|
8449 |
+
"learning_rate": 1.7270447979712056e-06,
|
8450 |
+
"loss": 0.3427,
|
8451 |
+
"step": 1206
|
8452 |
+
},
|
8453 |
+
{
|
8454 |
+
"epoch": 0.9114593165942987,
|
8455 |
+
"grad_norm": 0.1901153177022934,
|
8456 |
+
"learning_rate": 1.6959646448392763e-06,
|
8457 |
+
"loss": 0.4145,
|
8458 |
+
"step": 1207
|
8459 |
+
},
|
8460 |
+
{
|
8461 |
+
"epoch": 0.9122144610156693,
|
8462 |
+
"grad_norm": 0.2029486745595932,
|
8463 |
+
"learning_rate": 1.6651618732554775e-06,
|
8464 |
+
"loss": 0.3926,
|
8465 |
+
"step": 1208
|
8466 |
+
},
|
8467 |
+
{
|
8468 |
+
"epoch": 0.9129696054370399,
|
8469 |
+
"grad_norm": 0.19846974313259125,
|
8470 |
+
"learning_rate": 1.6346366601019347e-06,
|
8471 |
+
"loss": 0.3617,
|
8472 |
+
"step": 1209
|
8473 |
+
},
|
8474 |
+
{
|
8475 |
+
"epoch": 0.9137247498584105,
|
8476 |
+
"grad_norm": 0.2179420292377472,
|
8477 |
+
"learning_rate": 1.6043891806669376e-06,
|
8478 |
+
"loss": 0.4296,
|
8479 |
+
"step": 1210
|
8480 |
+
},
|
8481 |
+
{
|
8482 |
+
"epoch": 0.914479894279781,
|
8483 |
+
"grad_norm": 0.20902778208255768,
|
8484 |
+
"learning_rate": 1.574419608643879e-06,
|
8485 |
+
"loss": 0.3957,
|
8486 |
+
"step": 1211
|
8487 |
+
},
|
8488 |
+
{
|
8489 |
+
"epoch": 0.9152350387011516,
|
8490 |
+
"grad_norm": 0.21325752139091492,
|
8491 |
+
"learning_rate": 1.544728116130323e-06,
|
8492 |
+
"loss": 0.3968,
|
8493 |
+
"step": 1212
|
8494 |
+
},
|
8495 |
+
{
|
8496 |
+
"epoch": 0.9159901831225222,
|
8497 |
+
"grad_norm": 0.22161689400672913,
|
8498 |
+
"learning_rate": 1.5153148736269719e-06,
|
8499 |
+
"loss": 0.3584,
|
8500 |
+
"step": 1213
|
8501 |
+
},
|
8502 |
+
{
|
8503 |
+
"epoch": 0.9167453275438928,
|
8504 |
+
"grad_norm": 0.22005094587802887,
|
8505 |
+
"learning_rate": 1.4861800500367008e-06,
|
8506 |
+
"loss": 0.4194,
|
8507 |
+
"step": 1214
|
8508 |
+
},
|
8509 |
+
{
|
8510 |
+
"epoch": 0.9175004719652634,
|
8511 |
+
"grad_norm": 0.2082773596048355,
|
8512 |
+
"learning_rate": 1.457323812663608e-06,
|
8513 |
+
"loss": 0.3759,
|
8514 |
+
"step": 1215
|
8515 |
+
},
|
8516 |
+
{
|
8517 |
+
"epoch": 0.9182556163866339,
|
8518 |
+
"grad_norm": 0.21454690396785736,
|
8519 |
+
"learning_rate": 1.4287463272120104e-06,
|
8520 |
+
"loss": 0.3494,
|
8521 |
+
"step": 1216
|
8522 |
+
},
|
8523 |
+
{
|
8524 |
+
"epoch": 0.9190107608080045,
|
8525 |
+
"grad_norm": 0.22656020522117615,
|
8526 |
+
"learning_rate": 1.400447757785539e-06,
|
8527 |
+
"loss": 0.4118,
|
8528 |
+
"step": 1217
|
8529 |
+
},
|
8530 |
+
{
|
8531 |
+
"epoch": 0.9197659052293751,
|
8532 |
+
"grad_norm": 0.2260131984949112,
|
8533 |
+
"learning_rate": 1.3724282668861665e-06,
|
8534 |
+
"loss": 0.4568,
|
8535 |
+
"step": 1218
|
8536 |
+
},
|
8537 |
+
{
|
8538 |
+
"epoch": 0.9205210496507457,
|
8539 |
+
"grad_norm": 0.2255602777004242,
|
8540 |
+
"learning_rate": 1.3446880154132757e-06,
|
8541 |
+
"loss": 0.4048,
|
8542 |
+
"step": 1219
|
8543 |
+
},
|
8544 |
+
{
|
8545 |
+
"epoch": 0.9212761940721163,
|
8546 |
+
"grad_norm": 0.23461206257343292,
|
8547 |
+
"learning_rate": 1.3172271626627486e-06,
|
8548 |
+
"loss": 0.4381,
|
8549 |
+
"step": 1220
|
8550 |
+
},
|
8551 |
+
{
|
8552 |
+
"epoch": 0.9220313384934868,
|
8553 |
+
"grad_norm": 0.239901602268219,
|
8554 |
+
"learning_rate": 1.2900458663260506e-06,
|
8555 |
+
"loss": 0.4144,
|
8556 |
+
"step": 1221
|
8557 |
+
},
|
8558 |
+
{
|
8559 |
+
"epoch": 0.9227864829148574,
|
8560 |
+
"grad_norm": 0.23655924201011658,
|
8561 |
+
"learning_rate": 1.2631442824893036e-06,
|
8562 |
+
"loss": 0.4315,
|
8563 |
+
"step": 1222
|
8564 |
+
},
|
8565 |
+
{
|
8566 |
+
"epoch": 0.923541627336228,
|
8567 |
+
"grad_norm": 0.22921998798847198,
|
8568 |
+
"learning_rate": 1.236522565632431e-06,
|
8569 |
+
"loss": 0.3893,
|
8570 |
+
"step": 1223
|
8571 |
+
},
|
8572 |
+
{
|
8573 |
+
"epoch": 0.9242967717575986,
|
8574 |
+
"grad_norm": 0.2575366795063019,
|
8575 |
+
"learning_rate": 1.210180868628219e-06,
|
8576 |
+
"loss": 0.4597,
|
8577 |
+
"step": 1224
|
8578 |
+
},
|
8579 |
+
{
|
8580 |
+
"epoch": 0.9250519161789692,
|
8581 |
+
"grad_norm": 0.2390882968902588,
|
8582 |
+
"learning_rate": 1.1841193427414798e-06,
|
8583 |
+
"loss": 0.3758,
|
8584 |
+
"step": 1225
|
8585 |
+
},
|
8586 |
+
{
|
8587 |
+
"epoch": 0.9258070606003398,
|
8588 |
+
"grad_norm": 0.2630118131637573,
|
8589 |
+
"learning_rate": 1.1583381376281731e-06,
|
8590 |
+
"loss": 0.4565,
|
8591 |
+
"step": 1226
|
8592 |
+
},
|
8593 |
+
{
|
8594 |
+
"epoch": 0.9265622050217104,
|
8595 |
+
"grad_norm": 0.27250075340270996,
|
8596 |
+
"learning_rate": 1.132837401334519e-06,
|
8597 |
+
"loss": 0.4368,
|
8598 |
+
"step": 1227
|
8599 |
+
},
|
8600 |
+
{
|
8601 |
+
"epoch": 0.927317349443081,
|
8602 |
+
"grad_norm": 0.2387106716632843,
|
8603 |
+
"learning_rate": 1.107617280296197e-06,
|
8604 |
+
"loss": 0.3618,
|
8605 |
+
"step": 1228
|
8606 |
+
},
|
8607 |
+
{
|
8608 |
+
"epoch": 0.9280724938644516,
|
8609 |
+
"grad_norm": 0.2761591374874115,
|
8610 |
+
"learning_rate": 1.0826779193374713e-06,
|
8611 |
+
"loss": 0.4607,
|
8612 |
+
"step": 1229
|
8613 |
+
},
|
8614 |
+
{
|
8615 |
+
"epoch": 0.9288276382858222,
|
8616 |
+
"grad_norm": 0.2529843747615814,
|
8617 |
+
"learning_rate": 1.0580194616703565e-06,
|
8618 |
+
"loss": 0.3403,
|
8619 |
+
"step": 1230
|
8620 |
+
},
|
8621 |
+
{
|
8622 |
+
"epoch": 0.9295827827071927,
|
8623 |
+
"grad_norm": 0.27074989676475525,
|
8624 |
+
"learning_rate": 1.033642048893807e-06,
|
8625 |
+
"loss": 0.4108,
|
8626 |
+
"step": 1231
|
8627 |
+
},
|
8628 |
+
{
|
8629 |
+
"epoch": 0.9303379271285633,
|
8630 |
+
"grad_norm": 0.28065839409828186,
|
8631 |
+
"learning_rate": 1.0095458209929242e-06,
|
8632 |
+
"loss": 0.4031,
|
8633 |
+
"step": 1232
|
8634 |
+
},
|
8635 |
+
{
|
8636 |
+
"epoch": 0.9310930715499339,
|
8637 |
+
"grad_norm": 0.28449884057044983,
|
8638 |
+
"learning_rate": 9.857309163380956e-07,
|
8639 |
+
"loss": 0.3624,
|
8640 |
+
"step": 1233
|
8641 |
+
},
|
8642 |
+
{
|
8643 |
+
"epoch": 0.9318482159713045,
|
8644 |
+
"grad_norm": 0.2827039361000061,
|
8645 |
+
"learning_rate": 9.621974716842674e-07,
|
8646 |
+
"loss": 0.3778,
|
8647 |
+
"step": 1234
|
8648 |
+
},
|
8649 |
+
{
|
8650 |
+
"epoch": 0.9326033603926751,
|
8651 |
+
"grad_norm": 0.2874571681022644,
|
8652 |
+
"learning_rate": 9.389456221701121e-07,
|
8653 |
+
"loss": 0.4096,
|
8654 |
+
"step": 1235
|
8655 |
+
},
|
8656 |
+
{
|
8657 |
+
"epoch": 0.9333585048140457,
|
8658 |
+
"grad_norm": 0.28555232286453247,
|
8659 |
+
"learning_rate": 9.159755013172622e-07,
|
8660 |
+
"loss": 0.3969,
|
8661 |
+
"step": 1236
|
8662 |
+
},
|
8663 |
+
{
|
8664 |
+
"epoch": 0.9341136492354163,
|
8665 |
+
"grad_norm": 0.29169705510139465,
|
8666 |
+
"learning_rate": 8.932872410295667e-07,
|
8667 |
+
"loss": 0.3701,
|
8668 |
+
"step": 1237
|
8669 |
+
},
|
8670 |
+
{
|
8671 |
+
"epoch": 0.9348687936567869,
|
8672 |
+
"grad_norm": 0.29561880230903625,
|
8673 |
+
"learning_rate": 8.708809715922972e-07,
|
8674 |
+
"loss": 0.3977,
|
8675 |
+
"step": 1238
|
8676 |
+
},
|
8677 |
+
{
|
8678 |
+
"epoch": 0.9356239380781575,
|
8679 |
+
"grad_norm": 0.30140790343284607,
|
8680 |
+
"learning_rate": 8.487568216714315e-07,
|
8681 |
+
"loss": 0.4149,
|
8682 |
+
"step": 1239
|
8683 |
+
},
|
8684 |
+
{
|
8685 |
+
"epoch": 0.9363790824995281,
|
8686 |
+
"grad_norm": 0.3187549114227295,
|
8687 |
+
"learning_rate": 8.269149183128988e-07,
|
8688 |
+
"loss": 0.3354,
|
8689 |
+
"step": 1240
|
8690 |
+
},
|
8691 |
+
{
|
8692 |
+
"epoch": 0.9371342269208986,
|
8693 |
+
"grad_norm": 0.33990204334259033,
|
8694 |
+
"learning_rate": 8.053553869418418e-07,
|
8695 |
+
"loss": 0.4578,
|
8696 |
+
"step": 1241
|
8697 |
+
},
|
8698 |
+
{
|
8699 |
+
"epoch": 0.9378893713422692,
|
8700 |
+
"grad_norm": 0.3531990051269531,
|
8701 |
+
"learning_rate": 7.840783513619221e-07,
|
8702 |
+
"loss": 0.4437,
|
8703 |
+
"step": 1242
|
8704 |
+
},
|
8705 |
+
{
|
8706 |
+
"epoch": 0.9386445157636398,
|
8707 |
+
"grad_norm": 0.34928497672080994,
|
8708 |
+
"learning_rate": 7.630839337545936e-07,
|
8709 |
+
"loss": 0.4052,
|
8710 |
+
"step": 1243
|
8711 |
+
},
|
8712 |
+
{
|
8713 |
+
"epoch": 0.9393996601850104,
|
8714 |
+
"grad_norm": 0.36756661534309387,
|
8715 |
+
"learning_rate": 7.423722546783918e-07,
|
8716 |
+
"loss": 0.3936,
|
8717 |
+
"step": 1244
|
8718 |
+
},
|
8719 |
+
{
|
8720 |
+
"epoch": 0.940154804606381,
|
8721 |
+
"grad_norm": 0.35124626755714417,
|
8722 |
+
"learning_rate": 7.219434330682618e-07,
|
8723 |
+
"loss": 0.3659,
|
8724 |
+
"step": 1245
|
8725 |
+
},
|
8726 |
+
{
|
8727 |
+
"epoch": 0.9409099490277516,
|
8728 |
+
"grad_norm": 0.3702350854873657,
|
8729 |
+
"learning_rate": 7.017975862348592e-07,
|
8730 |
+
"loss": 0.2996,
|
8731 |
+
"step": 1246
|
8732 |
+
},
|
8733 |
+
{
|
8734 |
+
"epoch": 0.9416650934491222,
|
8735 |
+
"grad_norm": 0.3857671320438385,
|
8736 |
+
"learning_rate": 6.819348298638839e-07,
|
8737 |
+
"loss": 0.4173,
|
8738 |
+
"step": 1247
|
8739 |
+
},
|
8740 |
+
{
|
8741 |
+
"epoch": 0.9424202378704928,
|
8742 |
+
"grad_norm": 0.35669073462486267,
|
8743 |
+
"learning_rate": 6.623552780154197e-07,
|
8744 |
+
"loss": 0.324,
|
8745 |
+
"step": 1248
|
8746 |
+
},
|
8747 |
+
{
|
8748 |
+
"epoch": 0.9431753822918634,
|
8749 |
+
"grad_norm": 0.47897228598594666,
|
8750 |
+
"learning_rate": 6.430590431232563e-07,
|
8751 |
+
"loss": 0.458,
|
8752 |
+
"step": 1249
|
8753 |
+
},
|
8754 |
+
{
|
8755 |
+
"epoch": 0.9439305267132339,
|
8756 |
+
"grad_norm": 0.5287113785743713,
|
8757 |
+
"learning_rate": 6.240462359942967e-07,
|
8758 |
+
"loss": 0.4152,
|
8759 |
+
"step": 1250
|
8760 |
+
},
|
8761 |
+
{
|
8762 |
+
"epoch": 0.9446856711346044,
|
8763 |
+
"grad_norm": 0.16385497152805328,
|
8764 |
+
"learning_rate": 6.053169658078506e-07,
|
8765 |
+
"loss": 0.316,
|
8766 |
+
"step": 1251
|
8767 |
+
},
|
8768 |
+
{
|
8769 |
+
"epoch": 0.945440815555975,
|
8770 |
+
"grad_norm": 0.16225773096084595,
|
8771 |
+
"learning_rate": 5.868713401150583e-07,
|
8772 |
+
"loss": 0.3335,
|
8773 |
+
"step": 1252
|
8774 |
+
},
|
8775 |
+
{
|
8776 |
+
"epoch": 0.9461959599773456,
|
8777 |
+
"grad_norm": 0.16285216808319092,
|
8778 |
+
"learning_rate": 5.687094648382518e-07,
|
8779 |
+
"loss": 0.3312,
|
8780 |
+
"step": 1253
|
8781 |
+
},
|
8782 |
+
{
|
8783 |
+
"epoch": 0.9469511043987162,
|
8784 |
+
"grad_norm": 0.17022986710071564,
|
8785 |
+
"learning_rate": 5.508314442703611e-07,
|
8786 |
+
"loss": 0.2996,
|
8787 |
+
"step": 1254
|
8788 |
+
},
|
8789 |
+
{
|
8790 |
+
"epoch": 0.9477062488200868,
|
8791 |
+
"grad_norm": 0.18027344346046448,
|
8792 |
+
"learning_rate": 5.332373810742919e-07,
|
8793 |
+
"loss": 0.3887,
|
8794 |
+
"step": 1255
|
8795 |
+
},
|
8796 |
+
{
|
8797 |
+
"epoch": 0.9484613932414574,
|
8798 |
+
"grad_norm": 0.1849389225244522,
|
8799 |
+
"learning_rate": 5.159273762823657e-07,
|
8800 |
+
"loss": 0.3787,
|
8801 |
+
"step": 1256
|
8802 |
+
},
|
8803 |
+
{
|
8804 |
+
"epoch": 0.949216537662828,
|
8805 |
+
"grad_norm": 0.20186136662960052,
|
8806 |
+
"learning_rate": 4.989015292957144e-07,
|
8807 |
+
"loss": 0.3687,
|
8808 |
+
"step": 1257
|
8809 |
+
},
|
8810 |
+
{
|
8811 |
+
"epoch": 0.9499716820841986,
|
8812 |
+
"grad_norm": 0.19465675950050354,
|
8813 |
+
"learning_rate": 4.821599378837249e-07,
|
8814 |
+
"loss": 0.3625,
|
8815 |
+
"step": 1258
|
8816 |
+
},
|
8817 |
+
{
|
8818 |
+
"epoch": 0.9507268265055692,
|
8819 |
+
"grad_norm": 0.19634687900543213,
|
8820 |
+
"learning_rate": 4.6570269818346224e-07,
|
8821 |
+
"loss": 0.4195,
|
8822 |
+
"step": 1259
|
8823 |
+
},
|
8824 |
+
{
|
8825 |
+
"epoch": 0.9514819709269398,
|
8826 |
+
"grad_norm": 0.204311341047287,
|
8827 |
+
"learning_rate": 4.495299046991419e-07,
|
8828 |
+
"loss": 0.3959,
|
8829 |
+
"step": 1260
|
8830 |
+
},
|
8831 |
+
{
|
8832 |
+
"epoch": 0.9522371153483103,
|
8833 |
+
"grad_norm": 0.19881460070610046,
|
8834 |
+
"learning_rate": 4.336416503015639e-07,
|
8835 |
+
"loss": 0.4098,
|
8836 |
+
"step": 1261
|
8837 |
+
},
|
8838 |
+
{
|
8839 |
+
"epoch": 0.9529922597696809,
|
8840 |
+
"grad_norm": 0.20822936296463013,
|
8841 |
+
"learning_rate": 4.180380262275907e-07,
|
8842 |
+
"loss": 0.3894,
|
8843 |
+
"step": 1262
|
8844 |
+
},
|
8845 |
+
{
|
8846 |
+
"epoch": 0.9537474041910515,
|
8847 |
+
"grad_norm": 0.2120690792798996,
|
8848 |
+
"learning_rate": 4.0271912207962e-07,
|
8849 |
+
"loss": 0.379,
|
8850 |
+
"step": 1263
|
8851 |
+
},
|
8852 |
+
{
|
8853 |
+
"epoch": 0.9545025486124221,
|
8854 |
+
"grad_norm": 0.19805222749710083,
|
8855 |
+
"learning_rate": 3.8768502582507415e-07,
|
8856 |
+
"loss": 0.3766,
|
8857 |
+
"step": 1264
|
8858 |
+
},
|
8859 |
+
{
|
8860 |
+
"epoch": 0.9552576930337927,
|
8861 |
+
"grad_norm": 0.19410884380340576,
|
8862 |
+
"learning_rate": 3.7293582379587797e-07,
|
8863 |
+
"loss": 0.3523,
|
8864 |
+
"step": 1265
|
8865 |
+
},
|
8866 |
+
{
|
8867 |
+
"epoch": 0.9560128374551633,
|
8868 |
+
"grad_norm": 0.21975958347320557,
|
8869 |
+
"learning_rate": 3.584716006879985e-07,
|
8870 |
+
"loss": 0.3801,
|
8871 |
+
"step": 1266
|
8872 |
+
},
|
8873 |
+
{
|
8874 |
+
"epoch": 0.9567679818765339,
|
8875 |
+
"grad_norm": 0.22401130199432373,
|
8876 |
+
"learning_rate": 3.4429243956092817e-07,
|
8877 |
+
"loss": 0.4142,
|
8878 |
+
"step": 1267
|
8879 |
+
},
|
8880 |
+
{
|
8881 |
+
"epoch": 0.9575231262979045,
|
8882 |
+
"grad_norm": 0.2196296602487564,
|
8883 |
+
"learning_rate": 3.303984218372136e-07,
|
8884 |
+
"loss": 0.4482,
|
8885 |
+
"step": 1268
|
8886 |
+
},
|
8887 |
+
{
|
8888 |
+
"epoch": 0.9582782707192751,
|
8889 |
+
"grad_norm": 0.21536298096179962,
|
8890 |
+
"learning_rate": 3.167896273019888e-07,
|
8891 |
+
"loss": 0.3774,
|
8892 |
+
"step": 1269
|
8893 |
+
},
|
8894 |
+
{
|
8895 |
+
"epoch": 0.9590334151406457,
|
8896 |
+
"grad_norm": 0.24629651010036469,
|
8897 |
+
"learning_rate": 3.034661341025258e-07,
|
8898 |
+
"loss": 0.3771,
|
8899 |
+
"step": 1270
|
8900 |
+
},
|
8901 |
+
{
|
8902 |
+
"epoch": 0.9597885595620163,
|
8903 |
+
"grad_norm": 0.23244929313659668,
|
8904 |
+
"learning_rate": 2.9042801874777927e-07,
|
8905 |
+
"loss": 0.4528,
|
8906 |
+
"step": 1271
|
8907 |
+
},
|
8908 |
+
{
|
8909 |
+
"epoch": 0.9605437039833868,
|
8910 |
+
"grad_norm": 0.2288171648979187,
|
8911 |
+
"learning_rate": 2.776753561079481e-07,
|
8912 |
+
"loss": 0.3626,
|
8913 |
+
"step": 1272
|
8914 |
+
},
|
8915 |
+
{
|
8916 |
+
"epoch": 0.9612988484047574,
|
8917 |
+
"grad_norm": 0.22338609397411346,
|
8918 |
+
"learning_rate": 2.6520821941403704e-07,
|
8919 |
+
"loss": 0.3966,
|
8920 |
+
"step": 1273
|
8921 |
+
},
|
8922 |
+
{
|
8923 |
+
"epoch": 0.962053992826128,
|
8924 |
+
"grad_norm": 0.23835286498069763,
|
8925 |
+
"learning_rate": 2.53026680257451e-07,
|
8926 |
+
"loss": 0.4144,
|
8927 |
+
"step": 1274
|
8928 |
+
},
|
8929 |
+
{
|
8930 |
+
"epoch": 0.9628091372474986,
|
8931 |
+
"grad_norm": 0.2332562953233719,
|
8932 |
+
"learning_rate": 2.4113080858957914e-07,
|
8933 |
+
"loss": 0.3221,
|
8934 |
+
"step": 1275
|
8935 |
+
},
|
8936 |
+
{
|
8937 |
+
"epoch": 0.9635642816688692,
|
8938 |
+
"grad_norm": 0.2693784534931183,
|
8939 |
+
"learning_rate": 2.2952067272137834e-07,
|
8940 |
+
"loss": 0.4761,
|
8941 |
+
"step": 1276
|
8942 |
+
},
|
8943 |
+
{
|
8944 |
+
"epoch": 0.9643194260902398,
|
8945 |
+
"grad_norm": 0.2431482970714569,
|
8946 |
+
"learning_rate": 2.1819633932301796e-07,
|
8947 |
+
"loss": 0.3944,
|
8948 |
+
"step": 1277
|
8949 |
+
},
|
8950 |
+
{
|
8951 |
+
"epoch": 0.9650745705116104,
|
8952 |
+
"grad_norm": 0.2623022794723511,
|
8953 |
+
"learning_rate": 2.0715787342343584e-07,
|
8954 |
+
"loss": 0.3733,
|
8955 |
+
"step": 1278
|
8956 |
+
},
|
8957 |
+
{
|
8958 |
+
"epoch": 0.965829714932981,
|
8959 |
+
"grad_norm": 0.24833761155605316,
|
8960 |
+
"learning_rate": 1.9640533841003838e-07,
|
8961 |
+
"loss": 0.3611,
|
8962 |
+
"step": 1279
|
8963 |
+
},
|
8964 |
+
{
|
8965 |
+
"epoch": 0.9665848593543516,
|
8966 |
+
"grad_norm": 0.2568053603172302,
|
8967 |
+
"learning_rate": 1.8593879602828435e-07,
|
8968 |
+
"loss": 0.4602,
|
8969 |
+
"step": 1280
|
8970 |
+
},
|
8971 |
+
{
|
8972 |
+
"epoch": 0.9673400037757222,
|
8973 |
+
"grad_norm": 0.25137168169021606,
|
8974 |
+
"learning_rate": 1.7575830638133506e-07,
|
8975 |
+
"loss": 0.3323,
|
8976 |
+
"step": 1281
|
8977 |
+
},
|
8978 |
+
{
|
8979 |
+
"epoch": 0.9680951481970927,
|
8980 |
+
"grad_norm": 0.26353737711906433,
|
8981 |
+
"learning_rate": 1.6586392792973805e-07,
|
8982 |
+
"loss": 0.3855,
|
8983 |
+
"step": 1282
|
8984 |
+
},
|
8985 |
+
{
|
8986 |
+
"epoch": 0.9688502926184632,
|
8987 |
+
"grad_norm": 0.27223828434944153,
|
8988 |
+
"learning_rate": 1.562557174910606e-07,
|
8989 |
+
"loss": 0.3631,
|
8990 |
+
"step": 1283
|
8991 |
+
},
|
8992 |
+
{
|
8993 |
+
"epoch": 0.9696054370398338,
|
8994 |
+
"grad_norm": 0.2705494165420532,
|
8995 |
+
"learning_rate": 1.4693373023958458e-07,
|
8996 |
+
"loss": 0.4229,
|
8997 |
+
"step": 1284
|
8998 |
+
},
|
8999 |
+
{
|
9000 |
+
"epoch": 0.9703605814612044,
|
9001 |
+
"grad_norm": 0.2955893278121948,
|
9002 |
+
"learning_rate": 1.3789801970597872e-07,
|
9003 |
+
"loss": 0.534,
|
9004 |
+
"step": 1285
|
9005 |
+
},
|
9006 |
+
{
|
9007 |
+
"epoch": 0.971115725882575,
|
9008 |
+
"grad_norm": 0.2767453193664551,
|
9009 |
+
"learning_rate": 1.2914863777698792e-07,
|
9010 |
+
"loss": 0.3949,
|
9011 |
+
"step": 1286
|
9012 |
+
},
|
9013 |
+
{
|
9014 |
+
"epoch": 0.9718708703039456,
|
9015 |
+
"grad_norm": 0.28201353549957275,
|
9016 |
+
"learning_rate": 1.2068563469515014e-07,
|
9017 |
+
"loss": 0.3584,
|
9018 |
+
"step": 1287
|
9019 |
+
},
|
9020 |
+
{
|
9021 |
+
"epoch": 0.9726260147253162,
|
9022 |
+
"grad_norm": 0.28768327832221985,
|
9023 |
+
"learning_rate": 1.1250905905849097e-07,
|
9024 |
+
"loss": 0.3976,
|
9025 |
+
"step": 1288
|
9026 |
+
},
|
9027 |
+
{
|
9028 |
+
"epoch": 0.9733811591466868,
|
9029 |
+
"grad_norm": 0.30336374044418335,
|
9030 |
+
"learning_rate": 1.0461895782025167e-07,
|
9031 |
+
"loss": 0.3949,
|
9032 |
+
"step": 1289
|
9033 |
+
},
|
9034 |
+
{
|
9035 |
+
"epoch": 0.9741363035680574,
|
9036 |
+
"grad_norm": 0.32189151644706726,
|
9037 |
+
"learning_rate": 9.701537628863389e-08,
|
9038 |
+
"loss": 0.4426,
|
9039 |
+
"step": 1290
|
9040 |
+
},
|
9041 |
+
{
|
9042 |
+
"epoch": 0.974891447989428,
|
9043 |
+
"grad_norm": 0.3040902316570282,
|
9044 |
+
"learning_rate": 8.96983581265054e-08,
|
9045 |
+
"loss": 0.4035,
|
9046 |
+
"step": 1291
|
9047 |
+
},
|
9048 |
+
{
|
9049 |
+
"epoch": 0.9756465924107985,
|
9050 |
+
"grad_norm": 0.3161725699901581,
|
9051 |
+
"learning_rate": 8.266794535118916e-08,
|
9052 |
+
"loss": 0.3984,
|
9053 |
+
"step": 1292
|
9054 |
+
},
|
9055 |
+
{
|
9056 |
+
"epoch": 0.9764017368321691,
|
9057 |
+
"grad_norm": 0.3418150842189789,
|
9058 |
+
"learning_rate": 7.59241783341913e-08,
|
9059 |
+
"loss": 0.4402,
|
9060 |
+
"step": 1293
|
9061 |
+
},
|
9062 |
+
{
|
9063 |
+
"epoch": 0.9771568812535397,
|
9064 |
+
"grad_norm": 0.3445986807346344,
|
9065 |
+
"learning_rate": 6.94670958009791e-08,
|
9066 |
+
"loss": 0.4631,
|
9067 |
+
"step": 1294
|
9068 |
+
},
|
9069 |
+
{
|
9070 |
+
"epoch": 0.9779120256749103,
|
9071 |
+
"grad_norm": 0.3366871774196625,
|
9072 |
+
"learning_rate": 6.329673483076448e-08,
|
9073 |
+
"loss": 0.3533,
|
9074 |
+
"step": 1295
|
9075 |
+
},
|
9076 |
+
{
|
9077 |
+
"epoch": 0.9786671700962809,
|
9078 |
+
"grad_norm": 0.3505009710788727,
|
9079 |
+
"learning_rate": 5.7413130856293076e-08,
|
9080 |
+
"loss": 0.368,
|
9081 |
+
"step": 1296
|
9082 |
+
},
|
9083 |
+
{
|
9084 |
+
"epoch": 0.9794223145176515,
|
9085 |
+
"grad_norm": 0.4207570254802704,
|
9086 |
+
"learning_rate": 5.181631766362216e-08,
|
9087 |
+
"loss": 0.4519,
|
9088 |
+
"step": 1297
|
9089 |
+
},
|
9090 |
+
{
|
9091 |
+
"epoch": 0.9801774589390221,
|
9092 |
+
"grad_norm": 0.36577296257019043,
|
9093 |
+
"learning_rate": 4.650632739194305e-08,
|
9094 |
+
"loss": 0.3788,
|
9095 |
+
"step": 1298
|
9096 |
+
},
|
9097 |
+
{
|
9098 |
+
"epoch": 0.9809326033603927,
|
9099 |
+
"grad_norm": 0.43589064478874207,
|
9100 |
+
"learning_rate": 4.148319053340344e-08,
|
9101 |
+
"loss": 0.4351,
|
9102 |
+
"step": 1299
|
9103 |
+
},
|
9104 |
+
{
|
9105 |
+
"epoch": 0.9816877477817633,
|
9106 |
+
"grad_norm": 0.5222182869911194,
|
9107 |
+
"learning_rate": 3.674693593290202e-08,
|
9108 |
+
"loss": 0.3479,
|
9109 |
+
"step": 1300
|
9110 |
+
},
|
9111 |
+
{
|
9112 |
+
"epoch": 0.9824428922031339,
|
9113 |
+
"grad_norm": 0.15054112672805786,
|
9114 |
+
"learning_rate": 3.229759078795524e-08,
|
9115 |
+
"loss": 0.3271,
|
9116 |
+
"step": 1301
|
9117 |
+
},
|
9118 |
+
{
|
9119 |
+
"epoch": 0.9831980366245044,
|
9120 |
+
"grad_norm": 0.16661597788333893,
|
9121 |
+
"learning_rate": 2.813518064852527e-08,
|
9122 |
+
"loss": 0.3463,
|
9123 |
+
"step": 1302
|
9124 |
+
},
|
9125 |
+
{
|
9126 |
+
"epoch": 0.983953181045875,
|
9127 |
+
"grad_norm": 0.18651586771011353,
|
9128 |
+
"learning_rate": 2.4259729416870047e-08,
|
9129 |
+
"loss": 0.4193,
|
9130 |
+
"step": 1303
|
9131 |
+
},
|
9132 |
+
{
|
9133 |
+
"epoch": 0.9847083254672456,
|
9134 |
+
"grad_norm": 0.18549008667469025,
|
9135 |
+
"learning_rate": 2.0671259347426753e-08,
|
9136 |
+
"loss": 0.3928,
|
9137 |
+
"step": 1304
|
9138 |
+
},
|
9139 |
+
{
|
9140 |
+
"epoch": 0.9854634698886162,
|
9141 |
+
"grad_norm": 0.20331433415412903,
|
9142 |
+
"learning_rate": 1.736979104665637e-08,
|
9143 |
+
"loss": 0.376,
|
9144 |
+
"step": 1305
|
9145 |
+
},
|
9146 |
+
{
|
9147 |
+
"epoch": 0.9862186143099868,
|
9148 |
+
"grad_norm": 0.21076466143131256,
|
9149 |
+
"learning_rate": 1.4355343472932659e-08,
|
9150 |
+
"loss": 0.4072,
|
9151 |
+
"step": 1306
|
9152 |
+
},
|
9153 |
+
{
|
9154 |
+
"epoch": 0.9869737587313574,
|
9155 |
+
"grad_norm": 0.22106704115867615,
|
9156 |
+
"learning_rate": 1.1627933936464441e-08,
|
9157 |
+
"loss": 0.4331,
|
9158 |
+
"step": 1307
|
9159 |
+
},
|
9160 |
+
{
|
9161 |
+
"epoch": 0.987728903152728,
|
9162 |
+
"grad_norm": 0.22276969254016876,
|
9163 |
+
"learning_rate": 9.187578099140171e-09,
|
9164 |
+
"loss": 0.4089,
|
9165 |
+
"step": 1308
|
9166 |
+
},
|
9167 |
+
{
|
9168 |
+
"epoch": 0.9884840475740986,
|
9169 |
+
"grad_norm": 0.2439909279346466,
|
9170 |
+
"learning_rate": 7.034289974483521e-09,
|
9171 |
+
"loss": 0.4358,
|
9172 |
+
"step": 1309
|
9173 |
+
},
|
9174 |
+
{
|
9175 |
+
"epoch": 0.9892391919954692,
|
9176 |
+
"grad_norm": 0.22721384465694427,
|
9177 |
+
"learning_rate": 5.168081927564572e-09,
|
9178 |
+
"loss": 0.3826,
|
9179 |
+
"step": 1310
|
9180 |
+
},
|
9181 |
+
{
|
9182 |
+
"epoch": 0.9899943364168398,
|
9183 |
+
"grad_norm": 0.2524462938308716,
|
9184 |
+
"learning_rate": 3.5889646749109886e-09,
|
9185 |
+
"loss": 0.4407,
|
9186 |
+
"step": 1311
|
9187 |
+
},
|
9188 |
+
{
|
9189 |
+
"epoch": 0.9907494808382104,
|
9190 |
+
"grad_norm": 0.25080129504203796,
|
9191 |
+
"learning_rate": 2.296947284452511e-09,
|
9192 |
+
"loss": 0.4524,
|
9193 |
+
"step": 1312
|
9194 |
+
},
|
9195 |
+
{
|
9196 |
+
"epoch": 0.9915046252595809,
|
9197 |
+
"grad_norm": 0.24871572852134705,
|
9198 |
+
"learning_rate": 1.2920371754931991e-09,
|
9199 |
+
"loss": 0.3901,
|
9200 |
+
"step": 1313
|
9201 |
+
},
|
9202 |
+
{
|
9203 |
+
"epoch": 0.9922597696809515,
|
9204 |
+
"grad_norm": 0.2635830342769623,
|
9205 |
+
"learning_rate": 5.742401186281665e-10,
|
9206 |
+
"loss": 0.4563,
|
9207 |
+
"step": 1314
|
9208 |
+
},
|
9209 |
+
{
|
9210 |
+
"epoch": 0.9930149141023221,
|
9211 |
+
"grad_norm": 0.2695348262786865,
|
9212 |
+
"learning_rate": 1.435602357546806e-10,
|
9213 |
+
"loss": 0.4125,
|
9214 |
+
"step": 1315
|
9215 |
+
},
|
9216 |
+
{
|
9217 |
+
"epoch": 0.9937700585236926,
|
9218 |
+
"grad_norm": 0.2821287214756012,
|
9219 |
+
"learning_rate": 0.0,
|
9220 |
+
"loss": 0.4672,
|
9221 |
+
"step": 1316
|
9222 |
}
|
9223 |
],
|
9224 |
"logging_steps": 1,
|
|
|
9233 |
"should_evaluate": false,
|
9234 |
"should_log": false,
|
9235 |
"should_save": true,
|
9236 |
+
"should_training_stop": true
|
9237 |
},
|
9238 |
"attributes": {}
|
9239 |
}
|
9240 |
},
|
9241 |
+
"total_flos": 1.6951719820787712e+18,
|
9242 |
"train_batch_size": 4,
|
9243 |
"trial_name": null,
|
9244 |
"trial_params": null
|