Training in progress, step 1325, 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 sha256:715f83572aa3f4ed79e33403950f80eee9fccf552b83ef4a77bf113947f7ac7d
|
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 sha256:99b71d30026df7e8e05cad7c4763e931cb252ef5a7a6768252eb9836319a5ed0
|
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 sha256:302845160bac916181403545ad8e36243268631aeb7740ad441789eaf18357d9
|
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:ba54fc85a19d04f74315f4fc03468ec40e7a52d310f34854d7b1cd3ab924f698
|
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":
|
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,
|
@@ -6979,6 +6979,2317 @@
|
|
6979 |
"learning_rate": 1.4560615642794517e-05,
|
6980 |
"loss": 0.3108,
|
6981 |
"step": 996
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
6982 |
}
|
6983 |
],
|
6984 |
"logging_steps": 1,
|
@@ -6993,12 +9304,12 @@
|
|
6993 |
"should_evaluate": false,
|
6994 |
"should_log": false,
|
6995 |
"should_save": true,
|
6996 |
-
"should_training_stop":
|
6997 |
},
|
6998 |
"attributes": {}
|
6999 |
}
|
7000 |
},
|
7001 |
-
"total_flos": 1.
|
7002 |
"train_batch_size": 4,
|
7003 |
"trial_name": null,
|
7004 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 1.000566358316028,
|
5 |
"eval_steps": 500,
|
6 |
+
"global_step": 1325,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
6979 |
"learning_rate": 1.4560615642794517e-05,
|
6980 |
"loss": 0.3108,
|
6981 |
"step": 996
|
6982 |
+
},
|
6983 |
+
{
|
6984 |
+
"epoch": 0.7528789881064754,
|
6985 |
+
"grad_norm": 0.39953500032424927,
|
6986 |
+
"learning_rate": 1.4476771086731567e-05,
|
6987 |
+
"loss": 0.4049,
|
6988 |
+
"step": 997
|
6989 |
+
},
|
6990 |
+
{
|
6991 |
+
"epoch": 0.753634132527846,
|
6992 |
+
"grad_norm": 0.35916373133659363,
|
6993 |
+
"learning_rate": 1.4393127747410417e-05,
|
6994 |
+
"loss": 0.335,
|
6995 |
+
"step": 998
|
6996 |
+
},
|
6997 |
+
{
|
6998 |
+
"epoch": 0.7543892769492165,
|
6999 |
+
"grad_norm": 0.3811121881008148,
|
7000 |
+
"learning_rate": 1.4309686098617975e-05,
|
7001 |
+
"loss": 0.3608,
|
7002 |
+
"step": 999
|
7003 |
+
},
|
7004 |
+
{
|
7005 |
+
"epoch": 0.7551444213705871,
|
7006 |
+
"grad_norm": 0.49154841899871826,
|
7007 |
+
"learning_rate": 1.4226446612998673e-05,
|
7008 |
+
"loss": 0.4254,
|
7009 |
+
"step": 1000
|
7010 |
+
},
|
7011 |
+
{
|
7012 |
+
"epoch": 0.7558995657919577,
|
7013 |
+
"grad_norm": 0.14996010065078735,
|
7014 |
+
"learning_rate": 1.414340976205183e-05,
|
7015 |
+
"loss": 0.3259,
|
7016 |
+
"step": 1001
|
7017 |
+
},
|
7018 |
+
{
|
7019 |
+
"epoch": 0.7566547102133283,
|
7020 |
+
"grad_norm": 0.15046906471252441,
|
7021 |
+
"learning_rate": 1.4060576016128974e-05,
|
7022 |
+
"loss": 0.3184,
|
7023 |
+
"step": 1002
|
7024 |
+
},
|
7025 |
+
{
|
7026 |
+
"epoch": 0.7574098546346989,
|
7027 |
+
"grad_norm": 0.1714065670967102,
|
7028 |
+
"learning_rate": 1.3977945844431118e-05,
|
7029 |
+
"loss": 0.3564,
|
7030 |
+
"step": 1003
|
7031 |
+
},
|
7032 |
+
{
|
7033 |
+
"epoch": 0.7581649990560695,
|
7034 |
+
"grad_norm": 0.18413116037845612,
|
7035 |
+
"learning_rate": 1.3895519715006238e-05,
|
7036 |
+
"loss": 0.3889,
|
7037 |
+
"step": 1004
|
7038 |
+
},
|
7039 |
+
{
|
7040 |
+
"epoch": 0.7589201434774401,
|
7041 |
+
"grad_norm": 0.19010891020298004,
|
7042 |
+
"learning_rate": 1.3813298094746491e-05,
|
7043 |
+
"loss": 0.3909,
|
7044 |
+
"step": 1005
|
7045 |
+
},
|
7046 |
+
{
|
7047 |
+
"epoch": 0.7596752878988107,
|
7048 |
+
"grad_norm": 0.1898065060377121,
|
7049 |
+
"learning_rate": 1.373128144938563e-05,
|
7050 |
+
"loss": 0.4388,
|
7051 |
+
"step": 1006
|
7052 |
+
},
|
7053 |
+
{
|
7054 |
+
"epoch": 0.7604304323201813,
|
7055 |
+
"grad_norm": 0.1871468871831894,
|
7056 |
+
"learning_rate": 1.3649470243496326e-05,
|
7057 |
+
"loss": 0.401,
|
7058 |
+
"step": 1007
|
7059 |
+
},
|
7060 |
+
{
|
7061 |
+
"epoch": 0.7611855767415519,
|
7062 |
+
"grad_norm": 0.18679498136043549,
|
7063 |
+
"learning_rate": 1.3567864940487584e-05,
|
7064 |
+
"loss": 0.4038,
|
7065 |
+
"step": 1008
|
7066 |
+
},
|
7067 |
+
{
|
7068 |
+
"epoch": 0.7619407211629224,
|
7069 |
+
"grad_norm": 0.19137872755527496,
|
7070 |
+
"learning_rate": 1.3486466002602133e-05,
|
7071 |
+
"loss": 0.3803,
|
7072 |
+
"step": 1009
|
7073 |
+
},
|
7074 |
+
{
|
7075 |
+
"epoch": 0.762695865584293,
|
7076 |
+
"grad_norm": 0.1978340446949005,
|
7077 |
+
"learning_rate": 1.340527389091374e-05,
|
7078 |
+
"loss": 0.388,
|
7079 |
+
"step": 1010
|
7080 |
+
},
|
7081 |
+
{
|
7082 |
+
"epoch": 0.7634510100056636,
|
7083 |
+
"grad_norm": 0.19116266071796417,
|
7084 |
+
"learning_rate": 1.3324289065324608e-05,
|
7085 |
+
"loss": 0.3728,
|
7086 |
+
"step": 1011
|
7087 |
+
},
|
7088 |
+
{
|
7089 |
+
"epoch": 0.7642061544270342,
|
7090 |
+
"grad_norm": 0.21141821146011353,
|
7091 |
+
"learning_rate": 1.3243511984562824e-05,
|
7092 |
+
"loss": 0.4367,
|
7093 |
+
"step": 1012
|
7094 |
+
},
|
7095 |
+
{
|
7096 |
+
"epoch": 0.7649612988484048,
|
7097 |
+
"grad_norm": 0.22845213115215302,
|
7098 |
+
"learning_rate": 1.3162943106179749e-05,
|
7099 |
+
"loss": 0.3907,
|
7100 |
+
"step": 1013
|
7101 |
+
},
|
7102 |
+
{
|
7103 |
+
"epoch": 0.7657164432697754,
|
7104 |
+
"grad_norm": 0.22269576787948608,
|
7105 |
+
"learning_rate": 1.3082582886547395e-05,
|
7106 |
+
"loss": 0.4779,
|
7107 |
+
"step": 1014
|
7108 |
+
},
|
7109 |
+
{
|
7110 |
+
"epoch": 0.766471587691146,
|
7111 |
+
"grad_norm": 0.21351350843906403,
|
7112 |
+
"learning_rate": 1.3002431780855817e-05,
|
7113 |
+
"loss": 0.4206,
|
7114 |
+
"step": 1015
|
7115 |
+
},
|
7116 |
+
{
|
7117 |
+
"epoch": 0.7672267321125166,
|
7118 |
+
"grad_norm": 0.2109295129776001,
|
7119 |
+
"learning_rate": 1.2922490243110614e-05,
|
7120 |
+
"loss": 0.3882,
|
7121 |
+
"step": 1016
|
7122 |
+
},
|
7123 |
+
{
|
7124 |
+
"epoch": 0.767981876533887,
|
7125 |
+
"grad_norm": 0.23167261481285095,
|
7126 |
+
"learning_rate": 1.2842758726130283e-05,
|
7127 |
+
"loss": 0.4386,
|
7128 |
+
"step": 1017
|
7129 |
+
},
|
7130 |
+
{
|
7131 |
+
"epoch": 0.7687370209552576,
|
7132 |
+
"grad_norm": 0.23334629833698273,
|
7133 |
+
"learning_rate": 1.2763237681543732e-05,
|
7134 |
+
"loss": 0.4477,
|
7135 |
+
"step": 1018
|
7136 |
+
},
|
7137 |
+
{
|
7138 |
+
"epoch": 0.7694921653766282,
|
7139 |
+
"grad_norm": 0.23084315657615662,
|
7140 |
+
"learning_rate": 1.2683927559787655e-05,
|
7141 |
+
"loss": 0.423,
|
7142 |
+
"step": 1019
|
7143 |
+
},
|
7144 |
+
{
|
7145 |
+
"epoch": 0.7702473097979988,
|
7146 |
+
"grad_norm": 0.2281108796596527,
|
7147 |
+
"learning_rate": 1.2604828810103957e-05,
|
7148 |
+
"loss": 0.4073,
|
7149 |
+
"step": 1020
|
7150 |
+
},
|
7151 |
+
{
|
7152 |
+
"epoch": 0.7710024542193694,
|
7153 |
+
"grad_norm": 0.2173498570919037,
|
7154 |
+
"learning_rate": 1.2525941880537307e-05,
|
7155 |
+
"loss": 0.3423,
|
7156 |
+
"step": 1021
|
7157 |
+
},
|
7158 |
+
{
|
7159 |
+
"epoch": 0.77175759864074,
|
7160 |
+
"grad_norm": 0.21901051700115204,
|
7161 |
+
"learning_rate": 1.2447267217932507e-05,
|
7162 |
+
"loss": 0.3601,
|
7163 |
+
"step": 1022
|
7164 |
+
},
|
7165 |
+
{
|
7166 |
+
"epoch": 0.7725127430621106,
|
7167 |
+
"grad_norm": 0.22421492636203766,
|
7168 |
+
"learning_rate": 1.236880526793207e-05,
|
7169 |
+
"loss": 0.3953,
|
7170 |
+
"step": 1023
|
7171 |
+
},
|
7172 |
+
{
|
7173 |
+
"epoch": 0.7732678874834812,
|
7174 |
+
"grad_norm": 0.2377631813287735,
|
7175 |
+
"learning_rate": 1.2290556474973536e-05,
|
7176 |
+
"loss": 0.3744,
|
7177 |
+
"step": 1024
|
7178 |
+
},
|
7179 |
+
{
|
7180 |
+
"epoch": 0.7740230319048518,
|
7181 |
+
"grad_norm": 0.23121589422225952,
|
7182 |
+
"learning_rate": 1.2212521282287092e-05,
|
7183 |
+
"loss": 0.3976,
|
7184 |
+
"step": 1025
|
7185 |
+
},
|
7186 |
+
{
|
7187 |
+
"epoch": 0.7747781763262224,
|
7188 |
+
"grad_norm": 0.255655437707901,
|
7189 |
+
"learning_rate": 1.2134700131893012e-05,
|
7190 |
+
"loss": 0.4125,
|
7191 |
+
"step": 1026
|
7192 |
+
},
|
7193 |
+
{
|
7194 |
+
"epoch": 0.775533320747593,
|
7195 |
+
"grad_norm": 0.2366100549697876,
|
7196 |
+
"learning_rate": 1.2057093464599157e-05,
|
7197 |
+
"loss": 0.3724,
|
7198 |
+
"step": 1027
|
7199 |
+
},
|
7200 |
+
{
|
7201 |
+
"epoch": 0.7762884651689635,
|
7202 |
+
"grad_norm": 0.2489083856344223,
|
7203 |
+
"learning_rate": 1.1979701719998453e-05,
|
7204 |
+
"loss": 0.4281,
|
7205 |
+
"step": 1028
|
7206 |
+
},
|
7207 |
+
{
|
7208 |
+
"epoch": 0.7770436095903341,
|
7209 |
+
"grad_norm": 0.26776382327079773,
|
7210 |
+
"learning_rate": 1.1902525336466464e-05,
|
7211 |
+
"loss": 0.4041,
|
7212 |
+
"step": 1029
|
7213 |
+
},
|
7214 |
+
{
|
7215 |
+
"epoch": 0.7777987540117047,
|
7216 |
+
"grad_norm": 0.24452626705169678,
|
7217 |
+
"learning_rate": 1.1825564751158823e-05,
|
7218 |
+
"loss": 0.4135,
|
7219 |
+
"step": 1030
|
7220 |
+
},
|
7221 |
+
{
|
7222 |
+
"epoch": 0.7785538984330753,
|
7223 |
+
"grad_norm": 0.2541411817073822,
|
7224 |
+
"learning_rate": 1.1748820400008843e-05,
|
7225 |
+
"loss": 0.4086,
|
7226 |
+
"step": 1031
|
7227 |
+
},
|
7228 |
+
{
|
7229 |
+
"epoch": 0.7793090428544459,
|
7230 |
+
"grad_norm": 0.27573496103286743,
|
7231 |
+
"learning_rate": 1.167229271772498e-05,
|
7232 |
+
"loss": 0.3883,
|
7233 |
+
"step": 1032
|
7234 |
+
},
|
7235 |
+
{
|
7236 |
+
"epoch": 0.7800641872758165,
|
7237 |
+
"grad_norm": 0.2743297219276428,
|
7238 |
+
"learning_rate": 1.1595982137788403e-05,
|
7239 |
+
"loss": 0.4083,
|
7240 |
+
"step": 1033
|
7241 |
+
},
|
7242 |
+
{
|
7243 |
+
"epoch": 0.7808193316971871,
|
7244 |
+
"grad_norm": 0.2688688039779663,
|
7245 |
+
"learning_rate": 1.1519889092450542e-05,
|
7246 |
+
"loss": 0.3992,
|
7247 |
+
"step": 1034
|
7248 |
+
},
|
7249 |
+
{
|
7250 |
+
"epoch": 0.7815744761185577,
|
7251 |
+
"grad_norm": 0.28381526470184326,
|
7252 |
+
"learning_rate": 1.144401401273062e-05,
|
7253 |
+
"loss": 0.3882,
|
7254 |
+
"step": 1035
|
7255 |
+
},
|
7256 |
+
{
|
7257 |
+
"epoch": 0.7823296205399283,
|
7258 |
+
"grad_norm": 0.26710647344589233,
|
7259 |
+
"learning_rate": 1.1368357328413242e-05,
|
7260 |
+
"loss": 0.3639,
|
7261 |
+
"step": 1036
|
7262 |
+
},
|
7263 |
+
{
|
7264 |
+
"epoch": 0.7830847649612989,
|
7265 |
+
"grad_norm": 0.3097337782382965,
|
7266 |
+
"learning_rate": 1.1292919468045877e-05,
|
7267 |
+
"loss": 0.4394,
|
7268 |
+
"step": 1037
|
7269 |
+
},
|
7270 |
+
{
|
7271 |
+
"epoch": 0.7838399093826695,
|
7272 |
+
"grad_norm": 0.28815412521362305,
|
7273 |
+
"learning_rate": 1.1217700858936587e-05,
|
7274 |
+
"loss": 0.4298,
|
7275 |
+
"step": 1038
|
7276 |
+
},
|
7277 |
+
{
|
7278 |
+
"epoch": 0.78459505380404,
|
7279 |
+
"grad_norm": 0.30151909589767456,
|
7280 |
+
"learning_rate": 1.1142701927151456e-05,
|
7281 |
+
"loss": 0.393,
|
7282 |
+
"step": 1039
|
7283 |
+
},
|
7284 |
+
{
|
7285 |
+
"epoch": 0.7853501982254106,
|
7286 |
+
"grad_norm": 0.30772241950035095,
|
7287 |
+
"learning_rate": 1.1067923097512256e-05,
|
7288 |
+
"loss": 0.3688,
|
7289 |
+
"step": 1040
|
7290 |
+
},
|
7291 |
+
{
|
7292 |
+
"epoch": 0.7861053426467812,
|
7293 |
+
"grad_norm": 0.31982895731925964,
|
7294 |
+
"learning_rate": 1.099336479359398e-05,
|
7295 |
+
"loss": 0.3815,
|
7296 |
+
"step": 1041
|
7297 |
+
},
|
7298 |
+
{
|
7299 |
+
"epoch": 0.7868604870681518,
|
7300 |
+
"grad_norm": 0.3298172950744629,
|
7301 |
+
"learning_rate": 1.0919027437722513e-05,
|
7302 |
+
"loss": 0.4153,
|
7303 |
+
"step": 1042
|
7304 |
+
},
|
7305 |
+
{
|
7306 |
+
"epoch": 0.7876156314895224,
|
7307 |
+
"grad_norm": 0.33412277698516846,
|
7308 |
+
"learning_rate": 1.0844911450972229e-05,
|
7309 |
+
"loss": 0.3972,
|
7310 |
+
"step": 1043
|
7311 |
+
},
|
7312 |
+
{
|
7313 |
+
"epoch": 0.788370775910893,
|
7314 |
+
"grad_norm": 0.3366442322731018,
|
7315 |
+
"learning_rate": 1.0771017253163568e-05,
|
7316 |
+
"loss": 0.3627,
|
7317 |
+
"step": 1044
|
7318 |
+
},
|
7319 |
+
{
|
7320 |
+
"epoch": 0.7891259203322636,
|
7321 |
+
"grad_norm": 0.3680926263332367,
|
7322 |
+
"learning_rate": 1.0697345262860636e-05,
|
7323 |
+
"loss": 0.4297,
|
7324 |
+
"step": 1045
|
7325 |
+
},
|
7326 |
+
{
|
7327 |
+
"epoch": 0.7898810647536342,
|
7328 |
+
"grad_norm": 0.3372995853424072,
|
7329 |
+
"learning_rate": 1.0623895897368913e-05,
|
7330 |
+
"loss": 0.3856,
|
7331 |
+
"step": 1046
|
7332 |
+
},
|
7333 |
+
{
|
7334 |
+
"epoch": 0.7906362091750048,
|
7335 |
+
"grad_norm": 0.34899917244911194,
|
7336 |
+
"learning_rate": 1.0550669572732863e-05,
|
7337 |
+
"loss": 0.2923,
|
7338 |
+
"step": 1047
|
7339 |
+
},
|
7340 |
+
{
|
7341 |
+
"epoch": 0.7913913535963754,
|
7342 |
+
"grad_norm": 0.4165075421333313,
|
7343 |
+
"learning_rate": 1.0477666703733541e-05,
|
7344 |
+
"loss": 0.3788,
|
7345 |
+
"step": 1048
|
7346 |
+
},
|
7347 |
+
{
|
7348 |
+
"epoch": 0.792146498017746,
|
7349 |
+
"grad_norm": 0.44895628094673157,
|
7350 |
+
"learning_rate": 1.0404887703886251e-05,
|
7351 |
+
"loss": 0.373,
|
7352 |
+
"step": 1049
|
7353 |
+
},
|
7354 |
+
{
|
7355 |
+
"epoch": 0.7929016424391164,
|
7356 |
+
"grad_norm": 0.4823060631752014,
|
7357 |
+
"learning_rate": 1.0332332985438248e-05,
|
7358 |
+
"loss": 0.3716,
|
7359 |
+
"step": 1050
|
7360 |
+
},
|
7361 |
+
{
|
7362 |
+
"epoch": 0.793656786860487,
|
7363 |
+
"grad_norm": 0.15826448798179626,
|
7364 |
+
"learning_rate": 1.0260002959366349e-05,
|
7365 |
+
"loss": 0.3269,
|
7366 |
+
"step": 1051
|
7367 |
+
},
|
7368 |
+
{
|
7369 |
+
"epoch": 0.7944119312818576,
|
7370 |
+
"grad_norm": 0.1592281609773636,
|
7371 |
+
"learning_rate": 1.0187898035374682e-05,
|
7372 |
+
"loss": 0.3417,
|
7373 |
+
"step": 1052
|
7374 |
+
},
|
7375 |
+
{
|
7376 |
+
"epoch": 0.7951670757032282,
|
7377 |
+
"grad_norm": 0.18132025003433228,
|
7378 |
+
"learning_rate": 1.0116018621892237e-05,
|
7379 |
+
"loss": 0.3531,
|
7380 |
+
"step": 1053
|
7381 |
+
},
|
7382 |
+
{
|
7383 |
+
"epoch": 0.7959222201245988,
|
7384 |
+
"grad_norm": 0.16262286901474,
|
7385 |
+
"learning_rate": 1.0044365126070682e-05,
|
7386 |
+
"loss": 0.3089,
|
7387 |
+
"step": 1054
|
7388 |
+
},
|
7389 |
+
{
|
7390 |
+
"epoch": 0.7966773645459694,
|
7391 |
+
"grad_norm": 0.17961286008358002,
|
7392 |
+
"learning_rate": 9.972937953781986e-06,
|
7393 |
+
"loss": 0.3534,
|
7394 |
+
"step": 1055
|
7395 |
+
},
|
7396 |
+
{
|
7397 |
+
"epoch": 0.79743250896734,
|
7398 |
+
"grad_norm": 0.17105191946029663,
|
7399 |
+
"learning_rate": 9.901737509616143e-06,
|
7400 |
+
"loss": 0.3361,
|
7401 |
+
"step": 1056
|
7402 |
+
},
|
7403 |
+
{
|
7404 |
+
"epoch": 0.7981876533887106,
|
7405 |
+
"grad_norm": 0.1858292818069458,
|
7406 |
+
"learning_rate": 9.830764196878872e-06,
|
7407 |
+
"loss": 0.354,
|
7408 |
+
"step": 1057
|
7409 |
+
},
|
7410 |
+
{
|
7411 |
+
"epoch": 0.7989427978100812,
|
7412 |
+
"grad_norm": 0.19711358845233917,
|
7413 |
+
"learning_rate": 9.760018417589334e-06,
|
7414 |
+
"loss": 0.3887,
|
7415 |
+
"step": 1058
|
7416 |
+
},
|
7417 |
+
{
|
7418 |
+
"epoch": 0.7996979422314517,
|
7419 |
+
"grad_norm": 0.20733587443828583,
|
7420 |
+
"learning_rate": 9.689500572477855e-06,
|
7421 |
+
"loss": 0.4724,
|
7422 |
+
"step": 1059
|
7423 |
+
},
|
7424 |
+
{
|
7425 |
+
"epoch": 0.8004530866528223,
|
7426 |
+
"grad_norm": 0.2481202483177185,
|
7427 |
+
"learning_rate": 9.619211060983675e-06,
|
7428 |
+
"loss": 0.4828,
|
7429 |
+
"step": 1060
|
7430 |
+
},
|
7431 |
+
{
|
7432 |
+
"epoch": 0.8012082310741929,
|
7433 |
+
"grad_norm": 0.191118523478508,
|
7434 |
+
"learning_rate": 9.549150281252633e-06,
|
7435 |
+
"loss": 0.3909,
|
7436 |
+
"step": 1061
|
7437 |
+
},
|
7438 |
+
{
|
7439 |
+
"epoch": 0.8019633754955635,
|
7440 |
+
"grad_norm": 0.19519171118736267,
|
7441 |
+
"learning_rate": 9.479318630134976e-06,
|
7442 |
+
"loss": 0.339,
|
7443 |
+
"step": 1062
|
7444 |
+
},
|
7445 |
+
{
|
7446 |
+
"epoch": 0.8027185199169341,
|
7447 |
+
"grad_norm": 0.19820590317249298,
|
7448 |
+
"learning_rate": 9.409716503183074e-06,
|
7449 |
+
"loss": 0.3476,
|
7450 |
+
"step": 1063
|
7451 |
+
},
|
7452 |
+
{
|
7453 |
+
"epoch": 0.8034736643383047,
|
7454 |
+
"grad_norm": 0.23485320806503296,
|
7455 |
+
"learning_rate": 9.340344294649184e-06,
|
7456 |
+
"loss": 0.4675,
|
7457 |
+
"step": 1064
|
7458 |
+
},
|
7459 |
+
{
|
7460 |
+
"epoch": 0.8042288087596753,
|
7461 |
+
"grad_norm": 0.20198017358779907,
|
7462 |
+
"learning_rate": 9.271202397483215e-06,
|
7463 |
+
"loss": 0.336,
|
7464 |
+
"step": 1065
|
7465 |
+
},
|
7466 |
+
{
|
7467 |
+
"epoch": 0.8049839531810459,
|
7468 |
+
"grad_norm": 0.19426412880420685,
|
7469 |
+
"learning_rate": 9.20229120333052e-06,
|
7470 |
+
"loss": 0.3578,
|
7471 |
+
"step": 1066
|
7472 |
+
},
|
7473 |
+
{
|
7474 |
+
"epoch": 0.8057390976024165,
|
7475 |
+
"grad_norm": 0.2336643636226654,
|
7476 |
+
"learning_rate": 9.133611102529654e-06,
|
7477 |
+
"loss": 0.4355,
|
7478 |
+
"step": 1067
|
7479 |
+
},
|
7480 |
+
{
|
7481 |
+
"epoch": 0.8064942420237871,
|
7482 |
+
"grad_norm": 0.2223149985074997,
|
7483 |
+
"learning_rate": 9.065162484110179e-06,
|
7484 |
+
"loss": 0.4256,
|
7485 |
+
"step": 1068
|
7486 |
+
},
|
7487 |
+
{
|
7488 |
+
"epoch": 0.8072493864451576,
|
7489 |
+
"grad_norm": 0.23664018511772156,
|
7490 |
+
"learning_rate": 8.996945735790447e-06,
|
7491 |
+
"loss": 0.4148,
|
7492 |
+
"step": 1069
|
7493 |
+
},
|
7494 |
+
{
|
7495 |
+
"epoch": 0.8080045308665282,
|
7496 |
+
"grad_norm": 0.22716124355793,
|
7497 |
+
"learning_rate": 8.928961243975437e-06,
|
7498 |
+
"loss": 0.3981,
|
7499 |
+
"step": 1070
|
7500 |
+
},
|
7501 |
+
{
|
7502 |
+
"epoch": 0.8087596752878988,
|
7503 |
+
"grad_norm": 0.230534628033638,
|
7504 |
+
"learning_rate": 8.861209393754477e-06,
|
7505 |
+
"loss": 0.4269,
|
7506 |
+
"step": 1071
|
7507 |
+
},
|
7508 |
+
{
|
7509 |
+
"epoch": 0.8095148197092694,
|
7510 |
+
"grad_norm": 0.21818408370018005,
|
7511 |
+
"learning_rate": 8.793690568899216e-06,
|
7512 |
+
"loss": 0.3498,
|
7513 |
+
"step": 1072
|
7514 |
+
},
|
7515 |
+
{
|
7516 |
+
"epoch": 0.81026996413064,
|
7517 |
+
"grad_norm": 0.2515822947025299,
|
7518 |
+
"learning_rate": 8.7264051518613e-06,
|
7519 |
+
"loss": 0.4829,
|
7520 |
+
"step": 1073
|
7521 |
+
},
|
7522 |
+
{
|
7523 |
+
"epoch": 0.8110251085520106,
|
7524 |
+
"grad_norm": 0.23883438110351562,
|
7525 |
+
"learning_rate": 8.659353523770297e-06,
|
7526 |
+
"loss": 0.3792,
|
7527 |
+
"step": 1074
|
7528 |
+
},
|
7529 |
+
{
|
7530 |
+
"epoch": 0.8117802529733812,
|
7531 |
+
"grad_norm": 0.25294432044029236,
|
7532 |
+
"learning_rate": 8.592536064431467e-06,
|
7533 |
+
"loss": 0.3966,
|
7534 |
+
"step": 1075
|
7535 |
+
},
|
7536 |
+
{
|
7537 |
+
"epoch": 0.8125353973947518,
|
7538 |
+
"grad_norm": 0.2528051435947418,
|
7539 |
+
"learning_rate": 8.525953152323684e-06,
|
7540 |
+
"loss": 0.4245,
|
7541 |
+
"step": 1076
|
7542 |
+
},
|
7543 |
+
{
|
7544 |
+
"epoch": 0.8132905418161224,
|
7545 |
+
"grad_norm": 0.25422972440719604,
|
7546 |
+
"learning_rate": 8.459605164597267e-06,
|
7547 |
+
"loss": 0.4256,
|
7548 |
+
"step": 1077
|
7549 |
+
},
|
7550 |
+
{
|
7551 |
+
"epoch": 0.814045686237493,
|
7552 |
+
"grad_norm": 0.2697378098964691,
|
7553 |
+
"learning_rate": 8.393492477071829e-06,
|
7554 |
+
"loss": 0.4137,
|
7555 |
+
"step": 1078
|
7556 |
+
},
|
7557 |
+
{
|
7558 |
+
"epoch": 0.8148008306588636,
|
7559 |
+
"grad_norm": 0.25492045283317566,
|
7560 |
+
"learning_rate": 8.327615464234129e-06,
|
7561 |
+
"loss": 0.4055,
|
7562 |
+
"step": 1079
|
7563 |
+
},
|
7564 |
+
{
|
7565 |
+
"epoch": 0.8155559750802341,
|
7566 |
+
"grad_norm": 0.26645827293395996,
|
7567 |
+
"learning_rate": 8.261974499235987e-06,
|
7568 |
+
"loss": 0.4531,
|
7569 |
+
"step": 1080
|
7570 |
+
},
|
7571 |
+
{
|
7572 |
+
"epoch": 0.8163111195016047,
|
7573 |
+
"grad_norm": 0.2661876082420349,
|
7574 |
+
"learning_rate": 8.196569953892202e-06,
|
7575 |
+
"loss": 0.3774,
|
7576 |
+
"step": 1081
|
7577 |
+
},
|
7578 |
+
{
|
7579 |
+
"epoch": 0.8170662639229753,
|
7580 |
+
"grad_norm": 0.2471131682395935,
|
7581 |
+
"learning_rate": 8.131402198678373e-06,
|
7582 |
+
"loss": 0.3474,
|
7583 |
+
"step": 1082
|
7584 |
+
},
|
7585 |
+
{
|
7586 |
+
"epoch": 0.8178214083443458,
|
7587 |
+
"grad_norm": 0.26696231961250305,
|
7588 |
+
"learning_rate": 8.066471602728803e-06,
|
7589 |
+
"loss": 0.3357,
|
7590 |
+
"step": 1083
|
7591 |
+
},
|
7592 |
+
{
|
7593 |
+
"epoch": 0.8185765527657164,
|
7594 |
+
"grad_norm": 0.26743122935295105,
|
7595 |
+
"learning_rate": 8.001778533834487e-06,
|
7596 |
+
"loss": 0.3404,
|
7597 |
+
"step": 1084
|
7598 |
+
},
|
7599 |
+
{
|
7600 |
+
"epoch": 0.819331697187087,
|
7601 |
+
"grad_norm": 0.28732678294181824,
|
7602 |
+
"learning_rate": 7.937323358440935e-06,
|
7603 |
+
"loss": 0.389,
|
7604 |
+
"step": 1085
|
7605 |
+
},
|
7606 |
+
{
|
7607 |
+
"epoch": 0.8200868416084576,
|
7608 |
+
"grad_norm": 0.30629798769950867,
|
7609 |
+
"learning_rate": 7.873106441646205e-06,
|
7610 |
+
"loss": 0.4185,
|
7611 |
+
"step": 1086
|
7612 |
+
},
|
7613 |
+
{
|
7614 |
+
"epoch": 0.8208419860298282,
|
7615 |
+
"grad_norm": 0.2828892469406128,
|
7616 |
+
"learning_rate": 7.809128147198691e-06,
|
7617 |
+
"loss": 0.3792,
|
7618 |
+
"step": 1087
|
7619 |
+
},
|
7620 |
+
{
|
7621 |
+
"epoch": 0.8215971304511988,
|
7622 |
+
"grad_norm": 0.28884345293045044,
|
7623 |
+
"learning_rate": 7.745388837495188e-06,
|
7624 |
+
"loss": 0.369,
|
7625 |
+
"step": 1088
|
7626 |
+
},
|
7627 |
+
{
|
7628 |
+
"epoch": 0.8223522748725693,
|
7629 |
+
"grad_norm": 0.30468007922172546,
|
7630 |
+
"learning_rate": 7.681888873578786e-06,
|
7631 |
+
"loss": 0.4518,
|
7632 |
+
"step": 1089
|
7633 |
+
},
|
7634 |
+
{
|
7635 |
+
"epoch": 0.8231074192939399,
|
7636 |
+
"grad_norm": 0.3138682246208191,
|
7637 |
+
"learning_rate": 7.618628615136825e-06,
|
7638 |
+
"loss": 0.3665,
|
7639 |
+
"step": 1090
|
7640 |
+
},
|
7641 |
+
{
|
7642 |
+
"epoch": 0.8238625637153105,
|
7643 |
+
"grad_norm": 0.2910728454589844,
|
7644 |
+
"learning_rate": 7.555608420498872e-06,
|
7645 |
+
"loss": 0.2928,
|
7646 |
+
"step": 1091
|
7647 |
+
},
|
7648 |
+
{
|
7649 |
+
"epoch": 0.8246177081366811,
|
7650 |
+
"grad_norm": 0.3152346611022949,
|
7651 |
+
"learning_rate": 7.4928286466346754e-06,
|
7652 |
+
"loss": 0.3834,
|
7653 |
+
"step": 1092
|
7654 |
+
},
|
7655 |
+
{
|
7656 |
+
"epoch": 0.8253728525580517,
|
7657 |
+
"grad_norm": 0.336488276720047,
|
7658 |
+
"learning_rate": 7.430289649152156e-06,
|
7659 |
+
"loss": 0.3728,
|
7660 |
+
"step": 1093
|
7661 |
+
},
|
7662 |
+
{
|
7663 |
+
"epoch": 0.8261279969794223,
|
7664 |
+
"grad_norm": 0.32753413915634155,
|
7665 |
+
"learning_rate": 7.367991782295391e-06,
|
7666 |
+
"loss": 0.3237,
|
7667 |
+
"step": 1094
|
7668 |
+
},
|
7669 |
+
{
|
7670 |
+
"epoch": 0.8268831414007929,
|
7671 |
+
"grad_norm": 0.33121833205223083,
|
7672 |
+
"learning_rate": 7.305935398942598e-06,
|
7673 |
+
"loss": 0.3403,
|
7674 |
+
"step": 1095
|
7675 |
+
},
|
7676 |
+
{
|
7677 |
+
"epoch": 0.8276382858221635,
|
7678 |
+
"grad_norm": 0.3293071389198303,
|
7679 |
+
"learning_rate": 7.244120850604141e-06,
|
7680 |
+
"loss": 0.3105,
|
7681 |
+
"step": 1096
|
7682 |
+
},
|
7683 |
+
{
|
7684 |
+
"epoch": 0.8283934302435341,
|
7685 |
+
"grad_norm": 0.3871884047985077,
|
7686 |
+
"learning_rate": 7.182548487420554e-06,
|
7687 |
+
"loss": 0.3617,
|
7688 |
+
"step": 1097
|
7689 |
+
},
|
7690 |
+
{
|
7691 |
+
"epoch": 0.8291485746649047,
|
7692 |
+
"grad_norm": 0.4038209915161133,
|
7693 |
+
"learning_rate": 7.121218658160527e-06,
|
7694 |
+
"loss": 0.4204,
|
7695 |
+
"step": 1098
|
7696 |
+
},
|
7697 |
+
{
|
7698 |
+
"epoch": 0.8299037190862752,
|
7699 |
+
"grad_norm": 0.41719168424606323,
|
7700 |
+
"learning_rate": 7.060131710218959e-06,
|
7701 |
+
"loss": 0.299,
|
7702 |
+
"step": 1099
|
7703 |
+
},
|
7704 |
+
{
|
7705 |
+
"epoch": 0.8306588635076458,
|
7706 |
+
"grad_norm": 0.4760392904281616,
|
7707 |
+
"learning_rate": 6.999287989614972e-06,
|
7708 |
+
"loss": 0.3683,
|
7709 |
+
"step": 1100
|
7710 |
+
},
|
7711 |
+
{
|
7712 |
+
"epoch": 0.8314140079290164,
|
7713 |
+
"grad_norm": 0.17138616740703583,
|
7714 |
+
"learning_rate": 6.9386878409899715e-06,
|
7715 |
+
"loss": 0.3231,
|
7716 |
+
"step": 1101
|
7717 |
+
},
|
7718 |
+
{
|
7719 |
+
"epoch": 0.832169152350387,
|
7720 |
+
"grad_norm": 0.16634538769721985,
|
7721 |
+
"learning_rate": 6.87833160760567e-06,
|
7722 |
+
"loss": 0.3349,
|
7723 |
+
"step": 1102
|
7724 |
+
},
|
7725 |
+
{
|
7726 |
+
"epoch": 0.8329242967717576,
|
7727 |
+
"grad_norm": 0.17278127372264862,
|
7728 |
+
"learning_rate": 6.818219631342149e-06,
|
7729 |
+
"loss": 0.3614,
|
7730 |
+
"step": 1103
|
7731 |
+
},
|
7732 |
+
{
|
7733 |
+
"epoch": 0.8336794411931282,
|
7734 |
+
"grad_norm": 0.16819556057453156,
|
7735 |
+
"learning_rate": 6.758352252695949e-06,
|
7736 |
+
"loss": 0.3442,
|
7737 |
+
"step": 1104
|
7738 |
+
},
|
7739 |
+
{
|
7740 |
+
"epoch": 0.8344345856144988,
|
7741 |
+
"grad_norm": 0.181631401181221,
|
7742 |
+
"learning_rate": 6.698729810778065e-06,
|
7743 |
+
"loss": 0.3663,
|
7744 |
+
"step": 1105
|
7745 |
+
},
|
7746 |
+
{
|
7747 |
+
"epoch": 0.8351897300358694,
|
7748 |
+
"grad_norm": 0.18373197317123413,
|
7749 |
+
"learning_rate": 6.639352643312164e-06,
|
7750 |
+
"loss": 0.3636,
|
7751 |
+
"step": 1106
|
7752 |
+
},
|
7753 |
+
{
|
7754 |
+
"epoch": 0.83594487445724,
|
7755 |
+
"grad_norm": 0.1827540099620819,
|
7756 |
+
"learning_rate": 6.580221086632516e-06,
|
7757 |
+
"loss": 0.3765,
|
7758 |
+
"step": 1107
|
7759 |
+
},
|
7760 |
+
{
|
7761 |
+
"epoch": 0.8367000188786106,
|
7762 |
+
"grad_norm": 0.1983499974012375,
|
7763 |
+
"learning_rate": 6.521335475682205e-06,
|
7764 |
+
"loss": 0.3806,
|
7765 |
+
"step": 1108
|
7766 |
+
},
|
7767 |
+
{
|
7768 |
+
"epoch": 0.8374551632999812,
|
7769 |
+
"grad_norm": 0.2072797417640686,
|
7770 |
+
"learning_rate": 6.462696144011149e-06,
|
7771 |
+
"loss": 0.4196,
|
7772 |
+
"step": 1109
|
7773 |
+
},
|
7774 |
+
{
|
7775 |
+
"epoch": 0.8382103077213517,
|
7776 |
+
"grad_norm": 0.20540378987789154,
|
7777 |
+
"learning_rate": 6.40430342377426e-06,
|
7778 |
+
"loss": 0.4063,
|
7779 |
+
"step": 1110
|
7780 |
+
},
|
7781 |
+
{
|
7782 |
+
"epoch": 0.8389654521427223,
|
7783 |
+
"grad_norm": 0.22013606131076813,
|
7784 |
+
"learning_rate": 6.346157645729589e-06,
|
7785 |
+
"loss": 0.4732,
|
7786 |
+
"step": 1111
|
7787 |
+
},
|
7788 |
+
{
|
7789 |
+
"epoch": 0.8397205965640929,
|
7790 |
+
"grad_norm": 0.2054942101240158,
|
7791 |
+
"learning_rate": 6.2882591392363795e-06,
|
7792 |
+
"loss": 0.3476,
|
7793 |
+
"step": 1112
|
7794 |
+
},
|
7795 |
+
{
|
7796 |
+
"epoch": 0.8404757409854635,
|
7797 |
+
"grad_norm": 0.22685834765434265,
|
7798 |
+
"learning_rate": 6.230608232253227e-06,
|
7799 |
+
"loss": 0.4091,
|
7800 |
+
"step": 1113
|
7801 |
+
},
|
7802 |
+
{
|
7803 |
+
"epoch": 0.8412308854068341,
|
7804 |
+
"grad_norm": 0.22038882970809937,
|
7805 |
+
"learning_rate": 6.173205251336239e-06,
|
7806 |
+
"loss": 0.4229,
|
7807 |
+
"step": 1114
|
7808 |
+
},
|
7809 |
+
{
|
7810 |
+
"epoch": 0.8419860298282047,
|
7811 |
+
"grad_norm": 0.20709578692913055,
|
7812 |
+
"learning_rate": 6.116050521637218e-06,
|
7813 |
+
"loss": 0.4012,
|
7814 |
+
"step": 1115
|
7815 |
+
},
|
7816 |
+
{
|
7817 |
+
"epoch": 0.8427411742495753,
|
7818 |
+
"grad_norm": 0.2158709317445755,
|
7819 |
+
"learning_rate": 6.059144366901736e-06,
|
7820 |
+
"loss": 0.3793,
|
7821 |
+
"step": 1116
|
7822 |
+
},
|
7823 |
+
{
|
7824 |
+
"epoch": 0.8434963186709458,
|
7825 |
+
"grad_norm": 0.21242888271808624,
|
7826 |
+
"learning_rate": 6.002487109467347e-06,
|
7827 |
+
"loss": 0.334,
|
7828 |
+
"step": 1117
|
7829 |
+
},
|
7830 |
+
{
|
7831 |
+
"epoch": 0.8442514630923164,
|
7832 |
+
"grad_norm": 0.23054109513759613,
|
7833 |
+
"learning_rate": 5.946079070261773e-06,
|
7834 |
+
"loss": 0.4508,
|
7835 |
+
"step": 1118
|
7836 |
+
},
|
7837 |
+
{
|
7838 |
+
"epoch": 0.845006607513687,
|
7839 |
+
"grad_norm": 0.22902311384677887,
|
7840 |
+
"learning_rate": 5.889920568801055e-06,
|
7841 |
+
"loss": 0.4533,
|
7842 |
+
"step": 1119
|
7843 |
+
},
|
7844 |
+
{
|
7845 |
+
"epoch": 0.8457617519350575,
|
7846 |
+
"grad_norm": 0.22745831310749054,
|
7847 |
+
"learning_rate": 5.834011923187805e-06,
|
7848 |
+
"loss": 0.4043,
|
7849 |
+
"step": 1120
|
7850 |
+
},
|
7851 |
+
{
|
7852 |
+
"epoch": 0.8465168963564281,
|
7853 |
+
"grad_norm": 0.22601962089538574,
|
7854 |
+
"learning_rate": 5.778353450109286e-06,
|
7855 |
+
"loss": 0.4465,
|
7856 |
+
"step": 1121
|
7857 |
+
},
|
7858 |
+
{
|
7859 |
+
"epoch": 0.8472720407777987,
|
7860 |
+
"grad_norm": 0.2339319884777069,
|
7861 |
+
"learning_rate": 5.722945464835749e-06,
|
7862 |
+
"loss": 0.3846,
|
7863 |
+
"step": 1122
|
7864 |
+
},
|
7865 |
+
{
|
7866 |
+
"epoch": 0.8480271851991693,
|
7867 |
+
"grad_norm": 0.23972941935062408,
|
7868 |
+
"learning_rate": 5.667788281218567e-06,
|
7869 |
+
"loss": 0.4077,
|
7870 |
+
"step": 1123
|
7871 |
+
},
|
7872 |
+
{
|
7873 |
+
"epoch": 0.8487823296205399,
|
7874 |
+
"grad_norm": 0.24830228090286255,
|
7875 |
+
"learning_rate": 5.61288221168848e-06,
|
7876 |
+
"loss": 0.4068,
|
7877 |
+
"step": 1124
|
7878 |
+
},
|
7879 |
+
{
|
7880 |
+
"epoch": 0.8495374740419105,
|
7881 |
+
"grad_norm": 0.24531161785125732,
|
7882 |
+
"learning_rate": 5.558227567253832e-06,
|
7883 |
+
"loss": 0.3847,
|
7884 |
+
"step": 1125
|
7885 |
+
},
|
7886 |
+
{
|
7887 |
+
"epoch": 0.8502926184632811,
|
7888 |
+
"grad_norm": 0.2520170509815216,
|
7889 |
+
"learning_rate": 5.503824657498785e-06,
|
7890 |
+
"loss": 0.3514,
|
7891 |
+
"step": 1126
|
7892 |
+
},
|
7893 |
+
{
|
7894 |
+
"epoch": 0.8510477628846517,
|
7895 |
+
"grad_norm": 0.24631567299365997,
|
7896 |
+
"learning_rate": 5.449673790581611e-06,
|
7897 |
+
"loss": 0.4191,
|
7898 |
+
"step": 1127
|
7899 |
+
},
|
7900 |
+
{
|
7901 |
+
"epoch": 0.8518029073060223,
|
7902 |
+
"grad_norm": 0.26101672649383545,
|
7903 |
+
"learning_rate": 5.39577527323289e-06,
|
7904 |
+
"loss": 0.4393,
|
7905 |
+
"step": 1128
|
7906 |
+
},
|
7907 |
+
{
|
7908 |
+
"epoch": 0.8525580517273929,
|
7909 |
+
"grad_norm": 0.2639968991279602,
|
7910 |
+
"learning_rate": 5.34212941075381e-06,
|
7911 |
+
"loss": 0.4323,
|
7912 |
+
"step": 1129
|
7913 |
+
},
|
7914 |
+
{
|
7915 |
+
"epoch": 0.8533131961487634,
|
7916 |
+
"grad_norm": 0.2551827132701874,
|
7917 |
+
"learning_rate": 5.288736507014435e-06,
|
7918 |
+
"loss": 0.3638,
|
7919 |
+
"step": 1130
|
7920 |
+
},
|
7921 |
+
{
|
7922 |
+
"epoch": 0.854068340570134,
|
7923 |
+
"grad_norm": 0.24728746712207794,
|
7924 |
+
"learning_rate": 5.235596864451975e-06,
|
7925 |
+
"loss": 0.3579,
|
7926 |
+
"step": 1131
|
7927 |
+
},
|
7928 |
+
{
|
7929 |
+
"epoch": 0.8548234849915046,
|
7930 |
+
"grad_norm": 0.2714408040046692,
|
7931 |
+
"learning_rate": 5.182710784069067e-06,
|
7932 |
+
"loss": 0.4218,
|
7933 |
+
"step": 1132
|
7934 |
+
},
|
7935 |
+
{
|
7936 |
+
"epoch": 0.8555786294128752,
|
7937 |
+
"grad_norm": 0.26579996943473816,
|
7938 |
+
"learning_rate": 5.13007856543209e-06,
|
7939 |
+
"loss": 0.3796,
|
7940 |
+
"step": 1133
|
7941 |
+
},
|
7942 |
+
{
|
7943 |
+
"epoch": 0.8563337738342458,
|
7944 |
+
"grad_norm": 0.2808961868286133,
|
7945 |
+
"learning_rate": 5.077700506669425e-06,
|
7946 |
+
"loss": 0.35,
|
7947 |
+
"step": 1134
|
7948 |
+
},
|
7949 |
+
{
|
7950 |
+
"epoch": 0.8570889182556164,
|
7951 |
+
"grad_norm": 0.30303752422332764,
|
7952 |
+
"learning_rate": 5.025576904469842e-06,
|
7953 |
+
"loss": 0.4059,
|
7954 |
+
"step": 1135
|
7955 |
+
},
|
7956 |
+
{
|
7957 |
+
"epoch": 0.857844062676987,
|
7958 |
+
"grad_norm": 0.30269986391067505,
|
7959 |
+
"learning_rate": 4.97370805408075e-06,
|
7960 |
+
"loss": 0.4018,
|
7961 |
+
"step": 1136
|
7962 |
+
},
|
7963 |
+
{
|
7964 |
+
"epoch": 0.8585992070983576,
|
7965 |
+
"grad_norm": 0.27650511264801025,
|
7966 |
+
"learning_rate": 4.922094249306558e-06,
|
7967 |
+
"loss": 0.3363,
|
7968 |
+
"step": 1137
|
7969 |
+
},
|
7970 |
+
{
|
7971 |
+
"epoch": 0.8593543515197282,
|
7972 |
+
"grad_norm": 0.28319036960601807,
|
7973 |
+
"learning_rate": 4.87073578250698e-06,
|
7974 |
+
"loss": 0.3566,
|
7975 |
+
"step": 1138
|
7976 |
+
},
|
7977 |
+
{
|
7978 |
+
"epoch": 0.8601094959410988,
|
7979 |
+
"grad_norm": 0.31123289465904236,
|
7980 |
+
"learning_rate": 4.819632944595415e-06,
|
7981 |
+
"loss": 0.3427,
|
7982 |
+
"step": 1139
|
7983 |
+
},
|
7984 |
+
{
|
7985 |
+
"epoch": 0.8608646403624693,
|
7986 |
+
"grad_norm": 0.3235510587692261,
|
7987 |
+
"learning_rate": 4.768786025037309e-06,
|
7988 |
+
"loss": 0.3882,
|
7989 |
+
"step": 1140
|
7990 |
+
},
|
7991 |
+
{
|
7992 |
+
"epoch": 0.8616197847838399,
|
7993 |
+
"grad_norm": 0.34232163429260254,
|
7994 |
+
"learning_rate": 4.7181953118484556e-06,
|
7995 |
+
"loss": 0.4307,
|
7996 |
+
"step": 1141
|
7997 |
+
},
|
7998 |
+
{
|
7999 |
+
"epoch": 0.8623749292052105,
|
8000 |
+
"grad_norm": 0.3100459575653076,
|
8001 |
+
"learning_rate": 4.667861091593434e-06,
|
8002 |
+
"loss": 0.3515,
|
8003 |
+
"step": 1142
|
8004 |
+
},
|
8005 |
+
{
|
8006 |
+
"epoch": 0.8631300736265811,
|
8007 |
+
"grad_norm": 0.33639565110206604,
|
8008 |
+
"learning_rate": 4.617783649383905e-06,
|
8009 |
+
"loss": 0.4251,
|
8010 |
+
"step": 1143
|
8011 |
+
},
|
8012 |
+
{
|
8013 |
+
"epoch": 0.8638852180479517,
|
8014 |
+
"grad_norm": 0.30844351649284363,
|
8015 |
+
"learning_rate": 4.567963268877079e-06,
|
8016 |
+
"loss": 0.336,
|
8017 |
+
"step": 1144
|
8018 |
+
},
|
8019 |
+
{
|
8020 |
+
"epoch": 0.8646403624693223,
|
8021 |
+
"grad_norm": 0.3265226483345032,
|
8022 |
+
"learning_rate": 4.5184002322740785e-06,
|
8023 |
+
"loss": 0.3545,
|
8024 |
+
"step": 1145
|
8025 |
+
},
|
8026 |
+
{
|
8027 |
+
"epoch": 0.8653955068906929,
|
8028 |
+
"grad_norm": 0.3475089967250824,
|
8029 |
+
"learning_rate": 4.4690948203183255e-06,
|
8030 |
+
"loss": 0.3436,
|
8031 |
+
"step": 1146
|
8032 |
+
},
|
8033 |
+
{
|
8034 |
+
"epoch": 0.8661506513120635,
|
8035 |
+
"grad_norm": 0.3850986957550049,
|
8036 |
+
"learning_rate": 4.4200473122939456e-06,
|
8037 |
+
"loss": 0.4274,
|
8038 |
+
"step": 1147
|
8039 |
+
},
|
8040 |
+
{
|
8041 |
+
"epoch": 0.8669057957334341,
|
8042 |
+
"grad_norm": 0.4068509340286255,
|
8043 |
+
"learning_rate": 4.371257986024202e-06,
|
8044 |
+
"loss": 0.4225,
|
8045 |
+
"step": 1148
|
8046 |
+
},
|
8047 |
+
{
|
8048 |
+
"epoch": 0.8676609401548047,
|
8049 |
+
"grad_norm": 0.41805300116539,
|
8050 |
+
"learning_rate": 4.322727117869951e-06,
|
8051 |
+
"loss": 0.4207,
|
8052 |
+
"step": 1149
|
8053 |
+
},
|
8054 |
+
{
|
8055 |
+
"epoch": 0.8684160845761751,
|
8056 |
+
"grad_norm": 0.4633561670780182,
|
8057 |
+
"learning_rate": 4.274454982728032e-06,
|
8058 |
+
"loss": 0.3865,
|
8059 |
+
"step": 1150
|
8060 |
+
},
|
8061 |
+
{
|
8062 |
+
"epoch": 0.8691712289975457,
|
8063 |
+
"grad_norm": 0.13895747065544128,
|
8064 |
+
"learning_rate": 4.2264418540297e-06,
|
8065 |
+
"loss": 0.2752,
|
8066 |
+
"step": 1151
|
8067 |
+
},
|
8068 |
+
{
|
8069 |
+
"epoch": 0.8699263734189163,
|
8070 |
+
"grad_norm": 0.16666154563426971,
|
8071 |
+
"learning_rate": 4.178688003739129e-06,
|
8072 |
+
"loss": 0.3396,
|
8073 |
+
"step": 1152
|
8074 |
+
},
|
8075 |
+
{
|
8076 |
+
"epoch": 0.8706815178402869,
|
8077 |
+
"grad_norm": 0.1707499921321869,
|
8078 |
+
"learning_rate": 4.131193702351827e-06,
|
8079 |
+
"loss": 0.3197,
|
8080 |
+
"step": 1153
|
8081 |
+
},
|
8082 |
+
{
|
8083 |
+
"epoch": 0.8714366622616575,
|
8084 |
+
"grad_norm": 0.176160529255867,
|
8085 |
+
"learning_rate": 4.0839592188931576e-06,
|
8086 |
+
"loss": 0.3482,
|
8087 |
+
"step": 1154
|
8088 |
+
},
|
8089 |
+
{
|
8090 |
+
"epoch": 0.8721918066830281,
|
8091 |
+
"grad_norm": 0.18616865575313568,
|
8092 |
+
"learning_rate": 4.036984820916723e-06,
|
8093 |
+
"loss": 0.3697,
|
8094 |
+
"step": 1155
|
8095 |
+
},
|
8096 |
+
{
|
8097 |
+
"epoch": 0.8729469511043987,
|
8098 |
+
"grad_norm": 0.19371068477630615,
|
8099 |
+
"learning_rate": 3.990270774502941e-06,
|
8100 |
+
"loss": 0.413,
|
8101 |
+
"step": 1156
|
8102 |
+
},
|
8103 |
+
{
|
8104 |
+
"epoch": 0.8737020955257693,
|
8105 |
+
"grad_norm": 0.19405323266983032,
|
8106 |
+
"learning_rate": 3.9438173442575e-06,
|
8107 |
+
"loss": 0.3545,
|
8108 |
+
"step": 1157
|
8109 |
+
},
|
8110 |
+
{
|
8111 |
+
"epoch": 0.8744572399471399,
|
8112 |
+
"grad_norm": 0.19808508455753326,
|
8113 |
+
"learning_rate": 3.897624793309846e-06,
|
8114 |
+
"loss": 0.3991,
|
8115 |
+
"step": 1158
|
8116 |
+
},
|
8117 |
+
{
|
8118 |
+
"epoch": 0.8752123843685105,
|
8119 |
+
"grad_norm": 0.19052360951900482,
|
8120 |
+
"learning_rate": 3.851693383311722e-06,
|
8121 |
+
"loss": 0.3765,
|
8122 |
+
"step": 1159
|
8123 |
+
},
|
8124 |
+
{
|
8125 |
+
"epoch": 0.875967528789881,
|
8126 |
+
"grad_norm": 0.1995311975479126,
|
8127 |
+
"learning_rate": 3.8060233744356633e-06,
|
8128 |
+
"loss": 0.4073,
|
8129 |
+
"step": 1160
|
8130 |
+
},
|
8131 |
+
{
|
8132 |
+
"epoch": 0.8767226732112516,
|
8133 |
+
"grad_norm": 0.19826874136924744,
|
8134 |
+
"learning_rate": 3.760615025373543e-06,
|
8135 |
+
"loss": 0.3841,
|
8136 |
+
"step": 1161
|
8137 |
+
},
|
8138 |
+
{
|
8139 |
+
"epoch": 0.8774778176326222,
|
8140 |
+
"grad_norm": 0.20532841980457306,
|
8141 |
+
"learning_rate": 3.7154685933350864e-06,
|
8142 |
+
"loss": 0.3538,
|
8143 |
+
"step": 1162
|
8144 |
+
},
|
8145 |
+
{
|
8146 |
+
"epoch": 0.8782329620539928,
|
8147 |
+
"grad_norm": 0.20944344997406006,
|
8148 |
+
"learning_rate": 3.6705843340464286e-06,
|
8149 |
+
"loss": 0.4038,
|
8150 |
+
"step": 1163
|
8151 |
+
},
|
8152 |
+
{
|
8153 |
+
"epoch": 0.8789881064753634,
|
8154 |
+
"grad_norm": 0.22046121954917908,
|
8155 |
+
"learning_rate": 3.625962501748653e-06,
|
8156 |
+
"loss": 0.4242,
|
8157 |
+
"step": 1164
|
8158 |
+
},
|
8159 |
+
{
|
8160 |
+
"epoch": 0.879743250896734,
|
8161 |
+
"grad_norm": 0.22591526806354523,
|
8162 |
+
"learning_rate": 3.581603349196372e-06,
|
8163 |
+
"loss": 0.4469,
|
8164 |
+
"step": 1165
|
8165 |
+
},
|
8166 |
+
{
|
8167 |
+
"epoch": 0.8804983953181046,
|
8168 |
+
"grad_norm": 0.22444604337215424,
|
8169 |
+
"learning_rate": 3.53750712765627e-06,
|
8170 |
+
"loss": 0.3994,
|
8171 |
+
"step": 1166
|
8172 |
+
},
|
8173 |
+
{
|
8174 |
+
"epoch": 0.8812535397394752,
|
8175 |
+
"grad_norm": 0.2247525304555893,
|
8176 |
+
"learning_rate": 3.4936740869057073e-06,
|
8177 |
+
"loss": 0.4276,
|
8178 |
+
"step": 1167
|
8179 |
+
},
|
8180 |
+
{
|
8181 |
+
"epoch": 0.8820086841608458,
|
8182 |
+
"grad_norm": 0.2309032678604126,
|
8183 |
+
"learning_rate": 3.4501044752312582e-06,
|
8184 |
+
"loss": 0.463,
|
8185 |
+
"step": 1168
|
8186 |
+
},
|
8187 |
+
{
|
8188 |
+
"epoch": 0.8827638285822164,
|
8189 |
+
"grad_norm": 0.22152245044708252,
|
8190 |
+
"learning_rate": 3.406798539427386e-06,
|
8191 |
+
"loss": 0.3662,
|
8192 |
+
"step": 1169
|
8193 |
+
},
|
8194 |
+
{
|
8195 |
+
"epoch": 0.883518973003587,
|
8196 |
+
"grad_norm": 0.2504233419895172,
|
8197 |
+
"learning_rate": 3.3637565247949588e-06,
|
8198 |
+
"loss": 0.4318,
|
8199 |
+
"step": 1170
|
8200 |
+
},
|
8201 |
+
{
|
8202 |
+
"epoch": 0.8842741174249575,
|
8203 |
+
"grad_norm": 0.23805570602416992,
|
8204 |
+
"learning_rate": 3.3209786751399187e-06,
|
8205 |
+
"loss": 0.4156,
|
8206 |
+
"step": 1171
|
8207 |
+
},
|
8208 |
+
{
|
8209 |
+
"epoch": 0.8850292618463281,
|
8210 |
+
"grad_norm": 0.23133568465709686,
|
8211 |
+
"learning_rate": 3.2784652327718547e-06,
|
8212 |
+
"loss": 0.3695,
|
8213 |
+
"step": 1172
|
8214 |
+
},
|
8215 |
+
{
|
8216 |
+
"epoch": 0.8857844062676987,
|
8217 |
+
"grad_norm": 0.2318771332502365,
|
8218 |
+
"learning_rate": 3.2362164385026706e-06,
|
8219 |
+
"loss": 0.3824,
|
8220 |
+
"step": 1173
|
8221 |
+
},
|
8222 |
+
{
|
8223 |
+
"epoch": 0.8865395506890693,
|
8224 |
+
"grad_norm": 0.22176909446716309,
|
8225 |
+
"learning_rate": 3.194232531645219e-06,
|
8226 |
+
"loss": 0.3575,
|
8227 |
+
"step": 1174
|
8228 |
+
},
|
8229 |
+
{
|
8230 |
+
"epoch": 0.8872946951104399,
|
8231 |
+
"grad_norm": 0.24867790937423706,
|
8232 |
+
"learning_rate": 3.1525137500119207e-06,
|
8233 |
+
"loss": 0.4419,
|
8234 |
+
"step": 1175
|
8235 |
+
},
|
8236 |
+
{
|
8237 |
+
"epoch": 0.8880498395318105,
|
8238 |
+
"grad_norm": 0.252105176448822,
|
8239 |
+
"learning_rate": 3.111060329913401e-06,
|
8240 |
+
"loss": 0.3854,
|
8241 |
+
"step": 1176
|
8242 |
+
},
|
8243 |
+
{
|
8244 |
+
"epoch": 0.8888049839531811,
|
8245 |
+
"grad_norm": 0.24969127774238586,
|
8246 |
+
"learning_rate": 3.069872506157212e-06,
|
8247 |
+
"loss": 0.3825,
|
8248 |
+
"step": 1177
|
8249 |
+
},
|
8250 |
+
{
|
8251 |
+
"epoch": 0.8895601283745517,
|
8252 |
+
"grad_norm": 0.2607312500476837,
|
8253 |
+
"learning_rate": 3.0289505120464743e-06,
|
8254 |
+
"loss": 0.3986,
|
8255 |
+
"step": 1178
|
8256 |
+
},
|
8257 |
+
{
|
8258 |
+
"epoch": 0.8903152727959223,
|
8259 |
+
"grad_norm": 0.2574225664138794,
|
8260 |
+
"learning_rate": 2.9882945793785367e-06,
|
8261 |
+
"loss": 0.3998,
|
8262 |
+
"step": 1179
|
8263 |
+
},
|
8264 |
+
{
|
8265 |
+
"epoch": 0.8910704172172929,
|
8266 |
+
"grad_norm": 0.25807490944862366,
|
8267 |
+
"learning_rate": 2.947904938443663e-06,
|
8268 |
+
"loss": 0.4147,
|
8269 |
+
"step": 1180
|
8270 |
+
},
|
8271 |
+
{
|
8272 |
+
"epoch": 0.8918255616386634,
|
8273 |
+
"grad_norm": 0.24891719222068787,
|
8274 |
+
"learning_rate": 2.9077818180237693e-06,
|
8275 |
+
"loss": 0.3648,
|
8276 |
+
"step": 1181
|
8277 |
+
},
|
8278 |
+
{
|
8279 |
+
"epoch": 0.892580706060034,
|
8280 |
+
"grad_norm": 0.27970343828201294,
|
8281 |
+
"learning_rate": 2.8679254453910785e-06,
|
8282 |
+
"loss": 0.4347,
|
8283 |
+
"step": 1182
|
8284 |
+
},
|
8285 |
+
{
|
8286 |
+
"epoch": 0.8933358504814045,
|
8287 |
+
"grad_norm": 0.26613345742225647,
|
8288 |
+
"learning_rate": 2.8283360463068785e-06,
|
8289 |
+
"loss": 0.3969,
|
8290 |
+
"step": 1183
|
8291 |
+
},
|
8292 |
+
{
|
8293 |
+
"epoch": 0.8940909949027751,
|
8294 |
+
"grad_norm": 0.3011374771595001,
|
8295 |
+
"learning_rate": 2.789013845020205e-06,
|
8296 |
+
"loss": 0.389,
|
8297 |
+
"step": 1184
|
8298 |
+
},
|
8299 |
+
{
|
8300 |
+
"epoch": 0.8948461393241457,
|
8301 |
+
"grad_norm": 0.26815304160118103,
|
8302 |
+
"learning_rate": 2.7499590642665774e-06,
|
8303 |
+
"loss": 0.3572,
|
8304 |
+
"step": 1185
|
8305 |
+
},
|
8306 |
+
{
|
8307 |
+
"epoch": 0.8956012837455163,
|
8308 |
+
"grad_norm": 0.27228736877441406,
|
8309 |
+
"learning_rate": 2.7111719252667647e-06,
|
8310 |
+
"loss": 0.3945,
|
8311 |
+
"step": 1186
|
8312 |
+
},
|
8313 |
+
{
|
8314 |
+
"epoch": 0.8963564281668869,
|
8315 |
+
"grad_norm": 0.28344476222991943,
|
8316 |
+
"learning_rate": 2.6726526477254987e-06,
|
8317 |
+
"loss": 0.4098,
|
8318 |
+
"step": 1187
|
8319 |
+
},
|
8320 |
+
{
|
8321 |
+
"epoch": 0.8971115725882575,
|
8322 |
+
"grad_norm": 0.3316936790943146,
|
8323 |
+
"learning_rate": 2.6344014498302704e-06,
|
8324 |
+
"loss": 0.4422,
|
8325 |
+
"step": 1188
|
8326 |
+
},
|
8327 |
+
{
|
8328 |
+
"epoch": 0.8978667170096281,
|
8329 |
+
"grad_norm": 0.3098110556602478,
|
8330 |
+
"learning_rate": 2.596418548250029e-06,
|
8331 |
+
"loss": 0.3844,
|
8332 |
+
"step": 1189
|
8333 |
+
},
|
8334 |
+
{
|
8335 |
+
"epoch": 0.8986218614309986,
|
8336 |
+
"grad_norm": 0.27956005930900574,
|
8337 |
+
"learning_rate": 2.5587041581340233e-06,
|
8338 |
+
"loss": 0.3017,
|
8339 |
+
"step": 1190
|
8340 |
+
},
|
8341 |
+
{
|
8342 |
+
"epoch": 0.8993770058523692,
|
8343 |
+
"grad_norm": 0.3119887709617615,
|
8344 |
+
"learning_rate": 2.52125849311054e-06,
|
8345 |
+
"loss": 0.4119,
|
8346 |
+
"step": 1191
|
8347 |
+
},
|
8348 |
+
{
|
8349 |
+
"epoch": 0.9001321502737398,
|
8350 |
+
"grad_norm": 0.3407526910305023,
|
8351 |
+
"learning_rate": 2.4840817652857172e-06,
|
8352 |
+
"loss": 0.4231,
|
8353 |
+
"step": 1192
|
8354 |
+
},
|
8355 |
+
{
|
8356 |
+
"epoch": 0.9008872946951104,
|
8357 |
+
"grad_norm": 0.34797540307044983,
|
8358 |
+
"learning_rate": 2.4471741852423237e-06,
|
8359 |
+
"loss": 0.394,
|
8360 |
+
"step": 1193
|
8361 |
+
},
|
8362 |
+
{
|
8363 |
+
"epoch": 0.901642439116481,
|
8364 |
+
"grad_norm": 0.3451668322086334,
|
8365 |
+
"learning_rate": 2.4105359620385847e-06,
|
8366 |
+
"loss": 0.3652,
|
8367 |
+
"step": 1194
|
8368 |
+
},
|
8369 |
+
{
|
8370 |
+
"epoch": 0.9023975835378516,
|
8371 |
+
"grad_norm": 0.3375682532787323,
|
8372 |
+
"learning_rate": 2.3741673032069756e-06,
|
8373 |
+
"loss": 0.3551,
|
8374 |
+
"step": 1195
|
8375 |
+
},
|
8376 |
+
{
|
8377 |
+
"epoch": 0.9031527279592222,
|
8378 |
+
"grad_norm": 0.36490514874458313,
|
8379 |
+
"learning_rate": 2.338068414753075e-06,
|
8380 |
+
"loss": 0.3753,
|
8381 |
+
"step": 1196
|
8382 |
+
},
|
8383 |
+
{
|
8384 |
+
"epoch": 0.9039078723805928,
|
8385 |
+
"grad_norm": 0.3733910322189331,
|
8386 |
+
"learning_rate": 2.3022395011543686e-06,
|
8387 |
+
"loss": 0.4036,
|
8388 |
+
"step": 1197
|
8389 |
+
},
|
8390 |
+
{
|
8391 |
+
"epoch": 0.9046630168019634,
|
8392 |
+
"grad_norm": 0.38239195942878723,
|
8393 |
+
"learning_rate": 2.2666807653591083e-06,
|
8394 |
+
"loss": 0.4228,
|
8395 |
+
"step": 1198
|
8396 |
+
},
|
8397 |
+
{
|
8398 |
+
"epoch": 0.905418161223334,
|
8399 |
+
"grad_norm": 0.46867436170578003,
|
8400 |
+
"learning_rate": 2.2313924087851656e-06,
|
8401 |
+
"loss": 0.4567,
|
8402 |
+
"step": 1199
|
8403 |
+
},
|
8404 |
+
{
|
8405 |
+
"epoch": 0.9061733056447046,
|
8406 |
+
"grad_norm": 0.6003281474113464,
|
8407 |
+
"learning_rate": 2.196374631318876e-06,
|
8408 |
+
"loss": 0.4376,
|
8409 |
+
"step": 1200
|
8410 |
+
},
|
8411 |
+
{
|
8412 |
+
"epoch": 0.9069284500660751,
|
8413 |
+
"grad_norm": 0.1415417641401291,
|
8414 |
+
"learning_rate": 2.161627631313923e-06,
|
8415 |
+
"loss": 0.2603,
|
8416 |
+
"step": 1201
|
8417 |
+
},
|
8418 |
+
{
|
8419 |
+
"epoch": 0.9076835944874457,
|
8420 |
+
"grad_norm": 0.1586138904094696,
|
8421 |
+
"learning_rate": 2.1271516055901777e-06,
|
8422 |
+
"loss": 0.3263,
|
8423 |
+
"step": 1202
|
8424 |
+
},
|
8425 |
+
{
|
8426 |
+
"epoch": 0.9084387389088163,
|
8427 |
+
"grad_norm": 0.16962364315986633,
|
8428 |
+
"learning_rate": 2.0929467494326614e-06,
|
8429 |
+
"loss": 0.3294,
|
8430 |
+
"step": 1203
|
8431 |
+
},
|
8432 |
+
{
|
8433 |
+
"epoch": 0.9091938833301869,
|
8434 |
+
"grad_norm": 0.16859561204910278,
|
8435 |
+
"learning_rate": 2.0590132565903476e-06,
|
8436 |
+
"loss": 0.3619,
|
8437 |
+
"step": 1204
|
8438 |
+
},
|
8439 |
+
{
|
8440 |
+
"epoch": 0.9099490277515575,
|
8441 |
+
"grad_norm": 0.1871105134487152,
|
8442 |
+
"learning_rate": 2.0253513192751373e-06,
|
8443 |
+
"loss": 0.3679,
|
8444 |
+
"step": 1205
|
8445 |
+
},
|
8446 |
+
{
|
8447 |
+
"epoch": 0.9107041721729281,
|
8448 |
+
"grad_norm": 0.18284808099269867,
|
8449 |
+
"learning_rate": 1.9919611281607077e-06,
|
8450 |
+
"loss": 0.3423,
|
8451 |
+
"step": 1206
|
8452 |
+
},
|
8453 |
+
{
|
8454 |
+
"epoch": 0.9114593165942987,
|
8455 |
+
"grad_norm": 0.19052286446094513,
|
8456 |
+
"learning_rate": 1.9588428723814946e-06,
|
8457 |
+
"loss": 0.4149,
|
8458 |
+
"step": 1207
|
8459 |
+
},
|
8460 |
+
{
|
8461 |
+
"epoch": 0.9122144610156693,
|
8462 |
+
"grad_norm": 0.20341211557388306,
|
8463 |
+
"learning_rate": 1.925996739531577e-06,
|
8464 |
+
"loss": 0.3938,
|
8465 |
+
"step": 1208
|
8466 |
+
},
|
8467 |
+
{
|
8468 |
+
"epoch": 0.9129696054370399,
|
8469 |
+
"grad_norm": 0.19892559945583344,
|
8470 |
+
"learning_rate": 1.8934229156636452e-06,
|
8471 |
+
"loss": 0.361,
|
8472 |
+
"step": 1209
|
8473 |
+
},
|
8474 |
+
{
|
8475 |
+
"epoch": 0.9137247498584105,
|
8476 |
+
"grad_norm": 0.21719536185264587,
|
8477 |
+
"learning_rate": 1.8611215852879005e-06,
|
8478 |
+
"loss": 0.429,
|
8479 |
+
"step": 1210
|
8480 |
+
},
|
8481 |
+
{
|
8482 |
+
"epoch": 0.914479894279781,
|
8483 |
+
"grad_norm": 0.20925089716911316,
|
8484 |
+
"learning_rate": 1.8290929313710513e-06,
|
8485 |
+
"loss": 0.3961,
|
8486 |
+
"step": 1211
|
8487 |
+
},
|
8488 |
+
{
|
8489 |
+
"epoch": 0.9152350387011516,
|
8490 |
+
"grad_norm": 0.21349644660949707,
|
8491 |
+
"learning_rate": 1.797337135335292e-06,
|
8492 |
+
"loss": 0.3969,
|
8493 |
+
"step": 1212
|
8494 |
+
},
|
8495 |
+
{
|
8496 |
+
"epoch": 0.9159901831225222,
|
8497 |
+
"grad_norm": 0.2184101939201355,
|
8498 |
+
"learning_rate": 1.7658543770572189e-06,
|
8499 |
+
"loss": 0.3583,
|
8500 |
+
"step": 1213
|
8501 |
+
},
|
8502 |
+
{
|
8503 |
+
"epoch": 0.9167453275438928,
|
8504 |
+
"grad_norm": 0.21886181831359863,
|
8505 |
+
"learning_rate": 1.7346448348668443e-06,
|
8506 |
+
"loss": 0.42,
|
8507 |
+
"step": 1214
|
8508 |
+
},
|
8509 |
+
{
|
8510 |
+
"epoch": 0.9175004719652634,
|
8511 |
+
"grad_norm": 0.20769384503364563,
|
8512 |
+
"learning_rate": 1.70370868554659e-06,
|
8513 |
+
"loss": 0.3761,
|
8514 |
+
"step": 1215
|
8515 |
+
},
|
8516 |
+
{
|
8517 |
+
"epoch": 0.9182556163866339,
|
8518 |
+
"grad_norm": 0.2125682830810547,
|
8519 |
+
"learning_rate": 1.6730461043302726e-06,
|
8520 |
+
"loss": 0.3481,
|
8521 |
+
"step": 1216
|
8522 |
+
},
|
8523 |
+
{
|
8524 |
+
"epoch": 0.9190107608080045,
|
8525 |
+
"grad_norm": 0.22738561034202576,
|
8526 |
+
"learning_rate": 1.6426572649021476e-06,
|
8527 |
+
"loss": 0.4114,
|
8528 |
+
"step": 1217
|
8529 |
+
},
|
8530 |
+
{
|
8531 |
+
"epoch": 0.9197659052293751,
|
8532 |
+
"grad_norm": 0.22645215690135956,
|
8533 |
+
"learning_rate": 1.612542339395845e-06,
|
8534 |
+
"loss": 0.4581,
|
8535 |
+
"step": 1218
|
8536 |
+
},
|
8537 |
+
{
|
8538 |
+
"epoch": 0.9205210496507457,
|
8539 |
+
"grad_norm": 0.2223154753446579,
|
8540 |
+
"learning_rate": 1.582701498393474e-06,
|
8541 |
+
"loss": 0.4042,
|
8542 |
+
"step": 1219
|
8543 |
+
},
|
8544 |
+
{
|
8545 |
+
"epoch": 0.9212761940721163,
|
8546 |
+
"grad_norm": 0.23326793313026428,
|
8547 |
+
"learning_rate": 1.5531349109246362e-06,
|
8548 |
+
"loss": 0.4387,
|
8549 |
+
"step": 1220
|
8550 |
+
},
|
8551 |
+
{
|
8552 |
+
"epoch": 0.9220313384934868,
|
8553 |
+
"grad_norm": 0.23844638466835022,
|
8554 |
+
"learning_rate": 1.523842744465437e-06,
|
8555 |
+
"loss": 0.4143,
|
8556 |
+
"step": 1221
|
8557 |
+
},
|
8558 |
+
{
|
8559 |
+
"epoch": 0.9227864829148574,
|
8560 |
+
"grad_norm": 0.23629891872406006,
|
8561 |
+
"learning_rate": 1.4948251649375745e-06,
|
8562 |
+
"loss": 0.4301,
|
8563 |
+
"step": 1222
|
8564 |
+
},
|
8565 |
+
{
|
8566 |
+
"epoch": 0.923541627336228,
|
8567 |
+
"grad_norm": 0.23292282223701477,
|
8568 |
+
"learning_rate": 1.4660823367073751e-06,
|
8569 |
+
"loss": 0.389,
|
8570 |
+
"step": 1223
|
8571 |
+
},
|
8572 |
+
{
|
8573 |
+
"epoch": 0.9242967717575986,
|
8574 |
+
"grad_norm": 0.25617265701293945,
|
8575 |
+
"learning_rate": 1.437614422584882e-06,
|
8576 |
+
"loss": 0.4615,
|
8577 |
+
"step": 1224
|
8578 |
+
},
|
8579 |
+
{
|
8580 |
+
"epoch": 0.9250519161789692,
|
8581 |
+
"grad_norm": 0.23757751286029816,
|
8582 |
+
"learning_rate": 1.4094215838229176e-06,
|
8583 |
+
"loss": 0.3766,
|
8584 |
+
"step": 1225
|
8585 |
+
},
|
8586 |
+
{
|
8587 |
+
"epoch": 0.9258070606003398,
|
8588 |
+
"grad_norm": 0.2631858289241791,
|
8589 |
+
"learning_rate": 1.3815039801161721e-06,
|
8590 |
+
"loss": 0.4561,
|
8591 |
+
"step": 1226
|
8592 |
+
},
|
8593 |
+
{
|
8594 |
+
"epoch": 0.9265622050217104,
|
8595 |
+
"grad_norm": 0.2697192132472992,
|
8596 |
+
"learning_rate": 1.3538617696003064e-06,
|
8597 |
+
"loss": 0.4365,
|
8598 |
+
"step": 1227
|
8599 |
+
},
|
8600 |
+
{
|
8601 |
+
"epoch": 0.927317349443081,
|
8602 |
+
"grad_norm": 0.23894578218460083,
|
8603 |
+
"learning_rate": 1.3264951088510502e-06,
|
8604 |
+
"loss": 0.3623,
|
8605 |
+
"step": 1228
|
8606 |
+
},
|
8607 |
+
{
|
8608 |
+
"epoch": 0.9280724938644516,
|
8609 |
+
"grad_norm": 0.27340683341026306,
|
8610 |
+
"learning_rate": 1.2994041528833266e-06,
|
8611 |
+
"loss": 0.4602,
|
8612 |
+
"step": 1229
|
8613 |
+
},
|
8614 |
+
{
|
8615 |
+
"epoch": 0.9288276382858222,
|
8616 |
+
"grad_norm": 0.2491341382265091,
|
8617 |
+
"learning_rate": 1.2725890551503472e-06,
|
8618 |
+
"loss": 0.3397,
|
8619 |
+
"step": 1230
|
8620 |
+
},
|
8621 |
+
{
|
8622 |
+
"epoch": 0.9295827827071927,
|
8623 |
+
"grad_norm": 0.26839953660964966,
|
8624 |
+
"learning_rate": 1.2460499675427729e-06,
|
8625 |
+
"loss": 0.4095,
|
8626 |
+
"step": 1231
|
8627 |
+
},
|
8628 |
+
{
|
8629 |
+
"epoch": 0.9303379271285633,
|
8630 |
+
"grad_norm": 0.28354331851005554,
|
8631 |
+
"learning_rate": 1.2197870403878375e-06,
|
8632 |
+
"loss": 0.4034,
|
8633 |
+
"step": 1232
|
8634 |
+
},
|
8635 |
+
{
|
8636 |
+
"epoch": 0.9310930715499339,
|
8637 |
+
"grad_norm": 0.2778373062610626,
|
8638 |
+
"learning_rate": 1.1938004224484988e-06,
|
8639 |
+
"loss": 0.3612,
|
8640 |
+
"step": 1233
|
8641 |
+
},
|
8642 |
+
{
|
8643 |
+
"epoch": 0.9318482159713045,
|
8644 |
+
"grad_norm": 0.28623783588409424,
|
8645 |
+
"learning_rate": 1.1680902609225941e-06,
|
8646 |
+
"loss": 0.3763,
|
8647 |
+
"step": 1234
|
8648 |
+
},
|
8649 |
+
{
|
8650 |
+
"epoch": 0.9326033603926751,
|
8651 |
+
"grad_norm": 0.2844613790512085,
|
8652 |
+
"learning_rate": 1.1426567014420297e-06,
|
8653 |
+
"loss": 0.4077,
|
8654 |
+
"step": 1235
|
8655 |
+
},
|
8656 |
+
{
|
8657 |
+
"epoch": 0.9333585048140457,
|
8658 |
+
"grad_norm": 0.2865941524505615,
|
8659 |
+
"learning_rate": 1.1174998880718935e-06,
|
8660 |
+
"loss": 0.3973,
|
8661 |
+
"step": 1236
|
8662 |
+
},
|
8663 |
+
{
|
8664 |
+
"epoch": 0.9341136492354163,
|
8665 |
+
"grad_norm": 0.2908569276332855,
|
8666 |
+
"learning_rate": 1.0926199633097157e-06,
|
8667 |
+
"loss": 0.3688,
|
8668 |
+
"step": 1237
|
8669 |
+
},
|
8670 |
+
{
|
8671 |
+
"epoch": 0.9348687936567869,
|
8672 |
+
"grad_norm": 0.3032941520214081,
|
8673 |
+
"learning_rate": 1.0680170680846259e-06,
|
8674 |
+
"loss": 0.3971,
|
8675 |
+
"step": 1238
|
8676 |
+
},
|
8677 |
+
{
|
8678 |
+
"epoch": 0.9356239380781575,
|
8679 |
+
"grad_norm": 0.29943525791168213,
|
8680 |
+
"learning_rate": 1.0436913417565365e-06,
|
8681 |
+
"loss": 0.4151,
|
8682 |
+
"step": 1239
|
8683 |
+
},
|
8684 |
+
{
|
8685 |
+
"epoch": 0.9363790824995281,
|
8686 |
+
"grad_norm": 0.3232915997505188,
|
8687 |
+
"learning_rate": 1.0196429221153824e-06,
|
8688 |
+
"loss": 0.336,
|
8689 |
+
"step": 1240
|
8690 |
+
},
|
8691 |
+
{
|
8692 |
+
"epoch": 0.9371342269208986,
|
8693 |
+
"grad_norm": 0.3355953097343445,
|
8694 |
+
"learning_rate": 9.958719453803278e-07,
|
8695 |
+
"loss": 0.4568,
|
8696 |
+
"step": 1241
|
8697 |
+
},
|
8698 |
+
{
|
8699 |
+
"epoch": 0.9378893713422692,
|
8700 |
+
"grad_norm": 0.35610586404800415,
|
8701 |
+
"learning_rate": 9.723785461990099e-07,
|
8702 |
+
"loss": 0.4446,
|
8703 |
+
"step": 1242
|
8704 |
+
},
|
8705 |
+
{
|
8706 |
+
"epoch": 0.9386445157636398,
|
8707 |
+
"grad_norm": 0.347074031829834,
|
8708 |
+
"learning_rate": 9.491628576467515e-07,
|
8709 |
+
"loss": 0.4065,
|
8710 |
+
"step": 1243
|
8711 |
+
},
|
8712 |
+
{
|
8713 |
+
"epoch": 0.9393996601850104,
|
8714 |
+
"grad_norm": 0.37643951177597046,
|
8715 |
+
"learning_rate": 9.26225011225812e-07,
|
8716 |
+
"loss": 0.3945,
|
8717 |
+
"step": 1244
|
8718 |
+
},
|
8719 |
+
{
|
8720 |
+
"epoch": 0.940154804606381,
|
8721 |
+
"grad_norm": 0.34759992361068726,
|
8722 |
+
"learning_rate": 9.035651368646648e-07,
|
8723 |
+
"loss": 0.3654,
|
8724 |
+
"step": 1245
|
8725 |
+
},
|
8726 |
+
{
|
8727 |
+
"epoch": 0.9409099490277516,
|
8728 |
+
"grad_norm": 0.3707546889781952,
|
8729 |
+
"learning_rate": 8.811833629172428e-07,
|
8730 |
+
"loss": 0.2976,
|
8731 |
+
"step": 1246
|
8732 |
+
},
|
8733 |
+
{
|
8734 |
+
"epoch": 0.9416650934491222,
|
8735 |
+
"grad_norm": 0.3839170038700104,
|
8736 |
+
"learning_rate": 8.590798161622227e-07,
|
8737 |
+
"loss": 0.4161,
|
8738 |
+
"step": 1247
|
8739 |
+
},
|
8740 |
+
{
|
8741 |
+
"epoch": 0.9424202378704928,
|
8742 |
+
"grad_norm": 0.3534461557865143,
|
8743 |
+
"learning_rate": 8.372546218022747e-07,
|
8744 |
+
"loss": 0.3231,
|
8745 |
+
"step": 1248
|
8746 |
+
},
|
8747 |
+
{
|
8748 |
+
"epoch": 0.9431753822918634,
|
8749 |
+
"grad_norm": 0.4794304370880127,
|
8750 |
+
"learning_rate": 8.157079034633974e-07,
|
8751 |
+
"loss": 0.4584,
|
8752 |
+
"step": 1249
|
8753 |
+
},
|
8754 |
+
{
|
8755 |
+
"epoch": 0.9439305267132339,
|
8756 |
+
"grad_norm": 0.5317772030830383,
|
8757 |
+
"learning_rate": 7.944397831941952e-07,
|
8758 |
+
"loss": 0.4143,
|
8759 |
+
"step": 1250
|
8760 |
+
},
|
8761 |
+
{
|
8762 |
+
"epoch": 0.9446856711346044,
|
8763 |
+
"grad_norm": 0.16856196522712708,
|
8764 |
+
"learning_rate": 7.734503814651906e-07,
|
8765 |
+
"loss": 0.3159,
|
8766 |
+
"step": 1251
|
8767 |
+
},
|
8768 |
+
{
|
8769 |
+
"epoch": 0.945440815555975,
|
8770 |
+
"grad_norm": 0.16190548241138458,
|
8771 |
+
"learning_rate": 7.527398171681354e-07,
|
8772 |
+
"loss": 0.3332,
|
8773 |
+
"step": 1252
|
8774 |
+
},
|
8775 |
+
{
|
8776 |
+
"epoch": 0.9461959599773456,
|
8777 |
+
"grad_norm": 0.16370098292827606,
|
8778 |
+
"learning_rate": 7.323082076153509e-07,
|
8779 |
+
"loss": 0.331,
|
8780 |
+
"step": 1253
|
8781 |
+
},
|
8782 |
+
{
|
8783 |
+
"epoch": 0.9469511043987162,
|
8784 |
+
"grad_norm": 0.16921286284923553,
|
8785 |
+
"learning_rate": 7.12155668539044e-07,
|
8786 |
+
"loss": 0.2992,
|
8787 |
+
"step": 1254
|
8788 |
+
},
|
8789 |
+
{
|
8790 |
+
"epoch": 0.9477062488200868,
|
8791 |
+
"grad_norm": 0.17997263371944427,
|
8792 |
+
"learning_rate": 6.922823140906753e-07,
|
8793 |
+
"loss": 0.3884,
|
8794 |
+
"step": 1255
|
8795 |
+
},
|
8796 |
+
{
|
8797 |
+
"epoch": 0.9484613932414574,
|
8798 |
+
"grad_norm": 0.18398523330688477,
|
8799 |
+
"learning_rate": 6.726882568402871e-07,
|
8800 |
+
"loss": 0.3779,
|
8801 |
+
"step": 1256
|
8802 |
+
},
|
8803 |
+
{
|
8804 |
+
"epoch": 0.949216537662828,
|
8805 |
+
"grad_norm": 0.19981886446475983,
|
8806 |
+
"learning_rate": 6.533736077758868e-07,
|
8807 |
+
"loss": 0.3687,
|
8808 |
+
"step": 1257
|
8809 |
+
},
|
8810 |
+
{
|
8811 |
+
"epoch": 0.9499716820841986,
|
8812 |
+
"grad_norm": 0.19422514736652374,
|
8813 |
+
"learning_rate": 6.343384763028148e-07,
|
8814 |
+
"loss": 0.3624,
|
8815 |
+
"step": 1258
|
8816 |
+
},
|
8817 |
+
{
|
8818 |
+
"epoch": 0.9507268265055692,
|
8819 |
+
"grad_norm": 0.19577431678771973,
|
8820 |
+
"learning_rate": 6.15582970243117e-07,
|
8821 |
+
"loss": 0.4185,
|
8822 |
+
"step": 1259
|
8823 |
+
},
|
8824 |
+
{
|
8825 |
+
"epoch": 0.9514819709269398,
|
8826 |
+
"grad_norm": 0.2101883441209793,
|
8827 |
+
"learning_rate": 5.971071958349228e-07,
|
8828 |
+
"loss": 0.3948,
|
8829 |
+
"step": 1260
|
8830 |
+
},
|
8831 |
+
{
|
8832 |
+
"epoch": 0.9522371153483103,
|
8833 |
+
"grad_norm": 0.19973735511302948,
|
8834 |
+
"learning_rate": 5.78911257731879e-07,
|
8835 |
+
"loss": 0.4095,
|
8836 |
+
"step": 1261
|
8837 |
+
},
|
8838 |
+
{
|
8839 |
+
"epoch": 0.9529922597696809,
|
8840 |
+
"grad_norm": 0.20807954668998718,
|
8841 |
+
"learning_rate": 5.609952590025224e-07,
|
8842 |
+
"loss": 0.3892,
|
8843 |
+
"step": 1262
|
8844 |
+
},
|
8845 |
+
{
|
8846 |
+
"epoch": 0.9537474041910515,
|
8847 |
+
"grad_norm": 0.2160942405462265,
|
8848 |
+
"learning_rate": 5.4335930112972e-07,
|
8849 |
+
"loss": 0.3782,
|
8850 |
+
"step": 1263
|
8851 |
+
},
|
8852 |
+
{
|
8853 |
+
"epoch": 0.9545025486124221,
|
8854 |
+
"grad_norm": 0.1981229931116104,
|
8855 |
+
"learning_rate": 5.260034840100736e-07,
|
8856 |
+
"loss": 0.3768,
|
8857 |
+
"step": 1264
|
8858 |
+
},
|
8859 |
+
{
|
8860 |
+
"epoch": 0.9552576930337927,
|
8861 |
+
"grad_norm": 0.19447720050811768,
|
8862 |
+
"learning_rate": 5.089279059533658e-07,
|
8863 |
+
"loss": 0.3527,
|
8864 |
+
"step": 1265
|
8865 |
+
},
|
8866 |
+
{
|
8867 |
+
"epoch": 0.9560128374551633,
|
8868 |
+
"grad_norm": 0.21603922545909882,
|
8869 |
+
"learning_rate": 4.92132663681999e-07,
|
8870 |
+
"loss": 0.3792,
|
8871 |
+
"step": 1266
|
8872 |
+
},
|
8873 |
+
{
|
8874 |
+
"epoch": 0.9567679818765339,
|
8875 |
+
"grad_norm": 0.22324316203594208,
|
8876 |
+
"learning_rate": 4.756178523304622e-07,
|
8877 |
+
"loss": 0.413,
|
8878 |
+
"step": 1267
|
8879 |
+
},
|
8880 |
+
{
|
8881 |
+
"epoch": 0.9575231262979045,
|
8882 |
+
"grad_norm": 0.21948575973510742,
|
8883 |
+
"learning_rate": 4.593835654447709e-07,
|
8884 |
+
"loss": 0.4481,
|
8885 |
+
"step": 1268
|
8886 |
+
},
|
8887 |
+
{
|
8888 |
+
"epoch": 0.9582782707192751,
|
8889 |
+
"grad_norm": 0.21984027326107025,
|
8890 |
+
"learning_rate": 4.434298949819449e-07,
|
8891 |
+
"loss": 0.3766,
|
8892 |
+
"step": 1269
|
8893 |
+
},
|
8894 |
+
{
|
8895 |
+
"epoch": 0.9590334151406457,
|
8896 |
+
"grad_norm": 0.24279214441776276,
|
8897 |
+
"learning_rate": 4.277569313094809e-07,
|
8898 |
+
"loss": 0.3761,
|
8899 |
+
"step": 1270
|
8900 |
+
},
|
8901 |
+
{
|
8902 |
+
"epoch": 0.9597885595620163,
|
8903 |
+
"grad_norm": 0.23110735416412354,
|
8904 |
+
"learning_rate": 4.123647632048644e-07,
|
8905 |
+
"loss": 0.4528,
|
8906 |
+
"step": 1271
|
8907 |
+
},
|
8908 |
+
{
|
8909 |
+
"epoch": 0.9605437039833868,
|
8910 |
+
"grad_norm": 0.2304868847131729,
|
8911 |
+
"learning_rate": 3.972534778550474e-07,
|
8912 |
+
"loss": 0.3633,
|
8913 |
+
"step": 1272
|
8914 |
+
},
|
8915 |
+
{
|
8916 |
+
"epoch": 0.9612988484047574,
|
8917 |
+
"grad_norm": 0.2233697474002838,
|
8918 |
+
"learning_rate": 3.824231608559492e-07,
|
8919 |
+
"loss": 0.397,
|
8920 |
+
"step": 1273
|
8921 |
+
},
|
8922 |
+
{
|
8923 |
+
"epoch": 0.962053992826128,
|
8924 |
+
"grad_norm": 0.23600885272026062,
|
8925 |
+
"learning_rate": 3.6787389621198987e-07,
|
8926 |
+
"loss": 0.4136,
|
8927 |
+
"step": 1274
|
8928 |
+
},
|
8929 |
+
{
|
8930 |
+
"epoch": 0.9628091372474986,
|
8931 |
+
"grad_norm": 0.23501838743686676,
|
8932 |
+
"learning_rate": 3.536057663355852e-07,
|
8933 |
+
"loss": 0.323,
|
8934 |
+
"step": 1275
|
8935 |
+
},
|
8936 |
+
{
|
8937 |
+
"epoch": 0.9635642816688692,
|
8938 |
+
"grad_norm": 0.2491409033536911,
|
8939 |
+
"learning_rate": 3.3961885204673026e-07,
|
8940 |
+
"loss": 0.4763,
|
8941 |
+
"step": 1276
|
8942 |
+
},
|
8943 |
+
{
|
8944 |
+
"epoch": 0.9643194260902398,
|
8945 |
+
"grad_norm": 0.2430865615606308,
|
8946 |
+
"learning_rate": 3.2591323257248893e-07,
|
8947 |
+
"loss": 0.3949,
|
8948 |
+
"step": 1277
|
8949 |
+
},
|
8950 |
+
{
|
8951 |
+
"epoch": 0.9650745705116104,
|
8952 |
+
"grad_norm": 0.2593831717967987,
|
8953 |
+
"learning_rate": 3.124889855465718e-07,
|
8954 |
+
"loss": 0.3738,
|
8955 |
+
"step": 1278
|
8956 |
+
},
|
8957 |
+
{
|
8958 |
+
"epoch": 0.965829714932981,
|
8959 |
+
"grad_norm": 0.24975533783435822,
|
8960 |
+
"learning_rate": 2.993461870088921e-07,
|
8961 |
+
"loss": 0.3614,
|
8962 |
+
"step": 1279
|
8963 |
+
},
|
8964 |
+
{
|
8965 |
+
"epoch": 0.9665848593543516,
|
8966 |
+
"grad_norm": 0.25487789511680603,
|
8967 |
+
"learning_rate": 2.8648491140513266e-07,
|
8968 |
+
"loss": 0.46,
|
8969 |
+
"step": 1280
|
8970 |
+
},
|
8971 |
+
{
|
8972 |
+
"epoch": 0.9673400037757222,
|
8973 |
+
"grad_norm": 0.2505668103694916,
|
8974 |
+
"learning_rate": 2.7390523158633554e-07,
|
8975 |
+
"loss": 0.3317,
|
8976 |
+
"step": 1281
|
8977 |
+
},
|
8978 |
+
{
|
8979 |
+
"epoch": 0.9680951481970927,
|
8980 |
+
"grad_norm": 0.261342853307724,
|
8981 |
+
"learning_rate": 2.616072188084628e-07,
|
8982 |
+
"loss": 0.3857,
|
8983 |
+
"step": 1282
|
8984 |
+
},
|
8985 |
+
{
|
8986 |
+
"epoch": 0.9688502926184632,
|
8987 |
+
"grad_norm": 0.2685639262199402,
|
8988 |
+
"learning_rate": 2.4959094273201977e-07,
|
8989 |
+
"loss": 0.3631,
|
8990 |
+
"step": 1283
|
8991 |
+
},
|
8992 |
+
{
|
8993 |
+
"epoch": 0.9696054370398338,
|
8994 |
+
"grad_norm": 0.26906096935272217,
|
8995 |
+
"learning_rate": 2.378564714216547e-07,
|
8996 |
+
"loss": 0.4212,
|
8997 |
+
"step": 1284
|
8998 |
+
},
|
8999 |
+
{
|
9000 |
+
"epoch": 0.9703605814612044,
|
9001 |
+
"grad_norm": 0.28708136081695557,
|
9002 |
+
"learning_rate": 2.2640387134577058e-07,
|
9003 |
+
"loss": 0.5326,
|
9004 |
+
"step": 1285
|
9005 |
+
},
|
9006 |
+
{
|
9007 |
+
"epoch": 0.971115725882575,
|
9008 |
+
"grad_norm": 0.2781200706958771,
|
9009 |
+
"learning_rate": 2.1523320737613095e-07,
|
9010 |
+
"loss": 0.3952,
|
9011 |
+
"step": 1286
|
9012 |
+
},
|
9013 |
+
{
|
9014 |
+
"epoch": 0.9718708703039456,
|
9015 |
+
"grad_norm": 0.2814629077911377,
|
9016 |
+
"learning_rate": 2.0434454278752123e-07,
|
9017 |
+
"loss": 0.3578,
|
9018 |
+
"step": 1287
|
9019 |
+
},
|
9020 |
+
{
|
9021 |
+
"epoch": 0.9726260147253162,
|
9022 |
+
"grad_norm": 0.2876596450805664,
|
9023 |
+
"learning_rate": 1.937379392573768e-07,
|
9024 |
+
"loss": 0.3978,
|
9025 |
+
"step": 1288
|
9026 |
+
},
|
9027 |
+
{
|
9028 |
+
"epoch": 0.9733811591466868,
|
9029 |
+
"grad_norm": 0.3008500635623932,
|
9030 |
+
"learning_rate": 1.8341345686543332e-07,
|
9031 |
+
"loss": 0.3936,
|
9032 |
+
"step": 1289
|
9033 |
+
},
|
9034 |
+
{
|
9035 |
+
"epoch": 0.9741363035680574,
|
9036 |
+
"grad_norm": 0.32256075739860535,
|
9037 |
+
"learning_rate": 1.7337115409338244e-07,
|
9038 |
+
"loss": 0.4423,
|
9039 |
+
"step": 1290
|
9040 |
+
},
|
9041 |
+
{
|
9042 |
+
"epoch": 0.974891447989428,
|
9043 |
+
"grad_norm": 0.3017309010028839,
|
9044 |
+
"learning_rate": 1.6361108782456113e-07,
|
9045 |
+
"loss": 0.4027,
|
9046 |
+
"step": 1291
|
9047 |
+
},
|
9048 |
+
{
|
9049 |
+
"epoch": 0.9756465924107985,
|
9050 |
+
"grad_norm": 0.3096626400947571,
|
9051 |
+
"learning_rate": 1.5413331334360182e-07,
|
9052 |
+
"loss": 0.3962,
|
9053 |
+
"step": 1292
|
9054 |
+
},
|
9055 |
+
{
|
9056 |
+
"epoch": 0.9764017368321691,
|
9057 |
+
"grad_norm": 0.3427668511867523,
|
9058 |
+
"learning_rate": 1.449378843361271e-07,
|
9059 |
+
"loss": 0.4388,
|
9060 |
+
"step": 1293
|
9061 |
+
},
|
9062 |
+
{
|
9063 |
+
"epoch": 0.9771568812535397,
|
9064 |
+
"grad_norm": 0.3453672528266907,
|
9065 |
+
"learning_rate": 1.360248528884611e-07,
|
9066 |
+
"loss": 0.4638,
|
9067 |
+
"step": 1294
|
9068 |
+
},
|
9069 |
+
{
|
9070 |
+
"epoch": 0.9779120256749103,
|
9071 |
+
"grad_norm": 0.33292150497436523,
|
9072 |
+
"learning_rate": 1.2739426948732424e-07,
|
9073 |
+
"loss": 0.3513,
|
9074 |
+
"step": 1295
|
9075 |
+
},
|
9076 |
+
{
|
9077 |
+
"epoch": 0.9786671700962809,
|
9078 |
+
"grad_norm": 0.34688884019851685,
|
9079 |
+
"learning_rate": 1.190461830195333e-07,
|
9080 |
+
"loss": 0.3666,
|
9081 |
+
"step": 1296
|
9082 |
+
},
|
9083 |
+
{
|
9084 |
+
"epoch": 0.9794223145176515,
|
9085 |
+
"grad_norm": 0.41783586144447327,
|
9086 |
+
"learning_rate": 1.109806407717462e-07,
|
9087 |
+
"loss": 0.4518,
|
9088 |
+
"step": 1297
|
9089 |
+
},
|
9090 |
+
{
|
9091 |
+
"epoch": 0.9801774589390221,
|
9092 |
+
"grad_norm": 0.36416101455688477,
|
9093 |
+
"learning_rate": 1.0319768843018996e-07,
|
9094 |
+
"loss": 0.3785,
|
9095 |
+
"step": 1298
|
9096 |
+
},
|
9097 |
+
{
|
9098 |
+
"epoch": 0.9809326033603927,
|
9099 |
+
"grad_norm": 0.4353952407836914,
|
9100 |
+
"learning_rate": 9.56973700803887e-08,
|
9101 |
+
"loss": 0.4347,
|
9102 |
+
"step": 1299
|
9103 |
+
},
|
9104 |
+
{
|
9105 |
+
"epoch": 0.9816877477817633,
|
9106 |
+
"grad_norm": 0.5201441645622253,
|
9107 |
+
"learning_rate": 8.847972820693051e-08,
|
9108 |
+
"loss": 0.3487,
|
9109 |
+
"step": 1300
|
9110 |
+
},
|
9111 |
+
{
|
9112 |
+
"epoch": 0.9824428922031339,
|
9113 |
+
"grad_norm": 0.14752991497516632,
|
9114 |
+
"learning_rate": 8.15448036932176e-08,
|
9115 |
+
"loss": 0.3271,
|
9116 |
+
"step": 1301
|
9117 |
+
},
|
9118 |
+
{
|
9119 |
+
"epoch": 0.9831980366245044,
|
9120 |
+
"grad_norm": 0.1663774847984314,
|
9121 |
+
"learning_rate": 7.489263582122763e-08,
|
9122 |
+
"loss": 0.3467,
|
9123 |
+
"step": 1302
|
9124 |
+
},
|
9125 |
+
{
|
9126 |
+
"epoch": 0.983953181045875,
|
9127 |
+
"grad_norm": 0.18630105257034302,
|
9128 |
+
"learning_rate": 6.852326227130834e-08,
|
9129 |
+
"loss": 0.4198,
|
9130 |
+
"step": 1303
|
9131 |
+
},
|
9132 |
+
{
|
9133 |
+
"epoch": 0.9847083254672456,
|
9134 |
+
"grad_norm": 0.18574944138526917,
|
9135 |
+
"learning_rate": 6.243671912194993e-08,
|
9136 |
+
"loss": 0.393,
|
9137 |
+
"step": 1304
|
9138 |
+
},
|
9139 |
+
{
|
9140 |
+
"epoch": 0.9854634698886162,
|
9141 |
+
"grad_norm": 0.20900239050388336,
|
9142 |
+
"learning_rate": 5.663304084960186e-08,
|
9143 |
+
"loss": 0.3765,
|
9144 |
+
"step": 1305
|
9145 |
+
},
|
9146 |
+
{
|
9147 |
+
"epoch": 0.9862186143099868,
|
9148 |
+
"grad_norm": 0.20655685663223267,
|
9149 |
+
"learning_rate": 5.111226032843974e-08,
|
9150 |
+
"loss": 0.4079,
|
9151 |
+
"step": 1306
|
9152 |
+
},
|
9153 |
+
{
|
9154 |
+
"epoch": 0.9869737587313574,
|
9155 |
+
"grad_norm": 0.22022481262683868,
|
9156 |
+
"learning_rate": 4.5874408830215434e-08,
|
9157 |
+
"loss": 0.4319,
|
9158 |
+
"step": 1307
|
9159 |
+
},
|
9160 |
+
{
|
9161 |
+
"epoch": 0.987728903152728,
|
9162 |
+
"grad_norm": 0.22208468616008759,
|
9163 |
+
"learning_rate": 4.0919516024057195e-08,
|
9164 |
+
"loss": 0.4083,
|
9165 |
+
"step": 1308
|
9166 |
+
},
|
9167 |
+
{
|
9168 |
+
"epoch": 0.9884840475740986,
|
9169 |
+
"grad_norm": 0.24290700256824493,
|
9170 |
+
"learning_rate": 3.624760997631982e-08,
|
9171 |
+
"loss": 0.4351,
|
9172 |
+
"step": 1309
|
9173 |
+
},
|
9174 |
+
{
|
9175 |
+
"epoch": 0.9892391919954692,
|
9176 |
+
"grad_norm": 0.22663183510303497,
|
9177 |
+
"learning_rate": 3.185871715041255e-08,
|
9178 |
+
"loss": 0.3828,
|
9179 |
+
"step": 1310
|
9180 |
+
},
|
9181 |
+
{
|
9182 |
+
"epoch": 0.9899943364168398,
|
9183 |
+
"grad_norm": 0.25170832872390747,
|
9184 |
+
"learning_rate": 2.7752862406654757e-08,
|
9185 |
+
"loss": 0.4401,
|
9186 |
+
"step": 1311
|
9187 |
+
},
|
9188 |
+
{
|
9189 |
+
"epoch": 0.9907494808382104,
|
9190 |
+
"grad_norm": 0.2576093077659607,
|
9191 |
+
"learning_rate": 2.393006900212047e-08,
|
9192 |
+
"loss": 0.4513,
|
9193 |
+
"step": 1312
|
9194 |
+
},
|
9195 |
+
{
|
9196 |
+
"epoch": 0.9915046252595809,
|
9197 |
+
"grad_norm": 0.2478175014257431,
|
9198 |
+
"learning_rate": 2.0390358590538504e-08,
|
9199 |
+
"loss": 0.3902,
|
9200 |
+
"step": 1313
|
9201 |
+
},
|
9202 |
+
{
|
9203 |
+
"epoch": 0.9922597696809515,
|
9204 |
+
"grad_norm": 0.26220056414604187,
|
9205 |
+
"learning_rate": 1.7133751222137007e-08,
|
9206 |
+
"loss": 0.4562,
|
9207 |
+
"step": 1314
|
9208 |
+
},
|
9209 |
+
{
|
9210 |
+
"epoch": 0.9930149141023221,
|
9211 |
+
"grad_norm": 0.272030234336853,
|
9212 |
+
"learning_rate": 1.4160265343549083e-08,
|
9213 |
+
"loss": 0.4128,
|
9214 |
+
"step": 1315
|
9215 |
+
},
|
9216 |
+
{
|
9217 |
+
"epoch": 0.9937700585236926,
|
9218 |
+
"grad_norm": 0.282482385635376,
|
9219 |
+
"learning_rate": 1.1469917797696239e-08,
|
9220 |
+
"loss": 0.4678,
|
9221 |
+
"step": 1316
|
9222 |
+
},
|
9223 |
+
{
|
9224 |
+
"epoch": 0.9945252029450632,
|
9225 |
+
"grad_norm": 0.29758918285369873,
|
9226 |
+
"learning_rate": 9.06272382371065e-09,
|
9227 |
+
"loss": 0.4549,
|
9228 |
+
"step": 1317
|
9229 |
+
},
|
9230 |
+
{
|
9231 |
+
"epoch": 0.9952803473664338,
|
9232 |
+
"grad_norm": 0.2753942012786865,
|
9233 |
+
"learning_rate": 6.9386970568297014e-09,
|
9234 |
+
"loss": 0.3572,
|
9235 |
+
"step": 1318
|
9236 |
+
},
|
9237 |
+
{
|
9238 |
+
"epoch": 0.9960354917878044,
|
9239 |
+
"grad_norm": 0.3179655075073242,
|
9240 |
+
"learning_rate": 5.097849528334919e-09,
|
9241 |
+
"loss": 0.4328,
|
9242 |
+
"step": 1319
|
9243 |
+
},
|
9244 |
+
{
|
9245 |
+
"epoch": 0.996790636209175,
|
9246 |
+
"grad_norm": 0.30263400077819824,
|
9247 |
+
"learning_rate": 3.540191665457604e-09,
|
9248 |
+
"loss": 0.3691,
|
9249 |
+
"step": 1320
|
9250 |
+
},
|
9251 |
+
{
|
9252 |
+
"epoch": 0.9975457806305456,
|
9253 |
+
"grad_norm": 0.3153160810470581,
|
9254 |
+
"learning_rate": 2.265732291356626e-09,
|
9255 |
+
"loss": 0.3449,
|
9256 |
+
"step": 1321
|
9257 |
+
},
|
9258 |
+
{
|
9259 |
+
"epoch": 0.9983009250519161,
|
9260 |
+
"grad_norm": 0.33442163467407227,
|
9261 |
+
"learning_rate": 1.2744786250407092e-09,
|
9262 |
+
"loss": 0.3951,
|
9263 |
+
"step": 1322
|
9264 |
+
},
|
9265 |
+
{
|
9266 |
+
"epoch": 0.9990560694732867,
|
9267 |
+
"grad_norm": 0.3670744299888611,
|
9268 |
+
"learning_rate": 5.664362813406765e-10,
|
9269 |
+
"loss": 0.3497,
|
9270 |
+
"step": 1323
|
9271 |
+
},
|
9272 |
+
{
|
9273 |
+
"epoch": 0.9998112138946573,
|
9274 |
+
"grad_norm": 0.5614824891090393,
|
9275 |
+
"learning_rate": 1.416092708650396e-10,
|
9276 |
+
"loss": 0.4963,
|
9277 |
+
"step": 1324
|
9278 |
+
},
|
9279 |
+
{
|
9280 |
+
"epoch": 0.9998112138946573,
|
9281 |
+
"eval_loss": 0.3858806788921356,
|
9282 |
+
"eval_runtime": 92.8566,
|
9283 |
+
"eval_samples_per_second": 12.008,
|
9284 |
+
"eval_steps_per_second": 3.005,
|
9285 |
+
"step": 1324
|
9286 |
+
},
|
9287 |
+
{
|
9288 |
+
"epoch": 1.000566358316028,
|
9289 |
+
"grad_norm": 3.6732022762298584,
|
9290 |
+
"learning_rate": 0.0,
|
9291 |
+
"loss": 1.1897,
|
9292 |
+
"step": 1325
|
9293 |
}
|
9294 |
],
|
9295 |
"logging_steps": 1,
|
|
|
9304 |
"should_evaluate": false,
|
9305 |
"should_log": false,
|
9306 |
"should_save": true,
|
9307 |
+
"should_training_stop": true
|
9308 |
},
|
9309 |
"attributes": {}
|
9310 |
}
|
9311 |
},
|
9312 |
+
"total_flos": 1.7073098196516864e+18,
|
9313 |
"train_batch_size": 4,
|
9314 |
"trial_name": null,
|
9315 |
"trial_params": null
|