Upload folder using huggingface_hub
Browse files- model-00001-of-00002.safetensors +1 -1
- model-00002-of-00002.safetensors +1 -1
- optimizer.pt +3 -0
- scheduler.pt +1 -1
- trainer_state.json +2761 -3
model-00001-of-00002.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 4965799096
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5e3f6db49daf04c799cca70cbdf059992d1bd0755597b15111d8082b45a79300
|
3 |
size 4965799096
|
model-00002-of-00002.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1459729952
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b1030409b786e4ad59908db299928e3bf990ecac00c86b5d9f8ce96b3483c94c
|
3 |
size 1459729952
|
optimizer.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a3066289f504748316c3ecef3fb988eb7b9fe8fec3dfeafca43f4986dd693ab4
|
3 |
+
size 6527220350
|
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:f0a809473a15ae22afdbbd0025a12ab2bcc1734f6d34859db412a76dd8f4a9d7
|
3 |
size 1064
|
trainer_state.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
-
"epoch": 1.
|
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,
|
@@ -8289,6 +8289,2764 @@
|
|
8289 |
"learning_rate": 5.039916677556075e-06,
|
8290 |
"loss": 1.1673,
|
8291 |
"step": 1182
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
8292 |
}
|
8293 |
],
|
8294 |
"logging_steps": 1,
|
@@ -8308,7 +11066,7 @@
|
|
8308 |
"attributes": {}
|
8309 |
}
|
8310 |
},
|
8311 |
-
"total_flos":
|
8312 |
"train_batch_size": 2,
|
8313 |
"trial_name": null,
|
8314 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 1.9993650793650795,
|
5 |
"eval_steps": 500,
|
6 |
+
"global_step": 1576,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
8289 |
"learning_rate": 5.039916677556075e-06,
|
8290 |
"loss": 1.1673,
|
8291 |
"step": 1182
|
8292 |
+
},
|
8293 |
+
{
|
8294 |
+
"epoch": 1.5003174603174603,
|
8295 |
+
"grad_norm": 0.94140625,
|
8296 |
+
"learning_rate": 5.036676316368923e-06,
|
8297 |
+
"loss": 1.1223,
|
8298 |
+
"step": 1183
|
8299 |
+
},
|
8300 |
+
{
|
8301 |
+
"epoch": 1.5015873015873016,
|
8302 |
+
"grad_norm": 0.95703125,
|
8303 |
+
"learning_rate": 5.033434322685568e-06,
|
8304 |
+
"loss": 1.111,
|
8305 |
+
"step": 1184
|
8306 |
+
},
|
8307 |
+
{
|
8308 |
+
"epoch": 1.502857142857143,
|
8309 |
+
"grad_norm": 0.96875,
|
8310 |
+
"learning_rate": 5.030190699950161e-06,
|
8311 |
+
"loss": 1.0827,
|
8312 |
+
"step": 1185
|
8313 |
+
},
|
8314 |
+
{
|
8315 |
+
"epoch": 1.5041269841269842,
|
8316 |
+
"grad_norm": 0.921875,
|
8317 |
+
"learning_rate": 5.026945451608582e-06,
|
8318 |
+
"loss": 1.1214,
|
8319 |
+
"step": 1186
|
8320 |
+
},
|
8321 |
+
{
|
8322 |
+
"epoch": 1.5053968253968253,
|
8323 |
+
"grad_norm": 0.98046875,
|
8324 |
+
"learning_rate": 5.023698581108435e-06,
|
8325 |
+
"loss": 1.1387,
|
8326 |
+
"step": 1187
|
8327 |
+
},
|
8328 |
+
{
|
8329 |
+
"epoch": 1.5066666666666668,
|
8330 |
+
"grad_norm": 0.92578125,
|
8331 |
+
"learning_rate": 5.020450091899054e-06,
|
8332 |
+
"loss": 1.0856,
|
8333 |
+
"step": 1188
|
8334 |
+
},
|
8335 |
+
{
|
8336 |
+
"epoch": 1.507936507936508,
|
8337 |
+
"grad_norm": 0.984375,
|
8338 |
+
"learning_rate": 5.017199987431486e-06,
|
8339 |
+
"loss": 1.1512,
|
8340 |
+
"step": 1189
|
8341 |
+
},
|
8342 |
+
{
|
8343 |
+
"epoch": 1.5092063492063492,
|
8344 |
+
"grad_norm": 0.9921875,
|
8345 |
+
"learning_rate": 5.013948271158496e-06,
|
8346 |
+
"loss": 1.1007,
|
8347 |
+
"step": 1190
|
8348 |
+
},
|
8349 |
+
{
|
8350 |
+
"epoch": 1.5104761904761905,
|
8351 |
+
"grad_norm": 0.96484375,
|
8352 |
+
"learning_rate": 5.010694946534565e-06,
|
8353 |
+
"loss": 1.1165,
|
8354 |
+
"step": 1191
|
8355 |
+
},
|
8356 |
+
{
|
8357 |
+
"epoch": 1.5117460317460316,
|
8358 |
+
"grad_norm": 1.0,
|
8359 |
+
"learning_rate": 5.007440017015877e-06,
|
8360 |
+
"loss": 1.1697,
|
8361 |
+
"step": 1192
|
8362 |
+
},
|
8363 |
+
{
|
8364 |
+
"epoch": 1.5130158730158731,
|
8365 |
+
"grad_norm": 0.984375,
|
8366 |
+
"learning_rate": 5.004183486060324e-06,
|
8367 |
+
"loss": 1.148,
|
8368 |
+
"step": 1193
|
8369 |
+
},
|
8370 |
+
{
|
8371 |
+
"epoch": 1.5142857142857142,
|
8372 |
+
"grad_norm": 0.9140625,
|
8373 |
+
"learning_rate": 5.0009253571275e-06,
|
8374 |
+
"loss": 1.1619,
|
8375 |
+
"step": 1194
|
8376 |
+
},
|
8377 |
+
{
|
8378 |
+
"epoch": 1.5155555555555555,
|
8379 |
+
"grad_norm": 0.9921875,
|
8380 |
+
"learning_rate": 4.997665633678694e-06,
|
8381 |
+
"loss": 1.1353,
|
8382 |
+
"step": 1195
|
8383 |
+
},
|
8384 |
+
{
|
8385 |
+
"epoch": 1.5168253968253969,
|
8386 |
+
"grad_norm": 1.0,
|
8387 |
+
"learning_rate": 4.994404319176892e-06,
|
8388 |
+
"loss": 1.1593,
|
8389 |
+
"step": 1196
|
8390 |
+
},
|
8391 |
+
{
|
8392 |
+
"epoch": 1.518095238095238,
|
8393 |
+
"grad_norm": 1.0390625,
|
8394 |
+
"learning_rate": 4.991141417086767e-06,
|
8395 |
+
"loss": 1.1655,
|
8396 |
+
"step": 1197
|
8397 |
+
},
|
8398 |
+
{
|
8399 |
+
"epoch": 1.5193650793650795,
|
8400 |
+
"grad_norm": 0.99609375,
|
8401 |
+
"learning_rate": 4.987876930874681e-06,
|
8402 |
+
"loss": 1.0915,
|
8403 |
+
"step": 1198
|
8404 |
+
},
|
8405 |
+
{
|
8406 |
+
"epoch": 1.5206349206349206,
|
8407 |
+
"grad_norm": 0.98046875,
|
8408 |
+
"learning_rate": 4.984610864008679e-06,
|
8409 |
+
"loss": 1.0815,
|
8410 |
+
"step": 1199
|
8411 |
+
},
|
8412 |
+
{
|
8413 |
+
"epoch": 1.5219047619047619,
|
8414 |
+
"grad_norm": 0.90625,
|
8415 |
+
"learning_rate": 4.981343219958484e-06,
|
8416 |
+
"loss": 1.0946,
|
8417 |
+
"step": 1200
|
8418 |
+
},
|
8419 |
+
{
|
8420 |
+
"epoch": 1.5231746031746032,
|
8421 |
+
"grad_norm": 1.0703125,
|
8422 |
+
"learning_rate": 4.978074002195494e-06,
|
8423 |
+
"loss": 1.1965,
|
8424 |
+
"step": 1201
|
8425 |
+
},
|
8426 |
+
{
|
8427 |
+
"epoch": 1.5244444444444445,
|
8428 |
+
"grad_norm": 1.0625,
|
8429 |
+
"learning_rate": 4.97480321419278e-06,
|
8430 |
+
"loss": 1.0397,
|
8431 |
+
"step": 1202
|
8432 |
+
},
|
8433 |
+
{
|
8434 |
+
"epoch": 1.5257142857142858,
|
8435 |
+
"grad_norm": 0.98046875,
|
8436 |
+
"learning_rate": 4.971530859425082e-06,
|
8437 |
+
"loss": 1.1509,
|
8438 |
+
"step": 1203
|
8439 |
+
},
|
8440 |
+
{
|
8441 |
+
"epoch": 1.5269841269841269,
|
8442 |
+
"grad_norm": 0.9453125,
|
8443 |
+
"learning_rate": 4.9682569413688014e-06,
|
8444 |
+
"loss": 1.2044,
|
8445 |
+
"step": 1204
|
8446 |
+
},
|
8447 |
+
{
|
8448 |
+
"epoch": 1.5282539682539684,
|
8449 |
+
"grad_norm": 0.98046875,
|
8450 |
+
"learning_rate": 4.964981463502005e-06,
|
8451 |
+
"loss": 1.1998,
|
8452 |
+
"step": 1205
|
8453 |
+
},
|
8454 |
+
{
|
8455 |
+
"epoch": 1.5295238095238095,
|
8456 |
+
"grad_norm": 1.046875,
|
8457 |
+
"learning_rate": 4.961704429304411e-06,
|
8458 |
+
"loss": 1.1526,
|
8459 |
+
"step": 1206
|
8460 |
+
},
|
8461 |
+
{
|
8462 |
+
"epoch": 1.5307936507936508,
|
8463 |
+
"grad_norm": 0.8984375,
|
8464 |
+
"learning_rate": 4.9584258422573966e-06,
|
8465 |
+
"loss": 1.1067,
|
8466 |
+
"step": 1207
|
8467 |
+
},
|
8468 |
+
{
|
8469 |
+
"epoch": 1.5320634920634921,
|
8470 |
+
"grad_norm": 0.9609375,
|
8471 |
+
"learning_rate": 4.955145705843985e-06,
|
8472 |
+
"loss": 1.0507,
|
8473 |
+
"step": 1208
|
8474 |
+
},
|
8475 |
+
{
|
8476 |
+
"epoch": 1.5333333333333332,
|
8477 |
+
"grad_norm": 0.96875,
|
8478 |
+
"learning_rate": 4.9518640235488464e-06,
|
8479 |
+
"loss": 1.1291,
|
8480 |
+
"step": 1209
|
8481 |
+
},
|
8482 |
+
{
|
8483 |
+
"epoch": 1.5346031746031747,
|
8484 |
+
"grad_norm": 0.96875,
|
8485 |
+
"learning_rate": 4.948580798858294e-06,
|
8486 |
+
"loss": 1.1125,
|
8487 |
+
"step": 1210
|
8488 |
+
},
|
8489 |
+
{
|
8490 |
+
"epoch": 1.5358730158730158,
|
8491 |
+
"grad_norm": 0.9609375,
|
8492 |
+
"learning_rate": 4.945296035260276e-06,
|
8493 |
+
"loss": 1.1414,
|
8494 |
+
"step": 1211
|
8495 |
+
},
|
8496 |
+
{
|
8497 |
+
"epoch": 1.5371428571428571,
|
8498 |
+
"grad_norm": 0.89453125,
|
8499 |
+
"learning_rate": 4.9420097362443834e-06,
|
8500 |
+
"loss": 1.096,
|
8501 |
+
"step": 1212
|
8502 |
+
},
|
8503 |
+
{
|
8504 |
+
"epoch": 1.5384126984126985,
|
8505 |
+
"grad_norm": 0.984375,
|
8506 |
+
"learning_rate": 4.938721905301829e-06,
|
8507 |
+
"loss": 1.1666,
|
8508 |
+
"step": 1213
|
8509 |
+
},
|
8510 |
+
{
|
8511 |
+
"epoch": 1.5396825396825395,
|
8512 |
+
"grad_norm": 0.9609375,
|
8513 |
+
"learning_rate": 4.93543254592546e-06,
|
8514 |
+
"loss": 1.1204,
|
8515 |
+
"step": 1214
|
8516 |
+
},
|
8517 |
+
{
|
8518 |
+
"epoch": 1.540952380952381,
|
8519 |
+
"grad_norm": 1.0546875,
|
8520 |
+
"learning_rate": 4.932141661609743e-06,
|
8521 |
+
"loss": 1.0977,
|
8522 |
+
"step": 1215
|
8523 |
+
},
|
8524 |
+
{
|
8525 |
+
"epoch": 1.5422222222222222,
|
8526 |
+
"grad_norm": 0.92578125,
|
8527 |
+
"learning_rate": 4.9288492558507664e-06,
|
8528 |
+
"loss": 1.1075,
|
8529 |
+
"step": 1216
|
8530 |
+
},
|
8531 |
+
{
|
8532 |
+
"epoch": 1.5434920634920635,
|
8533 |
+
"grad_norm": 0.98828125,
|
8534 |
+
"learning_rate": 4.925555332146236e-06,
|
8535 |
+
"loss": 1.0444,
|
8536 |
+
"step": 1217
|
8537 |
+
},
|
8538 |
+
{
|
8539 |
+
"epoch": 1.5447619047619048,
|
8540 |
+
"grad_norm": 0.9296875,
|
8541 |
+
"learning_rate": 4.922259893995469e-06,
|
8542 |
+
"loss": 1.0876,
|
8543 |
+
"step": 1218
|
8544 |
+
},
|
8545 |
+
{
|
8546 |
+
"epoch": 1.5460317460317459,
|
8547 |
+
"grad_norm": 0.98046875,
|
8548 |
+
"learning_rate": 4.918962944899391e-06,
|
8549 |
+
"loss": 1.1154,
|
8550 |
+
"step": 1219
|
8551 |
+
},
|
8552 |
+
{
|
8553 |
+
"epoch": 1.5473015873015874,
|
8554 |
+
"grad_norm": 1.0078125,
|
8555 |
+
"learning_rate": 4.915664488360532e-06,
|
8556 |
+
"loss": 1.161,
|
8557 |
+
"step": 1220
|
8558 |
+
},
|
8559 |
+
{
|
8560 |
+
"epoch": 1.5485714285714285,
|
8561 |
+
"grad_norm": 0.921875,
|
8562 |
+
"learning_rate": 4.912364527883026e-06,
|
8563 |
+
"loss": 1.1021,
|
8564 |
+
"step": 1221
|
8565 |
+
},
|
8566 |
+
{
|
8567 |
+
"epoch": 1.5498412698412698,
|
8568 |
+
"grad_norm": 0.99609375,
|
8569 |
+
"learning_rate": 4.9090630669726045e-06,
|
8570 |
+
"loss": 1.115,
|
8571 |
+
"step": 1222
|
8572 |
+
},
|
8573 |
+
{
|
8574 |
+
"epoch": 1.551111111111111,
|
8575 |
+
"grad_norm": 0.88671875,
|
8576 |
+
"learning_rate": 4.905760109136591e-06,
|
8577 |
+
"loss": 1.0934,
|
8578 |
+
"step": 1223
|
8579 |
+
},
|
8580 |
+
{
|
8581 |
+
"epoch": 1.5523809523809524,
|
8582 |
+
"grad_norm": 1.0078125,
|
8583 |
+
"learning_rate": 4.9024556578839e-06,
|
8584 |
+
"loss": 1.0865,
|
8585 |
+
"step": 1224
|
8586 |
+
},
|
8587 |
+
{
|
8588 |
+
"epoch": 1.5536507936507937,
|
8589 |
+
"grad_norm": 0.9375,
|
8590 |
+
"learning_rate": 4.899149716725033e-06,
|
8591 |
+
"loss": 1.164,
|
8592 |
+
"step": 1225
|
8593 |
+
},
|
8594 |
+
{
|
8595 |
+
"epoch": 1.5549206349206348,
|
8596 |
+
"grad_norm": 1.0234375,
|
8597 |
+
"learning_rate": 4.895842289172073e-06,
|
8598 |
+
"loss": 1.1473,
|
8599 |
+
"step": 1226
|
8600 |
+
},
|
8601 |
+
{
|
8602 |
+
"epoch": 1.5561904761904763,
|
8603 |
+
"grad_norm": 0.90234375,
|
8604 |
+
"learning_rate": 4.892533378738685e-06,
|
8605 |
+
"loss": 1.1032,
|
8606 |
+
"step": 1227
|
8607 |
+
},
|
8608 |
+
{
|
8609 |
+
"epoch": 1.5574603174603174,
|
8610 |
+
"grad_norm": 1.1171875,
|
8611 |
+
"learning_rate": 4.8892229889401054e-06,
|
8612 |
+
"loss": 1.0927,
|
8613 |
+
"step": 1228
|
8614 |
+
},
|
8615 |
+
{
|
8616 |
+
"epoch": 1.5587301587301587,
|
8617 |
+
"grad_norm": 1.0390625,
|
8618 |
+
"learning_rate": 4.885911123293146e-06,
|
8619 |
+
"loss": 1.0953,
|
8620 |
+
"step": 1229
|
8621 |
+
},
|
8622 |
+
{
|
8623 |
+
"epoch": 1.56,
|
8624 |
+
"grad_norm": 0.91796875,
|
8625 |
+
"learning_rate": 4.882597785316183e-06,
|
8626 |
+
"loss": 1.1072,
|
8627 |
+
"step": 1230
|
8628 |
+
},
|
8629 |
+
{
|
8630 |
+
"epoch": 1.5612698412698411,
|
8631 |
+
"grad_norm": 1.015625,
|
8632 |
+
"learning_rate": 4.879282978529162e-06,
|
8633 |
+
"loss": 1.1429,
|
8634 |
+
"step": 1231
|
8635 |
+
},
|
8636 |
+
{
|
8637 |
+
"epoch": 1.5625396825396827,
|
8638 |
+
"grad_norm": 0.97265625,
|
8639 |
+
"learning_rate": 4.875966706453582e-06,
|
8640 |
+
"loss": 1.0791,
|
8641 |
+
"step": 1232
|
8642 |
+
},
|
8643 |
+
{
|
8644 |
+
"epoch": 1.5638095238095238,
|
8645 |
+
"grad_norm": 1.0078125,
|
8646 |
+
"learning_rate": 4.872648972612502e-06,
|
8647 |
+
"loss": 1.1152,
|
8648 |
+
"step": 1233
|
8649 |
+
},
|
8650 |
+
{
|
8651 |
+
"epoch": 1.565079365079365,
|
8652 |
+
"grad_norm": 0.94921875,
|
8653 |
+
"learning_rate": 4.8693297805305375e-06,
|
8654 |
+
"loss": 1.0872,
|
8655 |
+
"step": 1234
|
8656 |
+
},
|
8657 |
+
{
|
8658 |
+
"epoch": 1.5663492063492064,
|
8659 |
+
"grad_norm": 0.99609375,
|
8660 |
+
"learning_rate": 4.866009133733848e-06,
|
8661 |
+
"loss": 1.0842,
|
8662 |
+
"step": 1235
|
8663 |
+
},
|
8664 |
+
{
|
8665 |
+
"epoch": 1.5676190476190475,
|
8666 |
+
"grad_norm": 0.98828125,
|
8667 |
+
"learning_rate": 4.862687035750137e-06,
|
8668 |
+
"loss": 1.1449,
|
8669 |
+
"step": 1236
|
8670 |
+
},
|
8671 |
+
{
|
8672 |
+
"epoch": 1.568888888888889,
|
8673 |
+
"grad_norm": 0.96484375,
|
8674 |
+
"learning_rate": 4.859363490108657e-06,
|
8675 |
+
"loss": 1.1102,
|
8676 |
+
"step": 1237
|
8677 |
+
},
|
8678 |
+
{
|
8679 |
+
"epoch": 1.57015873015873,
|
8680 |
+
"grad_norm": 0.921875,
|
8681 |
+
"learning_rate": 4.856038500340194e-06,
|
8682 |
+
"loss": 1.0849,
|
8683 |
+
"step": 1238
|
8684 |
+
},
|
8685 |
+
{
|
8686 |
+
"epoch": 1.5714285714285714,
|
8687 |
+
"grad_norm": 1.0078125,
|
8688 |
+
"learning_rate": 4.852712069977067e-06,
|
8689 |
+
"loss": 1.1458,
|
8690 |
+
"step": 1239
|
8691 |
+
},
|
8692 |
+
{
|
8693 |
+
"epoch": 1.5726984126984127,
|
8694 |
+
"grad_norm": 1.0390625,
|
8695 |
+
"learning_rate": 4.849384202553127e-06,
|
8696 |
+
"loss": 1.1222,
|
8697 |
+
"step": 1240
|
8698 |
+
},
|
8699 |
+
{
|
8700 |
+
"epoch": 1.573968253968254,
|
8701 |
+
"grad_norm": 1.015625,
|
8702 |
+
"learning_rate": 4.84605490160375e-06,
|
8703 |
+
"loss": 1.123,
|
8704 |
+
"step": 1241
|
8705 |
+
},
|
8706 |
+
{
|
8707 |
+
"epoch": 1.5752380952380953,
|
8708 |
+
"grad_norm": 1.125,
|
8709 |
+
"learning_rate": 4.8427241706658395e-06,
|
8710 |
+
"loss": 1.1793,
|
8711 |
+
"step": 1242
|
8712 |
+
},
|
8713 |
+
{
|
8714 |
+
"epoch": 1.5765079365079364,
|
8715 |
+
"grad_norm": 0.95703125,
|
8716 |
+
"learning_rate": 4.8393920132778144e-06,
|
8717 |
+
"loss": 1.0034,
|
8718 |
+
"step": 1243
|
8719 |
+
},
|
8720 |
+
{
|
8721 |
+
"epoch": 1.5777777777777777,
|
8722 |
+
"grad_norm": 0.92578125,
|
8723 |
+
"learning_rate": 4.836058432979608e-06,
|
8724 |
+
"loss": 1.1062,
|
8725 |
+
"step": 1244
|
8726 |
+
},
|
8727 |
+
{
|
8728 |
+
"epoch": 1.579047619047619,
|
8729 |
+
"grad_norm": 0.96484375,
|
8730 |
+
"learning_rate": 4.83272343331267e-06,
|
8731 |
+
"loss": 1.045,
|
8732 |
+
"step": 1245
|
8733 |
+
},
|
8734 |
+
{
|
8735 |
+
"epoch": 1.5803174603174603,
|
8736 |
+
"grad_norm": 0.984375,
|
8737 |
+
"learning_rate": 4.829387017819951e-06,
|
8738 |
+
"loss": 1.1662,
|
8739 |
+
"step": 1246
|
8740 |
+
},
|
8741 |
+
{
|
8742 |
+
"epoch": 1.5815873015873017,
|
8743 |
+
"grad_norm": 1.03125,
|
8744 |
+
"learning_rate": 4.826049190045912e-06,
|
8745 |
+
"loss": 1.1569,
|
8746 |
+
"step": 1247
|
8747 |
+
},
|
8748 |
+
{
|
8749 |
+
"epoch": 1.5828571428571427,
|
8750 |
+
"grad_norm": 1.046875,
|
8751 |
+
"learning_rate": 4.822709953536511e-06,
|
8752 |
+
"loss": 1.1727,
|
8753 |
+
"step": 1248
|
8754 |
+
},
|
8755 |
+
{
|
8756 |
+
"epoch": 1.5841269841269843,
|
8757 |
+
"grad_norm": 1.015625,
|
8758 |
+
"learning_rate": 4.819369311839206e-06,
|
8759 |
+
"loss": 1.0916,
|
8760 |
+
"step": 1249
|
8761 |
+
},
|
8762 |
+
{
|
8763 |
+
"epoch": 1.5853968253968254,
|
8764 |
+
"grad_norm": 1.03125,
|
8765 |
+
"learning_rate": 4.8160272685029414e-06,
|
8766 |
+
"loss": 1.0817,
|
8767 |
+
"step": 1250
|
8768 |
+
},
|
8769 |
+
{
|
8770 |
+
"epoch": 1.5866666666666667,
|
8771 |
+
"grad_norm": 1.0234375,
|
8772 |
+
"learning_rate": 4.812683827078157e-06,
|
8773 |
+
"loss": 1.1218,
|
8774 |
+
"step": 1251
|
8775 |
+
},
|
8776 |
+
{
|
8777 |
+
"epoch": 1.587936507936508,
|
8778 |
+
"grad_norm": 1.0625,
|
8779 |
+
"learning_rate": 4.8093389911167766e-06,
|
8780 |
+
"loss": 1.1506,
|
8781 |
+
"step": 1252
|
8782 |
+
},
|
8783 |
+
{
|
8784 |
+
"epoch": 1.589206349206349,
|
8785 |
+
"grad_norm": 1.03125,
|
8786 |
+
"learning_rate": 4.805992764172202e-06,
|
8787 |
+
"loss": 1.1086,
|
8788 |
+
"step": 1253
|
8789 |
+
},
|
8790 |
+
{
|
8791 |
+
"epoch": 1.5904761904761906,
|
8792 |
+
"grad_norm": 1.046875,
|
8793 |
+
"learning_rate": 4.802645149799315e-06,
|
8794 |
+
"loss": 1.1163,
|
8795 |
+
"step": 1254
|
8796 |
+
},
|
8797 |
+
{
|
8798 |
+
"epoch": 1.5917460317460317,
|
8799 |
+
"grad_norm": 1.1328125,
|
8800 |
+
"learning_rate": 4.799296151554472e-06,
|
8801 |
+
"loss": 1.1357,
|
8802 |
+
"step": 1255
|
8803 |
+
},
|
8804 |
+
{
|
8805 |
+
"epoch": 1.593015873015873,
|
8806 |
+
"grad_norm": 1.03125,
|
8807 |
+
"learning_rate": 4.795945772995502e-06,
|
8808 |
+
"loss": 1.0616,
|
8809 |
+
"step": 1256
|
8810 |
+
},
|
8811 |
+
{
|
8812 |
+
"epoch": 1.5942857142857143,
|
8813 |
+
"grad_norm": 1.0,
|
8814 |
+
"learning_rate": 4.792594017681692e-06,
|
8815 |
+
"loss": 1.1926,
|
8816 |
+
"step": 1257
|
8817 |
+
},
|
8818 |
+
{
|
8819 |
+
"epoch": 1.5955555555555554,
|
8820 |
+
"grad_norm": 1.0390625,
|
8821 |
+
"learning_rate": 4.789240889173801e-06,
|
8822 |
+
"loss": 1.1069,
|
8823 |
+
"step": 1258
|
8824 |
+
},
|
8825 |
+
{
|
8826 |
+
"epoch": 1.596825396825397,
|
8827 |
+
"grad_norm": 1.0390625,
|
8828 |
+
"learning_rate": 4.785886391034041e-06,
|
8829 |
+
"loss": 1.1115,
|
8830 |
+
"step": 1259
|
8831 |
+
},
|
8832 |
+
{
|
8833 |
+
"epoch": 1.598095238095238,
|
8834 |
+
"grad_norm": 0.99609375,
|
8835 |
+
"learning_rate": 4.78253052682608e-06,
|
8836 |
+
"loss": 1.1455,
|
8837 |
+
"step": 1260
|
8838 |
+
},
|
8839 |
+
{
|
8840 |
+
"epoch": 1.5993650793650793,
|
8841 |
+
"grad_norm": 0.9765625,
|
8842 |
+
"learning_rate": 4.7791733001150415e-06,
|
8843 |
+
"loss": 1.0381,
|
8844 |
+
"step": 1261
|
8845 |
+
},
|
8846 |
+
{
|
8847 |
+
"epoch": 1.6006349206349206,
|
8848 |
+
"grad_norm": 1.0390625,
|
8849 |
+
"learning_rate": 4.775814714467489e-06,
|
8850 |
+
"loss": 1.1659,
|
8851 |
+
"step": 1262
|
8852 |
+
},
|
8853 |
+
{
|
8854 |
+
"epoch": 1.601904761904762,
|
8855 |
+
"grad_norm": 1.046875,
|
8856 |
+
"learning_rate": 4.772454773451437e-06,
|
8857 |
+
"loss": 1.0763,
|
8858 |
+
"step": 1263
|
8859 |
+
},
|
8860 |
+
{
|
8861 |
+
"epoch": 1.6031746031746033,
|
8862 |
+
"grad_norm": 1.0703125,
|
8863 |
+
"learning_rate": 4.769093480636334e-06,
|
8864 |
+
"loss": 1.1287,
|
8865 |
+
"step": 1264
|
8866 |
+
},
|
8867 |
+
{
|
8868 |
+
"epoch": 1.6044444444444443,
|
8869 |
+
"grad_norm": 0.93359375,
|
8870 |
+
"learning_rate": 4.7657308395930685e-06,
|
8871 |
+
"loss": 1.0678,
|
8872 |
+
"step": 1265
|
8873 |
+
},
|
8874 |
+
{
|
8875 |
+
"epoch": 1.6057142857142859,
|
8876 |
+
"grad_norm": 1.0234375,
|
8877 |
+
"learning_rate": 4.7623668538939605e-06,
|
8878 |
+
"loss": 1.1395,
|
8879 |
+
"step": 1266
|
8880 |
+
},
|
8881 |
+
{
|
8882 |
+
"epoch": 1.606984126984127,
|
8883 |
+
"grad_norm": 1.140625,
|
8884 |
+
"learning_rate": 4.759001527112757e-06,
|
8885 |
+
"loss": 1.1187,
|
8886 |
+
"step": 1267
|
8887 |
+
},
|
8888 |
+
{
|
8889 |
+
"epoch": 1.6082539682539683,
|
8890 |
+
"grad_norm": 1.0703125,
|
8891 |
+
"learning_rate": 4.75563486282463e-06,
|
8892 |
+
"loss": 1.058,
|
8893 |
+
"step": 1268
|
8894 |
+
},
|
8895 |
+
{
|
8896 |
+
"epoch": 1.6095238095238096,
|
8897 |
+
"grad_norm": 0.9609375,
|
8898 |
+
"learning_rate": 4.752266864606174e-06,
|
8899 |
+
"loss": 1.1288,
|
8900 |
+
"step": 1269
|
8901 |
+
},
|
8902 |
+
{
|
8903 |
+
"epoch": 1.6107936507936507,
|
8904 |
+
"grad_norm": 0.92578125,
|
8905 |
+
"learning_rate": 4.748897536035402e-06,
|
8906 |
+
"loss": 1.0998,
|
8907 |
+
"step": 1270
|
8908 |
+
},
|
8909 |
+
{
|
8910 |
+
"epoch": 1.6120634920634922,
|
8911 |
+
"grad_norm": 1.0625,
|
8912 |
+
"learning_rate": 4.745526880691734e-06,
|
8913 |
+
"loss": 1.0686,
|
8914 |
+
"step": 1271
|
8915 |
+
},
|
8916 |
+
{
|
8917 |
+
"epoch": 1.6133333333333333,
|
8918 |
+
"grad_norm": 1.0390625,
|
8919 |
+
"learning_rate": 4.742154902156005e-06,
|
8920 |
+
"loss": 1.0893,
|
8921 |
+
"step": 1272
|
8922 |
+
},
|
8923 |
+
{
|
8924 |
+
"epoch": 1.6146031746031746,
|
8925 |
+
"grad_norm": 0.9765625,
|
8926 |
+
"learning_rate": 4.738781604010454e-06,
|
8927 |
+
"loss": 1.1056,
|
8928 |
+
"step": 1273
|
8929 |
+
},
|
8930 |
+
{
|
8931 |
+
"epoch": 1.615873015873016,
|
8932 |
+
"grad_norm": 0.88671875,
|
8933 |
+
"learning_rate": 4.735406989838723e-06,
|
8934 |
+
"loss": 1.0968,
|
8935 |
+
"step": 1274
|
8936 |
+
},
|
8937 |
+
{
|
8938 |
+
"epoch": 1.617142857142857,
|
8939 |
+
"grad_norm": 0.97265625,
|
8940 |
+
"learning_rate": 4.7320310632258515e-06,
|
8941 |
+
"loss": 1.0764,
|
8942 |
+
"step": 1275
|
8943 |
+
},
|
8944 |
+
{
|
8945 |
+
"epoch": 1.6184126984126985,
|
8946 |
+
"grad_norm": 1.0546875,
|
8947 |
+
"learning_rate": 4.728653827758271e-06,
|
8948 |
+
"loss": 1.0824,
|
8949 |
+
"step": 1276
|
8950 |
+
},
|
8951 |
+
{
|
8952 |
+
"epoch": 1.6196825396825396,
|
8953 |
+
"grad_norm": 1.0078125,
|
8954 |
+
"learning_rate": 4.725275287023808e-06,
|
8955 |
+
"loss": 1.0422,
|
8956 |
+
"step": 1277
|
8957 |
+
},
|
8958 |
+
{
|
8959 |
+
"epoch": 1.620952380952381,
|
8960 |
+
"grad_norm": 0.92578125,
|
8961 |
+
"learning_rate": 4.721895444611671e-06,
|
8962 |
+
"loss": 1.1258,
|
8963 |
+
"step": 1278
|
8964 |
+
},
|
8965 |
+
{
|
8966 |
+
"epoch": 1.6222222222222222,
|
8967 |
+
"grad_norm": 1.0625,
|
8968 |
+
"learning_rate": 4.718514304112455e-06,
|
8969 |
+
"loss": 1.1382,
|
8970 |
+
"step": 1279
|
8971 |
+
},
|
8972 |
+
{
|
8973 |
+
"epoch": 1.6234920634920635,
|
8974 |
+
"grad_norm": 0.9296875,
|
8975 |
+
"learning_rate": 4.71513186911813e-06,
|
8976 |
+
"loss": 1.1232,
|
8977 |
+
"step": 1280
|
8978 |
+
},
|
8979 |
+
{
|
8980 |
+
"epoch": 1.6247619047619049,
|
8981 |
+
"grad_norm": 1.0078125,
|
8982 |
+
"learning_rate": 4.711748143222045e-06,
|
8983 |
+
"loss": 1.0685,
|
8984 |
+
"step": 1281
|
8985 |
+
},
|
8986 |
+
{
|
8987 |
+
"epoch": 1.626031746031746,
|
8988 |
+
"grad_norm": 1.015625,
|
8989 |
+
"learning_rate": 4.708363130018921e-06,
|
8990 |
+
"loss": 1.1688,
|
8991 |
+
"step": 1282
|
8992 |
+
},
|
8993 |
+
{
|
8994 |
+
"epoch": 1.6273015873015872,
|
8995 |
+
"grad_norm": 1.0,
|
8996 |
+
"learning_rate": 4.704976833104843e-06,
|
8997 |
+
"loss": 1.1588,
|
8998 |
+
"step": 1283
|
8999 |
+
},
|
9000 |
+
{
|
9001 |
+
"epoch": 1.6285714285714286,
|
9002 |
+
"grad_norm": 0.94140625,
|
9003 |
+
"learning_rate": 4.701589256077261e-06,
|
9004 |
+
"loss": 1.1249,
|
9005 |
+
"step": 1284
|
9006 |
+
},
|
9007 |
+
{
|
9008 |
+
"epoch": 1.6298412698412699,
|
9009 |
+
"grad_norm": 0.95703125,
|
9010 |
+
"learning_rate": 4.698200402534986e-06,
|
9011 |
+
"loss": 1.1255,
|
9012 |
+
"step": 1285
|
9013 |
+
},
|
9014 |
+
{
|
9015 |
+
"epoch": 1.6311111111111112,
|
9016 |
+
"grad_norm": 0.95703125,
|
9017 |
+
"learning_rate": 4.694810276078184e-06,
|
9018 |
+
"loss": 1.1134,
|
9019 |
+
"step": 1286
|
9020 |
+
},
|
9021 |
+
{
|
9022 |
+
"epoch": 1.6323809523809523,
|
9023 |
+
"grad_norm": 1.0078125,
|
9024 |
+
"learning_rate": 4.691418880308375e-06,
|
9025 |
+
"loss": 1.087,
|
9026 |
+
"step": 1287
|
9027 |
+
},
|
9028 |
+
{
|
9029 |
+
"epoch": 1.6336507936507938,
|
9030 |
+
"grad_norm": 0.953125,
|
9031 |
+
"learning_rate": 4.6880262188284235e-06,
|
9032 |
+
"loss": 1.1309,
|
9033 |
+
"step": 1288
|
9034 |
+
},
|
9035 |
+
{
|
9036 |
+
"epoch": 1.6349206349206349,
|
9037 |
+
"grad_norm": 0.94921875,
|
9038 |
+
"learning_rate": 4.684632295242544e-06,
|
9039 |
+
"loss": 1.1354,
|
9040 |
+
"step": 1289
|
9041 |
+
},
|
9042 |
+
{
|
9043 |
+
"epoch": 1.6361904761904762,
|
9044 |
+
"grad_norm": 0.98046875,
|
9045 |
+
"learning_rate": 4.681237113156287e-06,
|
9046 |
+
"loss": 1.0666,
|
9047 |
+
"step": 1290
|
9048 |
+
},
|
9049 |
+
{
|
9050 |
+
"epoch": 1.6374603174603175,
|
9051 |
+
"grad_norm": 0.890625,
|
9052 |
+
"learning_rate": 4.677840676176545e-06,
|
9053 |
+
"loss": 1.0729,
|
9054 |
+
"step": 1291
|
9055 |
+
},
|
9056 |
+
{
|
9057 |
+
"epoch": 1.6387301587301586,
|
9058 |
+
"grad_norm": 0.9921875,
|
9059 |
+
"learning_rate": 4.674442987911539e-06,
|
9060 |
+
"loss": 1.0809,
|
9061 |
+
"step": 1292
|
9062 |
+
},
|
9063 |
+
{
|
9064 |
+
"epoch": 1.6400000000000001,
|
9065 |
+
"grad_norm": 1.0,
|
9066 |
+
"learning_rate": 4.6710440519708205e-06,
|
9067 |
+
"loss": 1.1709,
|
9068 |
+
"step": 1293
|
9069 |
+
},
|
9070 |
+
{
|
9071 |
+
"epoch": 1.6412698412698412,
|
9072 |
+
"grad_norm": 0.99609375,
|
9073 |
+
"learning_rate": 4.667643871965267e-06,
|
9074 |
+
"loss": 1.1142,
|
9075 |
+
"step": 1294
|
9076 |
+
},
|
9077 |
+
{
|
9078 |
+
"epoch": 1.6425396825396825,
|
9079 |
+
"grad_norm": 0.953125,
|
9080 |
+
"learning_rate": 4.66424245150708e-06,
|
9081 |
+
"loss": 1.0243,
|
9082 |
+
"step": 1295
|
9083 |
+
},
|
9084 |
+
{
|
9085 |
+
"epoch": 1.6438095238095238,
|
9086 |
+
"grad_norm": 1.015625,
|
9087 |
+
"learning_rate": 4.660839794209774e-06,
|
9088 |
+
"loss": 1.0941,
|
9089 |
+
"step": 1296
|
9090 |
+
},
|
9091 |
+
{
|
9092 |
+
"epoch": 1.645079365079365,
|
9093 |
+
"grad_norm": 0.97265625,
|
9094 |
+
"learning_rate": 4.657435903688182e-06,
|
9095 |
+
"loss": 1.1224,
|
9096 |
+
"step": 1297
|
9097 |
+
},
|
9098 |
+
{
|
9099 |
+
"epoch": 1.6463492063492065,
|
9100 |
+
"grad_norm": 0.98828125,
|
9101 |
+
"learning_rate": 4.654030783558442e-06,
|
9102 |
+
"loss": 1.2249,
|
9103 |
+
"step": 1298
|
9104 |
+
},
|
9105 |
+
{
|
9106 |
+
"epoch": 1.6476190476190475,
|
9107 |
+
"grad_norm": 0.984375,
|
9108 |
+
"learning_rate": 4.650624437438005e-06,
|
9109 |
+
"loss": 1.1019,
|
9110 |
+
"step": 1299
|
9111 |
+
},
|
9112 |
+
{
|
9113 |
+
"epoch": 1.6488888888888888,
|
9114 |
+
"grad_norm": 1.015625,
|
9115 |
+
"learning_rate": 4.647216868945618e-06,
|
9116 |
+
"loss": 1.0937,
|
9117 |
+
"step": 1300
|
9118 |
+
},
|
9119 |
+
{
|
9120 |
+
"epoch": 1.6501587301587302,
|
9121 |
+
"grad_norm": 0.97265625,
|
9122 |
+
"learning_rate": 4.643808081701332e-06,
|
9123 |
+
"loss": 1.1725,
|
9124 |
+
"step": 1301
|
9125 |
+
},
|
9126 |
+
{
|
9127 |
+
"epoch": 1.6514285714285715,
|
9128 |
+
"grad_norm": 0.953125,
|
9129 |
+
"learning_rate": 4.640398079326487e-06,
|
9130 |
+
"loss": 1.0583,
|
9131 |
+
"step": 1302
|
9132 |
+
},
|
9133 |
+
{
|
9134 |
+
"epoch": 1.6526984126984128,
|
9135 |
+
"grad_norm": 0.921875,
|
9136 |
+
"learning_rate": 4.636986865443719e-06,
|
9137 |
+
"loss": 1.1474,
|
9138 |
+
"step": 1303
|
9139 |
+
},
|
9140 |
+
{
|
9141 |
+
"epoch": 1.6539682539682539,
|
9142 |
+
"grad_norm": 1.03125,
|
9143 |
+
"learning_rate": 4.63357444367695e-06,
|
9144 |
+
"loss": 1.1133,
|
9145 |
+
"step": 1304
|
9146 |
+
},
|
9147 |
+
{
|
9148 |
+
"epoch": 1.6552380952380954,
|
9149 |
+
"grad_norm": 0.96875,
|
9150 |
+
"learning_rate": 4.6301608176513825e-06,
|
9151 |
+
"loss": 1.0726,
|
9152 |
+
"step": 1305
|
9153 |
+
},
|
9154 |
+
{
|
9155 |
+
"epoch": 1.6565079365079365,
|
9156 |
+
"grad_norm": 0.95703125,
|
9157 |
+
"learning_rate": 4.6267459909935e-06,
|
9158 |
+
"loss": 1.1477,
|
9159 |
+
"step": 1306
|
9160 |
+
},
|
9161 |
+
{
|
9162 |
+
"epoch": 1.6577777777777778,
|
9163 |
+
"grad_norm": 0.95703125,
|
9164 |
+
"learning_rate": 4.623329967331062e-06,
|
9165 |
+
"loss": 1.0958,
|
9166 |
+
"step": 1307
|
9167 |
+
},
|
9168 |
+
{
|
9169 |
+
"epoch": 1.659047619047619,
|
9170 |
+
"grad_norm": 0.90625,
|
9171 |
+
"learning_rate": 4.6199127502931e-06,
|
9172 |
+
"loss": 1.0743,
|
9173 |
+
"step": 1308
|
9174 |
+
},
|
9175 |
+
{
|
9176 |
+
"epoch": 1.6603174603174602,
|
9177 |
+
"grad_norm": 0.90234375,
|
9178 |
+
"learning_rate": 4.616494343509912e-06,
|
9179 |
+
"loss": 1.1027,
|
9180 |
+
"step": 1309
|
9181 |
+
},
|
9182 |
+
{
|
9183 |
+
"epoch": 1.6615873015873017,
|
9184 |
+
"grad_norm": 0.90234375,
|
9185 |
+
"learning_rate": 4.613074750613062e-06,
|
9186 |
+
"loss": 1.0481,
|
9187 |
+
"step": 1310
|
9188 |
+
},
|
9189 |
+
{
|
9190 |
+
"epoch": 1.6628571428571428,
|
9191 |
+
"grad_norm": 1.046875,
|
9192 |
+
"learning_rate": 4.60965397523537e-06,
|
9193 |
+
"loss": 1.1293,
|
9194 |
+
"step": 1311
|
9195 |
+
},
|
9196 |
+
{
|
9197 |
+
"epoch": 1.6641269841269841,
|
9198 |
+
"grad_norm": 0.99609375,
|
9199 |
+
"learning_rate": 4.606232021010916e-06,
|
9200 |
+
"loss": 1.132,
|
9201 |
+
"step": 1312
|
9202 |
+
},
|
9203 |
+
{
|
9204 |
+
"epoch": 1.6653968253968254,
|
9205 |
+
"grad_norm": 0.9609375,
|
9206 |
+
"learning_rate": 4.6028088915750316e-06,
|
9207 |
+
"loss": 1.1219,
|
9208 |
+
"step": 1313
|
9209 |
+
},
|
9210 |
+
{
|
9211 |
+
"epoch": 1.6666666666666665,
|
9212 |
+
"grad_norm": 0.98828125,
|
9213 |
+
"learning_rate": 4.5993845905642965e-06,
|
9214 |
+
"loss": 1.1903,
|
9215 |
+
"step": 1314
|
9216 |
+
},
|
9217 |
+
{
|
9218 |
+
"epoch": 1.667936507936508,
|
9219 |
+
"grad_norm": 0.98046875,
|
9220 |
+
"learning_rate": 4.595959121616532e-06,
|
9221 |
+
"loss": 1.0881,
|
9222 |
+
"step": 1315
|
9223 |
+
},
|
9224 |
+
{
|
9225 |
+
"epoch": 1.6692063492063491,
|
9226 |
+
"grad_norm": 0.96484375,
|
9227 |
+
"learning_rate": 4.592532488370808e-06,
|
9228 |
+
"loss": 1.1348,
|
9229 |
+
"step": 1316
|
9230 |
+
},
|
9231 |
+
{
|
9232 |
+
"epoch": 1.6704761904761904,
|
9233 |
+
"grad_norm": 0.94140625,
|
9234 |
+
"learning_rate": 4.589104694467424e-06,
|
9235 |
+
"loss": 1.0883,
|
9236 |
+
"step": 1317
|
9237 |
+
},
|
9238 |
+
{
|
9239 |
+
"epoch": 1.6717460317460318,
|
9240 |
+
"grad_norm": 0.97265625,
|
9241 |
+
"learning_rate": 4.585675743547915e-06,
|
9242 |
+
"loss": 1.1143,
|
9243 |
+
"step": 1318
|
9244 |
+
},
|
9245 |
+
{
|
9246 |
+
"epoch": 1.673015873015873,
|
9247 |
+
"grad_norm": 0.98828125,
|
9248 |
+
"learning_rate": 4.582245639255044e-06,
|
9249 |
+
"loss": 1.1538,
|
9250 |
+
"step": 1319
|
9251 |
+
},
|
9252 |
+
{
|
9253 |
+
"epoch": 1.6742857142857144,
|
9254 |
+
"grad_norm": 0.91015625,
|
9255 |
+
"learning_rate": 4.578814385232802e-06,
|
9256 |
+
"loss": 1.096,
|
9257 |
+
"step": 1320
|
9258 |
+
},
|
9259 |
+
{
|
9260 |
+
"epoch": 1.6755555555555555,
|
9261 |
+
"grad_norm": 1.0703125,
|
9262 |
+
"learning_rate": 4.5753819851264e-06,
|
9263 |
+
"loss": 1.1107,
|
9264 |
+
"step": 1321
|
9265 |
+
},
|
9266 |
+
{
|
9267 |
+
"epoch": 1.6768253968253968,
|
9268 |
+
"grad_norm": 1.0234375,
|
9269 |
+
"learning_rate": 4.571948442582265e-06,
|
9270 |
+
"loss": 1.1652,
|
9271 |
+
"step": 1322
|
9272 |
+
},
|
9273 |
+
{
|
9274 |
+
"epoch": 1.678095238095238,
|
9275 |
+
"grad_norm": 1.0234375,
|
9276 |
+
"learning_rate": 4.56851376124804e-06,
|
9277 |
+
"loss": 1.076,
|
9278 |
+
"step": 1323
|
9279 |
+
},
|
9280 |
+
{
|
9281 |
+
"epoch": 1.6793650793650794,
|
9282 |
+
"grad_norm": 1.1328125,
|
9283 |
+
"learning_rate": 4.565077944772576e-06,
|
9284 |
+
"loss": 1.1744,
|
9285 |
+
"step": 1324
|
9286 |
+
},
|
9287 |
+
{
|
9288 |
+
"epoch": 1.6806349206349207,
|
9289 |
+
"grad_norm": 0.921875,
|
9290 |
+
"learning_rate": 4.561640996805933e-06,
|
9291 |
+
"loss": 1.1625,
|
9292 |
+
"step": 1325
|
9293 |
+
},
|
9294 |
+
{
|
9295 |
+
"epoch": 1.6819047619047618,
|
9296 |
+
"grad_norm": 1.0078125,
|
9297 |
+
"learning_rate": 4.558202920999368e-06,
|
9298 |
+
"loss": 1.137,
|
9299 |
+
"step": 1326
|
9300 |
+
},
|
9301 |
+
{
|
9302 |
+
"epoch": 1.6831746031746033,
|
9303 |
+
"grad_norm": 0.9453125,
|
9304 |
+
"learning_rate": 4.5547637210053405e-06,
|
9305 |
+
"loss": 1.0616,
|
9306 |
+
"step": 1327
|
9307 |
+
},
|
9308 |
+
{
|
9309 |
+
"epoch": 1.6844444444444444,
|
9310 |
+
"grad_norm": 0.9375,
|
9311 |
+
"learning_rate": 4.551323400477504e-06,
|
9312 |
+
"loss": 1.1175,
|
9313 |
+
"step": 1328
|
9314 |
+
},
|
9315 |
+
{
|
9316 |
+
"epoch": 1.6857142857142857,
|
9317 |
+
"grad_norm": 0.984375,
|
9318 |
+
"learning_rate": 4.547881963070698e-06,
|
9319 |
+
"loss": 1.0922,
|
9320 |
+
"step": 1329
|
9321 |
+
},
|
9322 |
+
{
|
9323 |
+
"epoch": 1.686984126984127,
|
9324 |
+
"grad_norm": 1.0,
|
9325 |
+
"learning_rate": 4.544439412440954e-06,
|
9326 |
+
"loss": 1.0534,
|
9327 |
+
"step": 1330
|
9328 |
+
},
|
9329 |
+
{
|
9330 |
+
"epoch": 1.6882539682539681,
|
9331 |
+
"grad_norm": 0.9609375,
|
9332 |
+
"learning_rate": 4.540995752245483e-06,
|
9333 |
+
"loss": 1.1033,
|
9334 |
+
"step": 1331
|
9335 |
+
},
|
9336 |
+
{
|
9337 |
+
"epoch": 1.6895238095238096,
|
9338 |
+
"grad_norm": 0.97265625,
|
9339 |
+
"learning_rate": 4.537550986142676e-06,
|
9340 |
+
"loss": 1.067,
|
9341 |
+
"step": 1332
|
9342 |
+
},
|
9343 |
+
{
|
9344 |
+
"epoch": 1.6907936507936507,
|
9345 |
+
"grad_norm": 0.89453125,
|
9346 |
+
"learning_rate": 4.534105117792098e-06,
|
9347 |
+
"loss": 1.1344,
|
9348 |
+
"step": 1333
|
9349 |
+
},
|
9350 |
+
{
|
9351 |
+
"epoch": 1.692063492063492,
|
9352 |
+
"grad_norm": 1.1015625,
|
9353 |
+
"learning_rate": 4.5306581508544865e-06,
|
9354 |
+
"loss": 1.1352,
|
9355 |
+
"step": 1334
|
9356 |
+
},
|
9357 |
+
{
|
9358 |
+
"epoch": 1.6933333333333334,
|
9359 |
+
"grad_norm": 0.890625,
|
9360 |
+
"learning_rate": 4.527210088991744e-06,
|
9361 |
+
"loss": 1.0163,
|
9362 |
+
"step": 1335
|
9363 |
+
},
|
9364 |
+
{
|
9365 |
+
"epoch": 1.6946031746031744,
|
9366 |
+
"grad_norm": 1.015625,
|
9367 |
+
"learning_rate": 4.523760935866939e-06,
|
9368 |
+
"loss": 1.0919,
|
9369 |
+
"step": 1336
|
9370 |
+
},
|
9371 |
+
{
|
9372 |
+
"epoch": 1.695873015873016,
|
9373 |
+
"grad_norm": 1.0390625,
|
9374 |
+
"learning_rate": 4.520310695144294e-06,
|
9375 |
+
"loss": 1.097,
|
9376 |
+
"step": 1337
|
9377 |
+
},
|
9378 |
+
{
|
9379 |
+
"epoch": 1.697142857142857,
|
9380 |
+
"grad_norm": 0.97265625,
|
9381 |
+
"learning_rate": 4.5168593704891926e-06,
|
9382 |
+
"loss": 1.1038,
|
9383 |
+
"step": 1338
|
9384 |
+
},
|
9385 |
+
{
|
9386 |
+
"epoch": 1.6984126984126984,
|
9387 |
+
"grad_norm": 0.98828125,
|
9388 |
+
"learning_rate": 4.513406965568169e-06,
|
9389 |
+
"loss": 1.1644,
|
9390 |
+
"step": 1339
|
9391 |
+
},
|
9392 |
+
{
|
9393 |
+
"epoch": 1.6996825396825397,
|
9394 |
+
"grad_norm": 0.8984375,
|
9395 |
+
"learning_rate": 4.509953484048903e-06,
|
9396 |
+
"loss": 1.0151,
|
9397 |
+
"step": 1340
|
9398 |
+
},
|
9399 |
+
{
|
9400 |
+
"epoch": 1.700952380952381,
|
9401 |
+
"grad_norm": 0.9453125,
|
9402 |
+
"learning_rate": 4.506498929600217e-06,
|
9403 |
+
"loss": 1.146,
|
9404 |
+
"step": 1341
|
9405 |
+
},
|
9406 |
+
{
|
9407 |
+
"epoch": 1.7022222222222223,
|
9408 |
+
"grad_norm": 0.94921875,
|
9409 |
+
"learning_rate": 4.503043305892077e-06,
|
9410 |
+
"loss": 1.0788,
|
9411 |
+
"step": 1342
|
9412 |
+
},
|
9413 |
+
{
|
9414 |
+
"epoch": 1.7034920634920634,
|
9415 |
+
"grad_norm": 0.96484375,
|
9416 |
+
"learning_rate": 4.499586616595583e-06,
|
9417 |
+
"loss": 1.0936,
|
9418 |
+
"step": 1343
|
9419 |
+
},
|
9420 |
+
{
|
9421 |
+
"epoch": 1.704761904761905,
|
9422 |
+
"grad_norm": 1.0234375,
|
9423 |
+
"learning_rate": 4.4961288653829665e-06,
|
9424 |
+
"loss": 1.1669,
|
9425 |
+
"step": 1344
|
9426 |
+
},
|
9427 |
+
{
|
9428 |
+
"epoch": 1.706031746031746,
|
9429 |
+
"grad_norm": 0.984375,
|
9430 |
+
"learning_rate": 4.492670055927589e-06,
|
9431 |
+
"loss": 1.0815,
|
9432 |
+
"step": 1345
|
9433 |
+
},
|
9434 |
+
{
|
9435 |
+
"epoch": 1.7073015873015873,
|
9436 |
+
"grad_norm": 0.98046875,
|
9437 |
+
"learning_rate": 4.489210191903933e-06,
|
9438 |
+
"loss": 1.1458,
|
9439 |
+
"step": 1346
|
9440 |
+
},
|
9441 |
+
{
|
9442 |
+
"epoch": 1.7085714285714286,
|
9443 |
+
"grad_norm": 1.0703125,
|
9444 |
+
"learning_rate": 4.4857492769876045e-06,
|
9445 |
+
"loss": 1.1193,
|
9446 |
+
"step": 1347
|
9447 |
+
},
|
9448 |
+
{
|
9449 |
+
"epoch": 1.7098412698412697,
|
9450 |
+
"grad_norm": 0.92578125,
|
9451 |
+
"learning_rate": 4.482287314855326e-06,
|
9452 |
+
"loss": 1.0507,
|
9453 |
+
"step": 1348
|
9454 |
+
},
|
9455 |
+
{
|
9456 |
+
"epoch": 1.7111111111111112,
|
9457 |
+
"grad_norm": 0.9375,
|
9458 |
+
"learning_rate": 4.4788243091849295e-06,
|
9459 |
+
"loss": 1.201,
|
9460 |
+
"step": 1349
|
9461 |
+
},
|
9462 |
+
{
|
9463 |
+
"epoch": 1.7123809523809523,
|
9464 |
+
"grad_norm": 0.91796875,
|
9465 |
+
"learning_rate": 4.475360263655357e-06,
|
9466 |
+
"loss": 1.0748,
|
9467 |
+
"step": 1350
|
9468 |
+
},
|
9469 |
+
{
|
9470 |
+
"epoch": 1.7136507936507936,
|
9471 |
+
"grad_norm": 1.0390625,
|
9472 |
+
"learning_rate": 4.471895181946656e-06,
|
9473 |
+
"loss": 1.1482,
|
9474 |
+
"step": 1351
|
9475 |
+
},
|
9476 |
+
{
|
9477 |
+
"epoch": 1.714920634920635,
|
9478 |
+
"grad_norm": 0.98828125,
|
9479 |
+
"learning_rate": 4.468429067739976e-06,
|
9480 |
+
"loss": 1.098,
|
9481 |
+
"step": 1352
|
9482 |
+
},
|
9483 |
+
{
|
9484 |
+
"epoch": 1.716190476190476,
|
9485 |
+
"grad_norm": 0.91015625,
|
9486 |
+
"learning_rate": 4.4649619247175605e-06,
|
9487 |
+
"loss": 1.135,
|
9488 |
+
"step": 1353
|
9489 |
+
},
|
9490 |
+
{
|
9491 |
+
"epoch": 1.7174603174603176,
|
9492 |
+
"grad_norm": 0.89453125,
|
9493 |
+
"learning_rate": 4.461493756562748e-06,
|
9494 |
+
"loss": 1.0311,
|
9495 |
+
"step": 1354
|
9496 |
+
},
|
9497 |
+
{
|
9498 |
+
"epoch": 1.7187301587301587,
|
9499 |
+
"grad_norm": 0.97265625,
|
9500 |
+
"learning_rate": 4.458024566959962e-06,
|
9501 |
+
"loss": 1.1186,
|
9502 |
+
"step": 1355
|
9503 |
+
},
|
9504 |
+
{
|
9505 |
+
"epoch": 1.72,
|
9506 |
+
"grad_norm": 0.9609375,
|
9507 |
+
"learning_rate": 4.454554359594718e-06,
|
9508 |
+
"loss": 1.096,
|
9509 |
+
"step": 1356
|
9510 |
+
},
|
9511 |
+
{
|
9512 |
+
"epoch": 1.7212698412698413,
|
9513 |
+
"grad_norm": 0.9609375,
|
9514 |
+
"learning_rate": 4.451083138153609e-06,
|
9515 |
+
"loss": 1.0987,
|
9516 |
+
"step": 1357
|
9517 |
+
},
|
9518 |
+
{
|
9519 |
+
"epoch": 1.7225396825396826,
|
9520 |
+
"grad_norm": 1.0234375,
|
9521 |
+
"learning_rate": 4.447610906324303e-06,
|
9522 |
+
"loss": 1.0812,
|
9523 |
+
"step": 1358
|
9524 |
+
},
|
9525 |
+
{
|
9526 |
+
"epoch": 1.723809523809524,
|
9527 |
+
"grad_norm": 1.0390625,
|
9528 |
+
"learning_rate": 4.444137667795546e-06,
|
9529 |
+
"loss": 1.1388,
|
9530 |
+
"step": 1359
|
9531 |
+
},
|
9532 |
+
{
|
9533 |
+
"epoch": 1.725079365079365,
|
9534 |
+
"grad_norm": 1.0078125,
|
9535 |
+
"learning_rate": 4.44066342625715e-06,
|
9536 |
+
"loss": 1.1375,
|
9537 |
+
"step": 1360
|
9538 |
+
},
|
9539 |
+
{
|
9540 |
+
"epoch": 1.7263492063492063,
|
9541 |
+
"grad_norm": 1.0,
|
9542 |
+
"learning_rate": 4.437188185399994e-06,
|
9543 |
+
"loss": 1.1185,
|
9544 |
+
"step": 1361
|
9545 |
+
},
|
9546 |
+
{
|
9547 |
+
"epoch": 1.7276190476190476,
|
9548 |
+
"grad_norm": 0.9765625,
|
9549 |
+
"learning_rate": 4.43371194891602e-06,
|
9550 |
+
"loss": 1.1783,
|
9551 |
+
"step": 1362
|
9552 |
+
},
|
9553 |
+
{
|
9554 |
+
"epoch": 1.728888888888889,
|
9555 |
+
"grad_norm": 0.88671875,
|
9556 |
+
"learning_rate": 4.430234720498224e-06,
|
9557 |
+
"loss": 1.0944,
|
9558 |
+
"step": 1363
|
9559 |
+
},
|
9560 |
+
{
|
9561 |
+
"epoch": 1.7301587301587302,
|
9562 |
+
"grad_norm": 1.0078125,
|
9563 |
+
"learning_rate": 4.426756503840659e-06,
|
9564 |
+
"loss": 1.1701,
|
9565 |
+
"step": 1364
|
9566 |
+
},
|
9567 |
+
{
|
9568 |
+
"epoch": 1.7314285714285713,
|
9569 |
+
"grad_norm": 1.015625,
|
9570 |
+
"learning_rate": 4.423277302638428e-06,
|
9571 |
+
"loss": 1.1493,
|
9572 |
+
"step": 1365
|
9573 |
+
},
|
9574 |
+
{
|
9575 |
+
"epoch": 1.7326984126984128,
|
9576 |
+
"grad_norm": 1.015625,
|
9577 |
+
"learning_rate": 4.4197971205876775e-06,
|
9578 |
+
"loss": 1.1577,
|
9579 |
+
"step": 1366
|
9580 |
+
},
|
9581 |
+
{
|
9582 |
+
"epoch": 1.733968253968254,
|
9583 |
+
"grad_norm": 1.0078125,
|
9584 |
+
"learning_rate": 4.416315961385597e-06,
|
9585 |
+
"loss": 1.1811,
|
9586 |
+
"step": 1367
|
9587 |
+
},
|
9588 |
+
{
|
9589 |
+
"epoch": 1.7352380952380952,
|
9590 |
+
"grad_norm": 0.97265625,
|
9591 |
+
"learning_rate": 4.412833828730414e-06,
|
9592 |
+
"loss": 1.115,
|
9593 |
+
"step": 1368
|
9594 |
+
},
|
9595 |
+
{
|
9596 |
+
"epoch": 1.7365079365079366,
|
9597 |
+
"grad_norm": 0.8984375,
|
9598 |
+
"learning_rate": 4.409350726321392e-06,
|
9599 |
+
"loss": 1.1033,
|
9600 |
+
"step": 1369
|
9601 |
+
},
|
9602 |
+
{
|
9603 |
+
"epoch": 1.7377777777777776,
|
9604 |
+
"grad_norm": 0.92578125,
|
9605 |
+
"learning_rate": 4.405866657858823e-06,
|
9606 |
+
"loss": 1.1589,
|
9607 |
+
"step": 1370
|
9608 |
+
},
|
9609 |
+
{
|
9610 |
+
"epoch": 1.7390476190476192,
|
9611 |
+
"grad_norm": 0.921875,
|
9612 |
+
"learning_rate": 4.4023816270440245e-06,
|
9613 |
+
"loss": 1.0999,
|
9614 |
+
"step": 1371
|
9615 |
+
},
|
9616 |
+
{
|
9617 |
+
"epoch": 1.7403174603174603,
|
9618 |
+
"grad_norm": 0.9375,
|
9619 |
+
"learning_rate": 4.398895637579338e-06,
|
9620 |
+
"loss": 1.0698,
|
9621 |
+
"step": 1372
|
9622 |
+
},
|
9623 |
+
{
|
9624 |
+
"epoch": 1.7415873015873016,
|
9625 |
+
"grad_norm": 1.0234375,
|
9626 |
+
"learning_rate": 4.395408693168123e-06,
|
9627 |
+
"loss": 1.1485,
|
9628 |
+
"step": 1373
|
9629 |
+
},
|
9630 |
+
{
|
9631 |
+
"epoch": 1.7428571428571429,
|
9632 |
+
"grad_norm": 0.96484375,
|
9633 |
+
"learning_rate": 4.391920797514753e-06,
|
9634 |
+
"loss": 1.0692,
|
9635 |
+
"step": 1374
|
9636 |
+
},
|
9637 |
+
{
|
9638 |
+
"epoch": 1.744126984126984,
|
9639 |
+
"grad_norm": 0.9296875,
|
9640 |
+
"learning_rate": 4.3884319543246125e-06,
|
9641 |
+
"loss": 1.1225,
|
9642 |
+
"step": 1375
|
9643 |
+
},
|
9644 |
+
{
|
9645 |
+
"epoch": 1.7453968253968255,
|
9646 |
+
"grad_norm": 0.94140625,
|
9647 |
+
"learning_rate": 4.384942167304093e-06,
|
9648 |
+
"loss": 1.1042,
|
9649 |
+
"step": 1376
|
9650 |
+
},
|
9651 |
+
{
|
9652 |
+
"epoch": 1.7466666666666666,
|
9653 |
+
"grad_norm": 0.98046875,
|
9654 |
+
"learning_rate": 4.381451440160587e-06,
|
9655 |
+
"loss": 1.1989,
|
9656 |
+
"step": 1377
|
9657 |
+
},
|
9658 |
+
{
|
9659 |
+
"epoch": 1.747936507936508,
|
9660 |
+
"grad_norm": 1.0078125,
|
9661 |
+
"learning_rate": 4.377959776602488e-06,
|
9662 |
+
"loss": 1.1128,
|
9663 |
+
"step": 1378
|
9664 |
+
},
|
9665 |
+
{
|
9666 |
+
"epoch": 1.7492063492063492,
|
9667 |
+
"grad_norm": 0.87890625,
|
9668 |
+
"learning_rate": 4.374467180339182e-06,
|
9669 |
+
"loss": 0.9888,
|
9670 |
+
"step": 1379
|
9671 |
+
},
|
9672 |
+
{
|
9673 |
+
"epoch": 1.7504761904761905,
|
9674 |
+
"grad_norm": 1.0625,
|
9675 |
+
"learning_rate": 4.370973655081048e-06,
|
9676 |
+
"loss": 1.1188,
|
9677 |
+
"step": 1380
|
9678 |
+
},
|
9679 |
+
{
|
9680 |
+
"epoch": 1.7517460317460318,
|
9681 |
+
"grad_norm": 0.96875,
|
9682 |
+
"learning_rate": 4.367479204539452e-06,
|
9683 |
+
"loss": 1.1025,
|
9684 |
+
"step": 1381
|
9685 |
+
},
|
9686 |
+
{
|
9687 |
+
"epoch": 1.753015873015873,
|
9688 |
+
"grad_norm": 0.9765625,
|
9689 |
+
"learning_rate": 4.363983832426739e-06,
|
9690 |
+
"loss": 1.1934,
|
9691 |
+
"step": 1382
|
9692 |
+
},
|
9693 |
+
{
|
9694 |
+
"epoch": 1.7542857142857144,
|
9695 |
+
"grad_norm": 0.9296875,
|
9696 |
+
"learning_rate": 4.36048754245624e-06,
|
9697 |
+
"loss": 0.9798,
|
9698 |
+
"step": 1383
|
9699 |
+
},
|
9700 |
+
{
|
9701 |
+
"epoch": 1.7555555555555555,
|
9702 |
+
"grad_norm": 0.984375,
|
9703 |
+
"learning_rate": 4.356990338342255e-06,
|
9704 |
+
"loss": 1.0822,
|
9705 |
+
"step": 1384
|
9706 |
+
},
|
9707 |
+
{
|
9708 |
+
"epoch": 1.7568253968253968,
|
9709 |
+
"grad_norm": 0.9453125,
|
9710 |
+
"learning_rate": 4.353492223800056e-06,
|
9711 |
+
"loss": 1.069,
|
9712 |
+
"step": 1385
|
9713 |
+
},
|
9714 |
+
{
|
9715 |
+
"epoch": 1.7580952380952382,
|
9716 |
+
"grad_norm": 0.96875,
|
9717 |
+
"learning_rate": 4.349993202545884e-06,
|
9718 |
+
"loss": 1.1006,
|
9719 |
+
"step": 1386
|
9720 |
+
},
|
9721 |
+
{
|
9722 |
+
"epoch": 1.7593650793650792,
|
9723 |
+
"grad_norm": 0.9453125,
|
9724 |
+
"learning_rate": 4.346493278296945e-06,
|
9725 |
+
"loss": 1.1396,
|
9726 |
+
"step": 1387
|
9727 |
+
},
|
9728 |
+
{
|
9729 |
+
"epoch": 1.7606349206349208,
|
9730 |
+
"grad_norm": 0.921875,
|
9731 |
+
"learning_rate": 4.342992454771401e-06,
|
9732 |
+
"loss": 1.0967,
|
9733 |
+
"step": 1388
|
9734 |
+
},
|
9735 |
+
{
|
9736 |
+
"epoch": 1.7619047619047619,
|
9737 |
+
"grad_norm": 1.0078125,
|
9738 |
+
"learning_rate": 4.33949073568837e-06,
|
9739 |
+
"loss": 1.172,
|
9740 |
+
"step": 1389
|
9741 |
+
},
|
9742 |
+
{
|
9743 |
+
"epoch": 1.7631746031746032,
|
9744 |
+
"grad_norm": 0.95703125,
|
9745 |
+
"learning_rate": 4.335988124767919e-06,
|
9746 |
+
"loss": 1.1301,
|
9747 |
+
"step": 1390
|
9748 |
+
},
|
9749 |
+
{
|
9750 |
+
"epoch": 1.7644444444444445,
|
9751 |
+
"grad_norm": 0.90625,
|
9752 |
+
"learning_rate": 4.332484625731069e-06,
|
9753 |
+
"loss": 1.1052,
|
9754 |
+
"step": 1391
|
9755 |
+
},
|
9756 |
+
{
|
9757 |
+
"epoch": 1.7657142857142856,
|
9758 |
+
"grad_norm": 0.96875,
|
9759 |
+
"learning_rate": 4.328980242299781e-06,
|
9760 |
+
"loss": 1.1134,
|
9761 |
+
"step": 1392
|
9762 |
+
},
|
9763 |
+
{
|
9764 |
+
"epoch": 1.766984126984127,
|
9765 |
+
"grad_norm": 1.015625,
|
9766 |
+
"learning_rate": 4.325474978196952e-06,
|
9767 |
+
"loss": 1.1061,
|
9768 |
+
"step": 1393
|
9769 |
+
},
|
9770 |
+
{
|
9771 |
+
"epoch": 1.7682539682539682,
|
9772 |
+
"grad_norm": 1.0,
|
9773 |
+
"learning_rate": 4.321968837146421e-06,
|
9774 |
+
"loss": 1.0766,
|
9775 |
+
"step": 1394
|
9776 |
+
},
|
9777 |
+
{
|
9778 |
+
"epoch": 1.7695238095238095,
|
9779 |
+
"grad_norm": 0.99609375,
|
9780 |
+
"learning_rate": 4.318461822872954e-06,
|
9781 |
+
"loss": 1.1707,
|
9782 |
+
"step": 1395
|
9783 |
+
},
|
9784 |
+
{
|
9785 |
+
"epoch": 1.7707936507936508,
|
9786 |
+
"grad_norm": 0.94921875,
|
9787 |
+
"learning_rate": 4.314953939102245e-06,
|
9788 |
+
"loss": 1.0879,
|
9789 |
+
"step": 1396
|
9790 |
+
},
|
9791 |
+
{
|
9792 |
+
"epoch": 1.7720634920634921,
|
9793 |
+
"grad_norm": 0.984375,
|
9794 |
+
"learning_rate": 4.311445189560915e-06,
|
9795 |
+
"loss": 1.0676,
|
9796 |
+
"step": 1397
|
9797 |
+
},
|
9798 |
+
{
|
9799 |
+
"epoch": 1.7733333333333334,
|
9800 |
+
"grad_norm": 0.953125,
|
9801 |
+
"learning_rate": 4.307935577976502e-06,
|
9802 |
+
"loss": 1.1716,
|
9803 |
+
"step": 1398
|
9804 |
+
},
|
9805 |
+
{
|
9806 |
+
"epoch": 1.7746031746031745,
|
9807 |
+
"grad_norm": 0.9296875,
|
9808 |
+
"learning_rate": 4.3044251080774585e-06,
|
9809 |
+
"loss": 1.0704,
|
9810 |
+
"step": 1399
|
9811 |
+
},
|
9812 |
+
{
|
9813 |
+
"epoch": 1.7758730158730158,
|
9814 |
+
"grad_norm": 0.98046875,
|
9815 |
+
"learning_rate": 4.300913783593154e-06,
|
9816 |
+
"loss": 1.1177,
|
9817 |
+
"step": 1400
|
9818 |
+
},
|
9819 |
+
{
|
9820 |
+
"epoch": 1.7771428571428571,
|
9821 |
+
"grad_norm": 0.9765625,
|
9822 |
+
"learning_rate": 4.2974016082538605e-06,
|
9823 |
+
"loss": 1.1566,
|
9824 |
+
"step": 1401
|
9825 |
+
},
|
9826 |
+
{
|
9827 |
+
"epoch": 1.7784126984126984,
|
9828 |
+
"grad_norm": 0.95703125,
|
9829 |
+
"learning_rate": 4.293888585790755e-06,
|
9830 |
+
"loss": 1.1727,
|
9831 |
+
"step": 1402
|
9832 |
+
},
|
9833 |
+
{
|
9834 |
+
"epoch": 1.7796825396825398,
|
9835 |
+
"grad_norm": 0.91796875,
|
9836 |
+
"learning_rate": 4.290374719935918e-06,
|
9837 |
+
"loss": 1.102,
|
9838 |
+
"step": 1403
|
9839 |
+
},
|
9840 |
+
{
|
9841 |
+
"epoch": 1.7809523809523808,
|
9842 |
+
"grad_norm": 0.93359375,
|
9843 |
+
"learning_rate": 4.286860014422323e-06,
|
9844 |
+
"loss": 1.0966,
|
9845 |
+
"step": 1404
|
9846 |
+
},
|
9847 |
+
{
|
9848 |
+
"epoch": 1.7822222222222224,
|
9849 |
+
"grad_norm": 1.015625,
|
9850 |
+
"learning_rate": 4.283344472983835e-06,
|
9851 |
+
"loss": 1.1441,
|
9852 |
+
"step": 1405
|
9853 |
+
},
|
9854 |
+
{
|
9855 |
+
"epoch": 1.7834920634920635,
|
9856 |
+
"grad_norm": 0.91015625,
|
9857 |
+
"learning_rate": 4.279828099355209e-06,
|
9858 |
+
"loss": 1.1356,
|
9859 |
+
"step": 1406
|
9860 |
+
},
|
9861 |
+
{
|
9862 |
+
"epoch": 1.7847619047619048,
|
9863 |
+
"grad_norm": 0.94921875,
|
9864 |
+
"learning_rate": 4.2763108972720815e-06,
|
9865 |
+
"loss": 1.0482,
|
9866 |
+
"step": 1407
|
9867 |
+
},
|
9868 |
+
{
|
9869 |
+
"epoch": 1.786031746031746,
|
9870 |
+
"grad_norm": 0.8828125,
|
9871 |
+
"learning_rate": 4.272792870470971e-06,
|
9872 |
+
"loss": 1.0668,
|
9873 |
+
"step": 1408
|
9874 |
+
},
|
9875 |
+
{
|
9876 |
+
"epoch": 1.7873015873015872,
|
9877 |
+
"grad_norm": 0.9296875,
|
9878 |
+
"learning_rate": 4.269274022689275e-06,
|
9879 |
+
"loss": 1.0832,
|
9880 |
+
"step": 1409
|
9881 |
+
},
|
9882 |
+
{
|
9883 |
+
"epoch": 1.7885714285714287,
|
9884 |
+
"grad_norm": 0.8828125,
|
9885 |
+
"learning_rate": 4.265754357665256e-06,
|
9886 |
+
"loss": 1.0726,
|
9887 |
+
"step": 1410
|
9888 |
+
},
|
9889 |
+
{
|
9890 |
+
"epoch": 1.7898412698412698,
|
9891 |
+
"grad_norm": 0.984375,
|
9892 |
+
"learning_rate": 4.26223387913805e-06,
|
9893 |
+
"loss": 1.106,
|
9894 |
+
"step": 1411
|
9895 |
+
},
|
9896 |
+
{
|
9897 |
+
"epoch": 1.791111111111111,
|
9898 |
+
"grad_norm": 1.0859375,
|
9899 |
+
"learning_rate": 4.258712590847656e-06,
|
9900 |
+
"loss": 1.0797,
|
9901 |
+
"step": 1412
|
9902 |
+
},
|
9903 |
+
{
|
9904 |
+
"epoch": 1.7923809523809524,
|
9905 |
+
"grad_norm": 0.96875,
|
9906 |
+
"learning_rate": 4.255190496534934e-06,
|
9907 |
+
"loss": 1.173,
|
9908 |
+
"step": 1413
|
9909 |
+
},
|
9910 |
+
{
|
9911 |
+
"epoch": 1.7936507936507935,
|
9912 |
+
"grad_norm": 0.93359375,
|
9913 |
+
"learning_rate": 4.251667599941599e-06,
|
9914 |
+
"loss": 1.1476,
|
9915 |
+
"step": 1414
|
9916 |
+
},
|
9917 |
+
{
|
9918 |
+
"epoch": 1.794920634920635,
|
9919 |
+
"grad_norm": 1.0390625,
|
9920 |
+
"learning_rate": 4.248143904810219e-06,
|
9921 |
+
"loss": 1.0581,
|
9922 |
+
"step": 1415
|
9923 |
+
},
|
9924 |
+
{
|
9925 |
+
"epoch": 1.7961904761904761,
|
9926 |
+
"grad_norm": 1.015625,
|
9927 |
+
"learning_rate": 4.2446194148842095e-06,
|
9928 |
+
"loss": 1.142,
|
9929 |
+
"step": 1416
|
9930 |
+
},
|
9931 |
+
{
|
9932 |
+
"epoch": 1.7974603174603174,
|
9933 |
+
"grad_norm": 0.99609375,
|
9934 |
+
"learning_rate": 4.241094133907832e-06,
|
9935 |
+
"loss": 1.1658,
|
9936 |
+
"step": 1417
|
9937 |
+
},
|
9938 |
+
{
|
9939 |
+
"epoch": 1.7987301587301587,
|
9940 |
+
"grad_norm": 0.953125,
|
9941 |
+
"learning_rate": 4.2375680656261864e-06,
|
9942 |
+
"loss": 1.0855,
|
9943 |
+
"step": 1418
|
9944 |
+
},
|
9945 |
+
{
|
9946 |
+
"epoch": 1.8,
|
9947 |
+
"grad_norm": 0.99609375,
|
9948 |
+
"learning_rate": 4.23404121378521e-06,
|
9949 |
+
"loss": 1.0656,
|
9950 |
+
"step": 1419
|
9951 |
+
},
|
9952 |
+
{
|
9953 |
+
"epoch": 1.8012698412698414,
|
9954 |
+
"grad_norm": 0.9453125,
|
9955 |
+
"learning_rate": 4.230513582131674e-06,
|
9956 |
+
"loss": 1.1448,
|
9957 |
+
"step": 1420
|
9958 |
+
},
|
9959 |
+
{
|
9960 |
+
"epoch": 1.8025396825396824,
|
9961 |
+
"grad_norm": 0.9609375,
|
9962 |
+
"learning_rate": 4.2269851744131735e-06,
|
9963 |
+
"loss": 1.1289,
|
9964 |
+
"step": 1421
|
9965 |
+
},
|
9966 |
+
{
|
9967 |
+
"epoch": 1.803809523809524,
|
9968 |
+
"grad_norm": 0.99609375,
|
9969 |
+
"learning_rate": 4.223455994378133e-06,
|
9970 |
+
"loss": 1.1863,
|
9971 |
+
"step": 1422
|
9972 |
+
},
|
9973 |
+
{
|
9974 |
+
"epoch": 1.805079365079365,
|
9975 |
+
"grad_norm": 1.015625,
|
9976 |
+
"learning_rate": 4.219926045775797e-06,
|
9977 |
+
"loss": 1.0682,
|
9978 |
+
"step": 1423
|
9979 |
+
},
|
9980 |
+
{
|
9981 |
+
"epoch": 1.8063492063492064,
|
9982 |
+
"grad_norm": 1.0,
|
9983 |
+
"learning_rate": 4.216395332356223e-06,
|
9984 |
+
"loss": 1.1456,
|
9985 |
+
"step": 1424
|
9986 |
+
},
|
9987 |
+
{
|
9988 |
+
"epoch": 1.8076190476190477,
|
9989 |
+
"grad_norm": 1.0234375,
|
9990 |
+
"learning_rate": 4.212863857870283e-06,
|
9991 |
+
"loss": 1.0884,
|
9992 |
+
"step": 1425
|
9993 |
+
},
|
9994 |
+
{
|
9995 |
+
"epoch": 1.8088888888888888,
|
9996 |
+
"grad_norm": 0.98828125,
|
9997 |
+
"learning_rate": 4.209331626069658e-06,
|
9998 |
+
"loss": 1.2046,
|
9999 |
+
"step": 1426
|
10000 |
+
},
|
10001 |
+
{
|
10002 |
+
"epoch": 1.8101587301587303,
|
10003 |
+
"grad_norm": 0.8984375,
|
10004 |
+
"learning_rate": 4.2057986407068335e-06,
|
10005 |
+
"loss": 1.113,
|
10006 |
+
"step": 1427
|
10007 |
+
},
|
10008 |
+
{
|
10009 |
+
"epoch": 1.8114285714285714,
|
10010 |
+
"grad_norm": 1.046875,
|
10011 |
+
"learning_rate": 4.202264905535095e-06,
|
10012 |
+
"loss": 1.0876,
|
10013 |
+
"step": 1428
|
10014 |
+
},
|
10015 |
+
{
|
10016 |
+
"epoch": 1.8126984126984127,
|
10017 |
+
"grad_norm": 0.9609375,
|
10018 |
+
"learning_rate": 4.198730424308525e-06,
|
10019 |
+
"loss": 1.0892,
|
10020 |
+
"step": 1429
|
10021 |
+
},
|
10022 |
+
{
|
10023 |
+
"epoch": 1.813968253968254,
|
10024 |
+
"grad_norm": 1.03125,
|
10025 |
+
"learning_rate": 4.195195200781997e-06,
|
10026 |
+
"loss": 1.1297,
|
10027 |
+
"step": 1430
|
10028 |
+
},
|
10029 |
+
{
|
10030 |
+
"epoch": 1.815238095238095,
|
10031 |
+
"grad_norm": 0.9453125,
|
10032 |
+
"learning_rate": 4.191659238711176e-06,
|
10033 |
+
"loss": 1.0367,
|
10034 |
+
"step": 1431
|
10035 |
+
},
|
10036 |
+
{
|
10037 |
+
"epoch": 1.8165079365079366,
|
10038 |
+
"grad_norm": 0.9296875,
|
10039 |
+
"learning_rate": 4.18812254185251e-06,
|
10040 |
+
"loss": 1.0728,
|
10041 |
+
"step": 1432
|
10042 |
+
},
|
10043 |
+
{
|
10044 |
+
"epoch": 1.8177777777777777,
|
10045 |
+
"grad_norm": 0.98046875,
|
10046 |
+
"learning_rate": 4.184585113963227e-06,
|
10047 |
+
"loss": 1.0955,
|
10048 |
+
"step": 1433
|
10049 |
+
},
|
10050 |
+
{
|
10051 |
+
"epoch": 1.819047619047619,
|
10052 |
+
"grad_norm": 0.9609375,
|
10053 |
+
"learning_rate": 4.181046958801329e-06,
|
10054 |
+
"loss": 1.0745,
|
10055 |
+
"step": 1434
|
10056 |
+
},
|
10057 |
+
{
|
10058 |
+
"epoch": 1.8203174603174603,
|
10059 |
+
"grad_norm": 0.98828125,
|
10060 |
+
"learning_rate": 4.1775080801256005e-06,
|
10061 |
+
"loss": 1.107,
|
10062 |
+
"step": 1435
|
10063 |
+
},
|
10064 |
+
{
|
10065 |
+
"epoch": 1.8215873015873014,
|
10066 |
+
"grad_norm": 0.99609375,
|
10067 |
+
"learning_rate": 4.173968481695583e-06,
|
10068 |
+
"loss": 1.1076,
|
10069 |
+
"step": 1436
|
10070 |
+
},
|
10071 |
+
{
|
10072 |
+
"epoch": 1.822857142857143,
|
10073 |
+
"grad_norm": 0.99609375,
|
10074 |
+
"learning_rate": 4.1704281672715885e-06,
|
10075 |
+
"loss": 1.0636,
|
10076 |
+
"step": 1437
|
10077 |
+
},
|
10078 |
+
{
|
10079 |
+
"epoch": 1.824126984126984,
|
10080 |
+
"grad_norm": 0.9375,
|
10081 |
+
"learning_rate": 4.1668871406146895e-06,
|
10082 |
+
"loss": 1.1531,
|
10083 |
+
"step": 1438
|
10084 |
+
},
|
10085 |
+
{
|
10086 |
+
"epoch": 1.8253968253968254,
|
10087 |
+
"grad_norm": 1.0,
|
10088 |
+
"learning_rate": 4.163345405486714e-06,
|
10089 |
+
"loss": 1.0968,
|
10090 |
+
"step": 1439
|
10091 |
+
},
|
10092 |
+
{
|
10093 |
+
"epoch": 1.8266666666666667,
|
10094 |
+
"grad_norm": 0.94140625,
|
10095 |
+
"learning_rate": 4.1598029656502435e-06,
|
10096 |
+
"loss": 1.1005,
|
10097 |
+
"step": 1440
|
10098 |
+
},
|
10099 |
+
{
|
10100 |
+
"epoch": 1.827936507936508,
|
10101 |
+
"grad_norm": 1.0,
|
10102 |
+
"learning_rate": 4.156259824868607e-06,
|
10103 |
+
"loss": 1.1411,
|
10104 |
+
"step": 1441
|
10105 |
+
},
|
10106 |
+
{
|
10107 |
+
"epoch": 1.8292063492063493,
|
10108 |
+
"grad_norm": 1.015625,
|
10109 |
+
"learning_rate": 4.15271598690588e-06,
|
10110 |
+
"loss": 1.1068,
|
10111 |
+
"step": 1442
|
10112 |
+
},
|
10113 |
+
{
|
10114 |
+
"epoch": 1.8304761904761904,
|
10115 |
+
"grad_norm": 0.96484375,
|
10116 |
+
"learning_rate": 4.149171455526874e-06,
|
10117 |
+
"loss": 1.1175,
|
10118 |
+
"step": 1443
|
10119 |
+
},
|
10120 |
+
{
|
10121 |
+
"epoch": 1.831746031746032,
|
10122 |
+
"grad_norm": 1.0,
|
10123 |
+
"learning_rate": 4.1456262344971455e-06,
|
10124 |
+
"loss": 1.102,
|
10125 |
+
"step": 1444
|
10126 |
+
},
|
10127 |
+
{
|
10128 |
+
"epoch": 1.833015873015873,
|
10129 |
+
"grad_norm": 1.046875,
|
10130 |
+
"learning_rate": 4.142080327582976e-06,
|
10131 |
+
"loss": 1.2087,
|
10132 |
+
"step": 1445
|
10133 |
+
},
|
10134 |
+
{
|
10135 |
+
"epoch": 1.8342857142857143,
|
10136 |
+
"grad_norm": 0.99609375,
|
10137 |
+
"learning_rate": 4.138533738551379e-06,
|
10138 |
+
"loss": 1.1595,
|
10139 |
+
"step": 1446
|
10140 |
+
},
|
10141 |
+
{
|
10142 |
+
"epoch": 1.8355555555555556,
|
10143 |
+
"grad_norm": 0.94140625,
|
10144 |
+
"learning_rate": 4.134986471170092e-06,
|
10145 |
+
"loss": 1.0973,
|
10146 |
+
"step": 1447
|
10147 |
+
},
|
10148 |
+
{
|
10149 |
+
"epoch": 1.8368253968253967,
|
10150 |
+
"grad_norm": 0.93359375,
|
10151 |
+
"learning_rate": 4.1314385292075725e-06,
|
10152 |
+
"loss": 1.0418,
|
10153 |
+
"step": 1448
|
10154 |
+
},
|
10155 |
+
{
|
10156 |
+
"epoch": 1.8380952380952382,
|
10157 |
+
"grad_norm": 1.0234375,
|
10158 |
+
"learning_rate": 4.127889916432997e-06,
|
10159 |
+
"loss": 1.1054,
|
10160 |
+
"step": 1449
|
10161 |
+
},
|
10162 |
+
{
|
10163 |
+
"epoch": 1.8393650793650793,
|
10164 |
+
"grad_norm": 0.9765625,
|
10165 |
+
"learning_rate": 4.124340636616252e-06,
|
10166 |
+
"loss": 1.0538,
|
10167 |
+
"step": 1450
|
10168 |
+
},
|
10169 |
+
{
|
10170 |
+
"epoch": 1.8406349206349206,
|
10171 |
+
"grad_norm": 0.98046875,
|
10172 |
+
"learning_rate": 4.120790693527933e-06,
|
10173 |
+
"loss": 1.1456,
|
10174 |
+
"step": 1451
|
10175 |
+
},
|
10176 |
+
{
|
10177 |
+
"epoch": 1.841904761904762,
|
10178 |
+
"grad_norm": 0.9921875,
|
10179 |
+
"learning_rate": 4.1172400909393414e-06,
|
10180 |
+
"loss": 1.1267,
|
10181 |
+
"step": 1452
|
10182 |
+
},
|
10183 |
+
{
|
10184 |
+
"epoch": 1.843174603174603,
|
10185 |
+
"grad_norm": 0.94921875,
|
10186 |
+
"learning_rate": 4.1136888326224805e-06,
|
10187 |
+
"loss": 1.1251,
|
10188 |
+
"step": 1453
|
10189 |
+
},
|
10190 |
+
{
|
10191 |
+
"epoch": 1.8444444444444446,
|
10192 |
+
"grad_norm": 0.98828125,
|
10193 |
+
"learning_rate": 4.110136922350045e-06,
|
10194 |
+
"loss": 1.1643,
|
10195 |
+
"step": 1454
|
10196 |
+
},
|
10197 |
+
{
|
10198 |
+
"epoch": 1.8457142857142856,
|
10199 |
+
"grad_norm": 1.046875,
|
10200 |
+
"learning_rate": 4.106584363895429e-06,
|
10201 |
+
"loss": 1.0467,
|
10202 |
+
"step": 1455
|
10203 |
+
},
|
10204 |
+
{
|
10205 |
+
"epoch": 1.846984126984127,
|
10206 |
+
"grad_norm": 1.109375,
|
10207 |
+
"learning_rate": 4.103031161032708e-06,
|
10208 |
+
"loss": 1.0191,
|
10209 |
+
"step": 1456
|
10210 |
+
},
|
10211 |
+
{
|
10212 |
+
"epoch": 1.8482539682539683,
|
10213 |
+
"grad_norm": 1.046875,
|
10214 |
+
"learning_rate": 4.099477317536649e-06,
|
10215 |
+
"loss": 1.1017,
|
10216 |
+
"step": 1457
|
10217 |
+
},
|
10218 |
+
{
|
10219 |
+
"epoch": 1.8495238095238096,
|
10220 |
+
"grad_norm": 0.96875,
|
10221 |
+
"learning_rate": 4.0959228371826955e-06,
|
10222 |
+
"loss": 1.1175,
|
10223 |
+
"step": 1458
|
10224 |
+
},
|
10225 |
+
{
|
10226 |
+
"epoch": 1.8507936507936509,
|
10227 |
+
"grad_norm": 0.91796875,
|
10228 |
+
"learning_rate": 4.092367723746968e-06,
|
10229 |
+
"loss": 1.0939,
|
10230 |
+
"step": 1459
|
10231 |
+
},
|
10232 |
+
{
|
10233 |
+
"epoch": 1.852063492063492,
|
10234 |
+
"grad_norm": 0.97265625,
|
10235 |
+
"learning_rate": 4.0888119810062594e-06,
|
10236 |
+
"loss": 1.1079,
|
10237 |
+
"step": 1460
|
10238 |
+
},
|
10239 |
+
{
|
10240 |
+
"epoch": 1.8533333333333335,
|
10241 |
+
"grad_norm": 1.03125,
|
10242 |
+
"learning_rate": 4.085255612738032e-06,
|
10243 |
+
"loss": 1.1314,
|
10244 |
+
"step": 1461
|
10245 |
+
},
|
10246 |
+
{
|
10247 |
+
"epoch": 1.8546031746031746,
|
10248 |
+
"grad_norm": 1.046875,
|
10249 |
+
"learning_rate": 4.081698622720414e-06,
|
10250 |
+
"loss": 1.0679,
|
10251 |
+
"step": 1462
|
10252 |
+
},
|
10253 |
+
{
|
10254 |
+
"epoch": 1.855873015873016,
|
10255 |
+
"grad_norm": 0.91015625,
|
10256 |
+
"learning_rate": 4.078141014732191e-06,
|
10257 |
+
"loss": 1.0438,
|
10258 |
+
"step": 1463
|
10259 |
+
},
|
10260 |
+
{
|
10261 |
+
"epoch": 1.8571428571428572,
|
10262 |
+
"grad_norm": 1.0234375,
|
10263 |
+
"learning_rate": 4.074582792552808e-06,
|
10264 |
+
"loss": 1.0791,
|
10265 |
+
"step": 1464
|
10266 |
+
},
|
10267 |
+
{
|
10268 |
+
"epoch": 1.8584126984126983,
|
10269 |
+
"grad_norm": 1.046875,
|
10270 |
+
"learning_rate": 4.0710239599623606e-06,
|
10271 |
+
"loss": 1.0715,
|
10272 |
+
"step": 1465
|
10273 |
+
},
|
10274 |
+
{
|
10275 |
+
"epoch": 1.8596825396825398,
|
10276 |
+
"grad_norm": 0.97265625,
|
10277 |
+
"learning_rate": 4.067464520741591e-06,
|
10278 |
+
"loss": 1.1739,
|
10279 |
+
"step": 1466
|
10280 |
+
},
|
10281 |
+
{
|
10282 |
+
"epoch": 1.860952380952381,
|
10283 |
+
"grad_norm": 0.92578125,
|
10284 |
+
"learning_rate": 4.0639044786718925e-06,
|
10285 |
+
"loss": 1.0591,
|
10286 |
+
"step": 1467
|
10287 |
+
},
|
10288 |
+
{
|
10289 |
+
"epoch": 1.8622222222222222,
|
10290 |
+
"grad_norm": 0.96875,
|
10291 |
+
"learning_rate": 4.060343837535292e-06,
|
10292 |
+
"loss": 1.1822,
|
10293 |
+
"step": 1468
|
10294 |
+
},
|
10295 |
+
{
|
10296 |
+
"epoch": 1.8634920634920635,
|
10297 |
+
"grad_norm": 0.953125,
|
10298 |
+
"learning_rate": 4.056782601114454e-06,
|
10299 |
+
"loss": 1.1192,
|
10300 |
+
"step": 1469
|
10301 |
+
},
|
10302 |
+
{
|
10303 |
+
"epoch": 1.8647619047619046,
|
10304 |
+
"grad_norm": 1.0,
|
10305 |
+
"learning_rate": 4.053220773192681e-06,
|
10306 |
+
"loss": 1.1677,
|
10307 |
+
"step": 1470
|
10308 |
+
},
|
10309 |
+
{
|
10310 |
+
"epoch": 1.8660317460317462,
|
10311 |
+
"grad_norm": 1.046875,
|
10312 |
+
"learning_rate": 4.049658357553897e-06,
|
10313 |
+
"loss": 1.0656,
|
10314 |
+
"step": 1471
|
10315 |
+
},
|
10316 |
+
{
|
10317 |
+
"epoch": 1.8673015873015872,
|
10318 |
+
"grad_norm": 0.93359375,
|
10319 |
+
"learning_rate": 4.046095357982652e-06,
|
10320 |
+
"loss": 1.0669,
|
10321 |
+
"step": 1472
|
10322 |
+
},
|
10323 |
+
{
|
10324 |
+
"epoch": 1.8685714285714285,
|
10325 |
+
"grad_norm": 0.9921875,
|
10326 |
+
"learning_rate": 4.042531778264121e-06,
|
10327 |
+
"loss": 1.1329,
|
10328 |
+
"step": 1473
|
10329 |
+
},
|
10330 |
+
{
|
10331 |
+
"epoch": 1.8698412698412699,
|
10332 |
+
"grad_norm": 0.8984375,
|
10333 |
+
"learning_rate": 4.038967622184088e-06,
|
10334 |
+
"loss": 1.0856,
|
10335 |
+
"step": 1474
|
10336 |
+
},
|
10337 |
+
{
|
10338 |
+
"epoch": 1.871111111111111,
|
10339 |
+
"grad_norm": 0.9453125,
|
10340 |
+
"learning_rate": 4.035402893528955e-06,
|
10341 |
+
"loss": 1.1338,
|
10342 |
+
"step": 1475
|
10343 |
+
},
|
10344 |
+
{
|
10345 |
+
"epoch": 1.8723809523809525,
|
10346 |
+
"grad_norm": 0.99609375,
|
10347 |
+
"learning_rate": 4.03183759608573e-06,
|
10348 |
+
"loss": 1.0468,
|
10349 |
+
"step": 1476
|
10350 |
+
},
|
10351 |
+
{
|
10352 |
+
"epoch": 1.8736507936507936,
|
10353 |
+
"grad_norm": 1.03125,
|
10354 |
+
"learning_rate": 4.0282717336420254e-06,
|
10355 |
+
"loss": 1.1713,
|
10356 |
+
"step": 1477
|
10357 |
+
},
|
10358 |
+
{
|
10359 |
+
"epoch": 1.8749206349206349,
|
10360 |
+
"grad_norm": 0.9765625,
|
10361 |
+
"learning_rate": 4.024705309986054e-06,
|
10362 |
+
"loss": 1.1247,
|
10363 |
+
"step": 1478
|
10364 |
+
},
|
10365 |
+
{
|
10366 |
+
"epoch": 1.8761904761904762,
|
10367 |
+
"grad_norm": 1.03125,
|
10368 |
+
"learning_rate": 4.021138328906626e-06,
|
10369 |
+
"loss": 1.0776,
|
10370 |
+
"step": 1479
|
10371 |
+
},
|
10372 |
+
{
|
10373 |
+
"epoch": 1.8774603174603175,
|
10374 |
+
"grad_norm": 0.98828125,
|
10375 |
+
"learning_rate": 4.017570794193141e-06,
|
10376 |
+
"loss": 1.0879,
|
10377 |
+
"step": 1480
|
10378 |
+
},
|
10379 |
+
{
|
10380 |
+
"epoch": 1.8787301587301588,
|
10381 |
+
"grad_norm": 0.9765625,
|
10382 |
+
"learning_rate": 4.014002709635589e-06,
|
10383 |
+
"loss": 1.13,
|
10384 |
+
"step": 1481
|
10385 |
+
},
|
10386 |
+
{
|
10387 |
+
"epoch": 1.88,
|
10388 |
+
"grad_norm": 1.0234375,
|
10389 |
+
"learning_rate": 4.010434079024543e-06,
|
10390 |
+
"loss": 1.1174,
|
10391 |
+
"step": 1482
|
10392 |
+
},
|
10393 |
+
{
|
10394 |
+
"epoch": 1.8812698412698414,
|
10395 |
+
"grad_norm": 0.953125,
|
10396 |
+
"learning_rate": 4.006864906151157e-06,
|
10397 |
+
"loss": 1.1313,
|
10398 |
+
"step": 1483
|
10399 |
+
},
|
10400 |
+
{
|
10401 |
+
"epoch": 1.8825396825396825,
|
10402 |
+
"grad_norm": 1.0,
|
10403 |
+
"learning_rate": 4.003295194807162e-06,
|
10404 |
+
"loss": 1.1686,
|
10405 |
+
"step": 1484
|
10406 |
+
},
|
10407 |
+
{
|
10408 |
+
"epoch": 1.8838095238095238,
|
10409 |
+
"grad_norm": 1.0234375,
|
10410 |
+
"learning_rate": 3.999724948784857e-06,
|
10411 |
+
"loss": 1.0579,
|
10412 |
+
"step": 1485
|
10413 |
+
},
|
10414 |
+
{
|
10415 |
+
"epoch": 1.8850793650793651,
|
10416 |
+
"grad_norm": 1.046875,
|
10417 |
+
"learning_rate": 3.996154171877113e-06,
|
10418 |
+
"loss": 1.1215,
|
10419 |
+
"step": 1486
|
10420 |
+
},
|
10421 |
+
{
|
10422 |
+
"epoch": 1.8863492063492062,
|
10423 |
+
"grad_norm": 1.0078125,
|
10424 |
+
"learning_rate": 3.992582867877363e-06,
|
10425 |
+
"loss": 1.1345,
|
10426 |
+
"step": 1487
|
10427 |
+
},
|
10428 |
+
{
|
10429 |
+
"epoch": 1.8876190476190478,
|
10430 |
+
"grad_norm": 0.95703125,
|
10431 |
+
"learning_rate": 3.989011040579602e-06,
|
10432 |
+
"loss": 1.0882,
|
10433 |
+
"step": 1488
|
10434 |
+
},
|
10435 |
+
{
|
10436 |
+
"epoch": 1.8888888888888888,
|
10437 |
+
"grad_norm": 1.1953125,
|
10438 |
+
"learning_rate": 3.98543869377838e-06,
|
10439 |
+
"loss": 1.1399,
|
10440 |
+
"step": 1489
|
10441 |
+
},
|
10442 |
+
{
|
10443 |
+
"epoch": 1.8901587301587301,
|
10444 |
+
"grad_norm": 1.0703125,
|
10445 |
+
"learning_rate": 3.981865831268797e-06,
|
10446 |
+
"loss": 1.1774,
|
10447 |
+
"step": 1490
|
10448 |
+
},
|
10449 |
+
{
|
10450 |
+
"epoch": 1.8914285714285715,
|
10451 |
+
"grad_norm": 0.97265625,
|
10452 |
+
"learning_rate": 3.978292456846502e-06,
|
10453 |
+
"loss": 1.1639,
|
10454 |
+
"step": 1491
|
10455 |
+
},
|
10456 |
+
{
|
10457 |
+
"epoch": 1.8926984126984125,
|
10458 |
+
"grad_norm": 0.99609375,
|
10459 |
+
"learning_rate": 3.974718574307689e-06,
|
10460 |
+
"loss": 1.1198,
|
10461 |
+
"step": 1492
|
10462 |
+
},
|
10463 |
+
{
|
10464 |
+
"epoch": 1.893968253968254,
|
10465 |
+
"grad_norm": 1.1171875,
|
10466 |
+
"learning_rate": 3.971144187449093e-06,
|
10467 |
+
"loss": 1.0223,
|
10468 |
+
"step": 1493
|
10469 |
+
},
|
10470 |
+
{
|
10471 |
+
"epoch": 1.8952380952380952,
|
10472 |
+
"grad_norm": 1.203125,
|
10473 |
+
"learning_rate": 3.967569300067979e-06,
|
10474 |
+
"loss": 1.1347,
|
10475 |
+
"step": 1494
|
10476 |
+
},
|
10477 |
+
{
|
10478 |
+
"epoch": 1.8965079365079365,
|
10479 |
+
"grad_norm": 1.1953125,
|
10480 |
+
"learning_rate": 3.96399391596215e-06,
|
10481 |
+
"loss": 1.1409,
|
10482 |
+
"step": 1495
|
10483 |
+
},
|
10484 |
+
{
|
10485 |
+
"epoch": 1.8977777777777778,
|
10486 |
+
"grad_norm": 0.9453125,
|
10487 |
+
"learning_rate": 3.960418038929933e-06,
|
10488 |
+
"loss": 1.0625,
|
10489 |
+
"step": 1496
|
10490 |
+
},
|
10491 |
+
{
|
10492 |
+
"epoch": 1.899047619047619,
|
10493 |
+
"grad_norm": 0.98828125,
|
10494 |
+
"learning_rate": 3.956841672770181e-06,
|
10495 |
+
"loss": 1.0833,
|
10496 |
+
"step": 1497
|
10497 |
+
},
|
10498 |
+
{
|
10499 |
+
"epoch": 1.9003174603174604,
|
10500 |
+
"grad_norm": 1.1015625,
|
10501 |
+
"learning_rate": 3.953264821282264e-06,
|
10502 |
+
"loss": 1.182,
|
10503 |
+
"step": 1498
|
10504 |
+
},
|
10505 |
+
{
|
10506 |
+
"epoch": 1.9015873015873015,
|
10507 |
+
"grad_norm": 1.0390625,
|
10508 |
+
"learning_rate": 3.949687488266071e-06,
|
10509 |
+
"loss": 1.1809,
|
10510 |
+
"step": 1499
|
10511 |
+
},
|
10512 |
+
{
|
10513 |
+
"epoch": 1.9028571428571428,
|
10514 |
+
"grad_norm": 1.1640625,
|
10515 |
+
"learning_rate": 3.946109677521999e-06,
|
10516 |
+
"loss": 1.2017,
|
10517 |
+
"step": 1500
|
10518 |
+
},
|
10519 |
+
{
|
10520 |
+
"epoch": 1.9041269841269841,
|
10521 |
+
"grad_norm": 1.0234375,
|
10522 |
+
"learning_rate": 3.942531392850953e-06,
|
10523 |
+
"loss": 1.1001,
|
10524 |
+
"step": 1501
|
10525 |
+
},
|
10526 |
+
{
|
10527 |
+
"epoch": 1.9053968253968254,
|
10528 |
+
"grad_norm": 0.9921875,
|
10529 |
+
"learning_rate": 3.938952638054347e-06,
|
10530 |
+
"loss": 1.1442,
|
10531 |
+
"step": 1502
|
10532 |
+
},
|
10533 |
+
{
|
10534 |
+
"epoch": 1.9066666666666667,
|
10535 |
+
"grad_norm": 0.9609375,
|
10536 |
+
"learning_rate": 3.935373416934086e-06,
|
10537 |
+
"loss": 1.1296,
|
10538 |
+
"step": 1503
|
10539 |
+
},
|
10540 |
+
{
|
10541 |
+
"epoch": 1.9079365079365078,
|
10542 |
+
"grad_norm": 1.15625,
|
10543 |
+
"learning_rate": 3.931793733292575e-06,
|
10544 |
+
"loss": 1.1029,
|
10545 |
+
"step": 1504
|
10546 |
+
},
|
10547 |
+
{
|
10548 |
+
"epoch": 1.9092063492063494,
|
10549 |
+
"grad_norm": 1.0859375,
|
10550 |
+
"learning_rate": 3.928213590932712e-06,
|
10551 |
+
"loss": 1.1349,
|
10552 |
+
"step": 1505
|
10553 |
+
},
|
10554 |
+
{
|
10555 |
+
"epoch": 1.9104761904761904,
|
10556 |
+
"grad_norm": 1.0234375,
|
10557 |
+
"learning_rate": 3.924632993657879e-06,
|
10558 |
+
"loss": 1.0712,
|
10559 |
+
"step": 1506
|
10560 |
+
},
|
10561 |
+
{
|
10562 |
+
"epoch": 1.9117460317460317,
|
10563 |
+
"grad_norm": 0.94921875,
|
10564 |
+
"learning_rate": 3.921051945271941e-06,
|
10565 |
+
"loss": 1.074,
|
10566 |
+
"step": 1507
|
10567 |
+
},
|
10568 |
+
{
|
10569 |
+
"epoch": 1.913015873015873,
|
10570 |
+
"grad_norm": 1.0546875,
|
10571 |
+
"learning_rate": 3.917470449579244e-06,
|
10572 |
+
"loss": 1.1432,
|
10573 |
+
"step": 1508
|
10574 |
+
},
|
10575 |
+
{
|
10576 |
+
"epoch": 1.9142857142857141,
|
10577 |
+
"grad_norm": 1.1640625,
|
10578 |
+
"learning_rate": 3.913888510384612e-06,
|
10579 |
+
"loss": 1.1314,
|
10580 |
+
"step": 1509
|
10581 |
+
},
|
10582 |
+
{
|
10583 |
+
"epoch": 1.9155555555555557,
|
10584 |
+
"grad_norm": 1.25,
|
10585 |
+
"learning_rate": 3.910306131493335e-06,
|
10586 |
+
"loss": 1.0772,
|
10587 |
+
"step": 1510
|
10588 |
+
},
|
10589 |
+
{
|
10590 |
+
"epoch": 1.9168253968253968,
|
10591 |
+
"grad_norm": 1.0859375,
|
10592 |
+
"learning_rate": 3.90672331671117e-06,
|
10593 |
+
"loss": 1.1526,
|
10594 |
+
"step": 1511
|
10595 |
+
},
|
10596 |
+
{
|
10597 |
+
"epoch": 1.918095238095238,
|
10598 |
+
"grad_norm": 0.9296875,
|
10599 |
+
"learning_rate": 3.9031400698443414e-06,
|
10600 |
+
"loss": 1.0993,
|
10601 |
+
"step": 1512
|
10602 |
+
},
|
10603 |
+
{
|
10604 |
+
"epoch": 1.9193650793650794,
|
10605 |
+
"grad_norm": 1.0703125,
|
10606 |
+
"learning_rate": 3.8995563946995305e-06,
|
10607 |
+
"loss": 1.0668,
|
10608 |
+
"step": 1513
|
10609 |
+
},
|
10610 |
+
{
|
10611 |
+
"epoch": 1.9206349206349205,
|
10612 |
+
"grad_norm": 1.296875,
|
10613 |
+
"learning_rate": 3.8959722950838745e-06,
|
10614 |
+
"loss": 1.1875,
|
10615 |
+
"step": 1514
|
10616 |
+
},
|
10617 |
+
{
|
10618 |
+
"epoch": 1.921904761904762,
|
10619 |
+
"grad_norm": 1.25,
|
10620 |
+
"learning_rate": 3.8923877748049575e-06,
|
10621 |
+
"loss": 1.1899,
|
10622 |
+
"step": 1515
|
10623 |
+
},
|
10624 |
+
{
|
10625 |
+
"epoch": 1.923174603174603,
|
10626 |
+
"grad_norm": 0.93359375,
|
10627 |
+
"learning_rate": 3.888802837670816e-06,
|
10628 |
+
"loss": 1.1052,
|
10629 |
+
"step": 1516
|
10630 |
+
},
|
10631 |
+
{
|
10632 |
+
"epoch": 1.9244444444444444,
|
10633 |
+
"grad_norm": 0.890625,
|
10634 |
+
"learning_rate": 3.885217487489926e-06,
|
10635 |
+
"loss": 1.0402,
|
10636 |
+
"step": 1517
|
10637 |
+
},
|
10638 |
+
{
|
10639 |
+
"epoch": 1.9257142857142857,
|
10640 |
+
"grad_norm": 1.0078125,
|
10641 |
+
"learning_rate": 3.8816317280712026e-06,
|
10642 |
+
"loss": 1.0479,
|
10643 |
+
"step": 1518
|
10644 |
+
},
|
10645 |
+
{
|
10646 |
+
"epoch": 1.926984126984127,
|
10647 |
+
"grad_norm": 1.0078125,
|
10648 |
+
"learning_rate": 3.878045563223999e-06,
|
10649 |
+
"loss": 1.1563,
|
10650 |
+
"step": 1519
|
10651 |
+
},
|
10652 |
+
{
|
10653 |
+
"epoch": 1.9282539682539683,
|
10654 |
+
"grad_norm": 1.078125,
|
10655 |
+
"learning_rate": 3.8744589967580935e-06,
|
10656 |
+
"loss": 1.1338,
|
10657 |
+
"step": 1520
|
10658 |
+
},
|
10659 |
+
{
|
10660 |
+
"epoch": 1.9295238095238094,
|
10661 |
+
"grad_norm": 0.95703125,
|
10662 |
+
"learning_rate": 3.870872032483695e-06,
|
10663 |
+
"loss": 1.0558,
|
10664 |
+
"step": 1521
|
10665 |
+
},
|
10666 |
+
{
|
10667 |
+
"epoch": 1.930793650793651,
|
10668 |
+
"grad_norm": 1.0390625,
|
10669 |
+
"learning_rate": 3.8672846742114325e-06,
|
10670 |
+
"loss": 1.0867,
|
10671 |
+
"step": 1522
|
10672 |
+
},
|
10673 |
+
{
|
10674 |
+
"epoch": 1.932063492063492,
|
10675 |
+
"grad_norm": 0.90625,
|
10676 |
+
"learning_rate": 3.863696925752357e-06,
|
10677 |
+
"loss": 1.1178,
|
10678 |
+
"step": 1523
|
10679 |
+
},
|
10680 |
+
{
|
10681 |
+
"epoch": 1.9333333333333333,
|
10682 |
+
"grad_norm": 1.078125,
|
10683 |
+
"learning_rate": 3.860108790917931e-06,
|
10684 |
+
"loss": 1.1162,
|
10685 |
+
"step": 1524
|
10686 |
+
},
|
10687 |
+
{
|
10688 |
+
"epoch": 1.9346031746031747,
|
10689 |
+
"grad_norm": 1.0234375,
|
10690 |
+
"learning_rate": 3.85652027352003e-06,
|
10691 |
+
"loss": 1.1278,
|
10692 |
+
"step": 1525
|
10693 |
+
},
|
10694 |
+
{
|
10695 |
+
"epoch": 1.9358730158730157,
|
10696 |
+
"grad_norm": 0.98046875,
|
10697 |
+
"learning_rate": 3.8529313773709324e-06,
|
10698 |
+
"loss": 1.0379,
|
10699 |
+
"step": 1526
|
10700 |
+
},
|
10701 |
+
{
|
10702 |
+
"epoch": 1.9371428571428573,
|
10703 |
+
"grad_norm": 0.9921875,
|
10704 |
+
"learning_rate": 3.849342106283322e-06,
|
10705 |
+
"loss": 1.0832,
|
10706 |
+
"step": 1527
|
10707 |
+
},
|
10708 |
+
{
|
10709 |
+
"epoch": 1.9384126984126984,
|
10710 |
+
"grad_norm": 0.97265625,
|
10711 |
+
"learning_rate": 3.84575246407028e-06,
|
10712 |
+
"loss": 1.153,
|
10713 |
+
"step": 1528
|
10714 |
+
},
|
10715 |
+
{
|
10716 |
+
"epoch": 1.9396825396825397,
|
10717 |
+
"grad_norm": 1.0078125,
|
10718 |
+
"learning_rate": 3.8421624545452815e-06,
|
10719 |
+
"loss": 1.1135,
|
10720 |
+
"step": 1529
|
10721 |
+
},
|
10722 |
+
{
|
10723 |
+
"epoch": 1.940952380952381,
|
10724 |
+
"grad_norm": 0.984375,
|
10725 |
+
"learning_rate": 3.838572081522191e-06,
|
10726 |
+
"loss": 1.029,
|
10727 |
+
"step": 1530
|
10728 |
+
},
|
10729 |
+
{
|
10730 |
+
"epoch": 1.942222222222222,
|
10731 |
+
"grad_norm": 1.0625,
|
10732 |
+
"learning_rate": 3.834981348815262e-06,
|
10733 |
+
"loss": 1.0614,
|
10734 |
+
"step": 1531
|
10735 |
+
},
|
10736 |
+
{
|
10737 |
+
"epoch": 1.9434920634920636,
|
10738 |
+
"grad_norm": 1.140625,
|
10739 |
+
"learning_rate": 3.831390260239128e-06,
|
10740 |
+
"loss": 1.1094,
|
10741 |
+
"step": 1532
|
10742 |
+
},
|
10743 |
+
{
|
10744 |
+
"epoch": 1.9447619047619047,
|
10745 |
+
"grad_norm": 0.9375,
|
10746 |
+
"learning_rate": 3.8277988196088e-06,
|
10747 |
+
"loss": 1.0785,
|
10748 |
+
"step": 1533
|
10749 |
+
},
|
10750 |
+
{
|
10751 |
+
"epoch": 1.946031746031746,
|
10752 |
+
"grad_norm": 0.95703125,
|
10753 |
+
"learning_rate": 3.824207030739665e-06,
|
10754 |
+
"loss": 1.0686,
|
10755 |
+
"step": 1534
|
10756 |
+
},
|
10757 |
+
{
|
10758 |
+
"epoch": 1.9473015873015873,
|
10759 |
+
"grad_norm": 1.0390625,
|
10760 |
+
"learning_rate": 3.820614897447478e-06,
|
10761 |
+
"loss": 1.065,
|
10762 |
+
"step": 1535
|
10763 |
+
},
|
10764 |
+
{
|
10765 |
+
"epoch": 1.9485714285714286,
|
10766 |
+
"grad_norm": 0.95703125,
|
10767 |
+
"learning_rate": 3.81702242354836e-06,
|
10768 |
+
"loss": 1.0723,
|
10769 |
+
"step": 1536
|
10770 |
+
},
|
10771 |
+
{
|
10772 |
+
"epoch": 1.94984126984127,
|
10773 |
+
"grad_norm": 1.0,
|
10774 |
+
"learning_rate": 3.813429612858798e-06,
|
10775 |
+
"loss": 1.1189,
|
10776 |
+
"step": 1537
|
10777 |
+
},
|
10778 |
+
{
|
10779 |
+
"epoch": 1.951111111111111,
|
10780 |
+
"grad_norm": 0.96484375,
|
10781 |
+
"learning_rate": 3.809836469195631e-06,
|
10782 |
+
"loss": 1.0674,
|
10783 |
+
"step": 1538
|
10784 |
+
},
|
10785 |
+
{
|
10786 |
+
"epoch": 1.9523809523809523,
|
10787 |
+
"grad_norm": 0.9375,
|
10788 |
+
"learning_rate": 3.806242996376052e-06,
|
10789 |
+
"loss": 1.0869,
|
10790 |
+
"step": 1539
|
10791 |
+
},
|
10792 |
+
{
|
10793 |
+
"epoch": 1.9536507936507936,
|
10794 |
+
"grad_norm": 1.0,
|
10795 |
+
"learning_rate": 3.8026491982176097e-06,
|
10796 |
+
"loss": 1.1377,
|
10797 |
+
"step": 1540
|
10798 |
+
},
|
10799 |
+
{
|
10800 |
+
"epoch": 1.954920634920635,
|
10801 |
+
"grad_norm": 1.03125,
|
10802 |
+
"learning_rate": 3.799055078538194e-06,
|
10803 |
+
"loss": 1.1337,
|
10804 |
+
"step": 1541
|
10805 |
+
},
|
10806 |
+
{
|
10807 |
+
"epoch": 1.9561904761904763,
|
10808 |
+
"grad_norm": 0.98828125,
|
10809 |
+
"learning_rate": 3.7954606411560347e-06,
|
10810 |
+
"loss": 1.1662,
|
10811 |
+
"step": 1542
|
10812 |
+
},
|
10813 |
+
{
|
10814 |
+
"epoch": 1.9574603174603173,
|
10815 |
+
"grad_norm": 0.9375,
|
10816 |
+
"learning_rate": 3.791865889889702e-06,
|
10817 |
+
"loss": 1.0977,
|
10818 |
+
"step": 1543
|
10819 |
+
},
|
10820 |
+
{
|
10821 |
+
"epoch": 1.9587301587301589,
|
10822 |
+
"grad_norm": 0.9375,
|
10823 |
+
"learning_rate": 3.788270828558101e-06,
|
10824 |
+
"loss": 1.0802,
|
10825 |
+
"step": 1544
|
10826 |
+
},
|
10827 |
+
{
|
10828 |
+
"epoch": 1.96,
|
10829 |
+
"grad_norm": 0.953125,
|
10830 |
+
"learning_rate": 3.784675460980462e-06,
|
10831 |
+
"loss": 1.0882,
|
10832 |
+
"step": 1545
|
10833 |
+
},
|
10834 |
+
{
|
10835 |
+
"epoch": 1.9612698412698413,
|
10836 |
+
"grad_norm": 1.0,
|
10837 |
+
"learning_rate": 3.781079790976344e-06,
|
10838 |
+
"loss": 1.0936,
|
10839 |
+
"step": 1546
|
10840 |
+
},
|
10841 |
+
{
|
10842 |
+
"epoch": 1.9625396825396826,
|
10843 |
+
"grad_norm": 0.9609375,
|
10844 |
+
"learning_rate": 3.777483822365624e-06,
|
10845 |
+
"loss": 1.0968,
|
10846 |
+
"step": 1547
|
10847 |
+
},
|
10848 |
+
{
|
10849 |
+
"epoch": 1.9638095238095237,
|
10850 |
+
"grad_norm": 0.97265625,
|
10851 |
+
"learning_rate": 3.7738875589685004e-06,
|
10852 |
+
"loss": 1.1422,
|
10853 |
+
"step": 1548
|
10854 |
+
},
|
10855 |
+
{
|
10856 |
+
"epoch": 1.9650793650793652,
|
10857 |
+
"grad_norm": 1.0078125,
|
10858 |
+
"learning_rate": 3.7702910046054817e-06,
|
10859 |
+
"loss": 1.0776,
|
10860 |
+
"step": 1549
|
10861 |
+
},
|
10862 |
+
{
|
10863 |
+
"epoch": 1.9663492063492063,
|
10864 |
+
"grad_norm": 0.96875,
|
10865 |
+
"learning_rate": 3.766694163097386e-06,
|
10866 |
+
"loss": 1.1129,
|
10867 |
+
"step": 1550
|
10868 |
+
},
|
10869 |
+
{
|
10870 |
+
"epoch": 1.9676190476190476,
|
10871 |
+
"grad_norm": 0.98828125,
|
10872 |
+
"learning_rate": 3.7630970382653382e-06,
|
10873 |
+
"loss": 1.1185,
|
10874 |
+
"step": 1551
|
10875 |
+
},
|
10876 |
+
{
|
10877 |
+
"epoch": 1.968888888888889,
|
10878 |
+
"grad_norm": 1.0546875,
|
10879 |
+
"learning_rate": 3.7594996339307617e-06,
|
10880 |
+
"loss": 1.1754,
|
10881 |
+
"step": 1552
|
10882 |
+
},
|
10883 |
+
{
|
10884 |
+
"epoch": 1.97015873015873,
|
10885 |
+
"grad_norm": 0.921875,
|
10886 |
+
"learning_rate": 3.7559019539153776e-06,
|
10887 |
+
"loss": 1.1038,
|
10888 |
+
"step": 1553
|
10889 |
+
},
|
10890 |
+
{
|
10891 |
+
"epoch": 1.9714285714285715,
|
10892 |
+
"grad_norm": 1.0,
|
10893 |
+
"learning_rate": 3.752304002041203e-06,
|
10894 |
+
"loss": 1.103,
|
10895 |
+
"step": 1554
|
10896 |
+
},
|
10897 |
+
{
|
10898 |
+
"epoch": 1.9726984126984126,
|
10899 |
+
"grad_norm": 0.94140625,
|
10900 |
+
"learning_rate": 3.748705782130538e-06,
|
10901 |
+
"loss": 1.0537,
|
10902 |
+
"step": 1555
|
10903 |
+
},
|
10904 |
+
{
|
10905 |
+
"epoch": 1.973968253968254,
|
10906 |
+
"grad_norm": 0.92578125,
|
10907 |
+
"learning_rate": 3.7451072980059715e-06,
|
10908 |
+
"loss": 1.1092,
|
10909 |
+
"step": 1556
|
10910 |
+
},
|
10911 |
+
{
|
10912 |
+
"epoch": 1.9752380952380952,
|
10913 |
+
"grad_norm": 0.91796875,
|
10914 |
+
"learning_rate": 3.7415085534903725e-06,
|
10915 |
+
"loss": 1.0675,
|
10916 |
+
"step": 1557
|
10917 |
+
},
|
10918 |
+
{
|
10919 |
+
"epoch": 1.9765079365079365,
|
10920 |
+
"grad_norm": 1.0546875,
|
10921 |
+
"learning_rate": 3.737909552406886e-06,
|
10922 |
+
"loss": 1.0737,
|
10923 |
+
"step": 1558
|
10924 |
+
},
|
10925 |
+
{
|
10926 |
+
"epoch": 1.9777777777777779,
|
10927 |
+
"grad_norm": 0.89453125,
|
10928 |
+
"learning_rate": 3.73431029857893e-06,
|
10929 |
+
"loss": 1.1477,
|
10930 |
+
"step": 1559
|
10931 |
+
},
|
10932 |
+
{
|
10933 |
+
"epoch": 1.979047619047619,
|
10934 |
+
"grad_norm": 1.015625,
|
10935 |
+
"learning_rate": 3.7307107958301905e-06,
|
10936 |
+
"loss": 1.1316,
|
10937 |
+
"step": 1560
|
10938 |
+
},
|
10939 |
+
{
|
10940 |
+
"epoch": 1.9803174603174605,
|
10941 |
+
"grad_norm": 0.98046875,
|
10942 |
+
"learning_rate": 3.727111047984618e-06,
|
10943 |
+
"loss": 1.1262,
|
10944 |
+
"step": 1561
|
10945 |
+
},
|
10946 |
+
{
|
10947 |
+
"epoch": 1.9815873015873016,
|
10948 |
+
"grad_norm": 1.078125,
|
10949 |
+
"learning_rate": 3.723511058866424e-06,
|
10950 |
+
"loss": 1.14,
|
10951 |
+
"step": 1562
|
10952 |
+
},
|
10953 |
+
{
|
10954 |
+
"epoch": 1.9828571428571429,
|
10955 |
+
"grad_norm": 1.0625,
|
10956 |
+
"learning_rate": 3.7199108323000767e-06,
|
10957 |
+
"loss": 1.1392,
|
10958 |
+
"step": 1563
|
10959 |
+
},
|
10960 |
+
{
|
10961 |
+
"epoch": 1.9841269841269842,
|
10962 |
+
"grad_norm": 0.984375,
|
10963 |
+
"learning_rate": 3.716310372110295e-06,
|
10964 |
+
"loss": 1.096,
|
10965 |
+
"step": 1564
|
10966 |
+
},
|
10967 |
+
{
|
10968 |
+
"epoch": 1.9853968253968253,
|
10969 |
+
"grad_norm": 0.96875,
|
10970 |
+
"learning_rate": 3.7127096821220467e-06,
|
10971 |
+
"loss": 1.0792,
|
10972 |
+
"step": 1565
|
10973 |
+
},
|
10974 |
+
{
|
10975 |
+
"epoch": 1.9866666666666668,
|
10976 |
+
"grad_norm": 1.0,
|
10977 |
+
"learning_rate": 3.7091087661605437e-06,
|
10978 |
+
"loss": 1.1489,
|
10979 |
+
"step": 1566
|
10980 |
+
},
|
10981 |
+
{
|
10982 |
+
"epoch": 1.987936507936508,
|
10983 |
+
"grad_norm": 0.9921875,
|
10984 |
+
"learning_rate": 3.7055076280512384e-06,
|
10985 |
+
"loss": 1.14,
|
10986 |
+
"step": 1567
|
10987 |
+
},
|
10988 |
+
{
|
10989 |
+
"epoch": 1.9892063492063492,
|
10990 |
+
"grad_norm": 1.0390625,
|
10991 |
+
"learning_rate": 3.701906271619819e-06,
|
10992 |
+
"loss": 1.1288,
|
10993 |
+
"step": 1568
|
10994 |
+
},
|
10995 |
+
{
|
10996 |
+
"epoch": 1.9904761904761905,
|
10997 |
+
"grad_norm": 0.98828125,
|
10998 |
+
"learning_rate": 3.6983047006922063e-06,
|
10999 |
+
"loss": 1.0907,
|
11000 |
+
"step": 1569
|
11001 |
+
},
|
11002 |
+
{
|
11003 |
+
"epoch": 1.9917460317460316,
|
11004 |
+
"grad_norm": 0.91015625,
|
11005 |
+
"learning_rate": 3.694702919094546e-06,
|
11006 |
+
"loss": 1.1328,
|
11007 |
+
"step": 1570
|
11008 |
+
},
|
11009 |
+
{
|
11010 |
+
"epoch": 1.9930158730158731,
|
11011 |
+
"grad_norm": 0.9765625,
|
11012 |
+
"learning_rate": 3.691100930653212e-06,
|
11013 |
+
"loss": 1.1367,
|
11014 |
+
"step": 1571
|
11015 |
+
},
|
11016 |
+
{
|
11017 |
+
"epoch": 1.9942857142857142,
|
11018 |
+
"grad_norm": 1.0859375,
|
11019 |
+
"learning_rate": 3.6874987391947947e-06,
|
11020 |
+
"loss": 1.043,
|
11021 |
+
"step": 1572
|
11022 |
+
},
|
11023 |
+
{
|
11024 |
+
"epoch": 1.9955555555555555,
|
11025 |
+
"grad_norm": 1.015625,
|
11026 |
+
"learning_rate": 3.683896348546102e-06,
|
11027 |
+
"loss": 1.1243,
|
11028 |
+
"step": 1573
|
11029 |
+
},
|
11030 |
+
{
|
11031 |
+
"epoch": 1.9968253968253968,
|
11032 |
+
"grad_norm": 1.0234375,
|
11033 |
+
"learning_rate": 3.6802937625341506e-06,
|
11034 |
+
"loss": 1.1258,
|
11035 |
+
"step": 1574
|
11036 |
+
},
|
11037 |
+
{
|
11038 |
+
"epoch": 1.9980952380952381,
|
11039 |
+
"grad_norm": 0.97265625,
|
11040 |
+
"learning_rate": 3.676690984986169e-06,
|
11041 |
+
"loss": 1.164,
|
11042 |
+
"step": 1575
|
11043 |
+
},
|
11044 |
+
{
|
11045 |
+
"epoch": 1.9993650793650795,
|
11046 |
+
"grad_norm": 1.0390625,
|
11047 |
+
"learning_rate": 3.673088019729587e-06,
|
11048 |
+
"loss": 1.1386,
|
11049 |
+
"step": 1576
|
11050 |
}
|
11051 |
],
|
11052 |
"logging_steps": 1,
|
|
|
11066 |
"attributes": {}
|
11067 |
}
|
11068 |
},
|
11069 |
+
"total_flos": 3.491386209651917e+18,
|
11070 |
"train_batch_size": 2,
|
11071 |
"trial_name": null,
|
11072 |
"trial_params": null
|