Upload folder using huggingface_hub
Browse files- adapter_model.safetensors +1 -1
- optimizer.pt +1 -1
- rng_state.pth +1 -1
- scheduler.pt +1 -1
- trainer_state.json +3503 -3
adapter_model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1852600
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:763a2e6211692a29489c663be78e38405ddeebee7123cf422bef0897660522b5
|
3 |
size 1852600
|
optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1108346
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:f2e89a0b2817e3604c09bf67f0d4eb755faa1736d27ad6237f97d28a1b19ceb5
|
3 |
size 1108346
|
rng_state.pth
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 14244
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:80f2397b349c3d7d219331d0f9f17be72806daa1aedfdb0787087c6d5818527b
|
3 |
size 14244
|
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 sha256:7c26800e6aad79542c71a265990c35470e0f0c258d88e99aef73f16b83c77e16
|
3 |
size 1064
|
trainer_state.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
-
"epoch":
|
5 |
"eval_steps": 500,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -7007,6 +7007,3506 @@
|
|
7007 |
"learning_rate": 6.950819672131148e-05,
|
7008 |
"loss": 0.2519,
|
7009 |
"step": 20000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
7010 |
}
|
7011 |
],
|
7012 |
"logging_steps": 20,
|
@@ -7026,7 +10526,7 @@
|
|
7026 |
"attributes": {}
|
7027 |
}
|
7028 |
},
|
7029 |
-
"total_flos":
|
7030 |
"train_batch_size": 1,
|
7031 |
"trial_name": null,
|
7032 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 146.4307504575961,
|
5 |
"eval_steps": 500,
|
6 |
+
"global_step": 30000,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
7007 |
"learning_rate": 6.950819672131148e-05,
|
7008 |
"loss": 0.2519,
|
7009 |
"step": 20000
|
7010 |
+
},
|
7011 |
+
{
|
7012 |
+
"epoch": 97.71812080536913,
|
7013 |
+
"grad_norm": 3.771855592727661,
|
7014 |
+
"learning_rate": 6.937704918032787e-05,
|
7015 |
+
"loss": 0.1967,
|
7016 |
+
"step": 20020
|
7017 |
+
},
|
7018 |
+
{
|
7019 |
+
"epoch": 97.8157413056742,
|
7020 |
+
"grad_norm": 4.431488990783691,
|
7021 |
+
"learning_rate": 6.924590163934427e-05,
|
7022 |
+
"loss": 0.2112,
|
7023 |
+
"step": 20040
|
7024 |
+
},
|
7025 |
+
{
|
7026 |
+
"epoch": 97.91336180597925,
|
7027 |
+
"grad_norm": 3.970080852508545,
|
7028 |
+
"learning_rate": 6.911475409836066e-05,
|
7029 |
+
"loss": 0.2295,
|
7030 |
+
"step": 20060
|
7031 |
+
},
|
7032 |
+
{
|
7033 |
+
"epoch": 98.01098230628432,
|
7034 |
+
"grad_norm": 2.521176338195801,
|
7035 |
+
"learning_rate": 6.898360655737705e-05,
|
7036 |
+
"loss": 0.2495,
|
7037 |
+
"step": 20080
|
7038 |
+
},
|
7039 |
+
{
|
7040 |
+
"epoch": 98.10860280658939,
|
7041 |
+
"grad_norm": 3.8167226314544678,
|
7042 |
+
"learning_rate": 6.885245901639344e-05,
|
7043 |
+
"loss": 0.2123,
|
7044 |
+
"step": 20100
|
7045 |
+
},
|
7046 |
+
{
|
7047 |
+
"epoch": 98.20622330689444,
|
7048 |
+
"grad_norm": 3.245234727859497,
|
7049 |
+
"learning_rate": 6.872131147540984e-05,
|
7050 |
+
"loss": 0.229,
|
7051 |
+
"step": 20120
|
7052 |
+
},
|
7053 |
+
{
|
7054 |
+
"epoch": 98.30384380719951,
|
7055 |
+
"grad_norm": 3.270099401473999,
|
7056 |
+
"learning_rate": 6.859016393442623e-05,
|
7057 |
+
"loss": 0.1986,
|
7058 |
+
"step": 20140
|
7059 |
+
},
|
7060 |
+
{
|
7061 |
+
"epoch": 98.40146430750458,
|
7062 |
+
"grad_norm": 3.133777379989624,
|
7063 |
+
"learning_rate": 6.845901639344262e-05,
|
7064 |
+
"loss": 0.2407,
|
7065 |
+
"step": 20160
|
7066 |
+
},
|
7067 |
+
{
|
7068 |
+
"epoch": 98.49908480780964,
|
7069 |
+
"grad_norm": 4.500607490539551,
|
7070 |
+
"learning_rate": 6.832786885245903e-05,
|
7071 |
+
"loss": 0.223,
|
7072 |
+
"step": 20180
|
7073 |
+
},
|
7074 |
+
{
|
7075 |
+
"epoch": 98.59670530811471,
|
7076 |
+
"grad_norm": 3.4543895721435547,
|
7077 |
+
"learning_rate": 6.819672131147542e-05,
|
7078 |
+
"loss": 0.2373,
|
7079 |
+
"step": 20200
|
7080 |
+
},
|
7081 |
+
{
|
7082 |
+
"epoch": 98.69432580841976,
|
7083 |
+
"grad_norm": 3.2081384658813477,
|
7084 |
+
"learning_rate": 6.80655737704918e-05,
|
7085 |
+
"loss": 0.2262,
|
7086 |
+
"step": 20220
|
7087 |
+
},
|
7088 |
+
{
|
7089 |
+
"epoch": 98.79194630872483,
|
7090 |
+
"grad_norm": 4.9004926681518555,
|
7091 |
+
"learning_rate": 6.79344262295082e-05,
|
7092 |
+
"loss": 0.2335,
|
7093 |
+
"step": 20240
|
7094 |
+
},
|
7095 |
+
{
|
7096 |
+
"epoch": 98.8895668090299,
|
7097 |
+
"grad_norm": 3.9303877353668213,
|
7098 |
+
"learning_rate": 6.78032786885246e-05,
|
7099 |
+
"loss": 0.2061,
|
7100 |
+
"step": 20260
|
7101 |
+
},
|
7102 |
+
{
|
7103 |
+
"epoch": 98.98718730933496,
|
7104 |
+
"grad_norm": 3.722957134246826,
|
7105 |
+
"learning_rate": 6.767213114754099e-05,
|
7106 |
+
"loss": 0.2518,
|
7107 |
+
"step": 20280
|
7108 |
+
},
|
7109 |
+
{
|
7110 |
+
"epoch": 99.08480780964003,
|
7111 |
+
"grad_norm": 4.127925872802734,
|
7112 |
+
"learning_rate": 6.754098360655739e-05,
|
7113 |
+
"loss": 0.2133,
|
7114 |
+
"step": 20300
|
7115 |
+
},
|
7116 |
+
{
|
7117 |
+
"epoch": 99.1824283099451,
|
7118 |
+
"grad_norm": 4.3639726638793945,
|
7119 |
+
"learning_rate": 6.740983606557378e-05,
|
7120 |
+
"loss": 0.2119,
|
7121 |
+
"step": 20320
|
7122 |
+
},
|
7123 |
+
{
|
7124 |
+
"epoch": 99.28004881025015,
|
7125 |
+
"grad_norm": 3.288351058959961,
|
7126 |
+
"learning_rate": 6.727868852459017e-05,
|
7127 |
+
"loss": 0.2307,
|
7128 |
+
"step": 20340
|
7129 |
+
},
|
7130 |
+
{
|
7131 |
+
"epoch": 99.37766931055522,
|
7132 |
+
"grad_norm": 3.5033068656921387,
|
7133 |
+
"learning_rate": 6.714754098360656e-05,
|
7134 |
+
"loss": 0.2227,
|
7135 |
+
"step": 20360
|
7136 |
+
},
|
7137 |
+
{
|
7138 |
+
"epoch": 99.47528981086027,
|
7139 |
+
"grad_norm": 3.1117262840270996,
|
7140 |
+
"learning_rate": 6.701639344262295e-05,
|
7141 |
+
"loss": 0.2257,
|
7142 |
+
"step": 20380
|
7143 |
+
},
|
7144 |
+
{
|
7145 |
+
"epoch": 99.57291031116534,
|
7146 |
+
"grad_norm": 4.022121906280518,
|
7147 |
+
"learning_rate": 6.688524590163935e-05,
|
7148 |
+
"loss": 0.2336,
|
7149 |
+
"step": 20400
|
7150 |
+
},
|
7151 |
+
{
|
7152 |
+
"epoch": 99.67053081147041,
|
7153 |
+
"grad_norm": 3.4611172676086426,
|
7154 |
+
"learning_rate": 6.675409836065574e-05,
|
7155 |
+
"loss": 0.2192,
|
7156 |
+
"step": 20420
|
7157 |
+
},
|
7158 |
+
{
|
7159 |
+
"epoch": 99.76815131177547,
|
7160 |
+
"grad_norm": 3.3648264408111572,
|
7161 |
+
"learning_rate": 6.662295081967214e-05,
|
7162 |
+
"loss": 0.2482,
|
7163 |
+
"step": 20440
|
7164 |
+
},
|
7165 |
+
{
|
7166 |
+
"epoch": 99.86577181208054,
|
7167 |
+
"grad_norm": 3.8535125255584717,
|
7168 |
+
"learning_rate": 6.649180327868853e-05,
|
7169 |
+
"loss": 0.2314,
|
7170 |
+
"step": 20460
|
7171 |
+
},
|
7172 |
+
{
|
7173 |
+
"epoch": 99.96339231238561,
|
7174 |
+
"grad_norm": 3.8780901432037354,
|
7175 |
+
"learning_rate": 6.636065573770492e-05,
|
7176 |
+
"loss": 0.2203,
|
7177 |
+
"step": 20480
|
7178 |
+
},
|
7179 |
+
{
|
7180 |
+
"epoch": 100.06101281269066,
|
7181 |
+
"grad_norm": 3.479278326034546,
|
7182 |
+
"learning_rate": 6.622950819672131e-05,
|
7183 |
+
"loss": 0.2003,
|
7184 |
+
"step": 20500
|
7185 |
+
},
|
7186 |
+
{
|
7187 |
+
"epoch": 100.15863331299573,
|
7188 |
+
"grad_norm": 3.530697822570801,
|
7189 |
+
"learning_rate": 6.609836065573771e-05,
|
7190 |
+
"loss": 0.2084,
|
7191 |
+
"step": 20520
|
7192 |
+
},
|
7193 |
+
{
|
7194 |
+
"epoch": 100.25625381330079,
|
7195 |
+
"grad_norm": 3.3677327632904053,
|
7196 |
+
"learning_rate": 6.59672131147541e-05,
|
7197 |
+
"loss": 0.2394,
|
7198 |
+
"step": 20540
|
7199 |
+
},
|
7200 |
+
{
|
7201 |
+
"epoch": 100.35387431360586,
|
7202 |
+
"grad_norm": 2.9369733333587646,
|
7203 |
+
"learning_rate": 6.58360655737705e-05,
|
7204 |
+
"loss": 0.2553,
|
7205 |
+
"step": 20560
|
7206 |
+
},
|
7207 |
+
{
|
7208 |
+
"epoch": 100.45149481391093,
|
7209 |
+
"grad_norm": 3.9915401935577393,
|
7210 |
+
"learning_rate": 6.57049180327869e-05,
|
7211 |
+
"loss": 0.2393,
|
7212 |
+
"step": 20580
|
7213 |
+
},
|
7214 |
+
{
|
7215 |
+
"epoch": 100.54911531421598,
|
7216 |
+
"grad_norm": 3.6804075241088867,
|
7217 |
+
"learning_rate": 6.557377049180327e-05,
|
7218 |
+
"loss": 0.2266,
|
7219 |
+
"step": 20600
|
7220 |
+
},
|
7221 |
+
{
|
7222 |
+
"epoch": 100.64673581452105,
|
7223 |
+
"grad_norm": 4.8830156326293945,
|
7224 |
+
"learning_rate": 6.544262295081967e-05,
|
7225 |
+
"loss": 0.19,
|
7226 |
+
"step": 20620
|
7227 |
+
},
|
7228 |
+
{
|
7229 |
+
"epoch": 100.74435631482612,
|
7230 |
+
"grad_norm": 4.2519731521606445,
|
7231 |
+
"learning_rate": 6.531147540983606e-05,
|
7232 |
+
"loss": 0.2416,
|
7233 |
+
"step": 20640
|
7234 |
+
},
|
7235 |
+
{
|
7236 |
+
"epoch": 100.84197681513118,
|
7237 |
+
"grad_norm": 3.413914203643799,
|
7238 |
+
"learning_rate": 6.518032786885247e-05,
|
7239 |
+
"loss": 0.2026,
|
7240 |
+
"step": 20660
|
7241 |
+
},
|
7242 |
+
{
|
7243 |
+
"epoch": 100.93959731543625,
|
7244 |
+
"grad_norm": 3.642609119415283,
|
7245 |
+
"learning_rate": 6.504918032786886e-05,
|
7246 |
+
"loss": 0.2227,
|
7247 |
+
"step": 20680
|
7248 |
+
},
|
7249 |
+
{
|
7250 |
+
"epoch": 101.0372178157413,
|
7251 |
+
"grad_norm": 4.770230770111084,
|
7252 |
+
"learning_rate": 6.491803278688526e-05,
|
7253 |
+
"loss": 0.2277,
|
7254 |
+
"step": 20700
|
7255 |
+
},
|
7256 |
+
{
|
7257 |
+
"epoch": 101.13483831604637,
|
7258 |
+
"grad_norm": 4.115867614746094,
|
7259 |
+
"learning_rate": 6.478688524590165e-05,
|
7260 |
+
"loss": 0.205,
|
7261 |
+
"step": 20720
|
7262 |
+
},
|
7263 |
+
{
|
7264 |
+
"epoch": 101.23245881635144,
|
7265 |
+
"grad_norm": 4.3617048263549805,
|
7266 |
+
"learning_rate": 6.465573770491804e-05,
|
7267 |
+
"loss": 0.2048,
|
7268 |
+
"step": 20740
|
7269 |
+
},
|
7270 |
+
{
|
7271 |
+
"epoch": 101.3300793166565,
|
7272 |
+
"grad_norm": 3.818500518798828,
|
7273 |
+
"learning_rate": 6.452459016393443e-05,
|
7274 |
+
"loss": 0.2245,
|
7275 |
+
"step": 20760
|
7276 |
+
},
|
7277 |
+
{
|
7278 |
+
"epoch": 101.42769981696156,
|
7279 |
+
"grad_norm": 3.3728935718536377,
|
7280 |
+
"learning_rate": 6.439344262295083e-05,
|
7281 |
+
"loss": 0.2342,
|
7282 |
+
"step": 20780
|
7283 |
+
},
|
7284 |
+
{
|
7285 |
+
"epoch": 101.52532031726662,
|
7286 |
+
"grad_norm": 3.2103302478790283,
|
7287 |
+
"learning_rate": 6.426229508196722e-05,
|
7288 |
+
"loss": 0.2024,
|
7289 |
+
"step": 20800
|
7290 |
+
},
|
7291 |
+
{
|
7292 |
+
"epoch": 101.62294081757169,
|
7293 |
+
"grad_norm": 3.5993459224700928,
|
7294 |
+
"learning_rate": 6.413114754098361e-05,
|
7295 |
+
"loss": 0.2272,
|
7296 |
+
"step": 20820
|
7297 |
+
},
|
7298 |
+
{
|
7299 |
+
"epoch": 101.72056131787676,
|
7300 |
+
"grad_norm": 3.478533983230591,
|
7301 |
+
"learning_rate": 6.400000000000001e-05,
|
7302 |
+
"loss": 0.236,
|
7303 |
+
"step": 20840
|
7304 |
+
},
|
7305 |
+
{
|
7306 |
+
"epoch": 101.81818181818181,
|
7307 |
+
"grad_norm": 4.245722770690918,
|
7308 |
+
"learning_rate": 6.386885245901639e-05,
|
7309 |
+
"loss": 0.2246,
|
7310 |
+
"step": 20860
|
7311 |
+
},
|
7312 |
+
{
|
7313 |
+
"epoch": 101.91580231848688,
|
7314 |
+
"grad_norm": 3.5051097869873047,
|
7315 |
+
"learning_rate": 6.373770491803279e-05,
|
7316 |
+
"loss": 0.2194,
|
7317 |
+
"step": 20880
|
7318 |
+
},
|
7319 |
+
{
|
7320 |
+
"epoch": 102.01342281879195,
|
7321 |
+
"grad_norm": 2.882301092147827,
|
7322 |
+
"learning_rate": 6.360655737704918e-05,
|
7323 |
+
"loss": 0.2196,
|
7324 |
+
"step": 20900
|
7325 |
+
},
|
7326 |
+
{
|
7327 |
+
"epoch": 102.111043319097,
|
7328 |
+
"grad_norm": 3.080702066421509,
|
7329 |
+
"learning_rate": 6.347540983606558e-05,
|
7330 |
+
"loss": 0.2109,
|
7331 |
+
"step": 20920
|
7332 |
+
},
|
7333 |
+
{
|
7334 |
+
"epoch": 102.20866381940208,
|
7335 |
+
"grad_norm": 2.9892094135284424,
|
7336 |
+
"learning_rate": 6.334426229508197e-05,
|
7337 |
+
"loss": 0.2086,
|
7338 |
+
"step": 20940
|
7339 |
+
},
|
7340 |
+
{
|
7341 |
+
"epoch": 102.30628431970713,
|
7342 |
+
"grad_norm": 3.401524782180786,
|
7343 |
+
"learning_rate": 6.321311475409837e-05,
|
7344 |
+
"loss": 0.202,
|
7345 |
+
"step": 20960
|
7346 |
+
},
|
7347 |
+
{
|
7348 |
+
"epoch": 102.4039048200122,
|
7349 |
+
"grad_norm": 3.0444400310516357,
|
7350 |
+
"learning_rate": 6.308196721311475e-05,
|
7351 |
+
"loss": 0.2071,
|
7352 |
+
"step": 20980
|
7353 |
+
},
|
7354 |
+
{
|
7355 |
+
"epoch": 102.50152532031727,
|
7356 |
+
"grad_norm": 3.028918743133545,
|
7357 |
+
"learning_rate": 6.295081967213115e-05,
|
7358 |
+
"loss": 0.226,
|
7359 |
+
"step": 21000
|
7360 |
+
},
|
7361 |
+
{
|
7362 |
+
"epoch": 102.59914582062233,
|
7363 |
+
"grad_norm": 3.2344133853912354,
|
7364 |
+
"learning_rate": 6.281967213114754e-05,
|
7365 |
+
"loss": 0.2417,
|
7366 |
+
"step": 21020
|
7367 |
+
},
|
7368 |
+
{
|
7369 |
+
"epoch": 102.6967663209274,
|
7370 |
+
"grad_norm": 3.1719000339508057,
|
7371 |
+
"learning_rate": 6.268852459016393e-05,
|
7372 |
+
"loss": 0.2363,
|
7373 |
+
"step": 21040
|
7374 |
+
},
|
7375 |
+
{
|
7376 |
+
"epoch": 102.79438682123246,
|
7377 |
+
"grad_norm": 3.474695920944214,
|
7378 |
+
"learning_rate": 6.255737704918033e-05,
|
7379 |
+
"loss": 0.2278,
|
7380 |
+
"step": 21060
|
7381 |
+
},
|
7382 |
+
{
|
7383 |
+
"epoch": 102.89200732153752,
|
7384 |
+
"grad_norm": 4.550293445587158,
|
7385 |
+
"learning_rate": 6.242622950819672e-05,
|
7386 |
+
"loss": 0.2087,
|
7387 |
+
"step": 21080
|
7388 |
+
},
|
7389 |
+
{
|
7390 |
+
"epoch": 102.98962782184259,
|
7391 |
+
"grad_norm": 3.7696001529693604,
|
7392 |
+
"learning_rate": 6.229508196721313e-05,
|
7393 |
+
"loss": 0.2305,
|
7394 |
+
"step": 21100
|
7395 |
+
},
|
7396 |
+
{
|
7397 |
+
"epoch": 103.08724832214764,
|
7398 |
+
"grad_norm": 3.388397216796875,
|
7399 |
+
"learning_rate": 6.21639344262295e-05,
|
7400 |
+
"loss": 0.2133,
|
7401 |
+
"step": 21120
|
7402 |
+
},
|
7403 |
+
{
|
7404 |
+
"epoch": 103.18486882245271,
|
7405 |
+
"grad_norm": 4.243392467498779,
|
7406 |
+
"learning_rate": 6.20327868852459e-05,
|
7407 |
+
"loss": 0.213,
|
7408 |
+
"step": 21140
|
7409 |
+
},
|
7410 |
+
{
|
7411 |
+
"epoch": 103.28248932275778,
|
7412 |
+
"grad_norm": 2.646786689758301,
|
7413 |
+
"learning_rate": 6.19016393442623e-05,
|
7414 |
+
"loss": 0.2132,
|
7415 |
+
"step": 21160
|
7416 |
+
},
|
7417 |
+
{
|
7418 |
+
"epoch": 103.38010982306284,
|
7419 |
+
"grad_norm": 3.157848596572876,
|
7420 |
+
"learning_rate": 6.17704918032787e-05,
|
7421 |
+
"loss": 0.1941,
|
7422 |
+
"step": 21180
|
7423 |
+
},
|
7424 |
+
{
|
7425 |
+
"epoch": 103.47773032336791,
|
7426 |
+
"grad_norm": 3.383357286453247,
|
7427 |
+
"learning_rate": 6.163934426229509e-05,
|
7428 |
+
"loss": 0.2211,
|
7429 |
+
"step": 21200
|
7430 |
+
},
|
7431 |
+
{
|
7432 |
+
"epoch": 103.57535082367298,
|
7433 |
+
"grad_norm": 2.940901279449463,
|
7434 |
+
"learning_rate": 6.150819672131148e-05,
|
7435 |
+
"loss": 0.1934,
|
7436 |
+
"step": 21220
|
7437 |
+
},
|
7438 |
+
{
|
7439 |
+
"epoch": 103.67297132397803,
|
7440 |
+
"grad_norm": 4.2973103523254395,
|
7441 |
+
"learning_rate": 6.137704918032787e-05,
|
7442 |
+
"loss": 0.2516,
|
7443 |
+
"step": 21240
|
7444 |
+
},
|
7445 |
+
{
|
7446 |
+
"epoch": 103.7705918242831,
|
7447 |
+
"grad_norm": 3.586219310760498,
|
7448 |
+
"learning_rate": 6.124590163934426e-05,
|
7449 |
+
"loss": 0.262,
|
7450 |
+
"step": 21260
|
7451 |
+
},
|
7452 |
+
{
|
7453 |
+
"epoch": 103.86821232458816,
|
7454 |
+
"grad_norm": 3.7726240158081055,
|
7455 |
+
"learning_rate": 6.111475409836066e-05,
|
7456 |
+
"loss": 0.1876,
|
7457 |
+
"step": 21280
|
7458 |
+
},
|
7459 |
+
{
|
7460 |
+
"epoch": 103.96583282489323,
|
7461 |
+
"grad_norm": 3.5038113594055176,
|
7462 |
+
"learning_rate": 6.098360655737705e-05,
|
7463 |
+
"loss": 0.2156,
|
7464 |
+
"step": 21300
|
7465 |
+
},
|
7466 |
+
{
|
7467 |
+
"epoch": 104.0634533251983,
|
7468 |
+
"grad_norm": 3.6055893898010254,
|
7469 |
+
"learning_rate": 6.085245901639345e-05,
|
7470 |
+
"loss": 0.2386,
|
7471 |
+
"step": 21320
|
7472 |
+
},
|
7473 |
+
{
|
7474 |
+
"epoch": 104.16107382550335,
|
7475 |
+
"grad_norm": 4.037646770477295,
|
7476 |
+
"learning_rate": 6.072131147540984e-05,
|
7477 |
+
"loss": 0.229,
|
7478 |
+
"step": 21340
|
7479 |
+
},
|
7480 |
+
{
|
7481 |
+
"epoch": 104.25869432580842,
|
7482 |
+
"grad_norm": 3.209284543991089,
|
7483 |
+
"learning_rate": 6.0590163934426236e-05,
|
7484 |
+
"loss": 0.2204,
|
7485 |
+
"step": 21360
|
7486 |
+
},
|
7487 |
+
{
|
7488 |
+
"epoch": 104.35631482611349,
|
7489 |
+
"grad_norm": 3.388456106185913,
|
7490 |
+
"learning_rate": 6.0459016393442625e-05,
|
7491 |
+
"loss": 0.2155,
|
7492 |
+
"step": 21380
|
7493 |
+
},
|
7494 |
+
{
|
7495 |
+
"epoch": 104.45393532641855,
|
7496 |
+
"grad_norm": 2.9914731979370117,
|
7497 |
+
"learning_rate": 6.032786885245902e-05,
|
7498 |
+
"loss": 0.2205,
|
7499 |
+
"step": 21400
|
7500 |
+
},
|
7501 |
+
{
|
7502 |
+
"epoch": 104.55155582672361,
|
7503 |
+
"grad_norm": 3.4753572940826416,
|
7504 |
+
"learning_rate": 6.019672131147541e-05,
|
7505 |
+
"loss": 0.226,
|
7506 |
+
"step": 21420
|
7507 |
+
},
|
7508 |
+
{
|
7509 |
+
"epoch": 104.64917632702867,
|
7510 |
+
"grad_norm": 3.400149345397949,
|
7511 |
+
"learning_rate": 6.00655737704918e-05,
|
7512 |
+
"loss": 0.2117,
|
7513 |
+
"step": 21440
|
7514 |
+
},
|
7515 |
+
{
|
7516 |
+
"epoch": 104.74679682733374,
|
7517 |
+
"grad_norm": 3.2313406467437744,
|
7518 |
+
"learning_rate": 5.99344262295082e-05,
|
7519 |
+
"loss": 0.216,
|
7520 |
+
"step": 21460
|
7521 |
+
},
|
7522 |
+
{
|
7523 |
+
"epoch": 104.84441732763881,
|
7524 |
+
"grad_norm": 4.102112770080566,
|
7525 |
+
"learning_rate": 5.9803278688524586e-05,
|
7526 |
+
"loss": 0.1962,
|
7527 |
+
"step": 21480
|
7528 |
+
},
|
7529 |
+
{
|
7530 |
+
"epoch": 104.94203782794386,
|
7531 |
+
"grad_norm": 3.6872830390930176,
|
7532 |
+
"learning_rate": 5.967213114754099e-05,
|
7533 |
+
"loss": 0.2212,
|
7534 |
+
"step": 21500
|
7535 |
+
},
|
7536 |
+
{
|
7537 |
+
"epoch": 105.03965832824893,
|
7538 |
+
"grad_norm": 3.1488096714019775,
|
7539 |
+
"learning_rate": 5.954098360655738e-05,
|
7540 |
+
"loss": 0.1919,
|
7541 |
+
"step": 21520
|
7542 |
+
},
|
7543 |
+
{
|
7544 |
+
"epoch": 105.137278828554,
|
7545 |
+
"grad_norm": 3.4388973712921143,
|
7546 |
+
"learning_rate": 5.9409836065573774e-05,
|
7547 |
+
"loss": 0.247,
|
7548 |
+
"step": 21540
|
7549 |
+
},
|
7550 |
+
{
|
7551 |
+
"epoch": 105.23489932885906,
|
7552 |
+
"grad_norm": 3.776465892791748,
|
7553 |
+
"learning_rate": 5.927868852459016e-05,
|
7554 |
+
"loss": 0.1856,
|
7555 |
+
"step": 21560
|
7556 |
+
},
|
7557 |
+
{
|
7558 |
+
"epoch": 105.33251982916413,
|
7559 |
+
"grad_norm": 3.552684783935547,
|
7560 |
+
"learning_rate": 5.9147540983606566e-05,
|
7561 |
+
"loss": 0.2042,
|
7562 |
+
"step": 21580
|
7563 |
+
},
|
7564 |
+
{
|
7565 |
+
"epoch": 105.43014032946918,
|
7566 |
+
"grad_norm": 2.8810436725616455,
|
7567 |
+
"learning_rate": 5.9016393442622956e-05,
|
7568 |
+
"loss": 0.2353,
|
7569 |
+
"step": 21600
|
7570 |
+
},
|
7571 |
+
{
|
7572 |
+
"epoch": 105.52776082977425,
|
7573 |
+
"grad_norm": 3.2408454418182373,
|
7574 |
+
"learning_rate": 5.888524590163935e-05,
|
7575 |
+
"loss": 0.2148,
|
7576 |
+
"step": 21620
|
7577 |
+
},
|
7578 |
+
{
|
7579 |
+
"epoch": 105.62538133007932,
|
7580 |
+
"grad_norm": 3.5531325340270996,
|
7581 |
+
"learning_rate": 5.875409836065574e-05,
|
7582 |
+
"loss": 0.2106,
|
7583 |
+
"step": 21640
|
7584 |
+
},
|
7585 |
+
{
|
7586 |
+
"epoch": 105.72300183038438,
|
7587 |
+
"grad_norm": 4.188174247741699,
|
7588 |
+
"learning_rate": 5.862295081967213e-05,
|
7589 |
+
"loss": 0.2211,
|
7590 |
+
"step": 21660
|
7591 |
+
},
|
7592 |
+
{
|
7593 |
+
"epoch": 105.82062233068945,
|
7594 |
+
"grad_norm": 4.232003688812256,
|
7595 |
+
"learning_rate": 5.849180327868853e-05,
|
7596 |
+
"loss": 0.2139,
|
7597 |
+
"step": 21680
|
7598 |
+
},
|
7599 |
+
{
|
7600 |
+
"epoch": 105.91824283099452,
|
7601 |
+
"grad_norm": 3.7038660049438477,
|
7602 |
+
"learning_rate": 5.8360655737704916e-05,
|
7603 |
+
"loss": 0.2048,
|
7604 |
+
"step": 21700
|
7605 |
+
},
|
7606 |
+
{
|
7607 |
+
"epoch": 106.01586333129957,
|
7608 |
+
"grad_norm": 2.9898719787597656,
|
7609 |
+
"learning_rate": 5.822950819672132e-05,
|
7610 |
+
"loss": 0.2184,
|
7611 |
+
"step": 21720
|
7612 |
+
},
|
7613 |
+
{
|
7614 |
+
"epoch": 106.11348383160464,
|
7615 |
+
"grad_norm": 3.482365846633911,
|
7616 |
+
"learning_rate": 5.80983606557377e-05,
|
7617 |
+
"loss": 0.2171,
|
7618 |
+
"step": 21740
|
7619 |
+
},
|
7620 |
+
{
|
7621 |
+
"epoch": 106.2111043319097,
|
7622 |
+
"grad_norm": 3.3595712184906006,
|
7623 |
+
"learning_rate": 5.7967213114754104e-05,
|
7624 |
+
"loss": 0.2041,
|
7625 |
+
"step": 21760
|
7626 |
+
},
|
7627 |
+
{
|
7628 |
+
"epoch": 106.30872483221476,
|
7629 |
+
"grad_norm": 3.7470903396606445,
|
7630 |
+
"learning_rate": 5.7836065573770494e-05,
|
7631 |
+
"loss": 0.1946,
|
7632 |
+
"step": 21780
|
7633 |
+
},
|
7634 |
+
{
|
7635 |
+
"epoch": 106.40634533251983,
|
7636 |
+
"grad_norm": 3.418549060821533,
|
7637 |
+
"learning_rate": 5.770491803278689e-05,
|
7638 |
+
"loss": 0.2013,
|
7639 |
+
"step": 21800
|
7640 |
+
},
|
7641 |
+
{
|
7642 |
+
"epoch": 106.50396583282489,
|
7643 |
+
"grad_norm": 3.115424394607544,
|
7644 |
+
"learning_rate": 5.757377049180328e-05,
|
7645 |
+
"loss": 0.2064,
|
7646 |
+
"step": 21820
|
7647 |
+
},
|
7648 |
+
{
|
7649 |
+
"epoch": 106.60158633312996,
|
7650 |
+
"grad_norm": 3.2718710899353027,
|
7651 |
+
"learning_rate": 5.744262295081968e-05,
|
7652 |
+
"loss": 0.2266,
|
7653 |
+
"step": 21840
|
7654 |
+
},
|
7655 |
+
{
|
7656 |
+
"epoch": 106.69920683343503,
|
7657 |
+
"grad_norm": 2.9478771686553955,
|
7658 |
+
"learning_rate": 5.731147540983607e-05,
|
7659 |
+
"loss": 0.2321,
|
7660 |
+
"step": 21860
|
7661 |
+
},
|
7662 |
+
{
|
7663 |
+
"epoch": 106.79682733374008,
|
7664 |
+
"grad_norm": 3.479456663131714,
|
7665 |
+
"learning_rate": 5.7180327868852454e-05,
|
7666 |
+
"loss": 0.2317,
|
7667 |
+
"step": 21880
|
7668 |
+
},
|
7669 |
+
{
|
7670 |
+
"epoch": 106.89444783404515,
|
7671 |
+
"grad_norm": 3.934882879257202,
|
7672 |
+
"learning_rate": 5.704918032786886e-05,
|
7673 |
+
"loss": 0.2074,
|
7674 |
+
"step": 21900
|
7675 |
+
},
|
7676 |
+
{
|
7677 |
+
"epoch": 106.99206833435021,
|
7678 |
+
"grad_norm": 3.40720272064209,
|
7679 |
+
"learning_rate": 5.6918032786885246e-05,
|
7680 |
+
"loss": 0.2123,
|
7681 |
+
"step": 21920
|
7682 |
+
},
|
7683 |
+
{
|
7684 |
+
"epoch": 107.08968883465528,
|
7685 |
+
"grad_norm": 2.825427770614624,
|
7686 |
+
"learning_rate": 5.678688524590164e-05,
|
7687 |
+
"loss": 0.1897,
|
7688 |
+
"step": 21940
|
7689 |
+
},
|
7690 |
+
{
|
7691 |
+
"epoch": 107.18730933496035,
|
7692 |
+
"grad_norm": 3.6434717178344727,
|
7693 |
+
"learning_rate": 5.665573770491803e-05,
|
7694 |
+
"loss": 0.207,
|
7695 |
+
"step": 21960
|
7696 |
+
},
|
7697 |
+
{
|
7698 |
+
"epoch": 107.2849298352654,
|
7699 |
+
"grad_norm": 4.119269847869873,
|
7700 |
+
"learning_rate": 5.6524590163934435e-05,
|
7701 |
+
"loss": 0.2086,
|
7702 |
+
"step": 21980
|
7703 |
+
},
|
7704 |
+
{
|
7705 |
+
"epoch": 107.38255033557047,
|
7706 |
+
"grad_norm": 2.835963487625122,
|
7707 |
+
"learning_rate": 5.639344262295082e-05,
|
7708 |
+
"loss": 0.1942,
|
7709 |
+
"step": 22000
|
7710 |
+
},
|
7711 |
+
{
|
7712 |
+
"epoch": 107.48017083587554,
|
7713 |
+
"grad_norm": 3.175858736038208,
|
7714 |
+
"learning_rate": 5.626229508196722e-05,
|
7715 |
+
"loss": 0.2089,
|
7716 |
+
"step": 22020
|
7717 |
+
},
|
7718 |
+
{
|
7719 |
+
"epoch": 107.5777913361806,
|
7720 |
+
"grad_norm": 3.6265509128570557,
|
7721 |
+
"learning_rate": 5.613114754098361e-05,
|
7722 |
+
"loss": 0.2164,
|
7723 |
+
"step": 22040
|
7724 |
+
},
|
7725 |
+
{
|
7726 |
+
"epoch": 107.67541183648567,
|
7727 |
+
"grad_norm": 2.90639591217041,
|
7728 |
+
"learning_rate": 5.6000000000000006e-05,
|
7729 |
+
"loss": 0.2121,
|
7730 |
+
"step": 22060
|
7731 |
+
},
|
7732 |
+
{
|
7733 |
+
"epoch": 107.77303233679072,
|
7734 |
+
"grad_norm": 4.155264377593994,
|
7735 |
+
"learning_rate": 5.5868852459016395e-05,
|
7736 |
+
"loss": 0.219,
|
7737 |
+
"step": 22080
|
7738 |
+
},
|
7739 |
+
{
|
7740 |
+
"epoch": 107.87065283709579,
|
7741 |
+
"grad_norm": 3.730433940887451,
|
7742 |
+
"learning_rate": 5.5737704918032785e-05,
|
7743 |
+
"loss": 0.2369,
|
7744 |
+
"step": 22100
|
7745 |
+
},
|
7746 |
+
{
|
7747 |
+
"epoch": 107.96827333740086,
|
7748 |
+
"grad_norm": 2.8690969944000244,
|
7749 |
+
"learning_rate": 5.560655737704919e-05,
|
7750 |
+
"loss": 0.2189,
|
7751 |
+
"step": 22120
|
7752 |
+
},
|
7753 |
+
{
|
7754 |
+
"epoch": 108.06589383770591,
|
7755 |
+
"grad_norm": 3.061427116394043,
|
7756 |
+
"learning_rate": 5.547540983606557e-05,
|
7757 |
+
"loss": 0.2203,
|
7758 |
+
"step": 22140
|
7759 |
+
},
|
7760 |
+
{
|
7761 |
+
"epoch": 108.16351433801098,
|
7762 |
+
"grad_norm": 3.6155471801757812,
|
7763 |
+
"learning_rate": 5.534426229508197e-05,
|
7764 |
+
"loss": 0.2043,
|
7765 |
+
"step": 22160
|
7766 |
+
},
|
7767 |
+
{
|
7768 |
+
"epoch": 108.26113483831605,
|
7769 |
+
"grad_norm": 3.1074283123016357,
|
7770 |
+
"learning_rate": 5.521311475409836e-05,
|
7771 |
+
"loss": 0.2095,
|
7772 |
+
"step": 22180
|
7773 |
+
},
|
7774 |
+
{
|
7775 |
+
"epoch": 108.35875533862111,
|
7776 |
+
"grad_norm": 3.6168534755706787,
|
7777 |
+
"learning_rate": 5.508196721311476e-05,
|
7778 |
+
"loss": 0.2339,
|
7779 |
+
"step": 22200
|
7780 |
+
},
|
7781 |
+
{
|
7782 |
+
"epoch": 108.45637583892618,
|
7783 |
+
"grad_norm": 2.9254798889160156,
|
7784 |
+
"learning_rate": 5.495081967213115e-05,
|
7785 |
+
"loss": 0.214,
|
7786 |
+
"step": 22220
|
7787 |
+
},
|
7788 |
+
{
|
7789 |
+
"epoch": 108.55399633923123,
|
7790 |
+
"grad_norm": 3.2898197174072266,
|
7791 |
+
"learning_rate": 5.481967213114755e-05,
|
7792 |
+
"loss": 0.1981,
|
7793 |
+
"step": 22240
|
7794 |
+
},
|
7795 |
+
{
|
7796 |
+
"epoch": 108.6516168395363,
|
7797 |
+
"grad_norm": 3.0724892616271973,
|
7798 |
+
"learning_rate": 5.4688524590163933e-05,
|
7799 |
+
"loss": 0.1997,
|
7800 |
+
"step": 22260
|
7801 |
+
},
|
7802 |
+
{
|
7803 |
+
"epoch": 108.74923733984137,
|
7804 |
+
"grad_norm": 3.3344459533691406,
|
7805 |
+
"learning_rate": 5.4557377049180336e-05,
|
7806 |
+
"loss": 0.2243,
|
7807 |
+
"step": 22280
|
7808 |
+
},
|
7809 |
+
{
|
7810 |
+
"epoch": 108.84685784014643,
|
7811 |
+
"grad_norm": 4.20386266708374,
|
7812 |
+
"learning_rate": 5.4426229508196726e-05,
|
7813 |
+
"loss": 0.1973,
|
7814 |
+
"step": 22300
|
7815 |
+
},
|
7816 |
+
{
|
7817 |
+
"epoch": 108.9444783404515,
|
7818 |
+
"grad_norm": 3.0647547245025635,
|
7819 |
+
"learning_rate": 5.4295081967213115e-05,
|
7820 |
+
"loss": 0.1984,
|
7821 |
+
"step": 22320
|
7822 |
+
},
|
7823 |
+
{
|
7824 |
+
"epoch": 109.04209884075657,
|
7825 |
+
"grad_norm": 3.2792584896087646,
|
7826 |
+
"learning_rate": 5.416393442622951e-05,
|
7827 |
+
"loss": 0.2209,
|
7828 |
+
"step": 22340
|
7829 |
+
},
|
7830 |
+
{
|
7831 |
+
"epoch": 109.13971934106162,
|
7832 |
+
"grad_norm": 2.900493860244751,
|
7833 |
+
"learning_rate": 5.40327868852459e-05,
|
7834 |
+
"loss": 0.2123,
|
7835 |
+
"step": 22360
|
7836 |
+
},
|
7837 |
+
{
|
7838 |
+
"epoch": 109.23733984136669,
|
7839 |
+
"grad_norm": 2.754514217376709,
|
7840 |
+
"learning_rate": 5.3901639344262304e-05,
|
7841 |
+
"loss": 0.2049,
|
7842 |
+
"step": 22380
|
7843 |
+
},
|
7844 |
+
{
|
7845 |
+
"epoch": 109.33496034167175,
|
7846 |
+
"grad_norm": 2.955946445465088,
|
7847 |
+
"learning_rate": 5.3770491803278686e-05,
|
7848 |
+
"loss": 0.2164,
|
7849 |
+
"step": 22400
|
7850 |
+
},
|
7851 |
+
{
|
7852 |
+
"epoch": 109.43258084197682,
|
7853 |
+
"grad_norm": 3.0447773933410645,
|
7854 |
+
"learning_rate": 5.363934426229509e-05,
|
7855 |
+
"loss": 0.1903,
|
7856 |
+
"step": 22420
|
7857 |
+
},
|
7858 |
+
{
|
7859 |
+
"epoch": 109.53020134228188,
|
7860 |
+
"grad_norm": 2.8788065910339355,
|
7861 |
+
"learning_rate": 5.350819672131148e-05,
|
7862 |
+
"loss": 0.2018,
|
7863 |
+
"step": 22440
|
7864 |
+
},
|
7865 |
+
{
|
7866 |
+
"epoch": 109.62782184258694,
|
7867 |
+
"grad_norm": 3.766073703765869,
|
7868 |
+
"learning_rate": 5.3377049180327875e-05,
|
7869 |
+
"loss": 0.2092,
|
7870 |
+
"step": 22460
|
7871 |
+
},
|
7872 |
+
{
|
7873 |
+
"epoch": 109.72544234289201,
|
7874 |
+
"grad_norm": 4.014832973480225,
|
7875 |
+
"learning_rate": 5.3245901639344264e-05,
|
7876 |
+
"loss": 0.2138,
|
7877 |
+
"step": 22480
|
7878 |
+
},
|
7879 |
+
{
|
7880 |
+
"epoch": 109.82306284319706,
|
7881 |
+
"grad_norm": 2.987813949584961,
|
7882 |
+
"learning_rate": 5.311475409836065e-05,
|
7883 |
+
"loss": 0.2253,
|
7884 |
+
"step": 22500
|
7885 |
+
},
|
7886 |
+
{
|
7887 |
+
"epoch": 109.92068334350213,
|
7888 |
+
"grad_norm": 2.980419158935547,
|
7889 |
+
"learning_rate": 5.298360655737705e-05,
|
7890 |
+
"loss": 0.2159,
|
7891 |
+
"step": 22520
|
7892 |
+
},
|
7893 |
+
{
|
7894 |
+
"epoch": 110.0183038438072,
|
7895 |
+
"grad_norm": 3.1005942821502686,
|
7896 |
+
"learning_rate": 5.285245901639344e-05,
|
7897 |
+
"loss": 0.1907,
|
7898 |
+
"step": 22540
|
7899 |
+
},
|
7900 |
+
{
|
7901 |
+
"epoch": 110.11592434411226,
|
7902 |
+
"grad_norm": 3.2539725303649902,
|
7903 |
+
"learning_rate": 5.272131147540984e-05,
|
7904 |
+
"loss": 0.2023,
|
7905 |
+
"step": 22560
|
7906 |
+
},
|
7907 |
+
{
|
7908 |
+
"epoch": 110.21354484441733,
|
7909 |
+
"grad_norm": 3.4975571632385254,
|
7910 |
+
"learning_rate": 5.259016393442623e-05,
|
7911 |
+
"loss": 0.2101,
|
7912 |
+
"step": 22580
|
7913 |
+
},
|
7914 |
+
{
|
7915 |
+
"epoch": 110.3111653447224,
|
7916 |
+
"grad_norm": 4.542675018310547,
|
7917 |
+
"learning_rate": 5.245901639344263e-05,
|
7918 |
+
"loss": 0.2071,
|
7919 |
+
"step": 22600
|
7920 |
+
},
|
7921 |
+
{
|
7922 |
+
"epoch": 110.40878584502745,
|
7923 |
+
"grad_norm": 2.5775978565216064,
|
7924 |
+
"learning_rate": 5.2327868852459017e-05,
|
7925 |
+
"loss": 0.2027,
|
7926 |
+
"step": 22620
|
7927 |
+
},
|
7928 |
+
{
|
7929 |
+
"epoch": 110.50640634533252,
|
7930 |
+
"grad_norm": 2.964486598968506,
|
7931 |
+
"learning_rate": 5.219672131147541e-05,
|
7932 |
+
"loss": 0.2281,
|
7933 |
+
"step": 22640
|
7934 |
+
},
|
7935 |
+
{
|
7936 |
+
"epoch": 110.60402684563758,
|
7937 |
+
"grad_norm": 3.883513927459717,
|
7938 |
+
"learning_rate": 5.20655737704918e-05,
|
7939 |
+
"loss": 0.2083,
|
7940 |
+
"step": 22660
|
7941 |
+
},
|
7942 |
+
{
|
7943 |
+
"epoch": 110.70164734594265,
|
7944 |
+
"grad_norm": 3.696744441986084,
|
7945 |
+
"learning_rate": 5.1934426229508205e-05,
|
7946 |
+
"loss": 0.2044,
|
7947 |
+
"step": 22680
|
7948 |
+
},
|
7949 |
+
{
|
7950 |
+
"epoch": 110.79926784624772,
|
7951 |
+
"grad_norm": 3.104335308074951,
|
7952 |
+
"learning_rate": 5.1803278688524594e-05,
|
7953 |
+
"loss": 0.216,
|
7954 |
+
"step": 22700
|
7955 |
+
},
|
7956 |
+
{
|
7957 |
+
"epoch": 110.89688834655277,
|
7958 |
+
"grad_norm": 3.6240875720977783,
|
7959 |
+
"learning_rate": 5.1672131147540984e-05,
|
7960 |
+
"loss": 0.2086,
|
7961 |
+
"step": 22720
|
7962 |
+
},
|
7963 |
+
{
|
7964 |
+
"epoch": 110.99450884685784,
|
7965 |
+
"grad_norm": 4.600063323974609,
|
7966 |
+
"learning_rate": 5.154098360655738e-05,
|
7967 |
+
"loss": 0.1983,
|
7968 |
+
"step": 22740
|
7969 |
+
},
|
7970 |
+
{
|
7971 |
+
"epoch": 111.09212934716291,
|
7972 |
+
"grad_norm": 4.165120601654053,
|
7973 |
+
"learning_rate": 5.140983606557377e-05,
|
7974 |
+
"loss": 0.207,
|
7975 |
+
"step": 22760
|
7976 |
+
},
|
7977 |
+
{
|
7978 |
+
"epoch": 111.18974984746797,
|
7979 |
+
"grad_norm": 3.1178033351898193,
|
7980 |
+
"learning_rate": 5.1278688524590165e-05,
|
7981 |
+
"loss": 0.1836,
|
7982 |
+
"step": 22780
|
7983 |
+
},
|
7984 |
+
{
|
7985 |
+
"epoch": 111.28737034777303,
|
7986 |
+
"grad_norm": 3.8547523021698,
|
7987 |
+
"learning_rate": 5.1147540983606555e-05,
|
7988 |
+
"loss": 0.213,
|
7989 |
+
"step": 22800
|
7990 |
+
},
|
7991 |
+
{
|
7992 |
+
"epoch": 111.38499084807809,
|
7993 |
+
"grad_norm": 3.626835346221924,
|
7994 |
+
"learning_rate": 5.101639344262296e-05,
|
7995 |
+
"loss": 0.1901,
|
7996 |
+
"step": 22820
|
7997 |
+
},
|
7998 |
+
{
|
7999 |
+
"epoch": 111.48261134838316,
|
8000 |
+
"grad_norm": 3.89408540725708,
|
8001 |
+
"learning_rate": 5.088524590163935e-05,
|
8002 |
+
"loss": 0.2151,
|
8003 |
+
"step": 22840
|
8004 |
+
},
|
8005 |
+
{
|
8006 |
+
"epoch": 111.58023184868823,
|
8007 |
+
"grad_norm": 3.3434460163116455,
|
8008 |
+
"learning_rate": 5.075409836065574e-05,
|
8009 |
+
"loss": 0.2165,
|
8010 |
+
"step": 22860
|
8011 |
+
},
|
8012 |
+
{
|
8013 |
+
"epoch": 111.67785234899328,
|
8014 |
+
"grad_norm": 3.1831305027008057,
|
8015 |
+
"learning_rate": 5.062295081967213e-05,
|
8016 |
+
"loss": 0.1911,
|
8017 |
+
"step": 22880
|
8018 |
+
},
|
8019 |
+
{
|
8020 |
+
"epoch": 111.77547284929835,
|
8021 |
+
"grad_norm": 3.6153972148895264,
|
8022 |
+
"learning_rate": 5.049180327868853e-05,
|
8023 |
+
"loss": 0.2076,
|
8024 |
+
"step": 22900
|
8025 |
+
},
|
8026 |
+
{
|
8027 |
+
"epoch": 111.87309334960342,
|
8028 |
+
"grad_norm": 2.87998628616333,
|
8029 |
+
"learning_rate": 5.036065573770492e-05,
|
8030 |
+
"loss": 0.2121,
|
8031 |
+
"step": 22920
|
8032 |
+
},
|
8033 |
+
{
|
8034 |
+
"epoch": 111.97071384990848,
|
8035 |
+
"grad_norm": 2.9351823329925537,
|
8036 |
+
"learning_rate": 5.022950819672131e-05,
|
8037 |
+
"loss": 0.2126,
|
8038 |
+
"step": 22940
|
8039 |
+
},
|
8040 |
+
{
|
8041 |
+
"epoch": 112.06833435021355,
|
8042 |
+
"grad_norm": 3.098568916320801,
|
8043 |
+
"learning_rate": 5.009836065573771e-05,
|
8044 |
+
"loss": 0.2169,
|
8045 |
+
"step": 22960
|
8046 |
+
},
|
8047 |
+
{
|
8048 |
+
"epoch": 112.1659548505186,
|
8049 |
+
"grad_norm": 4.321131229400635,
|
8050 |
+
"learning_rate": 4.99672131147541e-05,
|
8051 |
+
"loss": 0.2028,
|
8052 |
+
"step": 22980
|
8053 |
+
},
|
8054 |
+
{
|
8055 |
+
"epoch": 112.26357535082367,
|
8056 |
+
"grad_norm": 2.1096384525299072,
|
8057 |
+
"learning_rate": 4.9836065573770496e-05,
|
8058 |
+
"loss": 0.2041,
|
8059 |
+
"step": 23000
|
8060 |
+
},
|
8061 |
+
{
|
8062 |
+
"epoch": 112.36119585112874,
|
8063 |
+
"grad_norm": 3.537277936935425,
|
8064 |
+
"learning_rate": 4.970491803278689e-05,
|
8065 |
+
"loss": 0.1988,
|
8066 |
+
"step": 23020
|
8067 |
+
},
|
8068 |
+
{
|
8069 |
+
"epoch": 112.4588163514338,
|
8070 |
+
"grad_norm": 4.58275842666626,
|
8071 |
+
"learning_rate": 4.957377049180328e-05,
|
8072 |
+
"loss": 0.1855,
|
8073 |
+
"step": 23040
|
8074 |
+
},
|
8075 |
+
{
|
8076 |
+
"epoch": 112.55643685173887,
|
8077 |
+
"grad_norm": 3.290548086166382,
|
8078 |
+
"learning_rate": 4.944262295081967e-05,
|
8079 |
+
"loss": 0.2041,
|
8080 |
+
"step": 23060
|
8081 |
+
},
|
8082 |
+
{
|
8083 |
+
"epoch": 112.65405735204394,
|
8084 |
+
"grad_norm": 4.11199951171875,
|
8085 |
+
"learning_rate": 4.931147540983607e-05,
|
8086 |
+
"loss": 0.2062,
|
8087 |
+
"step": 23080
|
8088 |
+
},
|
8089 |
+
{
|
8090 |
+
"epoch": 112.75167785234899,
|
8091 |
+
"grad_norm": 4.122567176818848,
|
8092 |
+
"learning_rate": 4.918032786885246e-05,
|
8093 |
+
"loss": 0.2003,
|
8094 |
+
"step": 23100
|
8095 |
+
},
|
8096 |
+
{
|
8097 |
+
"epoch": 112.84929835265406,
|
8098 |
+
"grad_norm": 2.775681495666504,
|
8099 |
+
"learning_rate": 4.904918032786885e-05,
|
8100 |
+
"loss": 0.2201,
|
8101 |
+
"step": 23120
|
8102 |
+
},
|
8103 |
+
{
|
8104 |
+
"epoch": 112.94691885295912,
|
8105 |
+
"grad_norm": 2.8952252864837646,
|
8106 |
+
"learning_rate": 4.891803278688525e-05,
|
8107 |
+
"loss": 0.2068,
|
8108 |
+
"step": 23140
|
8109 |
+
},
|
8110 |
+
{
|
8111 |
+
"epoch": 113.04453935326418,
|
8112 |
+
"grad_norm": 2.8353431224823,
|
8113 |
+
"learning_rate": 4.8786885245901645e-05,
|
8114 |
+
"loss": 0.2087,
|
8115 |
+
"step": 23160
|
8116 |
+
},
|
8117 |
+
{
|
8118 |
+
"epoch": 113.14215985356925,
|
8119 |
+
"grad_norm": 4.125601768493652,
|
8120 |
+
"learning_rate": 4.8655737704918034e-05,
|
8121 |
+
"loss": 0.1973,
|
8122 |
+
"step": 23180
|
8123 |
+
},
|
8124 |
+
{
|
8125 |
+
"epoch": 113.23978035387431,
|
8126 |
+
"grad_norm": 3.5090670585632324,
|
8127 |
+
"learning_rate": 4.852459016393443e-05,
|
8128 |
+
"loss": 0.1847,
|
8129 |
+
"step": 23200
|
8130 |
+
},
|
8131 |
+
{
|
8132 |
+
"epoch": 113.33740085417938,
|
8133 |
+
"grad_norm": 2.688890218734741,
|
8134 |
+
"learning_rate": 4.8393442622950826e-05,
|
8135 |
+
"loss": 0.1842,
|
8136 |
+
"step": 23220
|
8137 |
+
},
|
8138 |
+
{
|
8139 |
+
"epoch": 113.43502135448445,
|
8140 |
+
"grad_norm": 4.937670707702637,
|
8141 |
+
"learning_rate": 4.8262295081967216e-05,
|
8142 |
+
"loss": 0.214,
|
8143 |
+
"step": 23240
|
8144 |
+
},
|
8145 |
+
{
|
8146 |
+
"epoch": 113.5326418547895,
|
8147 |
+
"grad_norm": 4.070579528808594,
|
8148 |
+
"learning_rate": 4.8131147540983605e-05,
|
8149 |
+
"loss": 0.2203,
|
8150 |
+
"step": 23260
|
8151 |
+
},
|
8152 |
+
{
|
8153 |
+
"epoch": 113.63026235509457,
|
8154 |
+
"grad_norm": 3.052457809448242,
|
8155 |
+
"learning_rate": 4.8e-05,
|
8156 |
+
"loss": 0.2116,
|
8157 |
+
"step": 23280
|
8158 |
+
},
|
8159 |
+
{
|
8160 |
+
"epoch": 113.72788285539963,
|
8161 |
+
"grad_norm": 3.2423150539398193,
|
8162 |
+
"learning_rate": 4.78688524590164e-05,
|
8163 |
+
"loss": 0.192,
|
8164 |
+
"step": 23300
|
8165 |
+
},
|
8166 |
+
{
|
8167 |
+
"epoch": 113.8255033557047,
|
8168 |
+
"grad_norm": 2.9941837787628174,
|
8169 |
+
"learning_rate": 4.773770491803279e-05,
|
8170 |
+
"loss": 0.1916,
|
8171 |
+
"step": 23320
|
8172 |
+
},
|
8173 |
+
{
|
8174 |
+
"epoch": 113.92312385600977,
|
8175 |
+
"grad_norm": 3.3031139373779297,
|
8176 |
+
"learning_rate": 4.760655737704918e-05,
|
8177 |
+
"loss": 0.2261,
|
8178 |
+
"step": 23340
|
8179 |
+
},
|
8180 |
+
{
|
8181 |
+
"epoch": 114.02074435631482,
|
8182 |
+
"grad_norm": 3.531646490097046,
|
8183 |
+
"learning_rate": 4.747540983606558e-05,
|
8184 |
+
"loss": 0.2186,
|
8185 |
+
"step": 23360
|
8186 |
+
},
|
8187 |
+
{
|
8188 |
+
"epoch": 114.11836485661989,
|
8189 |
+
"grad_norm": 2.753261089324951,
|
8190 |
+
"learning_rate": 4.734426229508197e-05,
|
8191 |
+
"loss": 0.1979,
|
8192 |
+
"step": 23380
|
8193 |
+
},
|
8194 |
+
{
|
8195 |
+
"epoch": 114.21598535692496,
|
8196 |
+
"grad_norm": 3.2900078296661377,
|
8197 |
+
"learning_rate": 4.7213114754098365e-05,
|
8198 |
+
"loss": 0.1892,
|
8199 |
+
"step": 23400
|
8200 |
+
},
|
8201 |
+
{
|
8202 |
+
"epoch": 114.31360585723002,
|
8203 |
+
"grad_norm": 3.6558027267456055,
|
8204 |
+
"learning_rate": 4.708196721311476e-05,
|
8205 |
+
"loss": 0.2202,
|
8206 |
+
"step": 23420
|
8207 |
+
},
|
8208 |
+
{
|
8209 |
+
"epoch": 114.41122635753509,
|
8210 |
+
"grad_norm": 2.9026436805725098,
|
8211 |
+
"learning_rate": 4.695081967213115e-05,
|
8212 |
+
"loss": 0.2193,
|
8213 |
+
"step": 23440
|
8214 |
+
},
|
8215 |
+
{
|
8216 |
+
"epoch": 114.50884685784014,
|
8217 |
+
"grad_norm": 3.0958666801452637,
|
8218 |
+
"learning_rate": 4.681967213114754e-05,
|
8219 |
+
"loss": 0.1861,
|
8220 |
+
"step": 23460
|
8221 |
+
},
|
8222 |
+
{
|
8223 |
+
"epoch": 114.60646735814521,
|
8224 |
+
"grad_norm": 3.4162192344665527,
|
8225 |
+
"learning_rate": 4.6688524590163936e-05,
|
8226 |
+
"loss": 0.1813,
|
8227 |
+
"step": 23480
|
8228 |
+
},
|
8229 |
+
{
|
8230 |
+
"epoch": 114.70408785845028,
|
8231 |
+
"grad_norm": 3.4320363998413086,
|
8232 |
+
"learning_rate": 4.655737704918033e-05,
|
8233 |
+
"loss": 0.1849,
|
8234 |
+
"step": 23500
|
8235 |
+
},
|
8236 |
+
{
|
8237 |
+
"epoch": 114.80170835875533,
|
8238 |
+
"grad_norm": 3.931405544281006,
|
8239 |
+
"learning_rate": 4.642622950819672e-05,
|
8240 |
+
"loss": 0.2091,
|
8241 |
+
"step": 23520
|
8242 |
+
},
|
8243 |
+
{
|
8244 |
+
"epoch": 114.8993288590604,
|
8245 |
+
"grad_norm": 2.463747024536133,
|
8246 |
+
"learning_rate": 4.629508196721312e-05,
|
8247 |
+
"loss": 0.2013,
|
8248 |
+
"step": 23540
|
8249 |
+
},
|
8250 |
+
{
|
8251 |
+
"epoch": 114.99694935936547,
|
8252 |
+
"grad_norm": 3.4332265853881836,
|
8253 |
+
"learning_rate": 4.616393442622951e-05,
|
8254 |
+
"loss": 0.2224,
|
8255 |
+
"step": 23560
|
8256 |
+
},
|
8257 |
+
{
|
8258 |
+
"epoch": 115.09456985967053,
|
8259 |
+
"grad_norm": 4.516382217407227,
|
8260 |
+
"learning_rate": 4.60327868852459e-05,
|
8261 |
+
"loss": 0.191,
|
8262 |
+
"step": 23580
|
8263 |
+
},
|
8264 |
+
{
|
8265 |
+
"epoch": 115.1921903599756,
|
8266 |
+
"grad_norm": 3.7761762142181396,
|
8267 |
+
"learning_rate": 4.59016393442623e-05,
|
8268 |
+
"loss": 0.1972,
|
8269 |
+
"step": 23600
|
8270 |
+
},
|
8271 |
+
{
|
8272 |
+
"epoch": 115.28981086028065,
|
8273 |
+
"grad_norm": 3.425079584121704,
|
8274 |
+
"learning_rate": 4.5770491803278695e-05,
|
8275 |
+
"loss": 0.2053,
|
8276 |
+
"step": 23620
|
8277 |
+
},
|
8278 |
+
{
|
8279 |
+
"epoch": 115.38743136058572,
|
8280 |
+
"grad_norm": 3.098958969116211,
|
8281 |
+
"learning_rate": 4.5639344262295084e-05,
|
8282 |
+
"loss": 0.1994,
|
8283 |
+
"step": 23640
|
8284 |
+
},
|
8285 |
+
{
|
8286 |
+
"epoch": 115.48505186089079,
|
8287 |
+
"grad_norm": 4.242363452911377,
|
8288 |
+
"learning_rate": 4.550819672131148e-05,
|
8289 |
+
"loss": 0.2062,
|
8290 |
+
"step": 23660
|
8291 |
+
},
|
8292 |
+
{
|
8293 |
+
"epoch": 115.58267236119585,
|
8294 |
+
"grad_norm": 3.9641380310058594,
|
8295 |
+
"learning_rate": 4.537704918032787e-05,
|
8296 |
+
"loss": 0.1892,
|
8297 |
+
"step": 23680
|
8298 |
+
},
|
8299 |
+
{
|
8300 |
+
"epoch": 115.68029286150092,
|
8301 |
+
"grad_norm": 3.817915678024292,
|
8302 |
+
"learning_rate": 4.524590163934426e-05,
|
8303 |
+
"loss": 0.2076,
|
8304 |
+
"step": 23700
|
8305 |
+
},
|
8306 |
+
{
|
8307 |
+
"epoch": 115.77791336180599,
|
8308 |
+
"grad_norm": 4.9167680740356445,
|
8309 |
+
"learning_rate": 4.5114754098360655e-05,
|
8310 |
+
"loss": 0.1991,
|
8311 |
+
"step": 23720
|
8312 |
+
},
|
8313 |
+
{
|
8314 |
+
"epoch": 115.87553386211104,
|
8315 |
+
"grad_norm": 4.05020809173584,
|
8316 |
+
"learning_rate": 4.498360655737705e-05,
|
8317 |
+
"loss": 0.2154,
|
8318 |
+
"step": 23740
|
8319 |
+
},
|
8320 |
+
{
|
8321 |
+
"epoch": 115.97315436241611,
|
8322 |
+
"grad_norm": 5.213686466217041,
|
8323 |
+
"learning_rate": 4.485245901639345e-05,
|
8324 |
+
"loss": 0.1893,
|
8325 |
+
"step": 23760
|
8326 |
+
},
|
8327 |
+
{
|
8328 |
+
"epoch": 116.07077486272117,
|
8329 |
+
"grad_norm": 2.9788458347320557,
|
8330 |
+
"learning_rate": 4.472131147540984e-05,
|
8331 |
+
"loss": 0.2015,
|
8332 |
+
"step": 23780
|
8333 |
+
},
|
8334 |
+
{
|
8335 |
+
"epoch": 116.16839536302624,
|
8336 |
+
"grad_norm": 3.5112173557281494,
|
8337 |
+
"learning_rate": 4.459016393442623e-05,
|
8338 |
+
"loss": 0.2058,
|
8339 |
+
"step": 23800
|
8340 |
+
},
|
8341 |
+
{
|
8342 |
+
"epoch": 116.2660158633313,
|
8343 |
+
"grad_norm": 3.3048148155212402,
|
8344 |
+
"learning_rate": 4.445901639344263e-05,
|
8345 |
+
"loss": 0.2107,
|
8346 |
+
"step": 23820
|
8347 |
+
},
|
8348 |
+
{
|
8349 |
+
"epoch": 116.36363636363636,
|
8350 |
+
"grad_norm": 3.895615339279175,
|
8351 |
+
"learning_rate": 4.432786885245902e-05,
|
8352 |
+
"loss": 0.1744,
|
8353 |
+
"step": 23840
|
8354 |
+
},
|
8355 |
+
{
|
8356 |
+
"epoch": 116.46125686394143,
|
8357 |
+
"grad_norm": 3.7658941745758057,
|
8358 |
+
"learning_rate": 4.4196721311475415e-05,
|
8359 |
+
"loss": 0.1935,
|
8360 |
+
"step": 23860
|
8361 |
+
},
|
8362 |
+
{
|
8363 |
+
"epoch": 116.5588773642465,
|
8364 |
+
"grad_norm": 3.4818522930145264,
|
8365 |
+
"learning_rate": 4.406557377049181e-05,
|
8366 |
+
"loss": 0.2028,
|
8367 |
+
"step": 23880
|
8368 |
+
},
|
8369 |
+
{
|
8370 |
+
"epoch": 116.65649786455155,
|
8371 |
+
"grad_norm": 2.971785068511963,
|
8372 |
+
"learning_rate": 4.3934426229508194e-05,
|
8373 |
+
"loss": 0.1993,
|
8374 |
+
"step": 23900
|
8375 |
+
},
|
8376 |
+
{
|
8377 |
+
"epoch": 116.75411836485662,
|
8378 |
+
"grad_norm": 4.34529447555542,
|
8379 |
+
"learning_rate": 4.380327868852459e-05,
|
8380 |
+
"loss": 0.1887,
|
8381 |
+
"step": 23920
|
8382 |
+
},
|
8383 |
+
{
|
8384 |
+
"epoch": 116.85173886516168,
|
8385 |
+
"grad_norm": 3.1781437397003174,
|
8386 |
+
"learning_rate": 4.3672131147540986e-05,
|
8387 |
+
"loss": 0.2241,
|
8388 |
+
"step": 23940
|
8389 |
+
},
|
8390 |
+
{
|
8391 |
+
"epoch": 116.94935936546675,
|
8392 |
+
"grad_norm": 3.8372061252593994,
|
8393 |
+
"learning_rate": 4.3540983606557375e-05,
|
8394 |
+
"loss": 0.2103,
|
8395 |
+
"step": 23960
|
8396 |
+
},
|
8397 |
+
{
|
8398 |
+
"epoch": 117.04697986577182,
|
8399 |
+
"grad_norm": 3.981776237487793,
|
8400 |
+
"learning_rate": 4.340983606557377e-05,
|
8401 |
+
"loss": 0.1859,
|
8402 |
+
"step": 23980
|
8403 |
+
},
|
8404 |
+
{
|
8405 |
+
"epoch": 117.14460036607687,
|
8406 |
+
"grad_norm": 4.0308685302734375,
|
8407 |
+
"learning_rate": 4.327868852459017e-05,
|
8408 |
+
"loss": 0.1841,
|
8409 |
+
"step": 24000
|
8410 |
+
},
|
8411 |
+
{
|
8412 |
+
"epoch": 117.24222086638194,
|
8413 |
+
"grad_norm": 3.2919681072235107,
|
8414 |
+
"learning_rate": 4.3147540983606564e-05,
|
8415 |
+
"loss": 0.1964,
|
8416 |
+
"step": 24020
|
8417 |
+
},
|
8418 |
+
{
|
8419 |
+
"epoch": 117.33984136668701,
|
8420 |
+
"grad_norm": 4.155374050140381,
|
8421 |
+
"learning_rate": 4.301639344262295e-05,
|
8422 |
+
"loss": 0.1896,
|
8423 |
+
"step": 24040
|
8424 |
+
},
|
8425 |
+
{
|
8426 |
+
"epoch": 117.43746186699207,
|
8427 |
+
"grad_norm": 3.5654046535491943,
|
8428 |
+
"learning_rate": 4.288524590163935e-05,
|
8429 |
+
"loss": 0.1913,
|
8430 |
+
"step": 24060
|
8431 |
+
},
|
8432 |
+
{
|
8433 |
+
"epoch": 117.53508236729714,
|
8434 |
+
"grad_norm": 2.805769681930542,
|
8435 |
+
"learning_rate": 4.2754098360655745e-05,
|
8436 |
+
"loss": 0.1995,
|
8437 |
+
"step": 24080
|
8438 |
+
},
|
8439 |
+
{
|
8440 |
+
"epoch": 117.63270286760219,
|
8441 |
+
"grad_norm": 2.990482807159424,
|
8442 |
+
"learning_rate": 4.262295081967213e-05,
|
8443 |
+
"loss": 0.2034,
|
8444 |
+
"step": 24100
|
8445 |
+
},
|
8446 |
+
{
|
8447 |
+
"epoch": 117.73032336790726,
|
8448 |
+
"grad_norm": 3.236433506011963,
|
8449 |
+
"learning_rate": 4.2491803278688524e-05,
|
8450 |
+
"loss": 0.1948,
|
8451 |
+
"step": 24120
|
8452 |
+
},
|
8453 |
+
{
|
8454 |
+
"epoch": 117.82794386821233,
|
8455 |
+
"grad_norm": 4.235846996307373,
|
8456 |
+
"learning_rate": 4.236065573770492e-05,
|
8457 |
+
"loss": 0.2014,
|
8458 |
+
"step": 24140
|
8459 |
+
},
|
8460 |
+
{
|
8461 |
+
"epoch": 117.92556436851739,
|
8462 |
+
"grad_norm": 3.238041400909424,
|
8463 |
+
"learning_rate": 4.222950819672131e-05,
|
8464 |
+
"loss": 0.2106,
|
8465 |
+
"step": 24160
|
8466 |
+
},
|
8467 |
+
{
|
8468 |
+
"epoch": 118.02318486882245,
|
8469 |
+
"grad_norm": 3.18979811668396,
|
8470 |
+
"learning_rate": 4.2098360655737706e-05,
|
8471 |
+
"loss": 0.1918,
|
8472 |
+
"step": 24180
|
8473 |
+
},
|
8474 |
+
{
|
8475 |
+
"epoch": 118.12080536912751,
|
8476 |
+
"grad_norm": 2.699910879135132,
|
8477 |
+
"learning_rate": 4.19672131147541e-05,
|
8478 |
+
"loss": 0.2121,
|
8479 |
+
"step": 24200
|
8480 |
+
},
|
8481 |
+
{
|
8482 |
+
"epoch": 118.21842586943258,
|
8483 |
+
"grad_norm": 3.6096503734588623,
|
8484 |
+
"learning_rate": 4.183606557377049e-05,
|
8485 |
+
"loss": 0.2165,
|
8486 |
+
"step": 24220
|
8487 |
+
},
|
8488 |
+
{
|
8489 |
+
"epoch": 118.31604636973765,
|
8490 |
+
"grad_norm": 3.384746789932251,
|
8491 |
+
"learning_rate": 4.170491803278689e-05,
|
8492 |
+
"loss": 0.1884,
|
8493 |
+
"step": 24240
|
8494 |
+
},
|
8495 |
+
{
|
8496 |
+
"epoch": 118.4136668700427,
|
8497 |
+
"grad_norm": 3.606003761291504,
|
8498 |
+
"learning_rate": 4.1573770491803283e-05,
|
8499 |
+
"loss": 0.1932,
|
8500 |
+
"step": 24260
|
8501 |
+
},
|
8502 |
+
{
|
8503 |
+
"epoch": 118.51128737034777,
|
8504 |
+
"grad_norm": 2.9768433570861816,
|
8505 |
+
"learning_rate": 4.144262295081967e-05,
|
8506 |
+
"loss": 0.1802,
|
8507 |
+
"step": 24280
|
8508 |
+
},
|
8509 |
+
{
|
8510 |
+
"epoch": 118.60890787065284,
|
8511 |
+
"grad_norm": 3.784608840942383,
|
8512 |
+
"learning_rate": 4.131147540983607e-05,
|
8513 |
+
"loss": 0.1919,
|
8514 |
+
"step": 24300
|
8515 |
+
},
|
8516 |
+
{
|
8517 |
+
"epoch": 118.7065283709579,
|
8518 |
+
"grad_norm": 3.2883870601654053,
|
8519 |
+
"learning_rate": 4.118032786885246e-05,
|
8520 |
+
"loss": 0.2015,
|
8521 |
+
"step": 24320
|
8522 |
+
},
|
8523 |
+
{
|
8524 |
+
"epoch": 118.80414887126297,
|
8525 |
+
"grad_norm": 4.078542232513428,
|
8526 |
+
"learning_rate": 4.1049180327868854e-05,
|
8527 |
+
"loss": 0.1994,
|
8528 |
+
"step": 24340
|
8529 |
+
},
|
8530 |
+
{
|
8531 |
+
"epoch": 118.90176937156802,
|
8532 |
+
"grad_norm": 3.7962749004364014,
|
8533 |
+
"learning_rate": 4.0918032786885244e-05,
|
8534 |
+
"loss": 0.1958,
|
8535 |
+
"step": 24360
|
8536 |
+
},
|
8537 |
+
{
|
8538 |
+
"epoch": 118.99938987187309,
|
8539 |
+
"grad_norm": 3.4484541416168213,
|
8540 |
+
"learning_rate": 4.078688524590164e-05,
|
8541 |
+
"loss": 0.1926,
|
8542 |
+
"step": 24380
|
8543 |
+
},
|
8544 |
+
{
|
8545 |
+
"epoch": 119.09701037217816,
|
8546 |
+
"grad_norm": 2.9271483421325684,
|
8547 |
+
"learning_rate": 4.0655737704918036e-05,
|
8548 |
+
"loss": 0.1818,
|
8549 |
+
"step": 24400
|
8550 |
+
},
|
8551 |
+
{
|
8552 |
+
"epoch": 119.19463087248322,
|
8553 |
+
"grad_norm": 5.304340839385986,
|
8554 |
+
"learning_rate": 4.0524590163934425e-05,
|
8555 |
+
"loss": 0.192,
|
8556 |
+
"step": 24420
|
8557 |
+
},
|
8558 |
+
{
|
8559 |
+
"epoch": 119.29225137278829,
|
8560 |
+
"grad_norm": 4.88720178604126,
|
8561 |
+
"learning_rate": 4.039344262295082e-05,
|
8562 |
+
"loss": 0.1975,
|
8563 |
+
"step": 24440
|
8564 |
+
},
|
8565 |
+
{
|
8566 |
+
"epoch": 119.38987187309336,
|
8567 |
+
"grad_norm": 3.7572522163391113,
|
8568 |
+
"learning_rate": 4.026229508196722e-05,
|
8569 |
+
"loss": 0.1856,
|
8570 |
+
"step": 24460
|
8571 |
+
},
|
8572 |
+
{
|
8573 |
+
"epoch": 119.48749237339841,
|
8574 |
+
"grad_norm": 3.3659627437591553,
|
8575 |
+
"learning_rate": 4.013114754098361e-05,
|
8576 |
+
"loss": 0.1961,
|
8577 |
+
"step": 24480
|
8578 |
+
},
|
8579 |
+
{
|
8580 |
+
"epoch": 119.58511287370348,
|
8581 |
+
"grad_norm": 3.1397533416748047,
|
8582 |
+
"learning_rate": 4e-05,
|
8583 |
+
"loss": 0.1991,
|
8584 |
+
"step": 24500
|
8585 |
+
},
|
8586 |
+
{
|
8587 |
+
"epoch": 119.68273337400854,
|
8588 |
+
"grad_norm": 3.9070587158203125,
|
8589 |
+
"learning_rate": 3.98688524590164e-05,
|
8590 |
+
"loss": 0.1946,
|
8591 |
+
"step": 24520
|
8592 |
+
},
|
8593 |
+
{
|
8594 |
+
"epoch": 119.7803538743136,
|
8595 |
+
"grad_norm": 3.3265631198883057,
|
8596 |
+
"learning_rate": 3.973770491803279e-05,
|
8597 |
+
"loss": 0.194,
|
8598 |
+
"step": 24540
|
8599 |
+
},
|
8600 |
+
{
|
8601 |
+
"epoch": 119.87797437461867,
|
8602 |
+
"grad_norm": 3.5021114349365234,
|
8603 |
+
"learning_rate": 3.960655737704918e-05,
|
8604 |
+
"loss": 0.2084,
|
8605 |
+
"step": 24560
|
8606 |
+
},
|
8607 |
+
{
|
8608 |
+
"epoch": 119.97559487492373,
|
8609 |
+
"grad_norm": 3.7133572101593018,
|
8610 |
+
"learning_rate": 3.9475409836065574e-05,
|
8611 |
+
"loss": 0.2102,
|
8612 |
+
"step": 24580
|
8613 |
+
},
|
8614 |
+
{
|
8615 |
+
"epoch": 120.0732153752288,
|
8616 |
+
"grad_norm": 3.27691912651062,
|
8617 |
+
"learning_rate": 3.934426229508197e-05,
|
8618 |
+
"loss": 0.1864,
|
8619 |
+
"step": 24600
|
8620 |
+
},
|
8621 |
+
{
|
8622 |
+
"epoch": 120.17083587553387,
|
8623 |
+
"grad_norm": 4.74030065536499,
|
8624 |
+
"learning_rate": 3.921311475409836e-05,
|
8625 |
+
"loss": 0.1967,
|
8626 |
+
"step": 24620
|
8627 |
+
},
|
8628 |
+
{
|
8629 |
+
"epoch": 120.26845637583892,
|
8630 |
+
"grad_norm": 4.080429553985596,
|
8631 |
+
"learning_rate": 3.9081967213114756e-05,
|
8632 |
+
"loss": 0.1888,
|
8633 |
+
"step": 24640
|
8634 |
+
},
|
8635 |
+
{
|
8636 |
+
"epoch": 120.36607687614399,
|
8637 |
+
"grad_norm": 3.5443646907806396,
|
8638 |
+
"learning_rate": 3.895081967213115e-05,
|
8639 |
+
"loss": 0.199,
|
8640 |
+
"step": 24660
|
8641 |
+
},
|
8642 |
+
{
|
8643 |
+
"epoch": 120.46369737644905,
|
8644 |
+
"grad_norm": 4.0403923988342285,
|
8645 |
+
"learning_rate": 3.881967213114754e-05,
|
8646 |
+
"loss": 0.2002,
|
8647 |
+
"step": 24680
|
8648 |
+
},
|
8649 |
+
{
|
8650 |
+
"epoch": 120.56131787675412,
|
8651 |
+
"grad_norm": 4.144064426422119,
|
8652 |
+
"learning_rate": 3.868852459016394e-05,
|
8653 |
+
"loss": 0.193,
|
8654 |
+
"step": 24700
|
8655 |
+
},
|
8656 |
+
{
|
8657 |
+
"epoch": 120.65893837705919,
|
8658 |
+
"grad_norm": 5.478573799133301,
|
8659 |
+
"learning_rate": 3.8557377049180334e-05,
|
8660 |
+
"loss": 0.19,
|
8661 |
+
"step": 24720
|
8662 |
+
},
|
8663 |
+
{
|
8664 |
+
"epoch": 120.75655887736424,
|
8665 |
+
"grad_norm": 3.799931287765503,
|
8666 |
+
"learning_rate": 3.842622950819672e-05,
|
8667 |
+
"loss": 0.1876,
|
8668 |
+
"step": 24740
|
8669 |
+
},
|
8670 |
+
{
|
8671 |
+
"epoch": 120.85417937766931,
|
8672 |
+
"grad_norm": 3.1754050254821777,
|
8673 |
+
"learning_rate": 3.829508196721311e-05,
|
8674 |
+
"loss": 0.1881,
|
8675 |
+
"step": 24760
|
8676 |
+
},
|
8677 |
+
{
|
8678 |
+
"epoch": 120.95179987797438,
|
8679 |
+
"grad_norm": 3.4417614936828613,
|
8680 |
+
"learning_rate": 3.816393442622951e-05,
|
8681 |
+
"loss": 0.2023,
|
8682 |
+
"step": 24780
|
8683 |
+
},
|
8684 |
+
{
|
8685 |
+
"epoch": 121.04942037827944,
|
8686 |
+
"grad_norm": 2.689237594604492,
|
8687 |
+
"learning_rate": 3.8032786885245905e-05,
|
8688 |
+
"loss": 0.1918,
|
8689 |
+
"step": 24800
|
8690 |
+
},
|
8691 |
+
{
|
8692 |
+
"epoch": 121.1470408785845,
|
8693 |
+
"grad_norm": 2.6631858348846436,
|
8694 |
+
"learning_rate": 3.7901639344262294e-05,
|
8695 |
+
"loss": 0.1681,
|
8696 |
+
"step": 24820
|
8697 |
+
},
|
8698 |
+
{
|
8699 |
+
"epoch": 121.24466137888956,
|
8700 |
+
"grad_norm": 4.315155982971191,
|
8701 |
+
"learning_rate": 3.777049180327869e-05,
|
8702 |
+
"loss": 0.2037,
|
8703 |
+
"step": 24840
|
8704 |
+
},
|
8705 |
+
{
|
8706 |
+
"epoch": 121.34228187919463,
|
8707 |
+
"grad_norm": 3.0183377265930176,
|
8708 |
+
"learning_rate": 3.7639344262295086e-05,
|
8709 |
+
"loss": 0.1901,
|
8710 |
+
"step": 24860
|
8711 |
+
},
|
8712 |
+
{
|
8713 |
+
"epoch": 121.4399023794997,
|
8714 |
+
"grad_norm": 2.389106035232544,
|
8715 |
+
"learning_rate": 3.7508196721311476e-05,
|
8716 |
+
"loss": 0.2162,
|
8717 |
+
"step": 24880
|
8718 |
+
},
|
8719 |
+
{
|
8720 |
+
"epoch": 121.53752287980475,
|
8721 |
+
"grad_norm": 3.5406622886657715,
|
8722 |
+
"learning_rate": 3.737704918032787e-05,
|
8723 |
+
"loss": 0.1887,
|
8724 |
+
"step": 24900
|
8725 |
+
},
|
8726 |
+
{
|
8727 |
+
"epoch": 121.63514338010982,
|
8728 |
+
"grad_norm": 3.272151231765747,
|
8729 |
+
"learning_rate": 3.724590163934427e-05,
|
8730 |
+
"loss": 0.2156,
|
8731 |
+
"step": 24920
|
8732 |
+
},
|
8733 |
+
{
|
8734 |
+
"epoch": 121.7327638804149,
|
8735 |
+
"grad_norm": 3.1407511234283447,
|
8736 |
+
"learning_rate": 3.711475409836066e-05,
|
8737 |
+
"loss": 0.1833,
|
8738 |
+
"step": 24940
|
8739 |
+
},
|
8740 |
+
{
|
8741 |
+
"epoch": 121.83038438071995,
|
8742 |
+
"grad_norm": 3.3532192707061768,
|
8743 |
+
"learning_rate": 3.698360655737705e-05,
|
8744 |
+
"loss": 0.1858,
|
8745 |
+
"step": 24960
|
8746 |
+
},
|
8747 |
+
{
|
8748 |
+
"epoch": 121.92800488102502,
|
8749 |
+
"grad_norm": 3.1784684658050537,
|
8750 |
+
"learning_rate": 3.685245901639344e-05,
|
8751 |
+
"loss": 0.2,
|
8752 |
+
"step": 24980
|
8753 |
+
},
|
8754 |
+
{
|
8755 |
+
"epoch": 122.02562538133007,
|
8756 |
+
"grad_norm": 3.5837671756744385,
|
8757 |
+
"learning_rate": 3.672131147540984e-05,
|
8758 |
+
"loss": 0.2073,
|
8759 |
+
"step": 25000
|
8760 |
+
},
|
8761 |
+
{
|
8762 |
+
"epoch": 122.12324588163514,
|
8763 |
+
"grad_norm": 3.2040927410125732,
|
8764 |
+
"learning_rate": 3.659016393442623e-05,
|
8765 |
+
"loss": 0.1712,
|
8766 |
+
"step": 25020
|
8767 |
+
},
|
8768 |
+
{
|
8769 |
+
"epoch": 122.22086638194021,
|
8770 |
+
"grad_norm": 3.3119306564331055,
|
8771 |
+
"learning_rate": 3.6459016393442625e-05,
|
8772 |
+
"loss": 0.1879,
|
8773 |
+
"step": 25040
|
8774 |
+
},
|
8775 |
+
{
|
8776 |
+
"epoch": 122.31848688224527,
|
8777 |
+
"grad_norm": 3.2154297828674316,
|
8778 |
+
"learning_rate": 3.632786885245902e-05,
|
8779 |
+
"loss": 0.1976,
|
8780 |
+
"step": 25060
|
8781 |
+
},
|
8782 |
+
{
|
8783 |
+
"epoch": 122.41610738255034,
|
8784 |
+
"grad_norm": 2.9070613384246826,
|
8785 |
+
"learning_rate": 3.619672131147541e-05,
|
8786 |
+
"loss": 0.2077,
|
8787 |
+
"step": 25080
|
8788 |
+
},
|
8789 |
+
{
|
8790 |
+
"epoch": 122.5137278828554,
|
8791 |
+
"grad_norm": 3.866682529449463,
|
8792 |
+
"learning_rate": 3.6065573770491806e-05,
|
8793 |
+
"loss": 0.1673,
|
8794 |
+
"step": 25100
|
8795 |
+
},
|
8796 |
+
{
|
8797 |
+
"epoch": 122.61134838316046,
|
8798 |
+
"grad_norm": 4.236519813537598,
|
8799 |
+
"learning_rate": 3.59344262295082e-05,
|
8800 |
+
"loss": 0.2096,
|
8801 |
+
"step": 25120
|
8802 |
+
},
|
8803 |
+
{
|
8804 |
+
"epoch": 122.70896888346553,
|
8805 |
+
"grad_norm": 3.386503219604492,
|
8806 |
+
"learning_rate": 3.580327868852459e-05,
|
8807 |
+
"loss": 0.1962,
|
8808 |
+
"step": 25140
|
8809 |
+
},
|
8810 |
+
{
|
8811 |
+
"epoch": 122.80658938377059,
|
8812 |
+
"grad_norm": 2.7001540660858154,
|
8813 |
+
"learning_rate": 3.567213114754099e-05,
|
8814 |
+
"loss": 0.1731,
|
8815 |
+
"step": 25160
|
8816 |
+
},
|
8817 |
+
{
|
8818 |
+
"epoch": 122.90420988407566,
|
8819 |
+
"grad_norm": 3.281818389892578,
|
8820 |
+
"learning_rate": 3.554098360655738e-05,
|
8821 |
+
"loss": 0.2225,
|
8822 |
+
"step": 25180
|
8823 |
+
},
|
8824 |
+
{
|
8825 |
+
"epoch": 123.00183038438072,
|
8826 |
+
"grad_norm": 4.1742095947265625,
|
8827 |
+
"learning_rate": 3.5409836065573773e-05,
|
8828 |
+
"loss": 0.1947,
|
8829 |
+
"step": 25200
|
8830 |
+
},
|
8831 |
+
{
|
8832 |
+
"epoch": 123.09945088468578,
|
8833 |
+
"grad_norm": 3.328521490097046,
|
8834 |
+
"learning_rate": 3.527868852459016e-05,
|
8835 |
+
"loss": 0.1607,
|
8836 |
+
"step": 25220
|
8837 |
+
},
|
8838 |
+
{
|
8839 |
+
"epoch": 123.19707138499085,
|
8840 |
+
"grad_norm": 4.815507888793945,
|
8841 |
+
"learning_rate": 3.514754098360656e-05,
|
8842 |
+
"loss": 0.1977,
|
8843 |
+
"step": 25240
|
8844 |
+
},
|
8845 |
+
{
|
8846 |
+
"epoch": 123.29469188529592,
|
8847 |
+
"grad_norm": 3.736438512802124,
|
8848 |
+
"learning_rate": 3.5016393442622955e-05,
|
8849 |
+
"loss": 0.2105,
|
8850 |
+
"step": 25260
|
8851 |
+
},
|
8852 |
+
{
|
8853 |
+
"epoch": 123.39231238560097,
|
8854 |
+
"grad_norm": 3.4552454948425293,
|
8855 |
+
"learning_rate": 3.4885245901639344e-05,
|
8856 |
+
"loss": 0.2002,
|
8857 |
+
"step": 25280
|
8858 |
+
},
|
8859 |
+
{
|
8860 |
+
"epoch": 123.48993288590604,
|
8861 |
+
"grad_norm": 2.3649439811706543,
|
8862 |
+
"learning_rate": 3.475409836065574e-05,
|
8863 |
+
"loss": 0.1823,
|
8864 |
+
"step": 25300
|
8865 |
+
},
|
8866 |
+
{
|
8867 |
+
"epoch": 123.5875533862111,
|
8868 |
+
"grad_norm": 3.0043177604675293,
|
8869 |
+
"learning_rate": 3.462295081967214e-05,
|
8870 |
+
"loss": 0.1922,
|
8871 |
+
"step": 25320
|
8872 |
+
},
|
8873 |
+
{
|
8874 |
+
"epoch": 123.68517388651617,
|
8875 |
+
"grad_norm": 4.2065253257751465,
|
8876 |
+
"learning_rate": 3.4491803278688526e-05,
|
8877 |
+
"loss": 0.1943,
|
8878 |
+
"step": 25340
|
8879 |
+
},
|
8880 |
+
{
|
8881 |
+
"epoch": 123.78279438682124,
|
8882 |
+
"grad_norm": 3.961331605911255,
|
8883 |
+
"learning_rate": 3.436065573770492e-05,
|
8884 |
+
"loss": 0.1859,
|
8885 |
+
"step": 25360
|
8886 |
+
},
|
8887 |
+
{
|
8888 |
+
"epoch": 123.88041488712629,
|
8889 |
+
"grad_norm": 4.434045791625977,
|
8890 |
+
"learning_rate": 3.422950819672131e-05,
|
8891 |
+
"loss": 0.1985,
|
8892 |
+
"step": 25380
|
8893 |
+
},
|
8894 |
+
{
|
8895 |
+
"epoch": 123.97803538743136,
|
8896 |
+
"grad_norm": 2.87196946144104,
|
8897 |
+
"learning_rate": 3.409836065573771e-05,
|
8898 |
+
"loss": 0.2046,
|
8899 |
+
"step": 25400
|
8900 |
+
},
|
8901 |
+
{
|
8902 |
+
"epoch": 124.07565588773643,
|
8903 |
+
"grad_norm": 2.9702465534210205,
|
8904 |
+
"learning_rate": 3.39672131147541e-05,
|
8905 |
+
"loss": 0.1665,
|
8906 |
+
"step": 25420
|
8907 |
+
},
|
8908 |
+
{
|
8909 |
+
"epoch": 124.17327638804149,
|
8910 |
+
"grad_norm": 4.106092929840088,
|
8911 |
+
"learning_rate": 3.383606557377049e-05,
|
8912 |
+
"loss": 0.1966,
|
8913 |
+
"step": 25440
|
8914 |
+
},
|
8915 |
+
{
|
8916 |
+
"epoch": 124.27089688834656,
|
8917 |
+
"grad_norm": 3.1091065406799316,
|
8918 |
+
"learning_rate": 3.370491803278689e-05,
|
8919 |
+
"loss": 0.2078,
|
8920 |
+
"step": 25460
|
8921 |
+
},
|
8922 |
+
{
|
8923 |
+
"epoch": 124.36851738865161,
|
8924 |
+
"grad_norm": 3.4036500453948975,
|
8925 |
+
"learning_rate": 3.357377049180328e-05,
|
8926 |
+
"loss": 0.1814,
|
8927 |
+
"step": 25480
|
8928 |
+
},
|
8929 |
+
{
|
8930 |
+
"epoch": 124.46613788895668,
|
8931 |
+
"grad_norm": 3.7814347743988037,
|
8932 |
+
"learning_rate": 3.3442622950819675e-05,
|
8933 |
+
"loss": 0.195,
|
8934 |
+
"step": 25500
|
8935 |
+
},
|
8936 |
+
{
|
8937 |
+
"epoch": 124.56375838926175,
|
8938 |
+
"grad_norm": 4.475528240203857,
|
8939 |
+
"learning_rate": 3.331147540983607e-05,
|
8940 |
+
"loss": 0.1984,
|
8941 |
+
"step": 25520
|
8942 |
+
},
|
8943 |
+
{
|
8944 |
+
"epoch": 124.6613788895668,
|
8945 |
+
"grad_norm": 3.35787296295166,
|
8946 |
+
"learning_rate": 3.318032786885246e-05,
|
8947 |
+
"loss": 0.1809,
|
8948 |
+
"step": 25540
|
8949 |
+
},
|
8950 |
+
{
|
8951 |
+
"epoch": 124.75899938987187,
|
8952 |
+
"grad_norm": 3.594639301300049,
|
8953 |
+
"learning_rate": 3.3049180327868857e-05,
|
8954 |
+
"loss": 0.1844,
|
8955 |
+
"step": 25560
|
8956 |
+
},
|
8957 |
+
{
|
8958 |
+
"epoch": 124.85661989017694,
|
8959 |
+
"grad_norm": 3.5354974269866943,
|
8960 |
+
"learning_rate": 3.291803278688525e-05,
|
8961 |
+
"loss": 0.1953,
|
8962 |
+
"step": 25580
|
8963 |
+
},
|
8964 |
+
{
|
8965 |
+
"epoch": 124.954240390482,
|
8966 |
+
"grad_norm": 2.810798168182373,
|
8967 |
+
"learning_rate": 3.2786885245901635e-05,
|
8968 |
+
"loss": 0.1911,
|
8969 |
+
"step": 25600
|
8970 |
+
},
|
8971 |
+
{
|
8972 |
+
"epoch": 125.05186089078707,
|
8973 |
+
"grad_norm": 3.3332059383392334,
|
8974 |
+
"learning_rate": 3.265573770491803e-05,
|
8975 |
+
"loss": 0.1593,
|
8976 |
+
"step": 25620
|
8977 |
+
},
|
8978 |
+
{
|
8979 |
+
"epoch": 125.14948139109212,
|
8980 |
+
"grad_norm": 2.8333628177642822,
|
8981 |
+
"learning_rate": 3.252459016393443e-05,
|
8982 |
+
"loss": 0.1709,
|
8983 |
+
"step": 25640
|
8984 |
+
},
|
8985 |
+
{
|
8986 |
+
"epoch": 125.2471018913972,
|
8987 |
+
"grad_norm": 2.9144675731658936,
|
8988 |
+
"learning_rate": 3.2393442622950824e-05,
|
8989 |
+
"loss": 0.2073,
|
8990 |
+
"step": 25660
|
8991 |
+
},
|
8992 |
+
{
|
8993 |
+
"epoch": 125.34472239170226,
|
8994 |
+
"grad_norm": 3.185001850128174,
|
8995 |
+
"learning_rate": 3.226229508196721e-05,
|
8996 |
+
"loss": 0.1665,
|
8997 |
+
"step": 25680
|
8998 |
+
},
|
8999 |
+
{
|
9000 |
+
"epoch": 125.44234289200732,
|
9001 |
+
"grad_norm": 2.8228659629821777,
|
9002 |
+
"learning_rate": 3.213114754098361e-05,
|
9003 |
+
"loss": 0.163,
|
9004 |
+
"step": 25700
|
9005 |
+
},
|
9006 |
+
{
|
9007 |
+
"epoch": 125.53996339231239,
|
9008 |
+
"grad_norm": 2.8857967853546143,
|
9009 |
+
"learning_rate": 3.2000000000000005e-05,
|
9010 |
+
"loss": 0.2057,
|
9011 |
+
"step": 25720
|
9012 |
+
},
|
9013 |
+
{
|
9014 |
+
"epoch": 125.63758389261746,
|
9015 |
+
"grad_norm": 2.9855597019195557,
|
9016 |
+
"learning_rate": 3.1868852459016395e-05,
|
9017 |
+
"loss": 0.1872,
|
9018 |
+
"step": 25740
|
9019 |
+
},
|
9020 |
+
{
|
9021 |
+
"epoch": 125.73520439292251,
|
9022 |
+
"grad_norm": 3.5702884197235107,
|
9023 |
+
"learning_rate": 3.173770491803279e-05,
|
9024 |
+
"loss": 0.1992,
|
9025 |
+
"step": 25760
|
9026 |
+
},
|
9027 |
+
{
|
9028 |
+
"epoch": 125.83282489322758,
|
9029 |
+
"grad_norm": 5.302943706512451,
|
9030 |
+
"learning_rate": 3.160655737704919e-05,
|
9031 |
+
"loss": 0.1995,
|
9032 |
+
"step": 25780
|
9033 |
+
},
|
9034 |
+
{
|
9035 |
+
"epoch": 125.93044539353264,
|
9036 |
+
"grad_norm": 3.9966931343078613,
|
9037 |
+
"learning_rate": 3.1475409836065576e-05,
|
9038 |
+
"loss": 0.2046,
|
9039 |
+
"step": 25800
|
9040 |
+
},
|
9041 |
+
{
|
9042 |
+
"epoch": 126.0280658938377,
|
9043 |
+
"grad_norm": 2.6985690593719482,
|
9044 |
+
"learning_rate": 3.1344262295081966e-05,
|
9045 |
+
"loss": 0.2084,
|
9046 |
+
"step": 25820
|
9047 |
+
},
|
9048 |
+
{
|
9049 |
+
"epoch": 126.12568639414278,
|
9050 |
+
"grad_norm": 3.317439079284668,
|
9051 |
+
"learning_rate": 3.121311475409836e-05,
|
9052 |
+
"loss": 0.1971,
|
9053 |
+
"step": 25840
|
9054 |
+
},
|
9055 |
+
{
|
9056 |
+
"epoch": 126.22330689444783,
|
9057 |
+
"grad_norm": 3.718867301940918,
|
9058 |
+
"learning_rate": 3.108196721311475e-05,
|
9059 |
+
"loss": 0.1759,
|
9060 |
+
"step": 25860
|
9061 |
+
},
|
9062 |
+
{
|
9063 |
+
"epoch": 126.3209273947529,
|
9064 |
+
"grad_norm": 3.7418856620788574,
|
9065 |
+
"learning_rate": 3.095081967213115e-05,
|
9066 |
+
"loss": 0.2067,
|
9067 |
+
"step": 25880
|
9068 |
+
},
|
9069 |
+
{
|
9070 |
+
"epoch": 126.41854789505797,
|
9071 |
+
"grad_norm": 3.9287869930267334,
|
9072 |
+
"learning_rate": 3.0819672131147544e-05,
|
9073 |
+
"loss": 0.2032,
|
9074 |
+
"step": 25900
|
9075 |
+
},
|
9076 |
+
{
|
9077 |
+
"epoch": 126.51616839536302,
|
9078 |
+
"grad_norm": 1.9036474227905273,
|
9079 |
+
"learning_rate": 3.068852459016393e-05,
|
9080 |
+
"loss": 0.1905,
|
9081 |
+
"step": 25920
|
9082 |
+
},
|
9083 |
+
{
|
9084 |
+
"epoch": 126.6137888956681,
|
9085 |
+
"grad_norm": 2.602092981338501,
|
9086 |
+
"learning_rate": 3.055737704918033e-05,
|
9087 |
+
"loss": 0.1851,
|
9088 |
+
"step": 25940
|
9089 |
+
},
|
9090 |
+
{
|
9091 |
+
"epoch": 126.71140939597315,
|
9092 |
+
"grad_norm": 3.2280685901641846,
|
9093 |
+
"learning_rate": 3.0426229508196725e-05,
|
9094 |
+
"loss": 0.1938,
|
9095 |
+
"step": 25960
|
9096 |
+
},
|
9097 |
+
{
|
9098 |
+
"epoch": 126.80902989627822,
|
9099 |
+
"grad_norm": 4.606971263885498,
|
9100 |
+
"learning_rate": 3.0295081967213118e-05,
|
9101 |
+
"loss": 0.1644,
|
9102 |
+
"step": 25980
|
9103 |
+
},
|
9104 |
+
{
|
9105 |
+
"epoch": 126.90665039658329,
|
9106 |
+
"grad_norm": 3.755833864212036,
|
9107 |
+
"learning_rate": 3.016393442622951e-05,
|
9108 |
+
"loss": 0.1836,
|
9109 |
+
"step": 26000
|
9110 |
+
},
|
9111 |
+
{
|
9112 |
+
"epoch": 127.00427089688834,
|
9113 |
+
"grad_norm": 3.572577953338623,
|
9114 |
+
"learning_rate": 3.00327868852459e-05,
|
9115 |
+
"loss": 0.1733,
|
9116 |
+
"step": 26020
|
9117 |
+
},
|
9118 |
+
{
|
9119 |
+
"epoch": 127.10189139719341,
|
9120 |
+
"grad_norm": 3.1435470581054688,
|
9121 |
+
"learning_rate": 2.9901639344262293e-05,
|
9122 |
+
"loss": 0.1815,
|
9123 |
+
"step": 26040
|
9124 |
+
},
|
9125 |
+
{
|
9126 |
+
"epoch": 127.19951189749847,
|
9127 |
+
"grad_norm": 3.3165206909179688,
|
9128 |
+
"learning_rate": 2.977049180327869e-05,
|
9129 |
+
"loss": 0.183,
|
9130 |
+
"step": 26060
|
9131 |
+
},
|
9132 |
+
{
|
9133 |
+
"epoch": 127.29713239780354,
|
9134 |
+
"grad_norm": 3.269935131072998,
|
9135 |
+
"learning_rate": 2.963934426229508e-05,
|
9136 |
+
"loss": 0.1854,
|
9137 |
+
"step": 26080
|
9138 |
+
},
|
9139 |
+
{
|
9140 |
+
"epoch": 127.3947528981086,
|
9141 |
+
"grad_norm": 3.6275577545166016,
|
9142 |
+
"learning_rate": 2.9508196721311478e-05,
|
9143 |
+
"loss": 0.1798,
|
9144 |
+
"step": 26100
|
9145 |
+
},
|
9146 |
+
{
|
9147 |
+
"epoch": 127.49237339841366,
|
9148 |
+
"grad_norm": 3.3832483291625977,
|
9149 |
+
"learning_rate": 2.937704918032787e-05,
|
9150 |
+
"loss": 0.1722,
|
9151 |
+
"step": 26120
|
9152 |
+
},
|
9153 |
+
{
|
9154 |
+
"epoch": 127.58999389871873,
|
9155 |
+
"grad_norm": 3.828364610671997,
|
9156 |
+
"learning_rate": 2.9245901639344263e-05,
|
9157 |
+
"loss": 0.1853,
|
9158 |
+
"step": 26140
|
9159 |
+
},
|
9160 |
+
{
|
9161 |
+
"epoch": 127.6876143990238,
|
9162 |
+
"grad_norm": 3.7207860946655273,
|
9163 |
+
"learning_rate": 2.911475409836066e-05,
|
9164 |
+
"loss": 0.1948,
|
9165 |
+
"step": 26160
|
9166 |
+
},
|
9167 |
+
{
|
9168 |
+
"epoch": 127.78523489932886,
|
9169 |
+
"grad_norm": 3.281031847000122,
|
9170 |
+
"learning_rate": 2.8983606557377052e-05,
|
9171 |
+
"loss": 0.2063,
|
9172 |
+
"step": 26180
|
9173 |
+
},
|
9174 |
+
{
|
9175 |
+
"epoch": 127.88285539963393,
|
9176 |
+
"grad_norm": 3.1116421222686768,
|
9177 |
+
"learning_rate": 2.8852459016393445e-05,
|
9178 |
+
"loss": 0.191,
|
9179 |
+
"step": 26200
|
9180 |
+
},
|
9181 |
+
{
|
9182 |
+
"epoch": 127.98047589993898,
|
9183 |
+
"grad_norm": 4.254022121429443,
|
9184 |
+
"learning_rate": 2.872131147540984e-05,
|
9185 |
+
"loss": 0.202,
|
9186 |
+
"step": 26220
|
9187 |
+
},
|
9188 |
+
{
|
9189 |
+
"epoch": 128.07809640024405,
|
9190 |
+
"grad_norm": 3.1279819011688232,
|
9191 |
+
"learning_rate": 2.8590163934426227e-05,
|
9192 |
+
"loss": 0.1934,
|
9193 |
+
"step": 26240
|
9194 |
+
},
|
9195 |
+
{
|
9196 |
+
"epoch": 128.1757169005491,
|
9197 |
+
"grad_norm": 3.3977596759796143,
|
9198 |
+
"learning_rate": 2.8459016393442623e-05,
|
9199 |
+
"loss": 0.1807,
|
9200 |
+
"step": 26260
|
9201 |
+
},
|
9202 |
+
{
|
9203 |
+
"epoch": 128.2733374008542,
|
9204 |
+
"grad_norm": 5.261218070983887,
|
9205 |
+
"learning_rate": 2.8327868852459016e-05,
|
9206 |
+
"loss": 0.17,
|
9207 |
+
"step": 26280
|
9208 |
+
},
|
9209 |
+
{
|
9210 |
+
"epoch": 128.37095790115924,
|
9211 |
+
"grad_norm": 4.153654098510742,
|
9212 |
+
"learning_rate": 2.819672131147541e-05,
|
9213 |
+
"loss": 0.1891,
|
9214 |
+
"step": 26300
|
9215 |
+
},
|
9216 |
+
{
|
9217 |
+
"epoch": 128.4685784014643,
|
9218 |
+
"grad_norm": 3.449397563934326,
|
9219 |
+
"learning_rate": 2.8065573770491805e-05,
|
9220 |
+
"loss": 0.1846,
|
9221 |
+
"step": 26320
|
9222 |
+
},
|
9223 |
+
{
|
9224 |
+
"epoch": 128.56619890176938,
|
9225 |
+
"grad_norm": 3.43442702293396,
|
9226 |
+
"learning_rate": 2.7934426229508198e-05,
|
9227 |
+
"loss": 0.1807,
|
9228 |
+
"step": 26340
|
9229 |
+
},
|
9230 |
+
{
|
9231 |
+
"epoch": 128.66381940207444,
|
9232 |
+
"grad_norm": 2.9243948459625244,
|
9233 |
+
"learning_rate": 2.7803278688524594e-05,
|
9234 |
+
"loss": 0.1704,
|
9235 |
+
"step": 26360
|
9236 |
+
},
|
9237 |
+
{
|
9238 |
+
"epoch": 128.7614399023795,
|
9239 |
+
"grad_norm": 3.9830613136291504,
|
9240 |
+
"learning_rate": 2.7672131147540987e-05,
|
9241 |
+
"loss": 0.1909,
|
9242 |
+
"step": 26380
|
9243 |
+
},
|
9244 |
+
{
|
9245 |
+
"epoch": 128.85906040268458,
|
9246 |
+
"grad_norm": 3.0765368938446045,
|
9247 |
+
"learning_rate": 2.754098360655738e-05,
|
9248 |
+
"loss": 0.1954,
|
9249 |
+
"step": 26400
|
9250 |
+
},
|
9251 |
+
{
|
9252 |
+
"epoch": 128.95668090298963,
|
9253 |
+
"grad_norm": 2.540853977203369,
|
9254 |
+
"learning_rate": 2.7409836065573775e-05,
|
9255 |
+
"loss": 0.2116,
|
9256 |
+
"step": 26420
|
9257 |
+
},
|
9258 |
+
{
|
9259 |
+
"epoch": 129.0543014032947,
|
9260 |
+
"grad_norm": 3.55985689163208,
|
9261 |
+
"learning_rate": 2.7278688524590168e-05,
|
9262 |
+
"loss": 0.193,
|
9263 |
+
"step": 26440
|
9264 |
+
},
|
9265 |
+
{
|
9266 |
+
"epoch": 129.15192190359974,
|
9267 |
+
"grad_norm": 3.974700450897217,
|
9268 |
+
"learning_rate": 2.7147540983606558e-05,
|
9269 |
+
"loss": 0.1673,
|
9270 |
+
"step": 26460
|
9271 |
+
},
|
9272 |
+
{
|
9273 |
+
"epoch": 129.24954240390483,
|
9274 |
+
"grad_norm": 4.614022731781006,
|
9275 |
+
"learning_rate": 2.701639344262295e-05,
|
9276 |
+
"loss": 0.1968,
|
9277 |
+
"step": 26480
|
9278 |
+
},
|
9279 |
+
{
|
9280 |
+
"epoch": 129.34716290420988,
|
9281 |
+
"grad_norm": 3.9277889728546143,
|
9282 |
+
"learning_rate": 2.6885245901639343e-05,
|
9283 |
+
"loss": 0.1643,
|
9284 |
+
"step": 26500
|
9285 |
+
},
|
9286 |
+
{
|
9287 |
+
"epoch": 129.44478340451494,
|
9288 |
+
"grad_norm": 3.4351913928985596,
|
9289 |
+
"learning_rate": 2.675409836065574e-05,
|
9290 |
+
"loss": 0.1903,
|
9291 |
+
"step": 26520
|
9292 |
+
},
|
9293 |
+
{
|
9294 |
+
"epoch": 129.54240390482002,
|
9295 |
+
"grad_norm": 3.3347392082214355,
|
9296 |
+
"learning_rate": 2.6622950819672132e-05,
|
9297 |
+
"loss": 0.1789,
|
9298 |
+
"step": 26540
|
9299 |
+
},
|
9300 |
+
{
|
9301 |
+
"epoch": 129.64002440512508,
|
9302 |
+
"grad_norm": 3.970414161682129,
|
9303 |
+
"learning_rate": 2.6491803278688525e-05,
|
9304 |
+
"loss": 0.1994,
|
9305 |
+
"step": 26560
|
9306 |
+
},
|
9307 |
+
{
|
9308 |
+
"epoch": 129.73764490543013,
|
9309 |
+
"grad_norm": 3.648883819580078,
|
9310 |
+
"learning_rate": 2.636065573770492e-05,
|
9311 |
+
"loss": 0.1877,
|
9312 |
+
"step": 26580
|
9313 |
+
},
|
9314 |
+
{
|
9315 |
+
"epoch": 129.8352654057352,
|
9316 |
+
"grad_norm": 3.3394792079925537,
|
9317 |
+
"learning_rate": 2.6229508196721314e-05,
|
9318 |
+
"loss": 0.1838,
|
9319 |
+
"step": 26600
|
9320 |
+
},
|
9321 |
+
{
|
9322 |
+
"epoch": 129.93288590604027,
|
9323 |
+
"grad_norm": 2.924798011779785,
|
9324 |
+
"learning_rate": 2.6098360655737706e-05,
|
9325 |
+
"loss": 0.1913,
|
9326 |
+
"step": 26620
|
9327 |
+
},
|
9328 |
+
{
|
9329 |
+
"epoch": 130.03050640634532,
|
9330 |
+
"grad_norm": 3.448457717895508,
|
9331 |
+
"learning_rate": 2.5967213114754103e-05,
|
9332 |
+
"loss": 0.1993,
|
9333 |
+
"step": 26640
|
9334 |
+
},
|
9335 |
+
{
|
9336 |
+
"epoch": 130.1281269066504,
|
9337 |
+
"grad_norm": 2.458868980407715,
|
9338 |
+
"learning_rate": 2.5836065573770492e-05,
|
9339 |
+
"loss": 0.1996,
|
9340 |
+
"step": 26660
|
9341 |
+
},
|
9342 |
+
{
|
9343 |
+
"epoch": 130.22574740695546,
|
9344 |
+
"grad_norm": 4.39287805557251,
|
9345 |
+
"learning_rate": 2.5704918032786885e-05,
|
9346 |
+
"loss": 0.1894,
|
9347 |
+
"step": 26680
|
9348 |
+
},
|
9349 |
+
{
|
9350 |
+
"epoch": 130.32336790726052,
|
9351 |
+
"grad_norm": 3.347745180130005,
|
9352 |
+
"learning_rate": 2.5573770491803277e-05,
|
9353 |
+
"loss": 0.1642,
|
9354 |
+
"step": 26700
|
9355 |
+
},
|
9356 |
+
{
|
9357 |
+
"epoch": 130.4209884075656,
|
9358 |
+
"grad_norm": 3.0466248989105225,
|
9359 |
+
"learning_rate": 2.5442622950819674e-05,
|
9360 |
+
"loss": 0.1976,
|
9361 |
+
"step": 26720
|
9362 |
+
},
|
9363 |
+
{
|
9364 |
+
"epoch": 130.51860890787066,
|
9365 |
+
"grad_norm": 2.580834150314331,
|
9366 |
+
"learning_rate": 2.5311475409836066e-05,
|
9367 |
+
"loss": 0.1764,
|
9368 |
+
"step": 26740
|
9369 |
+
},
|
9370 |
+
{
|
9371 |
+
"epoch": 130.6162294081757,
|
9372 |
+
"grad_norm": 3.6775128841400146,
|
9373 |
+
"learning_rate": 2.518032786885246e-05,
|
9374 |
+
"loss": 0.1725,
|
9375 |
+
"step": 26760
|
9376 |
+
},
|
9377 |
+
{
|
9378 |
+
"epoch": 130.71384990848077,
|
9379 |
+
"grad_norm": 3.829058885574341,
|
9380 |
+
"learning_rate": 2.5049180327868855e-05,
|
9381 |
+
"loss": 0.1776,
|
9382 |
+
"step": 26780
|
9383 |
+
},
|
9384 |
+
{
|
9385 |
+
"epoch": 130.81147040878585,
|
9386 |
+
"grad_norm": 4.047943592071533,
|
9387 |
+
"learning_rate": 2.4918032786885248e-05,
|
9388 |
+
"loss": 0.1943,
|
9389 |
+
"step": 26800
|
9390 |
+
},
|
9391 |
+
{
|
9392 |
+
"epoch": 130.9090909090909,
|
9393 |
+
"grad_norm": 3.2828705310821533,
|
9394 |
+
"learning_rate": 2.478688524590164e-05,
|
9395 |
+
"loss": 0.1864,
|
9396 |
+
"step": 26820
|
9397 |
+
},
|
9398 |
+
{
|
9399 |
+
"epoch": 131.00671140939596,
|
9400 |
+
"grad_norm": 4.2563958168029785,
|
9401 |
+
"learning_rate": 2.4655737704918033e-05,
|
9402 |
+
"loss": 0.1838,
|
9403 |
+
"step": 26840
|
9404 |
+
},
|
9405 |
+
{
|
9406 |
+
"epoch": 131.10433190970105,
|
9407 |
+
"grad_norm": 3.279503583908081,
|
9408 |
+
"learning_rate": 2.4524590163934426e-05,
|
9409 |
+
"loss": 0.2027,
|
9410 |
+
"step": 26860
|
9411 |
+
},
|
9412 |
+
{
|
9413 |
+
"epoch": 131.2019524100061,
|
9414 |
+
"grad_norm": 3.8052897453308105,
|
9415 |
+
"learning_rate": 2.4393442622950822e-05,
|
9416 |
+
"loss": 0.1916,
|
9417 |
+
"step": 26880
|
9418 |
+
},
|
9419 |
+
{
|
9420 |
+
"epoch": 131.29957291031116,
|
9421 |
+
"grad_norm": 3.12294602394104,
|
9422 |
+
"learning_rate": 2.4262295081967215e-05,
|
9423 |
+
"loss": 0.1542,
|
9424 |
+
"step": 26900
|
9425 |
+
},
|
9426 |
+
{
|
9427 |
+
"epoch": 131.39719341061624,
|
9428 |
+
"grad_norm": 4.636548042297363,
|
9429 |
+
"learning_rate": 2.4131147540983608e-05,
|
9430 |
+
"loss": 0.1677,
|
9431 |
+
"step": 26920
|
9432 |
+
},
|
9433 |
+
{
|
9434 |
+
"epoch": 131.4948139109213,
|
9435 |
+
"grad_norm": 2.8608415126800537,
|
9436 |
+
"learning_rate": 2.4e-05,
|
9437 |
+
"loss": 0.194,
|
9438 |
+
"step": 26940
|
9439 |
+
},
|
9440 |
+
{
|
9441 |
+
"epoch": 131.59243441122635,
|
9442 |
+
"grad_norm": 3.7946908473968506,
|
9443 |
+
"learning_rate": 2.3868852459016393e-05,
|
9444 |
+
"loss": 0.1764,
|
9445 |
+
"step": 26960
|
9446 |
+
},
|
9447 |
+
{
|
9448 |
+
"epoch": 131.69005491153143,
|
9449 |
+
"grad_norm": 3.1568832397460938,
|
9450 |
+
"learning_rate": 2.373770491803279e-05,
|
9451 |
+
"loss": 0.1865,
|
9452 |
+
"step": 26980
|
9453 |
+
},
|
9454 |
+
{
|
9455 |
+
"epoch": 131.7876754118365,
|
9456 |
+
"grad_norm": 4.103198528289795,
|
9457 |
+
"learning_rate": 2.3606557377049182e-05,
|
9458 |
+
"loss": 0.1753,
|
9459 |
+
"step": 27000
|
9460 |
+
},
|
9461 |
+
{
|
9462 |
+
"epoch": 131.88529591214154,
|
9463 |
+
"grad_norm": 3.655327796936035,
|
9464 |
+
"learning_rate": 2.3475409836065575e-05,
|
9465 |
+
"loss": 0.1965,
|
9466 |
+
"step": 27020
|
9467 |
+
},
|
9468 |
+
{
|
9469 |
+
"epoch": 131.98291641244663,
|
9470 |
+
"grad_norm": 3.908200263977051,
|
9471 |
+
"learning_rate": 2.3344262295081968e-05,
|
9472 |
+
"loss": 0.1791,
|
9473 |
+
"step": 27040
|
9474 |
+
},
|
9475 |
+
{
|
9476 |
+
"epoch": 132.08053691275168,
|
9477 |
+
"grad_norm": 3.2557129859924316,
|
9478 |
+
"learning_rate": 2.321311475409836e-05,
|
9479 |
+
"loss": 0.1715,
|
9480 |
+
"step": 27060
|
9481 |
+
},
|
9482 |
+
{
|
9483 |
+
"epoch": 132.17815741305674,
|
9484 |
+
"grad_norm": 3.102268695831299,
|
9485 |
+
"learning_rate": 2.3081967213114757e-05,
|
9486 |
+
"loss": 0.191,
|
9487 |
+
"step": 27080
|
9488 |
+
},
|
9489 |
+
{
|
9490 |
+
"epoch": 132.2757779133618,
|
9491 |
+
"grad_norm": 2.7028493881225586,
|
9492 |
+
"learning_rate": 2.295081967213115e-05,
|
9493 |
+
"loss": 0.1802,
|
9494 |
+
"step": 27100
|
9495 |
+
},
|
9496 |
+
{
|
9497 |
+
"epoch": 132.37339841366688,
|
9498 |
+
"grad_norm": 2.5240046977996826,
|
9499 |
+
"learning_rate": 2.2819672131147542e-05,
|
9500 |
+
"loss": 0.1618,
|
9501 |
+
"step": 27120
|
9502 |
+
},
|
9503 |
+
{
|
9504 |
+
"epoch": 132.47101891397193,
|
9505 |
+
"grad_norm": 3.0410265922546387,
|
9506 |
+
"learning_rate": 2.2688524590163935e-05,
|
9507 |
+
"loss": 0.1833,
|
9508 |
+
"step": 27140
|
9509 |
+
},
|
9510 |
+
{
|
9511 |
+
"epoch": 132.568639414277,
|
9512 |
+
"grad_norm": 3.677824020385742,
|
9513 |
+
"learning_rate": 2.2557377049180328e-05,
|
9514 |
+
"loss": 0.1804,
|
9515 |
+
"step": 27160
|
9516 |
+
},
|
9517 |
+
{
|
9518 |
+
"epoch": 132.66625991458207,
|
9519 |
+
"grad_norm": 2.826828718185425,
|
9520 |
+
"learning_rate": 2.2426229508196724e-05,
|
9521 |
+
"loss": 0.1915,
|
9522 |
+
"step": 27180
|
9523 |
+
},
|
9524 |
+
{
|
9525 |
+
"epoch": 132.76388041488713,
|
9526 |
+
"grad_norm": 4.393260955810547,
|
9527 |
+
"learning_rate": 2.2295081967213117e-05,
|
9528 |
+
"loss": 0.1952,
|
9529 |
+
"step": 27200
|
9530 |
+
},
|
9531 |
+
{
|
9532 |
+
"epoch": 132.86150091519218,
|
9533 |
+
"grad_norm": 3.1146349906921387,
|
9534 |
+
"learning_rate": 2.216393442622951e-05,
|
9535 |
+
"loss": 0.1973,
|
9536 |
+
"step": 27220
|
9537 |
+
},
|
9538 |
+
{
|
9539 |
+
"epoch": 132.95912141549726,
|
9540 |
+
"grad_norm": 3.343693494796753,
|
9541 |
+
"learning_rate": 2.2032786885245905e-05,
|
9542 |
+
"loss": 0.1822,
|
9543 |
+
"step": 27240
|
9544 |
+
},
|
9545 |
+
{
|
9546 |
+
"epoch": 133.05674191580232,
|
9547 |
+
"grad_norm": 3.0667552947998047,
|
9548 |
+
"learning_rate": 2.1901639344262295e-05,
|
9549 |
+
"loss": 0.1807,
|
9550 |
+
"step": 27260
|
9551 |
+
},
|
9552 |
+
{
|
9553 |
+
"epoch": 133.15436241610738,
|
9554 |
+
"grad_norm": 3.497859001159668,
|
9555 |
+
"learning_rate": 2.1770491803278688e-05,
|
9556 |
+
"loss": 0.1525,
|
9557 |
+
"step": 27280
|
9558 |
+
},
|
9559 |
+
{
|
9560 |
+
"epoch": 133.25198291641246,
|
9561 |
+
"grad_norm": 3.295478343963623,
|
9562 |
+
"learning_rate": 2.1639344262295084e-05,
|
9563 |
+
"loss": 0.1971,
|
9564 |
+
"step": 27300
|
9565 |
+
},
|
9566 |
+
{
|
9567 |
+
"epoch": 133.3496034167175,
|
9568 |
+
"grad_norm": 3.3247010707855225,
|
9569 |
+
"learning_rate": 2.1508196721311476e-05,
|
9570 |
+
"loss": 0.1963,
|
9571 |
+
"step": 27320
|
9572 |
+
},
|
9573 |
+
{
|
9574 |
+
"epoch": 133.44722391702257,
|
9575 |
+
"grad_norm": 4.269167900085449,
|
9576 |
+
"learning_rate": 2.1377049180327873e-05,
|
9577 |
+
"loss": 0.1967,
|
9578 |
+
"step": 27340
|
9579 |
+
},
|
9580 |
+
{
|
9581 |
+
"epoch": 133.54484441732765,
|
9582 |
+
"grad_norm": 3.7575721740722656,
|
9583 |
+
"learning_rate": 2.1245901639344262e-05,
|
9584 |
+
"loss": 0.1788,
|
9585 |
+
"step": 27360
|
9586 |
+
},
|
9587 |
+
{
|
9588 |
+
"epoch": 133.6424649176327,
|
9589 |
+
"grad_norm": 4.189979553222656,
|
9590 |
+
"learning_rate": 2.1114754098360655e-05,
|
9591 |
+
"loss": 0.1922,
|
9592 |
+
"step": 27380
|
9593 |
+
},
|
9594 |
+
{
|
9595 |
+
"epoch": 133.74008541793776,
|
9596 |
+
"grad_norm": 3.48610782623291,
|
9597 |
+
"learning_rate": 2.098360655737705e-05,
|
9598 |
+
"loss": 0.1749,
|
9599 |
+
"step": 27400
|
9600 |
+
},
|
9601 |
+
{
|
9602 |
+
"epoch": 133.83770591824282,
|
9603 |
+
"grad_norm": 3.320037364959717,
|
9604 |
+
"learning_rate": 2.0852459016393444e-05,
|
9605 |
+
"loss": 0.1886,
|
9606 |
+
"step": 27420
|
9607 |
+
},
|
9608 |
+
{
|
9609 |
+
"epoch": 133.9353264185479,
|
9610 |
+
"grad_norm": 3.347099781036377,
|
9611 |
+
"learning_rate": 2.0721311475409836e-05,
|
9612 |
+
"loss": 0.1717,
|
9613 |
+
"step": 27440
|
9614 |
+
},
|
9615 |
+
{
|
9616 |
+
"epoch": 134.03294691885296,
|
9617 |
+
"grad_norm": 2.9406418800354004,
|
9618 |
+
"learning_rate": 2.059016393442623e-05,
|
9619 |
+
"loss": 0.1702,
|
9620 |
+
"step": 27460
|
9621 |
+
},
|
9622 |
+
{
|
9623 |
+
"epoch": 134.130567419158,
|
9624 |
+
"grad_norm": 2.3794620037078857,
|
9625 |
+
"learning_rate": 2.0459016393442622e-05,
|
9626 |
+
"loss": 0.1872,
|
9627 |
+
"step": 27480
|
9628 |
+
},
|
9629 |
+
{
|
9630 |
+
"epoch": 134.2281879194631,
|
9631 |
+
"grad_norm": 3.1150660514831543,
|
9632 |
+
"learning_rate": 2.0327868852459018e-05,
|
9633 |
+
"loss": 0.2023,
|
9634 |
+
"step": 27500
|
9635 |
+
},
|
9636 |
+
{
|
9637 |
+
"epoch": 134.32580841976815,
|
9638 |
+
"grad_norm": 3.679694414138794,
|
9639 |
+
"learning_rate": 2.019672131147541e-05,
|
9640 |
+
"loss": 0.1858,
|
9641 |
+
"step": 27520
|
9642 |
+
},
|
9643 |
+
{
|
9644 |
+
"epoch": 134.4234289200732,
|
9645 |
+
"grad_norm": 2.665882110595703,
|
9646 |
+
"learning_rate": 2.0065573770491804e-05,
|
9647 |
+
"loss": 0.1782,
|
9648 |
+
"step": 27540
|
9649 |
+
},
|
9650 |
+
{
|
9651 |
+
"epoch": 134.5210494203783,
|
9652 |
+
"grad_norm": 3.0053212642669678,
|
9653 |
+
"learning_rate": 1.99344262295082e-05,
|
9654 |
+
"loss": 0.1773,
|
9655 |
+
"step": 27560
|
9656 |
+
},
|
9657 |
+
{
|
9658 |
+
"epoch": 134.61866992068335,
|
9659 |
+
"grad_norm": 2.689307451248169,
|
9660 |
+
"learning_rate": 1.980327868852459e-05,
|
9661 |
+
"loss": 0.1588,
|
9662 |
+
"step": 27580
|
9663 |
+
},
|
9664 |
+
{
|
9665 |
+
"epoch": 134.7162904209884,
|
9666 |
+
"grad_norm": 4.318088054656982,
|
9667 |
+
"learning_rate": 1.9672131147540985e-05,
|
9668 |
+
"loss": 0.1436,
|
9669 |
+
"step": 27600
|
9670 |
+
},
|
9671 |
+
{
|
9672 |
+
"epoch": 134.81391092129348,
|
9673 |
+
"grad_norm": 3.3378102779388428,
|
9674 |
+
"learning_rate": 1.9540983606557378e-05,
|
9675 |
+
"loss": 0.1919,
|
9676 |
+
"step": 27620
|
9677 |
+
},
|
9678 |
+
{
|
9679 |
+
"epoch": 134.91153142159854,
|
9680 |
+
"grad_norm": 3.2355871200561523,
|
9681 |
+
"learning_rate": 1.940983606557377e-05,
|
9682 |
+
"loss": 0.1828,
|
9683 |
+
"step": 27640
|
9684 |
+
},
|
9685 |
+
{
|
9686 |
+
"epoch": 135.0091519219036,
|
9687 |
+
"grad_norm": 3.1335229873657227,
|
9688 |
+
"learning_rate": 1.9278688524590167e-05,
|
9689 |
+
"loss": 0.1982,
|
9690 |
+
"step": 27660
|
9691 |
+
},
|
9692 |
+
{
|
9693 |
+
"epoch": 135.10677242220865,
|
9694 |
+
"grad_norm": 4.066319465637207,
|
9695 |
+
"learning_rate": 1.9147540983606556e-05,
|
9696 |
+
"loss": 0.1526,
|
9697 |
+
"step": 27680
|
9698 |
+
},
|
9699 |
+
{
|
9700 |
+
"epoch": 135.20439292251373,
|
9701 |
+
"grad_norm": 3.631089925765991,
|
9702 |
+
"learning_rate": 1.9016393442622952e-05,
|
9703 |
+
"loss": 0.1776,
|
9704 |
+
"step": 27700
|
9705 |
+
},
|
9706 |
+
{
|
9707 |
+
"epoch": 135.3020134228188,
|
9708 |
+
"grad_norm": 3.5840327739715576,
|
9709 |
+
"learning_rate": 1.8885245901639345e-05,
|
9710 |
+
"loss": 0.1922,
|
9711 |
+
"step": 27720
|
9712 |
+
},
|
9713 |
+
{
|
9714 |
+
"epoch": 135.39963392312384,
|
9715 |
+
"grad_norm": 2.926558256149292,
|
9716 |
+
"learning_rate": 1.8754098360655738e-05,
|
9717 |
+
"loss": 0.1847,
|
9718 |
+
"step": 27740
|
9719 |
+
},
|
9720 |
+
{
|
9721 |
+
"epoch": 135.49725442342893,
|
9722 |
+
"grad_norm": 4.487957000732422,
|
9723 |
+
"learning_rate": 1.8622950819672134e-05,
|
9724 |
+
"loss": 0.1896,
|
9725 |
+
"step": 27760
|
9726 |
+
},
|
9727 |
+
{
|
9728 |
+
"epoch": 135.59487492373398,
|
9729 |
+
"grad_norm": 3.209500789642334,
|
9730 |
+
"learning_rate": 1.8491803278688523e-05,
|
9731 |
+
"loss": 0.1827,
|
9732 |
+
"step": 27780
|
9733 |
+
},
|
9734 |
+
{
|
9735 |
+
"epoch": 135.69249542403904,
|
9736 |
+
"grad_norm": 2.8735058307647705,
|
9737 |
+
"learning_rate": 1.836065573770492e-05,
|
9738 |
+
"loss": 0.1671,
|
9739 |
+
"step": 27800
|
9740 |
+
},
|
9741 |
+
{
|
9742 |
+
"epoch": 135.79011592434412,
|
9743 |
+
"grad_norm": 3.221266508102417,
|
9744 |
+
"learning_rate": 1.8229508196721312e-05,
|
9745 |
+
"loss": 0.1861,
|
9746 |
+
"step": 27820
|
9747 |
+
},
|
9748 |
+
{
|
9749 |
+
"epoch": 135.88773642464918,
|
9750 |
+
"grad_norm": 3.1269659996032715,
|
9751 |
+
"learning_rate": 1.8098360655737705e-05,
|
9752 |
+
"loss": 0.1857,
|
9753 |
+
"step": 27840
|
9754 |
+
},
|
9755 |
+
{
|
9756 |
+
"epoch": 135.98535692495423,
|
9757 |
+
"grad_norm": 3.3478143215179443,
|
9758 |
+
"learning_rate": 1.79672131147541e-05,
|
9759 |
+
"loss": 0.1853,
|
9760 |
+
"step": 27860
|
9761 |
+
},
|
9762 |
+
{
|
9763 |
+
"epoch": 136.08297742525932,
|
9764 |
+
"grad_norm": 2.8102643489837646,
|
9765 |
+
"learning_rate": 1.7836065573770494e-05,
|
9766 |
+
"loss": 0.1691,
|
9767 |
+
"step": 27880
|
9768 |
+
},
|
9769 |
+
{
|
9770 |
+
"epoch": 136.18059792556437,
|
9771 |
+
"grad_norm": 2.9172909259796143,
|
9772 |
+
"learning_rate": 1.7704918032786887e-05,
|
9773 |
+
"loss": 0.1805,
|
9774 |
+
"step": 27900
|
9775 |
+
},
|
9776 |
+
{
|
9777 |
+
"epoch": 136.27821842586943,
|
9778 |
+
"grad_norm": 3.6803557872772217,
|
9779 |
+
"learning_rate": 1.757377049180328e-05,
|
9780 |
+
"loss": 0.1808,
|
9781 |
+
"step": 27920
|
9782 |
+
},
|
9783 |
+
{
|
9784 |
+
"epoch": 136.3758389261745,
|
9785 |
+
"grad_norm": 3.6122121810913086,
|
9786 |
+
"learning_rate": 1.7442622950819672e-05,
|
9787 |
+
"loss": 0.1825,
|
9788 |
+
"step": 27940
|
9789 |
+
},
|
9790 |
+
{
|
9791 |
+
"epoch": 136.47345942647956,
|
9792 |
+
"grad_norm": 2.836901903152466,
|
9793 |
+
"learning_rate": 1.731147540983607e-05,
|
9794 |
+
"loss": 0.1762,
|
9795 |
+
"step": 27960
|
9796 |
+
},
|
9797 |
+
{
|
9798 |
+
"epoch": 136.57107992678462,
|
9799 |
+
"grad_norm": 4.931893348693848,
|
9800 |
+
"learning_rate": 1.718032786885246e-05,
|
9801 |
+
"loss": 0.2097,
|
9802 |
+
"step": 27980
|
9803 |
+
},
|
9804 |
+
{
|
9805 |
+
"epoch": 136.66870042708968,
|
9806 |
+
"grad_norm": 3.634223461151123,
|
9807 |
+
"learning_rate": 1.7049180327868854e-05,
|
9808 |
+
"loss": 0.1764,
|
9809 |
+
"step": 28000
|
9810 |
+
},
|
9811 |
+
{
|
9812 |
+
"epoch": 136.76632092739476,
|
9813 |
+
"grad_norm": 4.417370319366455,
|
9814 |
+
"learning_rate": 1.6918032786885247e-05,
|
9815 |
+
"loss": 0.1695,
|
9816 |
+
"step": 28020
|
9817 |
+
},
|
9818 |
+
{
|
9819 |
+
"epoch": 136.8639414276998,
|
9820 |
+
"grad_norm": 3.6770431995391846,
|
9821 |
+
"learning_rate": 1.678688524590164e-05,
|
9822 |
+
"loss": 0.1762,
|
9823 |
+
"step": 28040
|
9824 |
+
},
|
9825 |
+
{
|
9826 |
+
"epoch": 136.96156192800487,
|
9827 |
+
"grad_norm": 3.134272813796997,
|
9828 |
+
"learning_rate": 1.6655737704918036e-05,
|
9829 |
+
"loss": 0.1603,
|
9830 |
+
"step": 28060
|
9831 |
+
},
|
9832 |
+
{
|
9833 |
+
"epoch": 137.05918242830995,
|
9834 |
+
"grad_norm": 3.993882179260254,
|
9835 |
+
"learning_rate": 1.6524590163934428e-05,
|
9836 |
+
"loss": 0.1927,
|
9837 |
+
"step": 28080
|
9838 |
+
},
|
9839 |
+
{
|
9840 |
+
"epoch": 137.156802928615,
|
9841 |
+
"grad_norm": 3.1614527702331543,
|
9842 |
+
"learning_rate": 1.6393442622950818e-05,
|
9843 |
+
"loss": 0.1739,
|
9844 |
+
"step": 28100
|
9845 |
+
},
|
9846 |
+
{
|
9847 |
+
"epoch": 137.25442342892006,
|
9848 |
+
"grad_norm": 2.9293642044067383,
|
9849 |
+
"learning_rate": 1.6262295081967214e-05,
|
9850 |
+
"loss": 0.1916,
|
9851 |
+
"step": 28120
|
9852 |
+
},
|
9853 |
+
{
|
9854 |
+
"epoch": 137.35204392922515,
|
9855 |
+
"grad_norm": 3.292917251586914,
|
9856 |
+
"learning_rate": 1.6131147540983607e-05,
|
9857 |
+
"loss": 0.1638,
|
9858 |
+
"step": 28140
|
9859 |
+
},
|
9860 |
+
{
|
9861 |
+
"epoch": 137.4496644295302,
|
9862 |
+
"grad_norm": 3.048471212387085,
|
9863 |
+
"learning_rate": 1.6000000000000003e-05,
|
9864 |
+
"loss": 0.1877,
|
9865 |
+
"step": 28160
|
9866 |
+
},
|
9867 |
+
{
|
9868 |
+
"epoch": 137.54728492983526,
|
9869 |
+
"grad_norm": 3.398252487182617,
|
9870 |
+
"learning_rate": 1.5868852459016395e-05,
|
9871 |
+
"loss": 0.1685,
|
9872 |
+
"step": 28180
|
9873 |
+
},
|
9874 |
+
{
|
9875 |
+
"epoch": 137.64490543014034,
|
9876 |
+
"grad_norm": 3.906764268875122,
|
9877 |
+
"learning_rate": 1.5737704918032788e-05,
|
9878 |
+
"loss": 0.1772,
|
9879 |
+
"step": 28200
|
9880 |
+
},
|
9881 |
+
{
|
9882 |
+
"epoch": 137.7425259304454,
|
9883 |
+
"grad_norm": 3.1852598190307617,
|
9884 |
+
"learning_rate": 1.560655737704918e-05,
|
9885 |
+
"loss": 0.168,
|
9886 |
+
"step": 28220
|
9887 |
+
},
|
9888 |
+
{
|
9889 |
+
"epoch": 137.84014643075045,
|
9890 |
+
"grad_norm": 2.9285385608673096,
|
9891 |
+
"learning_rate": 1.5475409836065574e-05,
|
9892 |
+
"loss": 0.1733,
|
9893 |
+
"step": 28240
|
9894 |
+
},
|
9895 |
+
{
|
9896 |
+
"epoch": 137.93776693105553,
|
9897 |
+
"grad_norm": 3.9846913814544678,
|
9898 |
+
"learning_rate": 1.5344262295081966e-05,
|
9899 |
+
"loss": 0.1906,
|
9900 |
+
"step": 28260
|
9901 |
+
},
|
9902 |
+
{
|
9903 |
+
"epoch": 138.0353874313606,
|
9904 |
+
"grad_norm": 3.3251185417175293,
|
9905 |
+
"learning_rate": 1.5213114754098363e-05,
|
9906 |
+
"loss": 0.1872,
|
9907 |
+
"step": 28280
|
9908 |
+
},
|
9909 |
+
{
|
9910 |
+
"epoch": 138.13300793166565,
|
9911 |
+
"grad_norm": 3.7904114723205566,
|
9912 |
+
"learning_rate": 1.5081967213114755e-05,
|
9913 |
+
"loss": 0.1811,
|
9914 |
+
"step": 28300
|
9915 |
+
},
|
9916 |
+
{
|
9917 |
+
"epoch": 138.2306284319707,
|
9918 |
+
"grad_norm": 3.5363810062408447,
|
9919 |
+
"learning_rate": 1.4950819672131146e-05,
|
9920 |
+
"loss": 0.1803,
|
9921 |
+
"step": 28320
|
9922 |
+
},
|
9923 |
+
{
|
9924 |
+
"epoch": 138.32824893227578,
|
9925 |
+
"grad_norm": 3.430577516555786,
|
9926 |
+
"learning_rate": 1.481967213114754e-05,
|
9927 |
+
"loss": 0.1619,
|
9928 |
+
"step": 28340
|
9929 |
+
},
|
9930 |
+
{
|
9931 |
+
"epoch": 138.42586943258084,
|
9932 |
+
"grad_norm": 3.1190154552459717,
|
9933 |
+
"learning_rate": 1.4688524590163935e-05,
|
9934 |
+
"loss": 0.1909,
|
9935 |
+
"step": 28360
|
9936 |
+
},
|
9937 |
+
{
|
9938 |
+
"epoch": 138.5234899328859,
|
9939 |
+
"grad_norm": 2.656212091445923,
|
9940 |
+
"learning_rate": 1.455737704918033e-05,
|
9941 |
+
"loss": 0.1682,
|
9942 |
+
"step": 28380
|
9943 |
+
},
|
9944 |
+
{
|
9945 |
+
"epoch": 138.62111043319098,
|
9946 |
+
"grad_norm": 3.1637041568756104,
|
9947 |
+
"learning_rate": 1.4426229508196722e-05,
|
9948 |
+
"loss": 0.1729,
|
9949 |
+
"step": 28400
|
9950 |
+
},
|
9951 |
+
{
|
9952 |
+
"epoch": 138.71873093349603,
|
9953 |
+
"grad_norm": 3.313136100769043,
|
9954 |
+
"learning_rate": 1.4295081967213114e-05,
|
9955 |
+
"loss": 0.1853,
|
9956 |
+
"step": 28420
|
9957 |
+
},
|
9958 |
+
{
|
9959 |
+
"epoch": 138.8163514338011,
|
9960 |
+
"grad_norm": 3.63885498046875,
|
9961 |
+
"learning_rate": 1.4163934426229508e-05,
|
9962 |
+
"loss": 0.1783,
|
9963 |
+
"step": 28440
|
9964 |
+
},
|
9965 |
+
{
|
9966 |
+
"epoch": 138.91397193410617,
|
9967 |
+
"grad_norm": 4.251205921173096,
|
9968 |
+
"learning_rate": 1.4032786885245902e-05,
|
9969 |
+
"loss": 0.1703,
|
9970 |
+
"step": 28460
|
9971 |
+
},
|
9972 |
+
{
|
9973 |
+
"epoch": 139.01159243441123,
|
9974 |
+
"grad_norm": 3.76887583732605,
|
9975 |
+
"learning_rate": 1.3901639344262297e-05,
|
9976 |
+
"loss": 0.1871,
|
9977 |
+
"step": 28480
|
9978 |
+
},
|
9979 |
+
{
|
9980 |
+
"epoch": 139.10921293471628,
|
9981 |
+
"grad_norm": 3.2986671924591064,
|
9982 |
+
"learning_rate": 1.377049180327869e-05,
|
9983 |
+
"loss": 0.1728,
|
9984 |
+
"step": 28500
|
9985 |
+
},
|
9986 |
+
{
|
9987 |
+
"epoch": 139.20683343502137,
|
9988 |
+
"grad_norm": 2.5212690830230713,
|
9989 |
+
"learning_rate": 1.3639344262295084e-05,
|
9990 |
+
"loss": 0.175,
|
9991 |
+
"step": 28520
|
9992 |
+
},
|
9993 |
+
{
|
9994 |
+
"epoch": 139.30445393532642,
|
9995 |
+
"grad_norm": 4.492109298706055,
|
9996 |
+
"learning_rate": 1.3508196721311475e-05,
|
9997 |
+
"loss": 0.1838,
|
9998 |
+
"step": 28540
|
9999 |
+
},
|
10000 |
+
{
|
10001 |
+
"epoch": 139.40207443563148,
|
10002 |
+
"grad_norm": 3.0226120948791504,
|
10003 |
+
"learning_rate": 1.337704918032787e-05,
|
10004 |
+
"loss": 0.1753,
|
10005 |
+
"step": 28560
|
10006 |
+
},
|
10007 |
+
{
|
10008 |
+
"epoch": 139.49969493593656,
|
10009 |
+
"grad_norm": 2.4843361377716064,
|
10010 |
+
"learning_rate": 1.3245901639344262e-05,
|
10011 |
+
"loss": 0.1915,
|
10012 |
+
"step": 28580
|
10013 |
+
},
|
10014 |
+
{
|
10015 |
+
"epoch": 139.59731543624162,
|
10016 |
+
"grad_norm": 3.4304590225219727,
|
10017 |
+
"learning_rate": 1.3114754098360657e-05,
|
10018 |
+
"loss": 0.1703,
|
10019 |
+
"step": 28600
|
10020 |
+
},
|
10021 |
+
{
|
10022 |
+
"epoch": 139.69493593654667,
|
10023 |
+
"grad_norm": 3.242751121520996,
|
10024 |
+
"learning_rate": 1.2983606557377051e-05,
|
10025 |
+
"loss": 0.176,
|
10026 |
+
"step": 28620
|
10027 |
+
},
|
10028 |
+
{
|
10029 |
+
"epoch": 139.79255643685173,
|
10030 |
+
"grad_norm": 3.2713284492492676,
|
10031 |
+
"learning_rate": 1.2852459016393442e-05,
|
10032 |
+
"loss": 0.1838,
|
10033 |
+
"step": 28640
|
10034 |
+
},
|
10035 |
+
{
|
10036 |
+
"epoch": 139.8901769371568,
|
10037 |
+
"grad_norm": 2.9921929836273193,
|
10038 |
+
"learning_rate": 1.2721311475409837e-05,
|
10039 |
+
"loss": 0.1764,
|
10040 |
+
"step": 28660
|
10041 |
+
},
|
10042 |
+
{
|
10043 |
+
"epoch": 139.98779743746186,
|
10044 |
+
"grad_norm": 3.6933350563049316,
|
10045 |
+
"learning_rate": 1.259016393442623e-05,
|
10046 |
+
"loss": 0.1803,
|
10047 |
+
"step": 28680
|
10048 |
+
},
|
10049 |
+
{
|
10050 |
+
"epoch": 140.08541793776692,
|
10051 |
+
"grad_norm": 2.947892665863037,
|
10052 |
+
"learning_rate": 1.2459016393442624e-05,
|
10053 |
+
"loss": 0.1732,
|
10054 |
+
"step": 28700
|
10055 |
+
},
|
10056 |
+
{
|
10057 |
+
"epoch": 140.183038438072,
|
10058 |
+
"grad_norm": 3.5047738552093506,
|
10059 |
+
"learning_rate": 1.2327868852459017e-05,
|
10060 |
+
"loss": 0.1751,
|
10061 |
+
"step": 28720
|
10062 |
+
},
|
10063 |
+
{
|
10064 |
+
"epoch": 140.28065893837706,
|
10065 |
+
"grad_norm": 2.9315600395202637,
|
10066 |
+
"learning_rate": 1.2196721311475411e-05,
|
10067 |
+
"loss": 0.1879,
|
10068 |
+
"step": 28740
|
10069 |
+
},
|
10070 |
+
{
|
10071 |
+
"epoch": 140.3782794386821,
|
10072 |
+
"grad_norm": 3.6878082752227783,
|
10073 |
+
"learning_rate": 1.2065573770491804e-05,
|
10074 |
+
"loss": 0.1551,
|
10075 |
+
"step": 28760
|
10076 |
+
},
|
10077 |
+
{
|
10078 |
+
"epoch": 140.4758999389872,
|
10079 |
+
"grad_norm": 3.2162342071533203,
|
10080 |
+
"learning_rate": 1.1934426229508197e-05,
|
10081 |
+
"loss": 0.183,
|
10082 |
+
"step": 28780
|
10083 |
+
},
|
10084 |
+
{
|
10085 |
+
"epoch": 140.57352043929225,
|
10086 |
+
"grad_norm": 3.6583456993103027,
|
10087 |
+
"learning_rate": 1.1803278688524591e-05,
|
10088 |
+
"loss": 0.1784,
|
10089 |
+
"step": 28800
|
10090 |
+
},
|
10091 |
+
{
|
10092 |
+
"epoch": 140.6711409395973,
|
10093 |
+
"grad_norm": 2.780412435531616,
|
10094 |
+
"learning_rate": 1.1672131147540984e-05,
|
10095 |
+
"loss": 0.1885,
|
10096 |
+
"step": 28820
|
10097 |
+
},
|
10098 |
+
{
|
10099 |
+
"epoch": 140.7687614399024,
|
10100 |
+
"grad_norm": 2.3773672580718994,
|
10101 |
+
"learning_rate": 1.1540983606557378e-05,
|
10102 |
+
"loss": 0.1611,
|
10103 |
+
"step": 28840
|
10104 |
+
},
|
10105 |
+
{
|
10106 |
+
"epoch": 140.86638194020745,
|
10107 |
+
"grad_norm": 3.6276142597198486,
|
10108 |
+
"learning_rate": 1.1409836065573771e-05,
|
10109 |
+
"loss": 0.1806,
|
10110 |
+
"step": 28860
|
10111 |
+
},
|
10112 |
+
{
|
10113 |
+
"epoch": 140.9640024405125,
|
10114 |
+
"grad_norm": 3.3313121795654297,
|
10115 |
+
"learning_rate": 1.1278688524590164e-05,
|
10116 |
+
"loss": 0.1741,
|
10117 |
+
"step": 28880
|
10118 |
+
},
|
10119 |
+
{
|
10120 |
+
"epoch": 141.06162294081759,
|
10121 |
+
"grad_norm": 3.1109941005706787,
|
10122 |
+
"learning_rate": 1.1147540983606558e-05,
|
10123 |
+
"loss": 0.1747,
|
10124 |
+
"step": 28900
|
10125 |
+
},
|
10126 |
+
{
|
10127 |
+
"epoch": 141.15924344112264,
|
10128 |
+
"grad_norm": 2.3083291053771973,
|
10129 |
+
"learning_rate": 1.1016393442622953e-05,
|
10130 |
+
"loss": 0.16,
|
10131 |
+
"step": 28920
|
10132 |
+
},
|
10133 |
+
{
|
10134 |
+
"epoch": 141.2568639414277,
|
10135 |
+
"grad_norm": 3.6427536010742188,
|
10136 |
+
"learning_rate": 1.0885245901639344e-05,
|
10137 |
+
"loss": 0.1792,
|
10138 |
+
"step": 28940
|
10139 |
+
},
|
10140 |
+
{
|
10141 |
+
"epoch": 141.35448444173275,
|
10142 |
+
"grad_norm": 3.2836146354675293,
|
10143 |
+
"learning_rate": 1.0754098360655738e-05,
|
10144 |
+
"loss": 0.1921,
|
10145 |
+
"step": 28960
|
10146 |
+
},
|
10147 |
+
{
|
10148 |
+
"epoch": 141.45210494203783,
|
10149 |
+
"grad_norm": 3.712411642074585,
|
10150 |
+
"learning_rate": 1.0622950819672131e-05,
|
10151 |
+
"loss": 0.1834,
|
10152 |
+
"step": 28980
|
10153 |
+
},
|
10154 |
+
{
|
10155 |
+
"epoch": 141.5497254423429,
|
10156 |
+
"grad_norm": 2.6064016819000244,
|
10157 |
+
"learning_rate": 1.0491803278688525e-05,
|
10158 |
+
"loss": 0.1975,
|
10159 |
+
"step": 29000
|
10160 |
+
},
|
10161 |
+
{
|
10162 |
+
"epoch": 141.64734594264795,
|
10163 |
+
"grad_norm": 3.3907470703125,
|
10164 |
+
"learning_rate": 1.0360655737704918e-05,
|
10165 |
+
"loss": 0.172,
|
10166 |
+
"step": 29020
|
10167 |
+
},
|
10168 |
+
{
|
10169 |
+
"epoch": 141.74496644295303,
|
10170 |
+
"grad_norm": 3.0713050365448,
|
10171 |
+
"learning_rate": 1.0229508196721311e-05,
|
10172 |
+
"loss": 0.1547,
|
10173 |
+
"step": 29040
|
10174 |
+
},
|
10175 |
+
{
|
10176 |
+
"epoch": 141.84258694325808,
|
10177 |
+
"grad_norm": 5.218588352203369,
|
10178 |
+
"learning_rate": 1.0098360655737705e-05,
|
10179 |
+
"loss": 0.1712,
|
10180 |
+
"step": 29060
|
10181 |
+
},
|
10182 |
+
{
|
10183 |
+
"epoch": 141.94020744356314,
|
10184 |
+
"grad_norm": 3.205132007598877,
|
10185 |
+
"learning_rate": 9.9672131147541e-06,
|
10186 |
+
"loss": 0.1629,
|
10187 |
+
"step": 29080
|
10188 |
+
},
|
10189 |
+
{
|
10190 |
+
"epoch": 142.03782794386822,
|
10191 |
+
"grad_norm": 3.580003499984741,
|
10192 |
+
"learning_rate": 9.836065573770493e-06,
|
10193 |
+
"loss": 0.1877,
|
10194 |
+
"step": 29100
|
10195 |
+
},
|
10196 |
+
{
|
10197 |
+
"epoch": 142.13544844417328,
|
10198 |
+
"grad_norm": 3.393789529800415,
|
10199 |
+
"learning_rate": 9.704918032786885e-06,
|
10200 |
+
"loss": 0.1928,
|
10201 |
+
"step": 29120
|
10202 |
+
},
|
10203 |
+
{
|
10204 |
+
"epoch": 142.23306894447833,
|
10205 |
+
"grad_norm": 2.592445135116577,
|
10206 |
+
"learning_rate": 9.573770491803278e-06,
|
10207 |
+
"loss": 0.1638,
|
10208 |
+
"step": 29140
|
10209 |
+
},
|
10210 |
+
{
|
10211 |
+
"epoch": 142.33068944478342,
|
10212 |
+
"grad_norm": 4.577868938446045,
|
10213 |
+
"learning_rate": 9.442622950819673e-06,
|
10214 |
+
"loss": 0.1868,
|
10215 |
+
"step": 29160
|
10216 |
+
},
|
10217 |
+
{
|
10218 |
+
"epoch": 142.42830994508847,
|
10219 |
+
"grad_norm": 3.1620144844055176,
|
10220 |
+
"learning_rate": 9.311475409836067e-06,
|
10221 |
+
"loss": 0.1635,
|
10222 |
+
"step": 29180
|
10223 |
+
},
|
10224 |
+
{
|
10225 |
+
"epoch": 142.52593044539353,
|
10226 |
+
"grad_norm": 3.382749319076538,
|
10227 |
+
"learning_rate": 9.18032786885246e-06,
|
10228 |
+
"loss": 0.1469,
|
10229 |
+
"step": 29200
|
10230 |
+
},
|
10231 |
+
{
|
10232 |
+
"epoch": 142.6235509456986,
|
10233 |
+
"grad_norm": 3.314983606338501,
|
10234 |
+
"learning_rate": 9.049180327868853e-06,
|
10235 |
+
"loss": 0.1719,
|
10236 |
+
"step": 29220
|
10237 |
+
},
|
10238 |
+
{
|
10239 |
+
"epoch": 142.72117144600367,
|
10240 |
+
"grad_norm": 2.643578290939331,
|
10241 |
+
"learning_rate": 8.918032786885247e-06,
|
10242 |
+
"loss": 0.1587,
|
10243 |
+
"step": 29240
|
10244 |
+
},
|
10245 |
+
{
|
10246 |
+
"epoch": 142.81879194630872,
|
10247 |
+
"grad_norm": 2.4660592079162598,
|
10248 |
+
"learning_rate": 8.78688524590164e-06,
|
10249 |
+
"loss": 0.2031,
|
10250 |
+
"step": 29260
|
10251 |
+
},
|
10252 |
+
{
|
10253 |
+
"epoch": 142.91641244661378,
|
10254 |
+
"grad_norm": 3.2867209911346436,
|
10255 |
+
"learning_rate": 8.655737704918034e-06,
|
10256 |
+
"loss": 0.1679,
|
10257 |
+
"step": 29280
|
10258 |
+
},
|
10259 |
+
{
|
10260 |
+
"epoch": 143.01403294691886,
|
10261 |
+
"grad_norm": 3.2089104652404785,
|
10262 |
+
"learning_rate": 8.524590163934427e-06,
|
10263 |
+
"loss": 0.171,
|
10264 |
+
"step": 29300
|
10265 |
+
},
|
10266 |
+
{
|
10267 |
+
"epoch": 143.11165344722392,
|
10268 |
+
"grad_norm": 2.9183740615844727,
|
10269 |
+
"learning_rate": 8.39344262295082e-06,
|
10270 |
+
"loss": 0.1642,
|
10271 |
+
"step": 29320
|
10272 |
+
},
|
10273 |
+
{
|
10274 |
+
"epoch": 143.20927394752897,
|
10275 |
+
"grad_norm": 2.9158482551574707,
|
10276 |
+
"learning_rate": 8.262295081967214e-06,
|
10277 |
+
"loss": 0.1743,
|
10278 |
+
"step": 29340
|
10279 |
+
},
|
10280 |
+
{
|
10281 |
+
"epoch": 143.30689444783405,
|
10282 |
+
"grad_norm": 3.256065607070923,
|
10283 |
+
"learning_rate": 8.131147540983607e-06,
|
10284 |
+
"loss": 0.1816,
|
10285 |
+
"step": 29360
|
10286 |
+
},
|
10287 |
+
{
|
10288 |
+
"epoch": 143.4045149481391,
|
10289 |
+
"grad_norm": 2.916098117828369,
|
10290 |
+
"learning_rate": 8.000000000000001e-06,
|
10291 |
+
"loss": 0.1575,
|
10292 |
+
"step": 29380
|
10293 |
+
},
|
10294 |
+
{
|
10295 |
+
"epoch": 143.50213544844416,
|
10296 |
+
"grad_norm": 3.414485454559326,
|
10297 |
+
"learning_rate": 7.868852459016394e-06,
|
10298 |
+
"loss": 0.1584,
|
10299 |
+
"step": 29400
|
10300 |
+
},
|
10301 |
+
{
|
10302 |
+
"epoch": 143.59975594874925,
|
10303 |
+
"grad_norm": 3.5028018951416016,
|
10304 |
+
"learning_rate": 7.737704918032787e-06,
|
10305 |
+
"loss": 0.1731,
|
10306 |
+
"step": 29420
|
10307 |
+
},
|
10308 |
+
{
|
10309 |
+
"epoch": 143.6973764490543,
|
10310 |
+
"grad_norm": 2.7573652267456055,
|
10311 |
+
"learning_rate": 7.606557377049181e-06,
|
10312 |
+
"loss": 0.1848,
|
10313 |
+
"step": 29440
|
10314 |
+
},
|
10315 |
+
{
|
10316 |
+
"epoch": 143.79499694935936,
|
10317 |
+
"grad_norm": 3.1906893253326416,
|
10318 |
+
"learning_rate": 7.475409836065573e-06,
|
10319 |
+
"loss": 0.1842,
|
10320 |
+
"step": 29460
|
10321 |
+
},
|
10322 |
+
{
|
10323 |
+
"epoch": 143.89261744966444,
|
10324 |
+
"grad_norm": 2.384742021560669,
|
10325 |
+
"learning_rate": 7.344262295081968e-06,
|
10326 |
+
"loss": 0.1852,
|
10327 |
+
"step": 29480
|
10328 |
+
},
|
10329 |
+
{
|
10330 |
+
"epoch": 143.9902379499695,
|
10331 |
+
"grad_norm": 3.7481918334960938,
|
10332 |
+
"learning_rate": 7.213114754098361e-06,
|
10333 |
+
"loss": 0.1794,
|
10334 |
+
"step": 29500
|
10335 |
+
},
|
10336 |
+
{
|
10337 |
+
"epoch": 144.08785845027455,
|
10338 |
+
"grad_norm": 2.903989791870117,
|
10339 |
+
"learning_rate": 7.081967213114754e-06,
|
10340 |
+
"loss": 0.1745,
|
10341 |
+
"step": 29520
|
10342 |
+
},
|
10343 |
+
{
|
10344 |
+
"epoch": 144.1854789505796,
|
10345 |
+
"grad_norm": 3.4449713230133057,
|
10346 |
+
"learning_rate": 6.9508196721311484e-06,
|
10347 |
+
"loss": 0.1874,
|
10348 |
+
"step": 29540
|
10349 |
+
},
|
10350 |
+
{
|
10351 |
+
"epoch": 144.2830994508847,
|
10352 |
+
"grad_norm": 2.9290127754211426,
|
10353 |
+
"learning_rate": 6.819672131147542e-06,
|
10354 |
+
"loss": 0.1736,
|
10355 |
+
"step": 29560
|
10356 |
+
},
|
10357 |
+
{
|
10358 |
+
"epoch": 144.38071995118975,
|
10359 |
+
"grad_norm": 2.805908203125,
|
10360 |
+
"learning_rate": 6.688524590163935e-06,
|
10361 |
+
"loss": 0.187,
|
10362 |
+
"step": 29580
|
10363 |
+
},
|
10364 |
+
{
|
10365 |
+
"epoch": 144.4783404514948,
|
10366 |
+
"grad_norm": 2.9539241790771484,
|
10367 |
+
"learning_rate": 6.557377049180328e-06,
|
10368 |
+
"loss": 0.1856,
|
10369 |
+
"step": 29600
|
10370 |
+
},
|
10371 |
+
{
|
10372 |
+
"epoch": 144.57596095179989,
|
10373 |
+
"grad_norm": 2.8198204040527344,
|
10374 |
+
"learning_rate": 6.426229508196721e-06,
|
10375 |
+
"loss": 0.1733,
|
10376 |
+
"step": 29620
|
10377 |
+
},
|
10378 |
+
{
|
10379 |
+
"epoch": 144.67358145210494,
|
10380 |
+
"grad_norm": 3.3926849365234375,
|
10381 |
+
"learning_rate": 6.295081967213115e-06,
|
10382 |
+
"loss": 0.18,
|
10383 |
+
"step": 29640
|
10384 |
+
},
|
10385 |
+
{
|
10386 |
+
"epoch": 144.77120195241,
|
10387 |
+
"grad_norm": 4.100579261779785,
|
10388 |
+
"learning_rate": 6.163934426229508e-06,
|
10389 |
+
"loss": 0.1568,
|
10390 |
+
"step": 29660
|
10391 |
+
},
|
10392 |
+
{
|
10393 |
+
"epoch": 144.86882245271508,
|
10394 |
+
"grad_norm": 3.2875492572784424,
|
10395 |
+
"learning_rate": 6.032786885245902e-06,
|
10396 |
+
"loss": 0.1561,
|
10397 |
+
"step": 29680
|
10398 |
+
},
|
10399 |
+
{
|
10400 |
+
"epoch": 144.96644295302013,
|
10401 |
+
"grad_norm": 2.626185417175293,
|
10402 |
+
"learning_rate": 5.9016393442622956e-06,
|
10403 |
+
"loss": 0.1714,
|
10404 |
+
"step": 29700
|
10405 |
+
},
|
10406 |
+
{
|
10407 |
+
"epoch": 145.0640634533252,
|
10408 |
+
"grad_norm": 4.3447265625,
|
10409 |
+
"learning_rate": 5.770491803278689e-06,
|
10410 |
+
"loss": 0.1951,
|
10411 |
+
"step": 29720
|
10412 |
+
},
|
10413 |
+
{
|
10414 |
+
"epoch": 145.16168395363027,
|
10415 |
+
"grad_norm": 4.056821346282959,
|
10416 |
+
"learning_rate": 5.639344262295082e-06,
|
10417 |
+
"loss": 0.1621,
|
10418 |
+
"step": 29740
|
10419 |
+
},
|
10420 |
+
{
|
10421 |
+
"epoch": 145.25930445393533,
|
10422 |
+
"grad_norm": 3.4116666316986084,
|
10423 |
+
"learning_rate": 5.508196721311476e-06,
|
10424 |
+
"loss": 0.179,
|
10425 |
+
"step": 29760
|
10426 |
+
},
|
10427 |
+
{
|
10428 |
+
"epoch": 145.35692495424038,
|
10429 |
+
"grad_norm": 2.810452699661255,
|
10430 |
+
"learning_rate": 5.377049180327869e-06,
|
10431 |
+
"loss": 0.1723,
|
10432 |
+
"step": 29780
|
10433 |
+
},
|
10434 |
+
{
|
10435 |
+
"epoch": 145.45454545454547,
|
10436 |
+
"grad_norm": 3.708115816116333,
|
10437 |
+
"learning_rate": 5.245901639344263e-06,
|
10438 |
+
"loss": 0.1675,
|
10439 |
+
"step": 29800
|
10440 |
+
},
|
10441 |
+
{
|
10442 |
+
"epoch": 145.55216595485052,
|
10443 |
+
"grad_norm": 4.000546455383301,
|
10444 |
+
"learning_rate": 5.1147540983606555e-06,
|
10445 |
+
"loss": 0.1716,
|
10446 |
+
"step": 29820
|
10447 |
+
},
|
10448 |
+
{
|
10449 |
+
"epoch": 145.64978645515558,
|
10450 |
+
"grad_norm": 3.2421109676361084,
|
10451 |
+
"learning_rate": 4.98360655737705e-06,
|
10452 |
+
"loss": 0.1656,
|
10453 |
+
"step": 29840
|
10454 |
+
},
|
10455 |
+
{
|
10456 |
+
"epoch": 145.74740695546063,
|
10457 |
+
"grad_norm": 3.13706111907959,
|
10458 |
+
"learning_rate": 4.852459016393443e-06,
|
10459 |
+
"loss": 0.1822,
|
10460 |
+
"step": 29860
|
10461 |
+
},
|
10462 |
+
{
|
10463 |
+
"epoch": 145.84502745576572,
|
10464 |
+
"grad_norm": 3.364842176437378,
|
10465 |
+
"learning_rate": 4.721311475409836e-06,
|
10466 |
+
"loss": 0.1772,
|
10467 |
+
"step": 29880
|
10468 |
+
},
|
10469 |
+
{
|
10470 |
+
"epoch": 145.94264795607077,
|
10471 |
+
"grad_norm": 3.2013063430786133,
|
10472 |
+
"learning_rate": 4.59016393442623e-06,
|
10473 |
+
"loss": 0.1717,
|
10474 |
+
"step": 29900
|
10475 |
+
},
|
10476 |
+
{
|
10477 |
+
"epoch": 146.04026845637583,
|
10478 |
+
"grad_norm": 4.174123287200928,
|
10479 |
+
"learning_rate": 4.4590163934426235e-06,
|
10480 |
+
"loss": 0.1731,
|
10481 |
+
"step": 29920
|
10482 |
+
},
|
10483 |
+
{
|
10484 |
+
"epoch": 146.1378889566809,
|
10485 |
+
"grad_norm": 2.8885281085968018,
|
10486 |
+
"learning_rate": 4.327868852459017e-06,
|
10487 |
+
"loss": 0.1421,
|
10488 |
+
"step": 29940
|
10489 |
+
},
|
10490 |
+
{
|
10491 |
+
"epoch": 146.23550945698597,
|
10492 |
+
"grad_norm": 2.6078240871429443,
|
10493 |
+
"learning_rate": 4.19672131147541e-06,
|
10494 |
+
"loss": 0.1898,
|
10495 |
+
"step": 29960
|
10496 |
+
},
|
10497 |
+
{
|
10498 |
+
"epoch": 146.33312995729102,
|
10499 |
+
"grad_norm": 4.43600606918335,
|
10500 |
+
"learning_rate": 4.0655737704918034e-06,
|
10501 |
+
"loss": 0.1787,
|
10502 |
+
"step": 29980
|
10503 |
+
},
|
10504 |
+
{
|
10505 |
+
"epoch": 146.4307504575961,
|
10506 |
+
"grad_norm": 2.9905123710632324,
|
10507 |
+
"learning_rate": 3.934426229508197e-06,
|
10508 |
+
"loss": 0.1866,
|
10509 |
+
"step": 30000
|
10510 |
}
|
10511 |
],
|
10512 |
"logging_steps": 20,
|
|
|
10526 |
"attributes": {}
|
10527 |
}
|
10528 |
},
|
10529 |
+
"total_flos": 3.859765296186163e+17,
|
10530 |
"train_batch_size": 1,
|
10531 |
"trial_name": null,
|
10532 |
"trial_params": null
|