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 +2103 -3
adapter_model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 3705288
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7eaf165321a6e07d20dfb4acf0b71c58ce8bd857e3963bdf551bd937c35ebe5e
|
3 |
size 3705288
|
optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 2213690
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:4fb0c49d2beff6cc27e40fe90e9eda9eb8e07a2d6b5742058d0507fc8e982a55
|
3 |
size 2213690
|
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 sha256:6d451be845325aaf5529b57ca9d6ff558484e3afd1212dcb496ff63ca3323c12
|
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:dd293d9ce0598f9847da19abf5d46d556cd79a0ffc0ab198425114ba155d0d8b
|
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,
|
@@ -3157,6 +3157,2106 @@
|
|
3157 |
"learning_rate": 0.0001123152709359606,
|
3158 |
"loss": 0.3774,
|
3159 |
"step": 9000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
3160 |
}
|
3161 |
],
|
3162 |
"logging_steps": 20,
|
@@ -3176,7 +5276,7 @@
|
|
3176 |
"attributes": {}
|
3177 |
}
|
3178 |
},
|
3179 |
-
"total_flos":
|
3180 |
"train_batch_size": 1,
|
3181 |
"trial_name": null,
|
3182 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 73.21537522879805,
|
5 |
"eval_steps": 500,
|
6 |
+
"global_step": 15000,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
3157 |
"learning_rate": 0.0001123152709359606,
|
3158 |
"loss": 0.3774,
|
3159 |
"step": 9000
|
3160 |
+
},
|
3161 |
+
{
|
3162 |
+
"epoch": 44.02684563758389,
|
3163 |
+
"grad_norm": 3.0228703022003174,
|
3164 |
+
"learning_rate": 0.00011211822660098524,
|
3165 |
+
"loss": 0.3479,
|
3166 |
+
"step": 9020
|
3167 |
+
},
|
3168 |
+
{
|
3169 |
+
"epoch": 44.12446613788896,
|
3170 |
+
"grad_norm": 4.042842864990234,
|
3171 |
+
"learning_rate": 0.00011192118226600985,
|
3172 |
+
"loss": 0.3567,
|
3173 |
+
"step": 9040
|
3174 |
+
},
|
3175 |
+
{
|
3176 |
+
"epoch": 44.22208663819402,
|
3177 |
+
"grad_norm": 2.5165748596191406,
|
3178 |
+
"learning_rate": 0.00011172413793103449,
|
3179 |
+
"loss": 0.357,
|
3180 |
+
"step": 9060
|
3181 |
+
},
|
3182 |
+
{
|
3183 |
+
"epoch": 44.31970713849908,
|
3184 |
+
"grad_norm": 2.9104301929473877,
|
3185 |
+
"learning_rate": 0.00011152709359605913,
|
3186 |
+
"loss": 0.3478,
|
3187 |
+
"step": 9080
|
3188 |
+
},
|
3189 |
+
{
|
3190 |
+
"epoch": 44.417327638804146,
|
3191 |
+
"grad_norm": 5.000180244445801,
|
3192 |
+
"learning_rate": 0.00011133004926108374,
|
3193 |
+
"loss": 0.3372,
|
3194 |
+
"step": 9100
|
3195 |
+
},
|
3196 |
+
{
|
3197 |
+
"epoch": 44.514948139109215,
|
3198 |
+
"grad_norm": 2.7573766708374023,
|
3199 |
+
"learning_rate": 0.00011113300492610838,
|
3200 |
+
"loss": 0.3574,
|
3201 |
+
"step": 9120
|
3202 |
+
},
|
3203 |
+
{
|
3204 |
+
"epoch": 44.61256863941428,
|
3205 |
+
"grad_norm": 3.473818778991699,
|
3206 |
+
"learning_rate": 0.000110935960591133,
|
3207 |
+
"loss": 0.3666,
|
3208 |
+
"step": 9140
|
3209 |
+
},
|
3210 |
+
{
|
3211 |
+
"epoch": 44.71018913971934,
|
3212 |
+
"grad_norm": 4.236100196838379,
|
3213 |
+
"learning_rate": 0.00011073891625615764,
|
3214 |
+
"loss": 0.3612,
|
3215 |
+
"step": 9160
|
3216 |
+
},
|
3217 |
+
{
|
3218 |
+
"epoch": 44.8078096400244,
|
3219 |
+
"grad_norm": 5.279041290283203,
|
3220 |
+
"learning_rate": 0.00011054187192118227,
|
3221 |
+
"loss": 0.3694,
|
3222 |
+
"step": 9180
|
3223 |
+
},
|
3224 |
+
{
|
3225 |
+
"epoch": 44.90543014032947,
|
3226 |
+
"grad_norm": 3.0009076595306396,
|
3227 |
+
"learning_rate": 0.0001103448275862069,
|
3228 |
+
"loss": 0.3629,
|
3229 |
+
"step": 9200
|
3230 |
+
},
|
3231 |
+
{
|
3232 |
+
"epoch": 45.003050640634534,
|
3233 |
+
"grad_norm": 3.358452796936035,
|
3234 |
+
"learning_rate": 0.00011014778325123153,
|
3235 |
+
"loss": 0.3584,
|
3236 |
+
"step": 9220
|
3237 |
+
},
|
3238 |
+
{
|
3239 |
+
"epoch": 45.100671140939596,
|
3240 |
+
"grad_norm": 2.9341399669647217,
|
3241 |
+
"learning_rate": 0.00010995073891625616,
|
3242 |
+
"loss": 0.3437,
|
3243 |
+
"step": 9240
|
3244 |
+
},
|
3245 |
+
{
|
3246 |
+
"epoch": 45.19829164124466,
|
3247 |
+
"grad_norm": 3.1249337196350098,
|
3248 |
+
"learning_rate": 0.0001097536945812808,
|
3249 |
+
"loss": 0.3551,
|
3250 |
+
"step": 9260
|
3251 |
+
},
|
3252 |
+
{
|
3253 |
+
"epoch": 45.29591214154973,
|
3254 |
+
"grad_norm": 2.4878969192504883,
|
3255 |
+
"learning_rate": 0.00010955665024630541,
|
3256 |
+
"loss": 0.3379,
|
3257 |
+
"step": 9280
|
3258 |
+
},
|
3259 |
+
{
|
3260 |
+
"epoch": 45.39353264185479,
|
3261 |
+
"grad_norm": 3.114165782928467,
|
3262 |
+
"learning_rate": 0.00010935960591133006,
|
3263 |
+
"loss": 0.3616,
|
3264 |
+
"step": 9300
|
3265 |
+
},
|
3266 |
+
{
|
3267 |
+
"epoch": 45.49115314215985,
|
3268 |
+
"grad_norm": 3.0727782249450684,
|
3269 |
+
"learning_rate": 0.0001091625615763547,
|
3270 |
+
"loss": 0.348,
|
3271 |
+
"step": 9320
|
3272 |
+
},
|
3273 |
+
{
|
3274 |
+
"epoch": 45.588773642464915,
|
3275 |
+
"grad_norm": 2.9487972259521484,
|
3276 |
+
"learning_rate": 0.00010896551724137931,
|
3277 |
+
"loss": 0.3397,
|
3278 |
+
"step": 9340
|
3279 |
+
},
|
3280 |
+
{
|
3281 |
+
"epoch": 45.686394142769984,
|
3282 |
+
"grad_norm": 3.0654473304748535,
|
3283 |
+
"learning_rate": 0.00010876847290640395,
|
3284 |
+
"loss": 0.3515,
|
3285 |
+
"step": 9360
|
3286 |
+
},
|
3287 |
+
{
|
3288 |
+
"epoch": 45.78401464307505,
|
3289 |
+
"grad_norm": 4.303600311279297,
|
3290 |
+
"learning_rate": 0.00010857142857142856,
|
3291 |
+
"loss": 0.3586,
|
3292 |
+
"step": 9380
|
3293 |
+
},
|
3294 |
+
{
|
3295 |
+
"epoch": 45.88163514338011,
|
3296 |
+
"grad_norm": 2.946246385574341,
|
3297 |
+
"learning_rate": 0.0001083743842364532,
|
3298 |
+
"loss": 0.3436,
|
3299 |
+
"step": 9400
|
3300 |
+
},
|
3301 |
+
{
|
3302 |
+
"epoch": 45.97925564368517,
|
3303 |
+
"grad_norm": 2.4360456466674805,
|
3304 |
+
"learning_rate": 0.00010817733990147785,
|
3305 |
+
"loss": 0.3766,
|
3306 |
+
"step": 9420
|
3307 |
+
},
|
3308 |
+
{
|
3309 |
+
"epoch": 46.07687614399024,
|
3310 |
+
"grad_norm": 2.8351433277130127,
|
3311 |
+
"learning_rate": 0.00010798029556650246,
|
3312 |
+
"loss": 0.3547,
|
3313 |
+
"step": 9440
|
3314 |
+
},
|
3315 |
+
{
|
3316 |
+
"epoch": 46.1744966442953,
|
3317 |
+
"grad_norm": 2.6005990505218506,
|
3318 |
+
"learning_rate": 0.0001077832512315271,
|
3319 |
+
"loss": 0.3333,
|
3320 |
+
"step": 9460
|
3321 |
+
},
|
3322 |
+
{
|
3323 |
+
"epoch": 46.272117144600365,
|
3324 |
+
"grad_norm": 2.52091121673584,
|
3325 |
+
"learning_rate": 0.00010758620689655173,
|
3326 |
+
"loss": 0.3507,
|
3327 |
+
"step": 9480
|
3328 |
+
},
|
3329 |
+
{
|
3330 |
+
"epoch": 46.36973764490543,
|
3331 |
+
"grad_norm": 3.0750203132629395,
|
3332 |
+
"learning_rate": 0.00010738916256157637,
|
3333 |
+
"loss": 0.3376,
|
3334 |
+
"step": 9500
|
3335 |
+
},
|
3336 |
+
{
|
3337 |
+
"epoch": 46.4673581452105,
|
3338 |
+
"grad_norm": 3.353597640991211,
|
3339 |
+
"learning_rate": 0.00010719211822660098,
|
3340 |
+
"loss": 0.3362,
|
3341 |
+
"step": 9520
|
3342 |
+
},
|
3343 |
+
{
|
3344 |
+
"epoch": 46.56497864551556,
|
3345 |
+
"grad_norm": 3.786407232284546,
|
3346 |
+
"learning_rate": 0.00010699507389162562,
|
3347 |
+
"loss": 0.3774,
|
3348 |
+
"step": 9540
|
3349 |
+
},
|
3350 |
+
{
|
3351 |
+
"epoch": 46.66259914582062,
|
3352 |
+
"grad_norm": 3.2476627826690674,
|
3353 |
+
"learning_rate": 0.00010679802955665026,
|
3354 |
+
"loss": 0.3423,
|
3355 |
+
"step": 9560
|
3356 |
+
},
|
3357 |
+
{
|
3358 |
+
"epoch": 46.760219646125684,
|
3359 |
+
"grad_norm": 2.966078281402588,
|
3360 |
+
"learning_rate": 0.00010660098522167488,
|
3361 |
+
"loss": 0.3382,
|
3362 |
+
"step": 9580
|
3363 |
+
},
|
3364 |
+
{
|
3365 |
+
"epoch": 46.85784014643075,
|
3366 |
+
"grad_norm": 3.7173826694488525,
|
3367 |
+
"learning_rate": 0.00010640394088669952,
|
3368 |
+
"loss": 0.3512,
|
3369 |
+
"step": 9600
|
3370 |
+
},
|
3371 |
+
{
|
3372 |
+
"epoch": 46.955460646735816,
|
3373 |
+
"grad_norm": 3.6152524948120117,
|
3374 |
+
"learning_rate": 0.00010620689655172413,
|
3375 |
+
"loss": 0.3499,
|
3376 |
+
"step": 9620
|
3377 |
+
},
|
3378 |
+
{
|
3379 |
+
"epoch": 47.05308114704088,
|
3380 |
+
"grad_norm": 3.6383986473083496,
|
3381 |
+
"learning_rate": 0.00010600985221674877,
|
3382 |
+
"loss": 0.3442,
|
3383 |
+
"step": 9640
|
3384 |
+
},
|
3385 |
+
{
|
3386 |
+
"epoch": 47.15070164734594,
|
3387 |
+
"grad_norm": 2.636918306350708,
|
3388 |
+
"learning_rate": 0.00010581280788177341,
|
3389 |
+
"loss": 0.3355,
|
3390 |
+
"step": 9660
|
3391 |
+
},
|
3392 |
+
{
|
3393 |
+
"epoch": 47.24832214765101,
|
3394 |
+
"grad_norm": 3.8844096660614014,
|
3395 |
+
"learning_rate": 0.00010561576354679802,
|
3396 |
+
"loss": 0.3389,
|
3397 |
+
"step": 9680
|
3398 |
+
},
|
3399 |
+
{
|
3400 |
+
"epoch": 47.34594264795607,
|
3401 |
+
"grad_norm": 4.149389743804932,
|
3402 |
+
"learning_rate": 0.00010541871921182267,
|
3403 |
+
"loss": 0.3168,
|
3404 |
+
"step": 9700
|
3405 |
+
},
|
3406 |
+
{
|
3407 |
+
"epoch": 47.443563148261134,
|
3408 |
+
"grad_norm": 3.205845832824707,
|
3409 |
+
"learning_rate": 0.00010522167487684729,
|
3410 |
+
"loss": 0.3247,
|
3411 |
+
"step": 9720
|
3412 |
+
},
|
3413 |
+
{
|
3414 |
+
"epoch": 47.5411836485662,
|
3415 |
+
"grad_norm": 3.4177889823913574,
|
3416 |
+
"learning_rate": 0.00010502463054187193,
|
3417 |
+
"loss": 0.3472,
|
3418 |
+
"step": 9740
|
3419 |
+
},
|
3420 |
+
{
|
3421 |
+
"epoch": 47.638804148871266,
|
3422 |
+
"grad_norm": 3.2508625984191895,
|
3423 |
+
"learning_rate": 0.00010482758620689656,
|
3424 |
+
"loss": 0.3354,
|
3425 |
+
"step": 9760
|
3426 |
+
},
|
3427 |
+
{
|
3428 |
+
"epoch": 47.73642464917633,
|
3429 |
+
"grad_norm": 3.2071492671966553,
|
3430 |
+
"learning_rate": 0.00010463054187192119,
|
3431 |
+
"loss": 0.3515,
|
3432 |
+
"step": 9780
|
3433 |
+
},
|
3434 |
+
{
|
3435 |
+
"epoch": 47.83404514948139,
|
3436 |
+
"grad_norm": 2.505859613418579,
|
3437 |
+
"learning_rate": 0.00010443349753694583,
|
3438 |
+
"loss": 0.3654,
|
3439 |
+
"step": 9800
|
3440 |
+
},
|
3441 |
+
{
|
3442 |
+
"epoch": 47.93166564978645,
|
3443 |
+
"grad_norm": 3.092602491378784,
|
3444 |
+
"learning_rate": 0.00010423645320197044,
|
3445 |
+
"loss": 0.3551,
|
3446 |
+
"step": 9820
|
3447 |
+
},
|
3448 |
+
{
|
3449 |
+
"epoch": 48.02928615009152,
|
3450 |
+
"grad_norm": 3.411740303039551,
|
3451 |
+
"learning_rate": 0.00010403940886699508,
|
3452 |
+
"loss": 0.3445,
|
3453 |
+
"step": 9840
|
3454 |
+
},
|
3455 |
+
{
|
3456 |
+
"epoch": 48.126906650396585,
|
3457 |
+
"grad_norm": 2.587663412094116,
|
3458 |
+
"learning_rate": 0.00010384236453201972,
|
3459 |
+
"loss": 0.3132,
|
3460 |
+
"step": 9860
|
3461 |
+
},
|
3462 |
+
{
|
3463 |
+
"epoch": 48.22452715070165,
|
3464 |
+
"grad_norm": 2.244938850402832,
|
3465 |
+
"learning_rate": 0.00010364532019704434,
|
3466 |
+
"loss": 0.3327,
|
3467 |
+
"step": 9880
|
3468 |
+
},
|
3469 |
+
{
|
3470 |
+
"epoch": 48.32214765100671,
|
3471 |
+
"grad_norm": 3.426699638366699,
|
3472 |
+
"learning_rate": 0.00010344827586206898,
|
3473 |
+
"loss": 0.3163,
|
3474 |
+
"step": 9900
|
3475 |
+
},
|
3476 |
+
{
|
3477 |
+
"epoch": 48.41976815131178,
|
3478 |
+
"grad_norm": 2.600964069366455,
|
3479 |
+
"learning_rate": 0.00010325123152709359,
|
3480 |
+
"loss": 0.3318,
|
3481 |
+
"step": 9920
|
3482 |
+
},
|
3483 |
+
{
|
3484 |
+
"epoch": 48.51738865161684,
|
3485 |
+
"grad_norm": 2.5745320320129395,
|
3486 |
+
"learning_rate": 0.00010305418719211823,
|
3487 |
+
"loss": 0.3302,
|
3488 |
+
"step": 9940
|
3489 |
+
},
|
3490 |
+
{
|
3491 |
+
"epoch": 48.6150091519219,
|
3492 |
+
"grad_norm": 2.9485421180725098,
|
3493 |
+
"learning_rate": 0.00010285714285714286,
|
3494 |
+
"loss": 0.3468,
|
3495 |
+
"step": 9960
|
3496 |
+
},
|
3497 |
+
{
|
3498 |
+
"epoch": 48.712629652226966,
|
3499 |
+
"grad_norm": 2.783953905105591,
|
3500 |
+
"learning_rate": 0.00010266009852216748,
|
3501 |
+
"loss": 0.3339,
|
3502 |
+
"step": 9980
|
3503 |
+
},
|
3504 |
+
{
|
3505 |
+
"epoch": 48.810250152532035,
|
3506 |
+
"grad_norm": 3.2114439010620117,
|
3507 |
+
"learning_rate": 0.00010246305418719213,
|
3508 |
+
"loss": 0.3496,
|
3509 |
+
"step": 10000
|
3510 |
+
},
|
3511 |
+
{
|
3512 |
+
"epoch": 48.9078706528371,
|
3513 |
+
"grad_norm": 4.33662748336792,
|
3514 |
+
"learning_rate": 0.00010226600985221675,
|
3515 |
+
"loss": 0.3358,
|
3516 |
+
"step": 10020
|
3517 |
+
},
|
3518 |
+
{
|
3519 |
+
"epoch": 49.00549115314216,
|
3520 |
+
"grad_norm": 2.714755058288574,
|
3521 |
+
"learning_rate": 0.0001020689655172414,
|
3522 |
+
"loss": 0.3677,
|
3523 |
+
"step": 10040
|
3524 |
+
},
|
3525 |
+
{
|
3526 |
+
"epoch": 49.10311165344722,
|
3527 |
+
"grad_norm": 2.1904876232147217,
|
3528 |
+
"learning_rate": 0.00010187192118226601,
|
3529 |
+
"loss": 0.2878,
|
3530 |
+
"step": 10060
|
3531 |
+
},
|
3532 |
+
{
|
3533 |
+
"epoch": 49.20073215375229,
|
3534 |
+
"grad_norm": 2.530484676361084,
|
3535 |
+
"learning_rate": 0.00010167487684729065,
|
3536 |
+
"loss": 0.3221,
|
3537 |
+
"step": 10080
|
3538 |
+
},
|
3539 |
+
{
|
3540 |
+
"epoch": 49.298352654057354,
|
3541 |
+
"grad_norm": 3.1762654781341553,
|
3542 |
+
"learning_rate": 0.00010147783251231529,
|
3543 |
+
"loss": 0.3427,
|
3544 |
+
"step": 10100
|
3545 |
+
},
|
3546 |
+
{
|
3547 |
+
"epoch": 49.395973154362416,
|
3548 |
+
"grad_norm": 3.0370638370513916,
|
3549 |
+
"learning_rate": 0.0001012807881773399,
|
3550 |
+
"loss": 0.3466,
|
3551 |
+
"step": 10120
|
3552 |
+
},
|
3553 |
+
{
|
3554 |
+
"epoch": 49.49359365466748,
|
3555 |
+
"grad_norm": 2.5626463890075684,
|
3556 |
+
"learning_rate": 0.00010108374384236454,
|
3557 |
+
"loss": 0.3218,
|
3558 |
+
"step": 10140
|
3559 |
+
},
|
3560 |
+
{
|
3561 |
+
"epoch": 49.59121415497255,
|
3562 |
+
"grad_norm": 3.4357545375823975,
|
3563 |
+
"learning_rate": 0.00010088669950738916,
|
3564 |
+
"loss": 0.3312,
|
3565 |
+
"step": 10160
|
3566 |
+
},
|
3567 |
+
{
|
3568 |
+
"epoch": 49.68883465527761,
|
3569 |
+
"grad_norm": 2.810955762863159,
|
3570 |
+
"learning_rate": 0.0001006896551724138,
|
3571 |
+
"loss": 0.3363,
|
3572 |
+
"step": 10180
|
3573 |
+
},
|
3574 |
+
{
|
3575 |
+
"epoch": 49.78645515558267,
|
3576 |
+
"grad_norm": 3.8722000122070312,
|
3577 |
+
"learning_rate": 0.00010049261083743844,
|
3578 |
+
"loss": 0.3251,
|
3579 |
+
"step": 10200
|
3580 |
+
},
|
3581 |
+
{
|
3582 |
+
"epoch": 49.884075655887735,
|
3583 |
+
"grad_norm": 3.185521364212036,
|
3584 |
+
"learning_rate": 0.00010029556650246305,
|
3585 |
+
"loss": 0.3429,
|
3586 |
+
"step": 10220
|
3587 |
+
},
|
3588 |
+
{
|
3589 |
+
"epoch": 49.981696156192804,
|
3590 |
+
"grad_norm": 2.707853078842163,
|
3591 |
+
"learning_rate": 0.00010009852216748769,
|
3592 |
+
"loss": 0.3548,
|
3593 |
+
"step": 10240
|
3594 |
+
},
|
3595 |
+
{
|
3596 |
+
"epoch": 50.079316656497866,
|
3597 |
+
"grad_norm": 2.749464511871338,
|
3598 |
+
"learning_rate": 9.990147783251232e-05,
|
3599 |
+
"loss": 0.3294,
|
3600 |
+
"step": 10260
|
3601 |
+
},
|
3602 |
+
{
|
3603 |
+
"epoch": 50.17693715680293,
|
3604 |
+
"grad_norm": 3.4640865325927734,
|
3605 |
+
"learning_rate": 9.970443349753696e-05,
|
3606 |
+
"loss": 0.3204,
|
3607 |
+
"step": 10280
|
3608 |
+
},
|
3609 |
+
{
|
3610 |
+
"epoch": 50.27455765710799,
|
3611 |
+
"grad_norm": 3.4412505626678467,
|
3612 |
+
"learning_rate": 9.950738916256159e-05,
|
3613 |
+
"loss": 0.3316,
|
3614 |
+
"step": 10300
|
3615 |
+
},
|
3616 |
+
{
|
3617 |
+
"epoch": 50.37217815741306,
|
3618 |
+
"grad_norm": 4.671158790588379,
|
3619 |
+
"learning_rate": 9.931034482758621e-05,
|
3620 |
+
"loss": 0.3092,
|
3621 |
+
"step": 10320
|
3622 |
+
},
|
3623 |
+
{
|
3624 |
+
"epoch": 50.46979865771812,
|
3625 |
+
"grad_norm": 2.812875986099243,
|
3626 |
+
"learning_rate": 9.911330049261084e-05,
|
3627 |
+
"loss": 0.3217,
|
3628 |
+
"step": 10340
|
3629 |
+
},
|
3630 |
+
{
|
3631 |
+
"epoch": 50.567419158023185,
|
3632 |
+
"grad_norm": 2.600764513015747,
|
3633 |
+
"learning_rate": 9.891625615763547e-05,
|
3634 |
+
"loss": 0.3525,
|
3635 |
+
"step": 10360
|
3636 |
+
},
|
3637 |
+
{
|
3638 |
+
"epoch": 50.66503965832825,
|
3639 |
+
"grad_norm": 2.8875558376312256,
|
3640 |
+
"learning_rate": 9.871921182266011e-05,
|
3641 |
+
"loss": 0.3267,
|
3642 |
+
"step": 10380
|
3643 |
+
},
|
3644 |
+
{
|
3645 |
+
"epoch": 50.76266015863331,
|
3646 |
+
"grad_norm": 2.479055643081665,
|
3647 |
+
"learning_rate": 9.852216748768474e-05,
|
3648 |
+
"loss": 0.3283,
|
3649 |
+
"step": 10400
|
3650 |
+
},
|
3651 |
+
{
|
3652 |
+
"epoch": 50.86028065893838,
|
3653 |
+
"grad_norm": 3.4580044746398926,
|
3654 |
+
"learning_rate": 9.832512315270936e-05,
|
3655 |
+
"loss": 0.3388,
|
3656 |
+
"step": 10420
|
3657 |
+
},
|
3658 |
+
{
|
3659 |
+
"epoch": 50.95790115924344,
|
3660 |
+
"grad_norm": 2.68265962600708,
|
3661 |
+
"learning_rate": 9.812807881773399e-05,
|
3662 |
+
"loss": 0.3309,
|
3663 |
+
"step": 10440
|
3664 |
+
},
|
3665 |
+
{
|
3666 |
+
"epoch": 51.0555216595485,
|
3667 |
+
"grad_norm": 2.545677661895752,
|
3668 |
+
"learning_rate": 9.793103448275862e-05,
|
3669 |
+
"loss": 0.3221,
|
3670 |
+
"step": 10460
|
3671 |
+
},
|
3672 |
+
{
|
3673 |
+
"epoch": 51.153142159853566,
|
3674 |
+
"grad_norm": 2.899627685546875,
|
3675 |
+
"learning_rate": 9.773399014778326e-05,
|
3676 |
+
"loss": 0.3084,
|
3677 |
+
"step": 10480
|
3678 |
+
},
|
3679 |
+
{
|
3680 |
+
"epoch": 51.250762660158635,
|
3681 |
+
"grad_norm": 2.948960781097412,
|
3682 |
+
"learning_rate": 9.753694581280788e-05,
|
3683 |
+
"loss": 0.3273,
|
3684 |
+
"step": 10500
|
3685 |
+
},
|
3686 |
+
{
|
3687 |
+
"epoch": 51.3483831604637,
|
3688 |
+
"grad_norm": 2.9379513263702393,
|
3689 |
+
"learning_rate": 9.733990147783252e-05,
|
3690 |
+
"loss": 0.3315,
|
3691 |
+
"step": 10520
|
3692 |
+
},
|
3693 |
+
{
|
3694 |
+
"epoch": 51.44600366076876,
|
3695 |
+
"grad_norm": 2.543419599533081,
|
3696 |
+
"learning_rate": 9.714285714285715e-05,
|
3697 |
+
"loss": 0.3258,
|
3698 |
+
"step": 10540
|
3699 |
+
},
|
3700 |
+
{
|
3701 |
+
"epoch": 51.54362416107382,
|
3702 |
+
"grad_norm": 2.7236459255218506,
|
3703 |
+
"learning_rate": 9.694581280788178e-05,
|
3704 |
+
"loss": 0.3129,
|
3705 |
+
"step": 10560
|
3706 |
+
},
|
3707 |
+
{
|
3708 |
+
"epoch": 51.64124466137889,
|
3709 |
+
"grad_norm": 3.11745548248291,
|
3710 |
+
"learning_rate": 9.67487684729064e-05,
|
3711 |
+
"loss": 0.3038,
|
3712 |
+
"step": 10580
|
3713 |
+
},
|
3714 |
+
{
|
3715 |
+
"epoch": 51.738865161683954,
|
3716 |
+
"grad_norm": 3.6259920597076416,
|
3717 |
+
"learning_rate": 9.655172413793105e-05,
|
3718 |
+
"loss": 0.3269,
|
3719 |
+
"step": 10600
|
3720 |
+
},
|
3721 |
+
{
|
3722 |
+
"epoch": 51.836485661989016,
|
3723 |
+
"grad_norm": 3.4961044788360596,
|
3724 |
+
"learning_rate": 9.635467980295567e-05,
|
3725 |
+
"loss": 0.336,
|
3726 |
+
"step": 10620
|
3727 |
+
},
|
3728 |
+
{
|
3729 |
+
"epoch": 51.93410616229408,
|
3730 |
+
"grad_norm": 3.01009202003479,
|
3731 |
+
"learning_rate": 9.61576354679803e-05,
|
3732 |
+
"loss": 0.3297,
|
3733 |
+
"step": 10640
|
3734 |
+
},
|
3735 |
+
{
|
3736 |
+
"epoch": 52.03172666259915,
|
3737 |
+
"grad_norm": 3.047903060913086,
|
3738 |
+
"learning_rate": 9.596059113300493e-05,
|
3739 |
+
"loss": 0.3295,
|
3740 |
+
"step": 10660
|
3741 |
+
},
|
3742 |
+
{
|
3743 |
+
"epoch": 52.12934716290421,
|
3744 |
+
"grad_norm": 2.8521170616149902,
|
3745 |
+
"learning_rate": 9.576354679802955e-05,
|
3746 |
+
"loss": 0.2952,
|
3747 |
+
"step": 10680
|
3748 |
+
},
|
3749 |
+
{
|
3750 |
+
"epoch": 52.22696766320927,
|
3751 |
+
"grad_norm": 2.8909034729003906,
|
3752 |
+
"learning_rate": 9.55665024630542e-05,
|
3753 |
+
"loss": 0.3128,
|
3754 |
+
"step": 10700
|
3755 |
+
},
|
3756 |
+
{
|
3757 |
+
"epoch": 52.324588163514335,
|
3758 |
+
"grad_norm": 3.2134296894073486,
|
3759 |
+
"learning_rate": 9.536945812807882e-05,
|
3760 |
+
"loss": 0.3175,
|
3761 |
+
"step": 10720
|
3762 |
+
},
|
3763 |
+
{
|
3764 |
+
"epoch": 52.422208663819404,
|
3765 |
+
"grad_norm": 3.113543748855591,
|
3766 |
+
"learning_rate": 9.517241379310345e-05,
|
3767 |
+
"loss": 0.3305,
|
3768 |
+
"step": 10740
|
3769 |
+
},
|
3770 |
+
{
|
3771 |
+
"epoch": 52.51982916412447,
|
3772 |
+
"grad_norm": 2.3091633319854736,
|
3773 |
+
"learning_rate": 9.497536945812808e-05,
|
3774 |
+
"loss": 0.3032,
|
3775 |
+
"step": 10760
|
3776 |
+
},
|
3777 |
+
{
|
3778 |
+
"epoch": 52.61744966442953,
|
3779 |
+
"grad_norm": 2.7626681327819824,
|
3780 |
+
"learning_rate": 9.477832512315272e-05,
|
3781 |
+
"loss": 0.3071,
|
3782 |
+
"step": 10780
|
3783 |
+
},
|
3784 |
+
{
|
3785 |
+
"epoch": 52.71507016473459,
|
3786 |
+
"grad_norm": 2.6978394985198975,
|
3787 |
+
"learning_rate": 9.458128078817734e-05,
|
3788 |
+
"loss": 0.3424,
|
3789 |
+
"step": 10800
|
3790 |
+
},
|
3791 |
+
{
|
3792 |
+
"epoch": 52.81269066503966,
|
3793 |
+
"grad_norm": 4.549131393432617,
|
3794 |
+
"learning_rate": 9.438423645320199e-05,
|
3795 |
+
"loss": 0.3086,
|
3796 |
+
"step": 10820
|
3797 |
+
},
|
3798 |
+
{
|
3799 |
+
"epoch": 52.91031116534472,
|
3800 |
+
"grad_norm": 3.3548974990844727,
|
3801 |
+
"learning_rate": 9.418719211822661e-05,
|
3802 |
+
"loss": 0.3414,
|
3803 |
+
"step": 10840
|
3804 |
+
},
|
3805 |
+
{
|
3806 |
+
"epoch": 53.007931665649785,
|
3807 |
+
"grad_norm": 2.191990852355957,
|
3808 |
+
"learning_rate": 9.399014778325124e-05,
|
3809 |
+
"loss": 0.3195,
|
3810 |
+
"step": 10860
|
3811 |
+
},
|
3812 |
+
{
|
3813 |
+
"epoch": 53.10555216595485,
|
3814 |
+
"grad_norm": 2.8169941902160645,
|
3815 |
+
"learning_rate": 9.379310344827587e-05,
|
3816 |
+
"loss": 0.2971,
|
3817 |
+
"step": 10880
|
3818 |
+
},
|
3819 |
+
{
|
3820 |
+
"epoch": 53.20317266625992,
|
3821 |
+
"grad_norm": 2.4809463024139404,
|
3822 |
+
"learning_rate": 9.35960591133005e-05,
|
3823 |
+
"loss": 0.3032,
|
3824 |
+
"step": 10900
|
3825 |
+
},
|
3826 |
+
{
|
3827 |
+
"epoch": 53.30079316656498,
|
3828 |
+
"grad_norm": 2.8981711864471436,
|
3829 |
+
"learning_rate": 9.339901477832512e-05,
|
3830 |
+
"loss": 0.3139,
|
3831 |
+
"step": 10920
|
3832 |
+
},
|
3833 |
+
{
|
3834 |
+
"epoch": 53.39841366687004,
|
3835 |
+
"grad_norm": 2.901442050933838,
|
3836 |
+
"learning_rate": 9.320197044334976e-05,
|
3837 |
+
"loss": 0.3197,
|
3838 |
+
"step": 10940
|
3839 |
+
},
|
3840 |
+
{
|
3841 |
+
"epoch": 53.496034167175104,
|
3842 |
+
"grad_norm": 3.1128933429718018,
|
3843 |
+
"learning_rate": 9.300492610837439e-05,
|
3844 |
+
"loss": 0.3109,
|
3845 |
+
"step": 10960
|
3846 |
+
},
|
3847 |
+
{
|
3848 |
+
"epoch": 53.59365466748017,
|
3849 |
+
"grad_norm": 2.6892173290252686,
|
3850 |
+
"learning_rate": 9.280788177339902e-05,
|
3851 |
+
"loss": 0.3153,
|
3852 |
+
"step": 10980
|
3853 |
+
},
|
3854 |
+
{
|
3855 |
+
"epoch": 53.691275167785236,
|
3856 |
+
"grad_norm": 3.1847739219665527,
|
3857 |
+
"learning_rate": 9.261083743842364e-05,
|
3858 |
+
"loss": 0.3135,
|
3859 |
+
"step": 11000
|
3860 |
+
},
|
3861 |
+
{
|
3862 |
+
"epoch": 53.7888956680903,
|
3863 |
+
"grad_norm": 3.1111955642700195,
|
3864 |
+
"learning_rate": 9.241379310344827e-05,
|
3865 |
+
"loss": 0.3472,
|
3866 |
+
"step": 11020
|
3867 |
+
},
|
3868 |
+
{
|
3869 |
+
"epoch": 53.88651616839536,
|
3870 |
+
"grad_norm": 2.667539119720459,
|
3871 |
+
"learning_rate": 9.221674876847291e-05,
|
3872 |
+
"loss": 0.3107,
|
3873 |
+
"step": 11040
|
3874 |
+
},
|
3875 |
+
{
|
3876 |
+
"epoch": 53.98413666870043,
|
3877 |
+
"grad_norm": 2.1500725746154785,
|
3878 |
+
"learning_rate": 9.201970443349755e-05,
|
3879 |
+
"loss": 0.3192,
|
3880 |
+
"step": 11060
|
3881 |
+
},
|
3882 |
+
{
|
3883 |
+
"epoch": 54.08175716900549,
|
3884 |
+
"grad_norm": 3.6513638496398926,
|
3885 |
+
"learning_rate": 9.182266009852218e-05,
|
3886 |
+
"loss": 0.2974,
|
3887 |
+
"step": 11080
|
3888 |
+
},
|
3889 |
+
{
|
3890 |
+
"epoch": 54.179377669310554,
|
3891 |
+
"grad_norm": 3.226287364959717,
|
3892 |
+
"learning_rate": 9.16256157635468e-05,
|
3893 |
+
"loss": 0.3216,
|
3894 |
+
"step": 11100
|
3895 |
+
},
|
3896 |
+
{
|
3897 |
+
"epoch": 54.27699816961562,
|
3898 |
+
"grad_norm": 3.4577550888061523,
|
3899 |
+
"learning_rate": 9.142857142857143e-05,
|
3900 |
+
"loss": 0.2999,
|
3901 |
+
"step": 11120
|
3902 |
+
},
|
3903 |
+
{
|
3904 |
+
"epoch": 54.374618669920686,
|
3905 |
+
"grad_norm": 2.047478199005127,
|
3906 |
+
"learning_rate": 9.123152709359606e-05,
|
3907 |
+
"loss": 0.3139,
|
3908 |
+
"step": 11140
|
3909 |
+
},
|
3910 |
+
{
|
3911 |
+
"epoch": 54.47223917022575,
|
3912 |
+
"grad_norm": 3.0338408946990967,
|
3913 |
+
"learning_rate": 9.10344827586207e-05,
|
3914 |
+
"loss": 0.2954,
|
3915 |
+
"step": 11160
|
3916 |
+
},
|
3917 |
+
{
|
3918 |
+
"epoch": 54.56985967053081,
|
3919 |
+
"grad_norm": 2.6099050045013428,
|
3920 |
+
"learning_rate": 9.083743842364533e-05,
|
3921 |
+
"loss": 0.3218,
|
3922 |
+
"step": 11180
|
3923 |
+
},
|
3924 |
+
{
|
3925 |
+
"epoch": 54.66748017083587,
|
3926 |
+
"grad_norm": 3.248973846435547,
|
3927 |
+
"learning_rate": 9.064039408866995e-05,
|
3928 |
+
"loss": 0.3243,
|
3929 |
+
"step": 11200
|
3930 |
+
},
|
3931 |
+
{
|
3932 |
+
"epoch": 54.76510067114094,
|
3933 |
+
"grad_norm": 4.767118453979492,
|
3934 |
+
"learning_rate": 9.044334975369458e-05,
|
3935 |
+
"loss": 0.315,
|
3936 |
+
"step": 11220
|
3937 |
+
},
|
3938 |
+
{
|
3939 |
+
"epoch": 54.862721171446005,
|
3940 |
+
"grad_norm": 2.872119188308716,
|
3941 |
+
"learning_rate": 9.024630541871921e-05,
|
3942 |
+
"loss": 0.3032,
|
3943 |
+
"step": 11240
|
3944 |
+
},
|
3945 |
+
{
|
3946 |
+
"epoch": 54.96034167175107,
|
3947 |
+
"grad_norm": 3.499648094177246,
|
3948 |
+
"learning_rate": 9.004926108374385e-05,
|
3949 |
+
"loss": 0.3141,
|
3950 |
+
"step": 11260
|
3951 |
+
},
|
3952 |
+
{
|
3953 |
+
"epoch": 55.05796217205613,
|
3954 |
+
"grad_norm": 3.0000522136688232,
|
3955 |
+
"learning_rate": 8.985221674876848e-05,
|
3956 |
+
"loss": 0.3153,
|
3957 |
+
"step": 11280
|
3958 |
+
},
|
3959 |
+
{
|
3960 |
+
"epoch": 55.1555826723612,
|
3961 |
+
"grad_norm": 2.2861599922180176,
|
3962 |
+
"learning_rate": 8.96551724137931e-05,
|
3963 |
+
"loss": 0.3258,
|
3964 |
+
"step": 11300
|
3965 |
+
},
|
3966 |
+
{
|
3967 |
+
"epoch": 55.25320317266626,
|
3968 |
+
"grad_norm": 2.980668306350708,
|
3969 |
+
"learning_rate": 8.945812807881774e-05,
|
3970 |
+
"loss": 0.3099,
|
3971 |
+
"step": 11320
|
3972 |
+
},
|
3973 |
+
{
|
3974 |
+
"epoch": 55.35082367297132,
|
3975 |
+
"grad_norm": 2.286050319671631,
|
3976 |
+
"learning_rate": 8.926108374384237e-05,
|
3977 |
+
"loss": 0.2931,
|
3978 |
+
"step": 11340
|
3979 |
+
},
|
3980 |
+
{
|
3981 |
+
"epoch": 55.448444173276386,
|
3982 |
+
"grad_norm": 4.078646659851074,
|
3983 |
+
"learning_rate": 8.9064039408867e-05,
|
3984 |
+
"loss": 0.3142,
|
3985 |
+
"step": 11360
|
3986 |
+
},
|
3987 |
+
{
|
3988 |
+
"epoch": 55.546064673581455,
|
3989 |
+
"grad_norm": 2.150973320007324,
|
3990 |
+
"learning_rate": 8.886699507389164e-05,
|
3991 |
+
"loss": 0.2839,
|
3992 |
+
"step": 11380
|
3993 |
+
},
|
3994 |
+
{
|
3995 |
+
"epoch": 55.64368517388652,
|
3996 |
+
"grad_norm": 2.671983242034912,
|
3997 |
+
"learning_rate": 8.866995073891627e-05,
|
3998 |
+
"loss": 0.2981,
|
3999 |
+
"step": 11400
|
4000 |
+
},
|
4001 |
+
{
|
4002 |
+
"epoch": 55.74130567419158,
|
4003 |
+
"grad_norm": 3.199276924133301,
|
4004 |
+
"learning_rate": 8.847290640394089e-05,
|
4005 |
+
"loss": 0.304,
|
4006 |
+
"step": 11420
|
4007 |
+
},
|
4008 |
+
{
|
4009 |
+
"epoch": 55.83892617449664,
|
4010 |
+
"grad_norm": 2.477468967437744,
|
4011 |
+
"learning_rate": 8.827586206896552e-05,
|
4012 |
+
"loss": 0.3288,
|
4013 |
+
"step": 11440
|
4014 |
+
},
|
4015 |
+
{
|
4016 |
+
"epoch": 55.93654667480171,
|
4017 |
+
"grad_norm": 2.3130173683166504,
|
4018 |
+
"learning_rate": 8.807881773399015e-05,
|
4019 |
+
"loss": 0.321,
|
4020 |
+
"step": 11460
|
4021 |
+
},
|
4022 |
+
{
|
4023 |
+
"epoch": 56.034167175106774,
|
4024 |
+
"grad_norm": 3.1496715545654297,
|
4025 |
+
"learning_rate": 8.788177339901477e-05,
|
4026 |
+
"loss": 0.2992,
|
4027 |
+
"step": 11480
|
4028 |
+
},
|
4029 |
+
{
|
4030 |
+
"epoch": 56.131787675411836,
|
4031 |
+
"grad_norm": 3.3296494483947754,
|
4032 |
+
"learning_rate": 8.768472906403941e-05,
|
4033 |
+
"loss": 0.3023,
|
4034 |
+
"step": 11500
|
4035 |
+
},
|
4036 |
+
{
|
4037 |
+
"epoch": 56.2294081757169,
|
4038 |
+
"grad_norm": 2.992814540863037,
|
4039 |
+
"learning_rate": 8.748768472906404e-05,
|
4040 |
+
"loss": 0.291,
|
4041 |
+
"step": 11520
|
4042 |
+
},
|
4043 |
+
{
|
4044 |
+
"epoch": 56.32702867602197,
|
4045 |
+
"grad_norm": 2.981858015060425,
|
4046 |
+
"learning_rate": 8.729064039408867e-05,
|
4047 |
+
"loss": 0.2908,
|
4048 |
+
"step": 11540
|
4049 |
+
},
|
4050 |
+
{
|
4051 |
+
"epoch": 56.42464917632703,
|
4052 |
+
"grad_norm": 3.968040704727173,
|
4053 |
+
"learning_rate": 8.709359605911331e-05,
|
4054 |
+
"loss": 0.2963,
|
4055 |
+
"step": 11560
|
4056 |
+
},
|
4057 |
+
{
|
4058 |
+
"epoch": 56.52226967663209,
|
4059 |
+
"grad_norm": 3.6845455169677734,
|
4060 |
+
"learning_rate": 8.689655172413794e-05,
|
4061 |
+
"loss": 0.3137,
|
4062 |
+
"step": 11580
|
4063 |
+
},
|
4064 |
+
{
|
4065 |
+
"epoch": 56.619890176937155,
|
4066 |
+
"grad_norm": 3.8928792476654053,
|
4067 |
+
"learning_rate": 8.669950738916258e-05,
|
4068 |
+
"loss": 0.2971,
|
4069 |
+
"step": 11600
|
4070 |
+
},
|
4071 |
+
{
|
4072 |
+
"epoch": 56.717510677242224,
|
4073 |
+
"grad_norm": 2.064180374145508,
|
4074 |
+
"learning_rate": 8.65024630541872e-05,
|
4075 |
+
"loss": 0.3067,
|
4076 |
+
"step": 11620
|
4077 |
+
},
|
4078 |
+
{
|
4079 |
+
"epoch": 56.815131177547286,
|
4080 |
+
"grad_norm": 2.8107266426086426,
|
4081 |
+
"learning_rate": 8.630541871921183e-05,
|
4082 |
+
"loss": 0.2972,
|
4083 |
+
"step": 11640
|
4084 |
+
},
|
4085 |
+
{
|
4086 |
+
"epoch": 56.91275167785235,
|
4087 |
+
"grad_norm": 2.747004270553589,
|
4088 |
+
"learning_rate": 8.610837438423646e-05,
|
4089 |
+
"loss": 0.3183,
|
4090 |
+
"step": 11660
|
4091 |
+
},
|
4092 |
+
{
|
4093 |
+
"epoch": 57.01037217815741,
|
4094 |
+
"grad_norm": 2.0700557231903076,
|
4095 |
+
"learning_rate": 8.591133004926109e-05,
|
4096 |
+
"loss": 0.3075,
|
4097 |
+
"step": 11680
|
4098 |
+
},
|
4099 |
+
{
|
4100 |
+
"epoch": 57.10799267846248,
|
4101 |
+
"grad_norm": 3.1093757152557373,
|
4102 |
+
"learning_rate": 8.571428571428571e-05,
|
4103 |
+
"loss": 0.2756,
|
4104 |
+
"step": 11700
|
4105 |
+
},
|
4106 |
+
{
|
4107 |
+
"epoch": 57.20561317876754,
|
4108 |
+
"grad_norm": 2.34448504447937,
|
4109 |
+
"learning_rate": 8.551724137931035e-05,
|
4110 |
+
"loss": 0.2898,
|
4111 |
+
"step": 11720
|
4112 |
+
},
|
4113 |
+
{
|
4114 |
+
"epoch": 57.303233679072605,
|
4115 |
+
"grad_norm": 3.3790042400360107,
|
4116 |
+
"learning_rate": 8.532019704433498e-05,
|
4117 |
+
"loss": 0.3081,
|
4118 |
+
"step": 11740
|
4119 |
+
},
|
4120 |
+
{
|
4121 |
+
"epoch": 57.40085417937767,
|
4122 |
+
"grad_norm": 2.700956106185913,
|
4123 |
+
"learning_rate": 8.512315270935961e-05,
|
4124 |
+
"loss": 0.2915,
|
4125 |
+
"step": 11760
|
4126 |
+
},
|
4127 |
+
{
|
4128 |
+
"epoch": 57.49847467968274,
|
4129 |
+
"grad_norm": 2.6353628635406494,
|
4130 |
+
"learning_rate": 8.492610837438423e-05,
|
4131 |
+
"loss": 0.3063,
|
4132 |
+
"step": 11780
|
4133 |
+
},
|
4134 |
+
{
|
4135 |
+
"epoch": 57.5960951799878,
|
4136 |
+
"grad_norm": 2.56706166267395,
|
4137 |
+
"learning_rate": 8.472906403940886e-05,
|
4138 |
+
"loss": 0.3005,
|
4139 |
+
"step": 11800
|
4140 |
+
},
|
4141 |
+
{
|
4142 |
+
"epoch": 57.69371568029286,
|
4143 |
+
"grad_norm": 4.074772357940674,
|
4144 |
+
"learning_rate": 8.45320197044335e-05,
|
4145 |
+
"loss": 0.3007,
|
4146 |
+
"step": 11820
|
4147 |
+
},
|
4148 |
+
{
|
4149 |
+
"epoch": 57.79133618059792,
|
4150 |
+
"grad_norm": 2.786485433578491,
|
4151 |
+
"learning_rate": 8.433497536945813e-05,
|
4152 |
+
"loss": 0.3141,
|
4153 |
+
"step": 11840
|
4154 |
+
},
|
4155 |
+
{
|
4156 |
+
"epoch": 57.88895668090299,
|
4157 |
+
"grad_norm": 2.9513659477233887,
|
4158 |
+
"learning_rate": 8.413793103448277e-05,
|
4159 |
+
"loss": 0.3216,
|
4160 |
+
"step": 11860
|
4161 |
+
},
|
4162 |
+
{
|
4163 |
+
"epoch": 57.986577181208055,
|
4164 |
+
"grad_norm": 3.126004219055176,
|
4165 |
+
"learning_rate": 8.39408866995074e-05,
|
4166 |
+
"loss": 0.3,
|
4167 |
+
"step": 11880
|
4168 |
+
},
|
4169 |
+
{
|
4170 |
+
"epoch": 58.08419768151312,
|
4171 |
+
"grad_norm": 2.20534348487854,
|
4172 |
+
"learning_rate": 8.374384236453202e-05,
|
4173 |
+
"loss": 0.2891,
|
4174 |
+
"step": 11900
|
4175 |
+
},
|
4176 |
+
{
|
4177 |
+
"epoch": 58.18181818181818,
|
4178 |
+
"grad_norm": 4.753482818603516,
|
4179 |
+
"learning_rate": 8.354679802955665e-05,
|
4180 |
+
"loss": 0.3019,
|
4181 |
+
"step": 11920
|
4182 |
+
},
|
4183 |
+
{
|
4184 |
+
"epoch": 58.27943868212325,
|
4185 |
+
"grad_norm": 3.1038873195648193,
|
4186 |
+
"learning_rate": 8.334975369458129e-05,
|
4187 |
+
"loss": 0.283,
|
4188 |
+
"step": 11940
|
4189 |
+
},
|
4190 |
+
{
|
4191 |
+
"epoch": 58.37705918242831,
|
4192 |
+
"grad_norm": 2.9366559982299805,
|
4193 |
+
"learning_rate": 8.315270935960592e-05,
|
4194 |
+
"loss": 0.302,
|
4195 |
+
"step": 11960
|
4196 |
+
},
|
4197 |
+
{
|
4198 |
+
"epoch": 58.474679682733374,
|
4199 |
+
"grad_norm": 3.008777379989624,
|
4200 |
+
"learning_rate": 8.295566502463055e-05,
|
4201 |
+
"loss": 0.3256,
|
4202 |
+
"step": 11980
|
4203 |
+
},
|
4204 |
+
{
|
4205 |
+
"epoch": 58.572300183038436,
|
4206 |
+
"grad_norm": 2.7105023860931396,
|
4207 |
+
"learning_rate": 8.275862068965517e-05,
|
4208 |
+
"loss": 0.2959,
|
4209 |
+
"step": 12000
|
4210 |
+
},
|
4211 |
+
{
|
4212 |
+
"epoch": 58.669920683343506,
|
4213 |
+
"grad_norm": 2.762347936630249,
|
4214 |
+
"learning_rate": 8.25615763546798e-05,
|
4215 |
+
"loss": 0.2826,
|
4216 |
+
"step": 12020
|
4217 |
+
},
|
4218 |
+
{
|
4219 |
+
"epoch": 58.76754118364857,
|
4220 |
+
"grad_norm": 2.8366870880126953,
|
4221 |
+
"learning_rate": 8.236453201970443e-05,
|
4222 |
+
"loss": 0.302,
|
4223 |
+
"step": 12040
|
4224 |
+
},
|
4225 |
+
{
|
4226 |
+
"epoch": 58.86516168395363,
|
4227 |
+
"grad_norm": 2.721994400024414,
|
4228 |
+
"learning_rate": 8.216748768472907e-05,
|
4229 |
+
"loss": 0.2966,
|
4230 |
+
"step": 12060
|
4231 |
+
},
|
4232 |
+
{
|
4233 |
+
"epoch": 58.96278218425869,
|
4234 |
+
"grad_norm": 2.988464117050171,
|
4235 |
+
"learning_rate": 8.19704433497537e-05,
|
4236 |
+
"loss": 0.2899,
|
4237 |
+
"step": 12080
|
4238 |
+
},
|
4239 |
+
{
|
4240 |
+
"epoch": 59.060402684563755,
|
4241 |
+
"grad_norm": 2.6657352447509766,
|
4242 |
+
"learning_rate": 8.177339901477834e-05,
|
4243 |
+
"loss": 0.2889,
|
4244 |
+
"step": 12100
|
4245 |
+
},
|
4246 |
+
{
|
4247 |
+
"epoch": 59.158023184868824,
|
4248 |
+
"grad_norm": 3.703511953353882,
|
4249 |
+
"learning_rate": 8.157635467980296e-05,
|
4250 |
+
"loss": 0.2794,
|
4251 |
+
"step": 12120
|
4252 |
+
},
|
4253 |
+
{
|
4254 |
+
"epoch": 59.25564368517389,
|
4255 |
+
"grad_norm": 2.9937832355499268,
|
4256 |
+
"learning_rate": 8.137931034482759e-05,
|
4257 |
+
"loss": 0.2896,
|
4258 |
+
"step": 12140
|
4259 |
+
},
|
4260 |
+
{
|
4261 |
+
"epoch": 59.35326418547895,
|
4262 |
+
"grad_norm": 3.188159704208374,
|
4263 |
+
"learning_rate": 8.118226600985223e-05,
|
4264 |
+
"loss": 0.2885,
|
4265 |
+
"step": 12160
|
4266 |
+
},
|
4267 |
+
{
|
4268 |
+
"epoch": 59.45088468578401,
|
4269 |
+
"grad_norm": 2.8724703788757324,
|
4270 |
+
"learning_rate": 8.098522167487686e-05,
|
4271 |
+
"loss": 0.2959,
|
4272 |
+
"step": 12180
|
4273 |
+
},
|
4274 |
+
{
|
4275 |
+
"epoch": 59.54850518608908,
|
4276 |
+
"grad_norm": 3.351435422897339,
|
4277 |
+
"learning_rate": 8.078817733990148e-05,
|
4278 |
+
"loss": 0.2867,
|
4279 |
+
"step": 12200
|
4280 |
+
},
|
4281 |
+
{
|
4282 |
+
"epoch": 59.64612568639414,
|
4283 |
+
"grad_norm": 2.5625758171081543,
|
4284 |
+
"learning_rate": 8.059113300492611e-05,
|
4285 |
+
"loss": 0.3042,
|
4286 |
+
"step": 12220
|
4287 |
+
},
|
4288 |
+
{
|
4289 |
+
"epoch": 59.743746186699205,
|
4290 |
+
"grad_norm": 3.3796396255493164,
|
4291 |
+
"learning_rate": 8.039408866995074e-05,
|
4292 |
+
"loss": 0.301,
|
4293 |
+
"step": 12240
|
4294 |
+
},
|
4295 |
+
{
|
4296 |
+
"epoch": 59.84136668700427,
|
4297 |
+
"grad_norm": 2.787851572036743,
|
4298 |
+
"learning_rate": 8.019704433497537e-05,
|
4299 |
+
"loss": 0.3072,
|
4300 |
+
"step": 12260
|
4301 |
+
},
|
4302 |
+
{
|
4303 |
+
"epoch": 59.93898718730934,
|
4304 |
+
"grad_norm": 2.9104974269866943,
|
4305 |
+
"learning_rate": 8e-05,
|
4306 |
+
"loss": 0.3059,
|
4307 |
+
"step": 12280
|
4308 |
+
},
|
4309 |
+
{
|
4310 |
+
"epoch": 60.0366076876144,
|
4311 |
+
"grad_norm": 2.957249879837036,
|
4312 |
+
"learning_rate": 7.980295566502463e-05,
|
4313 |
+
"loss": 0.2965,
|
4314 |
+
"step": 12300
|
4315 |
+
},
|
4316 |
+
{
|
4317 |
+
"epoch": 60.13422818791946,
|
4318 |
+
"grad_norm": 2.2982118129730225,
|
4319 |
+
"learning_rate": 7.960591133004926e-05,
|
4320 |
+
"loss": 0.2703,
|
4321 |
+
"step": 12320
|
4322 |
+
},
|
4323 |
+
{
|
4324 |
+
"epoch": 60.231848688224524,
|
4325 |
+
"grad_norm": 3.548534870147705,
|
4326 |
+
"learning_rate": 7.940886699507389e-05,
|
4327 |
+
"loss": 0.2843,
|
4328 |
+
"step": 12340
|
4329 |
+
},
|
4330 |
+
{
|
4331 |
+
"epoch": 60.32946918852959,
|
4332 |
+
"grad_norm": 2.3399384021759033,
|
4333 |
+
"learning_rate": 7.921182266009853e-05,
|
4334 |
+
"loss": 0.2855,
|
4335 |
+
"step": 12360
|
4336 |
+
},
|
4337 |
+
{
|
4338 |
+
"epoch": 60.427089688834656,
|
4339 |
+
"grad_norm": 3.4186365604400635,
|
4340 |
+
"learning_rate": 7.901477832512316e-05,
|
4341 |
+
"loss": 0.2942,
|
4342 |
+
"step": 12380
|
4343 |
+
},
|
4344 |
+
{
|
4345 |
+
"epoch": 60.52471018913972,
|
4346 |
+
"grad_norm": 2.572951316833496,
|
4347 |
+
"learning_rate": 7.88177339901478e-05,
|
4348 |
+
"loss": 0.2918,
|
4349 |
+
"step": 12400
|
4350 |
+
},
|
4351 |
+
{
|
4352 |
+
"epoch": 60.62233068944478,
|
4353 |
+
"grad_norm": 2.1056010723114014,
|
4354 |
+
"learning_rate": 7.862068965517242e-05,
|
4355 |
+
"loss": 0.3051,
|
4356 |
+
"step": 12420
|
4357 |
+
},
|
4358 |
+
{
|
4359 |
+
"epoch": 60.71995118974985,
|
4360 |
+
"grad_norm": 4.122783184051514,
|
4361 |
+
"learning_rate": 7.842364532019705e-05,
|
4362 |
+
"loss": 0.2811,
|
4363 |
+
"step": 12440
|
4364 |
+
},
|
4365 |
+
{
|
4366 |
+
"epoch": 60.81757169005491,
|
4367 |
+
"grad_norm": 2.3634865283966064,
|
4368 |
+
"learning_rate": 7.822660098522168e-05,
|
4369 |
+
"loss": 0.3063,
|
4370 |
+
"step": 12460
|
4371 |
+
},
|
4372 |
+
{
|
4373 |
+
"epoch": 60.915192190359974,
|
4374 |
+
"grad_norm": 3.362290143966675,
|
4375 |
+
"learning_rate": 7.80295566502463e-05,
|
4376 |
+
"loss": 0.2954,
|
4377 |
+
"step": 12480
|
4378 |
+
},
|
4379 |
+
{
|
4380 |
+
"epoch": 61.01281269066504,
|
4381 |
+
"grad_norm": 4.63106632232666,
|
4382 |
+
"learning_rate": 7.783251231527095e-05,
|
4383 |
+
"loss": 0.2855,
|
4384 |
+
"step": 12500
|
4385 |
+
},
|
4386 |
+
{
|
4387 |
+
"epoch": 61.110433190970106,
|
4388 |
+
"grad_norm": 3.6261041164398193,
|
4389 |
+
"learning_rate": 7.763546798029557e-05,
|
4390 |
+
"loss": 0.2792,
|
4391 |
+
"step": 12520
|
4392 |
+
},
|
4393 |
+
{
|
4394 |
+
"epoch": 61.20805369127517,
|
4395 |
+
"grad_norm": 2.869415760040283,
|
4396 |
+
"learning_rate": 7.74384236453202e-05,
|
4397 |
+
"loss": 0.2833,
|
4398 |
+
"step": 12540
|
4399 |
+
},
|
4400 |
+
{
|
4401 |
+
"epoch": 61.30567419158023,
|
4402 |
+
"grad_norm": 2.7370972633361816,
|
4403 |
+
"learning_rate": 7.724137931034483e-05,
|
4404 |
+
"loss": 0.2997,
|
4405 |
+
"step": 12560
|
4406 |
+
},
|
4407 |
+
{
|
4408 |
+
"epoch": 61.40329469188529,
|
4409 |
+
"grad_norm": 3.5397825241088867,
|
4410 |
+
"learning_rate": 7.704433497536945e-05,
|
4411 |
+
"loss": 0.2799,
|
4412 |
+
"step": 12580
|
4413 |
+
},
|
4414 |
+
{
|
4415 |
+
"epoch": 61.50091519219036,
|
4416 |
+
"grad_norm": 2.3903191089630127,
|
4417 |
+
"learning_rate": 7.684729064039408e-05,
|
4418 |
+
"loss": 0.2857,
|
4419 |
+
"step": 12600
|
4420 |
+
},
|
4421 |
+
{
|
4422 |
+
"epoch": 61.598535692495425,
|
4423 |
+
"grad_norm": 3.3589389324188232,
|
4424 |
+
"learning_rate": 7.665024630541872e-05,
|
4425 |
+
"loss": 0.2823,
|
4426 |
+
"step": 12620
|
4427 |
+
},
|
4428 |
+
{
|
4429 |
+
"epoch": 61.69615619280049,
|
4430 |
+
"grad_norm": 4.420291423797607,
|
4431 |
+
"learning_rate": 7.645320197044336e-05,
|
4432 |
+
"loss": 0.2895,
|
4433 |
+
"step": 12640
|
4434 |
+
},
|
4435 |
+
{
|
4436 |
+
"epoch": 61.79377669310555,
|
4437 |
+
"grad_norm": 3.060859441757202,
|
4438 |
+
"learning_rate": 7.625615763546799e-05,
|
4439 |
+
"loss": 0.2859,
|
4440 |
+
"step": 12660
|
4441 |
+
},
|
4442 |
+
{
|
4443 |
+
"epoch": 61.89139719341062,
|
4444 |
+
"grad_norm": 3.5927321910858154,
|
4445 |
+
"learning_rate": 7.605911330049262e-05,
|
4446 |
+
"loss": 0.2954,
|
4447 |
+
"step": 12680
|
4448 |
+
},
|
4449 |
+
{
|
4450 |
+
"epoch": 61.98901769371568,
|
4451 |
+
"grad_norm": 2.7577738761901855,
|
4452 |
+
"learning_rate": 7.586206896551724e-05,
|
4453 |
+
"loss": 0.2832,
|
4454 |
+
"step": 12700
|
4455 |
+
},
|
4456 |
+
{
|
4457 |
+
"epoch": 62.08663819402074,
|
4458 |
+
"grad_norm": 4.519462585449219,
|
4459 |
+
"learning_rate": 7.566502463054188e-05,
|
4460 |
+
"loss": 0.2695,
|
4461 |
+
"step": 12720
|
4462 |
+
},
|
4463 |
+
{
|
4464 |
+
"epoch": 62.184258694325806,
|
4465 |
+
"grad_norm": 2.231842279434204,
|
4466 |
+
"learning_rate": 7.546798029556651e-05,
|
4467 |
+
"loss": 0.2894,
|
4468 |
+
"step": 12740
|
4469 |
+
},
|
4470 |
+
{
|
4471 |
+
"epoch": 62.281879194630875,
|
4472 |
+
"grad_norm": 3.5176825523376465,
|
4473 |
+
"learning_rate": 7.527093596059114e-05,
|
4474 |
+
"loss": 0.2749,
|
4475 |
+
"step": 12760
|
4476 |
+
},
|
4477 |
+
{
|
4478 |
+
"epoch": 62.37949969493594,
|
4479 |
+
"grad_norm": 3.319891929626465,
|
4480 |
+
"learning_rate": 7.507389162561577e-05,
|
4481 |
+
"loss": 0.2909,
|
4482 |
+
"step": 12780
|
4483 |
+
},
|
4484 |
+
{
|
4485 |
+
"epoch": 62.477120195241,
|
4486 |
+
"grad_norm": 2.778862237930298,
|
4487 |
+
"learning_rate": 7.487684729064039e-05,
|
4488 |
+
"loss": 0.2816,
|
4489 |
+
"step": 12800
|
4490 |
+
},
|
4491 |
+
{
|
4492 |
+
"epoch": 62.57474069554606,
|
4493 |
+
"grad_norm": 2.7136170864105225,
|
4494 |
+
"learning_rate": 7.467980295566502e-05,
|
4495 |
+
"loss": 0.286,
|
4496 |
+
"step": 12820
|
4497 |
+
},
|
4498 |
+
{
|
4499 |
+
"epoch": 62.67236119585113,
|
4500 |
+
"grad_norm": 2.841850519180298,
|
4501 |
+
"learning_rate": 7.448275862068966e-05,
|
4502 |
+
"loss": 0.3078,
|
4503 |
+
"step": 12840
|
4504 |
+
},
|
4505 |
+
{
|
4506 |
+
"epoch": 62.769981696156194,
|
4507 |
+
"grad_norm": 3.159632682800293,
|
4508 |
+
"learning_rate": 7.428571428571429e-05,
|
4509 |
+
"loss": 0.2693,
|
4510 |
+
"step": 12860
|
4511 |
+
},
|
4512 |
+
{
|
4513 |
+
"epoch": 62.867602196461256,
|
4514 |
+
"grad_norm": 2.638611078262329,
|
4515 |
+
"learning_rate": 7.408866995073891e-05,
|
4516 |
+
"loss": 0.2838,
|
4517 |
+
"step": 12880
|
4518 |
+
},
|
4519 |
+
{
|
4520 |
+
"epoch": 62.96522269676632,
|
4521 |
+
"grad_norm": 3.453857421875,
|
4522 |
+
"learning_rate": 7.389162561576355e-05,
|
4523 |
+
"loss": 0.2892,
|
4524 |
+
"step": 12900
|
4525 |
+
},
|
4526 |
+
{
|
4527 |
+
"epoch": 63.06284319707139,
|
4528 |
+
"grad_norm": 3.6586861610412598,
|
4529 |
+
"learning_rate": 7.369458128078818e-05,
|
4530 |
+
"loss": 0.2626,
|
4531 |
+
"step": 12920
|
4532 |
+
},
|
4533 |
+
{
|
4534 |
+
"epoch": 63.16046369737645,
|
4535 |
+
"grad_norm": 3.8204469680786133,
|
4536 |
+
"learning_rate": 7.349753694581281e-05,
|
4537 |
+
"loss": 0.2834,
|
4538 |
+
"step": 12940
|
4539 |
+
},
|
4540 |
+
{
|
4541 |
+
"epoch": 63.25808419768151,
|
4542 |
+
"grad_norm": 1.7463505268096924,
|
4543 |
+
"learning_rate": 7.330049261083745e-05,
|
4544 |
+
"loss": 0.2909,
|
4545 |
+
"step": 12960
|
4546 |
+
},
|
4547 |
+
{
|
4548 |
+
"epoch": 63.355704697986575,
|
4549 |
+
"grad_norm": 1.687853217124939,
|
4550 |
+
"learning_rate": 7.310344827586208e-05,
|
4551 |
+
"loss": 0.2892,
|
4552 |
+
"step": 12980
|
4553 |
+
},
|
4554 |
+
{
|
4555 |
+
"epoch": 63.453325198291644,
|
4556 |
+
"grad_norm": 2.835196018218994,
|
4557 |
+
"learning_rate": 7.29064039408867e-05,
|
4558 |
+
"loss": 0.2763,
|
4559 |
+
"step": 13000
|
4560 |
+
},
|
4561 |
+
{
|
4562 |
+
"epoch": 63.550945698596706,
|
4563 |
+
"grad_norm": 3.77742862701416,
|
4564 |
+
"learning_rate": 7.270935960591133e-05,
|
4565 |
+
"loss": 0.2834,
|
4566 |
+
"step": 13020
|
4567 |
+
},
|
4568 |
+
{
|
4569 |
+
"epoch": 63.64856619890177,
|
4570 |
+
"grad_norm": 2.1246883869171143,
|
4571 |
+
"learning_rate": 7.251231527093596e-05,
|
4572 |
+
"loss": 0.2859,
|
4573 |
+
"step": 13040
|
4574 |
+
},
|
4575 |
+
{
|
4576 |
+
"epoch": 63.74618669920683,
|
4577 |
+
"grad_norm": 3.592597246170044,
|
4578 |
+
"learning_rate": 7.23152709359606e-05,
|
4579 |
+
"loss": 0.2865,
|
4580 |
+
"step": 13060
|
4581 |
+
},
|
4582 |
+
{
|
4583 |
+
"epoch": 63.8438071995119,
|
4584 |
+
"grad_norm": 2.8954873085021973,
|
4585 |
+
"learning_rate": 7.211822660098523e-05,
|
4586 |
+
"loss": 0.2855,
|
4587 |
+
"step": 13080
|
4588 |
+
},
|
4589 |
+
{
|
4590 |
+
"epoch": 63.94142769981696,
|
4591 |
+
"grad_norm": 2.266686201095581,
|
4592 |
+
"learning_rate": 7.192118226600985e-05,
|
4593 |
+
"loss": 0.2814,
|
4594 |
+
"step": 13100
|
4595 |
+
},
|
4596 |
+
{
|
4597 |
+
"epoch": 64.03904820012202,
|
4598 |
+
"grad_norm": 1.9330942630767822,
|
4599 |
+
"learning_rate": 7.172413793103448e-05,
|
4600 |
+
"loss": 0.2832,
|
4601 |
+
"step": 13120
|
4602 |
+
},
|
4603 |
+
{
|
4604 |
+
"epoch": 64.1366687004271,
|
4605 |
+
"grad_norm": 4.008347511291504,
|
4606 |
+
"learning_rate": 7.152709359605912e-05,
|
4607 |
+
"loss": 0.2762,
|
4608 |
+
"step": 13140
|
4609 |
+
},
|
4610 |
+
{
|
4611 |
+
"epoch": 64.23428920073215,
|
4612 |
+
"grad_norm": 2.2452552318573,
|
4613 |
+
"learning_rate": 7.133004926108375e-05,
|
4614 |
+
"loss": 0.269,
|
4615 |
+
"step": 13160
|
4616 |
+
},
|
4617 |
+
{
|
4618 |
+
"epoch": 64.33190970103722,
|
4619 |
+
"grad_norm": 7.247570991516113,
|
4620 |
+
"learning_rate": 7.113300492610839e-05,
|
4621 |
+
"loss": 0.2652,
|
4622 |
+
"step": 13180
|
4623 |
+
},
|
4624 |
+
{
|
4625 |
+
"epoch": 64.42953020134229,
|
4626 |
+
"grad_norm": 4.846076488494873,
|
4627 |
+
"learning_rate": 7.093596059113302e-05,
|
4628 |
+
"loss": 0.2766,
|
4629 |
+
"step": 13200
|
4630 |
+
},
|
4631 |
+
{
|
4632 |
+
"epoch": 64.52715070164734,
|
4633 |
+
"grad_norm": 3.444746732711792,
|
4634 |
+
"learning_rate": 7.073891625615764e-05,
|
4635 |
+
"loss": 0.2789,
|
4636 |
+
"step": 13220
|
4637 |
+
},
|
4638 |
+
{
|
4639 |
+
"epoch": 64.62477120195241,
|
4640 |
+
"grad_norm": 2.506460428237915,
|
4641 |
+
"learning_rate": 7.054187192118227e-05,
|
4642 |
+
"loss": 0.279,
|
4643 |
+
"step": 13240
|
4644 |
+
},
|
4645 |
+
{
|
4646 |
+
"epoch": 64.72239170225747,
|
4647 |
+
"grad_norm": 3.3973569869995117,
|
4648 |
+
"learning_rate": 7.03448275862069e-05,
|
4649 |
+
"loss": 0.2887,
|
4650 |
+
"step": 13260
|
4651 |
+
},
|
4652 |
+
{
|
4653 |
+
"epoch": 64.82001220256254,
|
4654 |
+
"grad_norm": 3.14697265625,
|
4655 |
+
"learning_rate": 7.014778325123154e-05,
|
4656 |
+
"loss": 0.2813,
|
4657 |
+
"step": 13280
|
4658 |
+
},
|
4659 |
+
{
|
4660 |
+
"epoch": 64.9176327028676,
|
4661 |
+
"grad_norm": 4.694430828094482,
|
4662 |
+
"learning_rate": 6.995073891625616e-05,
|
4663 |
+
"loss": 0.3026,
|
4664 |
+
"step": 13300
|
4665 |
+
},
|
4666 |
+
{
|
4667 |
+
"epoch": 65.01525320317266,
|
4668 |
+
"grad_norm": 2.2463550567626953,
|
4669 |
+
"learning_rate": 6.975369458128079e-05,
|
4670 |
+
"loss": 0.2739,
|
4671 |
+
"step": 13320
|
4672 |
+
},
|
4673 |
+
{
|
4674 |
+
"epoch": 65.11287370347773,
|
4675 |
+
"grad_norm": 2.907592535018921,
|
4676 |
+
"learning_rate": 6.955665024630542e-05,
|
4677 |
+
"loss": 0.2783,
|
4678 |
+
"step": 13340
|
4679 |
+
},
|
4680 |
+
{
|
4681 |
+
"epoch": 65.2104942037828,
|
4682 |
+
"grad_norm": 2.9708614349365234,
|
4683 |
+
"learning_rate": 6.935960591133005e-05,
|
4684 |
+
"loss": 0.2718,
|
4685 |
+
"step": 13360
|
4686 |
+
},
|
4687 |
+
{
|
4688 |
+
"epoch": 65.30811470408786,
|
4689 |
+
"grad_norm": 2.7227044105529785,
|
4690 |
+
"learning_rate": 6.916256157635467e-05,
|
4691 |
+
"loss": 0.2615,
|
4692 |
+
"step": 13380
|
4693 |
+
},
|
4694 |
+
{
|
4695 |
+
"epoch": 65.40573520439293,
|
4696 |
+
"grad_norm": 2.3960001468658447,
|
4697 |
+
"learning_rate": 6.896551724137931e-05,
|
4698 |
+
"loss": 0.2822,
|
4699 |
+
"step": 13400
|
4700 |
+
},
|
4701 |
+
{
|
4702 |
+
"epoch": 65.50335570469798,
|
4703 |
+
"grad_norm": 2.032240629196167,
|
4704 |
+
"learning_rate": 6.876847290640394e-05,
|
4705 |
+
"loss": 0.282,
|
4706 |
+
"step": 13420
|
4707 |
+
},
|
4708 |
+
{
|
4709 |
+
"epoch": 65.60097620500305,
|
4710 |
+
"grad_norm": 2.5334010124206543,
|
4711 |
+
"learning_rate": 6.857142857142858e-05,
|
4712 |
+
"loss": 0.2771,
|
4713 |
+
"step": 13440
|
4714 |
+
},
|
4715 |
+
{
|
4716 |
+
"epoch": 65.69859670530812,
|
4717 |
+
"grad_norm": 7.930431842803955,
|
4718 |
+
"learning_rate": 6.837438423645321e-05,
|
4719 |
+
"loss": 0.2878,
|
4720 |
+
"step": 13460
|
4721 |
+
},
|
4722 |
+
{
|
4723 |
+
"epoch": 65.79621720561317,
|
4724 |
+
"grad_norm": 2.709092378616333,
|
4725 |
+
"learning_rate": 6.817733990147784e-05,
|
4726 |
+
"loss": 0.2797,
|
4727 |
+
"step": 13480
|
4728 |
+
},
|
4729 |
+
{
|
4730 |
+
"epoch": 65.89383770591824,
|
4731 |
+
"grad_norm": 4.455546855926514,
|
4732 |
+
"learning_rate": 6.798029556650246e-05,
|
4733 |
+
"loss": 0.2803,
|
4734 |
+
"step": 13500
|
4735 |
+
},
|
4736 |
+
{
|
4737 |
+
"epoch": 65.99145820622331,
|
4738 |
+
"grad_norm": 4.6384077072143555,
|
4739 |
+
"learning_rate": 6.77832512315271e-05,
|
4740 |
+
"loss": 0.2764,
|
4741 |
+
"step": 13520
|
4742 |
+
},
|
4743 |
+
{
|
4744 |
+
"epoch": 66.08907870652837,
|
4745 |
+
"grad_norm": 2.7529897689819336,
|
4746 |
+
"learning_rate": 6.758620689655173e-05,
|
4747 |
+
"loss": 0.2614,
|
4748 |
+
"step": 13540
|
4749 |
+
},
|
4750 |
+
{
|
4751 |
+
"epoch": 66.18669920683344,
|
4752 |
+
"grad_norm": 2.0837860107421875,
|
4753 |
+
"learning_rate": 6.738916256157636e-05,
|
4754 |
+
"loss": 0.2696,
|
4755 |
+
"step": 13560
|
4756 |
+
},
|
4757 |
+
{
|
4758 |
+
"epoch": 66.2843197071385,
|
4759 |
+
"grad_norm": 1.6655378341674805,
|
4760 |
+
"learning_rate": 6.719211822660098e-05,
|
4761 |
+
"loss": 0.2781,
|
4762 |
+
"step": 13580
|
4763 |
+
},
|
4764 |
+
{
|
4765 |
+
"epoch": 66.38194020744356,
|
4766 |
+
"grad_norm": 1.8926398754119873,
|
4767 |
+
"learning_rate": 6.699507389162561e-05,
|
4768 |
+
"loss": 0.273,
|
4769 |
+
"step": 13600
|
4770 |
+
},
|
4771 |
+
{
|
4772 |
+
"epoch": 66.47956070774863,
|
4773 |
+
"grad_norm": 1.8903833627700806,
|
4774 |
+
"learning_rate": 6.679802955665025e-05,
|
4775 |
+
"loss": 0.2683,
|
4776 |
+
"step": 13620
|
4777 |
+
},
|
4778 |
+
{
|
4779 |
+
"epoch": 66.57718120805369,
|
4780 |
+
"grad_norm": 3.0182383060455322,
|
4781 |
+
"learning_rate": 6.660098522167488e-05,
|
4782 |
+
"loss": 0.2685,
|
4783 |
+
"step": 13640
|
4784 |
+
},
|
4785 |
+
{
|
4786 |
+
"epoch": 66.67480170835876,
|
4787 |
+
"grad_norm": 3.0081100463867188,
|
4788 |
+
"learning_rate": 6.64039408866995e-05,
|
4789 |
+
"loss": 0.2915,
|
4790 |
+
"step": 13660
|
4791 |
+
},
|
4792 |
+
{
|
4793 |
+
"epoch": 66.77242220866383,
|
4794 |
+
"grad_norm": 2.345440149307251,
|
4795 |
+
"learning_rate": 6.620689655172415e-05,
|
4796 |
+
"loss": 0.2707,
|
4797 |
+
"step": 13680
|
4798 |
+
},
|
4799 |
+
{
|
4800 |
+
"epoch": 66.87004270896888,
|
4801 |
+
"grad_norm": 2.430608034133911,
|
4802 |
+
"learning_rate": 6.600985221674877e-05,
|
4803 |
+
"loss": 0.2675,
|
4804 |
+
"step": 13700
|
4805 |
+
},
|
4806 |
+
{
|
4807 |
+
"epoch": 66.96766320927395,
|
4808 |
+
"grad_norm": 4.09646463394165,
|
4809 |
+
"learning_rate": 6.58128078817734e-05,
|
4810 |
+
"loss": 0.2886,
|
4811 |
+
"step": 13720
|
4812 |
+
},
|
4813 |
+
{
|
4814 |
+
"epoch": 67.065283709579,
|
4815 |
+
"grad_norm": 2.696843147277832,
|
4816 |
+
"learning_rate": 6.561576354679804e-05,
|
4817 |
+
"loss": 0.2743,
|
4818 |
+
"step": 13740
|
4819 |
+
},
|
4820 |
+
{
|
4821 |
+
"epoch": 67.16290420988408,
|
4822 |
+
"grad_norm": 1.8098782300949097,
|
4823 |
+
"learning_rate": 6.541871921182267e-05,
|
4824 |
+
"loss": 0.2629,
|
4825 |
+
"step": 13760
|
4826 |
+
},
|
4827 |
+
{
|
4828 |
+
"epoch": 67.26052471018915,
|
4829 |
+
"grad_norm": 2.604454278945923,
|
4830 |
+
"learning_rate": 6.52216748768473e-05,
|
4831 |
+
"loss": 0.2701,
|
4832 |
+
"step": 13780
|
4833 |
+
},
|
4834 |
+
{
|
4835 |
+
"epoch": 67.3581452104942,
|
4836 |
+
"grad_norm": 2.6400327682495117,
|
4837 |
+
"learning_rate": 6.502463054187192e-05,
|
4838 |
+
"loss": 0.2791,
|
4839 |
+
"step": 13800
|
4840 |
+
},
|
4841 |
+
{
|
4842 |
+
"epoch": 67.45576571079927,
|
4843 |
+
"grad_norm": 2.6029961109161377,
|
4844 |
+
"learning_rate": 6.482758620689655e-05,
|
4845 |
+
"loss": 0.2753,
|
4846 |
+
"step": 13820
|
4847 |
+
},
|
4848 |
+
{
|
4849 |
+
"epoch": 67.55338621110434,
|
4850 |
+
"grad_norm": 2.493805170059204,
|
4851 |
+
"learning_rate": 6.463054187192119e-05,
|
4852 |
+
"loss": 0.2654,
|
4853 |
+
"step": 13840
|
4854 |
+
},
|
4855 |
+
{
|
4856 |
+
"epoch": 67.6510067114094,
|
4857 |
+
"grad_norm": 3.1555075645446777,
|
4858 |
+
"learning_rate": 6.443349753694582e-05,
|
4859 |
+
"loss": 0.2701,
|
4860 |
+
"step": 13860
|
4861 |
+
},
|
4862 |
+
{
|
4863 |
+
"epoch": 67.74862721171446,
|
4864 |
+
"grad_norm": 4.280105113983154,
|
4865 |
+
"learning_rate": 6.423645320197044e-05,
|
4866 |
+
"loss": 0.2732,
|
4867 |
+
"step": 13880
|
4868 |
+
},
|
4869 |
+
{
|
4870 |
+
"epoch": 67.84624771201952,
|
4871 |
+
"grad_norm": 2.8167061805725098,
|
4872 |
+
"learning_rate": 6.403940886699507e-05,
|
4873 |
+
"loss": 0.2755,
|
4874 |
+
"step": 13900
|
4875 |
+
},
|
4876 |
+
{
|
4877 |
+
"epoch": 67.94386821232459,
|
4878 |
+
"grad_norm": 3.5046565532684326,
|
4879 |
+
"learning_rate": 6.38423645320197e-05,
|
4880 |
+
"loss": 0.2831,
|
4881 |
+
"step": 13920
|
4882 |
+
},
|
4883 |
+
{
|
4884 |
+
"epoch": 68.04148871262966,
|
4885 |
+
"grad_norm": 2.4737610816955566,
|
4886 |
+
"learning_rate": 6.364532019704434e-05,
|
4887 |
+
"loss": 0.2737,
|
4888 |
+
"step": 13940
|
4889 |
+
},
|
4890 |
+
{
|
4891 |
+
"epoch": 68.13910921293471,
|
4892 |
+
"grad_norm": 1.996193766593933,
|
4893 |
+
"learning_rate": 6.344827586206897e-05,
|
4894 |
+
"loss": 0.2637,
|
4895 |
+
"step": 13960
|
4896 |
+
},
|
4897 |
+
{
|
4898 |
+
"epoch": 68.23672971323978,
|
4899 |
+
"grad_norm": 2.7088236808776855,
|
4900 |
+
"learning_rate": 6.325123152709361e-05,
|
4901 |
+
"loss": 0.2683,
|
4902 |
+
"step": 13980
|
4903 |
+
},
|
4904 |
+
{
|
4905 |
+
"epoch": 68.33435021354484,
|
4906 |
+
"grad_norm": 2.344050168991089,
|
4907 |
+
"learning_rate": 6.305418719211823e-05,
|
4908 |
+
"loss": 0.2685,
|
4909 |
+
"step": 14000
|
4910 |
+
},
|
4911 |
+
{
|
4912 |
+
"epoch": 68.4319707138499,
|
4913 |
+
"grad_norm": 3.3628969192504883,
|
4914 |
+
"learning_rate": 6.285714285714286e-05,
|
4915 |
+
"loss": 0.2728,
|
4916 |
+
"step": 14020
|
4917 |
+
},
|
4918 |
+
{
|
4919 |
+
"epoch": 68.52959121415498,
|
4920 |
+
"grad_norm": 2.8613572120666504,
|
4921 |
+
"learning_rate": 6.266009852216749e-05,
|
4922 |
+
"loss": 0.2668,
|
4923 |
+
"step": 14040
|
4924 |
+
},
|
4925 |
+
{
|
4926 |
+
"epoch": 68.62721171446003,
|
4927 |
+
"grad_norm": 2.752930164337158,
|
4928 |
+
"learning_rate": 6.246305418719212e-05,
|
4929 |
+
"loss": 0.2753,
|
4930 |
+
"step": 14060
|
4931 |
+
},
|
4932 |
+
{
|
4933 |
+
"epoch": 68.7248322147651,
|
4934 |
+
"grad_norm": 2.426806926727295,
|
4935 |
+
"learning_rate": 6.226600985221676e-05,
|
4936 |
+
"loss": 0.2545,
|
4937 |
+
"step": 14080
|
4938 |
+
},
|
4939 |
+
{
|
4940 |
+
"epoch": 68.82245271507017,
|
4941 |
+
"grad_norm": 2.4970877170562744,
|
4942 |
+
"learning_rate": 6.206896551724138e-05,
|
4943 |
+
"loss": 0.2733,
|
4944 |
+
"step": 14100
|
4945 |
+
},
|
4946 |
+
{
|
4947 |
+
"epoch": 68.92007321537523,
|
4948 |
+
"grad_norm": 2.6764674186706543,
|
4949 |
+
"learning_rate": 6.187192118226601e-05,
|
4950 |
+
"loss": 0.2726,
|
4951 |
+
"step": 14120
|
4952 |
+
},
|
4953 |
+
{
|
4954 |
+
"epoch": 69.0176937156803,
|
4955 |
+
"grad_norm": 2.3702871799468994,
|
4956 |
+
"learning_rate": 6.167487684729064e-05,
|
4957 |
+
"loss": 0.2701,
|
4958 |
+
"step": 14140
|
4959 |
+
},
|
4960 |
+
{
|
4961 |
+
"epoch": 69.11531421598535,
|
4962 |
+
"grad_norm": 3.5141944885253906,
|
4963 |
+
"learning_rate": 6.147783251231526e-05,
|
4964 |
+
"loss": 0.2643,
|
4965 |
+
"step": 14160
|
4966 |
+
},
|
4967 |
+
{
|
4968 |
+
"epoch": 69.21293471629042,
|
4969 |
+
"grad_norm": 2.7750911712646484,
|
4970 |
+
"learning_rate": 6.12807881773399e-05,
|
4971 |
+
"loss": 0.248,
|
4972 |
+
"step": 14180
|
4973 |
+
},
|
4974 |
+
{
|
4975 |
+
"epoch": 69.31055521659549,
|
4976 |
+
"grad_norm": 4.1003217697143555,
|
4977 |
+
"learning_rate": 6.108374384236453e-05,
|
4978 |
+
"loss": 0.2618,
|
4979 |
+
"step": 14200
|
4980 |
+
},
|
4981 |
+
{
|
4982 |
+
"epoch": 69.40817571690054,
|
4983 |
+
"grad_norm": 2.183353900909424,
|
4984 |
+
"learning_rate": 6.0886699507389166e-05,
|
4985 |
+
"loss": 0.2618,
|
4986 |
+
"step": 14220
|
4987 |
+
},
|
4988 |
+
{
|
4989 |
+
"epoch": 69.50579621720561,
|
4990 |
+
"grad_norm": 2.447449207305908,
|
4991 |
+
"learning_rate": 6.068965517241379e-05,
|
4992 |
+
"loss": 0.284,
|
4993 |
+
"step": 14240
|
4994 |
+
},
|
4995 |
+
{
|
4996 |
+
"epoch": 69.60341671751068,
|
4997 |
+
"grad_norm": 2.466543674468994,
|
4998 |
+
"learning_rate": 6.049261083743843e-05,
|
4999 |
+
"loss": 0.269,
|
5000 |
+
"step": 14260
|
5001 |
+
},
|
5002 |
+
{
|
5003 |
+
"epoch": 69.70103721781574,
|
5004 |
+
"grad_norm": 3.8052902221679688,
|
5005 |
+
"learning_rate": 6.0295566502463054e-05,
|
5006 |
+
"loss": 0.2681,
|
5007 |
+
"step": 14280
|
5008 |
+
},
|
5009 |
+
{
|
5010 |
+
"epoch": 69.79865771812081,
|
5011 |
+
"grad_norm": 3.1913719177246094,
|
5012 |
+
"learning_rate": 6.0098522167487695e-05,
|
5013 |
+
"loss": 0.2677,
|
5014 |
+
"step": 14300
|
5015 |
+
},
|
5016 |
+
{
|
5017 |
+
"epoch": 69.89627821842586,
|
5018 |
+
"grad_norm": 1.6767873764038086,
|
5019 |
+
"learning_rate": 5.990147783251232e-05,
|
5020 |
+
"loss": 0.2739,
|
5021 |
+
"step": 14320
|
5022 |
+
},
|
5023 |
+
{
|
5024 |
+
"epoch": 69.99389871873093,
|
5025 |
+
"grad_norm": 2.805734634399414,
|
5026 |
+
"learning_rate": 5.970443349753695e-05,
|
5027 |
+
"loss": 0.2686,
|
5028 |
+
"step": 14340
|
5029 |
+
},
|
5030 |
+
{
|
5031 |
+
"epoch": 70.091519219036,
|
5032 |
+
"grad_norm": 2.671316146850586,
|
5033 |
+
"learning_rate": 5.9507389162561576e-05,
|
5034 |
+
"loss": 0.2558,
|
5035 |
+
"step": 14360
|
5036 |
+
},
|
5037 |
+
{
|
5038 |
+
"epoch": 70.18913971934106,
|
5039 |
+
"grad_norm": 2.5105350017547607,
|
5040 |
+
"learning_rate": 5.93103448275862e-05,
|
5041 |
+
"loss": 0.2692,
|
5042 |
+
"step": 14380
|
5043 |
+
},
|
5044 |
+
{
|
5045 |
+
"epoch": 70.28676021964613,
|
5046 |
+
"grad_norm": 2.0773072242736816,
|
5047 |
+
"learning_rate": 5.9113300492610844e-05,
|
5048 |
+
"loss": 0.266,
|
5049 |
+
"step": 14400
|
5050 |
+
},
|
5051 |
+
{
|
5052 |
+
"epoch": 70.3843807199512,
|
5053 |
+
"grad_norm": 2.2632055282592773,
|
5054 |
+
"learning_rate": 5.891625615763548e-05,
|
5055 |
+
"loss": 0.2624,
|
5056 |
+
"step": 14420
|
5057 |
+
},
|
5058 |
+
{
|
5059 |
+
"epoch": 70.48200122025625,
|
5060 |
+
"grad_norm": 3.4696826934814453,
|
5061 |
+
"learning_rate": 5.8719211822660105e-05,
|
5062 |
+
"loss": 0.2616,
|
5063 |
+
"step": 14440
|
5064 |
+
},
|
5065 |
+
{
|
5066 |
+
"epoch": 70.57962172056132,
|
5067 |
+
"grad_norm": 2.471937417984009,
|
5068 |
+
"learning_rate": 5.852216748768473e-05,
|
5069 |
+
"loss": 0.2534,
|
5070 |
+
"step": 14460
|
5071 |
+
},
|
5072 |
+
{
|
5073 |
+
"epoch": 70.67724222086638,
|
5074 |
+
"grad_norm": 2.4318599700927734,
|
5075 |
+
"learning_rate": 5.832512315270936e-05,
|
5076 |
+
"loss": 0.2661,
|
5077 |
+
"step": 14480
|
5078 |
+
},
|
5079 |
+
{
|
5080 |
+
"epoch": 70.77486272117144,
|
5081 |
+
"grad_norm": 2.773090362548828,
|
5082 |
+
"learning_rate": 5.8128078817733986e-05,
|
5083 |
+
"loss": 0.283,
|
5084 |
+
"step": 14500
|
5085 |
+
},
|
5086 |
+
{
|
5087 |
+
"epoch": 70.87248322147651,
|
5088 |
+
"grad_norm": 2.120820999145508,
|
5089 |
+
"learning_rate": 5.7931034482758627e-05,
|
5090 |
+
"loss": 0.2668,
|
5091 |
+
"step": 14520
|
5092 |
+
},
|
5093 |
+
{
|
5094 |
+
"epoch": 70.97010372178157,
|
5095 |
+
"grad_norm": 2.614382028579712,
|
5096 |
+
"learning_rate": 5.7733990147783254e-05,
|
5097 |
+
"loss": 0.2722,
|
5098 |
+
"step": 14540
|
5099 |
+
},
|
5100 |
+
{
|
5101 |
+
"epoch": 71.06772422208664,
|
5102 |
+
"grad_norm": 2.954516649246216,
|
5103 |
+
"learning_rate": 5.753694581280789e-05,
|
5104 |
+
"loss": 0.2571,
|
5105 |
+
"step": 14560
|
5106 |
+
},
|
5107 |
+
{
|
5108 |
+
"epoch": 71.16534472239171,
|
5109 |
+
"grad_norm": 2.9351367950439453,
|
5110 |
+
"learning_rate": 5.7339901477832515e-05,
|
5111 |
+
"loss": 0.2659,
|
5112 |
+
"step": 14580
|
5113 |
+
},
|
5114 |
+
{
|
5115 |
+
"epoch": 71.26296522269676,
|
5116 |
+
"grad_norm": 2.757805347442627,
|
5117 |
+
"learning_rate": 5.714285714285714e-05,
|
5118 |
+
"loss": 0.2461,
|
5119 |
+
"step": 14600
|
5120 |
+
},
|
5121 |
+
{
|
5122 |
+
"epoch": 71.36058572300183,
|
5123 |
+
"grad_norm": 3.4546825885772705,
|
5124 |
+
"learning_rate": 5.694581280788177e-05,
|
5125 |
+
"loss": 0.2655,
|
5126 |
+
"step": 14620
|
5127 |
+
},
|
5128 |
+
{
|
5129 |
+
"epoch": 71.45820622330689,
|
5130 |
+
"grad_norm": 2.822056531906128,
|
5131 |
+
"learning_rate": 5.674876847290641e-05,
|
5132 |
+
"loss": 0.2542,
|
5133 |
+
"step": 14640
|
5134 |
+
},
|
5135 |
+
{
|
5136 |
+
"epoch": 71.55582672361196,
|
5137 |
+
"grad_norm": 2.4004786014556885,
|
5138 |
+
"learning_rate": 5.6551724137931037e-05,
|
5139 |
+
"loss": 0.2489,
|
5140 |
+
"step": 14660
|
5141 |
+
},
|
5142 |
+
{
|
5143 |
+
"epoch": 71.65344722391703,
|
5144 |
+
"grad_norm": 3.2715816497802734,
|
5145 |
+
"learning_rate": 5.635467980295567e-05,
|
5146 |
+
"loss": 0.2669,
|
5147 |
+
"step": 14680
|
5148 |
+
},
|
5149 |
+
{
|
5150 |
+
"epoch": 71.75106772422208,
|
5151 |
+
"grad_norm": 4.031295299530029,
|
5152 |
+
"learning_rate": 5.61576354679803e-05,
|
5153 |
+
"loss": 0.2729,
|
5154 |
+
"step": 14700
|
5155 |
+
},
|
5156 |
+
{
|
5157 |
+
"epoch": 71.84868822452715,
|
5158 |
+
"grad_norm": 3.0305051803588867,
|
5159 |
+
"learning_rate": 5.5960591133004925e-05,
|
5160 |
+
"loss": 0.272,
|
5161 |
+
"step": 14720
|
5162 |
+
},
|
5163 |
+
{
|
5164 |
+
"epoch": 71.94630872483222,
|
5165 |
+
"grad_norm": 2.170488119125366,
|
5166 |
+
"learning_rate": 5.5763546798029565e-05,
|
5167 |
+
"loss": 0.2771,
|
5168 |
+
"step": 14740
|
5169 |
+
},
|
5170 |
+
{
|
5171 |
+
"epoch": 72.04392922513728,
|
5172 |
+
"grad_norm": 2.89032244682312,
|
5173 |
+
"learning_rate": 5.556650246305419e-05,
|
5174 |
+
"loss": 0.2665,
|
5175 |
+
"step": 14760
|
5176 |
+
},
|
5177 |
+
{
|
5178 |
+
"epoch": 72.14154972544235,
|
5179 |
+
"grad_norm": 2.4803104400634766,
|
5180 |
+
"learning_rate": 5.536945812807882e-05,
|
5181 |
+
"loss": 0.2653,
|
5182 |
+
"step": 14780
|
5183 |
+
},
|
5184 |
+
{
|
5185 |
+
"epoch": 72.2391702257474,
|
5186 |
+
"grad_norm": 2.525521755218506,
|
5187 |
+
"learning_rate": 5.517241379310345e-05,
|
5188 |
+
"loss": 0.2595,
|
5189 |
+
"step": 14800
|
5190 |
+
},
|
5191 |
+
{
|
5192 |
+
"epoch": 72.33679072605247,
|
5193 |
+
"grad_norm": 2.121696710586548,
|
5194 |
+
"learning_rate": 5.497536945812808e-05,
|
5195 |
+
"loss": 0.2557,
|
5196 |
+
"step": 14820
|
5197 |
+
},
|
5198 |
+
{
|
5199 |
+
"epoch": 72.43441122635754,
|
5200 |
+
"grad_norm": 1.8344529867172241,
|
5201 |
+
"learning_rate": 5.477832512315271e-05,
|
5202 |
+
"loss": 0.2512,
|
5203 |
+
"step": 14840
|
5204 |
+
},
|
5205 |
+
{
|
5206 |
+
"epoch": 72.5320317266626,
|
5207 |
+
"grad_norm": 2.196624517440796,
|
5208 |
+
"learning_rate": 5.458128078817735e-05,
|
5209 |
+
"loss": 0.2474,
|
5210 |
+
"step": 14860
|
5211 |
+
},
|
5212 |
+
{
|
5213 |
+
"epoch": 72.62965222696766,
|
5214 |
+
"grad_norm": 3.387305974960327,
|
5215 |
+
"learning_rate": 5.4384236453201975e-05,
|
5216 |
+
"loss": 0.2696,
|
5217 |
+
"step": 14880
|
5218 |
+
},
|
5219 |
+
{
|
5220 |
+
"epoch": 72.72727272727273,
|
5221 |
+
"grad_norm": 2.481462240219116,
|
5222 |
+
"learning_rate": 5.41871921182266e-05,
|
5223 |
+
"loss": 0.2681,
|
5224 |
+
"step": 14900
|
5225 |
+
},
|
5226 |
+
{
|
5227 |
+
"epoch": 72.82489322757779,
|
5228 |
+
"grad_norm": 2.6742024421691895,
|
5229 |
+
"learning_rate": 5.399014778325123e-05,
|
5230 |
+
"loss": 0.2553,
|
5231 |
+
"step": 14920
|
5232 |
+
},
|
5233 |
+
{
|
5234 |
+
"epoch": 72.92251372788286,
|
5235 |
+
"grad_norm": 2.590111494064331,
|
5236 |
+
"learning_rate": 5.379310344827586e-05,
|
5237 |
+
"loss": 0.265,
|
5238 |
+
"step": 14940
|
5239 |
+
},
|
5240 |
+
{
|
5241 |
+
"epoch": 73.02013422818791,
|
5242 |
+
"grad_norm": 2.311305046081543,
|
5243 |
+
"learning_rate": 5.359605911330049e-05,
|
5244 |
+
"loss": 0.2644,
|
5245 |
+
"step": 14960
|
5246 |
+
},
|
5247 |
+
{
|
5248 |
+
"epoch": 73.11775472849298,
|
5249 |
+
"grad_norm": 2.502192974090576,
|
5250 |
+
"learning_rate": 5.339901477832513e-05,
|
5251 |
+
"loss": 0.2634,
|
5252 |
+
"step": 14980
|
5253 |
+
},
|
5254 |
+
{
|
5255 |
+
"epoch": 73.21537522879805,
|
5256 |
+
"grad_norm": 2.5767018795013428,
|
5257 |
+
"learning_rate": 5.320197044334976e-05,
|
5258 |
+
"loss": 0.2625,
|
5259 |
+
"step": 15000
|
5260 |
}
|
5261 |
],
|
5262 |
"logging_steps": 20,
|
|
|
5276 |
"attributes": {}
|
5277 |
}
|
5278 |
},
|
5279 |
+
"total_flos": 2.988930606140621e+17,
|
5280 |
"train_batch_size": 1,
|
5281 |
"trial_name": null,
|
5282 |
"trial_params": null
|