Commit
•
2a0c344
1
Parent(s):
6fa4453
Upload 11 files
Browse files- README.md +7 -1
- adapter_config.json +22 -0
- adapter_model.bin +3 -0
- optimizer (3).pt +3 -0
- rng_state (3).pth +3 -0
- scheduler (4).pt +3 -0
- special_tokens_map (6).json +7 -0
- tokenizer_config (6).json +5 -0
- trainer_state (3).json +1779 -0
- training_args (3).bin +3 -0
- vocab (6).txt +33 -0
README.md
CHANGED
|
@@ -1,3 +1,9 @@
|
|
| 1 |
---
|
| 2 |
-
|
| 3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
+
library_name: peft
|
| 3 |
---
|
| 4 |
+
## Training procedure
|
| 5 |
+
|
| 6 |
+
### Framework versions
|
| 7 |
+
|
| 8 |
+
|
| 9 |
+
- PEFT 0.5.0
|
adapter_config.json
ADDED
|
@@ -0,0 +1,22 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"auto_mapping": null,
|
| 3 |
+
"base_model_name_or_path": "facebook/esm2_t12_35M_UR50D",
|
| 4 |
+
"bias": "none",
|
| 5 |
+
"fan_in_fan_out": false,
|
| 6 |
+
"inference_mode": true,
|
| 7 |
+
"init_lora_weights": true,
|
| 8 |
+
"layers_pattern": null,
|
| 9 |
+
"layers_to_transform": null,
|
| 10 |
+
"lora_alpha": 1,
|
| 11 |
+
"lora_dropout": 0.2,
|
| 12 |
+
"modules_to_save": null,
|
| 13 |
+
"peft_type": "LORA",
|
| 14 |
+
"r": 2,
|
| 15 |
+
"revision": null,
|
| 16 |
+
"target_modules": [
|
| 17 |
+
"query",
|
| 18 |
+
"key",
|
| 19 |
+
"value"
|
| 20 |
+
],
|
| 21 |
+
"task_type": "TOKEN_CLS"
|
| 22 |
+
}
|
adapter_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a01f9ab0c7791ac8d9408740883f1b3267a15696f32e2bf1693bb76687206976
|
| 3 |
+
size 307151
|
optimizer (3).pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:05261168043270dab3324841624c1620dece04af673dd3bd900152c21cc5a274
|
| 3 |
+
size 622661
|
rng_state (3).pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7b0e7343e53950faee5bbd21ff23f65d19afc16fea38b3587c0fa23b173f0a4d
|
| 3 |
+
size 14575
|
scheduler (4).pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3fe8d20b8d75ad644ef533365859fd66ea86dde6d9371102aa318f521883b275
|
| 3 |
+
size 627
|
special_tokens_map (6).json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"cls_token": "<cls>",
|
| 3 |
+
"eos_token": "<eos>",
|
| 4 |
+
"mask_token": "<mask>",
|
| 5 |
+
"pad_token": "<pad>",
|
| 6 |
+
"unk_token": "<unk>"
|
| 7 |
+
}
|
tokenizer_config (6).json
ADDED
|
@@ -0,0 +1,5 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"clean_up_tokenization_spaces": true,
|
| 3 |
+
"model_max_length": 1024,
|
| 4 |
+
"tokenizer_class": "EsmTokenizer"
|
| 5 |
+
}
|
trainer_state (3).json
ADDED
|
@@ -0,0 +1,1779 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.3995036053381367,
|
| 3 |
+
"best_model_checkpoint": "esm2_t12_35M-lora-binding-sites_2023-09-11_16-28-57/checkpoint-57940",
|
| 4 |
+
"epoch": 2.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 57940,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.01,
|
| 13 |
+
"learning_rate": 0.0005701554508678478,
|
| 14 |
+
"loss": 0.4954,
|
| 15 |
+
"step": 200
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.01,
|
| 19 |
+
"learning_rate": 0.000570151359481901,
|
| 20 |
+
"loss": 0.3627,
|
| 21 |
+
"step": 400
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.02,
|
| 25 |
+
"learning_rate": 0.0005701445314265366,
|
| 26 |
+
"loss": 0.325,
|
| 27 |
+
"step": 600
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.03,
|
| 31 |
+
"learning_rate": 0.0005701349667673032,
|
| 32 |
+
"loss": 0.2797,
|
| 33 |
+
"step": 800
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.03,
|
| 37 |
+
"learning_rate": 0.0005701226655960202,
|
| 38 |
+
"loss": 0.2708,
|
| 39 |
+
"step": 1000
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.04,
|
| 43 |
+
"learning_rate": 0.0005701076280307777,
|
| 44 |
+
"loss": 0.2547,
|
| 45 |
+
"step": 1200
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.05,
|
| 49 |
+
"learning_rate": 0.0005700898542159346,
|
| 50 |
+
"loss": 0.2395,
|
| 51 |
+
"step": 1400
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.06,
|
| 55 |
+
"learning_rate": 0.0005700693443221178,
|
| 56 |
+
"loss": 0.2348,
|
| 57 |
+
"step": 1600
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.06,
|
| 61 |
+
"learning_rate": 0.0005700460985462198,
|
| 62 |
+
"loss": 0.218,
|
| 63 |
+
"step": 1800
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.07,
|
| 67 |
+
"learning_rate": 0.0005700202538231217,
|
| 68 |
+
"loss": 0.2284,
|
| 69 |
+
"step": 2000
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.08,
|
| 73 |
+
"learning_rate": 0.000569991550655178,
|
| 74 |
+
"loss": 0.2307,
|
| 75 |
+
"step": 2200
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 0.08,
|
| 79 |
+
"learning_rate": 0.000569960112351964,
|
| 80 |
+
"loss": 0.22,
|
| 81 |
+
"step": 2400
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.09,
|
| 85 |
+
"learning_rate": 0.000569925939215284,
|
| 86 |
+
"loss": 0.2117,
|
| 87 |
+
"step": 2600
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.1,
|
| 91 |
+
"learning_rate": 0.0005698890315731962,
|
| 92 |
+
"loss": 0.2246,
|
| 93 |
+
"step": 2800
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.1,
|
| 97 |
+
"learning_rate": 0.0005698493897800102,
|
| 98 |
+
"loss": 0.2392,
|
| 99 |
+
"step": 3000
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.11,
|
| 103 |
+
"learning_rate": 0.0005698070142162829,
|
| 104 |
+
"loss": 0.2078,
|
| 105 |
+
"step": 3200
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.12,
|
| 109 |
+
"learning_rate": 0.0005697619052888148,
|
| 110 |
+
"loss": 0.2098,
|
| 111 |
+
"step": 3400
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 0.12,
|
| 115 |
+
"learning_rate": 0.0005697140634306472,
|
| 116 |
+
"loss": 0.1919,
|
| 117 |
+
"step": 3600
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.13,
|
| 121 |
+
"learning_rate": 0.0005696634891010563,
|
| 122 |
+
"loss": 0.1957,
|
| 123 |
+
"step": 3800
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 0.14,
|
| 127 |
+
"learning_rate": 0.0005696101827855504,
|
| 128 |
+
"loss": 0.2156,
|
| 129 |
+
"step": 4000
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.14,
|
| 133 |
+
"learning_rate": 0.0005695544319784758,
|
| 134 |
+
"loss": 0.1983,
|
| 135 |
+
"step": 4200
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.15,
|
| 139 |
+
"learning_rate": 0.0005694956769058657,
|
| 140 |
+
"loss": 0.2104,
|
| 141 |
+
"step": 4400
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 0.16,
|
| 145 |
+
"learning_rate": 0.0005694341914583195,
|
| 146 |
+
"loss": 0.1789,
|
| 147 |
+
"step": 4600
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 0.17,
|
| 151 |
+
"learning_rate": 0.0005693699762260905,
|
| 152 |
+
"loss": 0.1944,
|
| 153 |
+
"step": 4800
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.17,
|
| 157 |
+
"learning_rate": 0.0005693030318256384,
|
| 158 |
+
"loss": 0.1895,
|
| 159 |
+
"step": 5000
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.18,
|
| 163 |
+
"learning_rate": 0.0005692333588996219,
|
| 164 |
+
"loss": 0.2007,
|
| 165 |
+
"step": 5200
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 0.19,
|
| 169 |
+
"learning_rate": 0.0005691609581168934,
|
| 170 |
+
"loss": 0.1852,
|
| 171 |
+
"step": 5400
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.19,
|
| 175 |
+
"learning_rate": 0.0005690858301724929,
|
| 176 |
+
"loss": 0.1728,
|
| 177 |
+
"step": 5600
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.2,
|
| 181 |
+
"learning_rate": 0.0005690079757876402,
|
| 182 |
+
"loss": 0.2076,
|
| 183 |
+
"step": 5800
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"epoch": 0.21,
|
| 187 |
+
"learning_rate": 0.0005689273957097291,
|
| 188 |
+
"loss": 0.1844,
|
| 189 |
+
"step": 6000
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.21,
|
| 193 |
+
"learning_rate": 0.0005688445140142318,
|
| 194 |
+
"loss": 0.173,
|
| 195 |
+
"step": 6200
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 0.22,
|
| 199 |
+
"learning_rate": 0.0005687584985156089,
|
| 200 |
+
"loss": 0.189,
|
| 201 |
+
"step": 6400
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 0.23,
|
| 205 |
+
"learning_rate": 0.0005686697597188824,
|
| 206 |
+
"loss": 0.1728,
|
| 207 |
+
"step": 6600
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.23,
|
| 211 |
+
"learning_rate": 0.0005685782984759349,
|
| 212 |
+
"loss": 0.1764,
|
| 213 |
+
"step": 6800
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.24,
|
| 217 |
+
"learning_rate": 0.0005684841156647846,
|
| 218 |
+
"loss": 0.1817,
|
| 219 |
+
"step": 7000
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.25,
|
| 223 |
+
"learning_rate": 0.0005683872121895759,
|
| 224 |
+
"loss": 0.184,
|
| 225 |
+
"step": 7200
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 0.26,
|
| 229 |
+
"learning_rate": 0.000568287588980572,
|
| 230 |
+
"loss": 0.1611,
|
| 231 |
+
"step": 7400
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"epoch": 0.26,
|
| 235 |
+
"learning_rate": 0.0005681852469941448,
|
| 236 |
+
"loss": 0.1782,
|
| 237 |
+
"step": 7600
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 0.27,
|
| 241 |
+
"learning_rate": 0.000568080187212766,
|
| 242 |
+
"loss": 0.1674,
|
| 243 |
+
"step": 7800
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.28,
|
| 247 |
+
"learning_rate": 0.0005679724106449985,
|
| 248 |
+
"loss": 0.1592,
|
| 249 |
+
"step": 8000
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.28,
|
| 253 |
+
"learning_rate": 0.0005678619183254852,
|
| 254 |
+
"loss": 0.188,
|
| 255 |
+
"step": 8200
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.29,
|
| 259 |
+
"learning_rate": 0.0005677487113149407,
|
| 260 |
+
"loss": 0.1724,
|
| 261 |
+
"step": 8400
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.3,
|
| 265 |
+
"learning_rate": 0.00056763279070014,
|
| 266 |
+
"loss": 0.1734,
|
| 267 |
+
"step": 8600
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"epoch": 0.3,
|
| 271 |
+
"learning_rate": 0.0005675141575939082,
|
| 272 |
+
"loss": 0.1802,
|
| 273 |
+
"step": 8800
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 0.31,
|
| 277 |
+
"learning_rate": 0.0005673934265999777,
|
| 278 |
+
"loss": 0.1735,
|
| 279 |
+
"step": 9000
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.32,
|
| 283 |
+
"learning_rate": 0.0005672693855015051,
|
| 284 |
+
"loss": 0.1768,
|
| 285 |
+
"step": 9200
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 0.32,
|
| 289 |
+
"learning_rate": 0.0005671426354002519,
|
| 290 |
+
"loss": 0.1759,
|
| 291 |
+
"step": 9400
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 0.33,
|
| 295 |
+
"learning_rate": 0.0005670131775130054,
|
| 296 |
+
"loss": 0.1646,
|
| 297 |
+
"step": 9600
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.34,
|
| 301 |
+
"learning_rate": 0.0005668810130825471,
|
| 302 |
+
"loss": 0.1684,
|
| 303 |
+
"step": 9800
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.35,
|
| 307 |
+
"learning_rate": 0.0005667461433776413,
|
| 308 |
+
"loss": 0.1636,
|
| 309 |
+
"step": 10000
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.35,
|
| 313 |
+
"learning_rate": 0.0005666085696930225,
|
| 314 |
+
"loss": 0.1492,
|
| 315 |
+
"step": 10200
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 0.36,
|
| 319 |
+
"learning_rate": 0.0005664682933493831,
|
| 320 |
+
"loss": 0.1636,
|
| 321 |
+
"step": 10400
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 0.37,
|
| 325 |
+
"learning_rate": 0.0005663253156933608,
|
| 326 |
+
"loss": 0.1647,
|
| 327 |
+
"step": 10600
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 0.37,
|
| 331 |
+
"learning_rate": 0.0005661796380975255,
|
| 332 |
+
"loss": 0.181,
|
| 333 |
+
"step": 10800
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.38,
|
| 337 |
+
"learning_rate": 0.0005660320105513283,
|
| 338 |
+
"loss": 0.1708,
|
| 339 |
+
"step": 11000
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.39,
|
| 343 |
+
"learning_rate": 0.0005658809507792387,
|
| 344 |
+
"loss": 0.1702,
|
| 345 |
+
"step": 11200
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.39,
|
| 349 |
+
"learning_rate": 0.0005657271953331912,
|
| 350 |
+
"loss": 0.1617,
|
| 351 |
+
"step": 11400
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.4,
|
| 355 |
+
"learning_rate": 0.000565570745689221,
|
| 356 |
+
"loss": 0.1722,
|
| 357 |
+
"step": 11600
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 0.41,
|
| 361 |
+
"learning_rate": 0.000565411603349228,
|
| 362 |
+
"loss": 0.1702,
|
| 363 |
+
"step": 11800
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 0.41,
|
| 367 |
+
"learning_rate": 0.0005652497698409615,
|
| 368 |
+
"loss": 0.1543,
|
| 369 |
+
"step": 12000
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.42,
|
| 373 |
+
"learning_rate": 0.0005650852467180052,
|
| 374 |
+
"loss": 0.1654,
|
| 375 |
+
"step": 12200
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 0.43,
|
| 379 |
+
"learning_rate": 0.0005649180355597633,
|
| 380 |
+
"loss": 0.1676,
|
| 381 |
+
"step": 12400
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.43,
|
| 385 |
+
"learning_rate": 0.000564748137971445,
|
| 386 |
+
"loss": 0.16,
|
| 387 |
+
"step": 12600
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.44,
|
| 391 |
+
"learning_rate": 0.0005645755555840486,
|
| 392 |
+
"loss": 0.1627,
|
| 393 |
+
"step": 12800
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 0.45,
|
| 397 |
+
"learning_rate": 0.0005644002900543462,
|
| 398 |
+
"loss": 0.1712,
|
| 399 |
+
"step": 13000
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.46,
|
| 403 |
+
"learning_rate": 0.0005642232394671313,
|
| 404 |
+
"loss": 0.1603,
|
| 405 |
+
"step": 13200
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 0.46,
|
| 409 |
+
"learning_rate": 0.0005640426261206144,
|
| 410 |
+
"loss": 0.1648,
|
| 411 |
+
"step": 13400
|
| 412 |
+
},
|
| 413 |
+
{
|
| 414 |
+
"epoch": 0.47,
|
| 415 |
+
"learning_rate": 0.0005638593347478565,
|
| 416 |
+
"loss": 0.1608,
|
| 417 |
+
"step": 13600
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 0.48,
|
| 421 |
+
"learning_rate": 0.0005636733671084348,
|
| 422 |
+
"loss": 0.162,
|
| 423 |
+
"step": 13800
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.48,
|
| 427 |
+
"learning_rate": 0.0005634847249876185,
|
| 428 |
+
"loss": 0.1516,
|
| 429 |
+
"step": 14000
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.49,
|
| 433 |
+
"learning_rate": 0.0005632934101963515,
|
| 434 |
+
"loss": 0.1638,
|
| 435 |
+
"step": 14200
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 0.5,
|
| 439 |
+
"learning_rate": 0.000563099424571235,
|
| 440 |
+
"loss": 0.1691,
|
| 441 |
+
"step": 14400
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 0.5,
|
| 445 |
+
"learning_rate": 0.0005629027699745098,
|
| 446 |
+
"loss": 0.149,
|
| 447 |
+
"step": 14600
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 0.51,
|
| 451 |
+
"learning_rate": 0.0005627034482940388,
|
| 452 |
+
"loss": 0.1713,
|
| 453 |
+
"step": 14800
|
| 454 |
+
},
|
| 455 |
+
{
|
| 456 |
+
"epoch": 0.52,
|
| 457 |
+
"learning_rate": 0.0005625014614432885,
|
| 458 |
+
"loss": 0.1615,
|
| 459 |
+
"step": 15000
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.52,
|
| 463 |
+
"learning_rate": 0.0005622978412332689,
|
| 464 |
+
"loss": 0.1618,
|
| 465 |
+
"step": 15200
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.53,
|
| 469 |
+
"learning_rate": 0.0005620905431860859,
|
| 470 |
+
"loss": 0.1695,
|
| 471 |
+
"step": 15400
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.54,
|
| 475 |
+
"learning_rate": 0.0005618805858524459,
|
| 476 |
+
"loss": 0.1575,
|
| 477 |
+
"step": 15600
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 0.55,
|
| 481 |
+
"learning_rate": 0.0005616679712479168,
|
| 482 |
+
"loss": 0.1494,
|
| 483 |
+
"step": 15800
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 0.55,
|
| 487 |
+
"learning_rate": 0.0005614527014135753,
|
| 488 |
+
"loss": 0.1441,
|
| 489 |
+
"step": 16000
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 0.56,
|
| 493 |
+
"learning_rate": 0.0005612347784159886,
|
| 494 |
+
"loss": 0.1554,
|
| 495 |
+
"step": 16200
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 0.57,
|
| 499 |
+
"learning_rate": 0.000561014204347194,
|
| 500 |
+
"loss": 0.1597,
|
| 501 |
+
"step": 16400
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 0.57,
|
| 505 |
+
"learning_rate": 0.0005607909813246783,
|
| 506 |
+
"loss": 0.1603,
|
| 507 |
+
"step": 16600
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.58,
|
| 511 |
+
"learning_rate": 0.0005605651114913586,
|
| 512 |
+
"loss": 0.1566,
|
| 513 |
+
"step": 16800
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.59,
|
| 517 |
+
"learning_rate": 0.0005603365970155604,
|
| 518 |
+
"loss": 0.148,
|
| 519 |
+
"step": 17000
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 0.59,
|
| 523 |
+
"learning_rate": 0.0005601054400909981,
|
| 524 |
+
"loss": 0.153,
|
| 525 |
+
"step": 17200
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"epoch": 0.6,
|
| 529 |
+
"learning_rate": 0.0005598728184863933,
|
| 530 |
+
"loss": 0.1551,
|
| 531 |
+
"step": 17400
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 0.61,
|
| 535 |
+
"learning_rate": 0.000559636396531194,
|
| 536 |
+
"loss": 0.1808,
|
| 537 |
+
"step": 17600
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 0.61,
|
| 541 |
+
"learning_rate": 0.0005593973388490791,
|
| 542 |
+
"loss": 0.1537,
|
| 543 |
+
"step": 17800
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 0.62,
|
| 547 |
+
"learning_rate": 0.0005591556477349764,
|
| 548 |
+
"loss": 0.1488,
|
| 549 |
+
"step": 18000
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.63,
|
| 553 |
+
"learning_rate": 0.0005589113255090943,
|
| 554 |
+
"loss": 0.1632,
|
| 555 |
+
"step": 18200
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.64,
|
| 559 |
+
"learning_rate": 0.0005586643745168998,
|
| 560 |
+
"loss": 0.1595,
|
| 561 |
+
"step": 18400
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 0.64,
|
| 565 |
+
"learning_rate": 0.0005584147971290952,
|
| 566 |
+
"loss": 0.1692,
|
| 567 |
+
"step": 18600
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 0.65,
|
| 571 |
+
"learning_rate": 0.0005581625957415966,
|
| 572 |
+
"loss": 0.1454,
|
| 573 |
+
"step": 18800
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 0.66,
|
| 577 |
+
"learning_rate": 0.0005579077727755101,
|
| 578 |
+
"loss": 0.1608,
|
| 579 |
+
"step": 19000
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 0.66,
|
| 583 |
+
"learning_rate": 0.0005576503306771081,
|
| 584 |
+
"loss": 0.1506,
|
| 585 |
+
"step": 19200
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 0.67,
|
| 589 |
+
"learning_rate": 0.0005573915787164281,
|
| 590 |
+
"loss": 0.1639,
|
| 591 |
+
"step": 19400
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 0.68,
|
| 595 |
+
"learning_rate": 0.000557128918857334,
|
| 596 |
+
"loss": 0.1463,
|
| 597 |
+
"step": 19600
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.68,
|
| 601 |
+
"learning_rate": 0.0005568636473428375,
|
| 602 |
+
"loss": 0.159,
|
| 603 |
+
"step": 19800
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 0.69,
|
| 607 |
+
"learning_rate": 0.0005565957667195164,
|
| 608 |
+
"loss": 0.1792,
|
| 609 |
+
"step": 20000
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 0.7,
|
| 613 |
+
"learning_rate": 0.0005563252795589956,
|
| 614 |
+
"loss": 0.1512,
|
| 615 |
+
"step": 20200
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 0.7,
|
| 619 |
+
"learning_rate": 0.0005560521884579225,
|
| 620 |
+
"loss": 0.1604,
|
| 621 |
+
"step": 20400
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 0.71,
|
| 625 |
+
"learning_rate": 0.000555776496037942,
|
| 626 |
+
"loss": 0.1563,
|
| 627 |
+
"step": 20600
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 0.72,
|
| 631 |
+
"learning_rate": 0.0005554982049456713,
|
| 632 |
+
"loss": 0.1598,
|
| 633 |
+
"step": 20800
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 0.72,
|
| 637 |
+
"learning_rate": 0.0005552173178526746,
|
| 638 |
+
"loss": 0.1429,
|
| 639 |
+
"step": 21000
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.73,
|
| 643 |
+
"learning_rate": 0.0005549338374554378,
|
| 644 |
+
"loss": 0.1544,
|
| 645 |
+
"step": 21200
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 0.74,
|
| 649 |
+
"learning_rate": 0.0005546477664753415,
|
| 650 |
+
"loss": 0.1435,
|
| 651 |
+
"step": 21400
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 0.75,
|
| 655 |
+
"learning_rate": 0.00055436055738539,
|
| 656 |
+
"loss": 0.1562,
|
| 657 |
+
"step": 21600
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 0.75,
|
| 661 |
+
"learning_rate": 0.0005540693264215625,
|
| 662 |
+
"loss": 0.1501,
|
| 663 |
+
"step": 21800
|
| 664 |
+
},
|
| 665 |
+
{
|
| 666 |
+
"epoch": 0.76,
|
| 667 |
+
"learning_rate": 0.0005537755131740878,
|
| 668 |
+
"loss": 0.1511,
|
| 669 |
+
"step": 22000
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 0.77,
|
| 673 |
+
"learning_rate": 0.0005534791204635411,
|
| 674 |
+
"loss": 0.1456,
|
| 675 |
+
"step": 22200
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.77,
|
| 679 |
+
"learning_rate": 0.0005531801511352604,
|
| 680 |
+
"loss": 0.1507,
|
| 681 |
+
"step": 22400
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.78,
|
| 685 |
+
"learning_rate": 0.0005528786080593183,
|
| 686 |
+
"loss": 0.1477,
|
| 687 |
+
"step": 22600
|
| 688 |
+
},
|
| 689 |
+
{
|
| 690 |
+
"epoch": 0.79,
|
| 691 |
+
"learning_rate": 0.000552574494130496,
|
| 692 |
+
"loss": 0.1507,
|
| 693 |
+
"step": 22800
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 0.79,
|
| 697 |
+
"learning_rate": 0.000552267812268254,
|
| 698 |
+
"loss": 0.1494,
|
| 699 |
+
"step": 23000
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 0.8,
|
| 703 |
+
"learning_rate": 0.0005519585654167048,
|
| 704 |
+
"loss": 0.1489,
|
| 705 |
+
"step": 23200
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 0.81,
|
| 709 |
+
"learning_rate": 0.0005516467565445848,
|
| 710 |
+
"loss": 0.1495,
|
| 711 |
+
"step": 23400
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 0.81,
|
| 715 |
+
"learning_rate": 0.0005513323886452253,
|
| 716 |
+
"loss": 0.1467,
|
| 717 |
+
"step": 23600
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 0.82,
|
| 721 |
+
"learning_rate": 0.0005510154647365238,
|
| 722 |
+
"loss": 0.1465,
|
| 723 |
+
"step": 23800
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 0.83,
|
| 727 |
+
"learning_rate": 0.0005506975915907354,
|
| 728 |
+
"loss": 0.1725,
|
| 729 |
+
"step": 24000
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 0.84,
|
| 733 |
+
"learning_rate": 0.0005503755775569938,
|
| 734 |
+
"loss": 0.1604,
|
| 735 |
+
"step": 24200
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 0.84,
|
| 739 |
+
"learning_rate": 0.0005500510166991929,
|
| 740 |
+
"loss": 0.1492,
|
| 741 |
+
"step": 24400
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 0.85,
|
| 745 |
+
"learning_rate": 0.0005497239121330815,
|
| 746 |
+
"loss": 0.1446,
|
| 747 |
+
"step": 24600
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 0.86,
|
| 751 |
+
"learning_rate": 0.0005493942669988278,
|
| 752 |
+
"loss": 0.1492,
|
| 753 |
+
"step": 24800
|
| 754 |
+
},
|
| 755 |
+
{
|
| 756 |
+
"epoch": 0.86,
|
| 757 |
+
"learning_rate": 0.0005490620844609896,
|
| 758 |
+
"loss": 0.152,
|
| 759 |
+
"step": 25000
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.87,
|
| 763 |
+
"learning_rate": 0.0005487273677084831,
|
| 764 |
+
"loss": 0.1601,
|
| 765 |
+
"step": 25200
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.88,
|
| 769 |
+
"learning_rate": 0.0005483901199545525,
|
| 770 |
+
"loss": 0.1432,
|
| 771 |
+
"step": 25400
|
| 772 |
+
},
|
| 773 |
+
{
|
| 774 |
+
"epoch": 0.88,
|
| 775 |
+
"learning_rate": 0.0005480503444367399,
|
| 776 |
+
"loss": 0.153,
|
| 777 |
+
"step": 25600
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"epoch": 0.89,
|
| 781 |
+
"learning_rate": 0.0005477080444168533,
|
| 782 |
+
"loss": 0.1446,
|
| 783 |
+
"step": 25800
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 0.9,
|
| 787 |
+
"learning_rate": 0.0005473649535531743,
|
| 788 |
+
"loss": 0.1543,
|
| 789 |
+
"step": 26000
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 0.9,
|
| 793 |
+
"learning_rate": 0.0005470176269927267,
|
| 794 |
+
"loss": 0.1652,
|
| 795 |
+
"step": 26200
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 0.91,
|
| 799 |
+
"learning_rate": 0.0005466677858441803,
|
| 800 |
+
"loss": 0.1463,
|
| 801 |
+
"step": 26400
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 0.92,
|
| 805 |
+
"learning_rate": 0.0005463154334659722,
|
| 806 |
+
"loss": 0.152,
|
| 807 |
+
"step": 26600
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.93,
|
| 811 |
+
"learning_rate": 0.0005459605732406469,
|
| 812 |
+
"loss": 0.1572,
|
| 813 |
+
"step": 26800
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"epoch": 0.93,
|
| 817 |
+
"learning_rate": 0.0005456032085748239,
|
| 818 |
+
"loss": 0.1439,
|
| 819 |
+
"step": 27000
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 0.94,
|
| 823 |
+
"learning_rate": 0.0005452433428991655,
|
| 824 |
+
"loss": 0.1483,
|
| 825 |
+
"step": 27200
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 0.95,
|
| 829 |
+
"learning_rate": 0.000544880979668343,
|
| 830 |
+
"loss": 0.1633,
|
| 831 |
+
"step": 27400
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 0.95,
|
| 835 |
+
"learning_rate": 0.0005445161223610037,
|
| 836 |
+
"loss": 0.1559,
|
| 837 |
+
"step": 27600
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 0.96,
|
| 841 |
+
"learning_rate": 0.0005441487744797385,
|
| 842 |
+
"loss": 0.1582,
|
| 843 |
+
"step": 27800
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 0.97,
|
| 847 |
+
"learning_rate": 0.000543778939551047,
|
| 848 |
+
"loss": 0.1607,
|
| 849 |
+
"step": 28000
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.97,
|
| 853 |
+
"learning_rate": 0.0005434084888892294,
|
| 854 |
+
"loss": 0.1444,
|
| 855 |
+
"step": 28200
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 0.98,
|
| 859 |
+
"learning_rate": 0.0005430337029313361,
|
| 860 |
+
"loss": 0.1411,
|
| 861 |
+
"step": 28400
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"epoch": 0.99,
|
| 865 |
+
"learning_rate": 0.0005426564406305822,
|
| 866 |
+
"loss": 0.145,
|
| 867 |
+
"step": 28600
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 0.99,
|
| 871 |
+
"learning_rate": 0.0005422767056086448,
|
| 872 |
+
"loss": 0.1543,
|
| 873 |
+
"step": 28800
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 1.0,
|
| 877 |
+
"eval_accuracy": 0.9266715807717144,
|
| 878 |
+
"eval_auc": 0.8547374969101048,
|
| 879 |
+
"eval_f1": 0.3476562916154915,
|
| 880 |
+
"eval_loss": 0.36462366580963135,
|
| 881 |
+
"eval_precision": 0.22375778821030992,
|
| 882 |
+
"eval_recall": 0.7790042059793111,
|
| 883 |
+
"eval_runtime": 1778.9025,
|
| 884 |
+
"eval_samples_per_second": 24.978,
|
| 885 |
+
"eval_steps_per_second": 4.163,
|
| 886 |
+
"step": 28970
|
| 887 |
+
},
|
| 888 |
+
{
|
| 889 |
+
"epoch": 1.0,
|
| 890 |
+
"learning_rate": 0.000541894501510939,
|
| 891 |
+
"loss": 0.1532,
|
| 892 |
+
"step": 29000
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 1.01,
|
| 896 |
+
"learning_rate": 0.0005415098320065826,
|
| 897 |
+
"loss": 0.146,
|
| 898 |
+
"step": 29200
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 1.01,
|
| 902 |
+
"learning_rate": 0.0005411227007883609,
|
| 903 |
+
"loss": 0.142,
|
| 904 |
+
"step": 29400
|
| 905 |
+
},
|
| 906 |
+
{
|
| 907 |
+
"epoch": 1.02,
|
| 908 |
+
"learning_rate": 0.0005407331115726918,
|
| 909 |
+
"loss": 0.1398,
|
| 910 |
+
"step": 29600
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"epoch": 1.03,
|
| 914 |
+
"learning_rate": 0.0005403410680995892,
|
| 915 |
+
"loss": 0.1609,
|
| 916 |
+
"step": 29800
|
| 917 |
+
},
|
| 918 |
+
{
|
| 919 |
+
"epoch": 1.04,
|
| 920 |
+
"learning_rate": 0.000539946574132628,
|
| 921 |
+
"loss": 0.1443,
|
| 922 |
+
"step": 30000
|
| 923 |
+
},
|
| 924 |
+
{
|
| 925 |
+
"epoch": 1.04,
|
| 926 |
+
"learning_rate": 0.0005395496334589074,
|
| 927 |
+
"loss": 0.1466,
|
| 928 |
+
"step": 30200
|
| 929 |
+
},
|
| 930 |
+
{
|
| 931 |
+
"epoch": 1.05,
|
| 932 |
+
"learning_rate": 0.0005391502498890146,
|
| 933 |
+
"loss": 0.1542,
|
| 934 |
+
"step": 30400
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 1.06,
|
| 938 |
+
"learning_rate": 0.0005387504424309447,
|
| 939 |
+
"loss": 0.1624,
|
| 940 |
+
"step": 30600
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 1.06,
|
| 944 |
+
"learning_rate": 0.0005383461967606286,
|
| 945 |
+
"loss": 0.1492,
|
| 946 |
+
"step": 30800
|
| 947 |
+
},
|
| 948 |
+
{
|
| 949 |
+
"epoch": 1.07,
|
| 950 |
+
"learning_rate": 0.000537939519747002,
|
| 951 |
+
"loss": 0.1481,
|
| 952 |
+
"step": 31000
|
| 953 |
+
},
|
| 954 |
+
{
|
| 955 |
+
"epoch": 1.08,
|
| 956 |
+
"learning_rate": 0.0005375304152941202,
|
| 957 |
+
"loss": 0.1534,
|
| 958 |
+
"step": 31200
|
| 959 |
+
},
|
| 960 |
+
{
|
| 961 |
+
"epoch": 1.08,
|
| 962 |
+
"learning_rate": 0.0005371188873293412,
|
| 963 |
+
"loss": 0.1469,
|
| 964 |
+
"step": 31400
|
| 965 |
+
},
|
| 966 |
+
{
|
| 967 |
+
"epoch": 1.09,
|
| 968 |
+
"learning_rate": 0.000536704939803289,
|
| 969 |
+
"loss": 0.1443,
|
| 970 |
+
"step": 31600
|
| 971 |
+
},
|
| 972 |
+
{
|
| 973 |
+
"epoch": 1.1,
|
| 974 |
+
"learning_rate": 0.0005362885766898149,
|
| 975 |
+
"loss": 0.148,
|
| 976 |
+
"step": 31800
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 1.1,
|
| 980 |
+
"learning_rate": 0.0005358698019859594,
|
| 981 |
+
"loss": 0.1589,
|
| 982 |
+
"step": 32000
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 1.11,
|
| 986 |
+
"learning_rate": 0.0005354486197119143,
|
| 987 |
+
"loss": 0.1525,
|
| 988 |
+
"step": 32200
|
| 989 |
+
},
|
| 990 |
+
{
|
| 991 |
+
"epoch": 1.12,
|
| 992 |
+
"learning_rate": 0.0005350250339109836,
|
| 993 |
+
"loss": 0.1514,
|
| 994 |
+
"step": 32400
|
| 995 |
+
},
|
| 996 |
+
{
|
| 997 |
+
"epoch": 1.13,
|
| 998 |
+
"learning_rate": 0.0005346011845377551,
|
| 999 |
+
"loss": 0.1421,
|
| 1000 |
+
"step": 32600
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"epoch": 1.13,
|
| 1004 |
+
"learning_rate": 0.0005341728158718659,
|
| 1005 |
+
"loss": 0.1384,
|
| 1006 |
+
"step": 32800
|
| 1007 |
+
},
|
| 1008 |
+
{
|
| 1009 |
+
"epoch": 1.14,
|
| 1010 |
+
"learning_rate": 0.0005337442156678716,
|
| 1011 |
+
"loss": 0.1629,
|
| 1012 |
+
"step": 33000
|
| 1013 |
+
},
|
| 1014 |
+
{
|
| 1015 |
+
"epoch": 1.15,
|
| 1016 |
+
"learning_rate": 0.0005333110805040125,
|
| 1017 |
+
"loss": 0.1488,
|
| 1018 |
+
"step": 33200
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 1.15,
|
| 1022 |
+
"learning_rate": 0.000532875562333412,
|
| 1023 |
+
"loss": 0.1424,
|
| 1024 |
+
"step": 33400
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 1.16,
|
| 1028 |
+
"learning_rate": 0.0005324376653369972,
|
| 1029 |
+
"loss": 0.1436,
|
| 1030 |
+
"step": 33600
|
| 1031 |
+
},
|
| 1032 |
+
{
|
| 1033 |
+
"epoch": 1.17,
|
| 1034 |
+
"learning_rate": 0.0005319973937185317,
|
| 1035 |
+
"loss": 0.1624,
|
| 1036 |
+
"step": 33800
|
| 1037 |
+
},
|
| 1038 |
+
{
|
| 1039 |
+
"epoch": 1.17,
|
| 1040 |
+
"learning_rate": 0.0005315547517045754,
|
| 1041 |
+
"loss": 0.1381,
|
| 1042 |
+
"step": 34000
|
| 1043 |
+
},
|
| 1044 |
+
{
|
| 1045 |
+
"epoch": 1.18,
|
| 1046 |
+
"learning_rate": 0.0005311097435444433,
|
| 1047 |
+
"loss": 0.1484,
|
| 1048 |
+
"step": 34200
|
| 1049 |
+
},
|
| 1050 |
+
{
|
| 1051 |
+
"epoch": 1.19,
|
| 1052 |
+
"learning_rate": 0.0005306623735101656,
|
| 1053 |
+
"loss": 0.1482,
|
| 1054 |
+
"step": 34400
|
| 1055 |
+
},
|
| 1056 |
+
{
|
| 1057 |
+
"epoch": 1.19,
|
| 1058 |
+
"learning_rate": 0.0005302126458964461,
|
| 1059 |
+
"loss": 0.1439,
|
| 1060 |
+
"step": 34600
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 1.2,
|
| 1064 |
+
"learning_rate": 0.0005297605650206207,
|
| 1065 |
+
"loss": 0.1519,
|
| 1066 |
+
"step": 34800
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 1.21,
|
| 1070 |
+
"learning_rate": 0.000529306135222617,
|
| 1071 |
+
"loss": 0.1298,
|
| 1072 |
+
"step": 35000
|
| 1073 |
+
},
|
| 1074 |
+
{
|
| 1075 |
+
"epoch": 1.22,
|
| 1076 |
+
"learning_rate": 0.0005288493608649115,
|
| 1077 |
+
"loss": 0.1469,
|
| 1078 |
+
"step": 35200
|
| 1079 |
+
},
|
| 1080 |
+
{
|
| 1081 |
+
"epoch": 1.22,
|
| 1082 |
+
"learning_rate": 0.0005283902463324885,
|
| 1083 |
+
"loss": 0.1431,
|
| 1084 |
+
"step": 35400
|
| 1085 |
+
},
|
| 1086 |
+
{
|
| 1087 |
+
"epoch": 1.23,
|
| 1088 |
+
"learning_rate": 0.0005279287960327977,
|
| 1089 |
+
"loss": 0.1501,
|
| 1090 |
+
"step": 35600
|
| 1091 |
+
},
|
| 1092 |
+
{
|
| 1093 |
+
"epoch": 1.24,
|
| 1094 |
+
"learning_rate": 0.0005274650143957118,
|
| 1095 |
+
"loss": 0.142,
|
| 1096 |
+
"step": 35800
|
| 1097 |
+
},
|
| 1098 |
+
{
|
| 1099 |
+
"epoch": 1.24,
|
| 1100 |
+
"learning_rate": 0.0005269989058734842,
|
| 1101 |
+
"loss": 0.1433,
|
| 1102 |
+
"step": 36000
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 1.25,
|
| 1106 |
+
"learning_rate": 0.0005265304749407059,
|
| 1107 |
+
"loss": 0.1398,
|
| 1108 |
+
"step": 36200
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 1.26,
|
| 1112 |
+
"learning_rate": 0.0005260597260942631,
|
| 1113 |
+
"loss": 0.1553,
|
| 1114 |
+
"step": 36400
|
| 1115 |
+
},
|
| 1116 |
+
{
|
| 1117 |
+
"epoch": 1.26,
|
| 1118 |
+
"learning_rate": 0.0005255866638532936,
|
| 1119 |
+
"loss": 0.1545,
|
| 1120 |
+
"step": 36600
|
| 1121 |
+
},
|
| 1122 |
+
{
|
| 1123 |
+
"epoch": 1.27,
|
| 1124 |
+
"learning_rate": 0.0005251112927591435,
|
| 1125 |
+
"loss": 0.1526,
|
| 1126 |
+
"step": 36800
|
| 1127 |
+
},
|
| 1128 |
+
{
|
| 1129 |
+
"epoch": 1.28,
|
| 1130 |
+
"learning_rate": 0.0005246336173753237,
|
| 1131 |
+
"loss": 0.1503,
|
| 1132 |
+
"step": 37000
|
| 1133 |
+
},
|
| 1134 |
+
{
|
| 1135 |
+
"epoch": 1.28,
|
| 1136 |
+
"learning_rate": 0.000524153642287466,
|
| 1137 |
+
"loss": 0.139,
|
| 1138 |
+
"step": 37200
|
| 1139 |
+
},
|
| 1140 |
+
{
|
| 1141 |
+
"epoch": 1.29,
|
| 1142 |
+
"learning_rate": 0.0005236737891556027,
|
| 1143 |
+
"loss": 0.1486,
|
| 1144 |
+
"step": 37400
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 1.3,
|
| 1148 |
+
"learning_rate": 0.000523189239945608,
|
| 1149 |
+
"loss": 0.1375,
|
| 1150 |
+
"step": 37600
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 1.3,
|
| 1154 |
+
"learning_rate": 0.0005227024048974417,
|
| 1155 |
+
"loss": 0.142,
|
| 1156 |
+
"step": 37800
|
| 1157 |
+
},
|
| 1158 |
+
{
|
| 1159 |
+
"epoch": 1.31,
|
| 1160 |
+
"learning_rate": 0.0005222132886846675,
|
| 1161 |
+
"loss": 0.1412,
|
| 1162 |
+
"step": 38000
|
| 1163 |
+
},
|
| 1164 |
+
{
|
| 1165 |
+
"epoch": 1.32,
|
| 1166 |
+
"learning_rate": 0.000521721896002748,
|
| 1167 |
+
"loss": 0.1485,
|
| 1168 |
+
"step": 38200
|
| 1169 |
+
},
|
| 1170 |
+
{
|
| 1171 |
+
"epoch": 1.33,
|
| 1172 |
+
"learning_rate": 0.0005212282315689993,
|
| 1173 |
+
"loss": 0.1416,
|
| 1174 |
+
"step": 38400
|
| 1175 |
+
},
|
| 1176 |
+
{
|
| 1177 |
+
"epoch": 1.33,
|
| 1178 |
+
"learning_rate": 0.0005207323001225466,
|
| 1179 |
+
"loss": 0.1385,
|
| 1180 |
+
"step": 38600
|
| 1181 |
+
},
|
| 1182 |
+
{
|
| 1183 |
+
"epoch": 1.34,
|
| 1184 |
+
"learning_rate": 0.0005202341064242777,
|
| 1185 |
+
"loss": 0.1493,
|
| 1186 |
+
"step": 38800
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 1.35,
|
| 1190 |
+
"learning_rate": 0.0005197336552567981,
|
| 1191 |
+
"loss": 0.1454,
|
| 1192 |
+
"step": 39000
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 1.35,
|
| 1196 |
+
"learning_rate": 0.0005192309514243847,
|
| 1197 |
+
"loss": 0.1472,
|
| 1198 |
+
"step": 39200
|
| 1199 |
+
},
|
| 1200 |
+
{
|
| 1201 |
+
"epoch": 1.36,
|
| 1202 |
+
"learning_rate": 0.0005187259997529396,
|
| 1203 |
+
"loss": 0.1443,
|
| 1204 |
+
"step": 39400
|
| 1205 |
+
},
|
| 1206 |
+
{
|
| 1207 |
+
"epoch": 1.37,
|
| 1208 |
+
"learning_rate": 0.000518218805089944,
|
| 1209 |
+
"loss": 0.147,
|
| 1210 |
+
"step": 39600
|
| 1211 |
+
},
|
| 1212 |
+
{
|
| 1213 |
+
"epoch": 1.37,
|
| 1214 |
+
"learning_rate": 0.0005177093723044116,
|
| 1215 |
+
"loss": 0.1565,
|
| 1216 |
+
"step": 39800
|
| 1217 |
+
},
|
| 1218 |
+
{
|
| 1219 |
+
"epoch": 1.38,
|
| 1220 |
+
"learning_rate": 0.000517200270163978,
|
| 1221 |
+
"loss": 0.153,
|
| 1222 |
+
"step": 40000
|
| 1223 |
+
},
|
| 1224 |
+
{
|
| 1225 |
+
"epoch": 1.39,
|
| 1226 |
+
"learning_rate": 0.0005166863869556556,
|
| 1227 |
+
"loss": 0.1404,
|
| 1228 |
+
"step": 40200
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 1.39,
|
| 1232 |
+
"learning_rate": 0.000516170280335844,
|
| 1233 |
+
"loss": 0.1326,
|
| 1234 |
+
"step": 40400
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 1.4,
|
| 1238 |
+
"learning_rate": 0.0005156519552591106,
|
| 1239 |
+
"loss": 0.147,
|
| 1240 |
+
"step": 40600
|
| 1241 |
+
},
|
| 1242 |
+
{
|
| 1243 |
+
"epoch": 1.41,
|
| 1244 |
+
"learning_rate": 0.0005151314167013202,
|
| 1245 |
+
"loss": 0.151,
|
| 1246 |
+
"step": 40800
|
| 1247 |
+
},
|
| 1248 |
+
{
|
| 1249 |
+
"epoch": 1.42,
|
| 1250 |
+
"learning_rate": 0.000514608669659586,
|
| 1251 |
+
"loss": 0.1479,
|
| 1252 |
+
"step": 41000
|
| 1253 |
+
},
|
| 1254 |
+
{
|
| 1255 |
+
"epoch": 1.42,
|
| 1256 |
+
"learning_rate": 0.0005140837191522229,
|
| 1257 |
+
"loss": 0.1621,
|
| 1258 |
+
"step": 41200
|
| 1259 |
+
},
|
| 1260 |
+
{
|
| 1261 |
+
"epoch": 1.43,
|
| 1262 |
+
"learning_rate": 0.0005135565702186988,
|
| 1263 |
+
"loss": 0.1531,
|
| 1264 |
+
"step": 41400
|
| 1265 |
+
},
|
| 1266 |
+
{
|
| 1267 |
+
"epoch": 1.44,
|
| 1268 |
+
"learning_rate": 0.0005130272279195864,
|
| 1269 |
+
"loss": 0.1364,
|
| 1270 |
+
"step": 41600
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 1.44,
|
| 1274 |
+
"learning_rate": 0.0005124956973365139,
|
| 1275 |
+
"loss": 0.1419,
|
| 1276 |
+
"step": 41800
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 1.45,
|
| 1280 |
+
"learning_rate": 0.0005119619835721174,
|
| 1281 |
+
"loss": 0.1305,
|
| 1282 |
+
"step": 42000
|
| 1283 |
+
},
|
| 1284 |
+
{
|
| 1285 |
+
"epoch": 1.46,
|
| 1286 |
+
"learning_rate": 0.000511426091749991,
|
| 1287 |
+
"loss": 0.1461,
|
| 1288 |
+
"step": 42200
|
| 1289 |
+
},
|
| 1290 |
+
{
|
| 1291 |
+
"epoch": 1.46,
|
| 1292 |
+
"learning_rate": 0.0005108880270146381,
|
| 1293 |
+
"loss": 0.1501,
|
| 1294 |
+
"step": 42400
|
| 1295 |
+
},
|
| 1296 |
+
{
|
| 1297 |
+
"epoch": 1.47,
|
| 1298 |
+
"learning_rate": 0.0005103505010775399,
|
| 1299 |
+
"loss": 0.1587,
|
| 1300 |
+
"step": 42600
|
| 1301 |
+
},
|
| 1302 |
+
{
|
| 1303 |
+
"epoch": 1.48,
|
| 1304 |
+
"learning_rate": 0.0005098081168325064,
|
| 1305 |
+
"loss": 0.1322,
|
| 1306 |
+
"step": 42800
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 1.48,
|
| 1310 |
+
"learning_rate": 0.0005092635752066296,
|
| 1311 |
+
"loss": 0.1418,
|
| 1312 |
+
"step": 43000
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 1.49,
|
| 1316 |
+
"learning_rate": 0.0005087196202411544,
|
| 1317 |
+
"loss": 0.1564,
|
| 1318 |
+
"step": 43200
|
| 1319 |
+
},
|
| 1320 |
+
{
|
| 1321 |
+
"epoch": 1.5,
|
| 1322 |
+
"learning_rate": 0.0005081707902783097,
|
| 1323 |
+
"loss": 0.1552,
|
| 1324 |
+
"step": 43400
|
| 1325 |
+
},
|
| 1326 |
+
{
|
| 1327 |
+
"epoch": 1.51,
|
| 1328 |
+
"learning_rate": 0.0005076198186527788,
|
| 1329 |
+
"loss": 0.1362,
|
| 1330 |
+
"step": 43600
|
| 1331 |
+
},
|
| 1332 |
+
{
|
| 1333 |
+
"epoch": 1.51,
|
| 1334 |
+
"learning_rate": 0.0005070667106538294,
|
| 1335 |
+
"loss": 0.1497,
|
| 1336 |
+
"step": 43800
|
| 1337 |
+
},
|
| 1338 |
+
{
|
| 1339 |
+
"epoch": 1.52,
|
| 1340 |
+
"learning_rate": 0.0005065114715912386,
|
| 1341 |
+
"loss": 0.143,
|
| 1342 |
+
"step": 44000
|
| 1343 |
+
},
|
| 1344 |
+
{
|
| 1345 |
+
"epoch": 1.53,
|
| 1346 |
+
"learning_rate": 0.0005059541067952407,
|
| 1347 |
+
"loss": 0.1527,
|
| 1348 |
+
"step": 44200
|
| 1349 |
+
},
|
| 1350 |
+
{
|
| 1351 |
+
"epoch": 1.53,
|
| 1352 |
+
"learning_rate": 0.0005053946216164774,
|
| 1353 |
+
"loss": 0.1384,
|
| 1354 |
+
"step": 44400
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 1.54,
|
| 1358 |
+
"learning_rate": 0.0005048330214259455,
|
| 1359 |
+
"loss": 0.1569,
|
| 1360 |
+
"step": 44600
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 1.55,
|
| 1364 |
+
"learning_rate": 0.0005042693116149461,
|
| 1365 |
+
"loss": 0.1349,
|
| 1366 |
+
"step": 44800
|
| 1367 |
+
},
|
| 1368 |
+
{
|
| 1369 |
+
"epoch": 1.55,
|
| 1370 |
+
"learning_rate": 0.0005037034975950318,
|
| 1371 |
+
"loss": 0.1458,
|
| 1372 |
+
"step": 45000
|
| 1373 |
+
},
|
| 1374 |
+
{
|
| 1375 |
+
"epoch": 1.56,
|
| 1376 |
+
"learning_rate": 0.0005031355847979559,
|
| 1377 |
+
"loss": 0.1448,
|
| 1378 |
+
"step": 45200
|
| 1379 |
+
},
|
| 1380 |
+
{
|
| 1381 |
+
"epoch": 1.57,
|
| 1382 |
+
"learning_rate": 0.0005025655786756196,
|
| 1383 |
+
"loss": 0.1386,
|
| 1384 |
+
"step": 45400
|
| 1385 |
+
},
|
| 1386 |
+
{
|
| 1387 |
+
"epoch": 1.57,
|
| 1388 |
+
"learning_rate": 0.0005019934847000197,
|
| 1389 |
+
"loss": 0.146,
|
| 1390 |
+
"step": 45600
|
| 1391 |
+
},
|
| 1392 |
+
{
|
| 1393 |
+
"epoch": 1.58,
|
| 1394 |
+
"learning_rate": 0.0005014193083631962,
|
| 1395 |
+
"loss": 0.1366,
|
| 1396 |
+
"step": 45800
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 1.59,
|
| 1400 |
+
"learning_rate": 0.0005008430551771795,
|
| 1401 |
+
"loss": 0.1389,
|
| 1402 |
+
"step": 46000
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 1.59,
|
| 1406 |
+
"learning_rate": 0.0005002647306739379,
|
| 1407 |
+
"loss": 0.1378,
|
| 1408 |
+
"step": 46200
|
| 1409 |
+
},
|
| 1410 |
+
{
|
| 1411 |
+
"epoch": 1.6,
|
| 1412 |
+
"learning_rate": 0.0004996843404053234,
|
| 1413 |
+
"loss": 0.1375,
|
| 1414 |
+
"step": 46400
|
| 1415 |
+
},
|
| 1416 |
+
{
|
| 1417 |
+
"epoch": 1.61,
|
| 1418 |
+
"learning_rate": 0.0004991018899430197,
|
| 1419 |
+
"loss": 0.1424,
|
| 1420 |
+
"step": 46600
|
| 1421 |
+
},
|
| 1422 |
+
{
|
| 1423 |
+
"epoch": 1.62,
|
| 1424 |
+
"learning_rate": 0.0004985173848784881,
|
| 1425 |
+
"loss": 0.1509,
|
| 1426 |
+
"step": 46800
|
| 1427 |
+
},
|
| 1428 |
+
{
|
| 1429 |
+
"epoch": 1.62,
|
| 1430 |
+
"learning_rate": 0.0004979308308229137,
|
| 1431 |
+
"loss": 0.1438,
|
| 1432 |
+
"step": 47000
|
| 1433 |
+
},
|
| 1434 |
+
{
|
| 1435 |
+
"epoch": 1.63,
|
| 1436 |
+
"learning_rate": 0.0004973422334071515,
|
| 1437 |
+
"loss": 0.1306,
|
| 1438 |
+
"step": 47200
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 1.64,
|
| 1442 |
+
"learning_rate": 0.000496754556516732,
|
| 1443 |
+
"loss": 0.1475,
|
| 1444 |
+
"step": 47400
|
| 1445 |
+
},
|
| 1446 |
+
{
|
| 1447 |
+
"epoch": 1.64,
|
| 1448 |
+
"learning_rate": 0.0004961618994976319,
|
| 1449 |
+
"loss": 0.1422,
|
| 1450 |
+
"step": 47600
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 1.65,
|
| 1454 |
+
"learning_rate": 0.0004955672160998928,
|
| 1455 |
+
"loss": 0.1411,
|
| 1456 |
+
"step": 47800
|
| 1457 |
+
},
|
| 1458 |
+
{
|
| 1459 |
+
"epoch": 1.66,
|
| 1460 |
+
"learning_rate": 0.0004949705120324106,
|
| 1461 |
+
"loss": 0.1491,
|
| 1462 |
+
"step": 48000
|
| 1463 |
+
},
|
| 1464 |
+
{
|
| 1465 |
+
"epoch": 1.66,
|
| 1466 |
+
"learning_rate": 0.0004943717930234795,
|
| 1467 |
+
"loss": 0.1391,
|
| 1468 |
+
"step": 48200
|
| 1469 |
+
},
|
| 1470 |
+
{
|
| 1471 |
+
"epoch": 1.67,
|
| 1472 |
+
"learning_rate": 0.0004937710648207372,
|
| 1473 |
+
"loss": 0.1339,
|
| 1474 |
+
"step": 48400
|
| 1475 |
+
},
|
| 1476 |
+
{
|
| 1477 |
+
"epoch": 1.68,
|
| 1478 |
+
"learning_rate": 0.0004931683331911087,
|
| 1479 |
+
"loss": 0.1494,
|
| 1480 |
+
"step": 48600
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 1.68,
|
| 1484 |
+
"learning_rate": 0.0004925636039207527,
|
| 1485 |
+
"loss": 0.154,
|
| 1486 |
+
"step": 48800
|
| 1487 |
+
},
|
| 1488 |
+
{
|
| 1489 |
+
"epoch": 1.69,
|
| 1490 |
+
"learning_rate": 0.0004919568828150041,
|
| 1491 |
+
"loss": 0.1423,
|
| 1492 |
+
"step": 49000
|
| 1493 |
+
},
|
| 1494 |
+
{
|
| 1495 |
+
"epoch": 1.7,
|
| 1496 |
+
"learning_rate": 0.0004913481756983199,
|
| 1497 |
+
"loss": 0.1407,
|
| 1498 |
+
"step": 49200
|
| 1499 |
+
},
|
| 1500 |
+
{
|
| 1501 |
+
"epoch": 1.71,
|
| 1502 |
+
"learning_rate": 0.0004907405467666192,
|
| 1503 |
+
"loss": 0.1452,
|
| 1504 |
+
"step": 49400
|
| 1505 |
+
},
|
| 1506 |
+
{
|
| 1507 |
+
"epoch": 1.71,
|
| 1508 |
+
"learning_rate": 0.0004901278950345495,
|
| 1509 |
+
"loss": 0.1368,
|
| 1510 |
+
"step": 49600
|
| 1511 |
+
},
|
| 1512 |
+
{
|
| 1513 |
+
"epoch": 1.72,
|
| 1514 |
+
"learning_rate": 0.0004895132748496281,
|
| 1515 |
+
"loss": 0.1264,
|
| 1516 |
+
"step": 49800
|
| 1517 |
+
},
|
| 1518 |
+
{
|
| 1519 |
+
"epoch": 1.73,
|
| 1520 |
+
"learning_rate": 0.0004888966921121421,
|
| 1521 |
+
"loss": 0.1478,
|
| 1522 |
+
"step": 50000
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 1.73,
|
| 1526 |
+
"learning_rate": 0.00048827815274121847,
|
| 1527 |
+
"loss": 0.1505,
|
| 1528 |
+
"step": 50200
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"epoch": 1.74,
|
| 1532 |
+
"learning_rate": 0.0004876576626747684,
|
| 1533 |
+
"loss": 0.1328,
|
| 1534 |
+
"step": 50400
|
| 1535 |
+
},
|
| 1536 |
+
{
|
| 1537 |
+
"epoch": 1.75,
|
| 1538 |
+
"learning_rate": 0.0004870352278694289,
|
| 1539 |
+
"loss": 0.1511,
|
| 1540 |
+
"step": 50600
|
| 1541 |
+
},
|
| 1542 |
+
{
|
| 1543 |
+
"epoch": 1.75,
|
| 1544 |
+
"learning_rate": 0.00048641085430050653,
|
| 1545 |
+
"loss": 0.1475,
|
| 1546 |
+
"step": 50800
|
| 1547 |
+
},
|
| 1548 |
+
{
|
| 1549 |
+
"epoch": 1.76,
|
| 1550 |
+
"learning_rate": 0.00048578454796191977,
|
| 1551 |
+
"loss": 0.1454,
|
| 1552 |
+
"step": 51000
|
| 1553 |
+
},
|
| 1554 |
+
{
|
| 1555 |
+
"epoch": 1.77,
|
| 1556 |
+
"learning_rate": 0.00048515631486614134,
|
| 1557 |
+
"loss": 0.1472,
|
| 1558 |
+
"step": 51200
|
| 1559 |
+
},
|
| 1560 |
+
{
|
| 1561 |
+
"epoch": 1.77,
|
| 1562 |
+
"learning_rate": 0.0004845261610441408,
|
| 1563 |
+
"loss": 0.1295,
|
| 1564 |
+
"step": 51400
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 1.78,
|
| 1568 |
+
"learning_rate": 0.0004838972576405494,
|
| 1569 |
+
"loss": 0.1372,
|
| 1570 |
+
"step": 51600
|
| 1571 |
+
},
|
| 1572 |
+
{
|
| 1573 |
+
"epoch": 1.79,
|
| 1574 |
+
"learning_rate": 0.00048326329006063116,
|
| 1575 |
+
"loss": 0.1442,
|
| 1576 |
+
"step": 51800
|
| 1577 |
+
},
|
| 1578 |
+
{
|
| 1579 |
+
"epoch": 1.79,
|
| 1580 |
+
"learning_rate": 0.0004826274199273233,
|
| 1581 |
+
"loss": 0.1379,
|
| 1582 |
+
"step": 52000
|
| 1583 |
+
},
|
| 1584 |
+
{
|
| 1585 |
+
"epoch": 1.8,
|
| 1586 |
+
"learning_rate": 0.00048198965334490996,
|
| 1587 |
+
"loss": 0.1423,
|
| 1588 |
+
"step": 52200
|
| 1589 |
+
},
|
| 1590 |
+
{
|
| 1591 |
+
"epoch": 1.81,
|
| 1592 |
+
"learning_rate": 0.0004813499964358812,
|
| 1593 |
+
"loss": 0.1413,
|
| 1594 |
+
"step": 52400
|
| 1595 |
+
},
|
| 1596 |
+
{
|
| 1597 |
+
"epoch": 1.82,
|
| 1598 |
+
"learning_rate": 0.0004807084553408737,
|
| 1599 |
+
"loss": 0.1423,
|
| 1600 |
+
"step": 52600
|
| 1601 |
+
},
|
| 1602 |
+
{
|
| 1603 |
+
"epoch": 1.82,
|
| 1604 |
+
"learning_rate": 0.0004800650362186124,
|
| 1605 |
+
"loss": 0.1455,
|
| 1606 |
+
"step": 52800
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 1.83,
|
| 1610 |
+
"learning_rate": 0.0004794197452458509,
|
| 1611 |
+
"loss": 0.1415,
|
| 1612 |
+
"step": 53000
|
| 1613 |
+
},
|
| 1614 |
+
{
|
| 1615 |
+
"epoch": 1.84,
|
| 1616 |
+
"learning_rate": 0.00047877258861731253,
|
| 1617 |
+
"loss": 0.1452,
|
| 1618 |
+
"step": 53200
|
| 1619 |
+
},
|
| 1620 |
+
{
|
| 1621 |
+
"epoch": 1.84,
|
| 1622 |
+
"learning_rate": 0.00047812357254563043,
|
| 1623 |
+
"loss": 0.1361,
|
| 1624 |
+
"step": 53400
|
| 1625 |
+
},
|
| 1626 |
+
{
|
| 1627 |
+
"epoch": 1.85,
|
| 1628 |
+
"learning_rate": 0.00047747596220724844,
|
| 1629 |
+
"loss": 0.1437,
|
| 1630 |
+
"step": 53600
|
| 1631 |
+
},
|
| 1632 |
+
{
|
| 1633 |
+
"epoch": 1.86,
|
| 1634 |
+
"learning_rate": 0.0004768232551777713,
|
| 1635 |
+
"loss": 0.1301,
|
| 1636 |
+
"step": 53800
|
| 1637 |
+
},
|
| 1638 |
+
{
|
| 1639 |
+
"epoch": 1.86,
|
| 1640 |
+
"learning_rate": 0.0004761687074185398,
|
| 1641 |
+
"loss": 0.1401,
|
| 1642 |
+
"step": 54000
|
| 1643 |
+
},
|
| 1644 |
+
{
|
| 1645 |
+
"epoch": 1.87,
|
| 1646 |
+
"learning_rate": 0.0004755123252131414,
|
| 1647 |
+
"loss": 0.1392,
|
| 1648 |
+
"step": 54200
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 1.88,
|
| 1652 |
+
"learning_rate": 0.00047485411486277384,
|
| 1653 |
+
"loss": 0.1334,
|
| 1654 |
+
"step": 54400
|
| 1655 |
+
},
|
| 1656 |
+
{
|
| 1657 |
+
"epoch": 1.88,
|
| 1658 |
+
"learning_rate": 0.00047419408268618515,
|
| 1659 |
+
"loss": 0.1486,
|
| 1660 |
+
"step": 54600
|
| 1661 |
+
},
|
| 1662 |
+
{
|
| 1663 |
+
"epoch": 1.89,
|
| 1664 |
+
"learning_rate": 0.0004735322350196125,
|
| 1665 |
+
"loss": 0.146,
|
| 1666 |
+
"step": 54800
|
| 1667 |
+
},
|
| 1668 |
+
{
|
| 1669 |
+
"epoch": 1.9,
|
| 1670 |
+
"learning_rate": 0.00047286857821672145,
|
| 1671 |
+
"loss": 0.1531,
|
| 1672 |
+
"step": 55000
|
| 1673 |
+
},
|
| 1674 |
+
{
|
| 1675 |
+
"epoch": 1.91,
|
| 1676 |
+
"learning_rate": 0.00047220311864854545,
|
| 1677 |
+
"loss": 0.1312,
|
| 1678 |
+
"step": 55200
|
| 1679 |
+
},
|
| 1680 |
+
{
|
| 1681 |
+
"epoch": 1.91,
|
| 1682 |
+
"learning_rate": 0.0004715358627034239,
|
| 1683 |
+
"loss": 0.1376,
|
| 1684 |
+
"step": 55400
|
| 1685 |
+
},
|
| 1686 |
+
{
|
| 1687 |
+
"epoch": 1.92,
|
| 1688 |
+
"learning_rate": 0.00047086681678694143,
|
| 1689 |
+
"loss": 0.138,
|
| 1690 |
+
"step": 55600
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"epoch": 1.93,
|
| 1694 |
+
"learning_rate": 0.00047019934589512296,
|
| 1695 |
+
"loss": 0.1476,
|
| 1696 |
+
"step": 55800
|
| 1697 |
+
},
|
| 1698 |
+
{
|
| 1699 |
+
"epoch": 1.93,
|
| 1700 |
+
"learning_rate": 0.000469526748190841,
|
| 1701 |
+
"loss": 0.1394,
|
| 1702 |
+
"step": 56000
|
| 1703 |
+
},
|
| 1704 |
+
{
|
| 1705 |
+
"epoch": 1.94,
|
| 1706 |
+
"learning_rate": 0.0004688523798024791,
|
| 1707 |
+
"loss": 0.1527,
|
| 1708 |
+
"step": 56200
|
| 1709 |
+
},
|
| 1710 |
+
{
|
| 1711 |
+
"epoch": 1.95,
|
| 1712 |
+
"learning_rate": 0.00046817624720390063,
|
| 1713 |
+
"loss": 0.1422,
|
| 1714 |
+
"step": 56400
|
| 1715 |
+
},
|
| 1716 |
+
{
|
| 1717 |
+
"epoch": 1.95,
|
| 1718 |
+
"learning_rate": 0.00046750175069906427,
|
| 1719 |
+
"loss": 0.1615,
|
| 1720 |
+
"step": 56600
|
| 1721 |
+
},
|
| 1722 |
+
{
|
| 1723 |
+
"epoch": 1.96,
|
| 1724 |
+
"learning_rate": 0.00046682211790916814,
|
| 1725 |
+
"loss": 0.1263,
|
| 1726 |
+
"step": 56800
|
| 1727 |
+
},
|
| 1728 |
+
{
|
| 1729 |
+
"epoch": 1.97,
|
| 1730 |
+
"learning_rate": 0.00046614074039934907,
|
| 1731 |
+
"loss": 0.1503,
|
| 1732 |
+
"step": 57000
|
| 1733 |
+
},
|
| 1734 |
+
{
|
| 1735 |
+
"epoch": 1.97,
|
| 1736 |
+
"learning_rate": 0.00046545762471075726,
|
| 1737 |
+
"loss": 0.1557,
|
| 1738 |
+
"step": 57200
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 1.98,
|
| 1742 |
+
"learning_rate": 0.0004647727774012289,
|
| 1743 |
+
"loss": 0.1367,
|
| 1744 |
+
"step": 57400
|
| 1745 |
+
},
|
| 1746 |
+
{
|
| 1747 |
+
"epoch": 1.99,
|
| 1748 |
+
"learning_rate": 0.0004640862050452239,
|
| 1749 |
+
"loss": 0.1341,
|
| 1750 |
+
"step": 57600
|
| 1751 |
+
},
|
| 1752 |
+
{
|
| 1753 |
+
"epoch": 2.0,
|
| 1754 |
+
"learning_rate": 0.00046339791423376226,
|
| 1755 |
+
"loss": 0.1449,
|
| 1756 |
+
"step": 57800
|
| 1757 |
+
},
|
| 1758 |
+
{
|
| 1759 |
+
"epoch": 2.0,
|
| 1760 |
+
"eval_accuracy": 0.9439565632043402,
|
| 1761 |
+
"eval_auc": 0.8461801873862141,
|
| 1762 |
+
"eval_f1": 0.3995036053381367,
|
| 1763 |
+
"eval_loss": 0.4224541485309601,
|
| 1764 |
+
"eval_precision": 0.27316793551875035,
|
| 1765 |
+
"eval_recall": 0.7432397408207343,
|
| 1766 |
+
"eval_runtime": 1778.6249,
|
| 1767 |
+
"eval_samples_per_second": 24.982,
|
| 1768 |
+
"eval_steps_per_second": 4.164,
|
| 1769 |
+
"step": 57940
|
| 1770 |
+
}
|
| 1771 |
+
],
|
| 1772 |
+
"logging_steps": 200,
|
| 1773 |
+
"max_steps": 202790,
|
| 1774 |
+
"num_train_epochs": 7,
|
| 1775 |
+
"save_steps": 500,
|
| 1776 |
+
"total_flos": 6.9510044394e+16,
|
| 1777 |
+
"trial_name": null,
|
| 1778 |
+
"trial_params": null
|
| 1779 |
+
}
|
training_args (3).bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:aabfc585a5b018aaa712111ce2f054ea567fe971ffe6acd839649a028f2c23bf
|
| 3 |
+
size 4091
|
vocab (6).txt
ADDED
|
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
<cls>
|
| 2 |
+
<pad>
|
| 3 |
+
<eos>
|
| 4 |
+
<unk>
|
| 5 |
+
L
|
| 6 |
+
A
|
| 7 |
+
G
|
| 8 |
+
V
|
| 9 |
+
S
|
| 10 |
+
E
|
| 11 |
+
R
|
| 12 |
+
T
|
| 13 |
+
I
|
| 14 |
+
D
|
| 15 |
+
P
|
| 16 |
+
K
|
| 17 |
+
Q
|
| 18 |
+
N
|
| 19 |
+
F
|
| 20 |
+
Y
|
| 21 |
+
M
|
| 22 |
+
H
|
| 23 |
+
W
|
| 24 |
+
C
|
| 25 |
+
X
|
| 26 |
+
B
|
| 27 |
+
U
|
| 28 |
+
Z
|
| 29 |
+
O
|
| 30 |
+
.
|
| 31 |
+
-
|
| 32 |
+
<null_1>
|
| 33 |
+
<mask>
|