Hugging Face's logo

CybersecurityLab-unibz
/
cross_domain_ID_USC

Safetensors
English
roberta
Model card Files
xet
Community
mmock commited on
Commit
3ea61d0
·
verified ·
1 Parent(s): 4028e6e

Upload folder using huggingface_hub

Browse files
Files changed (11) hide show
  1. config.json +29 -0
  2. merges.txt +0 -0
  3. model.safetensors +3 -0
  4. optimizer.pt +3 -0
  5. rng_state.pth +3 -0
  6. scheduler.pt +3 -0
  7. special_tokens_map.json +51 -0
  8. tokenizer_config.json +56 -0
  9. trainer_state.json +3493 -0
  10. training_args.bin +3 -0
  11. vocab.json +0 -0
config.json ADDED
@@ -0,0 +1,29 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "microsoft/codebert-base",
3
+ "architectures": [
4
+ "RobertaForSequenceClassification"
5
+ ],
6
+ "attention_probs_dropout_prob": 0.1,
7
+ "bos_token_id": 0,
8
+ "classifier_dropout": null,
9
+ "eos_token_id": 2,
10
+ "hidden_act": "gelu",
11
+ "hidden_dropout_prob": 0.1,
12
+ "hidden_size": 768,
13
+ "initializer_range": 0.02,
14
+ "intermediate_size": 3072,
15
+ "layer_norm_eps": 1e-05,
16
+ "max_position_embeddings": 514,
17
+ "model_type": "roberta",
18
+ "num_attention_heads": 12,
19
+ "num_hidden_layers": 12,
20
+ "output_past": true,
21
+ "pad_token_id": 1,
22
+ "position_embedding_type": "absolute",
23
+ "problem_type": "single_label_classification",
24
+ "torch_dtype": "float32",
25
+ "transformers_version": "4.35.0",
26
+ "type_vocab_size": 1,
27
+ "use_cache": true,
28
+ "vocab_size": 50265
29
+ }
merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8d0484f87e5afe26654d7fc31434e4b120ca83437b1f87d2373fe3fe19f18fe0
3
+ size 498612824
optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a32fdb580e6aae235bbab52dc9434f859e7162da1475368b9ebe9ce6056d1e7f
3
+ size 997345530
rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1d5356fe88300223f466b71f24db6d2e399d790212a898710426f7d4021ddbba
3
+ size 14244
scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1eb89208d1d38fa9034e319e859a405b4a244a854a3f806d7cf46d0c4cc8e7e5
3
+ size 1064
special_tokens_map.json ADDED
@@ -0,0 +1,51 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": true,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "cls_token": {
10
+ "content": "<s>",
11
+ "lstrip": false,
12
+ "normalized": true,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "eos_token": {
17
+ "content": "</s>",
18
+ "lstrip": false,
19
+ "normalized": true,
20
+ "rstrip": false,
21
+ "single_word": false
22
+ },
23
+ "mask_token": {
24
+ "content": "<mask>",
25
+ "lstrip": true,
26
+ "normalized": false,
27
+ "rstrip": false,
28
+ "single_word": false
29
+ },
30
+ "pad_token": {
31
+ "content": "<pad>",
32
+ "lstrip": false,
33
+ "normalized": true,
34
+ "rstrip": false,
35
+ "single_word": false
36
+ },
37
+ "sep_token": {
38
+ "content": "</s>",
39
+ "lstrip": false,
40
+ "normalized": true,
41
+ "rstrip": false,
42
+ "single_word": false
43
+ },
44
+ "unk_token": {
45
+ "content": "<unk>",
46
+ "lstrip": false,
47
+ "normalized": true,
48
+ "rstrip": false,
49
+ "single_word": false
50
+ }
51
+ }
tokenizer_config.json ADDED
@@ -0,0 +1,56 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": false,
3
+ "added_tokens_decoder": {
4
+ "0": {
5
+ "content": "<s>",
6
+ "lstrip": false,
7
+ "normalized": true,
8
+ "rstrip": false,
9
+ "single_word": false,
10
+ "special": true
11
+ },
12
+ "1": {
13
+ "content": "<pad>",
14
+ "lstrip": false,
15
+ "normalized": true,
16
+ "rstrip": false,
17
+ "single_word": false,
18
+ "special": true
19
+ },
20
+ "2": {
21
+ "content": "</s>",
22
+ "lstrip": false,
23
+ "normalized": true,
24
+ "rstrip": false,
25
+ "single_word": false,
26
+ "special": true
27
+ },
28
+ "3": {
29
+ "content": "<unk>",
30
+ "lstrip": false,
31
+ "normalized": true,
32
+ "rstrip": false,
33
+ "single_word": false,
34
+ "special": true
35
+ },
36
+ "50264": {
37
+ "content": "<mask>",
38
+ "lstrip": true,
39
+ "normalized": false,
40
+ "rstrip": false,
41
+ "single_word": false,
42
+ "special": true
43
+ }
44
+ },
45
+ "bos_token": "<s>",
46
+ "clean_up_tokenization_spaces": true,
47
+ "cls_token": "<s>",
48
+ "eos_token": "</s>",
49
+ "errors": "replace",
50
+ "mask_token": "<mask>",
51
+ "model_max_length": 512,
52
+ "pad_token": "<pad>",
53
+ "sep_token": "</s>",
54
+ "tokenizer_class": "RobertaTokenizer",
55
+ "unk_token": "<unk>"
56
+ }
trainer_state.json ADDED
@@ -0,0 +1,3493 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.19916516542434692,
3
+ "best_model_checkpoint": "/leonardo_work/IscrC_AGENT/PROFES2025/results/balanced/industry/checkpoint-617",
4
+ "epoch": 9.0,
5
+ "eval_steps": 500,
6
+ "global_step": 5553,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.02,
13
+ "learning_rate": 1.9967585089141006e-05,
14
+ "loss": 0.672,
15
+ "step": 10
16
+ },
17
+ {
18
+ "epoch": 0.03,
19
+ "learning_rate": 1.993517017828201e-05,
20
+ "loss": 0.5205,
21
+ "step": 20
22
+ },
23
+ {
24
+ "epoch": 0.05,
25
+ "learning_rate": 1.9902755267423015e-05,
26
+ "loss": 0.5417,
27
+ "step": 30
28
+ },
29
+ {
30
+ "epoch": 0.06,
31
+ "learning_rate": 1.9870340356564023e-05,
32
+ "loss": 0.5525,
33
+ "step": 40
34
+ },
35
+ {
36
+ "epoch": 0.08,
37
+ "learning_rate": 1.9837925445705027e-05,
38
+ "loss": 0.5327,
39
+ "step": 50
40
+ },
41
+ {
42
+ "epoch": 0.1,
43
+ "learning_rate": 1.980551053484603e-05,
44
+ "loss": 0.5127,
45
+ "step": 60
46
+ },
47
+ {
48
+ "epoch": 0.11,
49
+ "learning_rate": 1.9773095623987036e-05,
50
+ "loss": 0.5114,
51
+ "step": 70
52
+ },
53
+ {
54
+ "epoch": 0.13,
55
+ "learning_rate": 1.974068071312804e-05,
56
+ "loss": 0.4299,
57
+ "step": 80
58
+ },
59
+ {
60
+ "epoch": 0.15,
61
+ "learning_rate": 1.9708265802269045e-05,
62
+ "loss": 0.4515,
63
+ "step": 90
64
+ },
65
+ {
66
+ "epoch": 0.16,
67
+ "learning_rate": 1.9675850891410053e-05,
68
+ "loss": 0.4393,
69
+ "step": 100
70
+ },
71
+ {
72
+ "epoch": 0.18,
73
+ "learning_rate": 1.9643435980551057e-05,
74
+ "loss": 0.3662,
75
+ "step": 110
76
+ },
77
+ {
78
+ "epoch": 0.19,
79
+ "learning_rate": 1.961102106969206e-05,
80
+ "loss": 0.5824,
81
+ "step": 120
82
+ },
83
+ {
84
+ "epoch": 0.21,
85
+ "learning_rate": 1.9578606158833066e-05,
86
+ "loss": 0.3448,
87
+ "step": 130
88
+ },
89
+ {
90
+ "epoch": 0.23,
91
+ "learning_rate": 1.954619124797407e-05,
92
+ "loss": 0.5221,
93
+ "step": 140
94
+ },
95
+ {
96
+ "epoch": 0.24,
97
+ "learning_rate": 1.9513776337115075e-05,
98
+ "loss": 0.4427,
99
+ "step": 150
100
+ },
101
+ {
102
+ "epoch": 0.26,
103
+ "learning_rate": 1.948136142625608e-05,
104
+ "loss": 0.3759,
105
+ "step": 160
106
+ },
107
+ {
108
+ "epoch": 0.28,
109
+ "learning_rate": 1.9448946515397084e-05,
110
+ "loss": 0.4701,
111
+ "step": 170
112
+ },
113
+ {
114
+ "epoch": 0.29,
115
+ "learning_rate": 1.9416531604538088e-05,
116
+ "loss": 0.3552,
117
+ "step": 180
118
+ },
119
+ {
120
+ "epoch": 0.31,
121
+ "learning_rate": 1.9384116693679093e-05,
122
+ "loss": 0.3537,
123
+ "step": 190
124
+ },
125
+ {
126
+ "epoch": 0.32,
127
+ "learning_rate": 1.9351701782820097e-05,
128
+ "loss": 0.4193,
129
+ "step": 200
130
+ },
131
+ {
132
+ "epoch": 0.34,
133
+ "learning_rate": 1.93192868719611e-05,
134
+ "loss": 0.3605,
135
+ "step": 210
136
+ },
137
+ {
138
+ "epoch": 0.36,
139
+ "learning_rate": 1.928687196110211e-05,
140
+ "loss": 0.3847,
141
+ "step": 220
142
+ },
143
+ {
144
+ "epoch": 0.37,
145
+ "learning_rate": 1.9254457050243114e-05,
146
+ "loss": 0.3656,
147
+ "step": 230
148
+ },
149
+ {
150
+ "epoch": 0.39,
151
+ "learning_rate": 1.9222042139384118e-05,
152
+ "loss": 0.4318,
153
+ "step": 240
154
+ },
155
+ {
156
+ "epoch": 0.41,
157
+ "learning_rate": 1.9189627228525122e-05,
158
+ "loss": 0.5195,
159
+ "step": 250
160
+ },
161
+ {
162
+ "epoch": 0.42,
163
+ "learning_rate": 1.9157212317666127e-05,
164
+ "loss": 0.3167,
165
+ "step": 260
166
+ },
167
+ {
168
+ "epoch": 0.44,
169
+ "learning_rate": 1.912479740680713e-05,
170
+ "loss": 0.426,
171
+ "step": 270
172
+ },
173
+ {
174
+ "epoch": 0.45,
175
+ "learning_rate": 1.909238249594814e-05,
176
+ "loss": 0.3715,
177
+ "step": 280
178
+ },
179
+ {
180
+ "epoch": 0.47,
181
+ "learning_rate": 1.9059967585089144e-05,
182
+ "loss": 0.3404,
183
+ "step": 290
184
+ },
185
+ {
186
+ "epoch": 0.49,
187
+ "learning_rate": 1.9027552674230148e-05,
188
+ "loss": 0.3938,
189
+ "step": 300
190
+ },
191
+ {
192
+ "epoch": 0.5,
193
+ "learning_rate": 1.8995137763371152e-05,
194
+ "loss": 0.3288,
195
+ "step": 310
196
+ },
197
+ {
198
+ "epoch": 0.52,
199
+ "learning_rate": 1.8962722852512157e-05,
200
+ "loss": 0.4818,
201
+ "step": 320
202
+ },
203
+ {
204
+ "epoch": 0.53,
205
+ "learning_rate": 1.893030794165316e-05,
206
+ "loss": 0.3244,
207
+ "step": 330
208
+ },
209
+ {
210
+ "epoch": 0.55,
211
+ "learning_rate": 1.889789303079417e-05,
212
+ "loss": 0.3841,
213
+ "step": 340
214
+ },
215
+ {
216
+ "epoch": 0.57,
217
+ "learning_rate": 1.8865478119935173e-05,
218
+ "loss": 0.3647,
219
+ "step": 350
220
+ },
221
+ {
222
+ "epoch": 0.58,
223
+ "learning_rate": 1.8833063209076178e-05,
224
+ "loss": 0.3362,
225
+ "step": 360
226
+ },
227
+ {
228
+ "epoch": 0.6,
229
+ "learning_rate": 1.8800648298217182e-05,
230
+ "loss": 0.2954,
231
+ "step": 370
232
+ },
233
+ {
234
+ "epoch": 0.62,
235
+ "learning_rate": 1.8768233387358187e-05,
236
+ "loss": 0.4099,
237
+ "step": 380
238
+ },
239
+ {
240
+ "epoch": 0.63,
241
+ "learning_rate": 1.873581847649919e-05,
242
+ "loss": 0.2331,
243
+ "step": 390
244
+ },
245
+ {
246
+ "epoch": 0.65,
247
+ "learning_rate": 1.8703403565640196e-05,
248
+ "loss": 0.4285,
249
+ "step": 400
250
+ },
251
+ {
252
+ "epoch": 0.66,
253
+ "learning_rate": 1.86709886547812e-05,
254
+ "loss": 0.3539,
255
+ "step": 410
256
+ },
257
+ {
258
+ "epoch": 0.68,
259
+ "learning_rate": 1.8638573743922204e-05,
260
+ "loss": 0.3132,
261
+ "step": 420
262
+ },
263
+ {
264
+ "epoch": 0.7,
265
+ "learning_rate": 1.860615883306321e-05,
266
+ "loss": 0.4059,
267
+ "step": 430
268
+ },
269
+ {
270
+ "epoch": 0.71,
271
+ "learning_rate": 1.8573743922204213e-05,
272
+ "loss": 0.3634,
273
+ "step": 440
274
+ },
275
+ {
276
+ "epoch": 0.73,
277
+ "learning_rate": 1.8541329011345218e-05,
278
+ "loss": 0.3523,
279
+ "step": 450
280
+ },
281
+ {
282
+ "epoch": 0.75,
283
+ "learning_rate": 1.8508914100486226e-05,
284
+ "loss": 0.3134,
285
+ "step": 460
286
+ },
287
+ {
288
+ "epoch": 0.76,
289
+ "learning_rate": 1.847649918962723e-05,
290
+ "loss": 0.3582,
291
+ "step": 470
292
+ },
293
+ {
294
+ "epoch": 0.78,
295
+ "learning_rate": 1.8444084278768234e-05,
296
+ "loss": 0.3172,
297
+ "step": 480
298
+ },
299
+ {
300
+ "epoch": 0.79,
301
+ "learning_rate": 1.841166936790924e-05,
302
+ "loss": 0.221,
303
+ "step": 490
304
+ },
305
+ {
306
+ "epoch": 0.81,
307
+ "learning_rate": 1.8379254457050243e-05,
308
+ "loss": 0.3974,
309
+ "step": 500
310
+ },
311
+ {
312
+ "epoch": 0.83,
313
+ "learning_rate": 1.8346839546191248e-05,
314
+ "loss": 0.276,
315
+ "step": 510
316
+ },
317
+ {
318
+ "epoch": 0.84,
319
+ "learning_rate": 1.8314424635332255e-05,
320
+ "loss": 0.3562,
321
+ "step": 520
322
+ },
323
+ {
324
+ "epoch": 0.86,
325
+ "learning_rate": 1.828200972447326e-05,
326
+ "loss": 0.3832,
327
+ "step": 530
328
+ },
329
+ {
330
+ "epoch": 0.88,
331
+ "learning_rate": 1.8249594813614264e-05,
332
+ "loss": 0.3265,
333
+ "step": 540
334
+ },
335
+ {
336
+ "epoch": 0.89,
337
+ "learning_rate": 1.821717990275527e-05,
338
+ "loss": 0.3384,
339
+ "step": 550
340
+ },
341
+ {
342
+ "epoch": 0.91,
343
+ "learning_rate": 1.8184764991896273e-05,
344
+ "loss": 0.274,
345
+ "step": 560
346
+ },
347
+ {
348
+ "epoch": 0.92,
349
+ "learning_rate": 1.8152350081037278e-05,
350
+ "loss": 0.368,
351
+ "step": 570
352
+ },
353
+ {
354
+ "epoch": 0.94,
355
+ "learning_rate": 1.8119935170178285e-05,
356
+ "loss": 0.2889,
357
+ "step": 580
358
+ },
359
+ {
360
+ "epoch": 0.96,
361
+ "learning_rate": 1.808752025931929e-05,
362
+ "loss": 0.315,
363
+ "step": 590
364
+ },
365
+ {
366
+ "epoch": 0.97,
367
+ "learning_rate": 1.8055105348460294e-05,
368
+ "loss": 0.3446,
369
+ "step": 600
370
+ },
371
+ {
372
+ "epoch": 0.99,
373
+ "learning_rate": 1.80226904376013e-05,
374
+ "loss": 0.2857,
375
+ "step": 610
376
+ },
377
+ {
378
+ "epoch": 1.0,
379
+ "eval_FN": 140,
380
+ "eval_FP": 433,
381
+ "eval_TN": 5405,
382
+ "eval_TP": 455,
383
+ "eval_accuracy": 0.9109280273589305,
384
+ "eval_f1": 0.6136210384356036,
385
+ "eval_loss": 0.19916516542434692,
386
+ "eval_precision": 0.5123873873873874,
387
+ "eval_recall": 0.7647058823529411,
388
+ "eval_runtime": 44.7866,
389
+ "eval_samples_per_second": 143.637,
390
+ "eval_steps_per_second": 8.998,
391
+ "step": 617
392
+ },
393
+ {
394
+ "epoch": 1.0,
395
+ "learning_rate": 1.7990275526742303e-05,
396
+ "loss": 0.3441,
397
+ "step": 620
398
+ },
399
+ {
400
+ "epoch": 1.02,
401
+ "learning_rate": 1.7957860615883308e-05,
402
+ "loss": 0.2048,
403
+ "step": 630
404
+ },
405
+ {
406
+ "epoch": 1.04,
407
+ "learning_rate": 1.7925445705024312e-05,
408
+ "loss": 0.3143,
409
+ "step": 640
410
+ },
411
+ {
412
+ "epoch": 1.05,
413
+ "learning_rate": 1.7893030794165316e-05,
414
+ "loss": 0.3601,
415
+ "step": 650
416
+ },
417
+ {
418
+ "epoch": 1.07,
419
+ "learning_rate": 1.7860615883306324e-05,
420
+ "loss": 0.2511,
421
+ "step": 660
422
+ },
423
+ {
424
+ "epoch": 1.09,
425
+ "learning_rate": 1.782820097244733e-05,
426
+ "loss": 0.2382,
427
+ "step": 670
428
+ },
429
+ {
430
+ "epoch": 1.1,
431
+ "learning_rate": 1.7795786061588333e-05,
432
+ "loss": 0.2701,
433
+ "step": 680
434
+ },
435
+ {
436
+ "epoch": 1.12,
437
+ "learning_rate": 1.7763371150729338e-05,
438
+ "loss": 0.2492,
439
+ "step": 690
440
+ },
441
+ {
442
+ "epoch": 1.13,
443
+ "learning_rate": 1.7730956239870342e-05,
444
+ "loss": 0.2658,
445
+ "step": 700
446
+ },
447
+ {
448
+ "epoch": 1.15,
449
+ "learning_rate": 1.7698541329011346e-05,
450
+ "loss": 0.4861,
451
+ "step": 710
452
+ },
453
+ {
454
+ "epoch": 1.17,
455
+ "learning_rate": 1.766612641815235e-05,
456
+ "loss": 0.2431,
457
+ "step": 720
458
+ },
459
+ {
460
+ "epoch": 1.18,
461
+ "learning_rate": 1.7633711507293355e-05,
462
+ "loss": 0.3409,
463
+ "step": 730
464
+ },
465
+ {
466
+ "epoch": 1.2,
467
+ "learning_rate": 1.760129659643436e-05,
468
+ "loss": 0.2178,
469
+ "step": 740
470
+ },
471
+ {
472
+ "epoch": 1.22,
473
+ "learning_rate": 1.7568881685575364e-05,
474
+ "loss": 0.1971,
475
+ "step": 750
476
+ },
477
+ {
478
+ "epoch": 1.23,
479
+ "learning_rate": 1.7536466774716372e-05,
480
+ "loss": 0.2586,
481
+ "step": 760
482
+ },
483
+ {
484
+ "epoch": 1.25,
485
+ "learning_rate": 1.7504051863857376e-05,
486
+ "loss": 0.2778,
487
+ "step": 770
488
+ },
489
+ {
490
+ "epoch": 1.26,
491
+ "learning_rate": 1.747163695299838e-05,
492
+ "loss": 0.2442,
493
+ "step": 780
494
+ },
495
+ {
496
+ "epoch": 1.28,
497
+ "learning_rate": 1.7439222042139385e-05,
498
+ "loss": 0.2361,
499
+ "step": 790
500
+ },
501
+ {
502
+ "epoch": 1.3,
503
+ "learning_rate": 1.740680713128039e-05,
504
+ "loss": 0.3689,
505
+ "step": 800
506
+ },
507
+ {
508
+ "epoch": 1.31,
509
+ "learning_rate": 1.7374392220421394e-05,
510
+ "loss": 0.3548,
511
+ "step": 810
512
+ },
513
+ {
514
+ "epoch": 1.33,
515
+ "learning_rate": 1.7341977309562402e-05,
516
+ "loss": 0.3304,
517
+ "step": 820
518
+ },
519
+ {
520
+ "epoch": 1.35,
521
+ "learning_rate": 1.7309562398703406e-05,
522
+ "loss": 0.1903,
523
+ "step": 830
524
+ },
525
+ {
526
+ "epoch": 1.36,
527
+ "learning_rate": 1.727714748784441e-05,
528
+ "loss": 0.2579,
529
+ "step": 840
530
+ },
531
+ {
532
+ "epoch": 1.38,
533
+ "learning_rate": 1.7244732576985415e-05,
534
+ "loss": 0.2982,
535
+ "step": 850
536
+ },
537
+ {
538
+ "epoch": 1.39,
539
+ "learning_rate": 1.721231766612642e-05,
540
+ "loss": 0.2881,
541
+ "step": 860
542
+ },
543
+ {
544
+ "epoch": 1.41,
545
+ "learning_rate": 1.7179902755267424e-05,
546
+ "loss": 0.2217,
547
+ "step": 870
548
+ },
549
+ {
550
+ "epoch": 1.43,
551
+ "learning_rate": 1.7147487844408432e-05,
552
+ "loss": 0.1942,
553
+ "step": 880
554
+ },
555
+ {
556
+ "epoch": 1.44,
557
+ "learning_rate": 1.7115072933549436e-05,
558
+ "loss": 0.1682,
559
+ "step": 890
560
+ },
561
+ {
562
+ "epoch": 1.46,
563
+ "learning_rate": 1.708265802269044e-05,
564
+ "loss": 0.2382,
565
+ "step": 900
566
+ },
567
+ {
568
+ "epoch": 1.47,
569
+ "learning_rate": 1.7050243111831445e-05,
570
+ "loss": 0.3257,
571
+ "step": 910
572
+ },
573
+ {
574
+ "epoch": 1.49,
575
+ "learning_rate": 1.701782820097245e-05,
576
+ "loss": 0.2064,
577
+ "step": 920
578
+ },
579
+ {
580
+ "epoch": 1.51,
581
+ "learning_rate": 1.6985413290113454e-05,
582
+ "loss": 0.3427,
583
+ "step": 930
584
+ },
585
+ {
586
+ "epoch": 1.52,
587
+ "learning_rate": 1.695299837925446e-05,
588
+ "loss": 0.2511,
589
+ "step": 940
590
+ },
591
+ {
592
+ "epoch": 1.54,
593
+ "learning_rate": 1.6920583468395463e-05,
594
+ "loss": 0.2737,
595
+ "step": 950
596
+ },
597
+ {
598
+ "epoch": 1.56,
599
+ "learning_rate": 1.6888168557536467e-05,
600
+ "loss": 0.2734,
601
+ "step": 960
602
+ },
603
+ {
604
+ "epoch": 1.57,
605
+ "learning_rate": 1.685575364667747e-05,
606
+ "loss": 0.3016,
607
+ "step": 970
608
+ },
609
+ {
610
+ "epoch": 1.59,
611
+ "learning_rate": 1.6823338735818476e-05,
612
+ "loss": 0.1923,
613
+ "step": 980
614
+ },
615
+ {
616
+ "epoch": 1.6,
617
+ "learning_rate": 1.679092382495948e-05,
618
+ "loss": 0.2098,
619
+ "step": 990
620
+ },
621
+ {
622
+ "epoch": 1.62,
623
+ "learning_rate": 1.6758508914100488e-05,
624
+ "loss": 0.3171,
625
+ "step": 1000
626
+ },
627
+ {
628
+ "epoch": 1.64,
629
+ "learning_rate": 1.6726094003241493e-05,
630
+ "loss": 0.1955,
631
+ "step": 1010
632
+ },
633
+ {
634
+ "epoch": 1.65,
635
+ "learning_rate": 1.6693679092382497e-05,
636
+ "loss": 0.303,
637
+ "step": 1020
638
+ },
639
+ {
640
+ "epoch": 1.67,
641
+ "learning_rate": 1.66612641815235e-05,
642
+ "loss": 0.2604,
643
+ "step": 1030
644
+ },
645
+ {
646
+ "epoch": 1.69,
647
+ "learning_rate": 1.6628849270664506e-05,
648
+ "loss": 0.2525,
649
+ "step": 1040
650
+ },
651
+ {
652
+ "epoch": 1.7,
653
+ "learning_rate": 1.659643435980551e-05,
654
+ "loss": 0.1941,
655
+ "step": 1050
656
+ },
657
+ {
658
+ "epoch": 1.72,
659
+ "learning_rate": 1.6564019448946518e-05,
660
+ "loss": 0.3828,
661
+ "step": 1060
662
+ },
663
+ {
664
+ "epoch": 1.73,
665
+ "learning_rate": 1.6531604538087523e-05,
666
+ "loss": 0.4138,
667
+ "step": 1070
668
+ },
669
+ {
670
+ "epoch": 1.75,
671
+ "learning_rate": 1.6499189627228527e-05,
672
+ "loss": 0.2303,
673
+ "step": 1080
674
+ },
675
+ {
676
+ "epoch": 1.77,
677
+ "learning_rate": 1.646677471636953e-05,
678
+ "loss": 0.2644,
679
+ "step": 1090
680
+ },
681
+ {
682
+ "epoch": 1.78,
683
+ "learning_rate": 1.6434359805510536e-05,
684
+ "loss": 0.3529,
685
+ "step": 1100
686
+ },
687
+ {
688
+ "epoch": 1.8,
689
+ "learning_rate": 1.640194489465154e-05,
690
+ "loss": 0.1834,
691
+ "step": 1110
692
+ },
693
+ {
694
+ "epoch": 1.82,
695
+ "learning_rate": 1.6369529983792548e-05,
696
+ "loss": 0.2402,
697
+ "step": 1120
698
+ },
699
+ {
700
+ "epoch": 1.83,
701
+ "learning_rate": 1.6337115072933553e-05,
702
+ "loss": 0.231,
703
+ "step": 1130
704
+ },
705
+ {
706
+ "epoch": 1.85,
707
+ "learning_rate": 1.6304700162074557e-05,
708
+ "loss": 0.1953,
709
+ "step": 1140
710
+ },
711
+ {
712
+ "epoch": 1.86,
713
+ "learning_rate": 1.627228525121556e-05,
714
+ "loss": 0.2261,
715
+ "step": 1150
716
+ },
717
+ {
718
+ "epoch": 1.88,
719
+ "learning_rate": 1.6239870340356566e-05,
720
+ "loss": 0.2854,
721
+ "step": 1160
722
+ },
723
+ {
724
+ "epoch": 1.9,
725
+ "learning_rate": 1.620745542949757e-05,
726
+ "loss": 0.2611,
727
+ "step": 1170
728
+ },
729
+ {
730
+ "epoch": 1.91,
731
+ "learning_rate": 1.6175040518638575e-05,
732
+ "loss": 0.2454,
733
+ "step": 1180
734
+ },
735
+ {
736
+ "epoch": 1.93,
737
+ "learning_rate": 1.614262560777958e-05,
738
+ "loss": 0.3354,
739
+ "step": 1190
740
+ },
741
+ {
742
+ "epoch": 1.94,
743
+ "learning_rate": 1.6110210696920584e-05,
744
+ "loss": 0.2082,
745
+ "step": 1200
746
+ },
747
+ {
748
+ "epoch": 1.96,
749
+ "learning_rate": 1.6077795786061588e-05,
750
+ "loss": 0.2637,
751
+ "step": 1210
752
+ },
753
+ {
754
+ "epoch": 1.98,
755
+ "learning_rate": 1.6045380875202596e-05,
756
+ "loss": 0.2994,
757
+ "step": 1220
758
+ },
759
+ {
760
+ "epoch": 1.99,
761
+ "learning_rate": 1.60129659643436e-05,
762
+ "loss": 0.2793,
763
+ "step": 1230
764
+ },
765
+ {
766
+ "epoch": 2.0,
767
+ "eval_FN": 69,
768
+ "eval_FP": 568,
769
+ "eval_TN": 5270,
770
+ "eval_TP": 526,
771
+ "eval_accuracy": 0.9009793253536452,
772
+ "eval_f1": 0.6228537596210776,
773
+ "eval_loss": 0.2388901710510254,
774
+ "eval_precision": 0.48080438756855576,
775
+ "eval_recall": 0.8840336134453781,
776
+ "eval_runtime": 44.786,
777
+ "eval_samples_per_second": 143.639,
778
+ "eval_steps_per_second": 8.998,
779
+ "step": 1234
780
+ },
781
+ {
782
+ "epoch": 2.01,
783
+ "learning_rate": 1.5980551053484605e-05,
784
+ "loss": 0.2449,
785
+ "step": 1240
786
+ },
787
+ {
788
+ "epoch": 2.03,
789
+ "learning_rate": 1.594813614262561e-05,
790
+ "loss": 0.192,
791
+ "step": 1250
792
+ },
793
+ {
794
+ "epoch": 2.04,
795
+ "learning_rate": 1.5915721231766613e-05,
796
+ "loss": 0.2042,
797
+ "step": 1260
798
+ },
799
+ {
800
+ "epoch": 2.06,
801
+ "learning_rate": 1.5883306320907618e-05,
802
+ "loss": 0.1986,
803
+ "step": 1270
804
+ },
805
+ {
806
+ "epoch": 2.07,
807
+ "learning_rate": 1.5850891410048622e-05,
808
+ "loss": 0.2182,
809
+ "step": 1280
810
+ },
811
+ {
812
+ "epoch": 2.09,
813
+ "learning_rate": 1.5818476499189627e-05,
814
+ "loss": 0.0889,
815
+ "step": 1290
816
+ },
817
+ {
818
+ "epoch": 2.11,
819
+ "learning_rate": 1.578606158833063e-05,
820
+ "loss": 0.1363,
821
+ "step": 1300
822
+ },
823
+ {
824
+ "epoch": 2.12,
825
+ "learning_rate": 1.575364667747164e-05,
826
+ "loss": 0.367,
827
+ "step": 1310
828
+ },
829
+ {
830
+ "epoch": 2.14,
831
+ "learning_rate": 1.5721231766612643e-05,
832
+ "loss": 0.2337,
833
+ "step": 1320
834
+ },
835
+ {
836
+ "epoch": 2.16,
837
+ "learning_rate": 1.5688816855753648e-05,
838
+ "loss": 0.1608,
839
+ "step": 1330
840
+ },
841
+ {
842
+ "epoch": 2.17,
843
+ "learning_rate": 1.5656401944894652e-05,
844
+ "loss": 0.2833,
845
+ "step": 1340
846
+ },
847
+ {
848
+ "epoch": 2.19,
849
+ "learning_rate": 1.5623987034035657e-05,
850
+ "loss": 0.1486,
851
+ "step": 1350
852
+ },
853
+ {
854
+ "epoch": 2.2,
855
+ "learning_rate": 1.559157212317666e-05,
856
+ "loss": 0.2531,
857
+ "step": 1360
858
+ },
859
+ {
860
+ "epoch": 2.22,
861
+ "learning_rate": 1.555915721231767e-05,
862
+ "loss": 0.1957,
863
+ "step": 1370
864
+ },
865
+ {
866
+ "epoch": 2.24,
867
+ "learning_rate": 1.5526742301458673e-05,
868
+ "loss": 0.1777,
869
+ "step": 1380
870
+ },
871
+ {
872
+ "epoch": 2.25,
873
+ "learning_rate": 1.5494327390599678e-05,
874
+ "loss": 0.3041,
875
+ "step": 1390
876
+ },
877
+ {
878
+ "epoch": 2.27,
879
+ "learning_rate": 1.5461912479740682e-05,
880
+ "loss": 0.2384,
881
+ "step": 1400
882
+ },
883
+ {
884
+ "epoch": 2.29,
885
+ "learning_rate": 1.5429497568881687e-05,
886
+ "loss": 0.1801,
887
+ "step": 1410
888
+ },
889
+ {
890
+ "epoch": 2.3,
891
+ "learning_rate": 1.539708265802269e-05,
892
+ "loss": 0.2493,
893
+ "step": 1420
894
+ },
895
+ {
896
+ "epoch": 2.32,
897
+ "learning_rate": 1.53646677471637e-05,
898
+ "loss": 0.1308,
899
+ "step": 1430
900
+ },
901
+ {
902
+ "epoch": 2.33,
903
+ "learning_rate": 1.5332252836304703e-05,
904
+ "loss": 0.3111,
905
+ "step": 1440
906
+ },
907
+ {
908
+ "epoch": 2.35,
909
+ "learning_rate": 1.5299837925445708e-05,
910
+ "loss": 0.3113,
911
+ "step": 1450
912
+ },
913
+ {
914
+ "epoch": 2.37,
915
+ "learning_rate": 1.5267423014586712e-05,
916
+ "loss": 0.1933,
917
+ "step": 1460
918
+ },
919
+ {
920
+ "epoch": 2.38,
921
+ "learning_rate": 1.5235008103727715e-05,
922
+ "loss": 0.1991,
923
+ "step": 1470
924
+ },
925
+ {
926
+ "epoch": 2.4,
927
+ "learning_rate": 1.520259319286872e-05,
928
+ "loss": 0.2317,
929
+ "step": 1480
930
+ },
931
+ {
932
+ "epoch": 2.41,
933
+ "learning_rate": 1.5170178282009727e-05,
934
+ "loss": 0.1928,
935
+ "step": 1490
936
+ },
937
+ {
938
+ "epoch": 2.43,
939
+ "learning_rate": 1.5137763371150732e-05,
940
+ "loss": 0.141,
941
+ "step": 1500
942
+ },
943
+ {
944
+ "epoch": 2.45,
945
+ "learning_rate": 1.5105348460291736e-05,
946
+ "loss": 0.1609,
947
+ "step": 1510
948
+ },
949
+ {
950
+ "epoch": 2.46,
951
+ "learning_rate": 1.507293354943274e-05,
952
+ "loss": 0.1643,
953
+ "step": 1520
954
+ },
955
+ {
956
+ "epoch": 2.48,
957
+ "learning_rate": 1.5040518638573745e-05,
958
+ "loss": 0.1728,
959
+ "step": 1530
960
+ },
961
+ {
962
+ "epoch": 2.5,
963
+ "learning_rate": 1.500810372771475e-05,
964
+ "loss": 0.1344,
965
+ "step": 1540
966
+ },
967
+ {
968
+ "epoch": 2.51,
969
+ "learning_rate": 1.4975688816855755e-05,
970
+ "loss": 0.2989,
971
+ "step": 1550
972
+ },
973
+ {
974
+ "epoch": 2.53,
975
+ "learning_rate": 1.494327390599676e-05,
976
+ "loss": 0.3097,
977
+ "step": 1560
978
+ },
979
+ {
980
+ "epoch": 2.54,
981
+ "learning_rate": 1.4910858995137764e-05,
982
+ "loss": 0.2205,
983
+ "step": 1570
984
+ },
985
+ {
986
+ "epoch": 2.56,
987
+ "learning_rate": 1.4878444084278769e-05,
988
+ "loss": 0.1469,
989
+ "step": 1580
990
+ },
991
+ {
992
+ "epoch": 2.58,
993
+ "learning_rate": 1.4846029173419773e-05,
994
+ "loss": 0.2065,
995
+ "step": 1590
996
+ },
997
+ {
998
+ "epoch": 2.59,
999
+ "learning_rate": 1.4813614262560778e-05,
1000
+ "loss": 0.3045,
1001
+ "step": 1600
1002
+ },
1003
+ {
1004
+ "epoch": 2.61,
1005
+ "learning_rate": 1.4781199351701785e-05,
1006
+ "loss": 0.3229,
1007
+ "step": 1610
1008
+ },
1009
+ {
1010
+ "epoch": 2.63,
1011
+ "learning_rate": 1.474878444084279e-05,
1012
+ "loss": 0.2143,
1013
+ "step": 1620
1014
+ },
1015
+ {
1016
+ "epoch": 2.64,
1017
+ "learning_rate": 1.4716369529983794e-05,
1018
+ "loss": 0.1367,
1019
+ "step": 1630
1020
+ },
1021
+ {
1022
+ "epoch": 2.66,
1023
+ "learning_rate": 1.4683954619124799e-05,
1024
+ "loss": 0.1539,
1025
+ "step": 1640
1026
+ },
1027
+ {
1028
+ "epoch": 2.67,
1029
+ "learning_rate": 1.4651539708265803e-05,
1030
+ "loss": 0.2721,
1031
+ "step": 1650
1032
+ },
1033
+ {
1034
+ "epoch": 2.69,
1035
+ "learning_rate": 1.4619124797406807e-05,
1036
+ "loss": 0.2805,
1037
+ "step": 1660
1038
+ },
1039
+ {
1040
+ "epoch": 2.71,
1041
+ "learning_rate": 1.4586709886547814e-05,
1042
+ "loss": 0.224,
1043
+ "step": 1670
1044
+ },
1045
+ {
1046
+ "epoch": 2.72,
1047
+ "learning_rate": 1.4554294975688818e-05,
1048
+ "loss": 0.1157,
1049
+ "step": 1680
1050
+ },
1051
+ {
1052
+ "epoch": 2.74,
1053
+ "learning_rate": 1.4521880064829822e-05,
1054
+ "loss": 0.2601,
1055
+ "step": 1690
1056
+ },
1057
+ {
1058
+ "epoch": 2.76,
1059
+ "learning_rate": 1.4489465153970827e-05,
1060
+ "loss": 0.1719,
1061
+ "step": 1700
1062
+ },
1063
+ {
1064
+ "epoch": 2.77,
1065
+ "learning_rate": 1.4457050243111831e-05,
1066
+ "loss": 0.1859,
1067
+ "step": 1710
1068
+ },
1069
+ {
1070
+ "epoch": 2.79,
1071
+ "learning_rate": 1.4424635332252836e-05,
1072
+ "loss": 0.2277,
1073
+ "step": 1720
1074
+ },
1075
+ {
1076
+ "epoch": 2.8,
1077
+ "learning_rate": 1.4392220421393844e-05,
1078
+ "loss": 0.2189,
1079
+ "step": 1730
1080
+ },
1081
+ {
1082
+ "epoch": 2.82,
1083
+ "learning_rate": 1.4359805510534848e-05,
1084
+ "loss": 0.3015,
1085
+ "step": 1740
1086
+ },
1087
+ {
1088
+ "epoch": 2.84,
1089
+ "learning_rate": 1.4327390599675852e-05,
1090
+ "loss": 0.0921,
1091
+ "step": 1750
1092
+ },
1093
+ {
1094
+ "epoch": 2.85,
1095
+ "learning_rate": 1.4294975688816857e-05,
1096
+ "loss": 0.1583,
1097
+ "step": 1760
1098
+ },
1099
+ {
1100
+ "epoch": 2.87,
1101
+ "learning_rate": 1.4262560777957861e-05,
1102
+ "loss": 0.1557,
1103
+ "step": 1770
1104
+ },
1105
+ {
1106
+ "epoch": 2.88,
1107
+ "learning_rate": 1.4230145867098866e-05,
1108
+ "loss": 0.2331,
1109
+ "step": 1780
1110
+ },
1111
+ {
1112
+ "epoch": 2.9,
1113
+ "learning_rate": 1.4197730956239872e-05,
1114
+ "loss": 0.1918,
1115
+ "step": 1790
1116
+ },
1117
+ {
1118
+ "epoch": 2.92,
1119
+ "learning_rate": 1.4165316045380876e-05,
1120
+ "loss": 0.2442,
1121
+ "step": 1800
1122
+ },
1123
+ {
1124
+ "epoch": 2.93,
1125
+ "learning_rate": 1.413290113452188e-05,
1126
+ "loss": 0.2091,
1127
+ "step": 1810
1128
+ },
1129
+ {
1130
+ "epoch": 2.95,
1131
+ "learning_rate": 1.4100486223662885e-05,
1132
+ "loss": 0.2063,
1133
+ "step": 1820
1134
+ },
1135
+ {
1136
+ "epoch": 2.97,
1137
+ "learning_rate": 1.4068071312803891e-05,
1138
+ "loss": 0.1886,
1139
+ "step": 1830
1140
+ },
1141
+ {
1142
+ "epoch": 2.98,
1143
+ "learning_rate": 1.4035656401944896e-05,
1144
+ "loss": 0.1629,
1145
+ "step": 1840
1146
+ },
1147
+ {
1148
+ "epoch": 3.0,
1149
+ "learning_rate": 1.4003241491085902e-05,
1150
+ "loss": 0.2031,
1151
+ "step": 1850
1152
+ },
1153
+ {
1154
+ "epoch": 3.0,
1155
+ "eval_FN": 76,
1156
+ "eval_FP": 503,
1157
+ "eval_TN": 5335,
1158
+ "eval_TP": 519,
1159
+ "eval_accuracy": 0.909995336545935,
1160
+ "eval_f1": 0.6419294990723562,
1161
+ "eval_loss": 0.2715374827384949,
1162
+ "eval_precision": 0.5078277886497065,
1163
+ "eval_recall": 0.8722689075630252,
1164
+ "eval_runtime": 44.7829,
1165
+ "eval_samples_per_second": 143.649,
1166
+ "eval_steps_per_second": 8.999,
1167
+ "step": 1851
1168
+ },
1169
+ {
1170
+ "epoch": 3.01,
1171
+ "learning_rate": 1.3970826580226906e-05,
1172
+ "loss": 0.1344,
1173
+ "step": 1860
1174
+ },
1175
+ {
1176
+ "epoch": 3.03,
1177
+ "learning_rate": 1.393841166936791e-05,
1178
+ "loss": 0.1297,
1179
+ "step": 1870
1180
+ },
1181
+ {
1182
+ "epoch": 3.05,
1183
+ "learning_rate": 1.3905996758508915e-05,
1184
+ "loss": 0.1366,
1185
+ "step": 1880
1186
+ },
1187
+ {
1188
+ "epoch": 3.06,
1189
+ "learning_rate": 1.387358184764992e-05,
1190
+ "loss": 0.1826,
1191
+ "step": 1890
1192
+ },
1193
+ {
1194
+ "epoch": 3.08,
1195
+ "learning_rate": 1.3841166936790924e-05,
1196
+ "loss": 0.1133,
1197
+ "step": 1900
1198
+ },
1199
+ {
1200
+ "epoch": 3.1,
1201
+ "learning_rate": 1.380875202593193e-05,
1202
+ "loss": 0.2307,
1203
+ "step": 1910
1204
+ },
1205
+ {
1206
+ "epoch": 3.11,
1207
+ "learning_rate": 1.3776337115072936e-05,
1208
+ "loss": 0.1424,
1209
+ "step": 1920
1210
+ },
1211
+ {
1212
+ "epoch": 3.13,
1213
+ "learning_rate": 1.374392220421394e-05,
1214
+ "loss": 0.0772,
1215
+ "step": 1930
1216
+ },
1217
+ {
1218
+ "epoch": 3.14,
1219
+ "learning_rate": 1.3711507293354945e-05,
1220
+ "loss": 0.0903,
1221
+ "step": 1940
1222
+ },
1223
+ {
1224
+ "epoch": 3.16,
1225
+ "learning_rate": 1.367909238249595e-05,
1226
+ "loss": 0.1002,
1227
+ "step": 1950
1228
+ },
1229
+ {
1230
+ "epoch": 3.18,
1231
+ "learning_rate": 1.3646677471636954e-05,
1232
+ "loss": 0.1644,
1233
+ "step": 1960
1234
+ },
1235
+ {
1236
+ "epoch": 3.19,
1237
+ "learning_rate": 1.361426256077796e-05,
1238
+ "loss": 0.122,
1239
+ "step": 1970
1240
+ },
1241
+ {
1242
+ "epoch": 3.21,
1243
+ "learning_rate": 1.3581847649918964e-05,
1244
+ "loss": 0.1097,
1245
+ "step": 1980
1246
+ },
1247
+ {
1248
+ "epoch": 3.23,
1249
+ "learning_rate": 1.3549432739059969e-05,
1250
+ "loss": 0.1299,
1251
+ "step": 1990
1252
+ },
1253
+ {
1254
+ "epoch": 3.24,
1255
+ "learning_rate": 1.3517017828200973e-05,
1256
+ "loss": 0.2256,
1257
+ "step": 2000
1258
+ },
1259
+ {
1260
+ "epoch": 3.26,
1261
+ "learning_rate": 1.3484602917341978e-05,
1262
+ "loss": 0.16,
1263
+ "step": 2010
1264
+ },
1265
+ {
1266
+ "epoch": 3.27,
1267
+ "learning_rate": 1.3452188006482982e-05,
1268
+ "loss": 0.1247,
1269
+ "step": 2020
1270
+ },
1271
+ {
1272
+ "epoch": 3.29,
1273
+ "learning_rate": 1.341977309562399e-05,
1274
+ "loss": 0.2167,
1275
+ "step": 2030
1276
+ },
1277
+ {
1278
+ "epoch": 3.31,
1279
+ "learning_rate": 1.3387358184764994e-05,
1280
+ "loss": 0.2039,
1281
+ "step": 2040
1282
+ },
1283
+ {
1284
+ "epoch": 3.32,
1285
+ "learning_rate": 1.3354943273905999e-05,
1286
+ "loss": 0.1226,
1287
+ "step": 2050
1288
+ },
1289
+ {
1290
+ "epoch": 3.34,
1291
+ "learning_rate": 1.3322528363047003e-05,
1292
+ "loss": 0.0925,
1293
+ "step": 2060
1294
+ },
1295
+ {
1296
+ "epoch": 3.35,
1297
+ "learning_rate": 1.3290113452188008e-05,
1298
+ "loss": 0.1474,
1299
+ "step": 2070
1300
+ },
1301
+ {
1302
+ "epoch": 3.37,
1303
+ "learning_rate": 1.3257698541329012e-05,
1304
+ "loss": 0.2396,
1305
+ "step": 2080
1306
+ },
1307
+ {
1308
+ "epoch": 3.39,
1309
+ "learning_rate": 1.3225283630470018e-05,
1310
+ "loss": 0.1968,
1311
+ "step": 2090
1312
+ },
1313
+ {
1314
+ "epoch": 3.4,
1315
+ "learning_rate": 1.3192868719611023e-05,
1316
+ "loss": 0.0448,
1317
+ "step": 2100
1318
+ },
1319
+ {
1320
+ "epoch": 3.42,
1321
+ "learning_rate": 1.3160453808752027e-05,
1322
+ "loss": 0.1013,
1323
+ "step": 2110
1324
+ },
1325
+ {
1326
+ "epoch": 3.44,
1327
+ "learning_rate": 1.3128038897893031e-05,
1328
+ "loss": 0.1572,
1329
+ "step": 2120
1330
+ },
1331
+ {
1332
+ "epoch": 3.45,
1333
+ "learning_rate": 1.3095623987034036e-05,
1334
+ "loss": 0.1201,
1335
+ "step": 2130
1336
+ },
1337
+ {
1338
+ "epoch": 3.47,
1339
+ "learning_rate": 1.306320907617504e-05,
1340
+ "loss": 0.1384,
1341
+ "step": 2140
1342
+ },
1343
+ {
1344
+ "epoch": 3.48,
1345
+ "learning_rate": 1.3030794165316048e-05,
1346
+ "loss": 0.2544,
1347
+ "step": 2150
1348
+ },
1349
+ {
1350
+ "epoch": 3.5,
1351
+ "learning_rate": 1.2998379254457052e-05,
1352
+ "loss": 0.1259,
1353
+ "step": 2160
1354
+ },
1355
+ {
1356
+ "epoch": 3.52,
1357
+ "learning_rate": 1.2965964343598057e-05,
1358
+ "loss": 0.0905,
1359
+ "step": 2170
1360
+ },
1361
+ {
1362
+ "epoch": 3.53,
1363
+ "learning_rate": 1.2933549432739061e-05,
1364
+ "loss": 0.2357,
1365
+ "step": 2180
1366
+ },
1367
+ {
1368
+ "epoch": 3.55,
1369
+ "learning_rate": 1.2901134521880066e-05,
1370
+ "loss": 0.161,
1371
+ "step": 2190
1372
+ },
1373
+ {
1374
+ "epoch": 3.57,
1375
+ "learning_rate": 1.286871961102107e-05,
1376
+ "loss": 0.2333,
1377
+ "step": 2200
1378
+ },
1379
+ {
1380
+ "epoch": 3.58,
1381
+ "learning_rate": 1.2836304700162076e-05,
1382
+ "loss": 0.2002,
1383
+ "step": 2210
1384
+ },
1385
+ {
1386
+ "epoch": 3.6,
1387
+ "learning_rate": 1.280388978930308e-05,
1388
+ "loss": 0.1036,
1389
+ "step": 2220
1390
+ },
1391
+ {
1392
+ "epoch": 3.61,
1393
+ "learning_rate": 1.2771474878444085e-05,
1394
+ "loss": 0.1025,
1395
+ "step": 2230
1396
+ },
1397
+ {
1398
+ "epoch": 3.63,
1399
+ "learning_rate": 1.273905996758509e-05,
1400
+ "loss": 0.1449,
1401
+ "step": 2240
1402
+ },
1403
+ {
1404
+ "epoch": 3.65,
1405
+ "learning_rate": 1.2706645056726094e-05,
1406
+ "loss": 0.062,
1407
+ "step": 2250
1408
+ },
1409
+ {
1410
+ "epoch": 3.66,
1411
+ "learning_rate": 1.2674230145867098e-05,
1412
+ "loss": 0.2617,
1413
+ "step": 2260
1414
+ },
1415
+ {
1416
+ "epoch": 3.68,
1417
+ "learning_rate": 1.2641815235008106e-05,
1418
+ "loss": 0.2288,
1419
+ "step": 2270
1420
+ },
1421
+ {
1422
+ "epoch": 3.7,
1423
+ "learning_rate": 1.260940032414911e-05,
1424
+ "loss": 0.113,
1425
+ "step": 2280
1426
+ },
1427
+ {
1428
+ "epoch": 3.71,
1429
+ "learning_rate": 1.2576985413290115e-05,
1430
+ "loss": 0.0937,
1431
+ "step": 2290
1432
+ },
1433
+ {
1434
+ "epoch": 3.73,
1435
+ "learning_rate": 1.254457050243112e-05,
1436
+ "loss": 0.116,
1437
+ "step": 2300
1438
+ },
1439
+ {
1440
+ "epoch": 3.74,
1441
+ "learning_rate": 1.2512155591572124e-05,
1442
+ "loss": 0.1854,
1443
+ "step": 2310
1444
+ },
1445
+ {
1446
+ "epoch": 3.76,
1447
+ "learning_rate": 1.2479740680713128e-05,
1448
+ "loss": 0.2072,
1449
+ "step": 2320
1450
+ },
1451
+ {
1452
+ "epoch": 3.78,
1453
+ "learning_rate": 1.2447325769854134e-05,
1454
+ "loss": 0.1775,
1455
+ "step": 2330
1456
+ },
1457
+ {
1458
+ "epoch": 3.79,
1459
+ "learning_rate": 1.2414910858995139e-05,
1460
+ "loss": 0.0767,
1461
+ "step": 2340
1462
+ },
1463
+ {
1464
+ "epoch": 3.81,
1465
+ "learning_rate": 1.2382495948136143e-05,
1466
+ "loss": 0.1084,
1467
+ "step": 2350
1468
+ },
1469
+ {
1470
+ "epoch": 3.82,
1471
+ "learning_rate": 1.2350081037277148e-05,
1472
+ "loss": 0.2725,
1473
+ "step": 2360
1474
+ },
1475
+ {
1476
+ "epoch": 3.84,
1477
+ "learning_rate": 1.2317666126418152e-05,
1478
+ "loss": 0.2558,
1479
+ "step": 2370
1480
+ },
1481
+ {
1482
+ "epoch": 3.86,
1483
+ "learning_rate": 1.2285251215559157e-05,
1484
+ "loss": 0.0938,
1485
+ "step": 2380
1486
+ },
1487
+ {
1488
+ "epoch": 3.87,
1489
+ "learning_rate": 1.2252836304700164e-05,
1490
+ "loss": 0.2513,
1491
+ "step": 2390
1492
+ },
1493
+ {
1494
+ "epoch": 3.89,
1495
+ "learning_rate": 1.2220421393841169e-05,
1496
+ "loss": 0.2215,
1497
+ "step": 2400
1498
+ },
1499
+ {
1500
+ "epoch": 3.91,
1501
+ "learning_rate": 1.2188006482982173e-05,
1502
+ "loss": 0.1163,
1503
+ "step": 2410
1504
+ },
1505
+ {
1506
+ "epoch": 3.92,
1507
+ "learning_rate": 1.2155591572123178e-05,
1508
+ "loss": 0.0894,
1509
+ "step": 2420
1510
+ },
1511
+ {
1512
+ "epoch": 3.94,
1513
+ "learning_rate": 1.2123176661264182e-05,
1514
+ "loss": 0.1823,
1515
+ "step": 2430
1516
+ },
1517
+ {
1518
+ "epoch": 3.95,
1519
+ "learning_rate": 1.2090761750405187e-05,
1520
+ "loss": 0.1217,
1521
+ "step": 2440
1522
+ },
1523
+ {
1524
+ "epoch": 3.97,
1525
+ "learning_rate": 1.2058346839546193e-05,
1526
+ "loss": 0.1846,
1527
+ "step": 2450
1528
+ },
1529
+ {
1530
+ "epoch": 3.99,
1531
+ "learning_rate": 1.2025931928687197e-05,
1532
+ "loss": 0.2361,
1533
+ "step": 2460
1534
+ },
1535
+ {
1536
+ "epoch": 4.0,
1537
+ "eval_FN": 61,
1538
+ "eval_FP": 589,
1539
+ "eval_TN": 5249,
1540
+ "eval_TP": 534,
1541
+ "eval_accuracy": 0.8989584952588217,
1542
+ "eval_f1": 0.6216530849825379,
1543
+ "eval_loss": 0.3128589689731598,
1544
+ "eval_precision": 0.4755120213713268,
1545
+ "eval_recall": 0.8974789915966387,
1546
+ "eval_runtime": 44.7987,
1547
+ "eval_samples_per_second": 143.598,
1548
+ "eval_steps_per_second": 8.996,
1549
+ "step": 2468
1550
+ },
1551
+ {
1552
+ "epoch": 4.0,
1553
+ "learning_rate": 1.1993517017828202e-05,
1554
+ "loss": 0.241,
1555
+ "step": 2470
1556
+ },
1557
+ {
1558
+ "epoch": 4.02,
1559
+ "learning_rate": 1.1961102106969208e-05,
1560
+ "loss": 0.1445,
1561
+ "step": 2480
1562
+ },
1563
+ {
1564
+ "epoch": 4.04,
1565
+ "learning_rate": 1.1928687196110212e-05,
1566
+ "loss": 0.1197,
1567
+ "step": 2490
1568
+ },
1569
+ {
1570
+ "epoch": 4.05,
1571
+ "learning_rate": 1.1896272285251216e-05,
1572
+ "loss": 0.0886,
1573
+ "step": 2500
1574
+ },
1575
+ {
1576
+ "epoch": 4.07,
1577
+ "learning_rate": 1.1863857374392221e-05,
1578
+ "loss": 0.1759,
1579
+ "step": 2510
1580
+ },
1581
+ {
1582
+ "epoch": 4.08,
1583
+ "learning_rate": 1.1831442463533227e-05,
1584
+ "loss": 0.0796,
1585
+ "step": 2520
1586
+ },
1587
+ {
1588
+ "epoch": 4.1,
1589
+ "learning_rate": 1.1799027552674231e-05,
1590
+ "loss": 0.1523,
1591
+ "step": 2530
1592
+ },
1593
+ {
1594
+ "epoch": 4.12,
1595
+ "learning_rate": 1.1766612641815236e-05,
1596
+ "loss": 0.1979,
1597
+ "step": 2540
1598
+ },
1599
+ {
1600
+ "epoch": 4.13,
1601
+ "learning_rate": 1.173419773095624e-05,
1602
+ "loss": 0.0158,
1603
+ "step": 2550
1604
+ },
1605
+ {
1606
+ "epoch": 4.15,
1607
+ "learning_rate": 1.1701782820097245e-05,
1608
+ "loss": 0.2161,
1609
+ "step": 2560
1610
+ },
1611
+ {
1612
+ "epoch": 4.17,
1613
+ "learning_rate": 1.166936790923825e-05,
1614
+ "loss": 0.2818,
1615
+ "step": 2570
1616
+ },
1617
+ {
1618
+ "epoch": 4.18,
1619
+ "learning_rate": 1.1636952998379257e-05,
1620
+ "loss": 0.0995,
1621
+ "step": 2580
1622
+ },
1623
+ {
1624
+ "epoch": 4.2,
1625
+ "learning_rate": 1.1604538087520261e-05,
1626
+ "loss": 0.1302,
1627
+ "step": 2590
1628
+ },
1629
+ {
1630
+ "epoch": 4.21,
1631
+ "learning_rate": 1.1572123176661266e-05,
1632
+ "loss": 0.1593,
1633
+ "step": 2600
1634
+ },
1635
+ {
1636
+ "epoch": 4.23,
1637
+ "learning_rate": 1.153970826580227e-05,
1638
+ "loss": 0.1202,
1639
+ "step": 2610
1640
+ },
1641
+ {
1642
+ "epoch": 4.25,
1643
+ "learning_rate": 1.1507293354943275e-05,
1644
+ "loss": 0.1004,
1645
+ "step": 2620
1646
+ },
1647
+ {
1648
+ "epoch": 4.26,
1649
+ "learning_rate": 1.1474878444084279e-05,
1650
+ "loss": 0.2116,
1651
+ "step": 2630
1652
+ },
1653
+ {
1654
+ "epoch": 4.28,
1655
+ "learning_rate": 1.1442463533225285e-05,
1656
+ "loss": 0.1037,
1657
+ "step": 2640
1658
+ },
1659
+ {
1660
+ "epoch": 4.29,
1661
+ "learning_rate": 1.141004862236629e-05,
1662
+ "loss": 0.0624,
1663
+ "step": 2650
1664
+ },
1665
+ {
1666
+ "epoch": 4.31,
1667
+ "learning_rate": 1.1377633711507294e-05,
1668
+ "loss": 0.121,
1669
+ "step": 2660
1670
+ },
1671
+ {
1672
+ "epoch": 4.33,
1673
+ "learning_rate": 1.1345218800648299e-05,
1674
+ "loss": 0.1265,
1675
+ "step": 2670
1676
+ },
1677
+ {
1678
+ "epoch": 4.34,
1679
+ "learning_rate": 1.1312803889789303e-05,
1680
+ "loss": 0.0984,
1681
+ "step": 2680
1682
+ },
1683
+ {
1684
+ "epoch": 4.36,
1685
+ "learning_rate": 1.1280388978930307e-05,
1686
+ "loss": 0.0773,
1687
+ "step": 2690
1688
+ },
1689
+ {
1690
+ "epoch": 4.38,
1691
+ "learning_rate": 1.1247974068071315e-05,
1692
+ "loss": 0.1514,
1693
+ "step": 2700
1694
+ },
1695
+ {
1696
+ "epoch": 4.39,
1697
+ "learning_rate": 1.121555915721232e-05,
1698
+ "loss": 0.1744,
1699
+ "step": 2710
1700
+ },
1701
+ {
1702
+ "epoch": 4.41,
1703
+ "learning_rate": 1.1183144246353324e-05,
1704
+ "loss": 0.1826,
1705
+ "step": 2720
1706
+ },
1707
+ {
1708
+ "epoch": 4.42,
1709
+ "learning_rate": 1.1150729335494328e-05,
1710
+ "loss": 0.0464,
1711
+ "step": 2730
1712
+ },
1713
+ {
1714
+ "epoch": 4.44,
1715
+ "learning_rate": 1.1118314424635333e-05,
1716
+ "loss": 0.1258,
1717
+ "step": 2740
1718
+ },
1719
+ {
1720
+ "epoch": 4.46,
1721
+ "learning_rate": 1.1085899513776337e-05,
1722
+ "loss": 0.2924,
1723
+ "step": 2750
1724
+ },
1725
+ {
1726
+ "epoch": 4.47,
1727
+ "learning_rate": 1.1053484602917343e-05,
1728
+ "loss": 0.1871,
1729
+ "step": 2760
1730
+ },
1731
+ {
1732
+ "epoch": 4.49,
1733
+ "learning_rate": 1.1021069692058348e-05,
1734
+ "loss": 0.0574,
1735
+ "step": 2770
1736
+ },
1737
+ {
1738
+ "epoch": 4.51,
1739
+ "learning_rate": 1.0988654781199352e-05,
1740
+ "loss": 0.1691,
1741
+ "step": 2780
1742
+ },
1743
+ {
1744
+ "epoch": 4.52,
1745
+ "learning_rate": 1.0956239870340357e-05,
1746
+ "loss": 0.0645,
1747
+ "step": 2790
1748
+ },
1749
+ {
1750
+ "epoch": 4.54,
1751
+ "learning_rate": 1.0923824959481361e-05,
1752
+ "loss": 0.0356,
1753
+ "step": 2800
1754
+ },
1755
+ {
1756
+ "epoch": 4.55,
1757
+ "learning_rate": 1.0891410048622366e-05,
1758
+ "loss": 0.2102,
1759
+ "step": 2810
1760
+ },
1761
+ {
1762
+ "epoch": 4.57,
1763
+ "learning_rate": 1.0858995137763373e-05,
1764
+ "loss": 0.1308,
1765
+ "step": 2820
1766
+ },
1767
+ {
1768
+ "epoch": 4.59,
1769
+ "learning_rate": 1.0826580226904378e-05,
1770
+ "loss": 0.0124,
1771
+ "step": 2830
1772
+ },
1773
+ {
1774
+ "epoch": 4.6,
1775
+ "learning_rate": 1.0794165316045382e-05,
1776
+ "loss": 0.1287,
1777
+ "step": 2840
1778
+ },
1779
+ {
1780
+ "epoch": 4.62,
1781
+ "learning_rate": 1.0761750405186387e-05,
1782
+ "loss": 0.1295,
1783
+ "step": 2850
1784
+ },
1785
+ {
1786
+ "epoch": 4.64,
1787
+ "learning_rate": 1.0729335494327391e-05,
1788
+ "loss": 0.0696,
1789
+ "step": 2860
1790
+ },
1791
+ {
1792
+ "epoch": 4.65,
1793
+ "learning_rate": 1.0696920583468395e-05,
1794
+ "loss": 0.1446,
1795
+ "step": 2870
1796
+ },
1797
+ {
1798
+ "epoch": 4.67,
1799
+ "learning_rate": 1.0664505672609402e-05,
1800
+ "loss": 0.0996,
1801
+ "step": 2880
1802
+ },
1803
+ {
1804
+ "epoch": 4.68,
1805
+ "learning_rate": 1.0632090761750406e-05,
1806
+ "loss": 0.0866,
1807
+ "step": 2890
1808
+ },
1809
+ {
1810
+ "epoch": 4.7,
1811
+ "learning_rate": 1.059967585089141e-05,
1812
+ "loss": 0.1139,
1813
+ "step": 2900
1814
+ },
1815
+ {
1816
+ "epoch": 4.72,
1817
+ "learning_rate": 1.0567260940032415e-05,
1818
+ "loss": 0.1281,
1819
+ "step": 2910
1820
+ },
1821
+ {
1822
+ "epoch": 4.73,
1823
+ "learning_rate": 1.053484602917342e-05,
1824
+ "loss": 0.1297,
1825
+ "step": 2920
1826
+ },
1827
+ {
1828
+ "epoch": 4.75,
1829
+ "learning_rate": 1.0502431118314424e-05,
1830
+ "loss": 0.1936,
1831
+ "step": 2930
1832
+ },
1833
+ {
1834
+ "epoch": 4.76,
1835
+ "learning_rate": 1.0470016207455432e-05,
1836
+ "loss": 0.1066,
1837
+ "step": 2940
1838
+ },
1839
+ {
1840
+ "epoch": 4.78,
1841
+ "learning_rate": 1.0437601296596436e-05,
1842
+ "loss": 0.0824,
1843
+ "step": 2950
1844
+ },
1845
+ {
1846
+ "epoch": 4.8,
1847
+ "learning_rate": 1.040518638573744e-05,
1848
+ "loss": 0.1188,
1849
+ "step": 2960
1850
+ },
1851
+ {
1852
+ "epoch": 4.81,
1853
+ "learning_rate": 1.0372771474878445e-05,
1854
+ "loss": 0.1891,
1855
+ "step": 2970
1856
+ },
1857
+ {
1858
+ "epoch": 4.83,
1859
+ "learning_rate": 1.034035656401945e-05,
1860
+ "loss": 0.0769,
1861
+ "step": 2980
1862
+ },
1863
+ {
1864
+ "epoch": 4.85,
1865
+ "learning_rate": 1.0307941653160454e-05,
1866
+ "loss": 0.039,
1867
+ "step": 2990
1868
+ },
1869
+ {
1870
+ "epoch": 4.86,
1871
+ "learning_rate": 1.027552674230146e-05,
1872
+ "loss": 0.1004,
1873
+ "step": 3000
1874
+ },
1875
+ {
1876
+ "epoch": 4.88,
1877
+ "learning_rate": 1.0243111831442464e-05,
1878
+ "loss": 0.0931,
1879
+ "step": 3010
1880
+ },
1881
+ {
1882
+ "epoch": 4.89,
1883
+ "learning_rate": 1.0210696920583469e-05,
1884
+ "loss": 0.0804,
1885
+ "step": 3020
1886
+ },
1887
+ {
1888
+ "epoch": 4.91,
1889
+ "learning_rate": 1.0178282009724473e-05,
1890
+ "loss": 0.2742,
1891
+ "step": 3030
1892
+ },
1893
+ {
1894
+ "epoch": 4.93,
1895
+ "learning_rate": 1.014586709886548e-05,
1896
+ "loss": 0.1597,
1897
+ "step": 3040
1898
+ },
1899
+ {
1900
+ "epoch": 4.94,
1901
+ "learning_rate": 1.0113452188006484e-05,
1902
+ "loss": 0.055,
1903
+ "step": 3050
1904
+ },
1905
+ {
1906
+ "epoch": 4.96,
1907
+ "learning_rate": 1.008103727714749e-05,
1908
+ "loss": 0.043,
1909
+ "step": 3060
1910
+ },
1911
+ {
1912
+ "epoch": 4.98,
1913
+ "learning_rate": 1.0048622366288494e-05,
1914
+ "loss": 0.2449,
1915
+ "step": 3070
1916
+ },
1917
+ {
1918
+ "epoch": 4.99,
1919
+ "learning_rate": 1.0016207455429499e-05,
1920
+ "loss": 0.2184,
1921
+ "step": 3080
1922
+ },
1923
+ {
1924
+ "epoch": 5.0,
1925
+ "eval_FN": 54,
1926
+ "eval_FP": 616,
1927
+ "eval_TN": 5222,
1928
+ "eval_TP": 541,
1929
+ "eval_accuracy": 0.89584952588217,
1930
+ "eval_f1": 0.617579908675799,
1931
+ "eval_loss": 0.4803202152252197,
1932
+ "eval_precision": 0.4675885911840968,
1933
+ "eval_recall": 0.9092436974789916,
1934
+ "eval_runtime": 44.7389,
1935
+ "eval_samples_per_second": 143.79,
1936
+ "eval_steps_per_second": 9.008,
1937
+ "step": 3085
1938
+ },
1939
+ {
1940
+ "epoch": 5.01,
1941
+ "learning_rate": 9.983792544570503e-06,
1942
+ "loss": 0.0754,
1943
+ "step": 3090
1944
+ },
1945
+ {
1946
+ "epoch": 5.02,
1947
+ "learning_rate": 9.951377633711507e-06,
1948
+ "loss": 0.0527,
1949
+ "step": 3100
1950
+ },
1951
+ {
1952
+ "epoch": 5.04,
1953
+ "learning_rate": 9.918962722852514e-06,
1954
+ "loss": 0.055,
1955
+ "step": 3110
1956
+ },
1957
+ {
1958
+ "epoch": 5.06,
1959
+ "learning_rate": 9.886547811993518e-06,
1960
+ "loss": 0.0489,
1961
+ "step": 3120
1962
+ },
1963
+ {
1964
+ "epoch": 5.07,
1965
+ "learning_rate": 9.854132901134522e-06,
1966
+ "loss": 0.0637,
1967
+ "step": 3130
1968
+ },
1969
+ {
1970
+ "epoch": 5.09,
1971
+ "learning_rate": 9.821717990275529e-06,
1972
+ "loss": 0.077,
1973
+ "step": 3140
1974
+ },
1975
+ {
1976
+ "epoch": 5.11,
1977
+ "learning_rate": 9.789303079416533e-06,
1978
+ "loss": 0.069,
1979
+ "step": 3150
1980
+ },
1981
+ {
1982
+ "epoch": 5.12,
1983
+ "learning_rate": 9.756888168557537e-06,
1984
+ "loss": 0.1144,
1985
+ "step": 3160
1986
+ },
1987
+ {
1988
+ "epoch": 5.14,
1989
+ "learning_rate": 9.724473257698542e-06,
1990
+ "loss": 0.0577,
1991
+ "step": 3170
1992
+ },
1993
+ {
1994
+ "epoch": 5.15,
1995
+ "learning_rate": 9.692058346839546e-06,
1996
+ "loss": 0.0753,
1997
+ "step": 3180
1998
+ },
1999
+ {
2000
+ "epoch": 5.17,
2001
+ "learning_rate": 9.65964343598055e-06,
2002
+ "loss": 0.0657,
2003
+ "step": 3190
2004
+ },
2005
+ {
2006
+ "epoch": 5.19,
2007
+ "learning_rate": 9.627228525121557e-06,
2008
+ "loss": 0.0783,
2009
+ "step": 3200
2010
+ },
2011
+ {
2012
+ "epoch": 5.2,
2013
+ "learning_rate": 9.594813614262561e-06,
2014
+ "loss": 0.0399,
2015
+ "step": 3210
2016
+ },
2017
+ {
2018
+ "epoch": 5.22,
2019
+ "learning_rate": 9.562398703403566e-06,
2020
+ "loss": 0.231,
2021
+ "step": 3220
2022
+ },
2023
+ {
2024
+ "epoch": 5.24,
2025
+ "learning_rate": 9.529983792544572e-06,
2026
+ "loss": 0.1075,
2027
+ "step": 3230
2028
+ },
2029
+ {
2030
+ "epoch": 5.25,
2031
+ "learning_rate": 9.497568881685576e-06,
2032
+ "loss": 0.1263,
2033
+ "step": 3240
2034
+ },
2035
+ {
2036
+ "epoch": 5.27,
2037
+ "learning_rate": 9.46515397082658e-06,
2038
+ "loss": 0.0937,
2039
+ "step": 3250
2040
+ },
2041
+ {
2042
+ "epoch": 5.28,
2043
+ "learning_rate": 9.432739059967587e-06,
2044
+ "loss": 0.0603,
2045
+ "step": 3260
2046
+ },
2047
+ {
2048
+ "epoch": 5.3,
2049
+ "learning_rate": 9.400324149108591e-06,
2050
+ "loss": 0.1018,
2051
+ "step": 3270
2052
+ },
2053
+ {
2054
+ "epoch": 5.32,
2055
+ "learning_rate": 9.367909238249596e-06,
2056
+ "loss": 0.076,
2057
+ "step": 3280
2058
+ },
2059
+ {
2060
+ "epoch": 5.33,
2061
+ "learning_rate": 9.3354943273906e-06,
2062
+ "loss": 0.133,
2063
+ "step": 3290
2064
+ },
2065
+ {
2066
+ "epoch": 5.35,
2067
+ "learning_rate": 9.303079416531604e-06,
2068
+ "loss": 0.1355,
2069
+ "step": 3300
2070
+ },
2071
+ {
2072
+ "epoch": 5.36,
2073
+ "learning_rate": 9.270664505672609e-06,
2074
+ "loss": 0.1191,
2075
+ "step": 3310
2076
+ },
2077
+ {
2078
+ "epoch": 5.38,
2079
+ "learning_rate": 9.238249594813615e-06,
2080
+ "loss": 0.0789,
2081
+ "step": 3320
2082
+ },
2083
+ {
2084
+ "epoch": 5.4,
2085
+ "learning_rate": 9.20583468395462e-06,
2086
+ "loss": 0.1554,
2087
+ "step": 3330
2088
+ },
2089
+ {
2090
+ "epoch": 5.41,
2091
+ "learning_rate": 9.173419773095624e-06,
2092
+ "loss": 0.1204,
2093
+ "step": 3340
2094
+ },
2095
+ {
2096
+ "epoch": 5.43,
2097
+ "learning_rate": 9.14100486223663e-06,
2098
+ "loss": 0.0855,
2099
+ "step": 3350
2100
+ },
2101
+ {
2102
+ "epoch": 5.45,
2103
+ "learning_rate": 9.108589951377634e-06,
2104
+ "loss": 0.1096,
2105
+ "step": 3360
2106
+ },
2107
+ {
2108
+ "epoch": 5.46,
2109
+ "learning_rate": 9.076175040518639e-06,
2110
+ "loss": 0.0559,
2111
+ "step": 3370
2112
+ },
2113
+ {
2114
+ "epoch": 5.48,
2115
+ "learning_rate": 9.043760129659645e-06,
2116
+ "loss": 0.0689,
2117
+ "step": 3380
2118
+ },
2119
+ {
2120
+ "epoch": 5.49,
2121
+ "learning_rate": 9.01134521880065e-06,
2122
+ "loss": 0.0885,
2123
+ "step": 3390
2124
+ },
2125
+ {
2126
+ "epoch": 5.51,
2127
+ "learning_rate": 8.978930307941654e-06,
2128
+ "loss": 0.0737,
2129
+ "step": 3400
2130
+ },
2131
+ {
2132
+ "epoch": 5.53,
2133
+ "learning_rate": 8.946515397082658e-06,
2134
+ "loss": 0.0501,
2135
+ "step": 3410
2136
+ },
2137
+ {
2138
+ "epoch": 5.54,
2139
+ "learning_rate": 8.914100486223664e-06,
2140
+ "loss": 0.0747,
2141
+ "step": 3420
2142
+ },
2143
+ {
2144
+ "epoch": 5.56,
2145
+ "learning_rate": 8.881685575364669e-06,
2146
+ "loss": 0.1595,
2147
+ "step": 3430
2148
+ },
2149
+ {
2150
+ "epoch": 5.58,
2151
+ "learning_rate": 8.849270664505673e-06,
2152
+ "loss": 0.0288,
2153
+ "step": 3440
2154
+ },
2155
+ {
2156
+ "epoch": 5.59,
2157
+ "learning_rate": 8.816855753646678e-06,
2158
+ "loss": 0.1112,
2159
+ "step": 3450
2160
+ },
2161
+ {
2162
+ "epoch": 5.61,
2163
+ "learning_rate": 8.784440842787682e-06,
2164
+ "loss": 0.1258,
2165
+ "step": 3460
2166
+ },
2167
+ {
2168
+ "epoch": 5.62,
2169
+ "learning_rate": 8.752025931928688e-06,
2170
+ "loss": 0.0343,
2171
+ "step": 3470
2172
+ },
2173
+ {
2174
+ "epoch": 5.64,
2175
+ "learning_rate": 8.719611021069693e-06,
2176
+ "loss": 0.055,
2177
+ "step": 3480
2178
+ },
2179
+ {
2180
+ "epoch": 5.66,
2181
+ "learning_rate": 8.687196110210697e-06,
2182
+ "loss": 0.1253,
2183
+ "step": 3490
2184
+ },
2185
+ {
2186
+ "epoch": 5.67,
2187
+ "learning_rate": 8.654781199351703e-06,
2188
+ "loss": 0.031,
2189
+ "step": 3500
2190
+ },
2191
+ {
2192
+ "epoch": 5.69,
2193
+ "learning_rate": 8.622366288492708e-06,
2194
+ "loss": 0.1125,
2195
+ "step": 3510
2196
+ },
2197
+ {
2198
+ "epoch": 5.71,
2199
+ "learning_rate": 8.589951377633712e-06,
2200
+ "loss": 0.1176,
2201
+ "step": 3520
2202
+ },
2203
+ {
2204
+ "epoch": 5.72,
2205
+ "learning_rate": 8.557536466774718e-06,
2206
+ "loss": 0.1003,
2207
+ "step": 3530
2208
+ },
2209
+ {
2210
+ "epoch": 5.74,
2211
+ "learning_rate": 8.525121555915723e-06,
2212
+ "loss": 0.1414,
2213
+ "step": 3540
2214
+ },
2215
+ {
2216
+ "epoch": 5.75,
2217
+ "learning_rate": 8.492706645056727e-06,
2218
+ "loss": 0.1375,
2219
+ "step": 3550
2220
+ },
2221
+ {
2222
+ "epoch": 5.77,
2223
+ "learning_rate": 8.460291734197731e-06,
2224
+ "loss": 0.0696,
2225
+ "step": 3560
2226
+ },
2227
+ {
2228
+ "epoch": 5.79,
2229
+ "learning_rate": 8.427876823338736e-06,
2230
+ "loss": 0.0804,
2231
+ "step": 3570
2232
+ },
2233
+ {
2234
+ "epoch": 5.8,
2235
+ "learning_rate": 8.39546191247974e-06,
2236
+ "loss": 0.1755,
2237
+ "step": 3580
2238
+ },
2239
+ {
2240
+ "epoch": 5.82,
2241
+ "learning_rate": 8.363047001620746e-06,
2242
+ "loss": 0.1475,
2243
+ "step": 3590
2244
+ },
2245
+ {
2246
+ "epoch": 5.83,
2247
+ "learning_rate": 8.33063209076175e-06,
2248
+ "loss": 0.0467,
2249
+ "step": 3600
2250
+ },
2251
+ {
2252
+ "epoch": 5.85,
2253
+ "learning_rate": 8.298217179902755e-06,
2254
+ "loss": 0.142,
2255
+ "step": 3610
2256
+ },
2257
+ {
2258
+ "epoch": 5.87,
2259
+ "learning_rate": 8.265802269043761e-06,
2260
+ "loss": 0.1949,
2261
+ "step": 3620
2262
+ },
2263
+ {
2264
+ "epoch": 5.88,
2265
+ "learning_rate": 8.233387358184766e-06,
2266
+ "loss": 0.0867,
2267
+ "step": 3630
2268
+ },
2269
+ {
2270
+ "epoch": 5.9,
2271
+ "learning_rate": 8.20097244732577e-06,
2272
+ "loss": 0.0826,
2273
+ "step": 3640
2274
+ },
2275
+ {
2276
+ "epoch": 5.92,
2277
+ "learning_rate": 8.168557536466776e-06,
2278
+ "loss": 0.0974,
2279
+ "step": 3650
2280
+ },
2281
+ {
2282
+ "epoch": 5.93,
2283
+ "learning_rate": 8.13614262560778e-06,
2284
+ "loss": 0.1217,
2285
+ "step": 3660
2286
+ },
2287
+ {
2288
+ "epoch": 5.95,
2289
+ "learning_rate": 8.103727714748785e-06,
2290
+ "loss": 0.1268,
2291
+ "step": 3670
2292
+ },
2293
+ {
2294
+ "epoch": 5.96,
2295
+ "learning_rate": 8.07131280388979e-06,
2296
+ "loss": 0.135,
2297
+ "step": 3680
2298
+ },
2299
+ {
2300
+ "epoch": 5.98,
2301
+ "learning_rate": 8.038897893030794e-06,
2302
+ "loss": 0.0423,
2303
+ "step": 3690
2304
+ },
2305
+ {
2306
+ "epoch": 6.0,
2307
+ "learning_rate": 8.0064829821718e-06,
2308
+ "loss": 0.0832,
2309
+ "step": 3700
2310
+ },
2311
+ {
2312
+ "epoch": 6.0,
2313
+ "eval_FN": 60,
2314
+ "eval_FP": 563,
2315
+ "eval_TN": 5275,
2316
+ "eval_TP": 535,
2317
+ "eval_accuracy": 0.9031556039173014,
2318
+ "eval_f1": 0.6320141760189014,
2319
+ "eval_loss": 0.44203639030456543,
2320
+ "eval_precision": 0.4872495446265938,
2321
+ "eval_recall": 0.8991596638655462,
2322
+ "eval_runtime": 44.7982,
2323
+ "eval_samples_per_second": 143.599,
2324
+ "eval_steps_per_second": 8.996,
2325
+ "step": 3702
2326
+ },
2327
+ {
2328
+ "epoch": 6.01,
2329
+ "learning_rate": 7.974068071312805e-06,
2330
+ "loss": 0.0439,
2331
+ "step": 3710
2332
+ },
2333
+ {
2334
+ "epoch": 6.03,
2335
+ "learning_rate": 7.941653160453809e-06,
2336
+ "loss": 0.0651,
2337
+ "step": 3720
2338
+ },
2339
+ {
2340
+ "epoch": 6.05,
2341
+ "learning_rate": 7.909238249594813e-06,
2342
+ "loss": 0.056,
2343
+ "step": 3730
2344
+ },
2345
+ {
2346
+ "epoch": 6.06,
2347
+ "learning_rate": 7.87682333873582e-06,
2348
+ "loss": 0.076,
2349
+ "step": 3740
2350
+ },
2351
+ {
2352
+ "epoch": 6.08,
2353
+ "learning_rate": 7.844408427876824e-06,
2354
+ "loss": 0.0366,
2355
+ "step": 3750
2356
+ },
2357
+ {
2358
+ "epoch": 6.09,
2359
+ "learning_rate": 7.811993517017828e-06,
2360
+ "loss": 0.0293,
2361
+ "step": 3760
2362
+ },
2363
+ {
2364
+ "epoch": 6.11,
2365
+ "learning_rate": 7.779578606158834e-06,
2366
+ "loss": 0.0501,
2367
+ "step": 3770
2368
+ },
2369
+ {
2370
+ "epoch": 6.13,
2371
+ "learning_rate": 7.747163695299839e-06,
2372
+ "loss": 0.0026,
2373
+ "step": 3780
2374
+ },
2375
+ {
2376
+ "epoch": 6.14,
2377
+ "learning_rate": 7.714748784440843e-06,
2378
+ "loss": 0.0379,
2379
+ "step": 3790
2380
+ },
2381
+ {
2382
+ "epoch": 6.16,
2383
+ "learning_rate": 7.68233387358185e-06,
2384
+ "loss": 0.0209,
2385
+ "step": 3800
2386
+ },
2387
+ {
2388
+ "epoch": 6.18,
2389
+ "learning_rate": 7.649918962722854e-06,
2390
+ "loss": 0.0498,
2391
+ "step": 3810
2392
+ },
2393
+ {
2394
+ "epoch": 6.19,
2395
+ "learning_rate": 7.6175040518638575e-06,
2396
+ "loss": 0.0587,
2397
+ "step": 3820
2398
+ },
2399
+ {
2400
+ "epoch": 6.21,
2401
+ "learning_rate": 7.5850891410048636e-06,
2402
+ "loss": 0.1034,
2403
+ "step": 3830
2404
+ },
2405
+ {
2406
+ "epoch": 6.22,
2407
+ "learning_rate": 7.552674230145868e-06,
2408
+ "loss": 0.0656,
2409
+ "step": 3840
2410
+ },
2411
+ {
2412
+ "epoch": 6.24,
2413
+ "learning_rate": 7.5202593192868724e-06,
2414
+ "loss": 0.0748,
2415
+ "step": 3850
2416
+ },
2417
+ {
2418
+ "epoch": 6.26,
2419
+ "learning_rate": 7.487844408427878e-06,
2420
+ "loss": 0.0762,
2421
+ "step": 3860
2422
+ },
2423
+ {
2424
+ "epoch": 6.27,
2425
+ "learning_rate": 7.455429497568882e-06,
2426
+ "loss": 0.0659,
2427
+ "step": 3870
2428
+ },
2429
+ {
2430
+ "epoch": 6.29,
2431
+ "learning_rate": 7.4230145867098865e-06,
2432
+ "loss": 0.0967,
2433
+ "step": 3880
2434
+ },
2435
+ {
2436
+ "epoch": 6.3,
2437
+ "learning_rate": 7.390599675850893e-06,
2438
+ "loss": 0.0277,
2439
+ "step": 3890
2440
+ },
2441
+ {
2442
+ "epoch": 6.32,
2443
+ "learning_rate": 7.358184764991897e-06,
2444
+ "loss": 0.0407,
2445
+ "step": 3900
2446
+ },
2447
+ {
2448
+ "epoch": 6.34,
2449
+ "learning_rate": 7.3257698541329015e-06,
2450
+ "loss": 0.0252,
2451
+ "step": 3910
2452
+ },
2453
+ {
2454
+ "epoch": 6.35,
2455
+ "learning_rate": 7.293354943273907e-06,
2456
+ "loss": 0.1699,
2457
+ "step": 3920
2458
+ },
2459
+ {
2460
+ "epoch": 6.37,
2461
+ "learning_rate": 7.260940032414911e-06,
2462
+ "loss": 0.0389,
2463
+ "step": 3930
2464
+ },
2465
+ {
2466
+ "epoch": 6.39,
2467
+ "learning_rate": 7.228525121555916e-06,
2468
+ "loss": 0.1974,
2469
+ "step": 3940
2470
+ },
2471
+ {
2472
+ "epoch": 6.4,
2473
+ "learning_rate": 7.196110210696922e-06,
2474
+ "loss": 0.067,
2475
+ "step": 3950
2476
+ },
2477
+ {
2478
+ "epoch": 6.42,
2479
+ "learning_rate": 7.163695299837926e-06,
2480
+ "loss": 0.0766,
2481
+ "step": 3960
2482
+ },
2483
+ {
2484
+ "epoch": 6.43,
2485
+ "learning_rate": 7.131280388978931e-06,
2486
+ "loss": 0.0624,
2487
+ "step": 3970
2488
+ },
2489
+ {
2490
+ "epoch": 6.45,
2491
+ "learning_rate": 7.098865478119936e-06,
2492
+ "loss": 0.0795,
2493
+ "step": 3980
2494
+ },
2495
+ {
2496
+ "epoch": 6.47,
2497
+ "learning_rate": 7.06645056726094e-06,
2498
+ "loss": 0.0439,
2499
+ "step": 3990
2500
+ },
2501
+ {
2502
+ "epoch": 6.48,
2503
+ "learning_rate": 7.034035656401946e-06,
2504
+ "loss": 0.0757,
2505
+ "step": 4000
2506
+ },
2507
+ {
2508
+ "epoch": 6.5,
2509
+ "learning_rate": 7.001620745542951e-06,
2510
+ "loss": 0.0487,
2511
+ "step": 4010
2512
+ },
2513
+ {
2514
+ "epoch": 6.52,
2515
+ "learning_rate": 6.969205834683955e-06,
2516
+ "loss": 0.152,
2517
+ "step": 4020
2518
+ },
2519
+ {
2520
+ "epoch": 6.53,
2521
+ "learning_rate": 6.93679092382496e-06,
2522
+ "loss": 0.0188,
2523
+ "step": 4030
2524
+ },
2525
+ {
2526
+ "epoch": 6.55,
2527
+ "learning_rate": 6.904376012965965e-06,
2528
+ "loss": 0.0358,
2529
+ "step": 4040
2530
+ },
2531
+ {
2532
+ "epoch": 6.56,
2533
+ "learning_rate": 6.87196110210697e-06,
2534
+ "loss": 0.1282,
2535
+ "step": 4050
2536
+ },
2537
+ {
2538
+ "epoch": 6.58,
2539
+ "learning_rate": 6.839546191247975e-06,
2540
+ "loss": 0.0009,
2541
+ "step": 4060
2542
+ },
2543
+ {
2544
+ "epoch": 6.6,
2545
+ "learning_rate": 6.80713128038898e-06,
2546
+ "loss": 0.132,
2547
+ "step": 4070
2548
+ },
2549
+ {
2550
+ "epoch": 6.61,
2551
+ "learning_rate": 6.774716369529984e-06,
2552
+ "loss": 0.0415,
2553
+ "step": 4080
2554
+ },
2555
+ {
2556
+ "epoch": 6.63,
2557
+ "learning_rate": 6.742301458670989e-06,
2558
+ "loss": 0.1097,
2559
+ "step": 4090
2560
+ },
2561
+ {
2562
+ "epoch": 6.65,
2563
+ "learning_rate": 6.709886547811995e-06,
2564
+ "loss": 0.0954,
2565
+ "step": 4100
2566
+ },
2567
+ {
2568
+ "epoch": 6.66,
2569
+ "learning_rate": 6.677471636952999e-06,
2570
+ "loss": 0.1012,
2571
+ "step": 4110
2572
+ },
2573
+ {
2574
+ "epoch": 6.68,
2575
+ "learning_rate": 6.645056726094004e-06,
2576
+ "loss": 0.0772,
2577
+ "step": 4120
2578
+ },
2579
+ {
2580
+ "epoch": 6.69,
2581
+ "learning_rate": 6.612641815235009e-06,
2582
+ "loss": 0.0773,
2583
+ "step": 4130
2584
+ },
2585
+ {
2586
+ "epoch": 6.71,
2587
+ "learning_rate": 6.5802269043760135e-06,
2588
+ "loss": 0.0972,
2589
+ "step": 4140
2590
+ },
2591
+ {
2592
+ "epoch": 6.73,
2593
+ "learning_rate": 6.547811993517018e-06,
2594
+ "loss": 0.0554,
2595
+ "step": 4150
2596
+ },
2597
+ {
2598
+ "epoch": 6.74,
2599
+ "learning_rate": 6.515397082658024e-06,
2600
+ "loss": 0.0312,
2601
+ "step": 4160
2602
+ },
2603
+ {
2604
+ "epoch": 6.76,
2605
+ "learning_rate": 6.4829821717990284e-06,
2606
+ "loss": 0.0158,
2607
+ "step": 4170
2608
+ },
2609
+ {
2610
+ "epoch": 6.77,
2611
+ "learning_rate": 6.450567260940033e-06,
2612
+ "loss": 0.0118,
2613
+ "step": 4180
2614
+ },
2615
+ {
2616
+ "epoch": 6.79,
2617
+ "learning_rate": 6.418152350081038e-06,
2618
+ "loss": 0.146,
2619
+ "step": 4190
2620
+ },
2621
+ {
2622
+ "epoch": 6.81,
2623
+ "learning_rate": 6.3857374392220426e-06,
2624
+ "loss": 0.064,
2625
+ "step": 4200
2626
+ },
2627
+ {
2628
+ "epoch": 6.82,
2629
+ "learning_rate": 6.353322528363047e-06,
2630
+ "loss": 0.1528,
2631
+ "step": 4210
2632
+ },
2633
+ {
2634
+ "epoch": 6.84,
2635
+ "learning_rate": 6.320907617504053e-06,
2636
+ "loss": 0.0069,
2637
+ "step": 4220
2638
+ },
2639
+ {
2640
+ "epoch": 6.86,
2641
+ "learning_rate": 6.2884927066450575e-06,
2642
+ "loss": 0.1694,
2643
+ "step": 4230
2644
+ },
2645
+ {
2646
+ "epoch": 6.87,
2647
+ "learning_rate": 6.256077795786062e-06,
2648
+ "loss": 0.1311,
2649
+ "step": 4240
2650
+ },
2651
+ {
2652
+ "epoch": 6.89,
2653
+ "learning_rate": 6.223662884927067e-06,
2654
+ "loss": 0.0219,
2655
+ "step": 4250
2656
+ },
2657
+ {
2658
+ "epoch": 6.9,
2659
+ "learning_rate": 6.191247974068072e-06,
2660
+ "loss": 0.0153,
2661
+ "step": 4260
2662
+ },
2663
+ {
2664
+ "epoch": 6.92,
2665
+ "learning_rate": 6.158833063209076e-06,
2666
+ "loss": 0.0918,
2667
+ "step": 4270
2668
+ },
2669
+ {
2670
+ "epoch": 6.94,
2671
+ "learning_rate": 6.126418152350082e-06,
2672
+ "loss": 0.1046,
2673
+ "step": 4280
2674
+ },
2675
+ {
2676
+ "epoch": 6.95,
2677
+ "learning_rate": 6.094003241491087e-06,
2678
+ "loss": 0.0357,
2679
+ "step": 4290
2680
+ },
2681
+ {
2682
+ "epoch": 6.97,
2683
+ "learning_rate": 6.061588330632091e-06,
2684
+ "loss": 0.0595,
2685
+ "step": 4300
2686
+ },
2687
+ {
2688
+ "epoch": 6.99,
2689
+ "learning_rate": 6.029173419773096e-06,
2690
+ "loss": 0.0342,
2691
+ "step": 4310
2692
+ },
2693
+ {
2694
+ "epoch": 7.0,
2695
+ "eval_FN": 55,
2696
+ "eval_FP": 646,
2697
+ "eval_TN": 5192,
2698
+ "eval_TP": 540,
2699
+ "eval_accuracy": 0.89103062334836,
2700
+ "eval_f1": 0.6064008983717013,
2701
+ "eval_loss": 0.6455559134483337,
2702
+ "eval_precision": 0.45531197301854975,
2703
+ "eval_recall": 0.907563025210084,
2704
+ "eval_runtime": 44.7662,
2705
+ "eval_samples_per_second": 143.702,
2706
+ "eval_steps_per_second": 9.002,
2707
+ "step": 4319
2708
+ },
2709
+ {
2710
+ "epoch": 7.0,
2711
+ "learning_rate": 5.996758508914101e-06,
2712
+ "loss": 0.095,
2713
+ "step": 4320
2714
+ },
2715
+ {
2716
+ "epoch": 7.02,
2717
+ "learning_rate": 5.964343598055106e-06,
2718
+ "loss": 0.0725,
2719
+ "step": 4330
2720
+ },
2721
+ {
2722
+ "epoch": 7.03,
2723
+ "learning_rate": 5.9319286871961105e-06,
2724
+ "loss": 0.0061,
2725
+ "step": 4340
2726
+ },
2727
+ {
2728
+ "epoch": 7.05,
2729
+ "learning_rate": 5.899513776337116e-06,
2730
+ "loss": 0.0213,
2731
+ "step": 4350
2732
+ },
2733
+ {
2734
+ "epoch": 7.07,
2735
+ "learning_rate": 5.86709886547812e-06,
2736
+ "loss": 0.0244,
2737
+ "step": 4360
2738
+ },
2739
+ {
2740
+ "epoch": 7.08,
2741
+ "learning_rate": 5.834683954619125e-06,
2742
+ "loss": 0.0615,
2743
+ "step": 4370
2744
+ },
2745
+ {
2746
+ "epoch": 7.1,
2747
+ "learning_rate": 5.802269043760131e-06,
2748
+ "loss": 0.051,
2749
+ "step": 4380
2750
+ },
2751
+ {
2752
+ "epoch": 7.12,
2753
+ "learning_rate": 5.769854132901135e-06,
2754
+ "loss": 0.1089,
2755
+ "step": 4390
2756
+ },
2757
+ {
2758
+ "epoch": 7.13,
2759
+ "learning_rate": 5.7374392220421396e-06,
2760
+ "loss": 0.0571,
2761
+ "step": 4400
2762
+ },
2763
+ {
2764
+ "epoch": 7.15,
2765
+ "learning_rate": 5.705024311183145e-06,
2766
+ "loss": 0.043,
2767
+ "step": 4410
2768
+ },
2769
+ {
2770
+ "epoch": 7.16,
2771
+ "learning_rate": 5.672609400324149e-06,
2772
+ "loss": 0.0648,
2773
+ "step": 4420
2774
+ },
2775
+ {
2776
+ "epoch": 7.18,
2777
+ "learning_rate": 5.640194489465154e-06,
2778
+ "loss": 0.0361,
2779
+ "step": 4430
2780
+ },
2781
+ {
2782
+ "epoch": 7.2,
2783
+ "learning_rate": 5.60777957860616e-06,
2784
+ "loss": 0.001,
2785
+ "step": 4440
2786
+ },
2787
+ {
2788
+ "epoch": 7.21,
2789
+ "learning_rate": 5.575364667747164e-06,
2790
+ "loss": 0.1136,
2791
+ "step": 4450
2792
+ },
2793
+ {
2794
+ "epoch": 7.23,
2795
+ "learning_rate": 5.542949756888169e-06,
2796
+ "loss": 0.0582,
2797
+ "step": 4460
2798
+ },
2799
+ {
2800
+ "epoch": 7.24,
2801
+ "learning_rate": 5.510534846029174e-06,
2802
+ "loss": 0.026,
2803
+ "step": 4470
2804
+ },
2805
+ {
2806
+ "epoch": 7.26,
2807
+ "learning_rate": 5.478119935170178e-06,
2808
+ "loss": 0.0304,
2809
+ "step": 4480
2810
+ },
2811
+ {
2812
+ "epoch": 7.28,
2813
+ "learning_rate": 5.445705024311183e-06,
2814
+ "loss": 0.0713,
2815
+ "step": 4490
2816
+ },
2817
+ {
2818
+ "epoch": 7.29,
2819
+ "learning_rate": 5.413290113452189e-06,
2820
+ "loss": 0.0292,
2821
+ "step": 4500
2822
+ },
2823
+ {
2824
+ "epoch": 7.31,
2825
+ "learning_rate": 5.380875202593193e-06,
2826
+ "loss": 0.0716,
2827
+ "step": 4510
2828
+ },
2829
+ {
2830
+ "epoch": 7.33,
2831
+ "learning_rate": 5.348460291734198e-06,
2832
+ "loss": 0.0349,
2833
+ "step": 4520
2834
+ },
2835
+ {
2836
+ "epoch": 7.34,
2837
+ "learning_rate": 5.316045380875203e-06,
2838
+ "loss": 0.0734,
2839
+ "step": 4530
2840
+ },
2841
+ {
2842
+ "epoch": 7.36,
2843
+ "learning_rate": 5.2836304700162074e-06,
2844
+ "loss": 0.022,
2845
+ "step": 4540
2846
+ },
2847
+ {
2848
+ "epoch": 7.37,
2849
+ "learning_rate": 5.251215559157212e-06,
2850
+ "loss": 0.1165,
2851
+ "step": 4550
2852
+ },
2853
+ {
2854
+ "epoch": 7.39,
2855
+ "learning_rate": 5.218800648298218e-06,
2856
+ "loss": 0.101,
2857
+ "step": 4560
2858
+ },
2859
+ {
2860
+ "epoch": 7.41,
2861
+ "learning_rate": 5.186385737439222e-06,
2862
+ "loss": 0.0832,
2863
+ "step": 4570
2864
+ },
2865
+ {
2866
+ "epoch": 7.42,
2867
+ "learning_rate": 5.153970826580227e-06,
2868
+ "loss": 0.0332,
2869
+ "step": 4580
2870
+ },
2871
+ {
2872
+ "epoch": 7.44,
2873
+ "learning_rate": 5.121555915721232e-06,
2874
+ "loss": 0.0535,
2875
+ "step": 4590
2876
+ },
2877
+ {
2878
+ "epoch": 7.46,
2879
+ "learning_rate": 5.0891410048622365e-06,
2880
+ "loss": 0.0232,
2881
+ "step": 4600
2882
+ },
2883
+ {
2884
+ "epoch": 7.47,
2885
+ "learning_rate": 5.056726094003242e-06,
2886
+ "loss": 0.008,
2887
+ "step": 4610
2888
+ },
2889
+ {
2890
+ "epoch": 7.49,
2891
+ "learning_rate": 5.024311183144247e-06,
2892
+ "loss": 0.1731,
2893
+ "step": 4620
2894
+ },
2895
+ {
2896
+ "epoch": 7.5,
2897
+ "learning_rate": 4.9918962722852515e-06,
2898
+ "loss": 0.0713,
2899
+ "step": 4630
2900
+ },
2901
+ {
2902
+ "epoch": 7.52,
2903
+ "learning_rate": 4.959481361426257e-06,
2904
+ "loss": 0.0028,
2905
+ "step": 4640
2906
+ },
2907
+ {
2908
+ "epoch": 7.54,
2909
+ "learning_rate": 4.927066450567261e-06,
2910
+ "loss": 0.0427,
2911
+ "step": 4650
2912
+ },
2913
+ {
2914
+ "epoch": 7.55,
2915
+ "learning_rate": 4.8946515397082665e-06,
2916
+ "loss": 0.0291,
2917
+ "step": 4660
2918
+ },
2919
+ {
2920
+ "epoch": 7.57,
2921
+ "learning_rate": 4.862236628849271e-06,
2922
+ "loss": 0.0315,
2923
+ "step": 4670
2924
+ },
2925
+ {
2926
+ "epoch": 7.59,
2927
+ "learning_rate": 4.829821717990275e-06,
2928
+ "loss": 0.0762,
2929
+ "step": 4680
2930
+ },
2931
+ {
2932
+ "epoch": 7.6,
2933
+ "learning_rate": 4.797406807131281e-06,
2934
+ "loss": 0.0319,
2935
+ "step": 4690
2936
+ },
2937
+ {
2938
+ "epoch": 7.62,
2939
+ "learning_rate": 4.764991896272286e-06,
2940
+ "loss": 0.0569,
2941
+ "step": 4700
2942
+ },
2943
+ {
2944
+ "epoch": 7.63,
2945
+ "learning_rate": 4.73257698541329e-06,
2946
+ "loss": 0.0437,
2947
+ "step": 4710
2948
+ },
2949
+ {
2950
+ "epoch": 7.65,
2951
+ "learning_rate": 4.700162074554296e-06,
2952
+ "loss": 0.077,
2953
+ "step": 4720
2954
+ },
2955
+ {
2956
+ "epoch": 7.67,
2957
+ "learning_rate": 4.6677471636953e-06,
2958
+ "loss": 0.0344,
2959
+ "step": 4730
2960
+ },
2961
+ {
2962
+ "epoch": 7.68,
2963
+ "learning_rate": 4.6353322528363044e-06,
2964
+ "loss": 0.1271,
2965
+ "step": 4740
2966
+ },
2967
+ {
2968
+ "epoch": 7.7,
2969
+ "learning_rate": 4.60291734197731e-06,
2970
+ "loss": 0.0286,
2971
+ "step": 4750
2972
+ },
2973
+ {
2974
+ "epoch": 7.71,
2975
+ "learning_rate": 4.570502431118315e-06,
2976
+ "loss": 0.039,
2977
+ "step": 4760
2978
+ },
2979
+ {
2980
+ "epoch": 7.73,
2981
+ "learning_rate": 4.538087520259319e-06,
2982
+ "loss": 0.0562,
2983
+ "step": 4770
2984
+ },
2985
+ {
2986
+ "epoch": 7.75,
2987
+ "learning_rate": 4.505672609400325e-06,
2988
+ "loss": 0.0415,
2989
+ "step": 4780
2990
+ },
2991
+ {
2992
+ "epoch": 7.76,
2993
+ "learning_rate": 4.473257698541329e-06,
2994
+ "loss": 0.0319,
2995
+ "step": 4790
2996
+ },
2997
+ {
2998
+ "epoch": 7.78,
2999
+ "learning_rate": 4.440842787682334e-06,
3000
+ "loss": 0.0464,
3001
+ "step": 4800
3002
+ },
3003
+ {
3004
+ "epoch": 7.8,
3005
+ "learning_rate": 4.408427876823339e-06,
3006
+ "loss": 0.0059,
3007
+ "step": 4810
3008
+ },
3009
+ {
3010
+ "epoch": 7.81,
3011
+ "learning_rate": 4.376012965964344e-06,
3012
+ "loss": 0.0504,
3013
+ "step": 4820
3014
+ },
3015
+ {
3016
+ "epoch": 7.83,
3017
+ "learning_rate": 4.3435980551053485e-06,
3018
+ "loss": 0.0406,
3019
+ "step": 4830
3020
+ },
3021
+ {
3022
+ "epoch": 7.84,
3023
+ "learning_rate": 4.311183144246354e-06,
3024
+ "loss": 0.0235,
3025
+ "step": 4840
3026
+ },
3027
+ {
3028
+ "epoch": 7.86,
3029
+ "learning_rate": 4.278768233387359e-06,
3030
+ "loss": 0.0055,
3031
+ "step": 4850
3032
+ },
3033
+ {
3034
+ "epoch": 7.88,
3035
+ "learning_rate": 4.2463533225283635e-06,
3036
+ "loss": 0.004,
3037
+ "step": 4860
3038
+ },
3039
+ {
3040
+ "epoch": 7.89,
3041
+ "learning_rate": 4.213938411669368e-06,
3042
+ "loss": 0.0492,
3043
+ "step": 4870
3044
+ },
3045
+ {
3046
+ "epoch": 7.91,
3047
+ "learning_rate": 4.181523500810373e-06,
3048
+ "loss": 0.0658,
3049
+ "step": 4880
3050
+ },
3051
+ {
3052
+ "epoch": 7.93,
3053
+ "learning_rate": 4.149108589951378e-06,
3054
+ "loss": 0.0013,
3055
+ "step": 4890
3056
+ },
3057
+ {
3058
+ "epoch": 7.94,
3059
+ "learning_rate": 4.116693679092383e-06,
3060
+ "loss": 0.0534,
3061
+ "step": 4900
3062
+ },
3063
+ {
3064
+ "epoch": 7.96,
3065
+ "learning_rate": 4.084278768233388e-06,
3066
+ "loss": 0.0399,
3067
+ "step": 4910
3068
+ },
3069
+ {
3070
+ "epoch": 7.97,
3071
+ "learning_rate": 4.0518638573743926e-06,
3072
+ "loss": 0.0045,
3073
+ "step": 4920
3074
+ },
3075
+ {
3076
+ "epoch": 7.99,
3077
+ "learning_rate": 4.019448946515397e-06,
3078
+ "loss": 0.1337,
3079
+ "step": 4930
3080
+ },
3081
+ {
3082
+ "epoch": 8.0,
3083
+ "eval_FN": 101,
3084
+ "eval_FP": 438,
3085
+ "eval_TN": 5400,
3086
+ "eval_TP": 494,
3087
+ "eval_accuracy": 0.9162132752992383,
3088
+ "eval_f1": 0.6470203012442698,
3089
+ "eval_loss": 0.5036618113517761,
3090
+ "eval_precision": 0.5300429184549357,
3091
+ "eval_recall": 0.8302521008403362,
3092
+ "eval_runtime": 44.7944,
3093
+ "eval_samples_per_second": 143.612,
3094
+ "eval_steps_per_second": 8.997,
3095
+ "step": 4936
3096
+ },
3097
+ {
3098
+ "epoch": 8.01,
3099
+ "learning_rate": 3.987034035656402e-06,
3100
+ "loss": 0.0456,
3101
+ "step": 4940
3102
+ },
3103
+ {
3104
+ "epoch": 8.02,
3105
+ "learning_rate": 3.954619124797407e-06,
3106
+ "loss": 0.1101,
3107
+ "step": 4950
3108
+ },
3109
+ {
3110
+ "epoch": 8.04,
3111
+ "learning_rate": 3.922204213938412e-06,
3112
+ "loss": 0.0013,
3113
+ "step": 4960
3114
+ },
3115
+ {
3116
+ "epoch": 8.06,
3117
+ "learning_rate": 3.889789303079417e-06,
3118
+ "loss": 0.0196,
3119
+ "step": 4970
3120
+ },
3121
+ {
3122
+ "epoch": 8.07,
3123
+ "learning_rate": 3.857374392220422e-06,
3124
+ "loss": 0.0665,
3125
+ "step": 4980
3126
+ },
3127
+ {
3128
+ "epoch": 8.09,
3129
+ "learning_rate": 3.824959481361427e-06,
3130
+ "loss": 0.001,
3131
+ "step": 4990
3132
+ },
3133
+ {
3134
+ "epoch": 8.1,
3135
+ "learning_rate": 3.7925445705024318e-06,
3136
+ "loss": 0.0069,
3137
+ "step": 5000
3138
+ },
3139
+ {
3140
+ "epoch": 8.12,
3141
+ "learning_rate": 3.7601296596434362e-06,
3142
+ "loss": 0.0028,
3143
+ "step": 5010
3144
+ },
3145
+ {
3146
+ "epoch": 8.14,
3147
+ "learning_rate": 3.727714748784441e-06,
3148
+ "loss": 0.0008,
3149
+ "step": 5020
3150
+ },
3151
+ {
3152
+ "epoch": 8.15,
3153
+ "learning_rate": 3.6952998379254463e-06,
3154
+ "loss": 0.0337,
3155
+ "step": 5030
3156
+ },
3157
+ {
3158
+ "epoch": 8.17,
3159
+ "learning_rate": 3.6628849270664508e-06,
3160
+ "loss": 0.0011,
3161
+ "step": 5040
3162
+ },
3163
+ {
3164
+ "epoch": 8.18,
3165
+ "learning_rate": 3.6304700162074556e-06,
3166
+ "loss": 0.037,
3167
+ "step": 5050
3168
+ },
3169
+ {
3170
+ "epoch": 8.2,
3171
+ "learning_rate": 3.598055105348461e-06,
3172
+ "loss": 0.0882,
3173
+ "step": 5060
3174
+ },
3175
+ {
3176
+ "epoch": 8.22,
3177
+ "learning_rate": 3.5656401944894653e-06,
3178
+ "loss": 0.0006,
3179
+ "step": 5070
3180
+ },
3181
+ {
3182
+ "epoch": 8.23,
3183
+ "learning_rate": 3.53322528363047e-06,
3184
+ "loss": 0.0006,
3185
+ "step": 5080
3186
+ },
3187
+ {
3188
+ "epoch": 8.25,
3189
+ "learning_rate": 3.5008103727714754e-06,
3190
+ "loss": 0.0067,
3191
+ "step": 5090
3192
+ },
3193
+ {
3194
+ "epoch": 8.27,
3195
+ "learning_rate": 3.46839546191248e-06,
3196
+ "loss": 0.0289,
3197
+ "step": 5100
3198
+ },
3199
+ {
3200
+ "epoch": 8.28,
3201
+ "learning_rate": 3.435980551053485e-06,
3202
+ "loss": 0.0096,
3203
+ "step": 5110
3204
+ },
3205
+ {
3206
+ "epoch": 8.3,
3207
+ "learning_rate": 3.40356564019449e-06,
3208
+ "loss": 0.0753,
3209
+ "step": 5120
3210
+ },
3211
+ {
3212
+ "epoch": 8.31,
3213
+ "learning_rate": 3.3711507293354944e-06,
3214
+ "loss": 0.0247,
3215
+ "step": 5130
3216
+ },
3217
+ {
3218
+ "epoch": 8.33,
3219
+ "learning_rate": 3.3387358184764997e-06,
3220
+ "loss": 0.0433,
3221
+ "step": 5140
3222
+ },
3223
+ {
3224
+ "epoch": 8.35,
3225
+ "learning_rate": 3.3063209076175045e-06,
3226
+ "loss": 0.1251,
3227
+ "step": 5150
3228
+ },
3229
+ {
3230
+ "epoch": 8.36,
3231
+ "learning_rate": 3.273905996758509e-06,
3232
+ "loss": 0.0194,
3233
+ "step": 5160
3234
+ },
3235
+ {
3236
+ "epoch": 8.38,
3237
+ "learning_rate": 3.2414910858995142e-06,
3238
+ "loss": 0.004,
3239
+ "step": 5170
3240
+ },
3241
+ {
3242
+ "epoch": 8.4,
3243
+ "learning_rate": 3.209076175040519e-06,
3244
+ "loss": 0.0011,
3245
+ "step": 5180
3246
+ },
3247
+ {
3248
+ "epoch": 8.41,
3249
+ "learning_rate": 3.1766612641815235e-06,
3250
+ "loss": 0.0995,
3251
+ "step": 5190
3252
+ },
3253
+ {
3254
+ "epoch": 8.43,
3255
+ "learning_rate": 3.1442463533225288e-06,
3256
+ "loss": 0.0005,
3257
+ "step": 5200
3258
+ },
3259
+ {
3260
+ "epoch": 8.44,
3261
+ "learning_rate": 3.1118314424635336e-06,
3262
+ "loss": 0.0203,
3263
+ "step": 5210
3264
+ },
3265
+ {
3266
+ "epoch": 8.46,
3267
+ "learning_rate": 3.079416531604538e-06,
3268
+ "loss": 0.0848,
3269
+ "step": 5220
3270
+ },
3271
+ {
3272
+ "epoch": 8.48,
3273
+ "learning_rate": 3.0470016207455433e-06,
3274
+ "loss": 0.0007,
3275
+ "step": 5230
3276
+ },
3277
+ {
3278
+ "epoch": 8.49,
3279
+ "learning_rate": 3.014586709886548e-06,
3280
+ "loss": 0.0992,
3281
+ "step": 5240
3282
+ },
3283
+ {
3284
+ "epoch": 8.51,
3285
+ "learning_rate": 2.982171799027553e-06,
3286
+ "loss": 0.0399,
3287
+ "step": 5250
3288
+ },
3289
+ {
3290
+ "epoch": 8.53,
3291
+ "learning_rate": 2.949756888168558e-06,
3292
+ "loss": 0.0005,
3293
+ "step": 5260
3294
+ },
3295
+ {
3296
+ "epoch": 8.54,
3297
+ "learning_rate": 2.9173419773095623e-06,
3298
+ "loss": 0.0265,
3299
+ "step": 5270
3300
+ },
3301
+ {
3302
+ "epoch": 8.56,
3303
+ "learning_rate": 2.8849270664505676e-06,
3304
+ "loss": 0.0405,
3305
+ "step": 5280
3306
+ },
3307
+ {
3308
+ "epoch": 8.57,
3309
+ "learning_rate": 2.8525121555915724e-06,
3310
+ "loss": 0.0052,
3311
+ "step": 5290
3312
+ },
3313
+ {
3314
+ "epoch": 8.59,
3315
+ "learning_rate": 2.820097244732577e-06,
3316
+ "loss": 0.0208,
3317
+ "step": 5300
3318
+ },
3319
+ {
3320
+ "epoch": 8.61,
3321
+ "learning_rate": 2.787682333873582e-06,
3322
+ "loss": 0.0336,
3323
+ "step": 5310
3324
+ },
3325
+ {
3326
+ "epoch": 8.62,
3327
+ "learning_rate": 2.755267423014587e-06,
3328
+ "loss": 0.0377,
3329
+ "step": 5320
3330
+ },
3331
+ {
3332
+ "epoch": 8.64,
3333
+ "learning_rate": 2.7228525121555914e-06,
3334
+ "loss": 0.0631,
3335
+ "step": 5330
3336
+ },
3337
+ {
3338
+ "epoch": 8.65,
3339
+ "learning_rate": 2.6904376012965967e-06,
3340
+ "loss": 0.0068,
3341
+ "step": 5340
3342
+ },
3343
+ {
3344
+ "epoch": 8.67,
3345
+ "learning_rate": 2.6580226904376015e-06,
3346
+ "loss": 0.0488,
3347
+ "step": 5350
3348
+ },
3349
+ {
3350
+ "epoch": 8.69,
3351
+ "learning_rate": 2.625607779578606e-06,
3352
+ "loss": 0.0006,
3353
+ "step": 5360
3354
+ },
3355
+ {
3356
+ "epoch": 8.7,
3357
+ "learning_rate": 2.593192868719611e-06,
3358
+ "loss": 0.0055,
3359
+ "step": 5370
3360
+ },
3361
+ {
3362
+ "epoch": 8.72,
3363
+ "learning_rate": 2.560777957860616e-06,
3364
+ "loss": 0.0019,
3365
+ "step": 5380
3366
+ },
3367
+ {
3368
+ "epoch": 8.74,
3369
+ "learning_rate": 2.528363047001621e-06,
3370
+ "loss": 0.0215,
3371
+ "step": 5390
3372
+ },
3373
+ {
3374
+ "epoch": 8.75,
3375
+ "learning_rate": 2.4959481361426258e-06,
3376
+ "loss": 0.0643,
3377
+ "step": 5400
3378
+ },
3379
+ {
3380
+ "epoch": 8.77,
3381
+ "learning_rate": 2.4635332252836306e-06,
3382
+ "loss": 0.0743,
3383
+ "step": 5410
3384
+ },
3385
+ {
3386
+ "epoch": 8.78,
3387
+ "learning_rate": 2.4311183144246355e-06,
3388
+ "loss": 0.0041,
3389
+ "step": 5420
3390
+ },
3391
+ {
3392
+ "epoch": 8.8,
3393
+ "learning_rate": 2.3987034035656403e-06,
3394
+ "loss": 0.0873,
3395
+ "step": 5430
3396
+ },
3397
+ {
3398
+ "epoch": 8.82,
3399
+ "learning_rate": 2.366288492706645e-06,
3400
+ "loss": 0.0015,
3401
+ "step": 5440
3402
+ },
3403
+ {
3404
+ "epoch": 8.83,
3405
+ "learning_rate": 2.33387358184765e-06,
3406
+ "loss": 0.1603,
3407
+ "step": 5450
3408
+ },
3409
+ {
3410
+ "epoch": 8.85,
3411
+ "learning_rate": 2.301458670988655e-06,
3412
+ "loss": 0.0468,
3413
+ "step": 5460
3414
+ },
3415
+ {
3416
+ "epoch": 8.87,
3417
+ "learning_rate": 2.2690437601296597e-06,
3418
+ "loss": 0.0219,
3419
+ "step": 5470
3420
+ },
3421
+ {
3422
+ "epoch": 8.88,
3423
+ "learning_rate": 2.2366288492706646e-06,
3424
+ "loss": 0.0029,
3425
+ "step": 5480
3426
+ },
3427
+ {
3428
+ "epoch": 8.9,
3429
+ "learning_rate": 2.2042139384116694e-06,
3430
+ "loss": 0.0422,
3431
+ "step": 5490
3432
+ },
3433
+ {
3434
+ "epoch": 8.91,
3435
+ "learning_rate": 2.1717990275526743e-06,
3436
+ "loss": 0.0023,
3437
+ "step": 5500
3438
+ },
3439
+ {
3440
+ "epoch": 8.93,
3441
+ "learning_rate": 2.1393841166936795e-06,
3442
+ "loss": 0.0005,
3443
+ "step": 5510
3444
+ },
3445
+ {
3446
+ "epoch": 8.95,
3447
+ "learning_rate": 2.106969205834684e-06,
3448
+ "loss": 0.0008,
3449
+ "step": 5520
3450
+ },
3451
+ {
3452
+ "epoch": 8.96,
3453
+ "learning_rate": 2.074554294975689e-06,
3454
+ "loss": 0.0513,
3455
+ "step": 5530
3456
+ },
3457
+ {
3458
+ "epoch": 8.98,
3459
+ "learning_rate": 2.042139384116694e-06,
3460
+ "loss": 0.0003,
3461
+ "step": 5540
3462
+ },
3463
+ {
3464
+ "epoch": 9.0,
3465
+ "learning_rate": 2.0097244732576985e-06,
3466
+ "loss": 0.0009,
3467
+ "step": 5550
3468
+ },
3469
+ {
3470
+ "epoch": 9.0,
3471
+ "eval_FN": 76,
3472
+ "eval_FP": 478,
3473
+ "eval_TN": 5360,
3474
+ "eval_TP": 519,
3475
+ "eval_accuracy": 0.9138815482667496,
3476
+ "eval_f1": 0.6520100502512564,
3477
+ "eval_loss": 0.6137506365776062,
3478
+ "eval_precision": 0.5205616850551655,
3479
+ "eval_recall": 0.8722689075630252,
3480
+ "eval_runtime": 44.7513,
3481
+ "eval_samples_per_second": 143.75,
3482
+ "eval_steps_per_second": 9.005,
3483
+ "step": 5553
3484
+ }
3485
+ ],
3486
+ "logging_steps": 10,
3487
+ "max_steps": 6170,
3488
+ "num_train_epochs": 10,
3489
+ "save_steps": 500,
3490
+ "total_flos": 2.336741904863232e+16,
3491
+ "trial_name": null,
3492
+ "trial_params": null
3493
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:05ad746cadd1b446b5fa3fb4525e42d41183a1e69e4572ef40a89cac990f4a13
3
+ size 4664
vocab.json ADDED
The diff for this file is too large to render. See raw diff