Hugging Face's logo

altsoph
/
chadpod

Text Classification
Transformers
Safetensors
deberta-v2
Model card Files
xet
Community
altsoph commited on
Commit
ce4155b
·
verified ·
1 Parent(s): 2334092

Upload folder using huggingface_hub

Browse files
Files changed (11) hide show
  1. added_tokens.json +3 -0
  2. config.json +1 -1
  3. optimizer.pt +3 -0
  4. rng_state.pth +3 -0
  5. scheduler.pt +3 -0
  6. special_tokens_map.json +8 -8
  7. spm.model +3 -0
  8. tokenizer.json +0 -0
  9. tokenizer_config.json +15 -14
  10. trainer_state.json +3713 -0
  11. training_args.bin +3 -0
added_tokens.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ {
2
+ "[MASK]": 128000
3
+ }
config.json CHANGED
@@ -1,5 +1,5 @@
1
  {
2
- "_name_or_path": "deberta-v3-large-lr55-e6-bs4-acc6-dsv3/checkpoint-512",
3
  "architectures": [
4
  "DebertaV2ForSequenceClassification"
5
  ],
 
1
  {
2
+ "_name_or_path": "microsoft/deberta-v3-large",
3
  "architectures": [
4
  "DebertaV2ForSequenceClassification"
5
  ],
optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f5662b941bfc8e9ecb05da2c6350d69ce954beb3f003c70abd87a2da926f6f75
3
+ size 3480840240
rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8d9fa451142d44ec6a3fc7afb84554ad03fd6ad0b403f44bbefe847f25f9612a
3
+ size 14244
scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:aca2fad444673dd2b910ffd31ca8b43be06cfe03841574578623912550c4bc24
3
+ size 1064
special_tokens_map.json CHANGED
@@ -2,14 +2,14 @@
2
  "bos_token": "[CLS]",
3
  "cls_token": "[CLS]",
4
  "eos_token": "[SEP]",
5
- "mask_token": {
6
- "content": "[MASK]",
7
- "lstrip": true,
8
- "normalized": false,
 
 
 
9
  "rstrip": false,
10
  "single_word": false
11
- },
12
- "pad_token": "<pad>",
13
- "sep_token": "[SEP]",
14
- "unk_token": "<unk>"
15
  }
 
2
  "bos_token": "[CLS]",
3
  "cls_token": "[CLS]",
4
  "eos_token": "[SEP]",
5
+ "mask_token": "[MASK]",
6
+ "pad_token": "[PAD]",
7
+ "sep_token": "[SEP]",
8
+ "unk_token": {
9
+ "content": "[UNK]",
10
+ "lstrip": false,
11
+ "normalized": true,
12
  "rstrip": false,
13
  "single_word": false
14
+ }
 
 
 
15
  }
spm.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c679fbf93643d19aab7ee10c0b99e460bdbc02fedf34b92b05af343b4af586fd
3
+ size 2464616
tokenizer.json CHANGED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json CHANGED
@@ -1,7 +1,7 @@
1
  {
2
  "added_tokens_decoder": {
3
  "0": {
4
- "content": "<pad>",
5
  "lstrip": false,
6
  "normalized": false,
7
  "rstrip": false,
@@ -9,7 +9,7 @@
9
  "special": true
10
  },
11
  "1": {
12
- "content": "<unk>",
13
  "lstrip": false,
14
  "normalized": false,
15
  "rstrip": false,
@@ -17,7 +17,7 @@
17
  "special": true
18
  },
19
  "2": {
20
- "content": "[CLS]",
21
  "lstrip": false,
22
  "normalized": false,
23
  "rstrip": false,
@@ -25,16 +25,16 @@
25
  "special": true
26
  },
27
  "3": {
28
- "content": "[SEP]",
29
  "lstrip": false,
30
- "normalized": false,
31
  "rstrip": false,
32
  "single_word": false,
33
  "special": true
34
  },
35
- "4": {
36
  "content": "[MASK]",
37
- "lstrip": true,
38
  "normalized": false,
39
  "rstrip": false,
40
  "single_word": false,
@@ -44,14 +44,15 @@
44
  "bos_token": "[CLS]",
45
  "clean_up_tokenization_spaces": true,
46
  "cls_token": "[CLS]",
47
- "do_lower_case": true,
48
  "eos_token": "[SEP]",
49
- "keep_accents": false,
50
  "mask_token": "[MASK]",
51
- "model_max_length": 512,
52
- "pad_token": "<pad>",
53
- "remove_space": true,
54
  "sep_token": "[SEP]",
55
- "tokenizer_class": "AlbertTokenizer",
56
- "unk_token": "<unk>"
 
 
 
57
  }
 
1
  {
2
  "added_tokens_decoder": {
3
  "0": {
4
+ "content": "[PAD]",
5
  "lstrip": false,
6
  "normalized": false,
7
  "rstrip": false,
 
9
  "special": true
10
  },
11
  "1": {
12
+ "content": "[CLS]",
13
  "lstrip": false,
14
  "normalized": false,
15
  "rstrip": false,
 
17
  "special": true
18
  },
19
  "2": {
20
+ "content": "[SEP]",
21
  "lstrip": false,
22
  "normalized": false,
23
  "rstrip": false,
 
25
  "special": true
26
  },
27
  "3": {
28
+ "content": "[UNK]",
29
  "lstrip": false,
30
+ "normalized": true,
31
  "rstrip": false,
32
  "single_word": false,
33
  "special": true
34
  },
35
+ "128000": {
36
  "content": "[MASK]",
37
+ "lstrip": false,
38
  "normalized": false,
39
  "rstrip": false,
40
  "single_word": false,
 
44
  "bos_token": "[CLS]",
45
  "clean_up_tokenization_spaces": true,
46
  "cls_token": "[CLS]",
47
+ "do_lower_case": false,
48
  "eos_token": "[SEP]",
 
49
  "mask_token": "[MASK]",
50
+ "model_max_length": 1000000000000000019884624838656,
51
+ "pad_token": "[PAD]",
 
52
  "sep_token": "[SEP]",
53
+ "sp_model_kwargs": {},
54
+ "split_by_punct": false,
55
+ "tokenizer_class": "DebertaV2Tokenizer",
56
+ "unk_token": "[UNK]",
57
+ "vocab_type": "spm"
58
  }
trainer_state.json ADDED
@@ -0,0 +1,3713 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.4087750315666199,
3
+ "best_model_checkpoint": "deberta-v3-large-lr55-e6-bs4-acc6-dsv3/checkpoint-213",
4
+ "epoch": 12.0,
5
+ "eval_steps": 500,
6
+ "global_step": 512,
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
+ "grad_norm": 1.8460181951522827,
14
+ "learning_rate": 5.496726190476191e-06,
15
+ "loss": 0.6818,
16
+ "step": 1
17
+ },
18
+ {
19
+ "epoch": 0.05,
20
+ "grad_norm": 1.9176546335220337,
21
+ "learning_rate": 5.493452380952381e-06,
22
+ "loss": 0.7102,
23
+ "step": 2
24
+ },
25
+ {
26
+ "epoch": 0.07,
27
+ "grad_norm": 2.076033353805542,
28
+ "learning_rate": 5.490178571428571e-06,
29
+ "loss": 0.712,
30
+ "step": 3
31
+ },
32
+ {
33
+ "epoch": 0.09,
34
+ "grad_norm": 0.9934174418449402,
35
+ "learning_rate": 5.486904761904762e-06,
36
+ "loss": 0.6877,
37
+ "step": 4
38
+ },
39
+ {
40
+ "epoch": 0.12,
41
+ "grad_norm": 3.206833600997925,
42
+ "learning_rate": 5.483630952380952e-06,
43
+ "loss": 0.7138,
44
+ "step": 5
45
+ },
46
+ {
47
+ "epoch": 0.14,
48
+ "grad_norm": 0.930253803730011,
49
+ "learning_rate": 5.480357142857143e-06,
50
+ "loss": 0.675,
51
+ "step": 6
52
+ },
53
+ {
54
+ "epoch": 0.16,
55
+ "grad_norm": 1.047216773033142,
56
+ "learning_rate": 5.477083333333333e-06,
57
+ "loss": 0.7025,
58
+ "step": 7
59
+ },
60
+ {
61
+ "epoch": 0.19,
62
+ "grad_norm": 2.5932674407958984,
63
+ "learning_rate": 5.473809523809524e-06,
64
+ "loss": 0.6736,
65
+ "step": 8
66
+ },
67
+ {
68
+ "epoch": 0.21,
69
+ "grad_norm": 1.1506260633468628,
70
+ "learning_rate": 5.4705357142857145e-06,
71
+ "loss": 0.6687,
72
+ "step": 9
73
+ },
74
+ {
75
+ "epoch": 0.23,
76
+ "grad_norm": 2.9675679206848145,
77
+ "learning_rate": 5.4672619047619045e-06,
78
+ "loss": 0.7021,
79
+ "step": 10
80
+ },
81
+ {
82
+ "epoch": 0.26,
83
+ "grad_norm": 3.620849132537842,
84
+ "learning_rate": 5.463988095238095e-06,
85
+ "loss": 0.682,
86
+ "step": 11
87
+ },
88
+ {
89
+ "epoch": 0.28,
90
+ "grad_norm": 2.2752580642700195,
91
+ "learning_rate": 5.460714285714286e-06,
92
+ "loss": 0.6633,
93
+ "step": 12
94
+ },
95
+ {
96
+ "epoch": 0.3,
97
+ "grad_norm": 3.6715457439422607,
98
+ "learning_rate": 5.457440476190476e-06,
99
+ "loss": 0.6926,
100
+ "step": 13
101
+ },
102
+ {
103
+ "epoch": 0.33,
104
+ "grad_norm": 3.4275119304656982,
105
+ "learning_rate": 5.454166666666667e-06,
106
+ "loss": 0.6945,
107
+ "step": 14
108
+ },
109
+ {
110
+ "epoch": 0.35,
111
+ "grad_norm": 2.6921539306640625,
112
+ "learning_rate": 5.450892857142857e-06,
113
+ "loss": 0.6876,
114
+ "step": 15
115
+ },
116
+ {
117
+ "epoch": 0.38,
118
+ "grad_norm": 3.6308276653289795,
119
+ "learning_rate": 5.4476190476190476e-06,
120
+ "loss": 0.7249,
121
+ "step": 16
122
+ },
123
+ {
124
+ "epoch": 0.4,
125
+ "grad_norm": 1.0595824718475342,
126
+ "learning_rate": 5.444345238095238e-06,
127
+ "loss": 0.6804,
128
+ "step": 17
129
+ },
130
+ {
131
+ "epoch": 0.42,
132
+ "grad_norm": 2.645174026489258,
133
+ "learning_rate": 5.441071428571428e-06,
134
+ "loss": 0.6768,
135
+ "step": 18
136
+ },
137
+ {
138
+ "epoch": 0.45,
139
+ "grad_norm": 1.631472110748291,
140
+ "learning_rate": 5.437797619047619e-06,
141
+ "loss": 0.6812,
142
+ "step": 19
143
+ },
144
+ {
145
+ "epoch": 0.47,
146
+ "grad_norm": 3.3360867500305176,
147
+ "learning_rate": 5.43452380952381e-06,
148
+ "loss": 0.6817,
149
+ "step": 20
150
+ },
151
+ {
152
+ "epoch": 0.49,
153
+ "grad_norm": 1.531082034111023,
154
+ "learning_rate": 5.43125e-06,
155
+ "loss": 0.6624,
156
+ "step": 21
157
+ },
158
+ {
159
+ "epoch": 0.52,
160
+ "grad_norm": 2.1303884983062744,
161
+ "learning_rate": 5.427976190476191e-06,
162
+ "loss": 0.6926,
163
+ "step": 22
164
+ },
165
+ {
166
+ "epoch": 0.54,
167
+ "grad_norm": 2.1816420555114746,
168
+ "learning_rate": 5.4247023809523815e-06,
169
+ "loss": 0.7173,
170
+ "step": 23
171
+ },
172
+ {
173
+ "epoch": 0.56,
174
+ "grad_norm": 3.382896900177002,
175
+ "learning_rate": 5.4214285714285714e-06,
176
+ "loss": 0.7172,
177
+ "step": 24
178
+ },
179
+ {
180
+ "epoch": 0.59,
181
+ "grad_norm": 4.177649021148682,
182
+ "learning_rate": 5.418154761904762e-06,
183
+ "loss": 0.6932,
184
+ "step": 25
185
+ },
186
+ {
187
+ "epoch": 0.61,
188
+ "grad_norm": 6.432918548583984,
189
+ "learning_rate": 5.414880952380952e-06,
190
+ "loss": 0.6679,
191
+ "step": 26
192
+ },
193
+ {
194
+ "epoch": 0.63,
195
+ "grad_norm": 1.7472172975540161,
196
+ "learning_rate": 5.411607142857142e-06,
197
+ "loss": 0.6704,
198
+ "step": 27
199
+ },
200
+ {
201
+ "epoch": 0.66,
202
+ "grad_norm": 2.1172163486480713,
203
+ "learning_rate": 5.408333333333333e-06,
204
+ "loss": 0.6637,
205
+ "step": 28
206
+ },
207
+ {
208
+ "epoch": 0.68,
209
+ "grad_norm": 1.6855010986328125,
210
+ "learning_rate": 5.405059523809524e-06,
211
+ "loss": 0.656,
212
+ "step": 29
213
+ },
214
+ {
215
+ "epoch": 0.7,
216
+ "grad_norm": 2.4389238357543945,
217
+ "learning_rate": 5.401785714285714e-06,
218
+ "loss": 0.6562,
219
+ "step": 30
220
+ },
221
+ {
222
+ "epoch": 0.73,
223
+ "grad_norm": 3.0723414421081543,
224
+ "learning_rate": 5.3985119047619045e-06,
225
+ "loss": 0.6812,
226
+ "step": 31
227
+ },
228
+ {
229
+ "epoch": 0.75,
230
+ "grad_norm": 2.1830461025238037,
231
+ "learning_rate": 5.3952380952380944e-06,
232
+ "loss": 0.6448,
233
+ "step": 32
234
+ },
235
+ {
236
+ "epoch": 0.77,
237
+ "grad_norm": 3.109797716140747,
238
+ "learning_rate": 5.391964285714285e-06,
239
+ "loss": 0.6841,
240
+ "step": 33
241
+ },
242
+ {
243
+ "epoch": 0.8,
244
+ "grad_norm": 2.5417990684509277,
245
+ "learning_rate": 5.388690476190476e-06,
246
+ "loss": 0.6428,
247
+ "step": 34
248
+ },
249
+ {
250
+ "epoch": 0.82,
251
+ "grad_norm": 2.0981853008270264,
252
+ "learning_rate": 5.385416666666666e-06,
253
+ "loss": 0.6493,
254
+ "step": 35
255
+ },
256
+ {
257
+ "epoch": 0.84,
258
+ "grad_norm": 1.927240014076233,
259
+ "learning_rate": 5.382142857142857e-06,
260
+ "loss": 0.6496,
261
+ "step": 36
262
+ },
263
+ {
264
+ "epoch": 0.87,
265
+ "grad_norm": 4.362425327301025,
266
+ "learning_rate": 5.378869047619048e-06,
267
+ "loss": 0.6666,
268
+ "step": 37
269
+ },
270
+ {
271
+ "epoch": 0.89,
272
+ "grad_norm": 6.46484899520874,
273
+ "learning_rate": 5.3755952380952376e-06,
274
+ "loss": 0.7947,
275
+ "step": 38
276
+ },
277
+ {
278
+ "epoch": 0.91,
279
+ "grad_norm": 3.2664661407470703,
280
+ "learning_rate": 5.372321428571428e-06,
281
+ "loss": 0.608,
282
+ "step": 39
283
+ },
284
+ {
285
+ "epoch": 0.94,
286
+ "grad_norm": 2.2718191146850586,
287
+ "learning_rate": 5.369047619047619e-06,
288
+ "loss": 0.6247,
289
+ "step": 40
290
+ },
291
+ {
292
+ "epoch": 0.96,
293
+ "grad_norm": 5.372130870819092,
294
+ "learning_rate": 5.365773809523809e-06,
295
+ "loss": 0.7885,
296
+ "step": 41
297
+ },
298
+ {
299
+ "epoch": 0.98,
300
+ "grad_norm": 1.7550398111343384,
301
+ "learning_rate": 5.3625e-06,
302
+ "loss": 0.6425,
303
+ "step": 42
304
+ },
305
+ {
306
+ "epoch": 0.98,
307
+ "eval_accuracy": 0.6681818181818182,
308
+ "eval_loss": 0.6295949816703796,
309
+ "eval_runtime": 1.9777,
310
+ "eval_samples_per_second": 111.242,
311
+ "eval_steps_per_second": 27.81,
312
+ "step": 42
313
+ },
314
+ {
315
+ "epoch": 1.01,
316
+ "grad_norm": 3.6129467487335205,
317
+ "learning_rate": 5.35922619047619e-06,
318
+ "loss": 0.7519,
319
+ "step": 43
320
+ },
321
+ {
322
+ "epoch": 1.03,
323
+ "grad_norm": 4.161266326904297,
324
+ "learning_rate": 5.355952380952381e-06,
325
+ "loss": 0.6534,
326
+ "step": 44
327
+ },
328
+ {
329
+ "epoch": 1.05,
330
+ "grad_norm": 2.0768189430236816,
331
+ "learning_rate": 5.3526785714285715e-06,
332
+ "loss": 0.6805,
333
+ "step": 45
334
+ },
335
+ {
336
+ "epoch": 1.08,
337
+ "grad_norm": 2.847532272338867,
338
+ "learning_rate": 5.349404761904761e-06,
339
+ "loss": 0.6691,
340
+ "step": 46
341
+ },
342
+ {
343
+ "epoch": 1.1,
344
+ "grad_norm": 2.2965738773345947,
345
+ "learning_rate": 5.346130952380952e-06,
346
+ "loss": 0.6695,
347
+ "step": 47
348
+ },
349
+ {
350
+ "epoch": 1.12,
351
+ "grad_norm": 2.5308315753936768,
352
+ "learning_rate": 5.342857142857143e-06,
353
+ "loss": 0.6215,
354
+ "step": 48
355
+ },
356
+ {
357
+ "epoch": 1.15,
358
+ "grad_norm": 2.581547975540161,
359
+ "learning_rate": 5.339583333333333e-06,
360
+ "loss": 0.5902,
361
+ "step": 49
362
+ },
363
+ {
364
+ "epoch": 1.17,
365
+ "grad_norm": 3.935713768005371,
366
+ "learning_rate": 5.336309523809524e-06,
367
+ "loss": 0.5963,
368
+ "step": 50
369
+ },
370
+ {
371
+ "epoch": 1.2,
372
+ "grad_norm": 3.4248504638671875,
373
+ "learning_rate": 5.333035714285714e-06,
374
+ "loss": 0.6649,
375
+ "step": 51
376
+ },
377
+ {
378
+ "epoch": 1.22,
379
+ "grad_norm": 7.237908840179443,
380
+ "learning_rate": 5.3297619047619045e-06,
381
+ "loss": 0.7107,
382
+ "step": 52
383
+ },
384
+ {
385
+ "epoch": 1.24,
386
+ "grad_norm": 2.9264256954193115,
387
+ "learning_rate": 5.326488095238095e-06,
388
+ "loss": 0.6806,
389
+ "step": 53
390
+ },
391
+ {
392
+ "epoch": 1.27,
393
+ "grad_norm": 2.699674606323242,
394
+ "learning_rate": 5.323214285714285e-06,
395
+ "loss": 0.6276,
396
+ "step": 54
397
+ },
398
+ {
399
+ "epoch": 1.29,
400
+ "grad_norm": 3.520171642303467,
401
+ "learning_rate": 5.319940476190476e-06,
402
+ "loss": 0.6278,
403
+ "step": 55
404
+ },
405
+ {
406
+ "epoch": 1.31,
407
+ "grad_norm": 2.0494229793548584,
408
+ "learning_rate": 5.316666666666667e-06,
409
+ "loss": 0.5886,
410
+ "step": 56
411
+ },
412
+ {
413
+ "epoch": 1.34,
414
+ "grad_norm": 2.4809341430664062,
415
+ "learning_rate": 5.313392857142857e-06,
416
+ "loss": 0.6001,
417
+ "step": 57
418
+ },
419
+ {
420
+ "epoch": 1.36,
421
+ "grad_norm": 5.9523820877075195,
422
+ "learning_rate": 5.310119047619048e-06,
423
+ "loss": 0.6717,
424
+ "step": 58
425
+ },
426
+ {
427
+ "epoch": 1.38,
428
+ "grad_norm": 4.1546502113342285,
429
+ "learning_rate": 5.306845238095238e-06,
430
+ "loss": 0.6483,
431
+ "step": 59
432
+ },
433
+ {
434
+ "epoch": 1.41,
435
+ "grad_norm": 3.3143973350524902,
436
+ "learning_rate": 5.303571428571428e-06,
437
+ "loss": 0.5775,
438
+ "step": 60
439
+ },
440
+ {
441
+ "epoch": 1.43,
442
+ "grad_norm": 7.167575359344482,
443
+ "learning_rate": 5.300297619047619e-06,
444
+ "loss": 0.6182,
445
+ "step": 61
446
+ },
447
+ {
448
+ "epoch": 1.45,
449
+ "grad_norm": 1.888744592666626,
450
+ "learning_rate": 5.297023809523809e-06,
451
+ "loss": 0.5694,
452
+ "step": 62
453
+ },
454
+ {
455
+ "epoch": 1.48,
456
+ "grad_norm": 4.127984046936035,
457
+ "learning_rate": 5.29375e-06,
458
+ "loss": 0.5242,
459
+ "step": 63
460
+ },
461
+ {
462
+ "epoch": 1.5,
463
+ "grad_norm": 3.9158403873443604,
464
+ "learning_rate": 5.290476190476191e-06,
465
+ "loss": 0.5515,
466
+ "step": 64
467
+ },
468
+ {
469
+ "epoch": 1.52,
470
+ "grad_norm": 3.5147557258605957,
471
+ "learning_rate": 5.287202380952381e-06,
472
+ "loss": 0.5724,
473
+ "step": 65
474
+ },
475
+ {
476
+ "epoch": 1.55,
477
+ "grad_norm": 4.157176494598389,
478
+ "learning_rate": 5.2839285714285715e-06,
479
+ "loss": 0.6203,
480
+ "step": 66
481
+ },
482
+ {
483
+ "epoch": 1.57,
484
+ "grad_norm": 5.852452754974365,
485
+ "learning_rate": 5.280654761904762e-06,
486
+ "loss": 0.6141,
487
+ "step": 67
488
+ },
489
+ {
490
+ "epoch": 1.59,
491
+ "grad_norm": 4.920362949371338,
492
+ "learning_rate": 5.277380952380952e-06,
493
+ "loss": 0.6591,
494
+ "step": 68
495
+ },
496
+ {
497
+ "epoch": 1.62,
498
+ "grad_norm": 13.502717018127441,
499
+ "learning_rate": 5.274107142857143e-06,
500
+ "loss": 0.5012,
501
+ "step": 69
502
+ },
503
+ {
504
+ "epoch": 1.64,
505
+ "grad_norm": 3.996752977371216,
506
+ "learning_rate": 5.270833333333333e-06,
507
+ "loss": 0.5382,
508
+ "step": 70
509
+ },
510
+ {
511
+ "epoch": 1.66,
512
+ "grad_norm": 6.804556369781494,
513
+ "learning_rate": 5.267559523809524e-06,
514
+ "loss": 0.6235,
515
+ "step": 71
516
+ },
517
+ {
518
+ "epoch": 1.69,
519
+ "grad_norm": 3.19149112701416,
520
+ "learning_rate": 5.264285714285715e-06,
521
+ "loss": 0.6962,
522
+ "step": 72
523
+ },
524
+ {
525
+ "epoch": 1.71,
526
+ "grad_norm": 5.843012809753418,
527
+ "learning_rate": 5.2610119047619045e-06,
528
+ "loss": 0.619,
529
+ "step": 73
530
+ },
531
+ {
532
+ "epoch": 1.73,
533
+ "grad_norm": 8.353483200073242,
534
+ "learning_rate": 5.257738095238095e-06,
535
+ "loss": 0.5508,
536
+ "step": 74
537
+ },
538
+ {
539
+ "epoch": 1.76,
540
+ "grad_norm": 4.157430648803711,
541
+ "learning_rate": 5.254464285714286e-06,
542
+ "loss": 0.5476,
543
+ "step": 75
544
+ },
545
+ {
546
+ "epoch": 1.78,
547
+ "grad_norm": 7.451327800750732,
548
+ "learning_rate": 5.251190476190476e-06,
549
+ "loss": 0.563,
550
+ "step": 76
551
+ },
552
+ {
553
+ "epoch": 1.8,
554
+ "grad_norm": 2.8933181762695312,
555
+ "learning_rate": 5.247916666666667e-06,
556
+ "loss": 0.537,
557
+ "step": 77
558
+ },
559
+ {
560
+ "epoch": 1.83,
561
+ "grad_norm": 5.481462478637695,
562
+ "learning_rate": 5.244642857142858e-06,
563
+ "loss": 0.4794,
564
+ "step": 78
565
+ },
566
+ {
567
+ "epoch": 1.85,
568
+ "grad_norm": 3.3026862144470215,
569
+ "learning_rate": 5.241369047619047e-06,
570
+ "loss": 0.5493,
571
+ "step": 79
572
+ },
573
+ {
574
+ "epoch": 1.88,
575
+ "grad_norm": 4.270678520202637,
576
+ "learning_rate": 5.238095238095238e-06,
577
+ "loss": 0.5098,
578
+ "step": 80
579
+ },
580
+ {
581
+ "epoch": 1.9,
582
+ "grad_norm": 6.741820812225342,
583
+ "learning_rate": 5.234821428571428e-06,
584
+ "loss": 0.587,
585
+ "step": 81
586
+ },
587
+ {
588
+ "epoch": 1.92,
589
+ "grad_norm": 12.857108116149902,
590
+ "learning_rate": 5.231547619047618e-06,
591
+ "loss": 0.6178,
592
+ "step": 82
593
+ },
594
+ {
595
+ "epoch": 1.95,
596
+ "grad_norm": 14.573826789855957,
597
+ "learning_rate": 5.228273809523809e-06,
598
+ "loss": 0.5698,
599
+ "step": 83
600
+ },
601
+ {
602
+ "epoch": 1.97,
603
+ "grad_norm": 16.281530380249023,
604
+ "learning_rate": 5.225e-06,
605
+ "loss": 0.5929,
606
+ "step": 84
607
+ },
608
+ {
609
+ "epoch": 1.99,
610
+ "grad_norm": 8.965470314025879,
611
+ "learning_rate": 5.22172619047619e-06,
612
+ "loss": 0.583,
613
+ "step": 85
614
+ },
615
+ {
616
+ "epoch": 1.99,
617
+ "eval_accuracy": 0.7,
618
+ "eval_loss": 0.5743765830993652,
619
+ "eval_runtime": 1.9707,
620
+ "eval_samples_per_second": 111.636,
621
+ "eval_steps_per_second": 27.909,
622
+ "step": 85
623
+ },
624
+ {
625
+ "epoch": 2.02,
626
+ "grad_norm": 7.075886249542236,
627
+ "learning_rate": 5.218452380952381e-06,
628
+ "loss": 0.5969,
629
+ "step": 86
630
+ },
631
+ {
632
+ "epoch": 2.04,
633
+ "grad_norm": 5.647829055786133,
634
+ "learning_rate": 5.2151785714285715e-06,
635
+ "loss": 0.5581,
636
+ "step": 87
637
+ },
638
+ {
639
+ "epoch": 2.06,
640
+ "grad_norm": 12.890501976013184,
641
+ "learning_rate": 5.2119047619047615e-06,
642
+ "loss": 0.5742,
643
+ "step": 88
644
+ },
645
+ {
646
+ "epoch": 2.09,
647
+ "grad_norm": 6.206398963928223,
648
+ "learning_rate": 5.208630952380952e-06,
649
+ "loss": 0.4763,
650
+ "step": 89
651
+ },
652
+ {
653
+ "epoch": 2.11,
654
+ "grad_norm": 15.82968521118164,
655
+ "learning_rate": 5.205357142857142e-06,
656
+ "loss": 0.5015,
657
+ "step": 90
658
+ },
659
+ {
660
+ "epoch": 2.13,
661
+ "grad_norm": 2.781341314315796,
662
+ "learning_rate": 5.202083333333333e-06,
663
+ "loss": 0.6048,
664
+ "step": 91
665
+ },
666
+ {
667
+ "epoch": 2.16,
668
+ "grad_norm": 31.130008697509766,
669
+ "learning_rate": 5.198809523809524e-06,
670
+ "loss": 0.6021,
671
+ "step": 92
672
+ },
673
+ {
674
+ "epoch": 2.18,
675
+ "grad_norm": 41.88652038574219,
676
+ "learning_rate": 5.195535714285714e-06,
677
+ "loss": 0.5294,
678
+ "step": 93
679
+ },
680
+ {
681
+ "epoch": 2.2,
682
+ "grad_norm": 13.205155372619629,
683
+ "learning_rate": 5.1922619047619046e-06,
684
+ "loss": 0.5571,
685
+ "step": 94
686
+ },
687
+ {
688
+ "epoch": 2.23,
689
+ "grad_norm": 4.5637054443359375,
690
+ "learning_rate": 5.188988095238095e-06,
691
+ "loss": 0.415,
692
+ "step": 95
693
+ },
694
+ {
695
+ "epoch": 2.25,
696
+ "grad_norm": 8.613974571228027,
697
+ "learning_rate": 5.185714285714285e-06,
698
+ "loss": 0.665,
699
+ "step": 96
700
+ },
701
+ {
702
+ "epoch": 2.27,
703
+ "grad_norm": 16.234405517578125,
704
+ "learning_rate": 5.182440476190476e-06,
705
+ "loss": 0.5181,
706
+ "step": 97
707
+ },
708
+ {
709
+ "epoch": 2.3,
710
+ "grad_norm": 6.523887634277344,
711
+ "learning_rate": 5.179166666666666e-06,
712
+ "loss": 0.4848,
713
+ "step": 98
714
+ },
715
+ {
716
+ "epoch": 2.32,
717
+ "grad_norm": 9.602444648742676,
718
+ "learning_rate": 5.175892857142857e-06,
719
+ "loss": 0.4226,
720
+ "step": 99
721
+ },
722
+ {
723
+ "epoch": 2.34,
724
+ "grad_norm": 5.141808032989502,
725
+ "learning_rate": 5.172619047619048e-06,
726
+ "loss": 0.4066,
727
+ "step": 100
728
+ },
729
+ {
730
+ "epoch": 2.37,
731
+ "grad_norm": 14.732641220092773,
732
+ "learning_rate": 5.169345238095238e-06,
733
+ "loss": 0.4309,
734
+ "step": 101
735
+ },
736
+ {
737
+ "epoch": 2.39,
738
+ "grad_norm": 13.681814193725586,
739
+ "learning_rate": 5.166071428571428e-06,
740
+ "loss": 0.5336,
741
+ "step": 102
742
+ },
743
+ {
744
+ "epoch": 2.41,
745
+ "grad_norm": 9.389954566955566,
746
+ "learning_rate": 5.162797619047619e-06,
747
+ "loss": 0.4979,
748
+ "step": 103
749
+ },
750
+ {
751
+ "epoch": 2.44,
752
+ "grad_norm": 3.370234727859497,
753
+ "learning_rate": 5.159523809523809e-06,
754
+ "loss": 0.5064,
755
+ "step": 104
756
+ },
757
+ {
758
+ "epoch": 2.46,
759
+ "grad_norm": 6.832080841064453,
760
+ "learning_rate": 5.15625e-06,
761
+ "loss": 0.5425,
762
+ "step": 105
763
+ },
764
+ {
765
+ "epoch": 2.48,
766
+ "grad_norm": 5.864181041717529,
767
+ "learning_rate": 5.152976190476191e-06,
768
+ "loss": 0.4536,
769
+ "step": 106
770
+ },
771
+ {
772
+ "epoch": 2.51,
773
+ "grad_norm": 4.189244747161865,
774
+ "learning_rate": 5.149702380952381e-06,
775
+ "loss": 0.5783,
776
+ "step": 107
777
+ },
778
+ {
779
+ "epoch": 2.53,
780
+ "grad_norm": 13.515569686889648,
781
+ "learning_rate": 5.1464285714285715e-06,
782
+ "loss": 0.6054,
783
+ "step": 108
784
+ },
785
+ {
786
+ "epoch": 2.55,
787
+ "grad_norm": 8.88909912109375,
788
+ "learning_rate": 5.1431547619047615e-06,
789
+ "loss": 0.4679,
790
+ "step": 109
791
+ },
792
+ {
793
+ "epoch": 2.58,
794
+ "grad_norm": 12.870144844055176,
795
+ "learning_rate": 5.139880952380952e-06,
796
+ "loss": 0.4515,
797
+ "step": 110
798
+ },
799
+ {
800
+ "epoch": 2.6,
801
+ "grad_norm": 7.6494927406311035,
802
+ "learning_rate": 5.136607142857143e-06,
803
+ "loss": 0.4118,
804
+ "step": 111
805
+ },
806
+ {
807
+ "epoch": 2.62,
808
+ "grad_norm": 5.379680633544922,
809
+ "learning_rate": 5.133333333333333e-06,
810
+ "loss": 0.4404,
811
+ "step": 112
812
+ },
813
+ {
814
+ "epoch": 2.65,
815
+ "grad_norm": 12.46003246307373,
816
+ "learning_rate": 5.130059523809524e-06,
817
+ "loss": 0.5413,
818
+ "step": 113
819
+ },
820
+ {
821
+ "epoch": 2.67,
822
+ "grad_norm": 16.497268676757812,
823
+ "learning_rate": 5.126785714285715e-06,
824
+ "loss": 0.4424,
825
+ "step": 114
826
+ },
827
+ {
828
+ "epoch": 2.7,
829
+ "grad_norm": 27.879886627197266,
830
+ "learning_rate": 5.123511904761905e-06,
831
+ "loss": 0.6332,
832
+ "step": 115
833
+ },
834
+ {
835
+ "epoch": 2.72,
836
+ "grad_norm": 24.27963638305664,
837
+ "learning_rate": 5.120238095238095e-06,
838
+ "loss": 0.6507,
839
+ "step": 116
840
+ },
841
+ {
842
+ "epoch": 2.74,
843
+ "grad_norm": 9.079182624816895,
844
+ "learning_rate": 5.116964285714285e-06,
845
+ "loss": 0.4821,
846
+ "step": 117
847
+ },
848
+ {
849
+ "epoch": 2.77,
850
+ "grad_norm": 8.210986137390137,
851
+ "learning_rate": 5.113690476190476e-06,
852
+ "loss": 0.4236,
853
+ "step": 118
854
+ },
855
+ {
856
+ "epoch": 2.79,
857
+ "grad_norm": 9.538209915161133,
858
+ "learning_rate": 5.110416666666667e-06,
859
+ "loss": 0.5458,
860
+ "step": 119
861
+ },
862
+ {
863
+ "epoch": 2.81,
864
+ "grad_norm": 11.242250442504883,
865
+ "learning_rate": 5.107142857142857e-06,
866
+ "loss": 0.3717,
867
+ "step": 120
868
+ },
869
+ {
870
+ "epoch": 2.84,
871
+ "grad_norm": 7.99606466293335,
872
+ "learning_rate": 5.103869047619048e-06,
873
+ "loss": 0.463,
874
+ "step": 121
875
+ },
876
+ {
877
+ "epoch": 2.86,
878
+ "grad_norm": 8.157175064086914,
879
+ "learning_rate": 5.1005952380952385e-06,
880
+ "loss": 0.4463,
881
+ "step": 122
882
+ },
883
+ {
884
+ "epoch": 2.88,
885
+ "grad_norm": 6.33786678314209,
886
+ "learning_rate": 5.0973214285714284e-06,
887
+ "loss": 0.4483,
888
+ "step": 123
889
+ },
890
+ {
891
+ "epoch": 2.91,
892
+ "grad_norm": 12.092632293701172,
893
+ "learning_rate": 5.094047619047619e-06,
894
+ "loss": 0.4729,
895
+ "step": 124
896
+ },
897
+ {
898
+ "epoch": 2.93,
899
+ "grad_norm": 6.666680335998535,
900
+ "learning_rate": 5.09077380952381e-06,
901
+ "loss": 0.4628,
902
+ "step": 125
903
+ },
904
+ {
905
+ "epoch": 2.95,
906
+ "grad_norm": 13.772384643554688,
907
+ "learning_rate": 5.0875e-06,
908
+ "loss": 0.2814,
909
+ "step": 126
910
+ },
911
+ {
912
+ "epoch": 2.98,
913
+ "grad_norm": 28.22318458557129,
914
+ "learning_rate": 5.084226190476191e-06,
915
+ "loss": 0.6727,
916
+ "step": 127
917
+ },
918
+ {
919
+ "epoch": 3.0,
920
+ "grad_norm": 25.690156936645508,
921
+ "learning_rate": 5.080952380952381e-06,
922
+ "loss": 0.4848,
923
+ "step": 128
924
+ },
925
+ {
926
+ "epoch": 3.0,
927
+ "eval_accuracy": 0.7227272727272728,
928
+ "eval_loss": 0.6313565373420715,
929
+ "eval_runtime": 1.9701,
930
+ "eval_samples_per_second": 111.669,
931
+ "eval_steps_per_second": 27.917,
932
+ "step": 128
933
+ },
934
+ {
935
+ "epoch": 3.02,
936
+ "grad_norm": 20.357358932495117,
937
+ "learning_rate": 5.0776785714285715e-06,
938
+ "loss": 0.6101,
939
+ "step": 129
940
+ },
941
+ {
942
+ "epoch": 3.05,
943
+ "grad_norm": 24.226192474365234,
944
+ "learning_rate": 5.074404761904762e-06,
945
+ "loss": 0.4901,
946
+ "step": 130
947
+ },
948
+ {
949
+ "epoch": 3.07,
950
+ "grad_norm": 21.727924346923828,
951
+ "learning_rate": 5.071130952380952e-06,
952
+ "loss": 0.3983,
953
+ "step": 131
954
+ },
955
+ {
956
+ "epoch": 3.09,
957
+ "grad_norm": 8.311666488647461,
958
+ "learning_rate": 5.067857142857142e-06,
959
+ "loss": 0.4098,
960
+ "step": 132
961
+ },
962
+ {
963
+ "epoch": 3.12,
964
+ "grad_norm": 8.897052764892578,
965
+ "learning_rate": 5.064583333333333e-06,
966
+ "loss": 0.4665,
967
+ "step": 133
968
+ },
969
+ {
970
+ "epoch": 3.14,
971
+ "grad_norm": 24.099576950073242,
972
+ "learning_rate": 5.061309523809523e-06,
973
+ "loss": 0.4198,
974
+ "step": 134
975
+ },
976
+ {
977
+ "epoch": 3.16,
978
+ "grad_norm": 10.50734806060791,
979
+ "learning_rate": 5.058035714285714e-06,
980
+ "loss": 0.3888,
981
+ "step": 135
982
+ },
983
+ {
984
+ "epoch": 3.19,
985
+ "grad_norm": 22.82213592529297,
986
+ "learning_rate": 5.054761904761905e-06,
987
+ "loss": 0.4525,
988
+ "step": 136
989
+ },
990
+ {
991
+ "epoch": 3.21,
992
+ "grad_norm": 37.29859161376953,
993
+ "learning_rate": 5.0514880952380946e-06,
994
+ "loss": 0.486,
995
+ "step": 137
996
+ },
997
+ {
998
+ "epoch": 3.23,
999
+ "grad_norm": 14.720842361450195,
1000
+ "learning_rate": 5.048214285714285e-06,
1001
+ "loss": 0.4094,
1002
+ "step": 138
1003
+ },
1004
+ {
1005
+ "epoch": 3.26,
1006
+ "grad_norm": 13.443496704101562,
1007
+ "learning_rate": 5.044940476190476e-06,
1008
+ "loss": 0.3444,
1009
+ "step": 139
1010
+ },
1011
+ {
1012
+ "epoch": 3.28,
1013
+ "grad_norm": 10.844743728637695,
1014
+ "learning_rate": 5.041666666666666e-06,
1015
+ "loss": 0.3291,
1016
+ "step": 140
1017
+ },
1018
+ {
1019
+ "epoch": 3.3,
1020
+ "grad_norm": 10.810824394226074,
1021
+ "learning_rate": 5.038392857142857e-06,
1022
+ "loss": 0.4249,
1023
+ "step": 141
1024
+ },
1025
+ {
1026
+ "epoch": 3.33,
1027
+ "grad_norm": 5.066943168640137,
1028
+ "learning_rate": 5.035119047619048e-06,
1029
+ "loss": 0.3591,
1030
+ "step": 142
1031
+ },
1032
+ {
1033
+ "epoch": 3.35,
1034
+ "grad_norm": 75.52027130126953,
1035
+ "learning_rate": 5.031845238095238e-06,
1036
+ "loss": 0.4469,
1037
+ "step": 143
1038
+ },
1039
+ {
1040
+ "epoch": 3.38,
1041
+ "grad_norm": 41.847652435302734,
1042
+ "learning_rate": 5.0285714285714285e-06,
1043
+ "loss": 0.4158,
1044
+ "step": 144
1045
+ },
1046
+ {
1047
+ "epoch": 3.4,
1048
+ "grad_norm": 17.77975082397461,
1049
+ "learning_rate": 5.025297619047618e-06,
1050
+ "loss": 0.2459,
1051
+ "step": 145
1052
+ },
1053
+ {
1054
+ "epoch": 3.42,
1055
+ "grad_norm": 41.41980743408203,
1056
+ "learning_rate": 5.022023809523809e-06,
1057
+ "loss": 0.5155,
1058
+ "step": 146
1059
+ },
1060
+ {
1061
+ "epoch": 3.45,
1062
+ "grad_norm": 42.26784896850586,
1063
+ "learning_rate": 5.01875e-06,
1064
+ "loss": 0.3646,
1065
+ "step": 147
1066
+ },
1067
+ {
1068
+ "epoch": 3.47,
1069
+ "grad_norm": 15.376642227172852,
1070
+ "learning_rate": 5.01547619047619e-06,
1071
+ "loss": 0.411,
1072
+ "step": 148
1073
+ },
1074
+ {
1075
+ "epoch": 3.49,
1076
+ "grad_norm": 17.662681579589844,
1077
+ "learning_rate": 5.012202380952381e-06,
1078
+ "loss": 0.3776,
1079
+ "step": 149
1080
+ },
1081
+ {
1082
+ "epoch": 3.52,
1083
+ "grad_norm": 12.808351516723633,
1084
+ "learning_rate": 5.0089285714285716e-06,
1085
+ "loss": 0.3035,
1086
+ "step": 150
1087
+ },
1088
+ {
1089
+ "epoch": 3.54,
1090
+ "grad_norm": 30.4917049407959,
1091
+ "learning_rate": 5.0056547619047615e-06,
1092
+ "loss": 0.5275,
1093
+ "step": 151
1094
+ },
1095
+ {
1096
+ "epoch": 3.56,
1097
+ "grad_norm": 9.420913696289062,
1098
+ "learning_rate": 5.002380952380952e-06,
1099
+ "loss": 0.4193,
1100
+ "step": 152
1101
+ },
1102
+ {
1103
+ "epoch": 3.59,
1104
+ "grad_norm": 20.13055419921875,
1105
+ "learning_rate": 4.999107142857142e-06,
1106
+ "loss": 0.5197,
1107
+ "step": 153
1108
+ },
1109
+ {
1110
+ "epoch": 3.61,
1111
+ "grad_norm": 3.9424262046813965,
1112
+ "learning_rate": 4.995833333333333e-06,
1113
+ "loss": 0.3119,
1114
+ "step": 154
1115
+ },
1116
+ {
1117
+ "epoch": 3.63,
1118
+ "grad_norm": 4.024766445159912,
1119
+ "learning_rate": 4.992559523809524e-06,
1120
+ "loss": 0.3072,
1121
+ "step": 155
1122
+ },
1123
+ {
1124
+ "epoch": 3.66,
1125
+ "grad_norm": 16.67523956298828,
1126
+ "learning_rate": 4.989285714285714e-06,
1127
+ "loss": 0.4243,
1128
+ "step": 156
1129
+ },
1130
+ {
1131
+ "epoch": 3.68,
1132
+ "grad_norm": 14.94045352935791,
1133
+ "learning_rate": 4.986011904761905e-06,
1134
+ "loss": 0.4539,
1135
+ "step": 157
1136
+ },
1137
+ {
1138
+ "epoch": 3.7,
1139
+ "grad_norm": 14.900248527526855,
1140
+ "learning_rate": 4.9827380952380954e-06,
1141
+ "loss": 0.3834,
1142
+ "step": 158
1143
+ },
1144
+ {
1145
+ "epoch": 3.73,
1146
+ "grad_norm": 59.21234893798828,
1147
+ "learning_rate": 4.979464285714285e-06,
1148
+ "loss": 0.6183,
1149
+ "step": 159
1150
+ },
1151
+ {
1152
+ "epoch": 3.75,
1153
+ "grad_norm": 9.785917282104492,
1154
+ "learning_rate": 4.976190476190476e-06,
1155
+ "loss": 0.6158,
1156
+ "step": 160
1157
+ },
1158
+ {
1159
+ "epoch": 3.77,
1160
+ "grad_norm": 47.627864837646484,
1161
+ "learning_rate": 4.972916666666667e-06,
1162
+ "loss": 0.7627,
1163
+ "step": 161
1164
+ },
1165
+ {
1166
+ "epoch": 3.8,
1167
+ "grad_norm": 33.516265869140625,
1168
+ "learning_rate": 4.969642857142857e-06,
1169
+ "loss": 0.6755,
1170
+ "step": 162
1171
+ },
1172
+ {
1173
+ "epoch": 3.82,
1174
+ "grad_norm": 31.58948516845703,
1175
+ "learning_rate": 4.966369047619048e-06,
1176
+ "loss": 0.3291,
1177
+ "step": 163
1178
+ },
1179
+ {
1180
+ "epoch": 3.84,
1181
+ "grad_norm": 6.718880653381348,
1182
+ "learning_rate": 4.963095238095238e-06,
1183
+ "loss": 0.3233,
1184
+ "step": 164
1185
+ },
1186
+ {
1187
+ "epoch": 3.87,
1188
+ "grad_norm": 17.761388778686523,
1189
+ "learning_rate": 4.9598214285714285e-06,
1190
+ "loss": 0.332,
1191
+ "step": 165
1192
+ },
1193
+ {
1194
+ "epoch": 3.89,
1195
+ "grad_norm": 24.450733184814453,
1196
+ "learning_rate": 4.956547619047619e-06,
1197
+ "loss": 0.4093,
1198
+ "step": 166
1199
+ },
1200
+ {
1201
+ "epoch": 3.91,
1202
+ "grad_norm": 5.664928913116455,
1203
+ "learning_rate": 4.953273809523809e-06,
1204
+ "loss": 0.458,
1205
+ "step": 167
1206
+ },
1207
+ {
1208
+ "epoch": 3.94,
1209
+ "grad_norm": 8.065629005432129,
1210
+ "learning_rate": 4.95e-06,
1211
+ "loss": 0.503,
1212
+ "step": 168
1213
+ },
1214
+ {
1215
+ "epoch": 3.96,
1216
+ "grad_norm": 8.75709056854248,
1217
+ "learning_rate": 4.946726190476191e-06,
1218
+ "loss": 0.4183,
1219
+ "step": 169
1220
+ },
1221
+ {
1222
+ "epoch": 3.98,
1223
+ "grad_norm": 29.46023941040039,
1224
+ "learning_rate": 4.943452380952381e-06,
1225
+ "loss": 0.4187,
1226
+ "step": 170
1227
+ },
1228
+ {
1229
+ "epoch": 3.98,
1230
+ "eval_accuracy": 0.7727272727272727,
1231
+ "eval_loss": 0.48904669284820557,
1232
+ "eval_runtime": 1.961,
1233
+ "eval_samples_per_second": 112.185,
1234
+ "eval_steps_per_second": 28.046,
1235
+ "step": 170
1236
+ },
1237
+ {
1238
+ "epoch": 4.01,
1239
+ "grad_norm": 2.7368314266204834,
1240
+ "learning_rate": 4.940178571428572e-06,
1241
+ "loss": 0.4246,
1242
+ "step": 171
1243
+ },
1244
+ {
1245
+ "epoch": 4.03,
1246
+ "grad_norm": 4.692036151885986,
1247
+ "learning_rate": 4.9369047619047615e-06,
1248
+ "loss": 0.303,
1249
+ "step": 172
1250
+ },
1251
+ {
1252
+ "epoch": 4.05,
1253
+ "grad_norm": 21.547483444213867,
1254
+ "learning_rate": 4.933630952380952e-06,
1255
+ "loss": 0.3867,
1256
+ "step": 173
1257
+ },
1258
+ {
1259
+ "epoch": 4.08,
1260
+ "grad_norm": 8.194944381713867,
1261
+ "learning_rate": 4.930357142857143e-06,
1262
+ "loss": 0.1695,
1263
+ "step": 174
1264
+ },
1265
+ {
1266
+ "epoch": 4.1,
1267
+ "grad_norm": 3.5077579021453857,
1268
+ "learning_rate": 4.927083333333333e-06,
1269
+ "loss": 0.3142,
1270
+ "step": 175
1271
+ },
1272
+ {
1273
+ "epoch": 4.12,
1274
+ "grad_norm": 33.11520004272461,
1275
+ "learning_rate": 4.923809523809524e-06,
1276
+ "loss": 0.2809,
1277
+ "step": 176
1278
+ },
1279
+ {
1280
+ "epoch": 4.15,
1281
+ "grad_norm": 22.623079299926758,
1282
+ "learning_rate": 4.920535714285715e-06,
1283
+ "loss": 0.5169,
1284
+ "step": 177
1285
+ },
1286
+ {
1287
+ "epoch": 4.17,
1288
+ "grad_norm": 50.30414581298828,
1289
+ "learning_rate": 4.917261904761905e-06,
1290
+ "loss": 0.5075,
1291
+ "step": 178
1292
+ },
1293
+ {
1294
+ "epoch": 4.2,
1295
+ "grad_norm": 22.926681518554688,
1296
+ "learning_rate": 4.9139880952380955e-06,
1297
+ "loss": 0.3153,
1298
+ "step": 179
1299
+ },
1300
+ {
1301
+ "epoch": 4.22,
1302
+ "grad_norm": 4.504119396209717,
1303
+ "learning_rate": 4.910714285714286e-06,
1304
+ "loss": 0.3218,
1305
+ "step": 180
1306
+ },
1307
+ {
1308
+ "epoch": 4.24,
1309
+ "grad_norm": 26.036670684814453,
1310
+ "learning_rate": 4.907440476190476e-06,
1311
+ "loss": 0.4977,
1312
+ "step": 181
1313
+ },
1314
+ {
1315
+ "epoch": 4.27,
1316
+ "grad_norm": 3.1559319496154785,
1317
+ "learning_rate": 4.904166666666667e-06,
1318
+ "loss": 0.2435,
1319
+ "step": 182
1320
+ },
1321
+ {
1322
+ "epoch": 4.29,
1323
+ "grad_norm": 10.12259292602539,
1324
+ "learning_rate": 4.900892857142857e-06,
1325
+ "loss": 0.2531,
1326
+ "step": 183
1327
+ },
1328
+ {
1329
+ "epoch": 4.31,
1330
+ "grad_norm": 14.153977394104004,
1331
+ "learning_rate": 4.897619047619047e-06,
1332
+ "loss": 0.2939,
1333
+ "step": 184
1334
+ },
1335
+ {
1336
+ "epoch": 4.34,
1337
+ "grad_norm": 27.15022850036621,
1338
+ "learning_rate": 4.894345238095238e-06,
1339
+ "loss": 0.3503,
1340
+ "step": 185
1341
+ },
1342
+ {
1343
+ "epoch": 4.36,
1344
+ "grad_norm": 14.092424392700195,
1345
+ "learning_rate": 4.8910714285714285e-06,
1346
+ "loss": 0.2999,
1347
+ "step": 186
1348
+ },
1349
+ {
1350
+ "epoch": 4.38,
1351
+ "grad_norm": 21.638959884643555,
1352
+ "learning_rate": 4.8877976190476185e-06,
1353
+ "loss": 0.2351,
1354
+ "step": 187
1355
+ },
1356
+ {
1357
+ "epoch": 4.41,
1358
+ "grad_norm": 10.01887321472168,
1359
+ "learning_rate": 4.884523809523809e-06,
1360
+ "loss": 0.313,
1361
+ "step": 188
1362
+ },
1363
+ {
1364
+ "epoch": 4.43,
1365
+ "grad_norm": 13.962733268737793,
1366
+ "learning_rate": 4.881249999999999e-06,
1367
+ "loss": 0.3914,
1368
+ "step": 189
1369
+ },
1370
+ {
1371
+ "epoch": 4.45,
1372
+ "grad_norm": 12.105456352233887,
1373
+ "learning_rate": 4.87797619047619e-06,
1374
+ "loss": 0.3625,
1375
+ "step": 190
1376
+ },
1377
+ {
1378
+ "epoch": 4.48,
1379
+ "grad_norm": 11.165689468383789,
1380
+ "learning_rate": 4.874702380952381e-06,
1381
+ "loss": 0.425,
1382
+ "step": 191
1383
+ },
1384
+ {
1385
+ "epoch": 4.5,
1386
+ "grad_norm": 13.687613487243652,
1387
+ "learning_rate": 4.871428571428571e-06,
1388
+ "loss": 0.3409,
1389
+ "step": 192
1390
+ },
1391
+ {
1392
+ "epoch": 4.52,
1393
+ "grad_norm": 16.423450469970703,
1394
+ "learning_rate": 4.8681547619047616e-06,
1395
+ "loss": 0.4593,
1396
+ "step": 193
1397
+ },
1398
+ {
1399
+ "epoch": 4.55,
1400
+ "grad_norm": 10.233711242675781,
1401
+ "learning_rate": 4.864880952380952e-06,
1402
+ "loss": 0.2118,
1403
+ "step": 194
1404
+ },
1405
+ {
1406
+ "epoch": 4.57,
1407
+ "grad_norm": 9.759559631347656,
1408
+ "learning_rate": 4.861607142857142e-06,
1409
+ "loss": 0.3881,
1410
+ "step": 195
1411
+ },
1412
+ {
1413
+ "epoch": 4.59,
1414
+ "grad_norm": 16.481748580932617,
1415
+ "learning_rate": 4.858333333333333e-06,
1416
+ "loss": 0.3053,
1417
+ "step": 196
1418
+ },
1419
+ {
1420
+ "epoch": 4.62,
1421
+ "grad_norm": 17.1108455657959,
1422
+ "learning_rate": 4.855059523809524e-06,
1423
+ "loss": 0.4328,
1424
+ "step": 197
1425
+ },
1426
+ {
1427
+ "epoch": 4.64,
1428
+ "grad_norm": 18.17218589782715,
1429
+ "learning_rate": 4.851785714285714e-06,
1430
+ "loss": 0.3355,
1431
+ "step": 198
1432
+ },
1433
+ {
1434
+ "epoch": 4.66,
1435
+ "grad_norm": 10.150218963623047,
1436
+ "learning_rate": 4.848511904761905e-06,
1437
+ "loss": 0.244,
1438
+ "step": 199
1439
+ },
1440
+ {
1441
+ "epoch": 4.69,
1442
+ "grad_norm": 8.062572479248047,
1443
+ "learning_rate": 4.845238095238095e-06,
1444
+ "loss": 0.2057,
1445
+ "step": 200
1446
+ },
1447
+ {
1448
+ "epoch": 4.71,
1449
+ "grad_norm": 11.194631576538086,
1450
+ "learning_rate": 4.8419642857142854e-06,
1451
+ "loss": 0.2364,
1452
+ "step": 201
1453
+ },
1454
+ {
1455
+ "epoch": 4.73,
1456
+ "grad_norm": 14.064760208129883,
1457
+ "learning_rate": 4.838690476190476e-06,
1458
+ "loss": 0.2463,
1459
+ "step": 202
1460
+ },
1461
+ {
1462
+ "epoch": 4.76,
1463
+ "grad_norm": 18.526512145996094,
1464
+ "learning_rate": 4.835416666666666e-06,
1465
+ "loss": 0.2456,
1466
+ "step": 203
1467
+ },
1468
+ {
1469
+ "epoch": 4.78,
1470
+ "grad_norm": 17.577804565429688,
1471
+ "learning_rate": 4.832142857142857e-06,
1472
+ "loss": 0.2601,
1473
+ "step": 204
1474
+ },
1475
+ {
1476
+ "epoch": 4.8,
1477
+ "grad_norm": 11.843798637390137,
1478
+ "learning_rate": 4.828869047619048e-06,
1479
+ "loss": 0.1637,
1480
+ "step": 205
1481
+ },
1482
+ {
1483
+ "epoch": 4.83,
1484
+ "grad_norm": 12.153199195861816,
1485
+ "learning_rate": 4.825595238095238e-06,
1486
+ "loss": 0.1723,
1487
+ "step": 206
1488
+ },
1489
+ {
1490
+ "epoch": 4.85,
1491
+ "grad_norm": 10.96668529510498,
1492
+ "learning_rate": 4.8223214285714285e-06,
1493
+ "loss": 0.3315,
1494
+ "step": 207
1495
+ },
1496
+ {
1497
+ "epoch": 4.88,
1498
+ "grad_norm": 8.555688858032227,
1499
+ "learning_rate": 4.8190476190476185e-06,
1500
+ "loss": 0.2321,
1501
+ "step": 208
1502
+ },
1503
+ {
1504
+ "epoch": 4.9,
1505
+ "grad_norm": 12.00627326965332,
1506
+ "learning_rate": 4.815773809523809e-06,
1507
+ "loss": 0.2672,
1508
+ "step": 209
1509
+ },
1510
+ {
1511
+ "epoch": 4.92,
1512
+ "grad_norm": 22.342628479003906,
1513
+ "learning_rate": 4.8125e-06,
1514
+ "loss": 0.4607,
1515
+ "step": 210
1516
+ },
1517
+ {
1518
+ "epoch": 4.95,
1519
+ "grad_norm": 25.33961296081543,
1520
+ "learning_rate": 4.80922619047619e-06,
1521
+ "loss": 0.2068,
1522
+ "step": 211
1523
+ },
1524
+ {
1525
+ "epoch": 4.97,
1526
+ "grad_norm": 22.976959228515625,
1527
+ "learning_rate": 4.805952380952381e-06,
1528
+ "loss": 0.3447,
1529
+ "step": 212
1530
+ },
1531
+ {
1532
+ "epoch": 4.99,
1533
+ "grad_norm": 19.871631622314453,
1534
+ "learning_rate": 4.802678571428572e-06,
1535
+ "loss": 0.1563,
1536
+ "step": 213
1537
+ },
1538
+ {
1539
+ "epoch": 4.99,
1540
+ "eval_accuracy": 0.8545454545454545,
1541
+ "eval_loss": 0.4087750315666199,
1542
+ "eval_runtime": 1.9518,
1543
+ "eval_samples_per_second": 112.716,
1544
+ "eval_steps_per_second": 28.179,
1545
+ "step": 213
1546
+ },
1547
+ {
1548
+ "epoch": 5.02,
1549
+ "grad_norm": 6.280614852905273,
1550
+ "learning_rate": 4.799404761904762e-06,
1551
+ "loss": 0.1367,
1552
+ "step": 214
1553
+ },
1554
+ {
1555
+ "epoch": 5.04,
1556
+ "grad_norm": 18.553333282470703,
1557
+ "learning_rate": 4.796130952380952e-06,
1558
+ "loss": 0.3876,
1559
+ "step": 215
1560
+ },
1561
+ {
1562
+ "epoch": 5.06,
1563
+ "grad_norm": 10.598409652709961,
1564
+ "learning_rate": 4.792857142857143e-06,
1565
+ "loss": 0.2598,
1566
+ "step": 216
1567
+ },
1568
+ {
1569
+ "epoch": 5.09,
1570
+ "grad_norm": 28.292373657226562,
1571
+ "learning_rate": 4.789583333333333e-06,
1572
+ "loss": 0.1943,
1573
+ "step": 217
1574
+ },
1575
+ {
1576
+ "epoch": 5.11,
1577
+ "grad_norm": 29.251375198364258,
1578
+ "learning_rate": 4.786309523809524e-06,
1579
+ "loss": 0.3162,
1580
+ "step": 218
1581
+ },
1582
+ {
1583
+ "epoch": 5.13,
1584
+ "grad_norm": 54.16463088989258,
1585
+ "learning_rate": 4.783035714285714e-06,
1586
+ "loss": 0.3793,
1587
+ "step": 219
1588
+ },
1589
+ {
1590
+ "epoch": 5.16,
1591
+ "grad_norm": 13.832413673400879,
1592
+ "learning_rate": 4.779761904761905e-06,
1593
+ "loss": 0.134,
1594
+ "step": 220
1595
+ },
1596
+ {
1597
+ "epoch": 5.18,
1598
+ "grad_norm": 17.206539154052734,
1599
+ "learning_rate": 4.7764880952380955e-06,
1600
+ "loss": 0.3387,
1601
+ "step": 221
1602
+ },
1603
+ {
1604
+ "epoch": 5.2,
1605
+ "grad_norm": 22.453044891357422,
1606
+ "learning_rate": 4.7732142857142855e-06,
1607
+ "loss": 0.2228,
1608
+ "step": 222
1609
+ },
1610
+ {
1611
+ "epoch": 5.23,
1612
+ "grad_norm": 34.940757751464844,
1613
+ "learning_rate": 4.769940476190476e-06,
1614
+ "loss": 0.2947,
1615
+ "step": 223
1616
+ },
1617
+ {
1618
+ "epoch": 5.25,
1619
+ "grad_norm": 10.684017181396484,
1620
+ "learning_rate": 4.766666666666667e-06,
1621
+ "loss": 0.303,
1622
+ "step": 224
1623
+ },
1624
+ {
1625
+ "epoch": 5.27,
1626
+ "grad_norm": 12.175122261047363,
1627
+ "learning_rate": 4.763392857142857e-06,
1628
+ "loss": 0.2265,
1629
+ "step": 225
1630
+ },
1631
+ {
1632
+ "epoch": 5.3,
1633
+ "grad_norm": 7.9341139793396,
1634
+ "learning_rate": 4.760119047619048e-06,
1635
+ "loss": 0.2761,
1636
+ "step": 226
1637
+ },
1638
+ {
1639
+ "epoch": 5.32,
1640
+ "grad_norm": 28.212831497192383,
1641
+ "learning_rate": 4.756845238095238e-06,
1642
+ "loss": 0.3097,
1643
+ "step": 227
1644
+ },
1645
+ {
1646
+ "epoch": 5.34,
1647
+ "grad_norm": 31.275775909423828,
1648
+ "learning_rate": 4.7535714285714286e-06,
1649
+ "loss": 0.3684,
1650
+ "step": 228
1651
+ },
1652
+ {
1653
+ "epoch": 5.37,
1654
+ "grad_norm": 23.49386978149414,
1655
+ "learning_rate": 4.750297619047619e-06,
1656
+ "loss": 0.221,
1657
+ "step": 229
1658
+ },
1659
+ {
1660
+ "epoch": 5.39,
1661
+ "grad_norm": 25.68425941467285,
1662
+ "learning_rate": 4.747023809523809e-06,
1663
+ "loss": 0.6258,
1664
+ "step": 230
1665
+ },
1666
+ {
1667
+ "epoch": 5.41,
1668
+ "grad_norm": 25.499523162841797,
1669
+ "learning_rate": 4.74375e-06,
1670
+ "loss": 0.3263,
1671
+ "step": 231
1672
+ },
1673
+ {
1674
+ "epoch": 5.44,
1675
+ "grad_norm": 17.70143699645996,
1676
+ "learning_rate": 4.740476190476191e-06,
1677
+ "loss": 0.2089,
1678
+ "step": 232
1679
+ },
1680
+ {
1681
+ "epoch": 5.46,
1682
+ "grad_norm": 14.780723571777344,
1683
+ "learning_rate": 4.737202380952381e-06,
1684
+ "loss": 0.1825,
1685
+ "step": 233
1686
+ },
1687
+ {
1688
+ "epoch": 5.48,
1689
+ "grad_norm": 5.02520227432251,
1690
+ "learning_rate": 4.733928571428572e-06,
1691
+ "loss": 0.2236,
1692
+ "step": 234
1693
+ },
1694
+ {
1695
+ "epoch": 5.51,
1696
+ "grad_norm": 26.148418426513672,
1697
+ "learning_rate": 4.7306547619047625e-06,
1698
+ "loss": 0.2121,
1699
+ "step": 235
1700
+ },
1701
+ {
1702
+ "epoch": 5.53,
1703
+ "grad_norm": 36.72833251953125,
1704
+ "learning_rate": 4.727380952380952e-06,
1705
+ "loss": 0.4922,
1706
+ "step": 236
1707
+ },
1708
+ {
1709
+ "epoch": 5.55,
1710
+ "grad_norm": 16.5570125579834,
1711
+ "learning_rate": 4.724107142857142e-06,
1712
+ "loss": 0.25,
1713
+ "step": 237
1714
+ },
1715
+ {
1716
+ "epoch": 5.58,
1717
+ "grad_norm": 36.25120544433594,
1718
+ "learning_rate": 4.720833333333333e-06,
1719
+ "loss": 0.4475,
1720
+ "step": 238
1721
+ },
1722
+ {
1723
+ "epoch": 5.6,
1724
+ "grad_norm": 13.410614013671875,
1725
+ "learning_rate": 4.717559523809523e-06,
1726
+ "loss": 0.2065,
1727
+ "step": 239
1728
+ },
1729
+ {
1730
+ "epoch": 5.62,
1731
+ "grad_norm": 18.893632888793945,
1732
+ "learning_rate": 4.714285714285714e-06,
1733
+ "loss": 0.2649,
1734
+ "step": 240
1735
+ },
1736
+ {
1737
+ "epoch": 5.65,
1738
+ "grad_norm": 9.750955581665039,
1739
+ "learning_rate": 4.711011904761905e-06,
1740
+ "loss": 0.2054,
1741
+ "step": 241
1742
+ },
1743
+ {
1744
+ "epoch": 5.67,
1745
+ "grad_norm": 9.189886093139648,
1746
+ "learning_rate": 4.707738095238095e-06,
1747
+ "loss": 0.1694,
1748
+ "step": 242
1749
+ },
1750
+ {
1751
+ "epoch": 5.7,
1752
+ "grad_norm": 6.946423530578613,
1753
+ "learning_rate": 4.7044642857142855e-06,
1754
+ "loss": 0.186,
1755
+ "step": 243
1756
+ },
1757
+ {
1758
+ "epoch": 5.72,
1759
+ "grad_norm": 12.687536239624023,
1760
+ "learning_rate": 4.701190476190476e-06,
1761
+ "loss": 0.1277,
1762
+ "step": 244
1763
+ },
1764
+ {
1765
+ "epoch": 5.74,
1766
+ "grad_norm": 30.949155807495117,
1767
+ "learning_rate": 4.697916666666666e-06,
1768
+ "loss": 0.2875,
1769
+ "step": 245
1770
+ },
1771
+ {
1772
+ "epoch": 5.77,
1773
+ "grad_norm": 11.202219009399414,
1774
+ "learning_rate": 4.694642857142857e-06,
1775
+ "loss": 0.2124,
1776
+ "step": 246
1777
+ },
1778
+ {
1779
+ "epoch": 5.79,
1780
+ "grad_norm": 2.138655185699463,
1781
+ "learning_rate": 4.691369047619047e-06,
1782
+ "loss": 0.067,
1783
+ "step": 247
1784
+ },
1785
+ {
1786
+ "epoch": 5.81,
1787
+ "grad_norm": 10.137479782104492,
1788
+ "learning_rate": 4.688095238095238e-06,
1789
+ "loss": 0.1081,
1790
+ "step": 248
1791
+ },
1792
+ {
1793
+ "epoch": 5.84,
1794
+ "grad_norm": 10.702754020690918,
1795
+ "learning_rate": 4.684821428571429e-06,
1796
+ "loss": 0.1297,
1797
+ "step": 249
1798
+ },
1799
+ {
1800
+ "epoch": 5.86,
1801
+ "grad_norm": 11.548544883728027,
1802
+ "learning_rate": 4.6815476190476185e-06,
1803
+ "loss": 0.1005,
1804
+ "step": 250
1805
+ },
1806
+ {
1807
+ "epoch": 5.88,
1808
+ "grad_norm": 22.75307273864746,
1809
+ "learning_rate": 4.678273809523809e-06,
1810
+ "loss": 0.1593,
1811
+ "step": 251
1812
+ },
1813
+ {
1814
+ "epoch": 5.91,
1815
+ "grad_norm": 9.360298156738281,
1816
+ "learning_rate": 4.675e-06,
1817
+ "loss": 0.0642,
1818
+ "step": 252
1819
+ },
1820
+ {
1821
+ "epoch": 5.93,
1822
+ "grad_norm": 19.112125396728516,
1823
+ "learning_rate": 4.67172619047619e-06,
1824
+ "loss": 0.2293,
1825
+ "step": 253
1826
+ },
1827
+ {
1828
+ "epoch": 5.95,
1829
+ "grad_norm": 28.329204559326172,
1830
+ "learning_rate": 4.668452380952381e-06,
1831
+ "loss": 0.2197,
1832
+ "step": 254
1833
+ },
1834
+ {
1835
+ "epoch": 5.98,
1836
+ "grad_norm": 18.537385940551758,
1837
+ "learning_rate": 4.665178571428571e-06,
1838
+ "loss": 0.3401,
1839
+ "step": 255
1840
+ },
1841
+ {
1842
+ "epoch": 6.0,
1843
+ "grad_norm": 38.293357849121094,
1844
+ "learning_rate": 4.661904761904762e-06,
1845
+ "loss": 0.3536,
1846
+ "step": 256
1847
+ },
1848
+ {
1849
+ "epoch": 6.0,
1850
+ "eval_accuracy": 0.8454545454545455,
1851
+ "eval_loss": 0.488827645778656,
1852
+ "eval_runtime": 1.962,
1853
+ "eval_samples_per_second": 112.13,
1854
+ "eval_steps_per_second": 28.033,
1855
+ "step": 256
1856
+ },
1857
+ {
1858
+ "epoch": 6.02,
1859
+ "grad_norm": 4.5169901847839355,
1860
+ "learning_rate": 4.6586309523809524e-06,
1861
+ "loss": 0.0597,
1862
+ "step": 257
1863
+ },
1864
+ {
1865
+ "epoch": 6.05,
1866
+ "grad_norm": 8.103499412536621,
1867
+ "learning_rate": 4.655357142857142e-06,
1868
+ "loss": 0.0922,
1869
+ "step": 258
1870
+ },
1871
+ {
1872
+ "epoch": 6.07,
1873
+ "grad_norm": 24.916545867919922,
1874
+ "learning_rate": 4.652083333333333e-06,
1875
+ "loss": 0.1184,
1876
+ "step": 259
1877
+ },
1878
+ {
1879
+ "epoch": 6.09,
1880
+ "grad_norm": 20.035037994384766,
1881
+ "learning_rate": 4.648809523809524e-06,
1882
+ "loss": 0.2025,
1883
+ "step": 260
1884
+ },
1885
+ {
1886
+ "epoch": 6.12,
1887
+ "grad_norm": 6.929393291473389,
1888
+ "learning_rate": 4.645535714285714e-06,
1889
+ "loss": 0.2023,
1890
+ "step": 261
1891
+ },
1892
+ {
1893
+ "epoch": 6.14,
1894
+ "grad_norm": 15.74714183807373,
1895
+ "learning_rate": 4.642261904761905e-06,
1896
+ "loss": 0.2039,
1897
+ "step": 262
1898
+ },
1899
+ {
1900
+ "epoch": 6.16,
1901
+ "grad_norm": 10.057015419006348,
1902
+ "learning_rate": 4.6389880952380955e-06,
1903
+ "loss": 0.0961,
1904
+ "step": 263
1905
+ },
1906
+ {
1907
+ "epoch": 6.19,
1908
+ "grad_norm": 1.1386276483535767,
1909
+ "learning_rate": 4.6357142857142855e-06,
1910
+ "loss": 0.0461,
1911
+ "step": 264
1912
+ },
1913
+ {
1914
+ "epoch": 6.21,
1915
+ "grad_norm": 3.349055767059326,
1916
+ "learning_rate": 4.632440476190476e-06,
1917
+ "loss": 0.181,
1918
+ "step": 265
1919
+ },
1920
+ {
1921
+ "epoch": 6.23,
1922
+ "grad_norm": 47.79439163208008,
1923
+ "learning_rate": 4.629166666666666e-06,
1924
+ "loss": 0.1307,
1925
+ "step": 266
1926
+ },
1927
+ {
1928
+ "epoch": 6.26,
1929
+ "grad_norm": 8.228836059570312,
1930
+ "learning_rate": 4.625892857142857e-06,
1931
+ "loss": 0.2784,
1932
+ "step": 267
1933
+ },
1934
+ {
1935
+ "epoch": 6.28,
1936
+ "grad_norm": 9.270495414733887,
1937
+ "learning_rate": 4.622619047619048e-06,
1938
+ "loss": 0.1872,
1939
+ "step": 268
1940
+ },
1941
+ {
1942
+ "epoch": 6.3,
1943
+ "grad_norm": 0.9865421652793884,
1944
+ "learning_rate": 4.619345238095238e-06,
1945
+ "loss": 0.0396,
1946
+ "step": 269
1947
+ },
1948
+ {
1949
+ "epoch": 6.33,
1950
+ "grad_norm": 46.00654602050781,
1951
+ "learning_rate": 4.616071428571429e-06,
1952
+ "loss": 0.2206,
1953
+ "step": 270
1954
+ },
1955
+ {
1956
+ "epoch": 6.35,
1957
+ "grad_norm": 35.620689392089844,
1958
+ "learning_rate": 4.612797619047619e-06,
1959
+ "loss": 0.0977,
1960
+ "step": 271
1961
+ },
1962
+ {
1963
+ "epoch": 6.38,
1964
+ "grad_norm": 16.84827995300293,
1965
+ "learning_rate": 4.609523809523809e-06,
1966
+ "loss": 0.0652,
1967
+ "step": 272
1968
+ },
1969
+ {
1970
+ "epoch": 6.4,
1971
+ "grad_norm": 11.778346061706543,
1972
+ "learning_rate": 4.60625e-06,
1973
+ "loss": 0.0619,
1974
+ "step": 273
1975
+ },
1976
+ {
1977
+ "epoch": 6.42,
1978
+ "grad_norm": 12.670008659362793,
1979
+ "learning_rate": 4.60297619047619e-06,
1980
+ "loss": 0.1638,
1981
+ "step": 274
1982
+ },
1983
+ {
1984
+ "epoch": 6.45,
1985
+ "grad_norm": 3.050675392150879,
1986
+ "learning_rate": 4.599702380952381e-06,
1987
+ "loss": 0.1787,
1988
+ "step": 275
1989
+ },
1990
+ {
1991
+ "epoch": 6.47,
1992
+ "grad_norm": 9.351134300231934,
1993
+ "learning_rate": 4.596428571428572e-06,
1994
+ "loss": 0.2906,
1995
+ "step": 276
1996
+ },
1997
+ {
1998
+ "epoch": 6.49,
1999
+ "grad_norm": 18.061622619628906,
2000
+ "learning_rate": 4.593154761904762e-06,
2001
+ "loss": 0.1441,
2002
+ "step": 277
2003
+ },
2004
+ {
2005
+ "epoch": 6.52,
2006
+ "grad_norm": 5.385902404785156,
2007
+ "learning_rate": 4.5898809523809525e-06,
2008
+ "loss": 0.0326,
2009
+ "step": 278
2010
+ },
2011
+ {
2012
+ "epoch": 6.54,
2013
+ "grad_norm": 1.4941680431365967,
2014
+ "learning_rate": 4.586607142857143e-06,
2015
+ "loss": 0.0304,
2016
+ "step": 279
2017
+ },
2018
+ {
2019
+ "epoch": 6.56,
2020
+ "grad_norm": 8.055174827575684,
2021
+ "learning_rate": 4.583333333333333e-06,
2022
+ "loss": 0.3047,
2023
+ "step": 280
2024
+ },
2025
+ {
2026
+ "epoch": 6.59,
2027
+ "grad_norm": 11.676218032836914,
2028
+ "learning_rate": 4.580059523809524e-06,
2029
+ "loss": 0.4685,
2030
+ "step": 281
2031
+ },
2032
+ {
2033
+ "epoch": 6.61,
2034
+ "grad_norm": 3.277010917663574,
2035
+ "learning_rate": 4.576785714285715e-06,
2036
+ "loss": 0.0433,
2037
+ "step": 282
2038
+ },
2039
+ {
2040
+ "epoch": 6.63,
2041
+ "grad_norm": 14.3773832321167,
2042
+ "learning_rate": 4.573511904761905e-06,
2043
+ "loss": 0.3162,
2044
+ "step": 283
2045
+ },
2046
+ {
2047
+ "epoch": 6.66,
2048
+ "grad_norm": 5.165549278259277,
2049
+ "learning_rate": 4.5702380952380956e-06,
2050
+ "loss": 0.1855,
2051
+ "step": 284
2052
+ },
2053
+ {
2054
+ "epoch": 6.68,
2055
+ "grad_norm": 23.437685012817383,
2056
+ "learning_rate": 4.5669642857142855e-06,
2057
+ "loss": 0.12,
2058
+ "step": 285
2059
+ },
2060
+ {
2061
+ "epoch": 6.7,
2062
+ "grad_norm": 3.0389316082000732,
2063
+ "learning_rate": 4.563690476190476e-06,
2064
+ "loss": 0.0439,
2065
+ "step": 286
2066
+ },
2067
+ {
2068
+ "epoch": 6.73,
2069
+ "grad_norm": 35.046241760253906,
2070
+ "learning_rate": 4.560416666666667e-06,
2071
+ "loss": 0.1244,
2072
+ "step": 287
2073
+ },
2074
+ {
2075
+ "epoch": 6.75,
2076
+ "grad_norm": 28.730819702148438,
2077
+ "learning_rate": 4.557142857142857e-06,
2078
+ "loss": 0.2398,
2079
+ "step": 288
2080
+ },
2081
+ {
2082
+ "epoch": 6.77,
2083
+ "grad_norm": 19.567346572875977,
2084
+ "learning_rate": 4.553869047619047e-06,
2085
+ "loss": 0.0952,
2086
+ "step": 289
2087
+ },
2088
+ {
2089
+ "epoch": 6.8,
2090
+ "grad_norm": 17.299070358276367,
2091
+ "learning_rate": 4.550595238095238e-06,
2092
+ "loss": 0.2421,
2093
+ "step": 290
2094
+ },
2095
+ {
2096
+ "epoch": 6.82,
2097
+ "grad_norm": 4.0813422203063965,
2098
+ "learning_rate": 4.547321428571428e-06,
2099
+ "loss": 0.1717,
2100
+ "step": 291
2101
+ },
2102
+ {
2103
+ "epoch": 6.84,
2104
+ "grad_norm": 15.247352600097656,
2105
+ "learning_rate": 4.544047619047619e-06,
2106
+ "loss": 0.1287,
2107
+ "step": 292
2108
+ },
2109
+ {
2110
+ "epoch": 6.87,
2111
+ "grad_norm": 22.969430923461914,
2112
+ "learning_rate": 4.540773809523809e-06,
2113
+ "loss": 0.0944,
2114
+ "step": 293
2115
+ },
2116
+ {
2117
+ "epoch": 6.89,
2118
+ "grad_norm": 14.135355949401855,
2119
+ "learning_rate": 4.537499999999999e-06,
2120
+ "loss": 0.0501,
2121
+ "step": 294
2122
+ },
2123
+ {
2124
+ "epoch": 6.91,
2125
+ "grad_norm": 6.1480607986450195,
2126
+ "learning_rate": 4.53422619047619e-06,
2127
+ "loss": 0.0491,
2128
+ "step": 295
2129
+ },
2130
+ {
2131
+ "epoch": 6.94,
2132
+ "grad_norm": 5.824121952056885,
2133
+ "learning_rate": 4.530952380952381e-06,
2134
+ "loss": 0.4394,
2135
+ "step": 296
2136
+ },
2137
+ {
2138
+ "epoch": 6.96,
2139
+ "grad_norm": 3.691617727279663,
2140
+ "learning_rate": 4.527678571428571e-06,
2141
+ "loss": 0.0313,
2142
+ "step": 297
2143
+ },
2144
+ {
2145
+ "epoch": 6.98,
2146
+ "grad_norm": 1.4261701107025146,
2147
+ "learning_rate": 4.524404761904762e-06,
2148
+ "loss": 0.1709,
2149
+ "step": 298
2150
+ },
2151
+ {
2152
+ "epoch": 6.98,
2153
+ "eval_accuracy": 0.85,
2154
+ "eval_loss": 0.5375058054924011,
2155
+ "eval_runtime": 1.9645,
2156
+ "eval_samples_per_second": 111.989,
2157
+ "eval_steps_per_second": 27.997,
2158
+ "step": 298
2159
+ },
2160
+ {
2161
+ "epoch": 7.01,
2162
+ "grad_norm": 37.57842254638672,
2163
+ "learning_rate": 4.5211309523809525e-06,
2164
+ "loss": 0.299,
2165
+ "step": 299
2166
+ },
2167
+ {
2168
+ "epoch": 7.03,
2169
+ "grad_norm": 19.698734283447266,
2170
+ "learning_rate": 4.5178571428571424e-06,
2171
+ "loss": 0.3081,
2172
+ "step": 300
2173
+ },
2174
+ {
2175
+ "epoch": 7.05,
2176
+ "grad_norm": 0.6386685371398926,
2177
+ "learning_rate": 4.514583333333333e-06,
2178
+ "loss": 0.024,
2179
+ "step": 301
2180
+ },
2181
+ {
2182
+ "epoch": 7.08,
2183
+ "grad_norm": 1.1291331052780151,
2184
+ "learning_rate": 4.511309523809523e-06,
2185
+ "loss": 0.0343,
2186
+ "step": 302
2187
+ },
2188
+ {
2189
+ "epoch": 7.1,
2190
+ "grad_norm": 3.5354042053222656,
2191
+ "learning_rate": 4.508035714285714e-06,
2192
+ "loss": 0.1509,
2193
+ "step": 303
2194
+ },
2195
+ {
2196
+ "epoch": 7.12,
2197
+ "grad_norm": 0.7393876314163208,
2198
+ "learning_rate": 4.504761904761905e-06,
2199
+ "loss": 0.0237,
2200
+ "step": 304
2201
+ },
2202
+ {
2203
+ "epoch": 7.15,
2204
+ "grad_norm": 2.4936347007751465,
2205
+ "learning_rate": 4.501488095238095e-06,
2206
+ "loss": 0.0299,
2207
+ "step": 305
2208
+ },
2209
+ {
2210
+ "epoch": 7.17,
2211
+ "grad_norm": 28.873003005981445,
2212
+ "learning_rate": 4.4982142857142855e-06,
2213
+ "loss": 0.0815,
2214
+ "step": 306
2215
+ },
2216
+ {
2217
+ "epoch": 7.2,
2218
+ "grad_norm": 53.65907669067383,
2219
+ "learning_rate": 4.494940476190476e-06,
2220
+ "loss": 0.1271,
2221
+ "step": 307
2222
+ },
2223
+ {
2224
+ "epoch": 7.22,
2225
+ "grad_norm": 0.4335859715938568,
2226
+ "learning_rate": 4.491666666666666e-06,
2227
+ "loss": 0.0183,
2228
+ "step": 308
2229
+ },
2230
+ {
2231
+ "epoch": 7.24,
2232
+ "grad_norm": 2.026947259902954,
2233
+ "learning_rate": 4.488392857142857e-06,
2234
+ "loss": 0.1534,
2235
+ "step": 309
2236
+ },
2237
+ {
2238
+ "epoch": 7.27,
2239
+ "grad_norm": 0.4929918348789215,
2240
+ "learning_rate": 4.485119047619047e-06,
2241
+ "loss": 0.0192,
2242
+ "step": 310
2243
+ },
2244
+ {
2245
+ "epoch": 7.29,
2246
+ "grad_norm": 24.899736404418945,
2247
+ "learning_rate": 4.481845238095238e-06,
2248
+ "loss": 0.0524,
2249
+ "step": 311
2250
+ },
2251
+ {
2252
+ "epoch": 7.31,
2253
+ "grad_norm": 0.6321877837181091,
2254
+ "learning_rate": 4.478571428571429e-06,
2255
+ "loss": 0.0205,
2256
+ "step": 312
2257
+ },
2258
+ {
2259
+ "epoch": 7.34,
2260
+ "grad_norm": 1.5749492645263672,
2261
+ "learning_rate": 4.475297619047619e-06,
2262
+ "loss": 0.1671,
2263
+ "step": 313
2264
+ },
2265
+ {
2266
+ "epoch": 7.36,
2267
+ "grad_norm": 0.4469403624534607,
2268
+ "learning_rate": 4.472023809523809e-06,
2269
+ "loss": 0.0154,
2270
+ "step": 314
2271
+ },
2272
+ {
2273
+ "epoch": 7.38,
2274
+ "grad_norm": 10.251415252685547,
2275
+ "learning_rate": 4.46875e-06,
2276
+ "loss": 0.1924,
2277
+ "step": 315
2278
+ },
2279
+ {
2280
+ "epoch": 7.41,
2281
+ "grad_norm": 4.918264865875244,
2282
+ "learning_rate": 4.46547619047619e-06,
2283
+ "loss": 0.499,
2284
+ "step": 316
2285
+ },
2286
+ {
2287
+ "epoch": 7.43,
2288
+ "grad_norm": 0.3266950845718384,
2289
+ "learning_rate": 4.462202380952381e-06,
2290
+ "loss": 0.0131,
2291
+ "step": 317
2292
+ },
2293
+ {
2294
+ "epoch": 7.45,
2295
+ "grad_norm": 11.629830360412598,
2296
+ "learning_rate": 4.458928571428572e-06,
2297
+ "loss": 0.3543,
2298
+ "step": 318
2299
+ },
2300
+ {
2301
+ "epoch": 7.48,
2302
+ "grad_norm": 35.164634704589844,
2303
+ "learning_rate": 4.455654761904762e-06,
2304
+ "loss": 0.1511,
2305
+ "step": 319
2306
+ },
2307
+ {
2308
+ "epoch": 7.5,
2309
+ "grad_norm": 0.5766786932945251,
2310
+ "learning_rate": 4.4523809523809525e-06,
2311
+ "loss": 0.0165,
2312
+ "step": 320
2313
+ },
2314
+ {
2315
+ "epoch": 7.52,
2316
+ "grad_norm": 30.383398056030273,
2317
+ "learning_rate": 4.4491071428571425e-06,
2318
+ "loss": 0.152,
2319
+ "step": 321
2320
+ },
2321
+ {
2322
+ "epoch": 7.55,
2323
+ "grad_norm": 26.80906867980957,
2324
+ "learning_rate": 4.445833333333333e-06,
2325
+ "loss": 0.2207,
2326
+ "step": 322
2327
+ },
2328
+ {
2329
+ "epoch": 7.57,
2330
+ "grad_norm": 0.6242391467094421,
2331
+ "learning_rate": 4.442559523809524e-06,
2332
+ "loss": 0.0129,
2333
+ "step": 323
2334
+ },
2335
+ {
2336
+ "epoch": 7.59,
2337
+ "grad_norm": 8.999675750732422,
2338
+ "learning_rate": 4.439285714285714e-06,
2339
+ "loss": 0.1625,
2340
+ "step": 324
2341
+ },
2342
+ {
2343
+ "epoch": 7.62,
2344
+ "grad_norm": 0.30194684863090515,
2345
+ "learning_rate": 4.436011904761905e-06,
2346
+ "loss": 0.0107,
2347
+ "step": 325
2348
+ },
2349
+ {
2350
+ "epoch": 7.64,
2351
+ "grad_norm": 11.341878890991211,
2352
+ "learning_rate": 4.432738095238096e-06,
2353
+ "loss": 0.1591,
2354
+ "step": 326
2355
+ },
2356
+ {
2357
+ "epoch": 7.66,
2358
+ "grad_norm": 1.6119003295898438,
2359
+ "learning_rate": 4.4294642857142856e-06,
2360
+ "loss": 0.0128,
2361
+ "step": 327
2362
+ },
2363
+ {
2364
+ "epoch": 7.69,
2365
+ "grad_norm": 58.44689178466797,
2366
+ "learning_rate": 4.426190476190476e-06,
2367
+ "loss": 0.1724,
2368
+ "step": 328
2369
+ },
2370
+ {
2371
+ "epoch": 7.71,
2372
+ "grad_norm": 1.108935832977295,
2373
+ "learning_rate": 4.422916666666666e-06,
2374
+ "loss": 0.0137,
2375
+ "step": 329
2376
+ },
2377
+ {
2378
+ "epoch": 7.73,
2379
+ "grad_norm": 0.3343135118484497,
2380
+ "learning_rate": 4.419642857142857e-06,
2381
+ "loss": 0.0115,
2382
+ "step": 330
2383
+ },
2384
+ {
2385
+ "epoch": 7.76,
2386
+ "grad_norm": 2.9117400646209717,
2387
+ "learning_rate": 4.416369047619048e-06,
2388
+ "loss": 0.0137,
2389
+ "step": 331
2390
+ },
2391
+ {
2392
+ "epoch": 7.78,
2393
+ "grad_norm": 0.3950476348400116,
2394
+ "learning_rate": 4.413095238095238e-06,
2395
+ "loss": 0.01,
2396
+ "step": 332
2397
+ },
2398
+ {
2399
+ "epoch": 7.8,
2400
+ "grad_norm": 0.46693718433380127,
2401
+ "learning_rate": 4.409821428571429e-06,
2402
+ "loss": 0.0109,
2403
+ "step": 333
2404
+ },
2405
+ {
2406
+ "epoch": 7.83,
2407
+ "grad_norm": 0.3062843680381775,
2408
+ "learning_rate": 4.4065476190476195e-06,
2409
+ "loss": 0.0082,
2410
+ "step": 334
2411
+ },
2412
+ {
2413
+ "epoch": 7.85,
2414
+ "grad_norm": 66.0848388671875,
2415
+ "learning_rate": 4.4032738095238094e-06,
2416
+ "loss": 0.0962,
2417
+ "step": 335
2418
+ },
2419
+ {
2420
+ "epoch": 7.88,
2421
+ "grad_norm": 0.5519910454750061,
2422
+ "learning_rate": 4.4e-06,
2423
+ "loss": 0.0084,
2424
+ "step": 336
2425
+ },
2426
+ {
2427
+ "epoch": 7.9,
2428
+ "grad_norm": 0.20682856440544128,
2429
+ "learning_rate": 4.396726190476191e-06,
2430
+ "loss": 0.0077,
2431
+ "step": 337
2432
+ },
2433
+ {
2434
+ "epoch": 7.92,
2435
+ "grad_norm": 23.140066146850586,
2436
+ "learning_rate": 4.393452380952381e-06,
2437
+ "loss": 0.0632,
2438
+ "step": 338
2439
+ },
2440
+ {
2441
+ "epoch": 7.95,
2442
+ "grad_norm": 1.7450965642929077,
2443
+ "learning_rate": 4.390178571428572e-06,
2444
+ "loss": 0.0097,
2445
+ "step": 339
2446
+ },
2447
+ {
2448
+ "epoch": 7.97,
2449
+ "grad_norm": 14.685700416564941,
2450
+ "learning_rate": 4.386904761904762e-06,
2451
+ "loss": 0.1616,
2452
+ "step": 340
2453
+ },
2454
+ {
2455
+ "epoch": 7.99,
2456
+ "grad_norm": 0.1476498693227768,
2457
+ "learning_rate": 4.3836309523809525e-06,
2458
+ "loss": 0.0067,
2459
+ "step": 341
2460
+ },
2461
+ {
2462
+ "epoch": 7.99,
2463
+ "eval_accuracy": 0.8863636363636364,
2464
+ "eval_loss": 0.5613145232200623,
2465
+ "eval_runtime": 1.9469,
2466
+ "eval_samples_per_second": 113.002,
2467
+ "eval_steps_per_second": 28.251,
2468
+ "step": 341
2469
+ },
2470
+ {
2471
+ "epoch": 8.02,
2472
+ "grad_norm": 0.2792944312095642,
2473
+ "learning_rate": 4.3803571428571425e-06,
2474
+ "loss": 0.0069,
2475
+ "step": 342
2476
+ },
2477
+ {
2478
+ "epoch": 8.04,
2479
+ "grad_norm": 0.5173578262329102,
2480
+ "learning_rate": 4.377083333333333e-06,
2481
+ "loss": 0.0073,
2482
+ "step": 343
2483
+ },
2484
+ {
2485
+ "epoch": 8.06,
2486
+ "grad_norm": 2.705310344696045,
2487
+ "learning_rate": 4.373809523809523e-06,
2488
+ "loss": 0.0116,
2489
+ "step": 344
2490
+ },
2491
+ {
2492
+ "epoch": 8.09,
2493
+ "grad_norm": 9.17689037322998,
2494
+ "learning_rate": 4.370535714285714e-06,
2495
+ "loss": 0.1767,
2496
+ "step": 345
2497
+ },
2498
+ {
2499
+ "epoch": 8.11,
2500
+ "grad_norm": 3.7868285179138184,
2501
+ "learning_rate": 4.367261904761904e-06,
2502
+ "loss": 0.0087,
2503
+ "step": 346
2504
+ },
2505
+ {
2506
+ "epoch": 8.13,
2507
+ "grad_norm": 0.3689323663711548,
2508
+ "learning_rate": 4.363988095238095e-06,
2509
+ "loss": 0.0061,
2510
+ "step": 347
2511
+ },
2512
+ {
2513
+ "epoch": 8.16,
2514
+ "grad_norm": 0.20344328880310059,
2515
+ "learning_rate": 4.360714285714286e-06,
2516
+ "loss": 0.0057,
2517
+ "step": 348
2518
+ },
2519
+ {
2520
+ "epoch": 8.18,
2521
+ "grad_norm": 37.69071578979492,
2522
+ "learning_rate": 4.3574404761904755e-06,
2523
+ "loss": 0.2596,
2524
+ "step": 349
2525
+ },
2526
+ {
2527
+ "epoch": 8.2,
2528
+ "grad_norm": 0.16801539063453674,
2529
+ "learning_rate": 4.354166666666666e-06,
2530
+ "loss": 0.0051,
2531
+ "step": 350
2532
+ },
2533
+ {
2534
+ "epoch": 8.23,
2535
+ "grad_norm": 5.060468673706055,
2536
+ "learning_rate": 4.350892857142857e-06,
2537
+ "loss": 0.2096,
2538
+ "step": 351
2539
+ },
2540
+ {
2541
+ "epoch": 8.25,
2542
+ "grad_norm": 1.643596887588501,
2543
+ "learning_rate": 4.347619047619047e-06,
2544
+ "loss": 0.0076,
2545
+ "step": 352
2546
+ },
2547
+ {
2548
+ "epoch": 8.27,
2549
+ "grad_norm": 0.7824279069900513,
2550
+ "learning_rate": 4.344345238095238e-06,
2551
+ "loss": 0.0059,
2552
+ "step": 353
2553
+ },
2554
+ {
2555
+ "epoch": 8.3,
2556
+ "grad_norm": 2.7976343631744385,
2557
+ "learning_rate": 4.341071428571429e-06,
2558
+ "loss": 0.0067,
2559
+ "step": 354
2560
+ },
2561
+ {
2562
+ "epoch": 8.32,
2563
+ "grad_norm": 13.010490417480469,
2564
+ "learning_rate": 4.337797619047619e-06,
2565
+ "loss": 0.3109,
2566
+ "step": 355
2567
+ },
2568
+ {
2569
+ "epoch": 8.34,
2570
+ "grad_norm": 0.10449977964162827,
2571
+ "learning_rate": 4.3345238095238094e-06,
2572
+ "loss": 0.0039,
2573
+ "step": 356
2574
+ },
2575
+ {
2576
+ "epoch": 8.37,
2577
+ "grad_norm": 0.4954240024089813,
2578
+ "learning_rate": 4.331249999999999e-06,
2579
+ "loss": 0.0064,
2580
+ "step": 357
2581
+ },
2582
+ {
2583
+ "epoch": 8.39,
2584
+ "grad_norm": 1.7399401664733887,
2585
+ "learning_rate": 4.32797619047619e-06,
2586
+ "loss": 0.0056,
2587
+ "step": 358
2588
+ },
2589
+ {
2590
+ "epoch": 8.41,
2591
+ "grad_norm": 7.1369147300720215,
2592
+ "learning_rate": 4.324702380952381e-06,
2593
+ "loss": 0.2385,
2594
+ "step": 359
2595
+ },
2596
+ {
2597
+ "epoch": 8.44,
2598
+ "grad_norm": 23.376176834106445,
2599
+ "learning_rate": 4.321428571428571e-06,
2600
+ "loss": 0.3003,
2601
+ "step": 360
2602
+ },
2603
+ {
2604
+ "epoch": 8.46,
2605
+ "grad_norm": 0.34425580501556396,
2606
+ "learning_rate": 4.318154761904762e-06,
2607
+ "loss": 0.0044,
2608
+ "step": 361
2609
+ },
2610
+ {
2611
+ "epoch": 8.48,
2612
+ "grad_norm": 0.21413090825080872,
2613
+ "learning_rate": 4.3148809523809526e-06,
2614
+ "loss": 0.004,
2615
+ "step": 362
2616
+ },
2617
+ {
2618
+ "epoch": 8.51,
2619
+ "grad_norm": 3.548433303833008,
2620
+ "learning_rate": 4.3116071428571425e-06,
2621
+ "loss": 0.123,
2622
+ "step": 363
2623
+ },
2624
+ {
2625
+ "epoch": 8.53,
2626
+ "grad_norm": 1.6423828601837158,
2627
+ "learning_rate": 4.308333333333333e-06,
2628
+ "loss": 0.2496,
2629
+ "step": 364
2630
+ },
2631
+ {
2632
+ "epoch": 8.55,
2633
+ "grad_norm": 2.283780813217163,
2634
+ "learning_rate": 4.305059523809523e-06,
2635
+ "loss": 0.2397,
2636
+ "step": 365
2637
+ },
2638
+ {
2639
+ "epoch": 8.58,
2640
+ "grad_norm": 0.362973153591156,
2641
+ "learning_rate": 4.301785714285714e-06,
2642
+ "loss": 0.0064,
2643
+ "step": 366
2644
+ },
2645
+ {
2646
+ "epoch": 8.6,
2647
+ "grad_norm": 19.265830993652344,
2648
+ "learning_rate": 4.298511904761905e-06,
2649
+ "loss": 0.0201,
2650
+ "step": 367
2651
+ },
2652
+ {
2653
+ "epoch": 8.62,
2654
+ "grad_norm": 0.20540741086006165,
2655
+ "learning_rate": 4.295238095238095e-06,
2656
+ "loss": 0.0047,
2657
+ "step": 368
2658
+ },
2659
+ {
2660
+ "epoch": 8.65,
2661
+ "grad_norm": 3.41279935836792,
2662
+ "learning_rate": 4.291964285714286e-06,
2663
+ "loss": 0.0088,
2664
+ "step": 369
2665
+ },
2666
+ {
2667
+ "epoch": 8.67,
2668
+ "grad_norm": 10.215970039367676,
2669
+ "learning_rate": 4.288690476190476e-06,
2670
+ "loss": 0.0162,
2671
+ "step": 370
2672
+ },
2673
+ {
2674
+ "epoch": 8.7,
2675
+ "grad_norm": 0.7728336453437805,
2676
+ "learning_rate": 4.285416666666666e-06,
2677
+ "loss": 0.008,
2678
+ "step": 371
2679
+ },
2680
+ {
2681
+ "epoch": 8.72,
2682
+ "grad_norm": 2.603362798690796,
2683
+ "learning_rate": 4.282142857142857e-06,
2684
+ "loss": 0.0122,
2685
+ "step": 372
2686
+ },
2687
+ {
2688
+ "epoch": 8.74,
2689
+ "grad_norm": 58.749488830566406,
2690
+ "learning_rate": 4.278869047619048e-06,
2691
+ "loss": 0.304,
2692
+ "step": 373
2693
+ },
2694
+ {
2695
+ "epoch": 8.77,
2696
+ "grad_norm": 8.352728843688965,
2697
+ "learning_rate": 4.275595238095238e-06,
2698
+ "loss": 0.1469,
2699
+ "step": 374
2700
+ },
2701
+ {
2702
+ "epoch": 8.79,
2703
+ "grad_norm": 0.2930574417114258,
2704
+ "learning_rate": 4.272321428571429e-06,
2705
+ "loss": 0.0067,
2706
+ "step": 375
2707
+ },
2708
+ {
2709
+ "epoch": 8.81,
2710
+ "grad_norm": 0.6395189166069031,
2711
+ "learning_rate": 4.269047619047619e-06,
2712
+ "loss": 0.0114,
2713
+ "step": 376
2714
+ },
2715
+ {
2716
+ "epoch": 8.84,
2717
+ "grad_norm": 21.4405517578125,
2718
+ "learning_rate": 4.2657738095238095e-06,
2719
+ "loss": 0.0245,
2720
+ "step": 377
2721
+ },
2722
+ {
2723
+ "epoch": 8.86,
2724
+ "grad_norm": 20.98697280883789,
2725
+ "learning_rate": 4.2625e-06,
2726
+ "loss": 0.2003,
2727
+ "step": 378
2728
+ },
2729
+ {
2730
+ "epoch": 8.88,
2731
+ "grad_norm": 0.16704310476779938,
2732
+ "learning_rate": 4.25922619047619e-06,
2733
+ "loss": 0.0045,
2734
+ "step": 379
2735
+ },
2736
+ {
2737
+ "epoch": 8.91,
2738
+ "grad_norm": 0.30824193358421326,
2739
+ "learning_rate": 4.255952380952381e-06,
2740
+ "loss": 0.0063,
2741
+ "step": 380
2742
+ },
2743
+ {
2744
+ "epoch": 8.93,
2745
+ "grad_norm": 38.134681701660156,
2746
+ "learning_rate": 4.252678571428572e-06,
2747
+ "loss": 0.0395,
2748
+ "step": 381
2749
+ },
2750
+ {
2751
+ "epoch": 8.95,
2752
+ "grad_norm": 0.32065123319625854,
2753
+ "learning_rate": 4.249404761904762e-06,
2754
+ "loss": 0.0055,
2755
+ "step": 382
2756
+ },
2757
+ {
2758
+ "epoch": 8.98,
2759
+ "grad_norm": 2.818316698074341,
2760
+ "learning_rate": 4.2461309523809526e-06,
2761
+ "loss": 0.0077,
2762
+ "step": 383
2763
+ },
2764
+ {
2765
+ "epoch": 9.0,
2766
+ "grad_norm": 8.902009010314941,
2767
+ "learning_rate": 4.2428571428571425e-06,
2768
+ "loss": 0.0103,
2769
+ "step": 384
2770
+ },
2771
+ {
2772
+ "epoch": 9.0,
2773
+ "eval_accuracy": 0.8772727272727273,
2774
+ "eval_loss": 0.6385995149612427,
2775
+ "eval_runtime": 1.9711,
2776
+ "eval_samples_per_second": 111.612,
2777
+ "eval_steps_per_second": 27.903,
2778
+ "step": 384
2779
+ },
2780
+ {
2781
+ "epoch": 9.02,
2782
+ "grad_norm": 50.5201301574707,
2783
+ "learning_rate": 4.239583333333333e-06,
2784
+ "loss": 0.1206,
2785
+ "step": 385
2786
+ },
2787
+ {
2788
+ "epoch": 9.05,
2789
+ "grad_norm": 0.10049311071634293,
2790
+ "learning_rate": 4.236309523809524e-06,
2791
+ "loss": 0.0038,
2792
+ "step": 386
2793
+ },
2794
+ {
2795
+ "epoch": 9.07,
2796
+ "grad_norm": 0.11850426346063614,
2797
+ "learning_rate": 4.233035714285714e-06,
2798
+ "loss": 0.0038,
2799
+ "step": 387
2800
+ },
2801
+ {
2802
+ "epoch": 9.09,
2803
+ "grad_norm": 0.10592693090438843,
2804
+ "learning_rate": 4.229761904761905e-06,
2805
+ "loss": 0.0037,
2806
+ "step": 388
2807
+ },
2808
+ {
2809
+ "epoch": 9.12,
2810
+ "grad_norm": 0.4136406481266022,
2811
+ "learning_rate": 4.226488095238096e-06,
2812
+ "loss": 0.0049,
2813
+ "step": 389
2814
+ },
2815
+ {
2816
+ "epoch": 9.14,
2817
+ "grad_norm": 0.09526017308235168,
2818
+ "learning_rate": 4.223214285714286e-06,
2819
+ "loss": 0.0034,
2820
+ "step": 390
2821
+ },
2822
+ {
2823
+ "epoch": 9.16,
2824
+ "grad_norm": 0.08932331204414368,
2825
+ "learning_rate": 4.2199404761904764e-06,
2826
+ "loss": 0.0034,
2827
+ "step": 391
2828
+ },
2829
+ {
2830
+ "epoch": 9.19,
2831
+ "grad_norm": 12.363450050354004,
2832
+ "learning_rate": 4.216666666666667e-06,
2833
+ "loss": 0.1276,
2834
+ "step": 392
2835
+ },
2836
+ {
2837
+ "epoch": 9.21,
2838
+ "grad_norm": 0.19276930391788483,
2839
+ "learning_rate": 4.213392857142857e-06,
2840
+ "loss": 0.0036,
2841
+ "step": 393
2842
+ },
2843
+ {
2844
+ "epoch": 9.23,
2845
+ "grad_norm": 0.10427949577569962,
2846
+ "learning_rate": 4.210119047619047e-06,
2847
+ "loss": 0.0041,
2848
+ "step": 394
2849
+ },
2850
+ {
2851
+ "epoch": 9.26,
2852
+ "grad_norm": 0.07943836599588394,
2853
+ "learning_rate": 4.206845238095238e-06,
2854
+ "loss": 0.0029,
2855
+ "step": 395
2856
+ },
2857
+ {
2858
+ "epoch": 9.28,
2859
+ "grad_norm": 0.18391607701778412,
2860
+ "learning_rate": 4.203571428571428e-06,
2861
+ "loss": 0.0036,
2862
+ "step": 396
2863
+ },
2864
+ {
2865
+ "epoch": 9.3,
2866
+ "grad_norm": 0.3423359990119934,
2867
+ "learning_rate": 4.200297619047619e-06,
2868
+ "loss": 0.0037,
2869
+ "step": 397
2870
+ },
2871
+ {
2872
+ "epoch": 9.33,
2873
+ "grad_norm": 41.31951904296875,
2874
+ "learning_rate": 4.1970238095238095e-06,
2875
+ "loss": 0.215,
2876
+ "step": 398
2877
+ },
2878
+ {
2879
+ "epoch": 9.35,
2880
+ "grad_norm": 0.1459679901599884,
2881
+ "learning_rate": 4.1937499999999994e-06,
2882
+ "loss": 0.0035,
2883
+ "step": 399
2884
+ },
2885
+ {
2886
+ "epoch": 9.38,
2887
+ "grad_norm": 0.0786140188574791,
2888
+ "learning_rate": 4.19047619047619e-06,
2889
+ "loss": 0.0027,
2890
+ "step": 400
2891
+ },
2892
+ {
2893
+ "epoch": 9.4,
2894
+ "grad_norm": 1.5436954498291016,
2895
+ "learning_rate": 4.18720238095238e-06,
2896
+ "loss": 0.086,
2897
+ "step": 401
2898
+ },
2899
+ {
2900
+ "epoch": 9.42,
2901
+ "grad_norm": 1.1575084924697876,
2902
+ "learning_rate": 4.183928571428571e-06,
2903
+ "loss": 0.0048,
2904
+ "step": 402
2905
+ },
2906
+ {
2907
+ "epoch": 9.45,
2908
+ "grad_norm": 0.07273612916469574,
2909
+ "learning_rate": 4.180654761904762e-06,
2910
+ "loss": 0.0027,
2911
+ "step": 403
2912
+ },
2913
+ {
2914
+ "epoch": 9.47,
2915
+ "grad_norm": 2.0359814167022705,
2916
+ "learning_rate": 4.177380952380952e-06,
2917
+ "loss": 0.2268,
2918
+ "step": 404
2919
+ },
2920
+ {
2921
+ "epoch": 9.49,
2922
+ "grad_norm": 0.06620334088802338,
2923
+ "learning_rate": 4.1741071428571426e-06,
2924
+ "loss": 0.0024,
2925
+ "step": 405
2926
+ },
2927
+ {
2928
+ "epoch": 9.52,
2929
+ "grad_norm": 0.9721193313598633,
2930
+ "learning_rate": 4.170833333333333e-06,
2931
+ "loss": 0.0038,
2932
+ "step": 406
2933
+ },
2934
+ {
2935
+ "epoch": 9.54,
2936
+ "grad_norm": 1.0129694938659668,
2937
+ "learning_rate": 4.167559523809523e-06,
2938
+ "loss": 0.0047,
2939
+ "step": 407
2940
+ },
2941
+ {
2942
+ "epoch": 9.56,
2943
+ "grad_norm": 2.4980669021606445,
2944
+ "learning_rate": 4.164285714285714e-06,
2945
+ "loss": 0.2284,
2946
+ "step": 408
2947
+ },
2948
+ {
2949
+ "epoch": 9.59,
2950
+ "grad_norm": 0.1482830047607422,
2951
+ "learning_rate": 4.161011904761905e-06,
2952
+ "loss": 0.0033,
2953
+ "step": 409
2954
+ },
2955
+ {
2956
+ "epoch": 9.61,
2957
+ "grad_norm": 96.48265075683594,
2958
+ "learning_rate": 4.157738095238095e-06,
2959
+ "loss": 0.0618,
2960
+ "step": 410
2961
+ },
2962
+ {
2963
+ "epoch": 9.63,
2964
+ "grad_norm": 0.1294354349374771,
2965
+ "learning_rate": 4.154464285714286e-06,
2966
+ "loss": 0.0031,
2967
+ "step": 411
2968
+ },
2969
+ {
2970
+ "epoch": 9.66,
2971
+ "grad_norm": 55.35332107543945,
2972
+ "learning_rate": 4.151190476190476e-06,
2973
+ "loss": 0.1618,
2974
+ "step": 412
2975
+ },
2976
+ {
2977
+ "epoch": 9.68,
2978
+ "grad_norm": 3.1412291526794434,
2979
+ "learning_rate": 4.147916666666666e-06,
2980
+ "loss": 0.1818,
2981
+ "step": 413
2982
+ },
2983
+ {
2984
+ "epoch": 9.7,
2985
+ "grad_norm": 2.8581583499908447,
2986
+ "learning_rate": 4.144642857142857e-06,
2987
+ "loss": 0.0895,
2988
+ "step": 414
2989
+ },
2990
+ {
2991
+ "epoch": 9.73,
2992
+ "grad_norm": 0.1566578447818756,
2993
+ "learning_rate": 4.141369047619047e-06,
2994
+ "loss": 0.004,
2995
+ "step": 415
2996
+ },
2997
+ {
2998
+ "epoch": 9.75,
2999
+ "grad_norm": 0.20300696790218353,
3000
+ "learning_rate": 4.138095238095238e-06,
3001
+ "loss": 0.0045,
3002
+ "step": 416
3003
+ },
3004
+ {
3005
+ "epoch": 9.77,
3006
+ "grad_norm": 28.00294303894043,
3007
+ "learning_rate": 4.134821428571429e-06,
3008
+ "loss": 0.0955,
3009
+ "step": 417
3010
+ },
3011
+ {
3012
+ "epoch": 9.8,
3013
+ "grad_norm": 4.9537553787231445,
3014
+ "learning_rate": 4.131547619047619e-06,
3015
+ "loss": 0.0108,
3016
+ "step": 418
3017
+ },
3018
+ {
3019
+ "epoch": 9.82,
3020
+ "grad_norm": 1.2664889097213745,
3021
+ "learning_rate": 4.1282738095238095e-06,
3022
+ "loss": 0.0144,
3023
+ "step": 419
3024
+ },
3025
+ {
3026
+ "epoch": 9.84,
3027
+ "grad_norm": 21.156566619873047,
3028
+ "learning_rate": 4.125e-06,
3029
+ "loss": 0.0984,
3030
+ "step": 420
3031
+ },
3032
+ {
3033
+ "epoch": 9.87,
3034
+ "grad_norm": 14.108423233032227,
3035
+ "learning_rate": 4.12172619047619e-06,
3036
+ "loss": 0.0181,
3037
+ "step": 421
3038
+ },
3039
+ {
3040
+ "epoch": 9.89,
3041
+ "grad_norm": 0.3327002227306366,
3042
+ "learning_rate": 4.118452380952381e-06,
3043
+ "loss": 0.0059,
3044
+ "step": 422
3045
+ },
3046
+ {
3047
+ "epoch": 9.91,
3048
+ "grad_norm": 0.26034343242645264,
3049
+ "learning_rate": 4.115178571428571e-06,
3050
+ "loss": 0.0047,
3051
+ "step": 423
3052
+ },
3053
+ {
3054
+ "epoch": 9.94,
3055
+ "grad_norm": 0.19648778438568115,
3056
+ "learning_rate": 4.111904761904762e-06,
3057
+ "loss": 0.0043,
3058
+ "step": 424
3059
+ },
3060
+ {
3061
+ "epoch": 9.96,
3062
+ "grad_norm": 95.51915740966797,
3063
+ "learning_rate": 4.108630952380953e-06,
3064
+ "loss": 0.0498,
3065
+ "step": 425
3066
+ },
3067
+ {
3068
+ "epoch": 9.98,
3069
+ "grad_norm": 0.3881193995475769,
3070
+ "learning_rate": 4.1053571428571426e-06,
3071
+ "loss": 0.0037,
3072
+ "step": 426
3073
+ },
3074
+ {
3075
+ "epoch": 9.98,
3076
+ "eval_accuracy": 0.8636363636363636,
3077
+ "eval_loss": 0.6795494556427002,
3078
+ "eval_runtime": 1.9632,
3079
+ "eval_samples_per_second": 112.063,
3080
+ "eval_steps_per_second": 28.016,
3081
+ "step": 426
3082
+ },
3083
+ {
3084
+ "epoch": 10.01,
3085
+ "grad_norm": 117.20142364501953,
3086
+ "learning_rate": 4.102083333333333e-06,
3087
+ "loss": 0.0613,
3088
+ "step": 427
3089
+ },
3090
+ {
3091
+ "epoch": 10.03,
3092
+ "grad_norm": 0.3005952537059784,
3093
+ "learning_rate": 4.098809523809524e-06,
3094
+ "loss": 0.0036,
3095
+ "step": 428
3096
+ },
3097
+ {
3098
+ "epoch": 10.05,
3099
+ "grad_norm": 0.22291749715805054,
3100
+ "learning_rate": 4.095535714285714e-06,
3101
+ "loss": 0.004,
3102
+ "step": 429
3103
+ },
3104
+ {
3105
+ "epoch": 10.08,
3106
+ "grad_norm": 0.16808579862117767,
3107
+ "learning_rate": 4.092261904761905e-06,
3108
+ "loss": 0.0035,
3109
+ "step": 430
3110
+ },
3111
+ {
3112
+ "epoch": 10.1,
3113
+ "grad_norm": 0.153959259390831,
3114
+ "learning_rate": 4.088988095238095e-06,
3115
+ "loss": 0.0036,
3116
+ "step": 431
3117
+ },
3118
+ {
3119
+ "epoch": 10.12,
3120
+ "grad_norm": 0.32211458683013916,
3121
+ "learning_rate": 4.085714285714286e-06,
3122
+ "loss": 0.0038,
3123
+ "step": 432
3124
+ },
3125
+ {
3126
+ "epoch": 10.15,
3127
+ "grad_norm": 1.331497311592102,
3128
+ "learning_rate": 4.0824404761904765e-06,
3129
+ "loss": 0.0044,
3130
+ "step": 433
3131
+ },
3132
+ {
3133
+ "epoch": 10.17,
3134
+ "grad_norm": 0.26392656564712524,
3135
+ "learning_rate": 4.0791666666666664e-06,
3136
+ "loss": 0.0032,
3137
+ "step": 434
3138
+ },
3139
+ {
3140
+ "epoch": 10.2,
3141
+ "grad_norm": 0.24153968691825867,
3142
+ "learning_rate": 4.075892857142857e-06,
3143
+ "loss": 0.0029,
3144
+ "step": 435
3145
+ },
3146
+ {
3147
+ "epoch": 10.22,
3148
+ "grad_norm": 52.36206817626953,
3149
+ "learning_rate": 4.072619047619048e-06,
3150
+ "loss": 0.0645,
3151
+ "step": 436
3152
+ },
3153
+ {
3154
+ "epoch": 10.24,
3155
+ "grad_norm": 0.12544922530651093,
3156
+ "learning_rate": 4.069345238095238e-06,
3157
+ "loss": 0.0025,
3158
+ "step": 437
3159
+ },
3160
+ {
3161
+ "epoch": 10.27,
3162
+ "grad_norm": 0.14890314638614655,
3163
+ "learning_rate": 4.066071428571429e-06,
3164
+ "loss": 0.0024,
3165
+ "step": 438
3166
+ },
3167
+ {
3168
+ "epoch": 10.29,
3169
+ "grad_norm": 0.050663262605667114,
3170
+ "learning_rate": 4.06279761904762e-06,
3171
+ "loss": 0.002,
3172
+ "step": 439
3173
+ },
3174
+ {
3175
+ "epoch": 10.31,
3176
+ "grad_norm": 0.07523107528686523,
3177
+ "learning_rate": 4.0595238095238095e-06,
3178
+ "loss": 0.0023,
3179
+ "step": 440
3180
+ },
3181
+ {
3182
+ "epoch": 10.34,
3183
+ "grad_norm": 0.04829227551817894,
3184
+ "learning_rate": 4.05625e-06,
3185
+ "loss": 0.0017,
3186
+ "step": 441
3187
+ },
3188
+ {
3189
+ "epoch": 10.36,
3190
+ "grad_norm": 0.0691770538687706,
3191
+ "learning_rate": 4.05297619047619e-06,
3192
+ "loss": 0.0023,
3193
+ "step": 442
3194
+ },
3195
+ {
3196
+ "epoch": 10.38,
3197
+ "grad_norm": 0.06425254791975021,
3198
+ "learning_rate": 4.049702380952381e-06,
3199
+ "loss": 0.002,
3200
+ "step": 443
3201
+ },
3202
+ {
3203
+ "epoch": 10.41,
3204
+ "grad_norm": 0.061768341809511185,
3205
+ "learning_rate": 4.046428571428572e-06,
3206
+ "loss": 0.0021,
3207
+ "step": 444
3208
+ },
3209
+ {
3210
+ "epoch": 10.43,
3211
+ "grad_norm": 2.5151658058166504,
3212
+ "learning_rate": 4.043154761904762e-06,
3213
+ "loss": 0.1136,
3214
+ "step": 445
3215
+ },
3216
+ {
3217
+ "epoch": 10.45,
3218
+ "grad_norm": 0.042002830654382706,
3219
+ "learning_rate": 4.039880952380953e-06,
3220
+ "loss": 0.0016,
3221
+ "step": 446
3222
+ },
3223
+ {
3224
+ "epoch": 10.48,
3225
+ "grad_norm": 0.064276322722435,
3226
+ "learning_rate": 4.036607142857143e-06,
3227
+ "loss": 0.0022,
3228
+ "step": 447
3229
+ },
3230
+ {
3231
+ "epoch": 10.5,
3232
+ "grad_norm": 0.06543342024087906,
3233
+ "learning_rate": 4.0333333333333326e-06,
3234
+ "loss": 0.0018,
3235
+ "step": 448
3236
+ },
3237
+ {
3238
+ "epoch": 10.52,
3239
+ "grad_norm": 0.04980894550681114,
3240
+ "learning_rate": 4.030059523809523e-06,
3241
+ "loss": 0.0017,
3242
+ "step": 449
3243
+ },
3244
+ {
3245
+ "epoch": 10.55,
3246
+ "grad_norm": 0.11392680555582047,
3247
+ "learning_rate": 4.026785714285714e-06,
3248
+ "loss": 0.0024,
3249
+ "step": 450
3250
+ },
3251
+ {
3252
+ "epoch": 10.57,
3253
+ "grad_norm": 0.15568284690380096,
3254
+ "learning_rate": 4.023511904761904e-06,
3255
+ "loss": 0.0021,
3256
+ "step": 451
3257
+ },
3258
+ {
3259
+ "epoch": 10.59,
3260
+ "grad_norm": 0.05342809110879898,
3261
+ "learning_rate": 4.020238095238095e-06,
3262
+ "loss": 0.0018,
3263
+ "step": 452
3264
+ },
3265
+ {
3266
+ "epoch": 10.62,
3267
+ "grad_norm": 9.609216690063477,
3268
+ "learning_rate": 4.016964285714286e-06,
3269
+ "loss": 0.1925,
3270
+ "step": 453
3271
+ },
3272
+ {
3273
+ "epoch": 10.64,
3274
+ "grad_norm": 18.55946922302246,
3275
+ "learning_rate": 4.013690476190476e-06,
3276
+ "loss": 0.0191,
3277
+ "step": 454
3278
+ },
3279
+ {
3280
+ "epoch": 10.66,
3281
+ "grad_norm": 1.1658024787902832,
3282
+ "learning_rate": 4.0104166666666665e-06,
3283
+ "loss": 0.0035,
3284
+ "step": 455
3285
+ },
3286
+ {
3287
+ "epoch": 10.69,
3288
+ "grad_norm": 0.062038108706474304,
3289
+ "learning_rate": 4.007142857142857e-06,
3290
+ "loss": 0.0017,
3291
+ "step": 456
3292
+ },
3293
+ {
3294
+ "epoch": 10.71,
3295
+ "grad_norm": 0.1960655152797699,
3296
+ "learning_rate": 4.003869047619047e-06,
3297
+ "loss": 0.0018,
3298
+ "step": 457
3299
+ },
3300
+ {
3301
+ "epoch": 10.73,
3302
+ "grad_norm": 0.9941339492797852,
3303
+ "learning_rate": 4.000595238095238e-06,
3304
+ "loss": 0.0802,
3305
+ "step": 458
3306
+ },
3307
+ {
3308
+ "epoch": 10.76,
3309
+ "grad_norm": 0.2770668864250183,
3310
+ "learning_rate": 3.997321428571428e-06,
3311
+ "loss": 0.002,
3312
+ "step": 459
3313
+ },
3314
+ {
3315
+ "epoch": 10.78,
3316
+ "grad_norm": 0.07205989211797714,
3317
+ "learning_rate": 3.994047619047619e-06,
3318
+ "loss": 0.0018,
3319
+ "step": 460
3320
+ },
3321
+ {
3322
+ "epoch": 10.8,
3323
+ "grad_norm": 0.17621174454689026,
3324
+ "learning_rate": 3.9907738095238096e-06,
3325
+ "loss": 0.0026,
3326
+ "step": 461
3327
+ },
3328
+ {
3329
+ "epoch": 10.83,
3330
+ "grad_norm": 0.6454585790634155,
3331
+ "learning_rate": 3.9874999999999995e-06,
3332
+ "loss": 0.0023,
3333
+ "step": 462
3334
+ },
3335
+ {
3336
+ "epoch": 10.85,
3337
+ "grad_norm": 0.9011671543121338,
3338
+ "learning_rate": 3.98422619047619e-06,
3339
+ "loss": 0.0775,
3340
+ "step": 463
3341
+ },
3342
+ {
3343
+ "epoch": 10.88,
3344
+ "grad_norm": 0.19559772312641144,
3345
+ "learning_rate": 3.980952380952381e-06,
3346
+ "loss": 0.0018,
3347
+ "step": 464
3348
+ },
3349
+ {
3350
+ "epoch": 10.9,
3351
+ "grad_norm": 0.06330468505620956,
3352
+ "learning_rate": 3.977678571428571e-06,
3353
+ "loss": 0.0016,
3354
+ "step": 465
3355
+ },
3356
+ {
3357
+ "epoch": 10.92,
3358
+ "grad_norm": 0.05253860726952553,
3359
+ "learning_rate": 3.974404761904762e-06,
3360
+ "loss": 0.0015,
3361
+ "step": 466
3362
+ },
3363
+ {
3364
+ "epoch": 10.95,
3365
+ "grad_norm": 0.16123701632022858,
3366
+ "learning_rate": 3.971130952380952e-06,
3367
+ "loss": 0.0019,
3368
+ "step": 467
3369
+ },
3370
+ {
3371
+ "epoch": 10.97,
3372
+ "grad_norm": 2.395909070968628,
3373
+ "learning_rate": 3.967857142857143e-06,
3374
+ "loss": 0.2863,
3375
+ "step": 468
3376
+ },
3377
+ {
3378
+ "epoch": 10.99,
3379
+ "grad_norm": 2.7706680297851562,
3380
+ "learning_rate": 3.964583333333333e-06,
3381
+ "loss": 0.2554,
3382
+ "step": 469
3383
+ },
3384
+ {
3385
+ "epoch": 10.99,
3386
+ "eval_accuracy": 0.8590909090909091,
3387
+ "eval_loss": 0.861154317855835,
3388
+ "eval_runtime": 1.9879,
3389
+ "eval_samples_per_second": 110.668,
3390
+ "eval_steps_per_second": 27.667,
3391
+ "step": 469
3392
+ },
3393
+ {
3394
+ "epoch": 11.02,
3395
+ "grad_norm": 0.11244534701108932,
3396
+ "learning_rate": 3.961309523809523e-06,
3397
+ "loss": 0.0019,
3398
+ "step": 470
3399
+ },
3400
+ {
3401
+ "epoch": 11.04,
3402
+ "grad_norm": 0.27306148409843445,
3403
+ "learning_rate": 3.958035714285714e-06,
3404
+ "loss": 0.0025,
3405
+ "step": 471
3406
+ },
3407
+ {
3408
+ "epoch": 11.06,
3409
+ "grad_norm": 0.06686495989561081,
3410
+ "learning_rate": 3.954761904761905e-06,
3411
+ "loss": 0.0017,
3412
+ "step": 472
3413
+ },
3414
+ {
3415
+ "epoch": 11.09,
3416
+ "grad_norm": 0.07426581531763077,
3417
+ "learning_rate": 3.951488095238095e-06,
3418
+ "loss": 0.002,
3419
+ "step": 473
3420
+ },
3421
+ {
3422
+ "epoch": 11.11,
3423
+ "grad_norm": 0.06276604533195496,
3424
+ "learning_rate": 3.948214285714286e-06,
3425
+ "loss": 0.0017,
3426
+ "step": 474
3427
+ },
3428
+ {
3429
+ "epoch": 11.13,
3430
+ "grad_norm": 114.83184051513672,
3431
+ "learning_rate": 3.9449404761904765e-06,
3432
+ "loss": 0.0446,
3433
+ "step": 475
3434
+ },
3435
+ {
3436
+ "epoch": 11.16,
3437
+ "grad_norm": 0.07929594069719315,
3438
+ "learning_rate": 3.9416666666666665e-06,
3439
+ "loss": 0.0019,
3440
+ "step": 476
3441
+ },
3442
+ {
3443
+ "epoch": 11.18,
3444
+ "grad_norm": 3.3116915225982666,
3445
+ "learning_rate": 3.938392857142857e-06,
3446
+ "loss": 0.2205,
3447
+ "step": 477
3448
+ },
3449
+ {
3450
+ "epoch": 11.2,
3451
+ "grad_norm": 0.2510410249233246,
3452
+ "learning_rate": 3.935119047619047e-06,
3453
+ "loss": 0.0028,
3454
+ "step": 478
3455
+ },
3456
+ {
3457
+ "epoch": 11.23,
3458
+ "grad_norm": 0.11125629395246506,
3459
+ "learning_rate": 3.931845238095238e-06,
3460
+ "loss": 0.0021,
3461
+ "step": 479
3462
+ },
3463
+ {
3464
+ "epoch": 11.25,
3465
+ "grad_norm": 0.33865758776664734,
3466
+ "learning_rate": 3.928571428571429e-06,
3467
+ "loss": 0.0036,
3468
+ "step": 480
3469
+ },
3470
+ {
3471
+ "epoch": 11.27,
3472
+ "grad_norm": 1.4806885719299316,
3473
+ "learning_rate": 3.925297619047619e-06,
3474
+ "loss": 0.0062,
3475
+ "step": 481
3476
+ },
3477
+ {
3478
+ "epoch": 11.3,
3479
+ "grad_norm": 0.4448486268520355,
3480
+ "learning_rate": 3.92202380952381e-06,
3481
+ "loss": 0.0026,
3482
+ "step": 482
3483
+ },
3484
+ {
3485
+ "epoch": 11.32,
3486
+ "grad_norm": 3.580939769744873,
3487
+ "learning_rate": 3.91875e-06,
3488
+ "loss": 0.1848,
3489
+ "step": 483
3490
+ },
3491
+ {
3492
+ "epoch": 11.34,
3493
+ "grad_norm": 0.07441940158605576,
3494
+ "learning_rate": 3.91547619047619e-06,
3495
+ "loss": 0.0019,
3496
+ "step": 484
3497
+ },
3498
+ {
3499
+ "epoch": 11.37,
3500
+ "grad_norm": 0.05166240781545639,
3501
+ "learning_rate": 3.912202380952381e-06,
3502
+ "loss": 0.0017,
3503
+ "step": 485
3504
+ },
3505
+ {
3506
+ "epoch": 11.39,
3507
+ "grad_norm": 0.08746666461229324,
3508
+ "learning_rate": 3.908928571428571e-06,
3509
+ "loss": 0.0021,
3510
+ "step": 486
3511
+ },
3512
+ {
3513
+ "epoch": 11.41,
3514
+ "grad_norm": 0.2311132699251175,
3515
+ "learning_rate": 3.905654761904762e-06,
3516
+ "loss": 0.0029,
3517
+ "step": 487
3518
+ },
3519
+ {
3520
+ "epoch": 11.44,
3521
+ "grad_norm": 0.10750468820333481,
3522
+ "learning_rate": 3.902380952380953e-06,
3523
+ "loss": 0.002,
3524
+ "step": 488
3525
+ },
3526
+ {
3527
+ "epoch": 11.46,
3528
+ "grad_norm": 0.29891398549079895,
3529
+ "learning_rate": 3.899107142857143e-06,
3530
+ "loss": 0.003,
3531
+ "step": 489
3532
+ },
3533
+ {
3534
+ "epoch": 11.48,
3535
+ "grad_norm": 0.13320636749267578,
3536
+ "learning_rate": 3.8958333333333334e-06,
3537
+ "loss": 0.0025,
3538
+ "step": 490
3539
+ },
3540
+ {
3541
+ "epoch": 11.51,
3542
+ "grad_norm": 0.11678657680749893,
3543
+ "learning_rate": 3.892559523809524e-06,
3544
+ "loss": 0.0028,
3545
+ "step": 491
3546
+ },
3547
+ {
3548
+ "epoch": 11.53,
3549
+ "grad_norm": 0.15591733157634735,
3550
+ "learning_rate": 3.889285714285714e-06,
3551
+ "loss": 0.0021,
3552
+ "step": 492
3553
+ },
3554
+ {
3555
+ "epoch": 11.55,
3556
+ "grad_norm": 35.02848434448242,
3557
+ "learning_rate": 3.886011904761905e-06,
3558
+ "loss": 0.0162,
3559
+ "step": 493
3560
+ },
3561
+ {
3562
+ "epoch": 11.58,
3563
+ "grad_norm": 14.46584415435791,
3564
+ "learning_rate": 3.882738095238096e-06,
3565
+ "loss": 0.0833,
3566
+ "step": 494
3567
+ },
3568
+ {
3569
+ "epoch": 11.6,
3570
+ "grad_norm": 0.19305245578289032,
3571
+ "learning_rate": 3.879464285714286e-06,
3572
+ "loss": 0.002,
3573
+ "step": 495
3574
+ },
3575
+ {
3576
+ "epoch": 11.62,
3577
+ "grad_norm": 0.814598023891449,
3578
+ "learning_rate": 3.8761904761904765e-06,
3579
+ "loss": 0.0697,
3580
+ "step": 496
3581
+ },
3582
+ {
3583
+ "epoch": 11.65,
3584
+ "grad_norm": 0.03925960883498192,
3585
+ "learning_rate": 3.8729166666666665e-06,
3586
+ "loss": 0.0015,
3587
+ "step": 497
3588
+ },
3589
+ {
3590
+ "epoch": 11.67,
3591
+ "grad_norm": 0.07029083371162415,
3592
+ "learning_rate": 3.869642857142857e-06,
3593
+ "loss": 0.0017,
3594
+ "step": 498
3595
+ },
3596
+ {
3597
+ "epoch": 11.7,
3598
+ "grad_norm": 0.048684656620025635,
3599
+ "learning_rate": 3.866369047619047e-06,
3600
+ "loss": 0.0016,
3601
+ "step": 499
3602
+ },
3603
+ {
3604
+ "epoch": 11.72,
3605
+ "grad_norm": 0.07016553729772568,
3606
+ "learning_rate": 3.863095238095238e-06,
3607
+ "loss": 0.002,
3608
+ "step": 500
3609
+ },
3610
+ {
3611
+ "epoch": 11.74,
3612
+ "grad_norm": 0.10889595001935959,
3613
+ "learning_rate": 3.859821428571428e-06,
3614
+ "loss": 0.0018,
3615
+ "step": 501
3616
+ },
3617
+ {
3618
+ "epoch": 11.77,
3619
+ "grad_norm": 0.04481940343976021,
3620
+ "learning_rate": 3.856547619047619e-06,
3621
+ "loss": 0.0015,
3622
+ "step": 502
3623
+ },
3624
+ {
3625
+ "epoch": 11.79,
3626
+ "grad_norm": 0.05054694786667824,
3627
+ "learning_rate": 3.853273809523809e-06,
3628
+ "loss": 0.0016,
3629
+ "step": 503
3630
+ },
3631
+ {
3632
+ "epoch": 11.81,
3633
+ "grad_norm": 0.07786347717046738,
3634
+ "learning_rate": 3.8499999999999996e-06,
3635
+ "loss": 0.0015,
3636
+ "step": 504
3637
+ },
3638
+ {
3639
+ "epoch": 11.84,
3640
+ "grad_norm": 0.10747917741537094,
3641
+ "learning_rate": 3.84672619047619e-06,
3642
+ "loss": 0.002,
3643
+ "step": 505
3644
+ },
3645
+ {
3646
+ "epoch": 11.86,
3647
+ "grad_norm": 3.685096263885498,
3648
+ "learning_rate": 3.84345238095238e-06,
3649
+ "loss": 0.0047,
3650
+ "step": 506
3651
+ },
3652
+ {
3653
+ "epoch": 11.88,
3654
+ "grad_norm": 0.8624734878540039,
3655
+ "learning_rate": 3.840178571428571e-06,
3656
+ "loss": 0.0707,
3657
+ "step": 507
3658
+ },
3659
+ {
3660
+ "epoch": 11.91,
3661
+ "grad_norm": 0.03694400563836098,
3662
+ "learning_rate": 3.836904761904762e-06,
3663
+ "loss": 0.0014,
3664
+ "step": 508
3665
+ },
3666
+ {
3667
+ "epoch": 11.93,
3668
+ "grad_norm": 0.08262762427330017,
3669
+ "learning_rate": 3.833630952380952e-06,
3670
+ "loss": 0.0015,
3671
+ "step": 509
3672
+ },
3673
+ {
3674
+ "epoch": 11.95,
3675
+ "grad_norm": 0.06174321845173836,
3676
+ "learning_rate": 3.830357142857143e-06,
3677
+ "loss": 0.0014,
3678
+ "step": 510
3679
+ },
3680
+ {
3681
+ "epoch": 11.98,
3682
+ "grad_norm": 0.04213571548461914,
3683
+ "learning_rate": 3.8270833333333335e-06,
3684
+ "loss": 0.0013,
3685
+ "step": 511
3686
+ },
3687
+ {
3688
+ "epoch": 12.0,
3689
+ "grad_norm": 4.956207275390625,
3690
+ "learning_rate": 3.823809523809523e-06,
3691
+ "loss": 0.2121,
3692
+ "step": 512
3693
+ },
3694
+ {
3695
+ "epoch": 12.0,
3696
+ "eval_accuracy": 0.8909090909090909,
3697
+ "eval_loss": 0.7056229114532471,
3698
+ "eval_runtime": 1.9577,
3699
+ "eval_samples_per_second": 112.375,
3700
+ "eval_steps_per_second": 28.094,
3701
+ "step": 512
3702
+ }
3703
+ ],
3704
+ "logging_steps": 1,
3705
+ "max_steps": 1680,
3706
+ "num_input_tokens_seen": 0,
3707
+ "num_train_epochs": 40,
3708
+ "save_steps": 500,
3709
+ "total_flos": 5202308973417984.0,
3710
+ "train_batch_size": 4,
3711
+ "trial_name": null,
3712
+ "trial_params": null
3713
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:88617d7ded71106a0f4d94ff5806dd9d0a807ed7e928b7186f4d27acbfce2a36
3
+ size 4920