Hugging Face's logo

nntoan209
/
bgem3-sft-msmarco-squadv2-e1

Sentence Similarity
sentence-transformers
Safetensors
xlm-roberta
feature-extraction
Inference Endpoints
text-embeddings-inference
Model card Files Community
nntoan209 commited on
Commit
989fa49
•
1 Parent(s): df21bfa

Upload folder using huggingface_hub

Browse files
Files changed (20) hide show
  1. .gitattributes +1 -0
  2. 1_Pooling/config.json +10 -0
  3. README.md +57 -0
  4. colbert_linear.pt +3 -0
  5. config.json +28 -0
  6. config_sentence_transformers.json +9 -0
  7. latest +1 -0
  8. model.safetensors +3 -0
  9. modules.json +20 -0
  10. rng_state_0.pth +3 -0
  11. rng_state_1.pth +3 -0
  12. sentence_bert_config.json +4 -0
  13. sentencepiece.bpe.model +3 -0
  14. sparse_linear.pt +3 -0
  15. special_tokens_map.json +51 -0
  16. tokenizer.json +3 -0
  17. tokenizer_config.json +55 -0
  18. trainer_state.json +3192 -0
  19. training_args.bin +3 -0
  20. zero_to_fp32.py +592 -0
.gitattributes CHANGED
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ tokenizer.json filter=lfs diff=lfs merge=lfs -text
1_Pooling/config.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "word_embedding_dimension": 1024,
3
+ "pooling_mode_cls_token": true,
4
+ "pooling_mode_mean_tokens": false,
5
+ "pooling_mode_max_tokens": false,
6
+ "pooling_mode_mean_sqrt_len_tokens": false,
7
+ "pooling_mode_weightedmean_tokens": false,
8
+ "pooling_mode_lasttoken": false,
9
+ "include_prompt": true
10
+ }
README.md ADDED
@@ -0,0 +1,57 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: sentence-transformers
3
+ pipeline_tag: sentence-similarity
4
+ tags:
5
+ - sentence-transformers
6
+ - feature-extraction
7
+ - sentence-similarity
8
+
9
+ ---
10
+
11
+ # {MODEL_NAME}
12
+
13
+ This is a [sentence-transformers](https://www.SBERT.net) model: It maps sentences & paragraphs to a 1024 dimensional dense vector space and can be used for tasks like clustering or semantic search.
14
+
15
+ <!--- Describe your model here -->
16
+
17
+ ## Usage (Sentence-Transformers)
18
+
19
+ Using this model becomes easy when you have [sentence-transformers](https://www.SBERT.net) installed:
20
+
21
+ ```
22
+ pip install -U sentence-transformers
23
+ ```
24
+
25
+ Then you can use the model like this:
26
+
27
+ ```python
28
+ from sentence_transformers import SentenceTransformer
29
+ sentences = ["This is an example sentence", "Each sentence is converted"]
30
+
31
+ model = SentenceTransformer('{MODEL_NAME}')
32
+ embeddings = model.encode(sentences)
33
+ print(embeddings)
34
+ ```
35
+
36
+
37
+
38
+ ## Evaluation Results
39
+
40
+ <!--- Describe how your model was evaluated -->
41
+
42
+ For an automated evaluation of this model, see the *Sentence Embeddings Benchmark*: [https://seb.sbert.net](https://seb.sbert.net?model_name={MODEL_NAME})
43
+
44
+
45
+
46
+ ## Full Model Architecture
47
+ ```
48
+ SentenceTransformer(
49
+ (0): Transformer({'max_seq_length': 8192, 'do_lower_case': False}) with Transformer model: XLMRobertaModel
50
+ (1): Pooling({'word_embedding_dimension': 1024, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})
51
+ (2): Normalize()
52
+ )
53
+ ```
54
+
55
+ ## Citing & Authors
56
+
57
+ <!--- Describe where people can find more information -->
colbert_linear.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:07fb4c11dbf7a594afafb841e4c0f2eabd9a0a44679b5d64407c038798e5cee7
3
+ size 2100674
config.json ADDED
@@ -0,0 +1,28 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "saved_models/bgem3_sft_20240528/tmp-checkpoint-4532",
3
+ "architectures": [
4
+ "XLMRobertaModel"
5
+ ],
6
+ "attention_probs_dropout_prob": 0.1,
7
+ "bos_token_id": 0,
8
+ "classifier_dropout": null,
9
+ "eos_token_id": 2,
10
+ "hidden_act": "gelu",
11
+ "hidden_dropout_prob": 0.1,
12
+ "hidden_size": 1024,
13
+ "initializer_range": 0.02,
14
+ "intermediate_size": 4096,
15
+ "layer_norm_eps": 1e-05,
16
+ "max_position_embeddings": 8194,
17
+ "model_type": "xlm-roberta",
18
+ "num_attention_heads": 16,
19
+ "num_hidden_layers": 24,
20
+ "output_past": true,
21
+ "pad_token_id": 1,
22
+ "position_embedding_type": "absolute",
23
+ "torch_dtype": "float32",
24
+ "transformers_version": "4.38.2",
25
+ "type_vocab_size": 1,
26
+ "use_cache": true,
27
+ "vocab_size": 250002
28
+ }
config_sentence_transformers.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "__version__": {
3
+ "sentence_transformers": "2.5.1",
4
+ "transformers": "4.38.2",
5
+ "pytorch": "2.1.0+cu121"
6
+ },
7
+ "prompts": {},
8
+ "default_prompt_name": null
9
+ }
latest ADDED
@@ -0,0 +1 @@
 
 
1
+ global_step4532
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:67be58be92274efa45c9e04c701e70f27270a8f02515bf18805994a96963f7f2
3
+ size 2271064456
modules.json ADDED
@@ -0,0 +1,20 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ [
2
+ {
3
+ "idx": 0,
4
+ "name": "0",
5
+ "path": "",
6
+ "type": "sentence_transformers.models.Transformer"
7
+ },
8
+ {
9
+ "idx": 1,
10
+ "name": "1",
11
+ "path": "1_Pooling",
12
+ "type": "sentence_transformers.models.Pooling"
13
+ },
14
+ {
15
+ "idx": 2,
16
+ "name": "2",
17
+ "path": "2_Normalize",
18
+ "type": "sentence_transformers.models.Normalize"
19
+ }
20
+ ]
rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fd17d388b5d19dbff8498fb5d5b9492821580b117eff968d193690cb88c3941b
3
+ size 14512
rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9b22f2ba964ee6780fa354c19ec2585c8483517d67068144dd1ac1c860817236
3
+ size 14512
sentence_bert_config.json ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ {
2
+ "max_seq_length": 8192,
3
+ "do_lower_case": false
4
+ }
sentencepiece.bpe.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cfc8146abe2a0488e9e2a0c56de7952f7c11ab059eca145a0a727afce0db2865
3
+ size 5069051
sparse_linear.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3edb1e33b5841b67152515c0d2bbdaec52e10f1c61faa4139978e0df8fa04d37
3
+ size 3516
special_tokens_map.json ADDED
@@ -0,0 +1,51 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "cls_token": {
10
+ "content": "<s>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "eos_token": {
17
+ "content": "</s>",
18
+ "lstrip": false,
19
+ "normalized": false,
20
+ "rstrip": false,
21
+ "single_word": false
22
+ },
23
+ "mask_token": {
24
+ "content": "<mask>",
25
+ "lstrip": true,
26
+ "normalized": false,
27
+ "rstrip": false,
28
+ "single_word": false
29
+ },
30
+ "pad_token": {
31
+ "content": "<pad>",
32
+ "lstrip": false,
33
+ "normalized": false,
34
+ "rstrip": false,
35
+ "single_word": false
36
+ },
37
+ "sep_token": {
38
+ "content": "</s>",
39
+ "lstrip": false,
40
+ "normalized": false,
41
+ "rstrip": false,
42
+ "single_word": false
43
+ },
44
+ "unk_token": {
45
+ "content": "<unk>",
46
+ "lstrip": false,
47
+ "normalized": false,
48
+ "rstrip": false,
49
+ "single_word": false
50
+ }
51
+ }
tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:69564b696052886ed0ac63fa393e928384e0f8caada38c1f4864a9bfbf379c15
3
+ size 17098273
tokenizer_config.json ADDED
@@ -0,0 +1,55 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "added_tokens_decoder": {
3
+ "0": {
4
+ "content": "<s>",
5
+ "lstrip": false,
6
+ "normalized": false,
7
+ "rstrip": false,
8
+ "single_word": false,
9
+ "special": true
10
+ },
11
+ "1": {
12
+ "content": "<pad>",
13
+ "lstrip": false,
14
+ "normalized": false,
15
+ "rstrip": false,
16
+ "single_word": false,
17
+ "special": true
18
+ },
19
+ "2": {
20
+ "content": "</s>",
21
+ "lstrip": false,
22
+ "normalized": false,
23
+ "rstrip": false,
24
+ "single_word": false,
25
+ "special": true
26
+ },
27
+ "3": {
28
+ "content": "<unk>",
29
+ "lstrip": false,
30
+ "normalized": false,
31
+ "rstrip": false,
32
+ "single_word": false,
33
+ "special": true
34
+ },
35
+ "250001": {
36
+ "content": "<mask>",
37
+ "lstrip": true,
38
+ "normalized": false,
39
+ "rstrip": false,
40
+ "single_word": false,
41
+ "special": true
42
+ }
43
+ },
44
+ "bos_token": "<s>",
45
+ "clean_up_tokenization_spaces": true,
46
+ "cls_token": "<s>",
47
+ "eos_token": "</s>",
48
+ "mask_token": "<mask>",
49
+ "model_max_length": 8192,
50
+ "pad_token": "<pad>",
51
+ "sep_token": "</s>",
52
+ "sp_model_kwargs": {},
53
+ "tokenizer_class": "XLMRobertaTokenizer",
54
+ "unk_token": "<unk>"
55
+ }
trainer_state.json ADDED
@@ -0,0 +1,3192 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 0.9181523500810372,
5
+ "eval_steps": 500,
6
+ "global_step": 4532,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.0,
13
+ "grad_norm": 10.73312117404433,
14
+ "learning_rate": 4.4048582995951427e-07,
15
+ "loss": 0.5001,
16
+ "step": 10
17
+ },
18
+ {
19
+ "epoch": 0.0,
20
+ "grad_norm": 10.320516533751473,
21
+ "learning_rate": 7.076923076923077e-07,
22
+ "loss": 0.4459,
23
+ "step": 20
24
+ },
25
+ {
26
+ "epoch": 0.01,
27
+ "grad_norm": 10.604032764201449,
28
+ "learning_rate": 9.748987854251014e-07,
29
+ "loss": 0.5337,
30
+ "step": 30
31
+ },
32
+ {
33
+ "epoch": 0.01,
34
+ "grad_norm": 9.501814819458504,
35
+ "learning_rate": 1.2421052631578948e-06,
36
+ "loss": 0.4867,
37
+ "step": 40
38
+ },
39
+ {
40
+ "epoch": 0.01,
41
+ "grad_norm": 7.2055395221098815,
42
+ "learning_rate": 1.5093117408906883e-06,
43
+ "loss": 0.4851,
44
+ "step": 50
45
+ },
46
+ {
47
+ "epoch": 0.01,
48
+ "grad_norm": 5.16903562483095,
49
+ "learning_rate": 1.776518218623482e-06,
50
+ "loss": 0.4368,
51
+ "step": 60
52
+ },
53
+ {
54
+ "epoch": 0.01,
55
+ "grad_norm": 5.746924199634763,
56
+ "learning_rate": 2.0437246963562754e-06,
57
+ "loss": 0.4289,
58
+ "step": 70
59
+ },
60
+ {
61
+ "epoch": 0.02,
62
+ "grad_norm": 5.827693974405843,
63
+ "learning_rate": 2.3109311740890693e-06,
64
+ "loss": 0.3924,
65
+ "step": 80
66
+ },
67
+ {
68
+ "epoch": 0.02,
69
+ "grad_norm": 7.020982629413032,
70
+ "learning_rate": 2.5781376518218628e-06,
71
+ "loss": 0.4108,
72
+ "step": 90
73
+ },
74
+ {
75
+ "epoch": 0.02,
76
+ "grad_norm": 6.01679137180817,
77
+ "learning_rate": 2.8453441295546562e-06,
78
+ "loss": 0.4146,
79
+ "step": 100
80
+ },
81
+ {
82
+ "epoch": 0.02,
83
+ "grad_norm": 5.448653259468705,
84
+ "learning_rate": 3.11255060728745e-06,
85
+ "loss": 0.3843,
86
+ "step": 110
87
+ },
88
+ {
89
+ "epoch": 0.02,
90
+ "grad_norm": 4.431377216444548,
91
+ "learning_rate": 3.379757085020243e-06,
92
+ "loss": 0.3752,
93
+ "step": 120
94
+ },
95
+ {
96
+ "epoch": 0.03,
97
+ "grad_norm": 4.706847685087678,
98
+ "learning_rate": 3.646963562753037e-06,
99
+ "loss": 0.373,
100
+ "step": 130
101
+ },
102
+ {
103
+ "epoch": 0.03,
104
+ "grad_norm": 4.800217647386447,
105
+ "learning_rate": 3.9141700404858305e-06,
106
+ "loss": 0.3189,
107
+ "step": 140
108
+ },
109
+ {
110
+ "epoch": 0.03,
111
+ "grad_norm": 4.55540918616468,
112
+ "learning_rate": 4.1813765182186235e-06,
113
+ "loss": 0.3337,
114
+ "step": 150
115
+ },
116
+ {
117
+ "epoch": 0.03,
118
+ "grad_norm": 5.728866550792321,
119
+ "learning_rate": 4.448582995951417e-06,
120
+ "loss": 0.2853,
121
+ "step": 160
122
+ },
123
+ {
124
+ "epoch": 0.03,
125
+ "grad_norm": 4.866232005178406,
126
+ "learning_rate": 4.71578947368421e-06,
127
+ "loss": 0.335,
128
+ "step": 170
129
+ },
130
+ {
131
+ "epoch": 0.04,
132
+ "grad_norm": 5.153227503990654,
133
+ "learning_rate": 4.982995951417004e-06,
134
+ "loss": 0.3563,
135
+ "step": 180
136
+ },
137
+ {
138
+ "epoch": 0.04,
139
+ "grad_norm": 4.833149588589071,
140
+ "learning_rate": 5.250202429149799e-06,
141
+ "loss": 0.3363,
142
+ "step": 190
143
+ },
144
+ {
145
+ "epoch": 0.04,
146
+ "grad_norm": 5.057205257076979,
147
+ "learning_rate": 5.517408906882591e-06,
148
+ "loss": 0.3718,
149
+ "step": 200
150
+ },
151
+ {
152
+ "epoch": 0.04,
153
+ "grad_norm": 5.623057559852214,
154
+ "learning_rate": 5.784615384615385e-06,
155
+ "loss": 0.3425,
156
+ "step": 210
157
+ },
158
+ {
159
+ "epoch": 0.04,
160
+ "grad_norm": 5.285732195613963,
161
+ "learning_rate": 6.05182186234818e-06,
162
+ "loss": 0.3323,
163
+ "step": 220
164
+ },
165
+ {
166
+ "epoch": 0.05,
167
+ "grad_norm": 5.287687993596724,
168
+ "learning_rate": 6.319028340080971e-06,
169
+ "loss": 0.3509,
170
+ "step": 230
171
+ },
172
+ {
173
+ "epoch": 0.05,
174
+ "grad_norm": 5.422212177972575,
175
+ "learning_rate": 6.586234817813766e-06,
176
+ "loss": 0.3967,
177
+ "step": 240
178
+ },
179
+ {
180
+ "epoch": 0.05,
181
+ "grad_norm": 4.898901033752465,
182
+ "learning_rate": 6.853441295546559e-06,
183
+ "loss": 0.304,
184
+ "step": 250
185
+ },
186
+ {
187
+ "epoch": 0.05,
188
+ "grad_norm": 5.807565871352443,
189
+ "learning_rate": 7.120647773279354e-06,
190
+ "loss": 0.3138,
191
+ "step": 260
192
+ },
193
+ {
194
+ "epoch": 0.05,
195
+ "grad_norm": 4.706658034348249,
196
+ "learning_rate": 7.387854251012147e-06,
197
+ "loss": 0.2711,
198
+ "step": 270
199
+ },
200
+ {
201
+ "epoch": 0.06,
202
+ "grad_norm": 4.330996445669558,
203
+ "learning_rate": 7.65506072874494e-06,
204
+ "loss": 0.3013,
205
+ "step": 280
206
+ },
207
+ {
208
+ "epoch": 0.06,
209
+ "grad_norm": 5.040245023856746,
210
+ "learning_rate": 7.922267206477734e-06,
211
+ "loss": 0.2845,
212
+ "step": 290
213
+ },
214
+ {
215
+ "epoch": 0.06,
216
+ "grad_norm": 5.377102462850796,
217
+ "learning_rate": 8.189473684210527e-06,
218
+ "loss": 0.2274,
219
+ "step": 300
220
+ },
221
+ {
222
+ "epoch": 0.06,
223
+ "grad_norm": 5.49394013615042,
224
+ "learning_rate": 8.45668016194332e-06,
225
+ "loss": 0.3002,
226
+ "step": 310
227
+ },
228
+ {
229
+ "epoch": 0.06,
230
+ "grad_norm": 7.592214672466295,
231
+ "learning_rate": 8.723886639676115e-06,
232
+ "loss": 0.2683,
233
+ "step": 320
234
+ },
235
+ {
236
+ "epoch": 0.07,
237
+ "grad_norm": 5.640000073588486,
238
+ "learning_rate": 8.991093117408907e-06,
239
+ "loss": 0.2753,
240
+ "step": 330
241
+ },
242
+ {
243
+ "epoch": 0.07,
244
+ "grad_norm": 4.527692569068709,
245
+ "learning_rate": 9.258299595141701e-06,
246
+ "loss": 0.2799,
247
+ "step": 340
248
+ },
249
+ {
250
+ "epoch": 0.07,
251
+ "grad_norm": 4.146002735657809,
252
+ "learning_rate": 9.525506072874495e-06,
253
+ "loss": 0.2735,
254
+ "step": 350
255
+ },
256
+ {
257
+ "epoch": 0.07,
258
+ "grad_norm": 7.711163580304015,
259
+ "learning_rate": 9.792712550607289e-06,
260
+ "loss": 0.2735,
261
+ "step": 360
262
+ },
263
+ {
264
+ "epoch": 0.07,
265
+ "grad_norm": 6.7114456570561005,
266
+ "learning_rate": 1.0059919028340081e-05,
267
+ "loss": 0.299,
268
+ "step": 370
269
+ },
270
+ {
271
+ "epoch": 0.08,
272
+ "grad_norm": 9.166917647191482,
273
+ "learning_rate": 1.0327125506072877e-05,
274
+ "loss": 0.2633,
275
+ "step": 380
276
+ },
277
+ {
278
+ "epoch": 0.08,
279
+ "grad_norm": 4.804784956773296,
280
+ "learning_rate": 1.0594331983805667e-05,
281
+ "loss": 0.3049,
282
+ "step": 390
283
+ },
284
+ {
285
+ "epoch": 0.08,
286
+ "grad_norm": 5.407897698128889,
287
+ "learning_rate": 1.0861538461538461e-05,
288
+ "loss": 0.2665,
289
+ "step": 400
290
+ },
291
+ {
292
+ "epoch": 0.08,
293
+ "grad_norm": 9.363037208047256,
294
+ "learning_rate": 1.1128744939271257e-05,
295
+ "loss": 0.2891,
296
+ "step": 410
297
+ },
298
+ {
299
+ "epoch": 0.09,
300
+ "grad_norm": 7.40785296512954,
301
+ "learning_rate": 1.1395951417004049e-05,
302
+ "loss": 0.305,
303
+ "step": 420
304
+ },
305
+ {
306
+ "epoch": 0.09,
307
+ "grad_norm": 4.832043881349699,
308
+ "learning_rate": 1.1663157894736843e-05,
309
+ "loss": 0.2275,
310
+ "step": 430
311
+ },
312
+ {
313
+ "epoch": 0.09,
314
+ "grad_norm": 5.819742592169462,
315
+ "learning_rate": 1.1930364372469638e-05,
316
+ "loss": 0.2919,
317
+ "step": 440
318
+ },
319
+ {
320
+ "epoch": 0.09,
321
+ "grad_norm": 4.641758430345508,
322
+ "learning_rate": 1.2197570850202429e-05,
323
+ "loss": 0.2943,
324
+ "step": 450
325
+ },
326
+ {
327
+ "epoch": 0.09,
328
+ "grad_norm": 6.619813436316814,
329
+ "learning_rate": 1.2464777327935223e-05,
330
+ "loss": 0.2852,
331
+ "step": 460
332
+ },
333
+ {
334
+ "epoch": 0.1,
335
+ "grad_norm": 4.1511414761990375,
336
+ "learning_rate": 1.2731983805668018e-05,
337
+ "loss": 0.2886,
338
+ "step": 470
339
+ },
340
+ {
341
+ "epoch": 0.1,
342
+ "grad_norm": 13.864446251944102,
343
+ "learning_rate": 1.299919028340081e-05,
344
+ "loss": 0.2418,
345
+ "step": 480
346
+ },
347
+ {
348
+ "epoch": 0.1,
349
+ "grad_norm": 4.2334846039024985,
350
+ "learning_rate": 1.3266396761133604e-05,
351
+ "loss": 0.3023,
352
+ "step": 490
353
+ },
354
+ {
355
+ "epoch": 0.1,
356
+ "grad_norm": 8.049684733976,
357
+ "learning_rate": 1.3533603238866397e-05,
358
+ "loss": 0.2475,
359
+ "step": 500
360
+ },
361
+ {
362
+ "epoch": 0.1,
363
+ "grad_norm": 10.458696782052067,
364
+ "learning_rate": 1.3800809716599192e-05,
365
+ "loss": 0.25,
366
+ "step": 510
367
+ },
368
+ {
369
+ "epoch": 0.11,
370
+ "grad_norm": 4.63729749376232,
371
+ "learning_rate": 1.4068016194331984e-05,
372
+ "loss": 0.2871,
373
+ "step": 520
374
+ },
375
+ {
376
+ "epoch": 0.11,
377
+ "grad_norm": 7.406867021700095,
378
+ "learning_rate": 1.4335222672064776e-05,
379
+ "loss": 0.2932,
380
+ "step": 530
381
+ },
382
+ {
383
+ "epoch": 0.11,
384
+ "grad_norm": 5.89092417799605,
385
+ "learning_rate": 1.4602429149797572e-05,
386
+ "loss": 0.2841,
387
+ "step": 540
388
+ },
389
+ {
390
+ "epoch": 0.11,
391
+ "grad_norm": 4.784763615906999,
392
+ "learning_rate": 1.4869635627530366e-05,
393
+ "loss": 0.2583,
394
+ "step": 550
395
+ },
396
+ {
397
+ "epoch": 0.11,
398
+ "grad_norm": 5.425616026327031,
399
+ "learning_rate": 1.5136842105263158e-05,
400
+ "loss": 0.2633,
401
+ "step": 560
402
+ },
403
+ {
404
+ "epoch": 0.12,
405
+ "grad_norm": 3.972773037284401,
406
+ "learning_rate": 1.5404048582995954e-05,
407
+ "loss": 0.2528,
408
+ "step": 570
409
+ },
410
+ {
411
+ "epoch": 0.12,
412
+ "grad_norm": 4.221584018079482,
413
+ "learning_rate": 1.5671255060728746e-05,
414
+ "loss": 0.2633,
415
+ "step": 580
416
+ },
417
+ {
418
+ "epoch": 0.12,
419
+ "grad_norm": 7.132176048070046,
420
+ "learning_rate": 1.5938461538461538e-05,
421
+ "loss": 0.2136,
422
+ "step": 590
423
+ },
424
+ {
425
+ "epoch": 0.12,
426
+ "grad_norm": 5.736662281189121,
427
+ "learning_rate": 1.6205668016194334e-05,
428
+ "loss": 0.2639,
429
+ "step": 600
430
+ },
431
+ {
432
+ "epoch": 0.12,
433
+ "grad_norm": 4.001622347847798,
434
+ "learning_rate": 1.6472874493927126e-05,
435
+ "loss": 0.315,
436
+ "step": 610
437
+ },
438
+ {
439
+ "epoch": 0.13,
440
+ "grad_norm": 4.526524467163638,
441
+ "learning_rate": 1.674008097165992e-05,
442
+ "loss": 0.2424,
443
+ "step": 620
444
+ },
445
+ {
446
+ "epoch": 0.13,
447
+ "grad_norm": 9.626596293663294,
448
+ "learning_rate": 1.7007287449392717e-05,
449
+ "loss": 0.2417,
450
+ "step": 630
451
+ },
452
+ {
453
+ "epoch": 0.13,
454
+ "grad_norm": 4.2292844007418555,
455
+ "learning_rate": 1.727449392712551e-05,
456
+ "loss": 0.2534,
457
+ "step": 640
458
+ },
459
+ {
460
+ "epoch": 0.13,
461
+ "grad_norm": 4.808396608849778,
462
+ "learning_rate": 1.75417004048583e-05,
463
+ "loss": 0.2501,
464
+ "step": 650
465
+ },
466
+ {
467
+ "epoch": 0.13,
468
+ "grad_norm": 3.79257297513555,
469
+ "learning_rate": 1.7808906882591094e-05,
470
+ "loss": 0.2583,
471
+ "step": 660
472
+ },
473
+ {
474
+ "epoch": 0.14,
475
+ "grad_norm": 2.8750772880448703,
476
+ "learning_rate": 1.807611336032389e-05,
477
+ "loss": 0.2296,
478
+ "step": 670
479
+ },
480
+ {
481
+ "epoch": 0.14,
482
+ "grad_norm": 4.439196839762044,
483
+ "learning_rate": 1.834331983805668e-05,
484
+ "loss": 0.2439,
485
+ "step": 680
486
+ },
487
+ {
488
+ "epoch": 0.14,
489
+ "grad_norm": 4.3898290815513965,
490
+ "learning_rate": 1.8610526315789473e-05,
491
+ "loss": 0.2899,
492
+ "step": 690
493
+ },
494
+ {
495
+ "epoch": 0.14,
496
+ "grad_norm": 4.42001400664859,
497
+ "learning_rate": 1.887773279352227e-05,
498
+ "loss": 0.2616,
499
+ "step": 700
500
+ },
501
+ {
502
+ "epoch": 0.14,
503
+ "grad_norm": 5.324629277322316,
504
+ "learning_rate": 1.9144939271255065e-05,
505
+ "loss": 0.2547,
506
+ "step": 710
507
+ },
508
+ {
509
+ "epoch": 0.15,
510
+ "grad_norm": 5.181807370455961,
511
+ "learning_rate": 1.9412145748987857e-05,
512
+ "loss": 0.2746,
513
+ "step": 720
514
+ },
515
+ {
516
+ "epoch": 0.15,
517
+ "grad_norm": 6.195963917912233,
518
+ "learning_rate": 1.967935222672065e-05,
519
+ "loss": 0.2269,
520
+ "step": 730
521
+ },
522
+ {
523
+ "epoch": 0.15,
524
+ "grad_norm": 21.9010566073324,
525
+ "learning_rate": 1.994655870445344e-05,
526
+ "loss": 0.2214,
527
+ "step": 740
528
+ },
529
+ {
530
+ "epoch": 0.15,
531
+ "grad_norm": 8.220773890093973,
532
+ "learning_rate": 1.9999979802007072e-05,
533
+ "loss": 0.3237,
534
+ "step": 750
535
+ },
536
+ {
537
+ "epoch": 0.15,
538
+ "grad_norm": 7.431578258197571,
539
+ "learning_rate": 1.9999909981889357e-05,
540
+ "loss": 0.2585,
541
+ "step": 760
542
+ },
543
+ {
544
+ "epoch": 0.16,
545
+ "grad_norm": 5.277053700798239,
546
+ "learning_rate": 1.999979029063708e-05,
547
+ "loss": 0.2865,
548
+ "step": 770
549
+ },
550
+ {
551
+ "epoch": 0.16,
552
+ "grad_norm": 4.434866755224634,
553
+ "learning_rate": 1.9999620728847215e-05,
554
+ "loss": 0.2355,
555
+ "step": 780
556
+ },
557
+ {
558
+ "epoch": 0.16,
559
+ "grad_norm": 5.028960471152081,
560
+ "learning_rate": 1.9999401297365485e-05,
561
+ "loss": 0.2896,
562
+ "step": 790
563
+ },
564
+ {
565
+ "epoch": 0.16,
566
+ "grad_norm": 4.533763530364985,
567
+ "learning_rate": 1.999913199728633e-05,
568
+ "loss": 0.3033,
569
+ "step": 800
570
+ },
571
+ {
572
+ "epoch": 0.16,
573
+ "grad_norm": 3.554515792556175,
574
+ "learning_rate": 1.9998812829952933e-05,
575
+ "loss": 0.2617,
576
+ "step": 810
577
+ },
578
+ {
579
+ "epoch": 0.17,
580
+ "grad_norm": 4.026912751246585,
581
+ "learning_rate": 1.999844379695719e-05,
582
+ "loss": 0.2924,
583
+ "step": 820
584
+ },
585
+ {
586
+ "epoch": 0.17,
587
+ "grad_norm": 8.551197472310484,
588
+ "learning_rate": 1.999802490013971e-05,
589
+ "loss": 0.2658,
590
+ "step": 830
591
+ },
592
+ {
593
+ "epoch": 0.17,
594
+ "grad_norm": 4.390661707823087,
595
+ "learning_rate": 1.9997556141589807e-05,
596
+ "loss": 0.2386,
597
+ "step": 840
598
+ },
599
+ {
600
+ "epoch": 0.17,
601
+ "grad_norm": 3.697035277012858,
602
+ "learning_rate": 1.9997037523645485e-05,
603
+ "loss": 0.2787,
604
+ "step": 850
605
+ },
606
+ {
607
+ "epoch": 0.17,
608
+ "grad_norm": 4.624963502482154,
609
+ "learning_rate": 1.9996469048893438e-05,
610
+ "loss": 0.2885,
611
+ "step": 860
612
+ },
613
+ {
614
+ "epoch": 0.18,
615
+ "grad_norm": 4.519073120214446,
616
+ "learning_rate": 1.999585072016902e-05,
617
+ "loss": 0.2184,
618
+ "step": 870
619
+ },
620
+ {
621
+ "epoch": 0.18,
622
+ "grad_norm": 4.45710292962464,
623
+ "learning_rate": 1.9995182540556242e-05,
624
+ "loss": 0.2465,
625
+ "step": 880
626
+ },
627
+ {
628
+ "epoch": 0.18,
629
+ "grad_norm": 4.907767674612729,
630
+ "learning_rate": 1.9994464513387758e-05,
631
+ "loss": 0.2579,
632
+ "step": 890
633
+ },
634
+ {
635
+ "epoch": 0.18,
636
+ "grad_norm": 4.569838800905319,
637
+ "learning_rate": 1.999369664224484e-05,
638
+ "loss": 0.3058,
639
+ "step": 900
640
+ },
641
+ {
642
+ "epoch": 0.18,
643
+ "grad_norm": 10.180778298594696,
644
+ "learning_rate": 1.9992878930957364e-05,
645
+ "loss": 0.2722,
646
+ "step": 910
647
+ },
648
+ {
649
+ "epoch": 0.19,
650
+ "grad_norm": 3.9044365907967737,
651
+ "learning_rate": 1.9992011383603794e-05,
652
+ "loss": 0.2905,
653
+ "step": 920
654
+ },
655
+ {
656
+ "epoch": 0.19,
657
+ "grad_norm": 4.204036344405577,
658
+ "learning_rate": 1.999109400451116e-05,
659
+ "loss": 0.2597,
660
+ "step": 930
661
+ },
662
+ {
663
+ "epoch": 0.19,
664
+ "grad_norm": 4.483086909407923,
665
+ "learning_rate": 1.9990126798255032e-05,
666
+ "loss": 0.2527,
667
+ "step": 940
668
+ },
669
+ {
670
+ "epoch": 0.19,
671
+ "grad_norm": 5.650984362708689,
672
+ "learning_rate": 1.9989109769659506e-05,
673
+ "loss": 0.2924,
674
+ "step": 950
675
+ },
676
+ {
677
+ "epoch": 0.19,
678
+ "grad_norm": 4.289005355593254,
679
+ "learning_rate": 1.9988042923797176e-05,
680
+ "loss": 0.2248,
681
+ "step": 960
682
+ },
683
+ {
684
+ "epoch": 0.2,
685
+ "grad_norm": 5.879170337714607,
686
+ "learning_rate": 1.9986926265989092e-05,
687
+ "loss": 0.2313,
688
+ "step": 970
689
+ },
690
+ {
691
+ "epoch": 0.2,
692
+ "grad_norm": 2.7369398894003716,
693
+ "learning_rate": 1.9985759801804768e-05,
694
+ "loss": 0.2655,
695
+ "step": 980
696
+ },
697
+ {
698
+ "epoch": 0.2,
699
+ "grad_norm": 3.9309313039975713,
700
+ "learning_rate": 1.998454353706213e-05,
701
+ "loss": 0.2413,
702
+ "step": 990
703
+ },
704
+ {
705
+ "epoch": 0.2,
706
+ "grad_norm": 3.853949683327624,
707
+ "learning_rate": 1.998327747782748e-05,
708
+ "loss": 0.2626,
709
+ "step": 1000
710
+ },
711
+ {
712
+ "epoch": 0.2,
713
+ "grad_norm": 4.233603769710417,
714
+ "learning_rate": 1.9981961630415495e-05,
715
+ "loss": 0.2813,
716
+ "step": 1010
717
+ },
718
+ {
719
+ "epoch": 0.21,
720
+ "grad_norm": 3.4712931614995775,
721
+ "learning_rate": 1.9980596001389173e-05,
722
+ "loss": 0.2804,
723
+ "step": 1020
724
+ },
725
+ {
726
+ "epoch": 0.21,
727
+ "grad_norm": 3.5161721036970564,
728
+ "learning_rate": 1.9979180597559795e-05,
729
+ "loss": 0.2498,
730
+ "step": 1030
731
+ },
732
+ {
733
+ "epoch": 0.21,
734
+ "grad_norm": 3.914986811551112,
735
+ "learning_rate": 1.997771542598691e-05,
736
+ "loss": 0.2638,
737
+ "step": 1040
738
+ },
739
+ {
740
+ "epoch": 0.21,
741
+ "grad_norm": 3.2174038340462405,
742
+ "learning_rate": 1.9976200493978302e-05,
743
+ "loss": 0.2188,
744
+ "step": 1050
745
+ },
746
+ {
747
+ "epoch": 0.21,
748
+ "grad_norm": 5.034094723003907,
749
+ "learning_rate": 1.9974635809089923e-05,
750
+ "loss": 0.2574,
751
+ "step": 1060
752
+ },
753
+ {
754
+ "epoch": 0.22,
755
+ "grad_norm": 3.9061787102869756,
756
+ "learning_rate": 1.9973021379125887e-05,
757
+ "loss": 0.2823,
758
+ "step": 1070
759
+ },
760
+ {
761
+ "epoch": 0.22,
762
+ "grad_norm": 3.749214471559017,
763
+ "learning_rate": 1.9971357212138418e-05,
764
+ "loss": 0.2158,
765
+ "step": 1080
766
+ },
767
+ {
768
+ "epoch": 0.22,
769
+ "grad_norm": 4.438077647261819,
770
+ "learning_rate": 1.9969643316427806e-05,
771
+ "loss": 0.2373,
772
+ "step": 1090
773
+ },
774
+ {
775
+ "epoch": 0.22,
776
+ "grad_norm": 3.7946665428127644,
777
+ "learning_rate": 1.9967879700542382e-05,
778
+ "loss": 0.266,
779
+ "step": 1100
780
+ },
781
+ {
782
+ "epoch": 0.22,
783
+ "grad_norm": 4.020417555953455,
784
+ "learning_rate": 1.996606637327846e-05,
785
+ "loss": 0.2692,
786
+ "step": 1110
787
+ },
788
+ {
789
+ "epoch": 0.23,
790
+ "grad_norm": 5.291843241151047,
791
+ "learning_rate": 1.9964203343680284e-05,
792
+ "loss": 0.2477,
793
+ "step": 1120
794
+ },
795
+ {
796
+ "epoch": 0.23,
797
+ "grad_norm": 3.3437519964764775,
798
+ "learning_rate": 1.996229062104001e-05,
799
+ "loss": 0.2507,
800
+ "step": 1130
801
+ },
802
+ {
803
+ "epoch": 0.23,
804
+ "grad_norm": 3.3879034836678033,
805
+ "learning_rate": 1.996032821489765e-05,
806
+ "loss": 0.2476,
807
+ "step": 1140
808
+ },
809
+ {
810
+ "epoch": 0.23,
811
+ "grad_norm": 3.7050532861499375,
812
+ "learning_rate": 1.9958316135041e-05,
813
+ "loss": 0.2224,
814
+ "step": 1150
815
+ },
816
+ {
817
+ "epoch": 0.24,
818
+ "grad_norm": 3.8739888502442037,
819
+ "learning_rate": 1.995625439150564e-05,
820
+ "loss": 0.2419,
821
+ "step": 1160
822
+ },
823
+ {
824
+ "epoch": 0.24,
825
+ "grad_norm": 10.982270952246383,
826
+ "learning_rate": 1.9954142994574825e-05,
827
+ "loss": 0.2242,
828
+ "step": 1170
829
+ },
830
+ {
831
+ "epoch": 0.24,
832
+ "grad_norm": 9.832505746272833,
833
+ "learning_rate": 1.9951981954779488e-05,
834
+ "loss": 0.2591,
835
+ "step": 1180
836
+ },
837
+ {
838
+ "epoch": 0.24,
839
+ "grad_norm": 4.369158741790222,
840
+ "learning_rate": 1.9949771282898153e-05,
841
+ "loss": 0.2374,
842
+ "step": 1190
843
+ },
844
+ {
845
+ "epoch": 0.24,
846
+ "grad_norm": 5.630104122791308,
847
+ "learning_rate": 1.994751098995689e-05,
848
+ "loss": 0.2442,
849
+ "step": 1200
850
+ },
851
+ {
852
+ "epoch": 0.25,
853
+ "grad_norm": 4.220933137387155,
854
+ "learning_rate": 1.9945201087229272e-05,
855
+ "loss": 0.2431,
856
+ "step": 1210
857
+ },
858
+ {
859
+ "epoch": 0.25,
860
+ "grad_norm": 5.346564014873171,
861
+ "learning_rate": 1.9942841586236297e-05,
862
+ "loss": 0.2787,
863
+ "step": 1220
864
+ },
865
+ {
866
+ "epoch": 0.25,
867
+ "grad_norm": 3.884615013459559,
868
+ "learning_rate": 1.9940432498746342e-05,
869
+ "loss": 0.1948,
870
+ "step": 1230
871
+ },
872
+ {
873
+ "epoch": 0.25,
874
+ "grad_norm": 4.199936121500256,
875
+ "learning_rate": 1.993797383677512e-05,
876
+ "loss": 0.2605,
877
+ "step": 1240
878
+ },
879
+ {
880
+ "epoch": 0.25,
881
+ "grad_norm": 3.7958971813092854,
882
+ "learning_rate": 1.9935465612585588e-05,
883
+ "loss": 0.2077,
884
+ "step": 1250
885
+ },
886
+ {
887
+ "epoch": 0.26,
888
+ "grad_norm": 4.513168141785257,
889
+ "learning_rate": 1.993290783868791e-05,
890
+ "loss": 0.2584,
891
+ "step": 1260
892
+ },
893
+ {
894
+ "epoch": 0.26,
895
+ "grad_norm": 4.7824378439198805,
896
+ "learning_rate": 1.993030052783938e-05,
897
+ "loss": 0.3022,
898
+ "step": 1270
899
+ },
900
+ {
901
+ "epoch": 0.26,
902
+ "grad_norm": 4.207953986768479,
903
+ "learning_rate": 1.992764369304438e-05,
904
+ "loss": 0.2308,
905
+ "step": 1280
906
+ },
907
+ {
908
+ "epoch": 0.26,
909
+ "grad_norm": 17.62536360994611,
910
+ "learning_rate": 1.9924937347554282e-05,
911
+ "loss": 0.2761,
912
+ "step": 1290
913
+ },
914
+ {
915
+ "epoch": 0.26,
916
+ "grad_norm": 4.270009899619283,
917
+ "learning_rate": 1.9922181504867414e-05,
918
+ "loss": 0.2427,
919
+ "step": 1300
920
+ },
921
+ {
922
+ "epoch": 0.27,
923
+ "grad_norm": 4.922976222292152,
924
+ "learning_rate": 1.9919376178728975e-05,
925
+ "loss": 0.2306,
926
+ "step": 1310
927
+ },
928
+ {
929
+ "epoch": 0.27,
930
+ "grad_norm": 4.215458242682613,
931
+ "learning_rate": 1.9916521383130965e-05,
932
+ "loss": 0.2339,
933
+ "step": 1320
934
+ },
935
+ {
936
+ "epoch": 0.27,
937
+ "grad_norm": 1.559875114528341,
938
+ "learning_rate": 1.9913617132312132e-05,
939
+ "loss": 0.1979,
940
+ "step": 1330
941
+ },
942
+ {
943
+ "epoch": 0.27,
944
+ "grad_norm": 3.911384564892276,
945
+ "learning_rate": 1.9910663440757878e-05,
946
+ "loss": 0.2474,
947
+ "step": 1340
948
+ },
949
+ {
950
+ "epoch": 0.27,
951
+ "grad_norm": 3.5797528557741036,
952
+ "learning_rate": 1.9907660323200207e-05,
953
+ "loss": 0.223,
954
+ "step": 1350
955
+ },
956
+ {
957
+ "epoch": 0.28,
958
+ "grad_norm": 7.031043833783013,
959
+ "learning_rate": 1.9904607794617635e-05,
960
+ "loss": 0.2141,
961
+ "step": 1360
962
+ },
963
+ {
964
+ "epoch": 0.28,
965
+ "grad_norm": 3.721340014798071,
966
+ "learning_rate": 1.9901505870235137e-05,
967
+ "loss": 0.2039,
968
+ "step": 1370
969
+ },
970
+ {
971
+ "epoch": 0.28,
972
+ "grad_norm": 3.889394419978848,
973
+ "learning_rate": 1.989835456552404e-05,
974
+ "loss": 0.3015,
975
+ "step": 1380
976
+ },
977
+ {
978
+ "epoch": 0.28,
979
+ "grad_norm": 4.368266346017717,
980
+ "learning_rate": 1.9895153896201977e-05,
981
+ "loss": 0.2278,
982
+ "step": 1390
983
+ },
984
+ {
985
+ "epoch": 0.28,
986
+ "grad_norm": 4.736620682008376,
987
+ "learning_rate": 1.9891903878232782e-05,
988
+ "loss": 0.246,
989
+ "step": 1400
990
+ },
991
+ {
992
+ "epoch": 0.29,
993
+ "grad_norm": 3.8257038790244744,
994
+ "learning_rate": 1.9888604527826435e-05,
995
+ "loss": 0.2758,
996
+ "step": 1410
997
+ },
998
+ {
999
+ "epoch": 0.29,
1000
+ "grad_norm": 3.1454697624115093,
1001
+ "learning_rate": 1.9885255861438966e-05,
1002
+ "loss": 0.2253,
1003
+ "step": 1420
1004
+ },
1005
+ {
1006
+ "epoch": 0.29,
1007
+ "grad_norm": 4.241144266291823,
1008
+ "learning_rate": 1.988185789577237e-05,
1009
+ "loss": 0.2545,
1010
+ "step": 1430
1011
+ },
1012
+ {
1013
+ "epoch": 0.29,
1014
+ "grad_norm": 4.303491443553284,
1015
+ "learning_rate": 1.987841064777454e-05,
1016
+ "loss": 0.2547,
1017
+ "step": 1440
1018
+ },
1019
+ {
1020
+ "epoch": 0.29,
1021
+ "grad_norm": 7.233119649704639,
1022
+ "learning_rate": 1.9874914134639163e-05,
1023
+ "loss": 0.2697,
1024
+ "step": 1450
1025
+ },
1026
+ {
1027
+ "epoch": 0.3,
1028
+ "grad_norm": 3.1627622593633133,
1029
+ "learning_rate": 1.987136837380565e-05,
1030
+ "loss": 0.2072,
1031
+ "step": 1460
1032
+ },
1033
+ {
1034
+ "epoch": 0.3,
1035
+ "grad_norm": 4.803567895467617,
1036
+ "learning_rate": 1.986777338295904e-05,
1037
+ "loss": 0.2504,
1038
+ "step": 1470
1039
+ },
1040
+ {
1041
+ "epoch": 0.3,
1042
+ "grad_norm": 4.645148028590643,
1043
+ "learning_rate": 1.9864129180029915e-05,
1044
+ "loss": 0.2957,
1045
+ "step": 1480
1046
+ },
1047
+ {
1048
+ "epoch": 0.3,
1049
+ "grad_norm": 4.372735663683347,
1050
+ "learning_rate": 1.9860435783194306e-05,
1051
+ "loss": 0.2342,
1052
+ "step": 1490
1053
+ },
1054
+ {
1055
+ "epoch": 0.3,
1056
+ "grad_norm": 4.244572708317798,
1057
+ "learning_rate": 1.9856693210873616e-05,
1058
+ "loss": 0.1971,
1059
+ "step": 1500
1060
+ },
1061
+ {
1062
+ "epoch": 0.31,
1063
+ "grad_norm": 3.150358082839134,
1064
+ "learning_rate": 1.9852901481734505e-05,
1065
+ "loss": 0.2601,
1066
+ "step": 1510
1067
+ },
1068
+ {
1069
+ "epoch": 0.31,
1070
+ "grad_norm": 3.8063005165293173,
1071
+ "learning_rate": 1.9849060614688825e-05,
1072
+ "loss": 0.2419,
1073
+ "step": 1520
1074
+ },
1075
+ {
1076
+ "epoch": 0.31,
1077
+ "grad_norm": 7.0543295261475105,
1078
+ "learning_rate": 1.98451706288935e-05,
1079
+ "loss": 0.2637,
1080
+ "step": 1530
1081
+ },
1082
+ {
1083
+ "epoch": 0.31,
1084
+ "grad_norm": 3.795035839087491,
1085
+ "learning_rate": 1.9841231543750446e-05,
1086
+ "loss": 0.2632,
1087
+ "step": 1540
1088
+ },
1089
+ {
1090
+ "epoch": 0.31,
1091
+ "grad_norm": 4.609631133641002,
1092
+ "learning_rate": 1.983724337890647e-05,
1093
+ "loss": 0.234,
1094
+ "step": 1550
1095
+ },
1096
+ {
1097
+ "epoch": 0.32,
1098
+ "grad_norm": 4.678952778756691,
1099
+ "learning_rate": 1.9833206154253165e-05,
1100
+ "loss": 0.2096,
1101
+ "step": 1560
1102
+ },
1103
+ {
1104
+ "epoch": 0.32,
1105
+ "grad_norm": 3.695528116851361,
1106
+ "learning_rate": 1.9829119889926836e-05,
1107
+ "loss": 0.2585,
1108
+ "step": 1570
1109
+ },
1110
+ {
1111
+ "epoch": 0.32,
1112
+ "grad_norm": 3.787900010945368,
1113
+ "learning_rate": 1.9824984606308356e-05,
1114
+ "loss": 0.2201,
1115
+ "step": 1580
1116
+ },
1117
+ {
1118
+ "epoch": 0.32,
1119
+ "grad_norm": 4.851549133184082,
1120
+ "learning_rate": 1.982080032402311e-05,
1121
+ "loss": 0.2625,
1122
+ "step": 1590
1123
+ },
1124
+ {
1125
+ "epoch": 0.32,
1126
+ "grad_norm": 4.285022938688077,
1127
+ "learning_rate": 1.9816567063940856e-05,
1128
+ "loss": 0.1898,
1129
+ "step": 1600
1130
+ },
1131
+ {
1132
+ "epoch": 0.33,
1133
+ "grad_norm": 4.576105603313343,
1134
+ "learning_rate": 1.981228484717565e-05,
1135
+ "loss": 0.3178,
1136
+ "step": 1610
1137
+ },
1138
+ {
1139
+ "epoch": 0.33,
1140
+ "grad_norm": 2.614430082371516,
1141
+ "learning_rate": 1.980795369508572e-05,
1142
+ "loss": 0.2022,
1143
+ "step": 1620
1144
+ },
1145
+ {
1146
+ "epoch": 0.33,
1147
+ "grad_norm": 4.348898710957631,
1148
+ "learning_rate": 1.9803573629273364e-05,
1149
+ "loss": 0.2426,
1150
+ "step": 1630
1151
+ },
1152
+ {
1153
+ "epoch": 0.33,
1154
+ "grad_norm": 3.7753032234127812,
1155
+ "learning_rate": 1.9799144671584853e-05,
1156
+ "loss": 0.2547,
1157
+ "step": 1640
1158
+ },
1159
+ {
1160
+ "epoch": 0.33,
1161
+ "grad_norm": 5.096970178030036,
1162
+ "learning_rate": 1.9794666844110303e-05,
1163
+ "loss": 0.209,
1164
+ "step": 1650
1165
+ },
1166
+ {
1167
+ "epoch": 0.34,
1168
+ "grad_norm": 3.8017117810837084,
1169
+ "learning_rate": 1.979014016918359e-05,
1170
+ "loss": 0.2149,
1171
+ "step": 1660
1172
+ },
1173
+ {
1174
+ "epoch": 0.34,
1175
+ "grad_norm": 4.186597128936523,
1176
+ "learning_rate": 1.978556466938221e-05,
1177
+ "loss": 0.2464,
1178
+ "step": 1670
1179
+ },
1180
+ {
1181
+ "epoch": 0.34,
1182
+ "grad_norm": 3.696513394158522,
1183
+ "learning_rate": 1.978094036752719e-05,
1184
+ "loss": 0.2568,
1185
+ "step": 1680
1186
+ },
1187
+ {
1188
+ "epoch": 0.34,
1189
+ "grad_norm": 4.675325784194562,
1190
+ "learning_rate": 1.9776267286682965e-05,
1191
+ "loss": 0.2323,
1192
+ "step": 1690
1193
+ },
1194
+ {
1195
+ "epoch": 0.34,
1196
+ "grad_norm": 5.427689218899514,
1197
+ "learning_rate": 1.9771545450157254e-05,
1198
+ "loss": 0.2631,
1199
+ "step": 1700
1200
+ },
1201
+ {
1202
+ "epoch": 0.35,
1203
+ "grad_norm": 3.1730771347081395,
1204
+ "learning_rate": 1.9766774881500958e-05,
1205
+ "loss": 0.216,
1206
+ "step": 1710
1207
+ },
1208
+ {
1209
+ "epoch": 0.35,
1210
+ "grad_norm": 3.581631337579299,
1211
+ "learning_rate": 1.9761955604508043e-05,
1212
+ "loss": 0.2327,
1213
+ "step": 1720
1214
+ },
1215
+ {
1216
+ "epoch": 0.35,
1217
+ "grad_norm": 4.216013154865277,
1218
+ "learning_rate": 1.975708764321541e-05,
1219
+ "loss": 0.2737,
1220
+ "step": 1730
1221
+ },
1222
+ {
1223
+ "epoch": 0.35,
1224
+ "grad_norm": 3.616181664241495,
1225
+ "learning_rate": 1.975217102190278e-05,
1226
+ "loss": 0.2531,
1227
+ "step": 1740
1228
+ },
1229
+ {
1230
+ "epoch": 0.35,
1231
+ "grad_norm": 4.944731426602014,
1232
+ "learning_rate": 1.974720576509257e-05,
1233
+ "loss": 0.2329,
1234
+ "step": 1750
1235
+ },
1236
+ {
1237
+ "epoch": 0.36,
1238
+ "grad_norm": 3.4547474359483075,
1239
+ "learning_rate": 1.9742191897549783e-05,
1240
+ "loss": 0.2082,
1241
+ "step": 1760
1242
+ },
1243
+ {
1244
+ "epoch": 0.36,
1245
+ "grad_norm": 16.606773079360572,
1246
+ "learning_rate": 1.973712944428187e-05,
1247
+ "loss": 0.2476,
1248
+ "step": 1770
1249
+ },
1250
+ {
1251
+ "epoch": 0.36,
1252
+ "grad_norm": 3.224499798212032,
1253
+ "learning_rate": 1.9732018430538613e-05,
1254
+ "loss": 0.2574,
1255
+ "step": 1780
1256
+ },
1257
+ {
1258
+ "epoch": 0.36,
1259
+ "grad_norm": 3.5823954456349805,
1260
+ "learning_rate": 1.9726858881811992e-05,
1261
+ "loss": 0.2242,
1262
+ "step": 1790
1263
+ },
1264
+ {
1265
+ "epoch": 0.36,
1266
+ "grad_norm": 4.249584626487707,
1267
+ "learning_rate": 1.9721650823836074e-05,
1268
+ "loss": 0.1894,
1269
+ "step": 1800
1270
+ },
1271
+ {
1272
+ "epoch": 0.37,
1273
+ "grad_norm": 2.805651655743975,
1274
+ "learning_rate": 1.971639428258686e-05,
1275
+ "loss": 0.2224,
1276
+ "step": 1810
1277
+ },
1278
+ {
1279
+ "epoch": 0.37,
1280
+ "grad_norm": 6.825631461083238,
1281
+ "learning_rate": 1.971108928428218e-05,
1282
+ "loss": 0.206,
1283
+ "step": 1820
1284
+ },
1285
+ {
1286
+ "epoch": 0.37,
1287
+ "grad_norm": 3.668175314712389,
1288
+ "learning_rate": 1.9705735855381544e-05,
1289
+ "loss": 0.228,
1290
+ "step": 1830
1291
+ },
1292
+ {
1293
+ "epoch": 0.37,
1294
+ "grad_norm": 2.8084463789135126,
1295
+ "learning_rate": 1.9700334022586016e-05,
1296
+ "loss": 0.2313,
1297
+ "step": 1840
1298
+ },
1299
+ {
1300
+ "epoch": 0.37,
1301
+ "grad_norm": 3.566250082150367,
1302
+ "learning_rate": 1.9694883812838095e-05,
1303
+ "loss": 0.2291,
1304
+ "step": 1850
1305
+ },
1306
+ {
1307
+ "epoch": 0.38,
1308
+ "grad_norm": 3.9864026464050437,
1309
+ "learning_rate": 1.9689385253321548e-05,
1310
+ "loss": 0.2312,
1311
+ "step": 1860
1312
+ },
1313
+ {
1314
+ "epoch": 0.38,
1315
+ "grad_norm": 4.0540053541677885,
1316
+ "learning_rate": 1.9683838371461315e-05,
1317
+ "loss": 0.2687,
1318
+ "step": 1870
1319
+ },
1320
+ {
1321
+ "epoch": 0.38,
1322
+ "grad_norm": 3.6121079968078234,
1323
+ "learning_rate": 1.9678243194923333e-05,
1324
+ "loss": 0.2403,
1325
+ "step": 1880
1326
+ },
1327
+ {
1328
+ "epoch": 0.38,
1329
+ "grad_norm": 4.062791315050937,
1330
+ "learning_rate": 1.9672599751614427e-05,
1331
+ "loss": 0.2225,
1332
+ "step": 1890
1333
+ },
1334
+ {
1335
+ "epoch": 0.38,
1336
+ "grad_norm": 11.916356720706089,
1337
+ "learning_rate": 1.966690806968216e-05,
1338
+ "loss": 0.2517,
1339
+ "step": 1900
1340
+ },
1341
+ {
1342
+ "epoch": 0.39,
1343
+ "grad_norm": 9.070397034086051,
1344
+ "learning_rate": 1.9661168177514683e-05,
1345
+ "loss": 0.2386,
1346
+ "step": 1910
1347
+ },
1348
+ {
1349
+ "epoch": 0.39,
1350
+ "grad_norm": 4.223253798926425,
1351
+ "learning_rate": 1.9655380103740618e-05,
1352
+ "loss": 0.2409,
1353
+ "step": 1920
1354
+ },
1355
+ {
1356
+ "epoch": 0.39,
1357
+ "grad_norm": 3.338130757023872,
1358
+ "learning_rate": 1.9649543877228886e-05,
1359
+ "loss": 0.2368,
1360
+ "step": 1930
1361
+ },
1362
+ {
1363
+ "epoch": 0.39,
1364
+ "grad_norm": 3.545188334725578,
1365
+ "learning_rate": 1.9643659527088587e-05,
1366
+ "loss": 0.1738,
1367
+ "step": 1940
1368
+ },
1369
+ {
1370
+ "epoch": 0.4,
1371
+ "grad_norm": 3.408623244835359,
1372
+ "learning_rate": 1.963772708266884e-05,
1373
+ "loss": 0.198,
1374
+ "step": 1950
1375
+ },
1376
+ {
1377
+ "epoch": 0.4,
1378
+ "grad_norm": 3.900722651898139,
1379
+ "learning_rate": 1.9631746573558646e-05,
1380
+ "loss": 0.1959,
1381
+ "step": 1960
1382
+ },
1383
+ {
1384
+ "epoch": 0.4,
1385
+ "grad_norm": 3.692501620096666,
1386
+ "learning_rate": 1.9625718029586732e-05,
1387
+ "loss": 0.2356,
1388
+ "step": 1970
1389
+ },
1390
+ {
1391
+ "epoch": 0.4,
1392
+ "grad_norm": 3.655015133723331,
1393
+ "learning_rate": 1.9619641480821407e-05,
1394
+ "loss": 0.242,
1395
+ "step": 1980
1396
+ },
1397
+ {
1398
+ "epoch": 0.4,
1399
+ "grad_norm": 13.072999667731198,
1400
+ "learning_rate": 1.9613516957570416e-05,
1401
+ "loss": 0.2206,
1402
+ "step": 1990
1403
+ },
1404
+ {
1405
+ "epoch": 0.41,
1406
+ "grad_norm": 3.6663532701131696,
1407
+ "learning_rate": 1.9607344490380778e-05,
1408
+ "loss": 0.2006,
1409
+ "step": 2000
1410
+ },
1411
+ {
1412
+ "epoch": 0.41,
1413
+ "grad_norm": 1.9175793161114592,
1414
+ "learning_rate": 1.9601124110038647e-05,
1415
+ "loss": 0.2138,
1416
+ "step": 2010
1417
+ },
1418
+ {
1419
+ "epoch": 0.41,
1420
+ "grad_norm": 2.244205112045336,
1421
+ "learning_rate": 1.9594855847569144e-05,
1422
+ "loss": 0.2025,
1423
+ "step": 2020
1424
+ },
1425
+ {
1426
+ "epoch": 0.41,
1427
+ "grad_norm": 1.896116682683314,
1428
+ "learning_rate": 1.9588539734236213e-05,
1429
+ "loss": 0.1844,
1430
+ "step": 2030
1431
+ },
1432
+ {
1433
+ "epoch": 0.41,
1434
+ "grad_norm": 2.0057399160737157,
1435
+ "learning_rate": 1.958217580154246e-05,
1436
+ "loss": 0.1576,
1437
+ "step": 2040
1438
+ },
1439
+ {
1440
+ "epoch": 0.42,
1441
+ "grad_norm": 2.1453450088649535,
1442
+ "learning_rate": 1.9575764081229004e-05,
1443
+ "loss": 0.2168,
1444
+ "step": 2050
1445
+ },
1446
+ {
1447
+ "epoch": 0.42,
1448
+ "grad_norm": 2.030575801761766,
1449
+ "learning_rate": 1.9569304605275304e-05,
1450
+ "loss": 0.1728,
1451
+ "step": 2060
1452
+ },
1453
+ {
1454
+ "epoch": 0.42,
1455
+ "grad_norm": 2.5496294537760775,
1456
+ "learning_rate": 1.9562797405899012e-05,
1457
+ "loss": 0.2134,
1458
+ "step": 2070
1459
+ },
1460
+ {
1461
+ "epoch": 0.42,
1462
+ "grad_norm": 2.242162724268332,
1463
+ "learning_rate": 1.955624251555581e-05,
1464
+ "loss": 0.201,
1465
+ "step": 2080
1466
+ },
1467
+ {
1468
+ "epoch": 0.42,
1469
+ "grad_norm": 1.9933952709556402,
1470
+ "learning_rate": 1.954963996693924e-05,
1471
+ "loss": 0.2179,
1472
+ "step": 2090
1473
+ },
1474
+ {
1475
+ "epoch": 0.43,
1476
+ "grad_norm": 2.240735158513416,
1477
+ "learning_rate": 1.954298979298055e-05,
1478
+ "loss": 0.2494,
1479
+ "step": 2100
1480
+ },
1481
+ {
1482
+ "epoch": 0.43,
1483
+ "grad_norm": 2.397210518680832,
1484
+ "learning_rate": 1.953629202684853e-05,
1485
+ "loss": 0.2077,
1486
+ "step": 2110
1487
+ },
1488
+ {
1489
+ "epoch": 0.43,
1490
+ "grad_norm": 4.51567146250967,
1491
+ "learning_rate": 1.9529546701949338e-05,
1492
+ "loss": 0.1859,
1493
+ "step": 2120
1494
+ },
1495
+ {
1496
+ "epoch": 0.43,
1497
+ "grad_norm": 1.6048501115252622,
1498
+ "learning_rate": 1.952275385192635e-05,
1499
+ "loss": 0.1858,
1500
+ "step": 2130
1501
+ },
1502
+ {
1503
+ "epoch": 0.43,
1504
+ "grad_norm": 3.266632654700684,
1505
+ "learning_rate": 1.951591351065996e-05,
1506
+ "loss": 0.2065,
1507
+ "step": 2140
1508
+ },
1509
+ {
1510
+ "epoch": 0.44,
1511
+ "grad_norm": 2.9963705041975857,
1512
+ "learning_rate": 1.950902571226745e-05,
1513
+ "loss": 0.2395,
1514
+ "step": 2150
1515
+ },
1516
+ {
1517
+ "epoch": 0.44,
1518
+ "grad_norm": 2.47269279368607,
1519
+ "learning_rate": 1.9502090491102805e-05,
1520
+ "loss": 0.2234,
1521
+ "step": 2160
1522
+ },
1523
+ {
1524
+ "epoch": 0.44,
1525
+ "grad_norm": 2.4172687657411283,
1526
+ "learning_rate": 1.949510788175652e-05,
1527
+ "loss": 0.2054,
1528
+ "step": 2170
1529
+ },
1530
+ {
1531
+ "epoch": 0.44,
1532
+ "grad_norm": 2.24131454859904,
1533
+ "learning_rate": 1.948807791905546e-05,
1534
+ "loss": 0.2036,
1535
+ "step": 2180
1536
+ },
1537
+ {
1538
+ "epoch": 0.44,
1539
+ "grad_norm": 2.7824059130997916,
1540
+ "learning_rate": 1.9481000638062667e-05,
1541
+ "loss": 0.211,
1542
+ "step": 2190
1543
+ },
1544
+ {
1545
+ "epoch": 0.45,
1546
+ "grad_norm": 2.5084669738687966,
1547
+ "learning_rate": 1.9473876074077193e-05,
1548
+ "loss": 0.1987,
1549
+ "step": 2200
1550
+ },
1551
+ {
1552
+ "epoch": 0.45,
1553
+ "grad_norm": 2.374685869021834,
1554
+ "learning_rate": 1.946670426263392e-05,
1555
+ "loss": 0.18,
1556
+ "step": 2210
1557
+ },
1558
+ {
1559
+ "epoch": 0.45,
1560
+ "grad_norm": 2.1982980430417665,
1561
+ "learning_rate": 1.9459485239503385e-05,
1562
+ "loss": 0.1883,
1563
+ "step": 2220
1564
+ },
1565
+ {
1566
+ "epoch": 0.45,
1567
+ "grad_norm": 2.2072115394074947,
1568
+ "learning_rate": 1.9452219040691604e-05,
1569
+ "loss": 0.1962,
1570
+ "step": 2230
1571
+ },
1572
+ {
1573
+ "epoch": 0.45,
1574
+ "grad_norm": 2.616971043987373,
1575
+ "learning_rate": 1.9444905702439874e-05,
1576
+ "loss": 0.2126,
1577
+ "step": 2240
1578
+ },
1579
+ {
1580
+ "epoch": 0.46,
1581
+ "grad_norm": 2.1674222240111525,
1582
+ "learning_rate": 1.943754526122463e-05,
1583
+ "loss": 0.1622,
1584
+ "step": 2250
1585
+ },
1586
+ {
1587
+ "epoch": 0.46,
1588
+ "grad_norm": 2.201552458849196,
1589
+ "learning_rate": 1.9430137753757222e-05,
1590
+ "loss": 0.2293,
1591
+ "step": 2260
1592
+ },
1593
+ {
1594
+ "epoch": 0.46,
1595
+ "grad_norm": 2.5879310691120314,
1596
+ "learning_rate": 1.9422683216983766e-05,
1597
+ "loss": 0.2008,
1598
+ "step": 2270
1599
+ },
1600
+ {
1601
+ "epoch": 0.46,
1602
+ "grad_norm": 2.113474838782104,
1603
+ "learning_rate": 1.9415181688084922e-05,
1604
+ "loss": 0.1622,
1605
+ "step": 2280
1606
+ },
1607
+ {
1608
+ "epoch": 0.46,
1609
+ "grad_norm": 2.05705061518402,
1610
+ "learning_rate": 1.9407633204475756e-05,
1611
+ "loss": 0.1674,
1612
+ "step": 2290
1613
+ },
1614
+ {
1615
+ "epoch": 0.47,
1616
+ "grad_norm": 3.3556666998600018,
1617
+ "learning_rate": 1.940003780380551e-05,
1618
+ "loss": 0.1765,
1619
+ "step": 2300
1620
+ },
1621
+ {
1622
+ "epoch": 0.47,
1623
+ "grad_norm": 2.332083901173591,
1624
+ "learning_rate": 1.9392395523957438e-05,
1625
+ "loss": 0.1656,
1626
+ "step": 2310
1627
+ },
1628
+ {
1629
+ "epoch": 0.47,
1630
+ "grad_norm": 2.3736488112606757,
1631
+ "learning_rate": 1.9384706403048618e-05,
1632
+ "loss": 0.2206,
1633
+ "step": 2320
1634
+ },
1635
+ {
1636
+ "epoch": 0.47,
1637
+ "grad_norm": 2.6382812391557318,
1638
+ "learning_rate": 1.937697047942974e-05,
1639
+ "loss": 0.1645,
1640
+ "step": 2330
1641
+ },
1642
+ {
1643
+ "epoch": 0.47,
1644
+ "grad_norm": 2.6013892863578194,
1645
+ "learning_rate": 1.9369187791684943e-05,
1646
+ "loss": 0.1705,
1647
+ "step": 2340
1648
+ },
1649
+ {
1650
+ "epoch": 0.48,
1651
+ "grad_norm": 2.1145392602796855,
1652
+ "learning_rate": 1.9361358378631604e-05,
1653
+ "loss": 0.206,
1654
+ "step": 2350
1655
+ },
1656
+ {
1657
+ "epoch": 0.48,
1658
+ "grad_norm": 2.200460676164746,
1659
+ "learning_rate": 1.9353482279320154e-05,
1660
+ "loss": 0.2172,
1661
+ "step": 2360
1662
+ },
1663
+ {
1664
+ "epoch": 0.48,
1665
+ "grad_norm": 2.3806064086412455,
1666
+ "learning_rate": 1.9345559533033867e-05,
1667
+ "loss": 0.1837,
1668
+ "step": 2370
1669
+ },
1670
+ {
1671
+ "epoch": 0.48,
1672
+ "grad_norm": 2.53378489217648,
1673
+ "learning_rate": 1.9337590179288694e-05,
1674
+ "loss": 0.1962,
1675
+ "step": 2380
1676
+ },
1677
+ {
1678
+ "epoch": 0.48,
1679
+ "grad_norm": 13.445648419685131,
1680
+ "learning_rate": 1.9329574257833035e-05,
1681
+ "loss": 0.1332,
1682
+ "step": 2390
1683
+ },
1684
+ {
1685
+ "epoch": 0.49,
1686
+ "grad_norm": 1.7839934654329992,
1687
+ "learning_rate": 1.932151180864756e-05,
1688
+ "loss": 0.1713,
1689
+ "step": 2400
1690
+ },
1691
+ {
1692
+ "epoch": 0.49,
1693
+ "grad_norm": 1.9367486050463985,
1694
+ "learning_rate": 1.9313402871945e-05,
1695
+ "loss": 0.207,
1696
+ "step": 2410
1697
+ },
1698
+ {
1699
+ "epoch": 0.49,
1700
+ "grad_norm": 3.9679375965918084,
1701
+ "learning_rate": 1.930524748816995e-05,
1702
+ "loss": 0.1766,
1703
+ "step": 2420
1704
+ },
1705
+ {
1706
+ "epoch": 0.49,
1707
+ "grad_norm": 2.12675067368969,
1708
+ "learning_rate": 1.9297045697998667e-05,
1709
+ "loss": 0.2119,
1710
+ "step": 2430
1711
+ },
1712
+ {
1713
+ "epoch": 0.49,
1714
+ "grad_norm": 2.330245927266684,
1715
+ "learning_rate": 1.9288797542338875e-05,
1716
+ "loss": 0.192,
1717
+ "step": 2440
1718
+ },
1719
+ {
1720
+ "epoch": 0.5,
1721
+ "grad_norm": 2.2758592638707986,
1722
+ "learning_rate": 1.9280503062329537e-05,
1723
+ "loss": 0.2218,
1724
+ "step": 2450
1725
+ },
1726
+ {
1727
+ "epoch": 0.5,
1728
+ "grad_norm": 3.420743432917777,
1729
+ "learning_rate": 1.9272162299340675e-05,
1730
+ "loss": 0.1517,
1731
+ "step": 2460
1732
+ },
1733
+ {
1734
+ "epoch": 0.5,
1735
+ "grad_norm": 1.4543338792444171,
1736
+ "learning_rate": 1.9263775294973168e-05,
1737
+ "loss": 0.1854,
1738
+ "step": 2470
1739
+ },
1740
+ {
1741
+ "epoch": 0.5,
1742
+ "grad_norm": 2.1078454017192123,
1743
+ "learning_rate": 1.92553420910585e-05,
1744
+ "loss": 0.2071,
1745
+ "step": 2480
1746
+ },
1747
+ {
1748
+ "epoch": 0.5,
1749
+ "grad_norm": 2.24945739985224,
1750
+ "learning_rate": 1.9246862729658616e-05,
1751
+ "loss": 0.1434,
1752
+ "step": 2490
1753
+ },
1754
+ {
1755
+ "epoch": 0.51,
1756
+ "grad_norm": 2.313736662246618,
1757
+ "learning_rate": 1.9238337253065655e-05,
1758
+ "loss": 0.2095,
1759
+ "step": 2500
1760
+ },
1761
+ {
1762
+ "epoch": 0.51,
1763
+ "grad_norm": 2.7303675467709847,
1764
+ "learning_rate": 1.922976570380177e-05,
1765
+ "loss": 0.2015,
1766
+ "step": 2510
1767
+ },
1768
+ {
1769
+ "epoch": 0.51,
1770
+ "grad_norm": 3.1027147281182703,
1771
+ "learning_rate": 1.9221148124618915e-05,
1772
+ "loss": 0.1902,
1773
+ "step": 2520
1774
+ },
1775
+ {
1776
+ "epoch": 0.51,
1777
+ "grad_norm": 1.9571047100967691,
1778
+ "learning_rate": 1.921248455849862e-05,
1779
+ "loss": 0.163,
1780
+ "step": 2530
1781
+ },
1782
+ {
1783
+ "epoch": 0.51,
1784
+ "grad_norm": 2.258815028020344,
1785
+ "learning_rate": 1.9203775048651776e-05,
1786
+ "loss": 0.159,
1787
+ "step": 2540
1788
+ },
1789
+ {
1790
+ "epoch": 0.52,
1791
+ "grad_norm": 3.119730206342172,
1792
+ "learning_rate": 1.9195019638518437e-05,
1793
+ "loss": 0.1766,
1794
+ "step": 2550
1795
+ },
1796
+ {
1797
+ "epoch": 0.52,
1798
+ "grad_norm": 4.155231092799811,
1799
+ "learning_rate": 1.9186218371767587e-05,
1800
+ "loss": 0.1999,
1801
+ "step": 2560
1802
+ },
1803
+ {
1804
+ "epoch": 0.52,
1805
+ "grad_norm": 3.938178594333825,
1806
+ "learning_rate": 1.9177371292296926e-05,
1807
+ "loss": 0.1967,
1808
+ "step": 2570
1809
+ },
1810
+ {
1811
+ "epoch": 0.52,
1812
+ "grad_norm": 2.264714656649641,
1813
+ "learning_rate": 1.916847844423265e-05,
1814
+ "loss": 0.1873,
1815
+ "step": 2580
1816
+ },
1817
+ {
1818
+ "epoch": 0.52,
1819
+ "grad_norm": 2.3278837142950835,
1820
+ "learning_rate": 1.915953987192924e-05,
1821
+ "loss": 0.198,
1822
+ "step": 2590
1823
+ },
1824
+ {
1825
+ "epoch": 0.53,
1826
+ "grad_norm": 2.3796897562558557,
1827
+ "learning_rate": 1.9150555619969228e-05,
1828
+ "loss": 0.1591,
1829
+ "step": 2600
1830
+ },
1831
+ {
1832
+ "epoch": 0.53,
1833
+ "grad_norm": 1.2146800627013359,
1834
+ "learning_rate": 1.914152573316298e-05,
1835
+ "loss": 0.1772,
1836
+ "step": 2610
1837
+ },
1838
+ {
1839
+ "epoch": 0.53,
1840
+ "grad_norm": 2.284220185885104,
1841
+ "learning_rate": 1.9132450256548482e-05,
1842
+ "loss": 0.1924,
1843
+ "step": 2620
1844
+ },
1845
+ {
1846
+ "epoch": 0.53,
1847
+ "grad_norm": 2.3132986803051954,
1848
+ "learning_rate": 1.912332923539109e-05,
1849
+ "loss": 0.1575,
1850
+ "step": 2630
1851
+ },
1852
+ {
1853
+ "epoch": 0.53,
1854
+ "grad_norm": 3.3347587557674214,
1855
+ "learning_rate": 1.9114162715183338e-05,
1856
+ "loss": 0.2016,
1857
+ "step": 2640
1858
+ },
1859
+ {
1860
+ "epoch": 0.54,
1861
+ "grad_norm": 4.1416219410825565,
1862
+ "learning_rate": 1.9104950741644682e-05,
1863
+ "loss": 0.1841,
1864
+ "step": 2650
1865
+ },
1866
+ {
1867
+ "epoch": 0.54,
1868
+ "grad_norm": 1.9784717611438265,
1869
+ "learning_rate": 1.9095693360721288e-05,
1870
+ "loss": 0.1988,
1871
+ "step": 2660
1872
+ },
1873
+ {
1874
+ "epoch": 0.54,
1875
+ "grad_norm": 1.9851485419245707,
1876
+ "learning_rate": 1.90863906185858e-05,
1877
+ "loss": 0.2306,
1878
+ "step": 2670
1879
+ },
1880
+ {
1881
+ "epoch": 0.54,
1882
+ "grad_norm": 1.7507079599778372,
1883
+ "learning_rate": 1.90770425616371e-05,
1884
+ "loss": 0.1812,
1885
+ "step": 2680
1886
+ },
1887
+ {
1888
+ "epoch": 0.54,
1889
+ "grad_norm": 1.4418225191285714,
1890
+ "learning_rate": 1.90676492365001e-05,
1891
+ "loss": 0.1724,
1892
+ "step": 2690
1893
+ },
1894
+ {
1895
+ "epoch": 0.55,
1896
+ "grad_norm": 6.358944590106823,
1897
+ "learning_rate": 1.905821069002548e-05,
1898
+ "loss": 0.2033,
1899
+ "step": 2700
1900
+ },
1901
+ {
1902
+ "epoch": 0.55,
1903
+ "grad_norm": 2.569265877186431,
1904
+ "learning_rate": 1.9048726969289472e-05,
1905
+ "loss": 0.1962,
1906
+ "step": 2710
1907
+ },
1908
+ {
1909
+ "epoch": 0.55,
1910
+ "grad_norm": 2.143027189864614,
1911
+ "learning_rate": 1.9039198121593623e-05,
1912
+ "loss": 0.2134,
1913
+ "step": 2720
1914
+ },
1915
+ {
1916
+ "epoch": 0.55,
1917
+ "grad_norm": 2.3661795002812593,
1918
+ "learning_rate": 1.9029624194464562e-05,
1919
+ "loss": 0.1594,
1920
+ "step": 2730
1921
+ },
1922
+ {
1923
+ "epoch": 0.56,
1924
+ "grad_norm": 2.2244765834008855,
1925
+ "learning_rate": 1.9020005235653752e-05,
1926
+ "loss": 0.1797,
1927
+ "step": 2740
1928
+ },
1929
+ {
1930
+ "epoch": 0.56,
1931
+ "grad_norm": 2.48042520800243,
1932
+ "learning_rate": 1.9010341293137265e-05,
1933
+ "loss": 0.1992,
1934
+ "step": 2750
1935
+ },
1936
+ {
1937
+ "epoch": 0.56,
1938
+ "grad_norm": 2.4141715312149854,
1939
+ "learning_rate": 1.9000632415115526e-05,
1940
+ "loss": 0.194,
1941
+ "step": 2760
1942
+ },
1943
+ {
1944
+ "epoch": 0.56,
1945
+ "grad_norm": 2.1311219610420453,
1946
+ "learning_rate": 1.8990878650013095e-05,
1947
+ "loss": 0.2152,
1948
+ "step": 2770
1949
+ },
1950
+ {
1951
+ "epoch": 0.56,
1952
+ "grad_norm": 2.2386424254232806,
1953
+ "learning_rate": 1.8981080046478408e-05,
1954
+ "loss": 0.1678,
1955
+ "step": 2780
1956
+ },
1957
+ {
1958
+ "epoch": 0.57,
1959
+ "grad_norm": 2.1290366036640167,
1960
+ "learning_rate": 1.8971236653383534e-05,
1961
+ "loss": 0.1815,
1962
+ "step": 2790
1963
+ },
1964
+ {
1965
+ "epoch": 0.57,
1966
+ "grad_norm": 2.332990235126333,
1967
+ "learning_rate": 1.896134851982395e-05,
1968
+ "loss": 0.1601,
1969
+ "step": 2800
1970
+ },
1971
+ {
1972
+ "epoch": 0.57,
1973
+ "grad_norm": 1.9313701752737014,
1974
+ "learning_rate": 1.895141569511827e-05,
1975
+ "loss": 0.1913,
1976
+ "step": 2810
1977
+ },
1978
+ {
1979
+ "epoch": 0.57,
1980
+ "grad_norm": 2.0079851484471387,
1981
+ "learning_rate": 1.8941438228808023e-05,
1982
+ "loss": 0.2147,
1983
+ "step": 2820
1984
+ },
1985
+ {
1986
+ "epoch": 0.57,
1987
+ "grad_norm": 3.300791940163902,
1988
+ "learning_rate": 1.8931416170657383e-05,
1989
+ "loss": 0.198,
1990
+ "step": 2830
1991
+ },
1992
+ {
1993
+ "epoch": 0.58,
1994
+ "grad_norm": 2.005476843621061,
1995
+ "learning_rate": 1.892134957065295e-05,
1996
+ "loss": 0.1835,
1997
+ "step": 2840
1998
+ },
1999
+ {
2000
+ "epoch": 0.58,
2001
+ "grad_norm": 2.6085877287350487,
2002
+ "learning_rate": 1.8911238479003464e-05,
2003
+ "loss": 0.1898,
2004
+ "step": 2850
2005
+ },
2006
+ {
2007
+ "epoch": 0.58,
2008
+ "grad_norm": 2.5393547946239683,
2009
+ "learning_rate": 1.8901082946139585e-05,
2010
+ "loss": 0.2094,
2011
+ "step": 2860
2012
+ },
2013
+ {
2014
+ "epoch": 0.58,
2015
+ "grad_norm": 2.1428525152610582,
2016
+ "learning_rate": 1.8890883022713635e-05,
2017
+ "loss": 0.1754,
2018
+ "step": 2870
2019
+ },
2020
+ {
2021
+ "epoch": 0.58,
2022
+ "grad_norm": 1.9149274467688195,
2023
+ "learning_rate": 1.8880638759599327e-05,
2024
+ "loss": 0.2039,
2025
+ "step": 2880
2026
+ },
2027
+ {
2028
+ "epoch": 0.59,
2029
+ "grad_norm": 3.51632507135593,
2030
+ "learning_rate": 1.8870350207891536e-05,
2031
+ "loss": 0.1857,
2032
+ "step": 2890
2033
+ },
2034
+ {
2035
+ "epoch": 0.59,
2036
+ "grad_norm": 2.4186516754116196,
2037
+ "learning_rate": 1.8860017418906028e-05,
2038
+ "loss": 0.1374,
2039
+ "step": 2900
2040
+ },
2041
+ {
2042
+ "epoch": 0.59,
2043
+ "grad_norm": 2.0258647249782333,
2044
+ "learning_rate": 1.884964044417921e-05,
2045
+ "loss": 0.2084,
2046
+ "step": 2910
2047
+ },
2048
+ {
2049
+ "epoch": 0.59,
2050
+ "grad_norm": 2.0985094320845574,
2051
+ "learning_rate": 1.8839219335467886e-05,
2052
+ "loss": 0.1484,
2053
+ "step": 2920
2054
+ },
2055
+ {
2056
+ "epoch": 0.59,
2057
+ "grad_norm": 2.19199552449932,
2058
+ "learning_rate": 1.8828754144748958e-05,
2059
+ "loss": 0.2029,
2060
+ "step": 2930
2061
+ },
2062
+ {
2063
+ "epoch": 0.6,
2064
+ "grad_norm": 1.9095904364534577,
2065
+ "learning_rate": 1.8818244924219217e-05,
2066
+ "loss": 0.1997,
2067
+ "step": 2940
2068
+ },
2069
+ {
2070
+ "epoch": 0.6,
2071
+ "grad_norm": 2.052834839544693,
2072
+ "learning_rate": 1.8807691726295053e-05,
2073
+ "loss": 0.1536,
2074
+ "step": 2950
2075
+ },
2076
+ {
2077
+ "epoch": 0.6,
2078
+ "grad_norm": 2.179267323237696,
2079
+ "learning_rate": 1.8797094603612192e-05,
2080
+ "loss": 0.2086,
2081
+ "step": 2960
2082
+ },
2083
+ {
2084
+ "epoch": 0.6,
2085
+ "grad_norm": 2.5098689789030355,
2086
+ "learning_rate": 1.878645360902546e-05,
2087
+ "loss": 0.1994,
2088
+ "step": 2970
2089
+ },
2090
+ {
2091
+ "epoch": 0.6,
2092
+ "grad_norm": 4.400137699746838,
2093
+ "learning_rate": 1.8775768795608472e-05,
2094
+ "loss": 0.1606,
2095
+ "step": 2980
2096
+ },
2097
+ {
2098
+ "epoch": 0.61,
2099
+ "grad_norm": 2.590590977646675,
2100
+ "learning_rate": 1.8765040216653427e-05,
2101
+ "loss": 0.1897,
2102
+ "step": 2990
2103
+ },
2104
+ {
2105
+ "epoch": 0.61,
2106
+ "grad_norm": 2.1321993774458954,
2107
+ "learning_rate": 1.8754267925670796e-05,
2108
+ "loss": 0.1531,
2109
+ "step": 3000
2110
+ },
2111
+ {
2112
+ "epoch": 0.61,
2113
+ "grad_norm": 2.5568642398269334,
2114
+ "learning_rate": 1.8743451976389068e-05,
2115
+ "loss": 0.178,
2116
+ "step": 3010
2117
+ },
2118
+ {
2119
+ "epoch": 0.61,
2120
+ "grad_norm": 1.9214830076333376,
2121
+ "learning_rate": 1.8732592422754495e-05,
2122
+ "loss": 0.1897,
2123
+ "step": 3020
2124
+ },
2125
+ {
2126
+ "epoch": 0.61,
2127
+ "grad_norm": 1.969751254803088,
2128
+ "learning_rate": 1.8721689318930806e-05,
2129
+ "loss": 0.1502,
2130
+ "step": 3030
2131
+ },
2132
+ {
2133
+ "epoch": 0.62,
2134
+ "grad_norm": 5.017433292114636,
2135
+ "learning_rate": 1.871074271929894e-05,
2136
+ "loss": 0.1982,
2137
+ "step": 3040
2138
+ },
2139
+ {
2140
+ "epoch": 0.62,
2141
+ "grad_norm": 1.9369614375603101,
2142
+ "learning_rate": 1.8699752678456788e-05,
2143
+ "loss": 0.1719,
2144
+ "step": 3050
2145
+ },
2146
+ {
2147
+ "epoch": 0.62,
2148
+ "grad_norm": 2.006594277192507,
2149
+ "learning_rate": 1.86887192512189e-05,
2150
+ "loss": 0.1932,
2151
+ "step": 3060
2152
+ },
2153
+ {
2154
+ "epoch": 0.62,
2155
+ "grad_norm": 1.715514310451893,
2156
+ "learning_rate": 1.8677642492616236e-05,
2157
+ "loss": 0.1801,
2158
+ "step": 3070
2159
+ },
2160
+ {
2161
+ "epoch": 0.62,
2162
+ "grad_norm": 1.9267545952361633,
2163
+ "learning_rate": 1.8666522457895862e-05,
2164
+ "loss": 0.1893,
2165
+ "step": 3080
2166
+ },
2167
+ {
2168
+ "epoch": 0.63,
2169
+ "grad_norm": 2.1874472475503106,
2170
+ "learning_rate": 1.86553592025207e-05,
2171
+ "loss": 0.1767,
2172
+ "step": 3090
2173
+ },
2174
+ {
2175
+ "epoch": 0.63,
2176
+ "grad_norm": 1.8843605357258664,
2177
+ "learning_rate": 1.8644152782169247e-05,
2178
+ "loss": 0.1802,
2179
+ "step": 3100
2180
+ },
2181
+ {
2182
+ "epoch": 0.63,
2183
+ "grad_norm": 2.3298848304608684,
2184
+ "learning_rate": 1.8632903252735276e-05,
2185
+ "loss": 0.1667,
2186
+ "step": 3110
2187
+ },
2188
+ {
2189
+ "epoch": 0.63,
2190
+ "grad_norm": 2.0245193476149312,
2191
+ "learning_rate": 1.862161067032759e-05,
2192
+ "loss": 0.1834,
2193
+ "step": 3120
2194
+ },
2195
+ {
2196
+ "epoch": 0.63,
2197
+ "grad_norm": 2.239252595358892,
2198
+ "learning_rate": 1.861027509126971e-05,
2199
+ "loss": 0.2083,
2200
+ "step": 3130
2201
+ },
2202
+ {
2203
+ "epoch": 0.64,
2204
+ "grad_norm": 2.000691294407405,
2205
+ "learning_rate": 1.8598896572099624e-05,
2206
+ "loss": 0.1739,
2207
+ "step": 3140
2208
+ },
2209
+ {
2210
+ "epoch": 0.64,
2211
+ "grad_norm": 1.656761882061858,
2212
+ "learning_rate": 1.8587475169569483e-05,
2213
+ "loss": 0.1815,
2214
+ "step": 3150
2215
+ },
2216
+ {
2217
+ "epoch": 0.64,
2218
+ "grad_norm": 2.1245215382192346,
2219
+ "learning_rate": 1.8576010940645325e-05,
2220
+ "loss": 0.1606,
2221
+ "step": 3160
2222
+ },
2223
+ {
2224
+ "epoch": 0.64,
2225
+ "grad_norm": 1.7762853803876042,
2226
+ "learning_rate": 1.856450394250679e-05,
2227
+ "loss": 0.1969,
2228
+ "step": 3170
2229
+ },
2230
+ {
2231
+ "epoch": 0.64,
2232
+ "grad_norm": 2.8830827622510697,
2233
+ "learning_rate": 1.855295423254685e-05,
2234
+ "loss": 0.1555,
2235
+ "step": 3180
2236
+ },
2237
+ {
2238
+ "epoch": 0.65,
2239
+ "grad_norm": 2.1861197340441234,
2240
+ "learning_rate": 1.854136186837149e-05,
2241
+ "loss": 0.1889,
2242
+ "step": 3190
2243
+ },
2244
+ {
2245
+ "epoch": 0.65,
2246
+ "grad_norm": 2.3198392751530705,
2247
+ "learning_rate": 1.8529726907799444e-05,
2248
+ "loss": 0.1943,
2249
+ "step": 3200
2250
+ },
2251
+ {
2252
+ "epoch": 0.65,
2253
+ "grad_norm": 1.9214231378576725,
2254
+ "learning_rate": 1.8518049408861915e-05,
2255
+ "loss": 0.1831,
2256
+ "step": 3210
2257
+ },
2258
+ {
2259
+ "epoch": 0.65,
2260
+ "grad_norm": 2.398048593011276,
2261
+ "learning_rate": 1.850632942980226e-05,
2262
+ "loss": 0.2029,
2263
+ "step": 3220
2264
+ },
2265
+ {
2266
+ "epoch": 0.65,
2267
+ "grad_norm": 2.2143080398207573,
2268
+ "learning_rate": 1.8494567029075714e-05,
2269
+ "loss": 0.1718,
2270
+ "step": 3230
2271
+ },
2272
+ {
2273
+ "epoch": 0.66,
2274
+ "grad_norm": 2.126646974386725,
2275
+ "learning_rate": 1.84827622653491e-05,
2276
+ "loss": 0.1302,
2277
+ "step": 3240
2278
+ },
2279
+ {
2280
+ "epoch": 0.66,
2281
+ "grad_norm": 11.599444777256583,
2282
+ "learning_rate": 1.847091519750053e-05,
2283
+ "loss": 0.1809,
2284
+ "step": 3250
2285
+ },
2286
+ {
2287
+ "epoch": 0.66,
2288
+ "grad_norm": 2.3991674250656403,
2289
+ "learning_rate": 1.8459025884619125e-05,
2290
+ "loss": 0.1865,
2291
+ "step": 3260
2292
+ },
2293
+ {
2294
+ "epoch": 0.66,
2295
+ "grad_norm": 6.4503963141704155,
2296
+ "learning_rate": 1.844709438600469e-05,
2297
+ "loss": 0.2251,
2298
+ "step": 3270
2299
+ },
2300
+ {
2301
+ "epoch": 0.66,
2302
+ "grad_norm": 2.4044007961395626,
2303
+ "learning_rate": 1.8435120761167453e-05,
2304
+ "loss": 0.1708,
2305
+ "step": 3280
2306
+ },
2307
+ {
2308
+ "epoch": 0.67,
2309
+ "grad_norm": 1.5993439730175965,
2310
+ "learning_rate": 1.8423105069827753e-05,
2311
+ "loss": 0.1651,
2312
+ "step": 3290
2313
+ },
2314
+ {
2315
+ "epoch": 0.67,
2316
+ "grad_norm": 1.9740878207799195,
2317
+ "learning_rate": 1.8411047371915737e-05,
2318
+ "loss": 0.1955,
2319
+ "step": 3300
2320
+ },
2321
+ {
2322
+ "epoch": 0.67,
2323
+ "grad_norm": 2.244083998109266,
2324
+ "learning_rate": 1.839894772757106e-05,
2325
+ "loss": 0.2106,
2326
+ "step": 3310
2327
+ },
2328
+ {
2329
+ "epoch": 0.67,
2330
+ "grad_norm": 2.2654343097283434,
2331
+ "learning_rate": 1.8386806197142607e-05,
2332
+ "loss": 0.1331,
2333
+ "step": 3320
2334
+ },
2335
+ {
2336
+ "epoch": 0.67,
2337
+ "grad_norm": 1.9373155321474749,
2338
+ "learning_rate": 1.837462284118817e-05,
2339
+ "loss": 0.143,
2340
+ "step": 3330
2341
+ },
2342
+ {
2343
+ "epoch": 0.68,
2344
+ "grad_norm": 2.041586413720046,
2345
+ "learning_rate": 1.8362397720474144e-05,
2346
+ "loss": 0.1805,
2347
+ "step": 3340
2348
+ },
2349
+ {
2350
+ "epoch": 0.68,
2351
+ "grad_norm": 2.311869767372086,
2352
+ "learning_rate": 1.8350130895975247e-05,
2353
+ "loss": 0.1748,
2354
+ "step": 3350
2355
+ },
2356
+ {
2357
+ "epoch": 0.68,
2358
+ "grad_norm": 2.0358596142782224,
2359
+ "learning_rate": 1.8337822428874187e-05,
2360
+ "loss": 0.197,
2361
+ "step": 3360
2362
+ },
2363
+ {
2364
+ "epoch": 0.68,
2365
+ "grad_norm": 12.765829395166001,
2366
+ "learning_rate": 1.8325472380561382e-05,
2367
+ "loss": 0.2043,
2368
+ "step": 3370
2369
+ },
2370
+ {
2371
+ "epoch": 0.68,
2372
+ "grad_norm": 2.283832499199408,
2373
+ "learning_rate": 1.831308081263464e-05,
2374
+ "loss": 0.1696,
2375
+ "step": 3380
2376
+ },
2377
+ {
2378
+ "epoch": 0.69,
2379
+ "grad_norm": 2.099264143128957,
2380
+ "learning_rate": 1.8300647786898843e-05,
2381
+ "loss": 0.1772,
2382
+ "step": 3390
2383
+ },
2384
+ {
2385
+ "epoch": 0.69,
2386
+ "grad_norm": 3.0418348048172117,
2387
+ "learning_rate": 1.8288173365365675e-05,
2388
+ "loss": 0.1679,
2389
+ "step": 3400
2390
+ },
2391
+ {
2392
+ "epoch": 0.69,
2393
+ "grad_norm": 2.928400270597656,
2394
+ "learning_rate": 1.827565761025326e-05,
2395
+ "loss": 0.1839,
2396
+ "step": 3410
2397
+ },
2398
+ {
2399
+ "epoch": 0.69,
2400
+ "grad_norm": 1.9489192388525756,
2401
+ "learning_rate": 1.82631005839859e-05,
2402
+ "loss": 0.1702,
2403
+ "step": 3420
2404
+ },
2405
+ {
2406
+ "epoch": 0.69,
2407
+ "grad_norm": 1.866673546449154,
2408
+ "learning_rate": 1.825050234919374e-05,
2409
+ "loss": 0.1913,
2410
+ "step": 3430
2411
+ },
2412
+ {
2413
+ "epoch": 0.7,
2414
+ "grad_norm": 1.9782480885566223,
2415
+ "learning_rate": 1.8237862968712442e-05,
2416
+ "loss": 0.189,
2417
+ "step": 3440
2418
+ },
2419
+ {
2420
+ "epoch": 0.7,
2421
+ "grad_norm": 2.072252048803903,
2422
+ "learning_rate": 1.8225182505582918e-05,
2423
+ "loss": 0.1908,
2424
+ "step": 3450
2425
+ },
2426
+ {
2427
+ "epoch": 0.7,
2428
+ "grad_norm": 2.1333806325715523,
2429
+ "learning_rate": 1.821246102305096e-05,
2430
+ "loss": 0.204,
2431
+ "step": 3460
2432
+ },
2433
+ {
2434
+ "epoch": 0.7,
2435
+ "grad_norm": 2.0590278992877113,
2436
+ "learning_rate": 1.8199698584566967e-05,
2437
+ "loss": 0.1833,
2438
+ "step": 3470
2439
+ },
2440
+ {
2441
+ "epoch": 0.71,
2442
+ "grad_norm": 2.2552945044942474,
2443
+ "learning_rate": 1.8186895253785603e-05,
2444
+ "loss": 0.2076,
2445
+ "step": 3480
2446
+ },
2447
+ {
2448
+ "epoch": 0.71,
2449
+ "grad_norm": 1.9634969960872868,
2450
+ "learning_rate": 1.8174051094565484e-05,
2451
+ "loss": 0.2097,
2452
+ "step": 3490
2453
+ },
2454
+ {
2455
+ "epoch": 0.71,
2456
+ "grad_norm": 2.2410044456550184,
2457
+ "learning_rate": 1.816116617096889e-05,
2458
+ "loss": 0.2217,
2459
+ "step": 3500
2460
+ },
2461
+ {
2462
+ "epoch": 0.71,
2463
+ "grad_norm": 2.1336586639575237,
2464
+ "learning_rate": 1.8148240547261387e-05,
2465
+ "loss": 0.18,
2466
+ "step": 3510
2467
+ },
2468
+ {
2469
+ "epoch": 0.71,
2470
+ "grad_norm": 2.173336388150898,
2471
+ "learning_rate": 1.813527428791156e-05,
2472
+ "loss": 0.1756,
2473
+ "step": 3520
2474
+ },
2475
+ {
2476
+ "epoch": 0.72,
2477
+ "grad_norm": 2.671364540282752,
2478
+ "learning_rate": 1.812226745759066e-05,
2479
+ "loss": 0.1863,
2480
+ "step": 3530
2481
+ },
2482
+ {
2483
+ "epoch": 0.72,
2484
+ "grad_norm": 1.9057012299641733,
2485
+ "learning_rate": 1.8109220121172306e-05,
2486
+ "loss": 0.2206,
2487
+ "step": 3540
2488
+ },
2489
+ {
2490
+ "epoch": 0.72,
2491
+ "grad_norm": 0.1813973982571034,
2492
+ "learning_rate": 1.8096132343732135e-05,
2493
+ "loss": 0.1462,
2494
+ "step": 3550
2495
+ },
2496
+ {
2497
+ "epoch": 0.72,
2498
+ "grad_norm": 1.8892866709987723,
2499
+ "learning_rate": 1.808300419054749e-05,
2500
+ "loss": 0.1803,
2501
+ "step": 3560
2502
+ },
2503
+ {
2504
+ "epoch": 0.72,
2505
+ "grad_norm": 2.003086926459966,
2506
+ "learning_rate": 1.80698357270971e-05,
2507
+ "loss": 0.2032,
2508
+ "step": 3570
2509
+ },
2510
+ {
2511
+ "epoch": 0.73,
2512
+ "grad_norm": 3.2221389262807874,
2513
+ "learning_rate": 1.8056627019060738e-05,
2514
+ "loss": 0.1631,
2515
+ "step": 3580
2516
+ },
2517
+ {
2518
+ "epoch": 0.73,
2519
+ "grad_norm": 1.2538102728217932,
2520
+ "learning_rate": 1.8043378132318927e-05,
2521
+ "loss": 0.1692,
2522
+ "step": 3590
2523
+ },
2524
+ {
2525
+ "epoch": 0.73,
2526
+ "grad_norm": 4.563350232475948,
2527
+ "learning_rate": 1.8030089132952557e-05,
2528
+ "loss": 0.1727,
2529
+ "step": 3600
2530
+ },
2531
+ {
2532
+ "epoch": 0.73,
2533
+ "grad_norm": 2.000374282147408,
2534
+ "learning_rate": 1.8016760087242605e-05,
2535
+ "loss": 0.1733,
2536
+ "step": 3610
2537
+ },
2538
+ {
2539
+ "epoch": 0.73,
2540
+ "grad_norm": 1.9278853800801559,
2541
+ "learning_rate": 1.800339106166978e-05,
2542
+ "loss": 0.1852,
2543
+ "step": 3620
2544
+ },
2545
+ {
2546
+ "epoch": 0.74,
2547
+ "grad_norm": 2.2241283295196514,
2548
+ "learning_rate": 1.79899821229142e-05,
2549
+ "loss": 0.143,
2550
+ "step": 3630
2551
+ },
2552
+ {
2553
+ "epoch": 0.74,
2554
+ "grad_norm": 2.4864913275556133,
2555
+ "learning_rate": 1.7976533337855053e-05,
2556
+ "loss": 0.1884,
2557
+ "step": 3640
2558
+ },
2559
+ {
2560
+ "epoch": 0.74,
2561
+ "grad_norm": 3.860042375322093,
2562
+ "learning_rate": 1.7963044773570265e-05,
2563
+ "loss": 0.1641,
2564
+ "step": 3650
2565
+ },
2566
+ {
2567
+ "epoch": 0.74,
2568
+ "grad_norm": 2.0454176745318677,
2569
+ "learning_rate": 1.7949516497336176e-05,
2570
+ "loss": 0.1864,
2571
+ "step": 3660
2572
+ },
2573
+ {
2574
+ "epoch": 0.74,
2575
+ "grad_norm": 2.46822670327846,
2576
+ "learning_rate": 1.793594857662718e-05,
2577
+ "loss": 0.1924,
2578
+ "step": 3670
2579
+ },
2580
+ {
2581
+ "epoch": 0.75,
2582
+ "grad_norm": 2.4872265651073158,
2583
+ "learning_rate": 1.792234107911542e-05,
2584
+ "loss": 0.1546,
2585
+ "step": 3680
2586
+ },
2587
+ {
2588
+ "epoch": 0.75,
2589
+ "grad_norm": 2.0379945728128663,
2590
+ "learning_rate": 1.7908694072670426e-05,
2591
+ "loss": 0.1711,
2592
+ "step": 3690
2593
+ },
2594
+ {
2595
+ "epoch": 0.75,
2596
+ "grad_norm": 2.407085620398544,
2597
+ "learning_rate": 1.7895007625358783e-05,
2598
+ "loss": 0.1701,
2599
+ "step": 3700
2600
+ },
2601
+ {
2602
+ "epoch": 0.75,
2603
+ "grad_norm": 4.042625051438552,
2604
+ "learning_rate": 1.7881281805443805e-05,
2605
+ "loss": 0.195,
2606
+ "step": 3710
2607
+ },
2608
+ {
2609
+ "epoch": 0.75,
2610
+ "grad_norm": 1.9781223218758097,
2611
+ "learning_rate": 1.786751668138517e-05,
2612
+ "loss": 0.1621,
2613
+ "step": 3720
2614
+ },
2615
+ {
2616
+ "epoch": 0.76,
2617
+ "grad_norm": 4.657693319193961,
2618
+ "learning_rate": 1.7853712321838602e-05,
2619
+ "loss": 0.1968,
2620
+ "step": 3730
2621
+ },
2622
+ {
2623
+ "epoch": 0.76,
2624
+ "grad_norm": 17.520636897246522,
2625
+ "learning_rate": 1.7839868795655507e-05,
2626
+ "loss": 0.1475,
2627
+ "step": 3740
2628
+ },
2629
+ {
2630
+ "epoch": 0.76,
2631
+ "grad_norm": 2.1710924893238626,
2632
+ "learning_rate": 1.782598617188265e-05,
2633
+ "loss": 0.1686,
2634
+ "step": 3750
2635
+ },
2636
+ {
2637
+ "epoch": 0.76,
2638
+ "grad_norm": 2.205220842093223,
2639
+ "learning_rate": 1.78120645197618e-05,
2640
+ "loss": 0.1587,
2641
+ "step": 3760
2642
+ },
2643
+ {
2644
+ "epoch": 0.76,
2645
+ "grad_norm": 1.95114988022513,
2646
+ "learning_rate": 1.7798103908729377e-05,
2647
+ "loss": 0.1743,
2648
+ "step": 3770
2649
+ },
2650
+ {
2651
+ "epoch": 0.77,
2652
+ "grad_norm": 2.7456615210938744,
2653
+ "learning_rate": 1.778410440841613e-05,
2654
+ "loss": 0.1468,
2655
+ "step": 3780
2656
+ },
2657
+ {
2658
+ "epoch": 0.77,
2659
+ "grad_norm": 2.8973758532017895,
2660
+ "learning_rate": 1.7770066088646767e-05,
2661
+ "loss": 0.1883,
2662
+ "step": 3790
2663
+ },
2664
+ {
2665
+ "epoch": 0.77,
2666
+ "grad_norm": 2.5445827161522616,
2667
+ "learning_rate": 1.7755989019439607e-05,
2668
+ "loss": 0.1895,
2669
+ "step": 3800
2670
+ },
2671
+ {
2672
+ "epoch": 0.77,
2673
+ "grad_norm": 2.169743981267003,
2674
+ "learning_rate": 1.774187327100625e-05,
2675
+ "loss": 0.1716,
2676
+ "step": 3810
2677
+ },
2678
+ {
2679
+ "epoch": 0.77,
2680
+ "grad_norm": 2.1135486144554387,
2681
+ "learning_rate": 1.7727718913751207e-05,
2682
+ "loss": 0.1164,
2683
+ "step": 3820
2684
+ },
2685
+ {
2686
+ "epoch": 0.78,
2687
+ "grad_norm": 2.0715545958907535,
2688
+ "learning_rate": 1.7713526018271558e-05,
2689
+ "loss": 0.1717,
2690
+ "step": 3830
2691
+ },
2692
+ {
2693
+ "epoch": 0.78,
2694
+ "grad_norm": 8.108185256384777,
2695
+ "learning_rate": 1.76992946553566e-05,
2696
+ "loss": 0.1662,
2697
+ "step": 3840
2698
+ },
2699
+ {
2700
+ "epoch": 0.78,
2701
+ "grad_norm": 2.4906125249035713,
2702
+ "learning_rate": 1.7685024895987494e-05,
2703
+ "loss": 0.184,
2704
+ "step": 3850
2705
+ },
2706
+ {
2707
+ "epoch": 0.78,
2708
+ "grad_norm": 3.009141821458063,
2709
+ "learning_rate": 1.7670716811336902e-05,
2710
+ "loss": 0.2086,
2711
+ "step": 3860
2712
+ },
2713
+ {
2714
+ "epoch": 0.78,
2715
+ "grad_norm": 1.935774373100715,
2716
+ "learning_rate": 1.7656370472768648e-05,
2717
+ "loss": 0.1566,
2718
+ "step": 3870
2719
+ },
2720
+ {
2721
+ "epoch": 0.79,
2722
+ "grad_norm": 6.368484758089139,
2723
+ "learning_rate": 1.7641985951837347e-05,
2724
+ "loss": 0.2127,
2725
+ "step": 3880
2726
+ },
2727
+ {
2728
+ "epoch": 0.79,
2729
+ "grad_norm": 2.2821667018184533,
2730
+ "learning_rate": 1.7627563320288056e-05,
2731
+ "loss": 0.1822,
2732
+ "step": 3890
2733
+ },
2734
+ {
2735
+ "epoch": 0.79,
2736
+ "grad_norm": 1.7481761010849746,
2737
+ "learning_rate": 1.7613102650055925e-05,
2738
+ "loss": 0.2264,
2739
+ "step": 3900
2740
+ },
2741
+ {
2742
+ "epoch": 0.79,
2743
+ "grad_norm": 2.111098426089184,
2744
+ "learning_rate": 1.759860401326581e-05,
2745
+ "loss": 0.1838,
2746
+ "step": 3910
2747
+ },
2748
+ {
2749
+ "epoch": 0.79,
2750
+ "grad_norm": 2.3865926806030204,
2751
+ "learning_rate": 1.758406748223194e-05,
2752
+ "loss": 0.1779,
2753
+ "step": 3920
2754
+ },
2755
+ {
2756
+ "epoch": 0.8,
2757
+ "grad_norm": 2.1732390806125,
2758
+ "learning_rate": 1.7569493129457554e-05,
2759
+ "loss": 0.1713,
2760
+ "step": 3930
2761
+ },
2762
+ {
2763
+ "epoch": 0.8,
2764
+ "grad_norm": 1.8814065202760815,
2765
+ "learning_rate": 1.7554881027634516e-05,
2766
+ "loss": 0.178,
2767
+ "step": 3940
2768
+ },
2769
+ {
2770
+ "epoch": 0.8,
2771
+ "grad_norm": 2.16916361006078,
2772
+ "learning_rate": 1.754023124964299e-05,
2773
+ "loss": 0.1475,
2774
+ "step": 3950
2775
+ },
2776
+ {
2777
+ "epoch": 0.8,
2778
+ "grad_norm": 2.27943306248829,
2779
+ "learning_rate": 1.7525543868551045e-05,
2780
+ "loss": 0.1997,
2781
+ "step": 3960
2782
+ },
2783
+ {
2784
+ "epoch": 0.8,
2785
+ "grad_norm": 3.117493119710199,
2786
+ "learning_rate": 1.7510818957614292e-05,
2787
+ "loss": 0.1475,
2788
+ "step": 3970
2789
+ },
2790
+ {
2791
+ "epoch": 0.81,
2792
+ "grad_norm": 2.3033259103584567,
2793
+ "learning_rate": 1.7496056590275546e-05,
2794
+ "loss": 0.1853,
2795
+ "step": 3980
2796
+ },
2797
+ {
2798
+ "epoch": 0.81,
2799
+ "grad_norm": 4.045509608812605,
2800
+ "learning_rate": 1.7481256840164436e-05,
2801
+ "loss": 0.171,
2802
+ "step": 3990
2803
+ },
2804
+ {
2805
+ "epoch": 0.81,
2806
+ "grad_norm": 1.7320172203917021,
2807
+ "learning_rate": 1.7466419781097038e-05,
2808
+ "loss": 0.1619,
2809
+ "step": 4000
2810
+ },
2811
+ {
2812
+ "epoch": 0.81,
2813
+ "grad_norm": 2.1721310910520772,
2814
+ "learning_rate": 1.745154548707551e-05,
2815
+ "loss": 0.1614,
2816
+ "step": 4010
2817
+ },
2818
+ {
2819
+ "epoch": 0.81,
2820
+ "grad_norm": 3.4362498003979374,
2821
+ "learning_rate": 1.7436634032287735e-05,
2822
+ "loss": 0.1885,
2823
+ "step": 4020
2824
+ },
2825
+ {
2826
+ "epoch": 0.82,
2827
+ "grad_norm": 2.203441191364378,
2828
+ "learning_rate": 1.7421685491106933e-05,
2829
+ "loss": 0.1746,
2830
+ "step": 4030
2831
+ },
2832
+ {
2833
+ "epoch": 0.82,
2834
+ "grad_norm": 2.2235513235782136,
2835
+ "learning_rate": 1.740669993809131e-05,
2836
+ "loss": 0.1958,
2837
+ "step": 4040
2838
+ },
2839
+ {
2840
+ "epoch": 0.82,
2841
+ "grad_norm": 1.662744062478203,
2842
+ "learning_rate": 1.7391677447983663e-05,
2843
+ "loss": 0.168,
2844
+ "step": 4050
2845
+ },
2846
+ {
2847
+ "epoch": 0.82,
2848
+ "grad_norm": 1.9923877091876279,
2849
+ "learning_rate": 1.7376618095711018e-05,
2850
+ "loss": 0.1718,
2851
+ "step": 4060
2852
+ },
2853
+ {
2854
+ "epoch": 0.82,
2855
+ "grad_norm": 2.426593736593661,
2856
+ "learning_rate": 1.7361521956384264e-05,
2857
+ "loss": 0.1741,
2858
+ "step": 4070
2859
+ },
2860
+ {
2861
+ "epoch": 0.83,
2862
+ "grad_norm": 1.7464227217148067,
2863
+ "learning_rate": 1.7346389105297766e-05,
2864
+ "loss": 0.1726,
2865
+ "step": 4080
2866
+ },
2867
+ {
2868
+ "epoch": 0.83,
2869
+ "grad_norm": 2.3749413734026383,
2870
+ "learning_rate": 1.7331219617928997e-05,
2871
+ "loss": 0.1583,
2872
+ "step": 4090
2873
+ },
2874
+ {
2875
+ "epoch": 0.83,
2876
+ "grad_norm": 2.114701165986187,
2877
+ "learning_rate": 1.7316013569938154e-05,
2878
+ "loss": 0.2066,
2879
+ "step": 4100
2880
+ },
2881
+ {
2882
+ "epoch": 0.83,
2883
+ "grad_norm": 1.8180037497973824,
2884
+ "learning_rate": 1.73007710371678e-05,
2885
+ "loss": 0.188,
2886
+ "step": 4110
2887
+ },
2888
+ {
2889
+ "epoch": 0.83,
2890
+ "grad_norm": 2.2556879416055726,
2891
+ "learning_rate": 1.7285492095642455e-05,
2892
+ "loss": 0.1824,
2893
+ "step": 4120
2894
+ },
2895
+ {
2896
+ "epoch": 0.84,
2897
+ "grad_norm": 1.9441561651729724,
2898
+ "learning_rate": 1.7270176821568244e-05,
2899
+ "loss": 0.1828,
2900
+ "step": 4130
2901
+ },
2902
+ {
2903
+ "epoch": 0.84,
2904
+ "grad_norm": 3.863082155472389,
2905
+ "learning_rate": 1.72548252913325e-05,
2906
+ "loss": 0.1929,
2907
+ "step": 4140
2908
+ },
2909
+ {
2910
+ "epoch": 0.84,
2911
+ "grad_norm": 2.1777659050408067,
2912
+ "learning_rate": 1.72394375815034e-05,
2913
+ "loss": 0.1872,
2914
+ "step": 4150
2915
+ },
2916
+ {
2917
+ "epoch": 0.84,
2918
+ "grad_norm": 2.289054063384357,
2919
+ "learning_rate": 1.722401376882955e-05,
2920
+ "loss": 0.1619,
2921
+ "step": 4160
2922
+ },
2923
+ {
2924
+ "epoch": 0.84,
2925
+ "grad_norm": 7.1273553535612,
2926
+ "learning_rate": 1.7208553930239655e-05,
2927
+ "loss": 0.1752,
2928
+ "step": 4170
2929
+ },
2930
+ {
2931
+ "epoch": 0.85,
2932
+ "grad_norm": 2.4202257016442523,
2933
+ "learning_rate": 1.7193058142842076e-05,
2934
+ "loss": 0.1966,
2935
+ "step": 4180
2936
+ },
2937
+ {
2938
+ "epoch": 0.85,
2939
+ "grad_norm": 2.599358985816695,
2940
+ "learning_rate": 1.7177526483924492e-05,
2941
+ "loss": 0.1739,
2942
+ "step": 4190
2943
+ },
2944
+ {
2945
+ "epoch": 0.85,
2946
+ "grad_norm": 2.145207088406254,
2947
+ "learning_rate": 1.7161959030953498e-05,
2948
+ "loss": 0.1606,
2949
+ "step": 4200
2950
+ },
2951
+ {
2952
+ "epoch": 0.85,
2953
+ "grad_norm": 2.4989297483292643,
2954
+ "learning_rate": 1.71463558615742e-05,
2955
+ "loss": 0.1441,
2956
+ "step": 4210
2957
+ },
2958
+ {
2959
+ "epoch": 0.85,
2960
+ "grad_norm": 2.333564701416749,
2961
+ "learning_rate": 1.713071705360987e-05,
2962
+ "loss": 0.1697,
2963
+ "step": 4220
2964
+ },
2965
+ {
2966
+ "epoch": 0.86,
2967
+ "grad_norm": 1.6007875381874954,
2968
+ "learning_rate": 1.7115042685061507e-05,
2969
+ "loss": 0.1801,
2970
+ "step": 4230
2971
+ },
2972
+ {
2973
+ "epoch": 0.86,
2974
+ "grad_norm": 2.0001777331534516,
2975
+ "learning_rate": 1.7099332834107497e-05,
2976
+ "loss": 0.1236,
2977
+ "step": 4240
2978
+ },
2979
+ {
2980
+ "epoch": 0.86,
2981
+ "grad_norm": 2.6336492006976187,
2982
+ "learning_rate": 1.7083587579103187e-05,
2983
+ "loss": 0.166,
2984
+ "step": 4250
2985
+ },
2986
+ {
2987
+ "epoch": 0.86,
2988
+ "grad_norm": 2.181960694578793,
2989
+ "learning_rate": 1.7067806998580507e-05,
2990
+ "loss": 0.1997,
2991
+ "step": 4260
2992
+ },
2993
+ {
2994
+ "epoch": 0.87,
2995
+ "grad_norm": 3.2024986585127366,
2996
+ "learning_rate": 1.7051991171247582e-05,
2997
+ "loss": 0.1508,
2998
+ "step": 4270
2999
+ },
3000
+ {
3001
+ "epoch": 0.87,
3002
+ "grad_norm": 2.0206614417941258,
3003
+ "learning_rate": 1.7036140175988344e-05,
3004
+ "loss": 0.1471,
3005
+ "step": 4280
3006
+ },
3007
+ {
3008
+ "epoch": 0.87,
3009
+ "grad_norm": 2.047393965997251,
3010
+ "learning_rate": 1.702025409186211e-05,
3011
+ "loss": 0.1777,
3012
+ "step": 4290
3013
+ },
3014
+ {
3015
+ "epoch": 0.87,
3016
+ "grad_norm": 2.6683142062789713,
3017
+ "learning_rate": 1.7004332998103232e-05,
3018
+ "loss": 0.1769,
3019
+ "step": 4300
3020
+ },
3021
+ {
3022
+ "epoch": 0.87,
3023
+ "grad_norm": 1.6732439515126907,
3024
+ "learning_rate": 1.698837697412066e-05,
3025
+ "loss": 0.1268,
3026
+ "step": 4310
3027
+ },
3028
+ {
3029
+ "epoch": 0.88,
3030
+ "grad_norm": 3.001202183493077,
3031
+ "learning_rate": 1.697238609949757e-05,
3032
+ "loss": 0.1489,
3033
+ "step": 4320
3034
+ },
3035
+ {
3036
+ "epoch": 0.88,
3037
+ "grad_norm": 1.9957859824768167,
3038
+ "learning_rate": 1.6956360453990964e-05,
3039
+ "loss": 0.1536,
3040
+ "step": 4330
3041
+ },
3042
+ {
3043
+ "epoch": 0.88,
3044
+ "grad_norm": 3.2825840735727154,
3045
+ "learning_rate": 1.694030011753127e-05,
3046
+ "loss": 0.2101,
3047
+ "step": 4340
3048
+ },
3049
+ {
3050
+ "epoch": 0.88,
3051
+ "grad_norm": 2.0765138274517088,
3052
+ "learning_rate": 1.6924205170221933e-05,
3053
+ "loss": 0.1811,
3054
+ "step": 4350
3055
+ },
3056
+ {
3057
+ "epoch": 0.88,
3058
+ "grad_norm": 2.1133828976030595,
3059
+ "learning_rate": 1.6908075692339035e-05,
3060
+ "loss": 0.1728,
3061
+ "step": 4360
3062
+ },
3063
+ {
3064
+ "epoch": 0.89,
3065
+ "grad_norm": 1.1354696745558366,
3066
+ "learning_rate": 1.6891911764330887e-05,
3067
+ "loss": 0.1663,
3068
+ "step": 4370
3069
+ },
3070
+ {
3071
+ "epoch": 0.89,
3072
+ "grad_norm": 2.131963092613327,
3073
+ "learning_rate": 1.6875713466817608e-05,
3074
+ "loss": 0.1971,
3075
+ "step": 4380
3076
+ },
3077
+ {
3078
+ "epoch": 0.89,
3079
+ "grad_norm": 6.080414407059132,
3080
+ "learning_rate": 1.6859480880590755e-05,
3081
+ "loss": 0.1518,
3082
+ "step": 4390
3083
+ },
3084
+ {
3085
+ "epoch": 0.89,
3086
+ "grad_norm": 1.9313104885467733,
3087
+ "learning_rate": 1.684321408661291e-05,
3088
+ "loss": 0.1726,
3089
+ "step": 4400
3090
+ },
3091
+ {
3092
+ "epoch": 0.89,
3093
+ "grad_norm": 1.891911281896888,
3094
+ "learning_rate": 1.6826913166017257e-05,
3095
+ "loss": 0.2049,
3096
+ "step": 4410
3097
+ },
3098
+ {
3099
+ "epoch": 0.9,
3100
+ "grad_norm": 1.8833660404212225,
3101
+ "learning_rate": 1.68105782001072e-05,
3102
+ "loss": 0.1628,
3103
+ "step": 4420
3104
+ },
3105
+ {
3106
+ "epoch": 0.9,
3107
+ "grad_norm": 2.043852343315347,
3108
+ "learning_rate": 1.6794209270355946e-05,
3109
+ "loss": 0.1975,
3110
+ "step": 4430
3111
+ },
3112
+ {
3113
+ "epoch": 0.9,
3114
+ "grad_norm": 1.4598525807667742,
3115
+ "learning_rate": 1.677780645840611e-05,
3116
+ "loss": 0.159,
3117
+ "step": 4440
3118
+ },
3119
+ {
3120
+ "epoch": 0.9,
3121
+ "grad_norm": 1.6015094283079794,
3122
+ "learning_rate": 1.6761369846069292e-05,
3123
+ "loss": 0.157,
3124
+ "step": 4450
3125
+ },
3126
+ {
3127
+ "epoch": 0.9,
3128
+ "grad_norm": 2.0628558921093125,
3129
+ "learning_rate": 1.6744899515325674e-05,
3130
+ "loss": 0.1748,
3131
+ "step": 4460
3132
+ },
3133
+ {
3134
+ "epoch": 0.91,
3135
+ "grad_norm": 2.076881323364807,
3136
+ "learning_rate": 1.672839554832362e-05,
3137
+ "loss": 0.1966,
3138
+ "step": 4470
3139
+ },
3140
+ {
3141
+ "epoch": 0.91,
3142
+ "grad_norm": 2.0325935028828135,
3143
+ "learning_rate": 1.671185802737926e-05,
3144
+ "loss": 0.1885,
3145
+ "step": 4480
3146
+ },
3147
+ {
3148
+ "epoch": 0.91,
3149
+ "grad_norm": 2.280549242220261,
3150
+ "learning_rate": 1.6695287034976078e-05,
3151
+ "loss": 0.1624,
3152
+ "step": 4490
3153
+ },
3154
+ {
3155
+ "epoch": 0.91,
3156
+ "grad_norm": 2.08609705396923,
3157
+ "learning_rate": 1.6678682653764502e-05,
3158
+ "loss": 0.1631,
3159
+ "step": 4500
3160
+ },
3161
+ {
3162
+ "epoch": 0.91,
3163
+ "grad_norm": 2.2526237237040903,
3164
+ "learning_rate": 1.666204496656149e-05,
3165
+ "loss": 0.145,
3166
+ "step": 4510
3167
+ },
3168
+ {
3169
+ "epoch": 0.92,
3170
+ "grad_norm": 1.9007711000245981,
3171
+ "learning_rate": 1.6645374056350128e-05,
3172
+ "loss": 0.173,
3173
+ "step": 4520
3174
+ },
3175
+ {
3176
+ "epoch": 0.92,
3177
+ "grad_norm": 1.9372171995443488,
3178
+ "learning_rate": 1.6628670006279194e-05,
3179
+ "loss": 0.142,
3180
+ "step": 4530
3181
+ }
3182
+ ],
3183
+ "logging_steps": 10,
3184
+ "max_steps": 14808,
3185
+ "num_input_tokens_seen": 0,
3186
+ "num_train_epochs": 3,
3187
+ "save_steps": 4532,
3188
+ "total_flos": 0.0,
3189
+ "train_batch_size": 1,
3190
+ "trial_name": null,
3191
+ "trial_params": null
3192
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e5a9a77cca320ca6b5e0711cab13402d6db40c25a953099f160649f67f0e98c2
3
+ size 6712
zero_to_fp32.py ADDED
@@ -0,0 +1,592 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #!/usr/bin/env python
2
+
3
+ # Copyright (c) Microsoft Corporation.
4
+ # SPDX-License-Identifier: Apache-2.0
5
+
6
+ # DeepSpeed Team
7
+
8
+ # This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
9
+ # copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
10
+ # the future. Once extracted, the weights don't require DeepSpeed and can be used in any
11
+ # application.
12
+ #
13
+ # example: python zero_to_fp32.py . pytorch_model.bin
14
+
15
+ import argparse
16
+ import torch
17
+ import glob
18
+ import math
19
+ import os
20
+ import re
21
+ from collections import OrderedDict
22
+ from dataclasses import dataclass
23
+
24
+ # while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
25
+ # DeepSpeed data structures it has to be available in the current python environment.
26
+ from deepspeed.utils import logger
27
+ from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
28
+ FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
29
+ FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
30
+
31
+
32
+ @dataclass
33
+ class zero_model_state:
34
+ buffers: dict()
35
+ param_shapes: dict()
36
+ shared_params: list
37
+ ds_version: int
38
+ frozen_param_shapes: dict()
39
+ frozen_param_fragments: dict()
40
+
41
+
42
+ debug = 0
43
+
44
+ # load to cpu
45
+ device = torch.device('cpu')
46
+
47
+
48
+ def atoi(text):
49
+ return int(text) if text.isdigit() else text
50
+
51
+
52
+ def natural_keys(text):
53
+ '''
54
+ alist.sort(key=natural_keys) sorts in human order
55
+ http://nedbatchelder.com/blog/200712/human_sorting.html
56
+ (See Toothy's implementation in the comments)
57
+ '''
58
+ return [atoi(c) for c in re.split(r'(\d+)', text)]
59
+
60
+
61
+ def get_model_state_file(checkpoint_dir, zero_stage):
62
+ if not os.path.isdir(checkpoint_dir):
63
+ raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
64
+
65
+ # there should be only one file
66
+ if zero_stage <= 2:
67
+ file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
68
+ elif zero_stage == 3:
69
+ file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
70
+
71
+ if not os.path.exists(file):
72
+ raise FileNotFoundError(f"can't find model states file at '{file}'")
73
+
74
+ return file
75
+
76
+
77
+ def get_checkpoint_files(checkpoint_dir, glob_pattern):
78
+ # XXX: need to test that this simple glob rule works for multi-node setup too
79
+ ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
80
+
81
+ if len(ckpt_files) == 0:
82
+ raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
83
+
84
+ return ckpt_files
85
+
86
+
87
+ def get_optim_files(checkpoint_dir):
88
+ return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
89
+
90
+
91
+ def get_model_state_files(checkpoint_dir):
92
+ return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
93
+
94
+
95
+ def parse_model_states(files):
96
+ zero_model_states = []
97
+ for file in files:
98
+ state_dict = torch.load(file, map_location=device)
99
+
100
+ if BUFFER_NAMES not in state_dict:
101
+ raise ValueError(f"{file} is not a model state checkpoint")
102
+ buffer_names = state_dict[BUFFER_NAMES]
103
+ if debug:
104
+ print("Found buffers:", buffer_names)
105
+
106
+ # recover just the buffers while restoring them to fp32 if they were saved in fp16
107
+ buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
108
+ param_shapes = state_dict[PARAM_SHAPES]
109
+
110
+ # collect parameters that are included in param_shapes
111
+ param_names = []
112
+ for s in param_shapes:
113
+ for name in s.keys():
114
+ param_names.append(name)
115
+
116
+ # update with frozen parameters
117
+ frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
118
+ if frozen_param_shapes is not None:
119
+ if debug:
120
+ print(f"Found frozen_param_shapes: {frozen_param_shapes}")
121
+ param_names += list(frozen_param_shapes.keys())
122
+
123
+ # handle shared params
124
+ shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
125
+
126
+ ds_version = state_dict.get(DS_VERSION, None)
127
+
128
+ frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
129
+
130
+ z_model_state = zero_model_state(buffers=buffers,
131
+ param_shapes=param_shapes,
132
+ shared_params=shared_params,
133
+ ds_version=ds_version,
134
+ frozen_param_shapes=frozen_param_shapes,
135
+ frozen_param_fragments=frozen_param_fragments)
136
+ zero_model_states.append(z_model_state)
137
+
138
+ return zero_model_states
139
+
140
+
141
+ def parse_optim_states(files, ds_checkpoint_dir):
142
+
143
+ total_files = len(files)
144
+ state_dicts = []
145
+ for f in files:
146
+ state_dict = torch.load(f, map_location=device)
147
+ # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
148
+ # and also handle the case where it was already removed by another helper script
149
+ state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
150
+ state_dicts.append(state_dict)
151
+
152
+ if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
153
+ raise ValueError(f"{files[0]} is not a zero checkpoint")
154
+ zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
155
+ world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
156
+
157
+ # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
158
+ # parameters can be different from data parallelism for non-expert parameters. So we can just
159
+ # use the max of the partition_count to get the dp world_size.
160
+
161
+ if type(world_size) is list:
162
+ world_size = max(world_size)
163
+
164
+ if world_size != total_files:
165
+ raise ValueError(
166
+ f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
167
+ "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
168
+ )
169
+
170
+ # the groups are named differently in each stage
171
+ if zero_stage <= 2:
172
+ fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
173
+ elif zero_stage == 3:
174
+ fp32_groups_key = FP32_FLAT_GROUPS
175
+ else:
176
+ raise ValueError(f"unknown zero stage {zero_stage}")
177
+
178
+ if zero_stage <= 2:
179
+ fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
180
+ elif zero_stage == 3:
181
+ # if there is more than one param group, there will be multiple flattened tensors - one
182
+ # flattened tensor per group - for simplicity merge them into a single tensor
183
+ #
184
+ # XXX: could make the script more memory efficient for when there are multiple groups - it
185
+ # will require matching the sub-lists of param_shapes for each param group flattened tensor
186
+
187
+ fp32_flat_groups = [
188
+ torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
189
+ ]
190
+
191
+ return zero_stage, world_size, fp32_flat_groups
192
+
193
+
194
+ def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir):
195
+ """
196
+ Returns fp32 state_dict reconstructed from ds checkpoint
197
+
198
+ Args:
199
+ - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
200
+
201
+ """
202
+ print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
203
+
204
+ optim_files = get_optim_files(ds_checkpoint_dir)
205
+ zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
206
+ print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
207
+
208
+ model_files = get_model_state_files(ds_checkpoint_dir)
209
+
210
+ zero_model_states = parse_model_states(model_files)
211
+ print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
212
+
213
+ if zero_stage <= 2:
214
+ return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states)
215
+ elif zero_stage == 3:
216
+ return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states)
217
+
218
+
219
+ def _zero2_merge_frozen_params(state_dict, zero_model_states):
220
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
221
+ return
222
+
223
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
224
+ frozen_param_fragments = zero_model_states[0].frozen_param_fragments
225
+
226
+ if debug:
227
+ num_elem = sum(s.numel() for s in frozen_param_shapes.values())
228
+ print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
229
+
230
+ wanted_params = len(frozen_param_shapes)
231
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
232
+ avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
233
+ print(f'Frozen params: Have {avail_numel} numels to process.')
234
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
235
+
236
+ total_params = 0
237
+ total_numel = 0
238
+ for name, shape in frozen_param_shapes.items():
239
+ total_params += 1
240
+ unpartitioned_numel = shape.numel()
241
+ total_numel += unpartitioned_numel
242
+
243
+ state_dict[name] = frozen_param_fragments[name]
244
+
245
+ if debug:
246
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
247
+
248
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
249
+
250
+
251
+ def _has_callable(obj, fn):
252
+ attr = getattr(obj, fn, None)
253
+ return callable(attr)
254
+
255
+
256
+ def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
257
+ param_shapes = zero_model_states[0].param_shapes
258
+
259
+ # Reconstruction protocol:
260
+ #
261
+ # XXX: document this
262
+
263
+ if debug:
264
+ for i in range(world_size):
265
+ for j in range(len(fp32_flat_groups[0])):
266
+ print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
267
+
268
+ # XXX: memory usage doubles here (zero2)
269
+ num_param_groups = len(fp32_flat_groups[0])
270
+ merged_single_partition_of_fp32_groups = []
271
+ for i in range(num_param_groups):
272
+ merged_partitions = [sd[i] for sd in fp32_flat_groups]
273
+ full_single_fp32_vector = torch.cat(merged_partitions, 0)
274
+ merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
275
+ avail_numel = sum(
276
+ [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
277
+
278
+ if debug:
279
+ wanted_params = sum([len(shapes) for shapes in param_shapes])
280
+ wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
281
+ # not asserting if there is a mismatch due to possible padding
282
+ print(f"Have {avail_numel} numels to process.")
283
+ print(f"Need {wanted_numel} numels in {wanted_params} params.")
284
+
285
+ # params
286
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
287
+ # out-of-core computing solution
288
+ total_numel = 0
289
+ total_params = 0
290
+ for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
291
+ offset = 0
292
+ avail_numel = full_single_fp32_vector.numel()
293
+ for name, shape in shapes.items():
294
+
295
+ unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape)
296
+ total_numel += unpartitioned_numel
297
+ total_params += 1
298
+
299
+ if debug:
300
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
301
+ state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
302
+ offset += unpartitioned_numel
303
+
304
+ # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
305
+ # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
306
+ # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
307
+ # live optimizer object, so we are checking that the numbers are within the right range
308
+ align_to = 2 * world_size
309
+
310
+ def zero2_align(x):
311
+ return align_to * math.ceil(x / align_to)
312
+
313
+ if debug:
314
+ print(f"original offset={offset}, avail_numel={avail_numel}")
315
+
316
+ offset = zero2_align(offset)
317
+ avail_numel = zero2_align(avail_numel)
318
+
319
+ if debug:
320
+ print(f"aligned offset={offset}, avail_numel={avail_numel}")
321
+
322
+ # Sanity check
323
+ if offset != avail_numel:
324
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
325
+
326
+ print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
327
+
328
+
329
+ def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states):
330
+ state_dict = OrderedDict()
331
+
332
+ # buffers
333
+ buffers = zero_model_states[0].buffers
334
+ state_dict.update(buffers)
335
+ if debug:
336
+ print(f"added {len(buffers)} buffers")
337
+
338
+ _zero2_merge_frozen_params(state_dict, zero_model_states)
339
+
340
+ _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
341
+
342
+ # recover shared parameters
343
+ for pair in zero_model_states[0].shared_params:
344
+ if pair[1] in state_dict:
345
+ state_dict[pair[0]] = state_dict[pair[1]]
346
+
347
+ return state_dict
348
+
349
+
350
+ def zero3_partitioned_param_info(unpartitioned_numel, world_size):
351
+ remainder = unpartitioned_numel % world_size
352
+ padding_numel = (world_size - remainder) if remainder else 0
353
+ partitioned_numel = math.ceil(unpartitioned_numel / world_size)
354
+ return partitioned_numel, padding_numel
355
+
356
+
357
+ def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
358
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
359
+ return
360
+
361
+ if debug:
362
+ for i in range(world_size):
363
+ num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
364
+ print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
365
+
366
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
367
+ wanted_params = len(frozen_param_shapes)
368
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
369
+ avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
370
+ print(f'Frozen params: Have {avail_numel} numels to process.')
371
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
372
+
373
+ total_params = 0
374
+ total_numel = 0
375
+ for name, shape in zero_model_states[0].frozen_param_shapes.items():
376
+ total_params += 1
377
+ unpartitioned_numel = shape.numel()
378
+ total_numel += unpartitioned_numel
379
+
380
+ param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
381
+ state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
382
+
383
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
384
+
385
+ if debug:
386
+ print(
387
+ f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
388
+ )
389
+
390
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
391
+
392
+
393
+ def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
394
+ param_shapes = zero_model_states[0].param_shapes
395
+ avail_numel = fp32_flat_groups[0].numel() * world_size
396
+ # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
397
+ # param, re-consolidating each param, while dealing with padding if any
398
+
399
+ # merge list of dicts, preserving order
400
+ param_shapes = {k: v for d in param_shapes for k, v in d.items()}
401
+
402
+ if debug:
403
+ for i in range(world_size):
404
+ print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
405
+
406
+ wanted_params = len(param_shapes)
407
+ wanted_numel = sum(shape.numel() for shape in param_shapes.values())
408
+ # not asserting if there is a mismatch due to possible padding
409
+ avail_numel = fp32_flat_groups[0].numel() * world_size
410
+ print(f"Trainable params: Have {avail_numel} numels to process.")
411
+ print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
412
+
413
+ # params
414
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
415
+ # out-of-core computing solution
416
+ offset = 0
417
+ total_numel = 0
418
+ total_params = 0
419
+ for name, shape in param_shapes.items():
420
+
421
+ unpartitioned_numel = shape.numel()
422
+ total_numel += unpartitioned_numel
423
+ total_params += 1
424
+
425
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
426
+
427
+ if debug:
428
+ print(
429
+ f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
430
+ )
431
+
432
+ # XXX: memory usage doubles here
433
+ state_dict[name] = torch.cat(
434
+ tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
435
+ 0).narrow(0, 0, unpartitioned_numel).view(shape)
436
+ offset += partitioned_numel
437
+
438
+ offset *= world_size
439
+
440
+ # Sanity check
441
+ if offset != avail_numel:
442
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
443
+
444
+ print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
445
+
446
+
447
+ def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states):
448
+ state_dict = OrderedDict()
449
+
450
+ # buffers
451
+ buffers = zero_model_states[0].buffers
452
+ state_dict.update(buffers)
453
+ if debug:
454
+ print(f"added {len(buffers)} buffers")
455
+
456
+ _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
457
+
458
+ _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
459
+
460
+ # recover shared parameters
461
+ for pair in zero_model_states[0].shared_params:
462
+ if pair[1] in state_dict:
463
+ state_dict[pair[0]] = state_dict[pair[1]]
464
+
465
+ return state_dict
466
+
467
+
468
+ def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None):
469
+ """
470
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
471
+ ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
472
+ via a model hub.
473
+
474
+ Args:
475
+ - ``checkpoint_dir``: path to the desired checkpoint folder
476
+ - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
477
+
478
+ Returns:
479
+ - pytorch ``state_dict``
480
+
481
+ Note: this approach may not work if your application doesn't have sufficient free CPU memory and
482
+ you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
483
+ the checkpoint.
484
+
485
+ A typical usage might be ::
486
+
487
+ from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
488
+ # do the training and checkpoint saving
489
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
490
+ model = model.cpu() # move to cpu
491
+ model.load_state_dict(state_dict)
492
+ # submit to model hub or save the model to share with others
493
+
494
+ In this example the ``model`` will no longer be usable in the deepspeed context of the same
495
+ application. i.e. you will need to re-initialize the deepspeed engine, since
496
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
497
+
498
+ If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
499
+
500
+ """
501
+ if tag is None:
502
+ latest_path = os.path.join(checkpoint_dir, 'latest')
503
+ if os.path.isfile(latest_path):
504
+ with open(latest_path, 'r') as fd:
505
+ tag = fd.read().strip()
506
+ else:
507
+ raise ValueError(f"Unable to find 'latest' file at {latest_path}")
508
+
509
+ ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
510
+
511
+ if not os.path.isdir(ds_checkpoint_dir):
512
+ raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
513
+
514
+ return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir)
515
+
516
+
517
+ def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None):
518
+ """
519
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
520
+ loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
521
+
522
+ Args:
523
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
524
+ - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin)
525
+ - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
526
+ """
527
+
528
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
529
+ print(f"Saving fp32 state dict to {output_file}")
530
+ torch.save(state_dict, output_file)
531
+
532
+
533
+ def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
534
+ """
535
+ 1. Put the provided model to cpu
536
+ 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
537
+ 3. Load it into the provided model
538
+
539
+ Args:
540
+ - ``model``: the model object to update
541
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
542
+ - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
543
+
544
+ Returns:
545
+ - ``model`: modified model
546
+
547
+ Make sure you have plenty of CPU memory available before you call this function. If you don't
548
+ have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
549
+ conveniently placed for you in the checkpoint folder.
550
+
551
+ A typical usage might be ::
552
+
553
+ from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
554
+ model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
555
+ # submit to model hub or save the model to share with others
556
+
557
+ Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
558
+ of the same application. i.e. you will need to re-initialize the deepspeed engine, since
559
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
560
+
561
+ """
562
+ logger.info(f"Extracting fp32 weights")
563
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
564
+
565
+ logger.info(f"Overwriting model with fp32 weights")
566
+ model = model.cpu()
567
+ model.load_state_dict(state_dict, strict=False)
568
+
569
+ return model
570
+
571
+
572
+ if __name__ == "__main__":
573
+
574
+ parser = argparse.ArgumentParser()
575
+ parser.add_argument("checkpoint_dir",
576
+ type=str,
577
+ help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
578
+ parser.add_argument(
579
+ "output_file",
580
+ type=str,
581
+ help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)")
582
+ parser.add_argument("-t",
583
+ "--tag",
584
+ type=str,
585
+ default=None,
586
+ help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
587
+ parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
588
+ args = parser.parse_args()
589
+
590
+ debug = args.debug
591
+
592
+ convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, args.output_file, tag=args.tag)