Hugging Face's logo

dophys
/
bge-m3_finetuned_for_physics

Sentence Similarity
sentence-transformers
TensorBoard
Safetensors
xlm-roberta
feature-extraction
text-embeddings-inference
Model card Files
xet
Metrics Community
dophys commited on
Commit
83f6dbf
·
verified ·
1 Parent(s): cdc9691

Upload folder using huggingface_hub

Browse files
Files changed (39) hide show
  1. .gitattributes +2 -0
  2. 1_Pooling/config.json +10 -0
  3. README.md +144 -0
  4. checkpoint-853/1_Pooling/config.json +10 -0
  5. checkpoint-853/README.md +144 -0
  6. checkpoint-853/config.json +28 -0
  7. checkpoint-853/config_sentence_transformers.json +10 -0
  8. checkpoint-853/global_step853/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt +3 -0
  9. checkpoint-853/global_step853/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt +3 -0
  10. checkpoint-853/global_step853/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt +3 -0
  11. checkpoint-853/global_step853/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt +3 -0
  12. checkpoint-853/global_step853/mp_rank_00_model_states.pt +3 -0
  13. checkpoint-853/latest +1 -0
  14. checkpoint-853/model.safetensors +3 -0
  15. checkpoint-853/modules.json +20 -0
  16. checkpoint-853/rng_state_0.pth +3 -0
  17. checkpoint-853/rng_state_1.pth +3 -0
  18. checkpoint-853/rng_state_2.pth +3 -0
  19. checkpoint-853/rng_state_3.pth +3 -0
  20. checkpoint-853/sentence_bert_config.json +4 -0
  21. checkpoint-853/sentencepiece.bpe.model +3 -0
  22. checkpoint-853/special_tokens_map.json +51 -0
  23. checkpoint-853/tokenizer.json +3 -0
  24. checkpoint-853/tokenizer_config.json +55 -0
  25. checkpoint-853/trainer_state.json +3015 -0
  26. checkpoint-853/training_args.bin +3 -0
  27. checkpoint-853/zero_to_fp32.py +604 -0
  28. config.json +28 -0
  29. config_sentence_transformers.json +10 -0
  30. model.safetensors +3 -0
  31. modules.json +20 -0
  32. runs/Aug22_17-17-24_autodl-container-c024408f5d-9bcd732d/events.out.tfevents.1724318254.autodl-container-c024408f5d-9bcd732d.5345.0 +3 -0
  33. runs/Aug22_17-18-40_autodl-container-c024408f5d-9bcd732d/events.out.tfevents.1724318333.autodl-container-c024408f5d-9bcd732d.6318.0 +3 -0
  34. sentence_bert_config.json +4 -0
  35. sentencepiece.bpe.model +3 -0
  36. special_tokens_map.json +51 -0
  37. tokenizer.json +3 -0
  38. tokenizer_config.json +55 -0
  39. training_args.bin +3 -0
.gitattributes CHANGED
@@ -33,3 +33,5 @@ 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
+ checkpoint-853/tokenizer.json filter=lfs diff=lfs merge=lfs -text
37
+ 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 +1,144 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ datasets: []
3
+ language: []
4
+ library_name: sentence-transformers
5
+ pipeline_tag: sentence-similarity
6
+ tags:
7
+ - sentence-transformers
8
+ - sentence-similarity
9
+ - feature-extraction
10
+ widget: []
11
+ ---
12
+
13
+ # SentenceTransformer
14
+
15
+ This is a [sentence-transformers](https://www.SBERT.net) model trained. It maps sentences & paragraphs to a 1024-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more.
16
+
17
+ ## Model Details
18
+
19
+ ### Model Description
20
+ - **Model Type:** Sentence Transformer
21
+ <!-- - **Base model:** [Unknown](https://huggingface.co/unknown) -->
22
+ - **Maximum Sequence Length:** 8192 tokens
23
+ - **Output Dimensionality:** 1024 tokens
24
+ - **Similarity Function:** Cosine Similarity
25
+ <!-- - **Training Dataset:** Unknown -->
26
+ <!-- - **Language:** Unknown -->
27
+ <!-- - **License:** Unknown -->
28
+
29
+ ### Model Sources
30
+
31
+ - **Documentation:** [Sentence Transformers Documentation](https://sbert.net)
32
+ - **Repository:** [Sentence Transformers on GitHub](https://github.com/UKPLab/sentence-transformers)
33
+ - **Hugging Face:** [Sentence Transformers on Hugging Face](https://huggingface.co/models?library=sentence-transformers)
34
+
35
+ ### Full Model Architecture
36
+
37
+ ```
38
+ SentenceTransformer(
39
+ (0): Transformer({'max_seq_length': 8192, 'do_lower_case': False}) with Transformer model: XLMRobertaModel
40
+ (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})
41
+ (2): Normalize()
42
+ )
43
+ ```
44
+
45
+ ## Usage
46
+
47
+ ### Direct Usage (Sentence Transformers)
48
+
49
+ First install the Sentence Transformers library:
50
+
51
+ ```bash
52
+ pip install -U sentence-transformers
53
+ ```
54
+
55
+ Then you can load this model and run inference.
56
+ ```python
57
+ from sentence_transformers import SentenceTransformer
58
+
59
+ # Download from the 🤗 Hub
60
+ model = SentenceTransformer("sentence_transformers_model_id")
61
+ # Run inference
62
+ sentences = [
63
+ 'The weather is lovely today.',
64
+ "It's so sunny outside!",
65
+ 'He drove to the stadium.',
66
+ ]
67
+ embeddings = model.encode(sentences)
68
+ print(embeddings.shape)
69
+ # [3, 1024]
70
+
71
+ # Get the similarity scores for the embeddings
72
+ similarities = model.similarity(embeddings, embeddings)
73
+ print(similarities.shape)
74
+ # [3, 3]
75
+ ```
76
+
77
+ <!--
78
+ ### Direct Usage (Transformers)
79
+
80
+ <details><summary>Click to see the direct usage in Transformers</summary>
81
+
82
+ </details>
83
+ -->
84
+
85
+ <!--
86
+ ### Downstream Usage (Sentence Transformers)
87
+
88
+ You can finetune this model on your own dataset.
89
+
90
+ <details><summary>Click to expand</summary>
91
+
92
+ </details>
93
+ -->
94
+
95
+ <!--
96
+ ### Out-of-Scope Use
97
+
98
+ *List how the model may foreseeably be misused and address what users ought not to do with the model.*
99
+ -->
100
+
101
+ <!--
102
+ ## Bias, Risks and Limitations
103
+
104
+ *What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model.*
105
+ -->
106
+
107
+ <!--
108
+ ### Recommendations
109
+
110
+ *What are recommendations with respect to the foreseeable issues? For example, filtering explicit content.*
111
+ -->
112
+
113
+ ## Training Details
114
+
115
+ ### Framework Versions
116
+ - Python: 3.12.3
117
+ - Sentence Transformers: 3.0.1
118
+ - Transformers: 4.42.1
119
+ - PyTorch: 2.3.0+cu121
120
+ - Accelerate: 0.31.0
121
+ - Datasets: 2.20.0
122
+ - Tokenizers: 0.19.1
123
+
124
+ ## Citation
125
+
126
+ ### BibTeX
127
+
128
+ <!--
129
+ ## Glossary
130
+
131
+ *Clearly define terms in order to be accessible across audiences.*
132
+ -->
133
+
134
+ <!--
135
+ ## Model Card Authors
136
+
137
+ *Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction.*
138
+ -->
139
+
140
+ <!--
141
+ ## Model Card Contact
142
+
143
+ *Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors.*
144
+ -->
checkpoint-853/1_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
+ }
checkpoint-853/README.md ADDED
@@ -0,0 +1,144 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ datasets: []
3
+ language: []
4
+ library_name: sentence-transformers
5
+ pipeline_tag: sentence-similarity
6
+ tags:
7
+ - sentence-transformers
8
+ - sentence-similarity
9
+ - feature-extraction
10
+ widget: []
11
+ ---
12
+
13
+ # SentenceTransformer
14
+
15
+ This is a [sentence-transformers](https://www.SBERT.net) model trained. It maps sentences & paragraphs to a 1024-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more.
16
+
17
+ ## Model Details
18
+
19
+ ### Model Description
20
+ - **Model Type:** Sentence Transformer
21
+ <!-- - **Base model:** [Unknown](https://huggingface.co/unknown) -->
22
+ - **Maximum Sequence Length:** 8192 tokens
23
+ - **Output Dimensionality:** 1024 tokens
24
+ - **Similarity Function:** Cosine Similarity
25
+ <!-- - **Training Dataset:** Unknown -->
26
+ <!-- - **Language:** Unknown -->
27
+ <!-- - **License:** Unknown -->
28
+
29
+ ### Model Sources
30
+
31
+ - **Documentation:** [Sentence Transformers Documentation](https://sbert.net)
32
+ - **Repository:** [Sentence Transformers on GitHub](https://github.com/UKPLab/sentence-transformers)
33
+ - **Hugging Face:** [Sentence Transformers on Hugging Face](https://huggingface.co/models?library=sentence-transformers)
34
+
35
+ ### Full Model Architecture
36
+
37
+ ```
38
+ SentenceTransformer(
39
+ (0): Transformer({'max_seq_length': 8192, 'do_lower_case': False}) with Transformer model: XLMRobertaModel
40
+ (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})
41
+ (2): Normalize()
42
+ )
43
+ ```
44
+
45
+ ## Usage
46
+
47
+ ### Direct Usage (Sentence Transformers)
48
+
49
+ First install the Sentence Transformers library:
50
+
51
+ ```bash
52
+ pip install -U sentence-transformers
53
+ ```
54
+
55
+ Then you can load this model and run inference.
56
+ ```python
57
+ from sentence_transformers import SentenceTransformer
58
+
59
+ # Download from the 🤗 Hub
60
+ model = SentenceTransformer("sentence_transformers_model_id")
61
+ # Run inference
62
+ sentences = [
63
+ 'The weather is lovely today.',
64
+ "It's so sunny outside!",
65
+ 'He drove to the stadium.',
66
+ ]
67
+ embeddings = model.encode(sentences)
68
+ print(embeddings.shape)
69
+ # [3, 1024]
70
+
71
+ # Get the similarity scores for the embeddings
72
+ similarities = model.similarity(embeddings, embeddings)
73
+ print(similarities.shape)
74
+ # [3, 3]
75
+ ```
76
+
77
+ <!--
78
+ ### Direct Usage (Transformers)
79
+
80
+ <details><summary>Click to see the direct usage in Transformers</summary>
81
+
82
+ </details>
83
+ -->
84
+
85
+ <!--
86
+ ### Downstream Usage (Sentence Transformers)
87
+
88
+ You can finetune this model on your own dataset.
89
+
90
+ <details><summary>Click to expand</summary>
91
+
92
+ </details>
93
+ -->
94
+
95
+ <!--
96
+ ### Out-of-Scope Use
97
+
98
+ *List how the model may foreseeably be misused and address what users ought not to do with the model.*
99
+ -->
100
+
101
+ <!--
102
+ ## Bias, Risks and Limitations
103
+
104
+ *What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model.*
105
+ -->
106
+
107
+ <!--
108
+ ### Recommendations
109
+
110
+ *What are recommendations with respect to the foreseeable issues? For example, filtering explicit content.*
111
+ -->
112
+
113
+ ## Training Details
114
+
115
+ ### Framework Versions
116
+ - Python: 3.12.3
117
+ - Sentence Transformers: 3.0.1
118
+ - Transformers: 4.42.1
119
+ - PyTorch: 2.3.0+cu121
120
+ - Accelerate: 0.31.0
121
+ - Datasets: 2.20.0
122
+ - Tokenizers: 0.19.1
123
+
124
+ ## Citation
125
+
126
+ ### BibTeX
127
+
128
+ <!--
129
+ ## Glossary
130
+
131
+ *Clearly define terms in order to be accessible across audiences.*
132
+ -->
133
+
134
+ <!--
135
+ ## Model Card Authors
136
+
137
+ *Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction.*
138
+ -->
139
+
140
+ <!--
141
+ ## Model Card Contact
142
+
143
+ *Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors.*
144
+ -->
checkpoint-853/config.json ADDED
@@ -0,0 +1,28 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "/root/autodl-tmp/bge-m3_r4/checkpoint-853",
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": "4.42.1",
25
+ "type_vocab_size": 1,
26
+ "use_cache": true,
27
+ "vocab_size": 250002
28
+ }
checkpoint-853/config_sentence_transformers.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "__version__": {
3
+ "sentence_transformers": "3.0.1",
4
+ "transformers": "4.42.1",
5
+ "pytorch": "2.3.0+cu121"
6
+ },
7
+ "prompts": {},
8
+ "default_prompt_name": null,
9
+ "similarity_fn_name": null
10
+ }
checkpoint-853/global_step853/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ac25212b0dd7c1c33af9d13534f23a2f21eab85bf4c9e2e0f1128572d543b424
3
+ size 1703267856
checkpoint-853/global_step853/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8bdff9ef70123a1f1f4dad2085d3a4b06230abdf3586968517c8dc6aa38cbf9f
3
+ size 1703270288
checkpoint-853/global_step853/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f1c03c434680e35f1b5c8e28cdcf1cd0576573eaca6204bf2443374bb66ee115
3
+ size 1703283152
checkpoint-853/global_step853/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9eb0bf36954e8652094b0ae7bddbb405e2158e63dbeaad5c4b3db557401ede14
3
+ size 1703283472
checkpoint-853/global_step853/mp_rank_00_model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:85e1bc9c6dfde3f944cdedf1fa2ff240aaba02f760d9b2d19a313a37ad62e470
3
+ size 1135627884
checkpoint-853/latest ADDED
@@ -0,0 +1 @@
 
 
1
+ global_step853
checkpoint-853/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5ff2b3be09c7552fc58248f097a32771e376f56eb50737f93e0f41cef389d71d
3
+ size 2271064456
checkpoint-853/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
+ ]
checkpoint-853/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ea33f4d3831c1aa0fff107f94bdb1e3f44f9720a90a553c21c0e244b5b8e09f0
3
+ size 14960
checkpoint-853/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:605ff35753da8ab8637f8d7e383db916a3b20e716fe7662e9f25bf8f312a8a16
3
+ size 14960
checkpoint-853/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b8113ead6f45e448461169ca78aec213478a5414d0f423bf84379df9f5363aea
3
+ size 14960
checkpoint-853/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:86bee34831933d6a7c1c2561a8aedecde5c4cd60815ddd9e2a1b33907cc7843f
3
+ size 14960
checkpoint-853/sentence_bert_config.json ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ {
2
+ "max_seq_length": 8192,
3
+ "do_lower_case": false
4
+ }
checkpoint-853/sentencepiece.bpe.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cfc8146abe2a0488e9e2a0c56de7952f7c11ab059eca145a0a727afce0db2865
3
+ size 5069051
checkpoint-853/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
+ }
checkpoint-853/tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b74659c780d49afad7a7b9799868f75cbd3014fb6c34956e85a793028d38094a
3
+ size 17098251
checkpoint-853/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
+ }
checkpoint-853/trainer_state.json ADDED
@@ -0,0 +1,3015 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 0.9994141769185706,
5
+ "eval_steps": 500,
6
+ "global_step": 853,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.0023432923257176333,
13
+ "grad_norm": 0.00011052378977183253,
14
+ "learning_rate": 5e-06,
15
+ "loss": 0.0,
16
+ "step": 2
17
+ },
18
+ {
19
+ "epoch": 0.0046865846514352666,
20
+ "grad_norm": 0.00020697808940894902,
21
+ "learning_rate": 4.9941245593419514e-06,
22
+ "loss": 0.0,
23
+ "step": 4
24
+ },
25
+ {
26
+ "epoch": 0.007029876977152899,
27
+ "grad_norm": 0.0012532881228253245,
28
+ "learning_rate": 4.982373678025853e-06,
29
+ "loss": 0.0,
30
+ "step": 6
31
+ },
32
+ {
33
+ "epoch": 0.009373169302870533,
34
+ "grad_norm": 0.0008086035377345979,
35
+ "learning_rate": 4.970622796709754e-06,
36
+ "loss": 0.0,
37
+ "step": 8
38
+ },
39
+ {
40
+ "epoch": 0.011716461628588167,
41
+ "grad_norm": 0.0021155672147870064,
42
+ "learning_rate": 4.958871915393655e-06,
43
+ "loss": 0.0,
44
+ "step": 10
45
+ },
46
+ {
47
+ "epoch": 0.014059753954305799,
48
+ "grad_norm": 0.0012233309680595994,
49
+ "learning_rate": 4.947121034077556e-06,
50
+ "loss": 0.0,
51
+ "step": 12
52
+ },
53
+ {
54
+ "epoch": 0.016403046280023433,
55
+ "grad_norm": 0.0027737286873161793,
56
+ "learning_rate": 4.9353701527614576e-06,
57
+ "loss": 0.0,
58
+ "step": 14
59
+ },
60
+ {
61
+ "epoch": 0.018746338605741066,
62
+ "grad_norm": 0.0042906939052045345,
63
+ "learning_rate": 4.923619271445359e-06,
64
+ "loss": 0.0,
65
+ "step": 16
66
+ },
67
+ {
68
+ "epoch": 0.0210896309314587,
69
+ "grad_norm": 0.0005172386299818754,
70
+ "learning_rate": 4.91186839012926e-06,
71
+ "loss": 0.0,
72
+ "step": 18
73
+ },
74
+ {
75
+ "epoch": 0.023432923257176334,
76
+ "grad_norm": 0.002410772955045104,
77
+ "learning_rate": 4.900117508813161e-06,
78
+ "loss": 0.0,
79
+ "step": 20
80
+ },
81
+ {
82
+ "epoch": 0.025776215582893967,
83
+ "grad_norm": 0.6443753242492676,
84
+ "learning_rate": 4.8883666274970625e-06,
85
+ "loss": 0.0027,
86
+ "step": 22
87
+ },
88
+ {
89
+ "epoch": 0.028119507908611598,
90
+ "grad_norm": 0.004394118674099445,
91
+ "learning_rate": 4.876615746180964e-06,
92
+ "loss": 0.0001,
93
+ "step": 24
94
+ },
95
+ {
96
+ "epoch": 0.03046280023432923,
97
+ "grad_norm": 0.006466630846261978,
98
+ "learning_rate": 4.864864864864866e-06,
99
+ "loss": 0.0001,
100
+ "step": 26
101
+ },
102
+ {
103
+ "epoch": 0.032806092560046865,
104
+ "grad_norm": 0.011924203485250473,
105
+ "learning_rate": 4.853113983548767e-06,
106
+ "loss": 0.0001,
107
+ "step": 28
108
+ },
109
+ {
110
+ "epoch": 0.0351493848857645,
111
+ "grad_norm": 0.23746930062770844,
112
+ "learning_rate": 4.841363102232668e-06,
113
+ "loss": 0.0001,
114
+ "step": 30
115
+ },
116
+ {
117
+ "epoch": 0.03749267721148213,
118
+ "grad_norm": 0.0031001348979771137,
119
+ "learning_rate": 4.8296122209165694e-06,
120
+ "loss": 0.0,
121
+ "step": 32
122
+ },
123
+ {
124
+ "epoch": 0.03983596953719976,
125
+ "grad_norm": 0.0029028633143752813,
126
+ "learning_rate": 4.817861339600471e-06,
127
+ "loss": 0.0,
128
+ "step": 34
129
+ },
130
+ {
131
+ "epoch": 0.0421792618629174,
132
+ "grad_norm": 0.014626468531787395,
133
+ "learning_rate": 4.806110458284372e-06,
134
+ "loss": 0.0001,
135
+ "step": 36
136
+ },
137
+ {
138
+ "epoch": 0.04452255418863503,
139
+ "grad_norm": 0.001155451056547463,
140
+ "learning_rate": 4.794359576968273e-06,
141
+ "loss": 0.0,
142
+ "step": 38
143
+ },
144
+ {
145
+ "epoch": 0.04686584651435267,
146
+ "grad_norm": 0.003476829966530204,
147
+ "learning_rate": 4.782608695652174e-06,
148
+ "loss": 0.0,
149
+ "step": 40
150
+ },
151
+ {
152
+ "epoch": 0.0492091388400703,
153
+ "grad_norm": 0.0002227002551080659,
154
+ "learning_rate": 4.7708578143360756e-06,
155
+ "loss": 0.0,
156
+ "step": 42
157
+ },
158
+ {
159
+ "epoch": 0.051552431165787935,
160
+ "grad_norm": 0.0001427282695658505,
161
+ "learning_rate": 4.759106933019977e-06,
162
+ "loss": 0.0,
163
+ "step": 44
164
+ },
165
+ {
166
+ "epoch": 0.053895723491505565,
167
+ "grad_norm": 0.0027408564928919077,
168
+ "learning_rate": 4.747356051703878e-06,
169
+ "loss": 0.0002,
170
+ "step": 46
171
+ },
172
+ {
173
+ "epoch": 0.056239015817223195,
174
+ "grad_norm": 0.0020253027323633432,
175
+ "learning_rate": 4.735605170387779e-06,
176
+ "loss": 0.0,
177
+ "step": 48
178
+ },
179
+ {
180
+ "epoch": 0.05858230814294083,
181
+ "grad_norm": 0.001760220737196505,
182
+ "learning_rate": 4.723854289071681e-06,
183
+ "loss": 0.0,
184
+ "step": 50
185
+ },
186
+ {
187
+ "epoch": 0.06092560046865846,
188
+ "grad_norm": 0.0010492791188880801,
189
+ "learning_rate": 4.7121034077555825e-06,
190
+ "loss": 0.0,
191
+ "step": 52
192
+ },
193
+ {
194
+ "epoch": 0.0632688927943761,
195
+ "grad_norm": 0.002001305343583226,
196
+ "learning_rate": 4.700352526439484e-06,
197
+ "loss": 0.0,
198
+ "step": 54
199
+ },
200
+ {
201
+ "epoch": 0.06561218512009373,
202
+ "grad_norm": 0.18566887080669403,
203
+ "learning_rate": 4.688601645123384e-06,
204
+ "loss": 0.0009,
205
+ "step": 56
206
+ },
207
+ {
208
+ "epoch": 0.06795547744581136,
209
+ "grad_norm": 0.0009072807151824236,
210
+ "learning_rate": 4.676850763807285e-06,
211
+ "loss": 0.0,
212
+ "step": 58
213
+ },
214
+ {
215
+ "epoch": 0.070298769771529,
216
+ "grad_norm": 0.003983665257692337,
217
+ "learning_rate": 4.665099882491187e-06,
218
+ "loss": 0.0006,
219
+ "step": 60
220
+ },
221
+ {
222
+ "epoch": 0.07264206209724663,
223
+ "grad_norm": 0.01946200616657734,
224
+ "learning_rate": 4.653349001175089e-06,
225
+ "loss": 0.0001,
226
+ "step": 62
227
+ },
228
+ {
229
+ "epoch": 0.07498535442296426,
230
+ "grad_norm": 0.004048655740916729,
231
+ "learning_rate": 4.64159811985899e-06,
232
+ "loss": 0.0,
233
+ "step": 64
234
+ },
235
+ {
236
+ "epoch": 0.0773286467486819,
237
+ "grad_norm": 0.0005872617475688457,
238
+ "learning_rate": 4.629847238542891e-06,
239
+ "loss": 0.0001,
240
+ "step": 66
241
+ },
242
+ {
243
+ "epoch": 0.07967193907439953,
244
+ "grad_norm": 0.008831903338432312,
245
+ "learning_rate": 4.618096357226792e-06,
246
+ "loss": 0.0001,
247
+ "step": 68
248
+ },
249
+ {
250
+ "epoch": 0.08201523140011717,
251
+ "grad_norm": 0.006819219794124365,
252
+ "learning_rate": 4.6063454759106936e-06,
253
+ "loss": 0.0001,
254
+ "step": 70
255
+ },
256
+ {
257
+ "epoch": 0.0843585237258348,
258
+ "grad_norm": 0.0007863900391384959,
259
+ "learning_rate": 4.594594594594596e-06,
260
+ "loss": 0.0,
261
+ "step": 72
262
+ },
263
+ {
264
+ "epoch": 0.08670181605155243,
265
+ "grad_norm": 0.032210394740104675,
266
+ "learning_rate": 4.582843713278496e-06,
267
+ "loss": 0.0001,
268
+ "step": 74
269
+ },
270
+ {
271
+ "epoch": 0.08904510837727006,
272
+ "grad_norm": 0.2614983916282654,
273
+ "learning_rate": 4.571092831962397e-06,
274
+ "loss": 0.0008,
275
+ "step": 76
276
+ },
277
+ {
278
+ "epoch": 0.0913884007029877,
279
+ "grad_norm": 0.0012551415711641312,
280
+ "learning_rate": 4.5593419506462985e-06,
281
+ "loss": 0.0,
282
+ "step": 78
283
+ },
284
+ {
285
+ "epoch": 0.09373169302870533,
286
+ "grad_norm": 0.0019108065171167254,
287
+ "learning_rate": 4.5475910693302e-06,
288
+ "loss": 0.0,
289
+ "step": 80
290
+ },
291
+ {
292
+ "epoch": 0.09607498535442296,
293
+ "grad_norm": 0.02294810675084591,
294
+ "learning_rate": 4.535840188014101e-06,
295
+ "loss": 0.0001,
296
+ "step": 82
297
+ },
298
+ {
299
+ "epoch": 0.0984182776801406,
300
+ "grad_norm": 0.0012388118775561452,
301
+ "learning_rate": 4.524089306698003e-06,
302
+ "loss": 0.0,
303
+ "step": 84
304
+ },
305
+ {
306
+ "epoch": 0.10076157000585823,
307
+ "grad_norm": 0.001227575121447444,
308
+ "learning_rate": 4.512338425381904e-06,
309
+ "loss": 0.0001,
310
+ "step": 86
311
+ },
312
+ {
313
+ "epoch": 0.10310486233157587,
314
+ "grad_norm": 0.004755712114274502,
315
+ "learning_rate": 4.5005875440658054e-06,
316
+ "loss": 0.0001,
317
+ "step": 88
318
+ },
319
+ {
320
+ "epoch": 0.1054481546572935,
321
+ "grad_norm": 0.00837083999067545,
322
+ "learning_rate": 4.488836662749707e-06,
323
+ "loss": 0.0001,
324
+ "step": 90
325
+ },
326
+ {
327
+ "epoch": 0.10779144698301113,
328
+ "grad_norm": 0.48219314217567444,
329
+ "learning_rate": 4.477085781433608e-06,
330
+ "loss": 0.0017,
331
+ "step": 92
332
+ },
333
+ {
334
+ "epoch": 0.11013473930872876,
335
+ "grad_norm": 0.022060217335820198,
336
+ "learning_rate": 4.465334900117509e-06,
337
+ "loss": 0.0001,
338
+ "step": 94
339
+ },
340
+ {
341
+ "epoch": 0.11247803163444639,
342
+ "grad_norm": 0.0019385352497920394,
343
+ "learning_rate": 4.45358401880141e-06,
344
+ "loss": 0.0,
345
+ "step": 96
346
+ },
347
+ {
348
+ "epoch": 0.11482132396016403,
349
+ "grad_norm": 0.01225442998111248,
350
+ "learning_rate": 4.4418331374853116e-06,
351
+ "loss": 0.0001,
352
+ "step": 98
353
+ },
354
+ {
355
+ "epoch": 0.11716461628588166,
356
+ "grad_norm": 0.0005759520572610199,
357
+ "learning_rate": 4.430082256169213e-06,
358
+ "loss": 0.0,
359
+ "step": 100
360
+ },
361
+ {
362
+ "epoch": 0.1195079086115993,
363
+ "grad_norm": 0.02452813647687435,
364
+ "learning_rate": 4.418331374853114e-06,
365
+ "loss": 0.0001,
366
+ "step": 102
367
+ },
368
+ {
369
+ "epoch": 0.12185120093731693,
370
+ "grad_norm": 0.0078084710985422134,
371
+ "learning_rate": 4.406580493537015e-06,
372
+ "loss": 0.0001,
373
+ "step": 104
374
+ },
375
+ {
376
+ "epoch": 0.12419449326303457,
377
+ "grad_norm": 0.004263446666300297,
378
+ "learning_rate": 4.394829612220917e-06,
379
+ "loss": 0.0001,
380
+ "step": 106
381
+ },
382
+ {
383
+ "epoch": 0.1265377855887522,
384
+ "grad_norm": 0.0016304058954119682,
385
+ "learning_rate": 4.3830787309048185e-06,
386
+ "loss": 0.0001,
387
+ "step": 108
388
+ },
389
+ {
390
+ "epoch": 0.12888107791446984,
391
+ "grad_norm": 0.011672005988657475,
392
+ "learning_rate": 4.37132784958872e-06,
393
+ "loss": 0.0002,
394
+ "step": 110
395
+ },
396
+ {
397
+ "epoch": 0.13122437024018746,
398
+ "grad_norm": 0.002603155327960849,
399
+ "learning_rate": 4.359576968272621e-06,
400
+ "loss": 0.0,
401
+ "step": 112
402
+ },
403
+ {
404
+ "epoch": 0.1335676625659051,
405
+ "grad_norm": 0.005059251096099615,
406
+ "learning_rate": 4.347826086956522e-06,
407
+ "loss": 0.0001,
408
+ "step": 114
409
+ },
410
+ {
411
+ "epoch": 0.13591095489162272,
412
+ "grad_norm": 0.0005816388293169439,
413
+ "learning_rate": 4.3360752056404234e-06,
414
+ "loss": 0.0001,
415
+ "step": 116
416
+ },
417
+ {
418
+ "epoch": 0.13825424721734036,
419
+ "grad_norm": 0.019756818190217018,
420
+ "learning_rate": 4.324324324324325e-06,
421
+ "loss": 0.0001,
422
+ "step": 118
423
+ },
424
+ {
425
+ "epoch": 0.140597539543058,
426
+ "grad_norm": 0.0023519208189100027,
427
+ "learning_rate": 4.312573443008226e-06,
428
+ "loss": 0.0,
429
+ "step": 120
430
+ },
431
+ {
432
+ "epoch": 0.14294083186877563,
433
+ "grad_norm": 0.0028086318634450436,
434
+ "learning_rate": 4.300822561692127e-06,
435
+ "loss": 0.0,
436
+ "step": 122
437
+ },
438
+ {
439
+ "epoch": 0.14528412419449327,
440
+ "grad_norm": 0.0022307527251541615,
441
+ "learning_rate": 4.289071680376028e-06,
442
+ "loss": 0.0,
443
+ "step": 124
444
+ },
445
+ {
446
+ "epoch": 0.14762741652021089,
447
+ "grad_norm": 0.014247684739530087,
448
+ "learning_rate": 4.2773207990599296e-06,
449
+ "loss": 0.0001,
450
+ "step": 126
451
+ },
452
+ {
453
+ "epoch": 0.14997070884592853,
454
+ "grad_norm": 0.00011139630805701017,
455
+ "learning_rate": 4.265569917743831e-06,
456
+ "loss": 0.0,
457
+ "step": 128
458
+ },
459
+ {
460
+ "epoch": 0.15231400117164617,
461
+ "grad_norm": 0.000514341751113534,
462
+ "learning_rate": 4.253819036427733e-06,
463
+ "loss": 0.0,
464
+ "step": 130
465
+ },
466
+ {
467
+ "epoch": 0.1546572934973638,
468
+ "grad_norm": 0.002176255453377962,
469
+ "learning_rate": 4.242068155111634e-06,
470
+ "loss": 0.0001,
471
+ "step": 132
472
+ },
473
+ {
474
+ "epoch": 0.15700058582308143,
475
+ "grad_norm": 0.018497969955205917,
476
+ "learning_rate": 4.230317273795535e-06,
477
+ "loss": 0.0001,
478
+ "step": 134
479
+ },
480
+ {
481
+ "epoch": 0.15934387814879905,
482
+ "grad_norm": 0.013157431036233902,
483
+ "learning_rate": 4.2185663924794365e-06,
484
+ "loss": 0.0001,
485
+ "step": 136
486
+ },
487
+ {
488
+ "epoch": 0.1616871704745167,
489
+ "grad_norm": 0.007630129344761372,
490
+ "learning_rate": 4.206815511163338e-06,
491
+ "loss": 0.0,
492
+ "step": 138
493
+ },
494
+ {
495
+ "epoch": 0.16403046280023434,
496
+ "grad_norm": 0.0008055138750933111,
497
+ "learning_rate": 4.195064629847239e-06,
498
+ "loss": 0.0001,
499
+ "step": 140
500
+ },
501
+ {
502
+ "epoch": 0.16637375512595196,
503
+ "grad_norm": 0.006306421477347612,
504
+ "learning_rate": 4.18331374853114e-06,
505
+ "loss": 0.0,
506
+ "step": 142
507
+ },
508
+ {
509
+ "epoch": 0.1687170474516696,
510
+ "grad_norm": 0.020266445353627205,
511
+ "learning_rate": 4.1715628672150414e-06,
512
+ "loss": 0.0001,
513
+ "step": 144
514
+ },
515
+ {
516
+ "epoch": 0.17106033977738722,
517
+ "grad_norm": 0.00037427974166348577,
518
+ "learning_rate": 4.159811985898943e-06,
519
+ "loss": 0.0,
520
+ "step": 146
521
+ },
522
+ {
523
+ "epoch": 0.17340363210310486,
524
+ "grad_norm": 0.004259356763213873,
525
+ "learning_rate": 4.148061104582844e-06,
526
+ "loss": 0.0001,
527
+ "step": 148
528
+ },
529
+ {
530
+ "epoch": 0.1757469244288225,
531
+ "grad_norm": 0.0010232679778710008,
532
+ "learning_rate": 4.136310223266745e-06,
533
+ "loss": 0.0001,
534
+ "step": 150
535
+ },
536
+ {
537
+ "epoch": 0.17809021675454012,
538
+ "grad_norm": 0.003952402155846357,
539
+ "learning_rate": 4.124559341950647e-06,
540
+ "loss": 0.0,
541
+ "step": 152
542
+ },
543
+ {
544
+ "epoch": 0.18043350908025776,
545
+ "grad_norm": 0.0013295585522428155,
546
+ "learning_rate": 4.112808460634548e-06,
547
+ "loss": 0.0,
548
+ "step": 154
549
+ },
550
+ {
551
+ "epoch": 0.1827768014059754,
552
+ "grad_norm": 0.013831949792802334,
553
+ "learning_rate": 4.10105757931845e-06,
554
+ "loss": 0.0001,
555
+ "step": 156
556
+ },
557
+ {
558
+ "epoch": 0.18512009373169303,
559
+ "grad_norm": 0.0036904062144458294,
560
+ "learning_rate": 4.089306698002351e-06,
561
+ "loss": 0.0,
562
+ "step": 158
563
+ },
564
+ {
565
+ "epoch": 0.18746338605741067,
566
+ "grad_norm": 0.002993196714669466,
567
+ "learning_rate": 4.077555816686252e-06,
568
+ "loss": 0.0,
569
+ "step": 160
570
+ },
571
+ {
572
+ "epoch": 0.18980667838312829,
573
+ "grad_norm": 0.0016740068094804883,
574
+ "learning_rate": 4.0658049353701525e-06,
575
+ "loss": 0.0001,
576
+ "step": 162
577
+ },
578
+ {
579
+ "epoch": 0.19214997070884593,
580
+ "grad_norm": 0.012307717464864254,
581
+ "learning_rate": 4.0540540540540545e-06,
582
+ "loss": 0.0001,
583
+ "step": 164
584
+ },
585
+ {
586
+ "epoch": 0.19449326303456357,
587
+ "grad_norm": 0.0012654109159484506,
588
+ "learning_rate": 4.042303172737956e-06,
589
+ "loss": 0.0,
590
+ "step": 166
591
+ },
592
+ {
593
+ "epoch": 0.1968365553602812,
594
+ "grad_norm": 0.12437883019447327,
595
+ "learning_rate": 4.030552291421857e-06,
596
+ "loss": 0.0006,
597
+ "step": 168
598
+ },
599
+ {
600
+ "epoch": 0.19917984768599883,
601
+ "grad_norm": 8.974138472694904e-05,
602
+ "learning_rate": 4.018801410105758e-06,
603
+ "loss": 0.0,
604
+ "step": 170
605
+ },
606
+ {
607
+ "epoch": 0.20152314001171645,
608
+ "grad_norm": 0.0011903212871402502,
609
+ "learning_rate": 4.007050528789659e-06,
610
+ "loss": 0.0001,
611
+ "step": 172
612
+ },
613
+ {
614
+ "epoch": 0.2038664323374341,
615
+ "grad_norm": 0.012350277975201607,
616
+ "learning_rate": 3.995299647473561e-06,
617
+ "loss": 0.0001,
618
+ "step": 174
619
+ },
620
+ {
621
+ "epoch": 0.20620972466315174,
622
+ "grad_norm": 0.01664598099887371,
623
+ "learning_rate": 3.983548766157463e-06,
624
+ "loss": 0.0001,
625
+ "step": 176
626
+ },
627
+ {
628
+ "epoch": 0.20855301698886936,
629
+ "grad_norm": 0.0064240009523928165,
630
+ "learning_rate": 3.971797884841364e-06,
631
+ "loss": 0.0001,
632
+ "step": 178
633
+ },
634
+ {
635
+ "epoch": 0.210896309314587,
636
+ "grad_norm": 0.0031362581066787243,
637
+ "learning_rate": 3.960047003525264e-06,
638
+ "loss": 0.0,
639
+ "step": 180
640
+ },
641
+ {
642
+ "epoch": 0.21323960164030462,
643
+ "grad_norm": 0.00012566300574690104,
644
+ "learning_rate": 3.9482961222091655e-06,
645
+ "loss": 0.0001,
646
+ "step": 182
647
+ },
648
+ {
649
+ "epoch": 0.21558289396602226,
650
+ "grad_norm": 0.0018261070363223553,
651
+ "learning_rate": 3.936545240893067e-06,
652
+ "loss": 0.0,
653
+ "step": 184
654
+ },
655
+ {
656
+ "epoch": 0.2179261862917399,
657
+ "grad_norm": 0.0010897299507632852,
658
+ "learning_rate": 3.924794359576969e-06,
659
+ "loss": 0.0,
660
+ "step": 186
661
+ },
662
+ {
663
+ "epoch": 0.22026947861745752,
664
+ "grad_norm": 0.006528445053845644,
665
+ "learning_rate": 3.91304347826087e-06,
666
+ "loss": 0.0,
667
+ "step": 188
668
+ },
669
+ {
670
+ "epoch": 0.22261277094317516,
671
+ "grad_norm": 0.4626096785068512,
672
+ "learning_rate": 3.901292596944771e-06,
673
+ "loss": 0.0009,
674
+ "step": 190
675
+ },
676
+ {
677
+ "epoch": 0.22495606326889278,
678
+ "grad_norm": 0.002359338803216815,
679
+ "learning_rate": 3.8895417156286725e-06,
680
+ "loss": 0.0,
681
+ "step": 192
682
+ },
683
+ {
684
+ "epoch": 0.22729935559461042,
685
+ "grad_norm": 0.004821418318897486,
686
+ "learning_rate": 3.877790834312574e-06,
687
+ "loss": 0.0,
688
+ "step": 194
689
+ },
690
+ {
691
+ "epoch": 0.22964264792032807,
692
+ "grad_norm": 0.0011465001152828336,
693
+ "learning_rate": 3.866039952996475e-06,
694
+ "loss": 0.0008,
695
+ "step": 196
696
+ },
697
+ {
698
+ "epoch": 0.23198594024604569,
699
+ "grad_norm": 0.0007381247123703361,
700
+ "learning_rate": 3.854289071680376e-06,
701
+ "loss": 0.0001,
702
+ "step": 198
703
+ },
704
+ {
705
+ "epoch": 0.23432923257176333,
706
+ "grad_norm": 0.0023091183975338936,
707
+ "learning_rate": 3.842538190364277e-06,
708
+ "loss": 0.0,
709
+ "step": 200
710
+ },
711
+ {
712
+ "epoch": 0.23667252489748097,
713
+ "grad_norm": 0.0005714365397579968,
714
+ "learning_rate": 3.830787309048179e-06,
715
+ "loss": 0.0,
716
+ "step": 202
717
+ },
718
+ {
719
+ "epoch": 0.2390158172231986,
720
+ "grad_norm": 0.00351692084223032,
721
+ "learning_rate": 3.81903642773208e-06,
722
+ "loss": 0.0,
723
+ "step": 204
724
+ },
725
+ {
726
+ "epoch": 0.24135910954891623,
727
+ "grad_norm": 5.926425728830509e-05,
728
+ "learning_rate": 3.8072855464159815e-06,
729
+ "loss": 0.0,
730
+ "step": 206
731
+ },
732
+ {
733
+ "epoch": 0.24370240187463385,
734
+ "grad_norm": 0.0016421001637354493,
735
+ "learning_rate": 3.7955346650998827e-06,
736
+ "loss": 0.0,
737
+ "step": 208
738
+ },
739
+ {
740
+ "epoch": 0.2460456942003515,
741
+ "grad_norm": 0.012118808925151825,
742
+ "learning_rate": 3.7837837837837844e-06,
743
+ "loss": 0.0001,
744
+ "step": 210
745
+ },
746
+ {
747
+ "epoch": 0.24838898652606914,
748
+ "grad_norm": 0.00024874648079276085,
749
+ "learning_rate": 3.7720329024676856e-06,
750
+ "loss": 0.0002,
751
+ "step": 212
752
+ },
753
+ {
754
+ "epoch": 0.2507322788517868,
755
+ "grad_norm": 0.0017625248292461038,
756
+ "learning_rate": 3.760282021151587e-06,
757
+ "loss": 0.0,
758
+ "step": 214
759
+ },
760
+ {
761
+ "epoch": 0.2530755711775044,
762
+ "grad_norm": 0.0007431196281686425,
763
+ "learning_rate": 3.748531139835488e-06,
764
+ "loss": 0.0,
765
+ "step": 216
766
+ },
767
+ {
768
+ "epoch": 0.255418863503222,
769
+ "grad_norm": 0.0007026457460597157,
770
+ "learning_rate": 3.7367802585193893e-06,
771
+ "loss": 0.0,
772
+ "step": 218
773
+ },
774
+ {
775
+ "epoch": 0.2577621558289397,
776
+ "grad_norm": 0.002397920237854123,
777
+ "learning_rate": 3.72502937720329e-06,
778
+ "loss": 0.0,
779
+ "step": 220
780
+ },
781
+ {
782
+ "epoch": 0.2601054481546573,
783
+ "grad_norm": 0.003177257487550378,
784
+ "learning_rate": 3.713278495887192e-06,
785
+ "loss": 0.0,
786
+ "step": 222
787
+ },
788
+ {
789
+ "epoch": 0.2624487404803749,
790
+ "grad_norm": 0.003142025787383318,
791
+ "learning_rate": 3.7015276145710934e-06,
792
+ "loss": 0.0001,
793
+ "step": 224
794
+ },
795
+ {
796
+ "epoch": 0.26479203280609254,
797
+ "grad_norm": 0.03788410872220993,
798
+ "learning_rate": 3.6897767332549946e-06,
799
+ "loss": 0.0002,
800
+ "step": 226
801
+ },
802
+ {
803
+ "epoch": 0.2671353251318102,
804
+ "grad_norm": 0.005685464479029179,
805
+ "learning_rate": 3.6780258519388954e-06,
806
+ "loss": 0.0003,
807
+ "step": 228
808
+ },
809
+ {
810
+ "epoch": 0.2694786174575278,
811
+ "grad_norm": 0.0010328789940103889,
812
+ "learning_rate": 3.6662749706227966e-06,
813
+ "loss": 0.0003,
814
+ "step": 230
815
+ },
816
+ {
817
+ "epoch": 0.27182190978324544,
818
+ "grad_norm": 0.0052024442702531815,
819
+ "learning_rate": 3.6545240893066987e-06,
820
+ "loss": 0.0,
821
+ "step": 232
822
+ },
823
+ {
824
+ "epoch": 0.2741652021089631,
825
+ "grad_norm": 0.006033598445355892,
826
+ "learning_rate": 3.6427732079906e-06,
827
+ "loss": 0.0,
828
+ "step": 234
829
+ },
830
+ {
831
+ "epoch": 0.27650849443468073,
832
+ "grad_norm": 0.00023948443413246423,
833
+ "learning_rate": 3.6310223266745007e-06,
834
+ "loss": 0.0001,
835
+ "step": 236
836
+ },
837
+ {
838
+ "epoch": 0.27885178676039835,
839
+ "grad_norm": 0.00016467843670397997,
840
+ "learning_rate": 3.619271445358402e-06,
841
+ "loss": 0.0,
842
+ "step": 238
843
+ },
844
+ {
845
+ "epoch": 0.281195079086116,
846
+ "grad_norm": 0.003566320287063718,
847
+ "learning_rate": 3.607520564042303e-06,
848
+ "loss": 0.0,
849
+ "step": 240
850
+ },
851
+ {
852
+ "epoch": 0.28353837141183363,
853
+ "grad_norm": 0.00033969045034609735,
854
+ "learning_rate": 3.5957696827262044e-06,
855
+ "loss": 0.0,
856
+ "step": 242
857
+ },
858
+ {
859
+ "epoch": 0.28588166373755125,
860
+ "grad_norm": 0.0033994223922491074,
861
+ "learning_rate": 3.5840188014101065e-06,
862
+ "loss": 0.0,
863
+ "step": 244
864
+ },
865
+ {
866
+ "epoch": 0.28822495606326887,
867
+ "grad_norm": 0.14746786653995514,
868
+ "learning_rate": 3.5722679200940073e-06,
869
+ "loss": 0.0008,
870
+ "step": 246
871
+ },
872
+ {
873
+ "epoch": 0.29056824838898654,
874
+ "grad_norm": 0.012470235116779804,
875
+ "learning_rate": 3.5605170387779085e-06,
876
+ "loss": 0.0,
877
+ "step": 248
878
+ },
879
+ {
880
+ "epoch": 0.29291154071470415,
881
+ "grad_norm": 0.08307931572198868,
882
+ "learning_rate": 3.5487661574618097e-06,
883
+ "loss": 0.0003,
884
+ "step": 250
885
+ },
886
+ {
887
+ "epoch": 0.29525483304042177,
888
+ "grad_norm": 0.00033245363738387823,
889
+ "learning_rate": 3.537015276145711e-06,
890
+ "loss": 0.0,
891
+ "step": 252
892
+ },
893
+ {
894
+ "epoch": 0.29759812536613944,
895
+ "grad_norm": 0.0018247144762426615,
896
+ "learning_rate": 3.525264394829612e-06,
897
+ "loss": 0.0,
898
+ "step": 254
899
+ },
900
+ {
901
+ "epoch": 0.29994141769185706,
902
+ "grad_norm": 0.0011103990254923701,
903
+ "learning_rate": 3.513513513513514e-06,
904
+ "loss": 0.0001,
905
+ "step": 256
906
+ },
907
+ {
908
+ "epoch": 0.3022847100175747,
909
+ "grad_norm": 0.0010811882093548775,
910
+ "learning_rate": 3.501762632197415e-06,
911
+ "loss": 0.0,
912
+ "step": 258
913
+ },
914
+ {
915
+ "epoch": 0.30462800234329235,
916
+ "grad_norm": 0.011172047816216946,
917
+ "learning_rate": 3.4900117508813163e-06,
918
+ "loss": 0.0001,
919
+ "step": 260
920
+ },
921
+ {
922
+ "epoch": 0.30697129466900996,
923
+ "grad_norm": 0.0013676233356818557,
924
+ "learning_rate": 3.4782608695652175e-06,
925
+ "loss": 0.0,
926
+ "step": 262
927
+ },
928
+ {
929
+ "epoch": 0.3093145869947276,
930
+ "grad_norm": 0.002147970488294959,
931
+ "learning_rate": 3.4665099882491187e-06,
932
+ "loss": 0.0,
933
+ "step": 264
934
+ },
935
+ {
936
+ "epoch": 0.31165787932044525,
937
+ "grad_norm": 0.0009826518362388015,
938
+ "learning_rate": 3.4547591069330204e-06,
939
+ "loss": 0.0,
940
+ "step": 266
941
+ },
942
+ {
943
+ "epoch": 0.31400117164616287,
944
+ "grad_norm": 0.001499099307693541,
945
+ "learning_rate": 3.4430082256169216e-06,
946
+ "loss": 0.0,
947
+ "step": 268
948
+ },
949
+ {
950
+ "epoch": 0.3163444639718805,
951
+ "grad_norm": 0.001323301112279296,
952
+ "learning_rate": 3.431257344300823e-06,
953
+ "loss": 0.0,
954
+ "step": 270
955
+ },
956
+ {
957
+ "epoch": 0.3186877562975981,
958
+ "grad_norm": 0.018010340631008148,
959
+ "learning_rate": 3.419506462984724e-06,
960
+ "loss": 0.0005,
961
+ "step": 272
962
+ },
963
+ {
964
+ "epoch": 0.3210310486233158,
965
+ "grad_norm": 0.0024064648896455765,
966
+ "learning_rate": 3.4077555816686253e-06,
967
+ "loss": 0.0,
968
+ "step": 274
969
+ },
970
+ {
971
+ "epoch": 0.3233743409490334,
972
+ "grad_norm": 0.02396260015666485,
973
+ "learning_rate": 3.3960047003525265e-06,
974
+ "loss": 0.0001,
975
+ "step": 276
976
+ },
977
+ {
978
+ "epoch": 0.325717633274751,
979
+ "grad_norm": 0.002070352202281356,
980
+ "learning_rate": 3.384253819036428e-06,
981
+ "loss": 0.0,
982
+ "step": 278
983
+ },
984
+ {
985
+ "epoch": 0.3280609256004687,
986
+ "grad_norm": 0.0003108434902969748,
987
+ "learning_rate": 3.3725029377203294e-06,
988
+ "loss": 0.0001,
989
+ "step": 280
990
+ },
991
+ {
992
+ "epoch": 0.3304042179261863,
993
+ "grad_norm": 0.006573045626282692,
994
+ "learning_rate": 3.3607520564042306e-06,
995
+ "loss": 0.0001,
996
+ "step": 282
997
+ },
998
+ {
999
+ "epoch": 0.3327475102519039,
1000
+ "grad_norm": 0.0004413512069731951,
1001
+ "learning_rate": 3.349001175088132e-06,
1002
+ "loss": 0.0001,
1003
+ "step": 284
1004
+ },
1005
+ {
1006
+ "epoch": 0.3350908025776216,
1007
+ "grad_norm": 0.0005645502242259681,
1008
+ "learning_rate": 3.337250293772033e-06,
1009
+ "loss": 0.0,
1010
+ "step": 286
1011
+ },
1012
+ {
1013
+ "epoch": 0.3374340949033392,
1014
+ "grad_norm": 0.00034579774364829063,
1015
+ "learning_rate": 3.3254994124559343e-06,
1016
+ "loss": 0.0,
1017
+ "step": 288
1018
+ },
1019
+ {
1020
+ "epoch": 0.3397773872290568,
1021
+ "grad_norm": 0.003136229468509555,
1022
+ "learning_rate": 3.313748531139836e-06,
1023
+ "loss": 0.0,
1024
+ "step": 290
1025
+ },
1026
+ {
1027
+ "epoch": 0.34212067955477443,
1028
+ "grad_norm": 0.0031148705165833235,
1029
+ "learning_rate": 3.301997649823737e-06,
1030
+ "loss": 0.0,
1031
+ "step": 292
1032
+ },
1033
+ {
1034
+ "epoch": 0.3444639718804921,
1035
+ "grad_norm": 0.0012612566351890564,
1036
+ "learning_rate": 3.2902467685076384e-06,
1037
+ "loss": 0.0,
1038
+ "step": 294
1039
+ },
1040
+ {
1041
+ "epoch": 0.3468072642062097,
1042
+ "grad_norm": 0.0007469533011317253,
1043
+ "learning_rate": 3.2784958871915396e-06,
1044
+ "loss": 0.0,
1045
+ "step": 296
1046
+ },
1047
+ {
1048
+ "epoch": 0.34915055653192734,
1049
+ "grad_norm": 0.04412250965833664,
1050
+ "learning_rate": 3.266745005875441e-06,
1051
+ "loss": 0.0003,
1052
+ "step": 298
1053
+ },
1054
+ {
1055
+ "epoch": 0.351493848857645,
1056
+ "grad_norm": 0.004462533164769411,
1057
+ "learning_rate": 3.2549941245593425e-06,
1058
+ "loss": 0.0088,
1059
+ "step": 300
1060
+ },
1061
+ {
1062
+ "epoch": 0.3538371411833626,
1063
+ "grad_norm": 0.002911294111981988,
1064
+ "learning_rate": 3.2432432432432437e-06,
1065
+ "loss": 0.0006,
1066
+ "step": 302
1067
+ },
1068
+ {
1069
+ "epoch": 0.35618043350908024,
1070
+ "grad_norm": 0.0015191801358014345,
1071
+ "learning_rate": 3.231492361927145e-06,
1072
+ "loss": 0.0,
1073
+ "step": 304
1074
+ },
1075
+ {
1076
+ "epoch": 0.3585237258347979,
1077
+ "grad_norm": 0.017380721867084503,
1078
+ "learning_rate": 3.219741480611046e-06,
1079
+ "loss": 0.0094,
1080
+ "step": 306
1081
+ },
1082
+ {
1083
+ "epoch": 0.36086701816051553,
1084
+ "grad_norm": 0.002749436302110553,
1085
+ "learning_rate": 3.2079905992949474e-06,
1086
+ "loss": 0.0001,
1087
+ "step": 308
1088
+ },
1089
+ {
1090
+ "epoch": 0.36321031048623315,
1091
+ "grad_norm": 0.0008673086995258927,
1092
+ "learning_rate": 3.1962397179788486e-06,
1093
+ "loss": 0.0,
1094
+ "step": 310
1095
+ },
1096
+ {
1097
+ "epoch": 0.3655536028119508,
1098
+ "grad_norm": 0.00361701101064682,
1099
+ "learning_rate": 3.1844888366627503e-06,
1100
+ "loss": 0.0,
1101
+ "step": 312
1102
+ },
1103
+ {
1104
+ "epoch": 0.36789689513766843,
1105
+ "grad_norm": 0.006906528025865555,
1106
+ "learning_rate": 3.1727379553466515e-06,
1107
+ "loss": 0.0,
1108
+ "step": 314
1109
+ },
1110
+ {
1111
+ "epoch": 0.37024018746338605,
1112
+ "grad_norm": 2.259305238723755,
1113
+ "learning_rate": 3.1609870740305527e-06,
1114
+ "loss": 0.0157,
1115
+ "step": 316
1116
+ },
1117
+ {
1118
+ "epoch": 0.37258347978910367,
1119
+ "grad_norm": 0.00017454673070460558,
1120
+ "learning_rate": 3.149236192714454e-06,
1121
+ "loss": 0.0,
1122
+ "step": 318
1123
+ },
1124
+ {
1125
+ "epoch": 0.37492677211482134,
1126
+ "grad_norm": 0.16197967529296875,
1127
+ "learning_rate": 3.137485311398355e-06,
1128
+ "loss": 0.0009,
1129
+ "step": 320
1130
+ },
1131
+ {
1132
+ "epoch": 0.37727006444053895,
1133
+ "grad_norm": 0.002247605938464403,
1134
+ "learning_rate": 3.1257344300822564e-06,
1135
+ "loss": 0.0,
1136
+ "step": 322
1137
+ },
1138
+ {
1139
+ "epoch": 0.37961335676625657,
1140
+ "grad_norm": 0.023727795109152794,
1141
+ "learning_rate": 3.113983548766158e-06,
1142
+ "loss": 0.0001,
1143
+ "step": 324
1144
+ },
1145
+ {
1146
+ "epoch": 0.38195664909197424,
1147
+ "grad_norm": 0.008455273695290089,
1148
+ "learning_rate": 3.1022326674500592e-06,
1149
+ "loss": 0.0001,
1150
+ "step": 326
1151
+ },
1152
+ {
1153
+ "epoch": 0.38429994141769186,
1154
+ "grad_norm": 0.00022873218404129148,
1155
+ "learning_rate": 3.0904817861339605e-06,
1156
+ "loss": 0.0,
1157
+ "step": 328
1158
+ },
1159
+ {
1160
+ "epoch": 0.3866432337434095,
1161
+ "grad_norm": 3.000872850418091,
1162
+ "learning_rate": 3.0787309048178617e-06,
1163
+ "loss": 0.055,
1164
+ "step": 330
1165
+ },
1166
+ {
1167
+ "epoch": 0.38898652606912715,
1168
+ "grad_norm": 0.002177221467718482,
1169
+ "learning_rate": 3.066980023501763e-06,
1170
+ "loss": 0.0,
1171
+ "step": 332
1172
+ },
1173
+ {
1174
+ "epoch": 0.39132981839484476,
1175
+ "grad_norm": 0.002786975121125579,
1176
+ "learning_rate": 3.0552291421856637e-06,
1177
+ "loss": 0.0,
1178
+ "step": 334
1179
+ },
1180
+ {
1181
+ "epoch": 0.3936731107205624,
1182
+ "grad_norm": 0.004335256293416023,
1183
+ "learning_rate": 3.043478260869566e-06,
1184
+ "loss": 0.0,
1185
+ "step": 336
1186
+ },
1187
+ {
1188
+ "epoch": 0.39601640304628,
1189
+ "grad_norm": 0.007627409417182207,
1190
+ "learning_rate": 3.031727379553467e-06,
1191
+ "loss": 0.0001,
1192
+ "step": 338
1193
+ },
1194
+ {
1195
+ "epoch": 0.39835969537199767,
1196
+ "grad_norm": 0.002631911775097251,
1197
+ "learning_rate": 3.0199764982373682e-06,
1198
+ "loss": 0.0,
1199
+ "step": 340
1200
+ },
1201
+ {
1202
+ "epoch": 0.4007029876977153,
1203
+ "grad_norm": 0.009561799466609955,
1204
+ "learning_rate": 3.008225616921269e-06,
1205
+ "loss": 0.0001,
1206
+ "step": 342
1207
+ },
1208
+ {
1209
+ "epoch": 0.4030462800234329,
1210
+ "grad_norm": 0.0026635443791747093,
1211
+ "learning_rate": 2.9964747356051703e-06,
1212
+ "loss": 0.0001,
1213
+ "step": 344
1214
+ },
1215
+ {
1216
+ "epoch": 0.4053895723491506,
1217
+ "grad_norm": 0.0001533351169200614,
1218
+ "learning_rate": 2.9847238542890723e-06,
1219
+ "loss": 0.0,
1220
+ "step": 346
1221
+ },
1222
+ {
1223
+ "epoch": 0.4077328646748682,
1224
+ "grad_norm": 0.0835270956158638,
1225
+ "learning_rate": 2.9729729729729736e-06,
1226
+ "loss": 0.0005,
1227
+ "step": 348
1228
+ },
1229
+ {
1230
+ "epoch": 0.4100761570005858,
1231
+ "grad_norm": 0.003761101048439741,
1232
+ "learning_rate": 2.9612220916568744e-06,
1233
+ "loss": 0.0,
1234
+ "step": 350
1235
+ },
1236
+ {
1237
+ "epoch": 0.4124194493263035,
1238
+ "grad_norm": 0.01136633288115263,
1239
+ "learning_rate": 2.9494712103407756e-06,
1240
+ "loss": 0.0002,
1241
+ "step": 352
1242
+ },
1243
+ {
1244
+ "epoch": 0.4147627416520211,
1245
+ "grad_norm": 0.007711971178650856,
1246
+ "learning_rate": 2.937720329024677e-06,
1247
+ "loss": 0.0001,
1248
+ "step": 354
1249
+ },
1250
+ {
1251
+ "epoch": 0.4171060339777387,
1252
+ "grad_norm": 0.0003854953683912754,
1253
+ "learning_rate": 2.925969447708578e-06,
1254
+ "loss": 0.0,
1255
+ "step": 356
1256
+ },
1257
+ {
1258
+ "epoch": 0.4194493263034564,
1259
+ "grad_norm": 0.019140860065817833,
1260
+ "learning_rate": 2.91421856639248e-06,
1261
+ "loss": 0.0001,
1262
+ "step": 358
1263
+ },
1264
+ {
1265
+ "epoch": 0.421792618629174,
1266
+ "grad_norm": 0.0013410028768703341,
1267
+ "learning_rate": 2.902467685076381e-06,
1268
+ "loss": 0.0003,
1269
+ "step": 360
1270
+ },
1271
+ {
1272
+ "epoch": 0.4241359109548916,
1273
+ "grad_norm": 0.0011243935441598296,
1274
+ "learning_rate": 2.890716803760282e-06,
1275
+ "loss": 0.0001,
1276
+ "step": 362
1277
+ },
1278
+ {
1279
+ "epoch": 0.42647920328060923,
1280
+ "grad_norm": 0.012134709395468235,
1281
+ "learning_rate": 2.8789659224441834e-06,
1282
+ "loss": 0.0001,
1283
+ "step": 364
1284
+ },
1285
+ {
1286
+ "epoch": 0.4288224956063269,
1287
+ "grad_norm": 0.0028234529308974743,
1288
+ "learning_rate": 2.8672150411280846e-06,
1289
+ "loss": 0.0,
1290
+ "step": 366
1291
+ },
1292
+ {
1293
+ "epoch": 0.4311657879320445,
1294
+ "grad_norm": 0.004319467581808567,
1295
+ "learning_rate": 2.855464159811986e-06,
1296
+ "loss": 0.0,
1297
+ "step": 368
1298
+ },
1299
+ {
1300
+ "epoch": 0.43350908025776214,
1301
+ "grad_norm": 0.0068093533627688885,
1302
+ "learning_rate": 2.8437132784958875e-06,
1303
+ "loss": 0.0001,
1304
+ "step": 370
1305
+ },
1306
+ {
1307
+ "epoch": 0.4358523725834798,
1308
+ "grad_norm": 0.016774361953139305,
1309
+ "learning_rate": 2.8319623971797887e-06,
1310
+ "loss": 0.0001,
1311
+ "step": 372
1312
+ },
1313
+ {
1314
+ "epoch": 0.4381956649091974,
1315
+ "grad_norm": 0.014978869818150997,
1316
+ "learning_rate": 2.82021151586369e-06,
1317
+ "loss": 0.0001,
1318
+ "step": 374
1319
+ },
1320
+ {
1321
+ "epoch": 0.44053895723491504,
1322
+ "grad_norm": 0.0010881100315600634,
1323
+ "learning_rate": 2.808460634547591e-06,
1324
+ "loss": 0.0004,
1325
+ "step": 376
1326
+ },
1327
+ {
1328
+ "epoch": 0.4428822495606327,
1329
+ "grad_norm": 0.05522293969988823,
1330
+ "learning_rate": 2.7967097532314924e-06,
1331
+ "loss": 0.0002,
1332
+ "step": 378
1333
+ },
1334
+ {
1335
+ "epoch": 0.44522554188635033,
1336
+ "grad_norm": 0.0027575818821787834,
1337
+ "learning_rate": 2.784958871915394e-06,
1338
+ "loss": 0.0,
1339
+ "step": 380
1340
+ },
1341
+ {
1342
+ "epoch": 0.44756883421206795,
1343
+ "grad_norm": 0.0006020054570399225,
1344
+ "learning_rate": 2.7732079905992952e-06,
1345
+ "loss": 0.0005,
1346
+ "step": 382
1347
+ },
1348
+ {
1349
+ "epoch": 0.44991212653778556,
1350
+ "grad_norm": 0.0025616425555199385,
1351
+ "learning_rate": 2.7614571092831965e-06,
1352
+ "loss": 0.0,
1353
+ "step": 384
1354
+ },
1355
+ {
1356
+ "epoch": 0.45225541886350323,
1357
+ "grad_norm": 0.0018823420396074653,
1358
+ "learning_rate": 2.7497062279670977e-06,
1359
+ "loss": 0.0,
1360
+ "step": 386
1361
+ },
1362
+ {
1363
+ "epoch": 0.45459871118922085,
1364
+ "grad_norm": 0.003241207217797637,
1365
+ "learning_rate": 2.737955346650999e-06,
1366
+ "loss": 0.0,
1367
+ "step": 388
1368
+ },
1369
+ {
1370
+ "epoch": 0.45694200351493847,
1371
+ "grad_norm": 0.0010485474485903978,
1372
+ "learning_rate": 2.7262044653349e-06,
1373
+ "loss": 0.0002,
1374
+ "step": 390
1375
+ },
1376
+ {
1377
+ "epoch": 0.45928529584065614,
1378
+ "grad_norm": 0.013366922736167908,
1379
+ "learning_rate": 2.714453584018802e-06,
1380
+ "loss": 0.0001,
1381
+ "step": 392
1382
+ },
1383
+ {
1384
+ "epoch": 0.46162858816637375,
1385
+ "grad_norm": 0.0005886501166969538,
1386
+ "learning_rate": 2.702702702702703e-06,
1387
+ "loss": 0.0,
1388
+ "step": 394
1389
+ },
1390
+ {
1391
+ "epoch": 0.46397188049209137,
1392
+ "grad_norm": 7.603697304148227e-05,
1393
+ "learning_rate": 2.6909518213866042e-06,
1394
+ "loss": 0.0,
1395
+ "step": 396
1396
+ },
1397
+ {
1398
+ "epoch": 0.46631517281780904,
1399
+ "grad_norm": 0.000614571908954531,
1400
+ "learning_rate": 2.6792009400705055e-06,
1401
+ "loss": 0.0023,
1402
+ "step": 398
1403
+ },
1404
+ {
1405
+ "epoch": 0.46865846514352666,
1406
+ "grad_norm": 0.046423882246017456,
1407
+ "learning_rate": 2.6674500587544067e-06,
1408
+ "loss": 0.0002,
1409
+ "step": 400
1410
+ },
1411
+ {
1412
+ "epoch": 0.4710017574692443,
1413
+ "grad_norm": 0.0005994020029902458,
1414
+ "learning_rate": 2.655699177438308e-06,
1415
+ "loss": 0.0,
1416
+ "step": 402
1417
+ },
1418
+ {
1419
+ "epoch": 0.47334504979496195,
1420
+ "grad_norm": 0.011609828099608421,
1421
+ "learning_rate": 2.6439482961222096e-06,
1422
+ "loss": 0.0001,
1423
+ "step": 404
1424
+ },
1425
+ {
1426
+ "epoch": 0.47568834212067956,
1427
+ "grad_norm": 0.007135775871574879,
1428
+ "learning_rate": 2.632197414806111e-06,
1429
+ "loss": 0.0002,
1430
+ "step": 406
1431
+ },
1432
+ {
1433
+ "epoch": 0.4780316344463972,
1434
+ "grad_norm": 0.0028773818630725145,
1435
+ "learning_rate": 2.620446533490012e-06,
1436
+ "loss": 0.0,
1437
+ "step": 408
1438
+ },
1439
+ {
1440
+ "epoch": 0.4803749267721148,
1441
+ "grad_norm": 0.13341404497623444,
1442
+ "learning_rate": 2.6086956521739132e-06,
1443
+ "loss": 0.0008,
1444
+ "step": 410
1445
+ },
1446
+ {
1447
+ "epoch": 0.48271821909783247,
1448
+ "grad_norm": 0.03130058944225311,
1449
+ "learning_rate": 2.5969447708578145e-06,
1450
+ "loss": 0.0001,
1451
+ "step": 412
1452
+ },
1453
+ {
1454
+ "epoch": 0.4850615114235501,
1455
+ "grad_norm": 0.006637818645685911,
1456
+ "learning_rate": 2.5851938895417157e-06,
1457
+ "loss": 0.0001,
1458
+ "step": 414
1459
+ },
1460
+ {
1461
+ "epoch": 0.4874048037492677,
1462
+ "grad_norm": 0.0006390800117515028,
1463
+ "learning_rate": 2.5734430082256173e-06,
1464
+ "loss": 0.0001,
1465
+ "step": 416
1466
+ },
1467
+ {
1468
+ "epoch": 0.4897480960749854,
1469
+ "grad_norm": 0.02106345072388649,
1470
+ "learning_rate": 2.5616921269095186e-06,
1471
+ "loss": 0.0002,
1472
+ "step": 418
1473
+ },
1474
+ {
1475
+ "epoch": 0.492091388400703,
1476
+ "grad_norm": 0.0009213433368131518,
1477
+ "learning_rate": 2.5499412455934198e-06,
1478
+ "loss": 0.0001,
1479
+ "step": 420
1480
+ },
1481
+ {
1482
+ "epoch": 0.4944346807264206,
1483
+ "grad_norm": 2.5962471961975098,
1484
+ "learning_rate": 2.538190364277321e-06,
1485
+ "loss": 0.1436,
1486
+ "step": 422
1487
+ },
1488
+ {
1489
+ "epoch": 0.4967779730521383,
1490
+ "grad_norm": 0.009386847727000713,
1491
+ "learning_rate": 2.5264394829612222e-06,
1492
+ "loss": 0.0001,
1493
+ "step": 424
1494
+ },
1495
+ {
1496
+ "epoch": 0.4991212653778559,
1497
+ "grad_norm": 0.01308267842978239,
1498
+ "learning_rate": 2.514688601645124e-06,
1499
+ "loss": 0.0001,
1500
+ "step": 426
1501
+ },
1502
+ {
1503
+ "epoch": 0.5014645577035736,
1504
+ "grad_norm": 0.006409250665456057,
1505
+ "learning_rate": 2.502937720329025e-06,
1506
+ "loss": 0.0,
1507
+ "step": 428
1508
+ },
1509
+ {
1510
+ "epoch": 0.5038078500292912,
1511
+ "grad_norm": 0.0018047624034807086,
1512
+ "learning_rate": 2.4911868390129263e-06,
1513
+ "loss": 0.0001,
1514
+ "step": 430
1515
+ },
1516
+ {
1517
+ "epoch": 0.5061511423550088,
1518
+ "grad_norm": 0.007056268397718668,
1519
+ "learning_rate": 2.4794359576968276e-06,
1520
+ "loss": 0.0,
1521
+ "step": 432
1522
+ },
1523
+ {
1524
+ "epoch": 0.5084944346807264,
1525
+ "grad_norm": 2.4651243686676025,
1526
+ "learning_rate": 2.4676850763807288e-06,
1527
+ "loss": 0.0245,
1528
+ "step": 434
1529
+ },
1530
+ {
1531
+ "epoch": 0.510837727006444,
1532
+ "grad_norm": 0.0025760605931282043,
1533
+ "learning_rate": 2.45593419506463e-06,
1534
+ "loss": 0.0,
1535
+ "step": 436
1536
+ },
1537
+ {
1538
+ "epoch": 0.5131810193321616,
1539
+ "grad_norm": 0.059660654515028,
1540
+ "learning_rate": 2.4441833137485312e-06,
1541
+ "loss": 0.0003,
1542
+ "step": 438
1543
+ },
1544
+ {
1545
+ "epoch": 0.5155243116578794,
1546
+ "grad_norm": 0.032668206840753555,
1547
+ "learning_rate": 2.432432432432433e-06,
1548
+ "loss": 0.0002,
1549
+ "step": 440
1550
+ },
1551
+ {
1552
+ "epoch": 0.517867603983597,
1553
+ "grad_norm": 0.002476097084581852,
1554
+ "learning_rate": 2.420681551116334e-06,
1555
+ "loss": 0.0,
1556
+ "step": 442
1557
+ },
1558
+ {
1559
+ "epoch": 0.5202108963093146,
1560
+ "grad_norm": 0.0005356927285902202,
1561
+ "learning_rate": 2.4089306698002353e-06,
1562
+ "loss": 0.0,
1563
+ "step": 444
1564
+ },
1565
+ {
1566
+ "epoch": 0.5225541886350322,
1567
+ "grad_norm": 0.01949264481663704,
1568
+ "learning_rate": 2.3971797884841366e-06,
1569
+ "loss": 0.0001,
1570
+ "step": 446
1571
+ },
1572
+ {
1573
+ "epoch": 0.5248974809607498,
1574
+ "grad_norm": 0.4609091281890869,
1575
+ "learning_rate": 2.3854289071680378e-06,
1576
+ "loss": 0.0013,
1577
+ "step": 448
1578
+ },
1579
+ {
1580
+ "epoch": 0.5272407732864675,
1581
+ "grad_norm": 0.002268969314172864,
1582
+ "learning_rate": 2.373678025851939e-06,
1583
+ "loss": 0.027,
1584
+ "step": 450
1585
+ },
1586
+ {
1587
+ "epoch": 0.5295840656121851,
1588
+ "grad_norm": 0.42679542303085327,
1589
+ "learning_rate": 2.3619271445358407e-06,
1590
+ "loss": 0.002,
1591
+ "step": 452
1592
+ },
1593
+ {
1594
+ "epoch": 0.5319273579379028,
1595
+ "grad_norm": 0.030775954946875572,
1596
+ "learning_rate": 2.350176263219742e-06,
1597
+ "loss": 0.0001,
1598
+ "step": 454
1599
+ },
1600
+ {
1601
+ "epoch": 0.5342706502636204,
1602
+ "grad_norm": 0.006208465900272131,
1603
+ "learning_rate": 2.3384253819036427e-06,
1604
+ "loss": 0.0001,
1605
+ "step": 456
1606
+ },
1607
+ {
1608
+ "epoch": 0.536613942589338,
1609
+ "grad_norm": 0.001203950378112495,
1610
+ "learning_rate": 2.3266745005875443e-06,
1611
+ "loss": 0.0,
1612
+ "step": 458
1613
+ },
1614
+ {
1615
+ "epoch": 0.5389572349150556,
1616
+ "grad_norm": 0.0013062539510428905,
1617
+ "learning_rate": 2.3149236192714456e-06,
1618
+ "loss": 0.0001,
1619
+ "step": 460
1620
+ },
1621
+ {
1622
+ "epoch": 0.5413005272407733,
1623
+ "grad_norm": 0.014242034405469894,
1624
+ "learning_rate": 2.3031727379553468e-06,
1625
+ "loss": 0.0001,
1626
+ "step": 462
1627
+ },
1628
+ {
1629
+ "epoch": 0.5436438195664909,
1630
+ "grad_norm": 0.0024558689910918474,
1631
+ "learning_rate": 2.291421856639248e-06,
1632
+ "loss": 0.0,
1633
+ "step": 464
1634
+ },
1635
+ {
1636
+ "epoch": 0.5459871118922085,
1637
+ "grad_norm": 0.006871205288916826,
1638
+ "learning_rate": 2.2796709753231492e-06,
1639
+ "loss": 0.0,
1640
+ "step": 466
1641
+ },
1642
+ {
1643
+ "epoch": 0.5483304042179262,
1644
+ "grad_norm": 0.016744021326303482,
1645
+ "learning_rate": 2.2679200940070505e-06,
1646
+ "loss": 0.0001,
1647
+ "step": 468
1648
+ },
1649
+ {
1650
+ "epoch": 0.5506736965436438,
1651
+ "grad_norm": 0.0025478950701653957,
1652
+ "learning_rate": 2.256169212690952e-06,
1653
+ "loss": 0.0,
1654
+ "step": 470
1655
+ },
1656
+ {
1657
+ "epoch": 0.5530169888693615,
1658
+ "grad_norm": 0.002553507685661316,
1659
+ "learning_rate": 2.2444183313748533e-06,
1660
+ "loss": 0.0,
1661
+ "step": 472
1662
+ },
1663
+ {
1664
+ "epoch": 0.5553602811950791,
1665
+ "grad_norm": 0.0018396044615656137,
1666
+ "learning_rate": 2.2326674500587546e-06,
1667
+ "loss": 0.0002,
1668
+ "step": 474
1669
+ },
1670
+ {
1671
+ "epoch": 0.5577035735207967,
1672
+ "grad_norm": 0.002036860678344965,
1673
+ "learning_rate": 2.2209165687426558e-06,
1674
+ "loss": 0.0,
1675
+ "step": 476
1676
+ },
1677
+ {
1678
+ "epoch": 0.5600468658465143,
1679
+ "grad_norm": 0.0024688418488949537,
1680
+ "learning_rate": 2.209165687426557e-06,
1681
+ "loss": 0.0,
1682
+ "step": 478
1683
+ },
1684
+ {
1685
+ "epoch": 0.562390158172232,
1686
+ "grad_norm": 0.0028820293955504894,
1687
+ "learning_rate": 2.1974148061104587e-06,
1688
+ "loss": 0.0001,
1689
+ "step": 480
1690
+ },
1691
+ {
1692
+ "epoch": 0.5647334504979497,
1693
+ "grad_norm": 0.00978305283933878,
1694
+ "learning_rate": 2.18566392479436e-06,
1695
+ "loss": 0.0001,
1696
+ "step": 482
1697
+ },
1698
+ {
1699
+ "epoch": 0.5670767428236673,
1700
+ "grad_norm": 0.147267147898674,
1701
+ "learning_rate": 2.173913043478261e-06,
1702
+ "loss": 0.0014,
1703
+ "step": 484
1704
+ },
1705
+ {
1706
+ "epoch": 0.5694200351493849,
1707
+ "grad_norm": 0.005025573540478945,
1708
+ "learning_rate": 2.1621621621621623e-06,
1709
+ "loss": 0.0006,
1710
+ "step": 486
1711
+ },
1712
+ {
1713
+ "epoch": 0.5717633274751025,
1714
+ "grad_norm": 0.0010051846038550138,
1715
+ "learning_rate": 2.1504112808460636e-06,
1716
+ "loss": 0.0003,
1717
+ "step": 488
1718
+ },
1719
+ {
1720
+ "epoch": 0.5741066198008201,
1721
+ "grad_norm": 0.009055075235664845,
1722
+ "learning_rate": 2.1386603995299648e-06,
1723
+ "loss": 0.0001,
1724
+ "step": 490
1725
+ },
1726
+ {
1727
+ "epoch": 0.5764499121265377,
1728
+ "grad_norm": 0.0077414545230567455,
1729
+ "learning_rate": 2.1269095182138664e-06,
1730
+ "loss": 0.0001,
1731
+ "step": 492
1732
+ },
1733
+ {
1734
+ "epoch": 0.5787932044522555,
1735
+ "grad_norm": 0.0059761228039860725,
1736
+ "learning_rate": 2.1151586368977677e-06,
1737
+ "loss": 0.0001,
1738
+ "step": 494
1739
+ },
1740
+ {
1741
+ "epoch": 0.5811364967779731,
1742
+ "grad_norm": 0.0014180493308231235,
1743
+ "learning_rate": 2.103407755581669e-06,
1744
+ "loss": 0.0,
1745
+ "step": 496
1746
+ },
1747
+ {
1748
+ "epoch": 0.5834797891036907,
1749
+ "grad_norm": 0.0022345769684761763,
1750
+ "learning_rate": 2.09165687426557e-06,
1751
+ "loss": 0.0,
1752
+ "step": 498
1753
+ },
1754
+ {
1755
+ "epoch": 0.5858230814294083,
1756
+ "grad_norm": 0.005645833443850279,
1757
+ "learning_rate": 2.0799059929494713e-06,
1758
+ "loss": 0.0001,
1759
+ "step": 500
1760
+ },
1761
+ {
1762
+ "epoch": 0.5881663737551259,
1763
+ "grad_norm": 0.011956258676946163,
1764
+ "learning_rate": 2.0681551116333726e-06,
1765
+ "loss": 0.0001,
1766
+ "step": 502
1767
+ },
1768
+ {
1769
+ "epoch": 0.5905096660808435,
1770
+ "grad_norm": 0.01774289458990097,
1771
+ "learning_rate": 2.056404230317274e-06,
1772
+ "loss": 0.0002,
1773
+ "step": 504
1774
+ },
1775
+ {
1776
+ "epoch": 0.5928529584065613,
1777
+ "grad_norm": 0.21751126646995544,
1778
+ "learning_rate": 2.0446533490011754e-06,
1779
+ "loss": 0.0012,
1780
+ "step": 506
1781
+ },
1782
+ {
1783
+ "epoch": 0.5951962507322789,
1784
+ "grad_norm": 0.00307491235435009,
1785
+ "learning_rate": 2.0329024676850762e-06,
1786
+ "loss": 0.0,
1787
+ "step": 508
1788
+ },
1789
+ {
1790
+ "epoch": 0.5975395430579965,
1791
+ "grad_norm": 0.021330738440155983,
1792
+ "learning_rate": 2.021151586368978e-06,
1793
+ "loss": 0.0002,
1794
+ "step": 510
1795
+ },
1796
+ {
1797
+ "epoch": 0.5998828353837141,
1798
+ "grad_norm": 0.020080704241991043,
1799
+ "learning_rate": 2.009400705052879e-06,
1800
+ "loss": 0.0001,
1801
+ "step": 512
1802
+ },
1803
+ {
1804
+ "epoch": 0.6022261277094317,
1805
+ "grad_norm": 0.020522406324744225,
1806
+ "learning_rate": 1.9976498237367803e-06,
1807
+ "loss": 0.0002,
1808
+ "step": 514
1809
+ },
1810
+ {
1811
+ "epoch": 0.6045694200351494,
1812
+ "grad_norm": 0.0004171329492237419,
1813
+ "learning_rate": 1.985898942420682e-06,
1814
+ "loss": 0.0,
1815
+ "step": 516
1816
+ },
1817
+ {
1818
+ "epoch": 0.606912712360867,
1819
+ "grad_norm": 0.0027696220204234123,
1820
+ "learning_rate": 1.9741480611045828e-06,
1821
+ "loss": 0.0,
1822
+ "step": 518
1823
+ },
1824
+ {
1825
+ "epoch": 0.6092560046865847,
1826
+ "grad_norm": 0.021467505022883415,
1827
+ "learning_rate": 1.9623971797884844e-06,
1828
+ "loss": 0.0002,
1829
+ "step": 520
1830
+ },
1831
+ {
1832
+ "epoch": 0.6115992970123023,
1833
+ "grad_norm": 0.011968536302447319,
1834
+ "learning_rate": 1.9506462984723856e-06,
1835
+ "loss": 0.0001,
1836
+ "step": 522
1837
+ },
1838
+ {
1839
+ "epoch": 0.6139425893380199,
1840
+ "grad_norm": 0.0011503971181809902,
1841
+ "learning_rate": 1.938895417156287e-06,
1842
+ "loss": 0.0004,
1843
+ "step": 524
1844
+ },
1845
+ {
1846
+ "epoch": 0.6162858816637375,
1847
+ "grad_norm": 0.02280554361641407,
1848
+ "learning_rate": 1.927144535840188e-06,
1849
+ "loss": 0.0002,
1850
+ "step": 526
1851
+ },
1852
+ {
1853
+ "epoch": 0.6186291739894552,
1854
+ "grad_norm": 0.008415359072387218,
1855
+ "learning_rate": 1.9153936545240893e-06,
1856
+ "loss": 0.0001,
1857
+ "step": 528
1858
+ },
1859
+ {
1860
+ "epoch": 0.6209724663151728,
1861
+ "grad_norm": 0.0024012764915823936,
1862
+ "learning_rate": 1.9036427732079908e-06,
1863
+ "loss": 0.0001,
1864
+ "step": 530
1865
+ },
1866
+ {
1867
+ "epoch": 0.6233157586408905,
1868
+ "grad_norm": 0.010776808485388756,
1869
+ "learning_rate": 1.8918918918918922e-06,
1870
+ "loss": 0.0001,
1871
+ "step": 532
1872
+ },
1873
+ {
1874
+ "epoch": 0.6256590509666081,
1875
+ "grad_norm": 0.017337538301944733,
1876
+ "learning_rate": 1.8801410105757934e-06,
1877
+ "loss": 0.0001,
1878
+ "step": 534
1879
+ },
1880
+ {
1881
+ "epoch": 0.6280023432923257,
1882
+ "grad_norm": 0.0019926901441067457,
1883
+ "learning_rate": 1.8683901292596946e-06,
1884
+ "loss": 0.0001,
1885
+ "step": 536
1886
+ },
1887
+ {
1888
+ "epoch": 0.6303456356180434,
1889
+ "grad_norm": 0.013480707071721554,
1890
+ "learning_rate": 1.856639247943596e-06,
1891
+ "loss": 0.0002,
1892
+ "step": 538
1893
+ },
1894
+ {
1895
+ "epoch": 0.632688927943761,
1896
+ "grad_norm": 0.005608106963336468,
1897
+ "learning_rate": 1.8448883666274973e-06,
1898
+ "loss": 0.0002,
1899
+ "step": 540
1900
+ },
1901
+ {
1902
+ "epoch": 0.6350322202694786,
1903
+ "grad_norm": 0.002639380283653736,
1904
+ "learning_rate": 1.8331374853113983e-06,
1905
+ "loss": 0.0001,
1906
+ "step": 542
1907
+ },
1908
+ {
1909
+ "epoch": 0.6373755125951962,
1910
+ "grad_norm": 0.0022652854677289724,
1911
+ "learning_rate": 1.8213866039953e-06,
1912
+ "loss": 0.0002,
1913
+ "step": 544
1914
+ },
1915
+ {
1916
+ "epoch": 0.6397188049209139,
1917
+ "grad_norm": 0.003624632954597473,
1918
+ "learning_rate": 1.809635722679201e-06,
1919
+ "loss": 0.0001,
1920
+ "step": 546
1921
+ },
1922
+ {
1923
+ "epoch": 0.6420620972466315,
1924
+ "grad_norm": 0.007647163700312376,
1925
+ "learning_rate": 1.7978848413631022e-06,
1926
+ "loss": 0.0004,
1927
+ "step": 548
1928
+ },
1929
+ {
1930
+ "epoch": 0.6444053895723492,
1931
+ "grad_norm": 0.012163680978119373,
1932
+ "learning_rate": 1.7861339600470036e-06,
1933
+ "loss": 0.0002,
1934
+ "step": 550
1935
+ },
1936
+ {
1937
+ "epoch": 0.6467486818980668,
1938
+ "grad_norm": 0.09023822844028473,
1939
+ "learning_rate": 1.7743830787309049e-06,
1940
+ "loss": 0.0009,
1941
+ "step": 552
1942
+ },
1943
+ {
1944
+ "epoch": 0.6490919742237844,
1945
+ "grad_norm": 0.006924999412149191,
1946
+ "learning_rate": 1.762632197414806e-06,
1947
+ "loss": 0.0001,
1948
+ "step": 554
1949
+ },
1950
+ {
1951
+ "epoch": 0.651435266549502,
1952
+ "grad_norm": 0.0006185275269672275,
1953
+ "learning_rate": 1.7508813160987075e-06,
1954
+ "loss": 0.0001,
1955
+ "step": 556
1956
+ },
1957
+ {
1958
+ "epoch": 0.6537785588752196,
1959
+ "grad_norm": 0.011605402454733849,
1960
+ "learning_rate": 1.7391304347826088e-06,
1961
+ "loss": 0.0006,
1962
+ "step": 558
1963
+ },
1964
+ {
1965
+ "epoch": 0.6561218512009374,
1966
+ "grad_norm": 0.024394473060965538,
1967
+ "learning_rate": 1.7273795534665102e-06,
1968
+ "loss": 0.0001,
1969
+ "step": 560
1970
+ },
1971
+ {
1972
+ "epoch": 0.658465143526655,
1973
+ "grad_norm": 0.023466341197490692,
1974
+ "learning_rate": 1.7156286721504114e-06,
1975
+ "loss": 0.0002,
1976
+ "step": 562
1977
+ },
1978
+ {
1979
+ "epoch": 0.6608084358523726,
1980
+ "grad_norm": 0.010153519921004772,
1981
+ "learning_rate": 1.7038777908343126e-06,
1982
+ "loss": 0.0004,
1983
+ "step": 564
1984
+ },
1985
+ {
1986
+ "epoch": 0.6631517281780902,
1987
+ "grad_norm": 0.43800845742225647,
1988
+ "learning_rate": 1.692126909518214e-06,
1989
+ "loss": 0.0012,
1990
+ "step": 566
1991
+ },
1992
+ {
1993
+ "epoch": 0.6654950205038078,
1994
+ "grad_norm": 0.008404972031712532,
1995
+ "learning_rate": 1.6803760282021153e-06,
1996
+ "loss": 0.0001,
1997
+ "step": 568
1998
+ },
1999
+ {
2000
+ "epoch": 0.6678383128295254,
2001
+ "grad_norm": 0.10615257918834686,
2002
+ "learning_rate": 1.6686251468860165e-06,
2003
+ "loss": 0.0005,
2004
+ "step": 570
2005
+ },
2006
+ {
2007
+ "epoch": 0.6701816051552432,
2008
+ "grad_norm": 0.019307592883706093,
2009
+ "learning_rate": 1.656874265569918e-06,
2010
+ "loss": 0.0003,
2011
+ "step": 572
2012
+ },
2013
+ {
2014
+ "epoch": 0.6725248974809608,
2015
+ "grad_norm": 0.012227280996739864,
2016
+ "learning_rate": 1.6451233842538192e-06,
2017
+ "loss": 0.0002,
2018
+ "step": 574
2019
+ },
2020
+ {
2021
+ "epoch": 0.6748681898066784,
2022
+ "grad_norm": 0.002821948379278183,
2023
+ "learning_rate": 1.6333725029377204e-06,
2024
+ "loss": 0.0,
2025
+ "step": 576
2026
+ },
2027
+ {
2028
+ "epoch": 0.677211482132396,
2029
+ "grad_norm": 0.010473825968801975,
2030
+ "learning_rate": 1.6216216216216219e-06,
2031
+ "loss": 0.0003,
2032
+ "step": 578
2033
+ },
2034
+ {
2035
+ "epoch": 0.6795547744581136,
2036
+ "grad_norm": 0.014046385884284973,
2037
+ "learning_rate": 1.609870740305523e-06,
2038
+ "loss": 0.0236,
2039
+ "step": 580
2040
+ },
2041
+ {
2042
+ "epoch": 0.6818980667838312,
2043
+ "grad_norm": 0.0017795696621760726,
2044
+ "learning_rate": 1.5981198589894243e-06,
2045
+ "loss": 0.0001,
2046
+ "step": 582
2047
+ },
2048
+ {
2049
+ "epoch": 0.6842413591095489,
2050
+ "grad_norm": 0.0006959863239899278,
2051
+ "learning_rate": 1.5863689776733257e-06,
2052
+ "loss": 0.0002,
2053
+ "step": 584
2054
+ },
2055
+ {
2056
+ "epoch": 0.6865846514352666,
2057
+ "grad_norm": 0.019652947783470154,
2058
+ "learning_rate": 1.574618096357227e-06,
2059
+ "loss": 0.0003,
2060
+ "step": 586
2061
+ },
2062
+ {
2063
+ "epoch": 0.6889279437609842,
2064
+ "grad_norm": 0.002340570092201233,
2065
+ "learning_rate": 1.5628672150411282e-06,
2066
+ "loss": 0.0,
2067
+ "step": 588
2068
+ },
2069
+ {
2070
+ "epoch": 0.6912712360867018,
2071
+ "grad_norm": 0.011190817691385746,
2072
+ "learning_rate": 1.5511163337250296e-06,
2073
+ "loss": 0.0002,
2074
+ "step": 590
2075
+ },
2076
+ {
2077
+ "epoch": 0.6936145284124194,
2078
+ "grad_norm": 0.001152676297351718,
2079
+ "learning_rate": 1.5393654524089308e-06,
2080
+ "loss": 0.0001,
2081
+ "step": 592
2082
+ },
2083
+ {
2084
+ "epoch": 0.6959578207381371,
2085
+ "grad_norm": 0.003393592080101371,
2086
+ "learning_rate": 1.5276145710928319e-06,
2087
+ "loss": 0.0001,
2088
+ "step": 594
2089
+ },
2090
+ {
2091
+ "epoch": 0.6983011130638547,
2092
+ "grad_norm": 0.007921353913843632,
2093
+ "learning_rate": 1.5158636897767335e-06,
2094
+ "loss": 0.0001,
2095
+ "step": 596
2096
+ },
2097
+ {
2098
+ "epoch": 0.7006444053895724,
2099
+ "grad_norm": 0.1039208471775055,
2100
+ "learning_rate": 1.5041128084606345e-06,
2101
+ "loss": 0.0002,
2102
+ "step": 598
2103
+ },
2104
+ {
2105
+ "epoch": 0.70298769771529,
2106
+ "grad_norm": 0.0011576958931982517,
2107
+ "learning_rate": 1.4923619271445362e-06,
2108
+ "loss": 0.0001,
2109
+ "step": 600
2110
+ },
2111
+ {
2112
+ "epoch": 0.7053309900410076,
2113
+ "grad_norm": 0.06407307088375092,
2114
+ "learning_rate": 1.4806110458284372e-06,
2115
+ "loss": 0.0003,
2116
+ "step": 602
2117
+ },
2118
+ {
2119
+ "epoch": 0.7076742823667252,
2120
+ "grad_norm": 0.012639104388654232,
2121
+ "learning_rate": 1.4688601645123384e-06,
2122
+ "loss": 0.0002,
2123
+ "step": 604
2124
+ },
2125
+ {
2126
+ "epoch": 0.7100175746924429,
2127
+ "grad_norm": 0.0019591290038079023,
2128
+ "learning_rate": 1.45710928319624e-06,
2129
+ "loss": 0.0068,
2130
+ "step": 606
2131
+ },
2132
+ {
2133
+ "epoch": 0.7123608670181605,
2134
+ "grad_norm": 0.0008327167597599328,
2135
+ "learning_rate": 1.445358401880141e-06,
2136
+ "loss": 0.0001,
2137
+ "step": 608
2138
+ },
2139
+ {
2140
+ "epoch": 0.7147041593438781,
2141
+ "grad_norm": 0.0013139324728399515,
2142
+ "learning_rate": 1.4336075205640423e-06,
2143
+ "loss": 0.0,
2144
+ "step": 610
2145
+ },
2146
+ {
2147
+ "epoch": 0.7170474516695958,
2148
+ "grad_norm": 0.00803992711007595,
2149
+ "learning_rate": 1.4218566392479437e-06,
2150
+ "loss": 0.0002,
2151
+ "step": 612
2152
+ },
2153
+ {
2154
+ "epoch": 0.7193907439953134,
2155
+ "grad_norm": 0.011399227194488049,
2156
+ "learning_rate": 1.410105757931845e-06,
2157
+ "loss": 0.0002,
2158
+ "step": 614
2159
+ },
2160
+ {
2161
+ "epoch": 0.7217340363210311,
2162
+ "grad_norm": 0.007171169854700565,
2163
+ "learning_rate": 1.3983548766157462e-06,
2164
+ "loss": 0.0002,
2165
+ "step": 616
2166
+ },
2167
+ {
2168
+ "epoch": 0.7240773286467487,
2169
+ "grad_norm": 0.7272996306419373,
2170
+ "learning_rate": 1.3866039952996476e-06,
2171
+ "loss": 0.0028,
2172
+ "step": 618
2173
+ },
2174
+ {
2175
+ "epoch": 0.7264206209724663,
2176
+ "grad_norm": 0.0037387118209153414,
2177
+ "learning_rate": 1.3748531139835488e-06,
2178
+ "loss": 0.0001,
2179
+ "step": 620
2180
+ },
2181
+ {
2182
+ "epoch": 0.7287639132981839,
2183
+ "grad_norm": 0.015048849396407604,
2184
+ "learning_rate": 1.36310223266745e-06,
2185
+ "loss": 0.0002,
2186
+ "step": 622
2187
+ },
2188
+ {
2189
+ "epoch": 0.7311072056239016,
2190
+ "grad_norm": 0.0023705060593783855,
2191
+ "learning_rate": 1.3513513513513515e-06,
2192
+ "loss": 0.0001,
2193
+ "step": 624
2194
+ },
2195
+ {
2196
+ "epoch": 0.7334504979496193,
2197
+ "grad_norm": 0.03966263309121132,
2198
+ "learning_rate": 1.3396004700352527e-06,
2199
+ "loss": 0.0003,
2200
+ "step": 626
2201
+ },
2202
+ {
2203
+ "epoch": 0.7357937902753369,
2204
+ "grad_norm": 0.0033043306320905685,
2205
+ "learning_rate": 1.327849588719154e-06,
2206
+ "loss": 0.0004,
2207
+ "step": 628
2208
+ },
2209
+ {
2210
+ "epoch": 0.7381370826010545,
2211
+ "grad_norm": 0.35459718108177185,
2212
+ "learning_rate": 1.3160987074030554e-06,
2213
+ "loss": 0.0034,
2214
+ "step": 630
2215
+ },
2216
+ {
2217
+ "epoch": 0.7404803749267721,
2218
+ "grad_norm": 0.016441915184259415,
2219
+ "learning_rate": 1.3043478260869566e-06,
2220
+ "loss": 0.0002,
2221
+ "step": 632
2222
+ },
2223
+ {
2224
+ "epoch": 0.7428236672524897,
2225
+ "grad_norm": 0.0045352657325565815,
2226
+ "learning_rate": 1.2925969447708578e-06,
2227
+ "loss": 0.0002,
2228
+ "step": 634
2229
+ },
2230
+ {
2231
+ "epoch": 0.7451669595782073,
2232
+ "grad_norm": 0.06311573088169098,
2233
+ "learning_rate": 1.2808460634547593e-06,
2234
+ "loss": 0.0005,
2235
+ "step": 636
2236
+ },
2237
+ {
2238
+ "epoch": 0.7475102519039251,
2239
+ "grad_norm": 0.11154340207576752,
2240
+ "learning_rate": 1.2690951821386605e-06,
2241
+ "loss": 0.0009,
2242
+ "step": 638
2243
+ },
2244
+ {
2245
+ "epoch": 0.7498535442296427,
2246
+ "grad_norm": 0.01816423609852791,
2247
+ "learning_rate": 1.257344300822562e-06,
2248
+ "loss": 0.0006,
2249
+ "step": 640
2250
+ },
2251
+ {
2252
+ "epoch": 0.7521968365553603,
2253
+ "grad_norm": 0.027273530140519142,
2254
+ "learning_rate": 1.2455934195064632e-06,
2255
+ "loss": 0.0005,
2256
+ "step": 642
2257
+ },
2258
+ {
2259
+ "epoch": 0.7545401288810779,
2260
+ "grad_norm": 0.006555743515491486,
2261
+ "learning_rate": 1.2338425381903644e-06,
2262
+ "loss": 0.0003,
2263
+ "step": 644
2264
+ },
2265
+ {
2266
+ "epoch": 0.7568834212067955,
2267
+ "grad_norm": 0.0030812753830105066,
2268
+ "learning_rate": 1.2220916568742656e-06,
2269
+ "loss": 0.0279,
2270
+ "step": 646
2271
+ },
2272
+ {
2273
+ "epoch": 0.7592267135325131,
2274
+ "grad_norm": 0.01702543906867504,
2275
+ "learning_rate": 1.210340775558167e-06,
2276
+ "loss": 0.0001,
2277
+ "step": 648
2278
+ },
2279
+ {
2280
+ "epoch": 0.7615700058582309,
2281
+ "grad_norm": 0.02607725001871586,
2282
+ "learning_rate": 1.1985898942420683e-06,
2283
+ "loss": 0.0001,
2284
+ "step": 650
2285
+ },
2286
+ {
2287
+ "epoch": 0.7639132981839485,
2288
+ "grad_norm": 0.006388965994119644,
2289
+ "learning_rate": 1.1868390129259695e-06,
2290
+ "loss": 0.0001,
2291
+ "step": 652
2292
+ },
2293
+ {
2294
+ "epoch": 0.7662565905096661,
2295
+ "grad_norm": 0.008253968320786953,
2296
+ "learning_rate": 1.175088131609871e-06,
2297
+ "loss": 0.0001,
2298
+ "step": 654
2299
+ },
2300
+ {
2301
+ "epoch": 0.7685998828353837,
2302
+ "grad_norm": 0.004699599463492632,
2303
+ "learning_rate": 1.1633372502937722e-06,
2304
+ "loss": 0.0002,
2305
+ "step": 656
2306
+ },
2307
+ {
2308
+ "epoch": 0.7709431751611013,
2309
+ "grad_norm": 0.0012458263663575053,
2310
+ "learning_rate": 1.1515863689776734e-06,
2311
+ "loss": 0.0122,
2312
+ "step": 658
2313
+ },
2314
+ {
2315
+ "epoch": 0.773286467486819,
2316
+ "grad_norm": 0.02383268252015114,
2317
+ "learning_rate": 1.1398354876615746e-06,
2318
+ "loss": 0.0003,
2319
+ "step": 660
2320
+ },
2321
+ {
2322
+ "epoch": 0.7756297598125366,
2323
+ "grad_norm": 0.015058089047670364,
2324
+ "learning_rate": 1.128084606345476e-06,
2325
+ "loss": 0.0001,
2326
+ "step": 662
2327
+ },
2328
+ {
2329
+ "epoch": 0.7779730521382543,
2330
+ "grad_norm": 0.01569475792348385,
2331
+ "learning_rate": 1.1163337250293773e-06,
2332
+ "loss": 0.0003,
2333
+ "step": 664
2334
+ },
2335
+ {
2336
+ "epoch": 0.7803163444639719,
2337
+ "grad_norm": 0.04253750294446945,
2338
+ "learning_rate": 1.1045828437132785e-06,
2339
+ "loss": 0.0002,
2340
+ "step": 666
2341
+ },
2342
+ {
2343
+ "epoch": 0.7826596367896895,
2344
+ "grad_norm": 0.015156907960772514,
2345
+ "learning_rate": 1.09283196239718e-06,
2346
+ "loss": 0.0002,
2347
+ "step": 668
2348
+ },
2349
+ {
2350
+ "epoch": 0.7850029291154071,
2351
+ "grad_norm": 0.03742622211575508,
2352
+ "learning_rate": 1.0810810810810812e-06,
2353
+ "loss": 0.0005,
2354
+ "step": 670
2355
+ },
2356
+ {
2357
+ "epoch": 0.7873462214411248,
2358
+ "grad_norm": 0.027262985706329346,
2359
+ "learning_rate": 1.0693301997649824e-06,
2360
+ "loss": 0.0002,
2361
+ "step": 672
2362
+ },
2363
+ {
2364
+ "epoch": 0.7896895137668424,
2365
+ "grad_norm": 0.007641313597559929,
2366
+ "learning_rate": 1.0575793184488838e-06,
2367
+ "loss": 0.0002,
2368
+ "step": 674
2369
+ },
2370
+ {
2371
+ "epoch": 0.79203280609256,
2372
+ "grad_norm": 0.04441560059785843,
2373
+ "learning_rate": 1.045828437132785e-06,
2374
+ "loss": 0.0005,
2375
+ "step": 676
2376
+ },
2377
+ {
2378
+ "epoch": 0.7943760984182777,
2379
+ "grad_norm": 0.020478103309869766,
2380
+ "learning_rate": 1.0340775558166863e-06,
2381
+ "loss": 0.0002,
2382
+ "step": 678
2383
+ },
2384
+ {
2385
+ "epoch": 0.7967193907439953,
2386
+ "grad_norm": 0.10936477035284042,
2387
+ "learning_rate": 1.0223266745005877e-06,
2388
+ "loss": 0.001,
2389
+ "step": 680
2390
+ },
2391
+ {
2392
+ "epoch": 0.799062683069713,
2393
+ "grad_norm": 0.01284460723400116,
2394
+ "learning_rate": 1.010575793184489e-06,
2395
+ "loss": 0.0015,
2396
+ "step": 682
2397
+ },
2398
+ {
2399
+ "epoch": 0.8014059753954306,
2400
+ "grad_norm": 0.003440434578806162,
2401
+ "learning_rate": 9.988249118683902e-07,
2402
+ "loss": 0.0,
2403
+ "step": 684
2404
+ },
2405
+ {
2406
+ "epoch": 0.8037492677211482,
2407
+ "grad_norm": 0.013081365264952183,
2408
+ "learning_rate": 9.870740305522914e-07,
2409
+ "loss": 0.0009,
2410
+ "step": 686
2411
+ },
2412
+ {
2413
+ "epoch": 0.8060925600468658,
2414
+ "grad_norm": 0.013380183838307858,
2415
+ "learning_rate": 9.753231492361928e-07,
2416
+ "loss": 0.0002,
2417
+ "step": 688
2418
+ },
2419
+ {
2420
+ "epoch": 0.8084358523725835,
2421
+ "grad_norm": 0.03771582618355751,
2422
+ "learning_rate": 9.63572267920094e-07,
2423
+ "loss": 0.0003,
2424
+ "step": 690
2425
+ },
2426
+ {
2427
+ "epoch": 0.8107791446983011,
2428
+ "grad_norm": 0.0009556732256896794,
2429
+ "learning_rate": 9.518213866039954e-07,
2430
+ "loss": 0.0005,
2431
+ "step": 692
2432
+ },
2433
+ {
2434
+ "epoch": 0.8131224370240188,
2435
+ "grad_norm": 0.0019481348572298884,
2436
+ "learning_rate": 9.400705052878967e-07,
2437
+ "loss": 0.0001,
2438
+ "step": 694
2439
+ },
2440
+ {
2441
+ "epoch": 0.8154657293497364,
2442
+ "grad_norm": 0.0021866948809474707,
2443
+ "learning_rate": 9.28319623971798e-07,
2444
+ "loss": 0.0002,
2445
+ "step": 696
2446
+ },
2447
+ {
2448
+ "epoch": 0.817809021675454,
2449
+ "grad_norm": 0.007546517997980118,
2450
+ "learning_rate": 9.165687426556992e-07,
2451
+ "loss": 0.0007,
2452
+ "step": 698
2453
+ },
2454
+ {
2455
+ "epoch": 0.8201523140011716,
2456
+ "grad_norm": 2.074432611465454,
2457
+ "learning_rate": 9.048178613396005e-07,
2458
+ "loss": 0.0251,
2459
+ "step": 700
2460
+ },
2461
+ {
2462
+ "epoch": 0.8224956063268892,
2463
+ "grad_norm": 0.003374068532139063,
2464
+ "learning_rate": 8.930669800235018e-07,
2465
+ "loss": 0.0001,
2466
+ "step": 702
2467
+ },
2468
+ {
2469
+ "epoch": 0.824838898652607,
2470
+ "grad_norm": 0.010109562426805496,
2471
+ "learning_rate": 8.81316098707403e-07,
2472
+ "loss": 0.0006,
2473
+ "step": 704
2474
+ },
2475
+ {
2476
+ "epoch": 0.8271821909783246,
2477
+ "grad_norm": 0.017352379858493805,
2478
+ "learning_rate": 8.695652173913044e-07,
2479
+ "loss": 0.0001,
2480
+ "step": 706
2481
+ },
2482
+ {
2483
+ "epoch": 0.8295254833040422,
2484
+ "grad_norm": 0.016872087493538857,
2485
+ "learning_rate": 8.578143360752057e-07,
2486
+ "loss": 0.0002,
2487
+ "step": 708
2488
+ },
2489
+ {
2490
+ "epoch": 0.8318687756297598,
2491
+ "grad_norm": 0.041937246918678284,
2492
+ "learning_rate": 8.46063454759107e-07,
2493
+ "loss": 0.0228,
2494
+ "step": 710
2495
+ },
2496
+ {
2497
+ "epoch": 0.8342120679554774,
2498
+ "grad_norm": 0.02908233553171158,
2499
+ "learning_rate": 8.343125734430083e-07,
2500
+ "loss": 0.0002,
2501
+ "step": 712
2502
+ },
2503
+ {
2504
+ "epoch": 0.836555360281195,
2505
+ "grad_norm": 0.0012463816674426198,
2506
+ "learning_rate": 8.225616921269096e-07,
2507
+ "loss": 0.0004,
2508
+ "step": 714
2509
+ },
2510
+ {
2511
+ "epoch": 0.8388986526069128,
2512
+ "grad_norm": 0.04300675913691521,
2513
+ "learning_rate": 8.108108108108109e-07,
2514
+ "loss": 0.0006,
2515
+ "step": 716
2516
+ },
2517
+ {
2518
+ "epoch": 0.8412419449326304,
2519
+ "grad_norm": 2.7622828483581543,
2520
+ "learning_rate": 7.990599294947122e-07,
2521
+ "loss": 0.149,
2522
+ "step": 718
2523
+ },
2524
+ {
2525
+ "epoch": 0.843585237258348,
2526
+ "grad_norm": 0.010049765929579735,
2527
+ "learning_rate": 7.873090481786135e-07,
2528
+ "loss": 0.0002,
2529
+ "step": 720
2530
+ },
2531
+ {
2532
+ "epoch": 0.8459285295840656,
2533
+ "grad_norm": 0.011876920238137245,
2534
+ "learning_rate": 7.755581668625148e-07,
2535
+ "loss": 0.0001,
2536
+ "step": 722
2537
+ },
2538
+ {
2539
+ "epoch": 0.8482718219097832,
2540
+ "grad_norm": 0.014826681464910507,
2541
+ "learning_rate": 7.638072855464159e-07,
2542
+ "loss": 0.0003,
2543
+ "step": 724
2544
+ },
2545
+ {
2546
+ "epoch": 0.8506151142355008,
2547
+ "grad_norm": 0.16368882358074188,
2548
+ "learning_rate": 7.520564042303173e-07,
2549
+ "loss": 0.0013,
2550
+ "step": 726
2551
+ },
2552
+ {
2553
+ "epoch": 0.8529584065612185,
2554
+ "grad_norm": 0.02603282406926155,
2555
+ "learning_rate": 7.403055229142186e-07,
2556
+ "loss": 0.0004,
2557
+ "step": 728
2558
+ },
2559
+ {
2560
+ "epoch": 0.8553016988869362,
2561
+ "grad_norm": 0.7740702629089355,
2562
+ "learning_rate": 7.2855464159812e-07,
2563
+ "loss": 0.0043,
2564
+ "step": 730
2565
+ },
2566
+ {
2567
+ "epoch": 0.8576449912126538,
2568
+ "grad_norm": 0.010226438753306866,
2569
+ "learning_rate": 7.168037602820211e-07,
2570
+ "loss": 0.0002,
2571
+ "step": 732
2572
+ },
2573
+ {
2574
+ "epoch": 0.8599882835383714,
2575
+ "grad_norm": 0.02008165791630745,
2576
+ "learning_rate": 7.050528789659225e-07,
2577
+ "loss": 0.0002,
2578
+ "step": 734
2579
+ },
2580
+ {
2581
+ "epoch": 0.862331575864089,
2582
+ "grad_norm": 0.09208586066961288,
2583
+ "learning_rate": 6.933019976498238e-07,
2584
+ "loss": 0.0008,
2585
+ "step": 736
2586
+ },
2587
+ {
2588
+ "epoch": 0.8646748681898067,
2589
+ "grad_norm": 0.01933148130774498,
2590
+ "learning_rate": 6.81551116333725e-07,
2591
+ "loss": 0.0011,
2592
+ "step": 738
2593
+ },
2594
+ {
2595
+ "epoch": 0.8670181605155243,
2596
+ "grad_norm": 0.04433580860495567,
2597
+ "learning_rate": 6.698002350176264e-07,
2598
+ "loss": 0.0003,
2599
+ "step": 740
2600
+ },
2601
+ {
2602
+ "epoch": 0.869361452841242,
2603
+ "grad_norm": 0.01631711982190609,
2604
+ "learning_rate": 6.580493537015277e-07,
2605
+ "loss": 0.0003,
2606
+ "step": 742
2607
+ },
2608
+ {
2609
+ "epoch": 0.8717047451669596,
2610
+ "grad_norm": 0.042307399213314056,
2611
+ "learning_rate": 6.462984723854289e-07,
2612
+ "loss": 0.0004,
2613
+ "step": 744
2614
+ },
2615
+ {
2616
+ "epoch": 0.8740480374926772,
2617
+ "grad_norm": 0.22414757311344147,
2618
+ "learning_rate": 6.345475910693303e-07,
2619
+ "loss": 0.0018,
2620
+ "step": 746
2621
+ },
2622
+ {
2623
+ "epoch": 0.8763913298183948,
2624
+ "grad_norm": 0.17513447999954224,
2625
+ "learning_rate": 6.227967097532316e-07,
2626
+ "loss": 0.0015,
2627
+ "step": 748
2628
+ },
2629
+ {
2630
+ "epoch": 0.8787346221441125,
2631
+ "grad_norm": 0.3218580186367035,
2632
+ "learning_rate": 6.110458284371328e-07,
2633
+ "loss": 0.0029,
2634
+ "step": 750
2635
+ },
2636
+ {
2637
+ "epoch": 0.8810779144698301,
2638
+ "grad_norm": 0.026706017553806305,
2639
+ "learning_rate": 5.992949471210341e-07,
2640
+ "loss": 0.0004,
2641
+ "step": 752
2642
+ },
2643
+ {
2644
+ "epoch": 0.8834212067955477,
2645
+ "grad_norm": 0.4114263951778412,
2646
+ "learning_rate": 5.875440658049355e-07,
2647
+ "loss": 0.0035,
2648
+ "step": 754
2649
+ },
2650
+ {
2651
+ "epoch": 0.8857644991212654,
2652
+ "grad_norm": 0.25009235739707947,
2653
+ "learning_rate": 5.757931844888367e-07,
2654
+ "loss": 0.0016,
2655
+ "step": 756
2656
+ },
2657
+ {
2658
+ "epoch": 0.888107791446983,
2659
+ "grad_norm": 1.2960833311080933,
2660
+ "learning_rate": 5.64042303172738e-07,
2661
+ "loss": 0.0059,
2662
+ "step": 758
2663
+ },
2664
+ {
2665
+ "epoch": 0.8904510837727007,
2666
+ "grad_norm": 0.28417083621025085,
2667
+ "learning_rate": 5.522914218566393e-07,
2668
+ "loss": 0.0059,
2669
+ "step": 760
2670
+ },
2671
+ {
2672
+ "epoch": 0.8927943760984183,
2673
+ "grad_norm": 0.2292051613330841,
2674
+ "learning_rate": 5.405405405405406e-07,
2675
+ "loss": 0.0015,
2676
+ "step": 762
2677
+ },
2678
+ {
2679
+ "epoch": 0.8951376684241359,
2680
+ "grad_norm": 0.012189504690468311,
2681
+ "learning_rate": 5.287896592244419e-07,
2682
+ "loss": 0.0007,
2683
+ "step": 764
2684
+ },
2685
+ {
2686
+ "epoch": 0.8974809607498535,
2687
+ "grad_norm": 0.09458251297473907,
2688
+ "learning_rate": 5.170387779083431e-07,
2689
+ "loss": 0.0004,
2690
+ "step": 766
2691
+ },
2692
+ {
2693
+ "epoch": 0.8998242530755711,
2694
+ "grad_norm": 0.027070222422480583,
2695
+ "learning_rate": 5.052878965922445e-07,
2696
+ "loss": 0.0012,
2697
+ "step": 768
2698
+ },
2699
+ {
2700
+ "epoch": 0.9021675454012889,
2701
+ "grad_norm": 0.047401878982782364,
2702
+ "learning_rate": 4.935370152761457e-07,
2703
+ "loss": 0.0003,
2704
+ "step": 770
2705
+ },
2706
+ {
2707
+ "epoch": 0.9045108377270065,
2708
+ "grad_norm": 0.06239737570285797,
2709
+ "learning_rate": 4.81786133960047e-07,
2710
+ "loss": 0.0012,
2711
+ "step": 772
2712
+ },
2713
+ {
2714
+ "epoch": 0.9068541300527241,
2715
+ "grad_norm": 2.6842846870422363,
2716
+ "learning_rate": 4.7003525264394836e-07,
2717
+ "loss": 0.1103,
2718
+ "step": 774
2719
+ },
2720
+ {
2721
+ "epoch": 0.9091974223784417,
2722
+ "grad_norm": 0.057395774871110916,
2723
+ "learning_rate": 4.582843713278496e-07,
2724
+ "loss": 0.0004,
2725
+ "step": 776
2726
+ },
2727
+ {
2728
+ "epoch": 0.9115407147041593,
2729
+ "grad_norm": 0.16248440742492676,
2730
+ "learning_rate": 4.465334900117509e-07,
2731
+ "loss": 0.0018,
2732
+ "step": 778
2733
+ },
2734
+ {
2735
+ "epoch": 0.9138840070298769,
2736
+ "grad_norm": 0.11067284643650055,
2737
+ "learning_rate": 4.347826086956522e-07,
2738
+ "loss": 0.0011,
2739
+ "step": 780
2740
+ },
2741
+ {
2742
+ "epoch": 0.9162272993555947,
2743
+ "grad_norm": 0.07208680361509323,
2744
+ "learning_rate": 4.230317273795535e-07,
2745
+ "loss": 0.0011,
2746
+ "step": 782
2747
+ },
2748
+ {
2749
+ "epoch": 0.9185705916813123,
2750
+ "grad_norm": 0.4830150604248047,
2751
+ "learning_rate": 4.112808460634548e-07,
2752
+ "loss": 0.0022,
2753
+ "step": 784
2754
+ },
2755
+ {
2756
+ "epoch": 0.9209138840070299,
2757
+ "grad_norm": 0.01794450171291828,
2758
+ "learning_rate": 3.995299647473561e-07,
2759
+ "loss": 0.0011,
2760
+ "step": 786
2761
+ },
2762
+ {
2763
+ "epoch": 0.9232571763327475,
2764
+ "grad_norm": 3.0485081672668457,
2765
+ "learning_rate": 3.877790834312574e-07,
2766
+ "loss": 0.0508,
2767
+ "step": 788
2768
+ },
2769
+ {
2770
+ "epoch": 0.9256004686584651,
2771
+ "grad_norm": 3.130112648010254,
2772
+ "learning_rate": 3.7602820211515863e-07,
2773
+ "loss": 0.0194,
2774
+ "step": 790
2775
+ },
2776
+ {
2777
+ "epoch": 0.9279437609841827,
2778
+ "grad_norm": 3.5992815494537354,
2779
+ "learning_rate": 3.6427732079906e-07,
2780
+ "loss": 0.1036,
2781
+ "step": 792
2782
+ },
2783
+ {
2784
+ "epoch": 0.9302870533099004,
2785
+ "grad_norm": 0.0751647800207138,
2786
+ "learning_rate": 3.5252643948296124e-07,
2787
+ "loss": 0.0003,
2788
+ "step": 794
2789
+ },
2790
+ {
2791
+ "epoch": 0.9326303456356181,
2792
+ "grad_norm": 0.03622612729668617,
2793
+ "learning_rate": 3.407755581668625e-07,
2794
+ "loss": 0.0011,
2795
+ "step": 796
2796
+ },
2797
+ {
2798
+ "epoch": 0.9349736379613357,
2799
+ "grad_norm": 0.22365981340408325,
2800
+ "learning_rate": 3.2902467685076385e-07,
2801
+ "loss": 0.0028,
2802
+ "step": 798
2803
+ },
2804
+ {
2805
+ "epoch": 0.9373169302870533,
2806
+ "grad_norm": 0.04666091129183769,
2807
+ "learning_rate": 3.172737955346651e-07,
2808
+ "loss": 0.0041,
2809
+ "step": 800
2810
+ },
2811
+ {
2812
+ "epoch": 0.9396602226127709,
2813
+ "grad_norm": 5.363467693328857,
2814
+ "learning_rate": 3.055229142185664e-07,
2815
+ "loss": 0.2217,
2816
+ "step": 802
2817
+ },
2818
+ {
2819
+ "epoch": 0.9420035149384886,
2820
+ "grad_norm": 0.06753694266080856,
2821
+ "learning_rate": 2.9377203290246774e-07,
2822
+ "loss": 0.0026,
2823
+ "step": 804
2824
+ },
2825
+ {
2826
+ "epoch": 0.9443468072642062,
2827
+ "grad_norm": 2.554419994354248,
2828
+ "learning_rate": 2.82021151586369e-07,
2829
+ "loss": 0.0791,
2830
+ "step": 806
2831
+ },
2832
+ {
2833
+ "epoch": 0.9466900995899239,
2834
+ "grad_norm": 0.14563411474227905,
2835
+ "learning_rate": 2.702702702702703e-07,
2836
+ "loss": 0.0208,
2837
+ "step": 808
2838
+ },
2839
+ {
2840
+ "epoch": 0.9490333919156415,
2841
+ "grad_norm": 2.30971360206604,
2842
+ "learning_rate": 2.5851938895417157e-07,
2843
+ "loss": 0.1119,
2844
+ "step": 810
2845
+ },
2846
+ {
2847
+ "epoch": 0.9513766842413591,
2848
+ "grad_norm": 4.073694229125977,
2849
+ "learning_rate": 2.4676850763807285e-07,
2850
+ "loss": 0.1057,
2851
+ "step": 812
2852
+ },
2853
+ {
2854
+ "epoch": 0.9537199765670767,
2855
+ "grad_norm": 2.3215789794921875,
2856
+ "learning_rate": 2.3501762632197418e-07,
2857
+ "loss": 0.0286,
2858
+ "step": 814
2859
+ },
2860
+ {
2861
+ "epoch": 0.9560632688927944,
2862
+ "grad_norm": 0.46727773547172546,
2863
+ "learning_rate": 2.2326674500587546e-07,
2864
+ "loss": 0.0714,
2865
+ "step": 816
2866
+ },
2867
+ {
2868
+ "epoch": 0.958406561218512,
2869
+ "grad_norm": 2.0026137828826904,
2870
+ "learning_rate": 2.1151586368977676e-07,
2871
+ "loss": 0.0455,
2872
+ "step": 818
2873
+ },
2874
+ {
2875
+ "epoch": 0.9607498535442296,
2876
+ "grad_norm": 3.2537143230438232,
2877
+ "learning_rate": 1.9976498237367804e-07,
2878
+ "loss": 0.0765,
2879
+ "step": 820
2880
+ },
2881
+ {
2882
+ "epoch": 0.9630931458699473,
2883
+ "grad_norm": 3.485633134841919,
2884
+ "learning_rate": 1.8801410105757932e-07,
2885
+ "loss": 0.0493,
2886
+ "step": 822
2887
+ },
2888
+ {
2889
+ "epoch": 0.9654364381956649,
2890
+ "grad_norm": 2.769423246383667,
2891
+ "learning_rate": 1.7626321974148062e-07,
2892
+ "loss": 0.0602,
2893
+ "step": 824
2894
+ },
2895
+ {
2896
+ "epoch": 0.9677797305213826,
2897
+ "grad_norm": 2.236210823059082,
2898
+ "learning_rate": 1.6451233842538192e-07,
2899
+ "loss": 0.1404,
2900
+ "step": 826
2901
+ },
2902
+ {
2903
+ "epoch": 0.9701230228471002,
2904
+ "grad_norm": 0.06197360157966614,
2905
+ "learning_rate": 1.527614571092832e-07,
2906
+ "loss": 0.0472,
2907
+ "step": 828
2908
+ },
2909
+ {
2910
+ "epoch": 0.9724663151728178,
2911
+ "grad_norm": 0.8206185698509216,
2912
+ "learning_rate": 1.410105757931845e-07,
2913
+ "loss": 0.0686,
2914
+ "step": 830
2915
+ },
2916
+ {
2917
+ "epoch": 0.9748096074985354,
2918
+ "grad_norm": 2.434030771255493,
2919
+ "learning_rate": 1.2925969447708578e-07,
2920
+ "loss": 0.1322,
2921
+ "step": 832
2922
+ },
2923
+ {
2924
+ "epoch": 0.9771528998242531,
2925
+ "grad_norm": 0.03143630549311638,
2926
+ "learning_rate": 1.1750881316098709e-07,
2927
+ "loss": 0.1134,
2928
+ "step": 834
2929
+ },
2930
+ {
2931
+ "epoch": 0.9794961921499707,
2932
+ "grad_norm": 0.1770186424255371,
2933
+ "learning_rate": 1.0575793184488838e-07,
2934
+ "loss": 0.0011,
2935
+ "step": 836
2936
+ },
2937
+ {
2938
+ "epoch": 0.9818394844756884,
2939
+ "grad_norm": 6.03350830078125,
2940
+ "learning_rate": 9.400705052878966e-08,
2941
+ "loss": 0.4193,
2942
+ "step": 838
2943
+ },
2944
+ {
2945
+ "epoch": 0.984182776801406,
2946
+ "grad_norm": 4.842612266540527,
2947
+ "learning_rate": 8.225616921269096e-08,
2948
+ "loss": 0.0951,
2949
+ "step": 840
2950
+ },
2951
+ {
2952
+ "epoch": 0.9865260691271236,
2953
+ "grad_norm": 3.111945629119873,
2954
+ "learning_rate": 7.050528789659225e-08,
2955
+ "loss": 0.1375,
2956
+ "step": 842
2957
+ },
2958
+ {
2959
+ "epoch": 0.9888693614528412,
2960
+ "grad_norm": 3.4468753337860107,
2961
+ "learning_rate": 5.8754406580493544e-08,
2962
+ "loss": 0.157,
2963
+ "step": 844
2964
+ },
2965
+ {
2966
+ "epoch": 0.9912126537785588,
2967
+ "grad_norm": 5.563467502593994,
2968
+ "learning_rate": 4.700352526439483e-08,
2969
+ "loss": 0.1989,
2970
+ "step": 846
2971
+ },
2972
+ {
2973
+ "epoch": 0.9935559461042766,
2974
+ "grad_norm": 0.20900146663188934,
2975
+ "learning_rate": 3.5252643948296127e-08,
2976
+ "loss": 0.169,
2977
+ "step": 848
2978
+ },
2979
+ {
2980
+ "epoch": 0.9958992384299942,
2981
+ "grad_norm": 2.651283025741577,
2982
+ "learning_rate": 2.3501762632197414e-08,
2983
+ "loss": 0.0203,
2984
+ "step": 850
2985
+ },
2986
+ {
2987
+ "epoch": 0.9982425307557118,
2988
+ "grad_norm": 3.192451000213623,
2989
+ "learning_rate": 1.1750881316098707e-08,
2990
+ "loss": 0.0786,
2991
+ "step": 852
2992
+ }
2993
+ ],
2994
+ "logging_steps": 2,
2995
+ "max_steps": 853,
2996
+ "num_input_tokens_seen": 0,
2997
+ "num_train_epochs": 1,
2998
+ "save_steps": 20000,
2999
+ "stateful_callbacks": {
3000
+ "TrainerControl": {
3001
+ "args": {
3002
+ "should_epoch_stop": false,
3003
+ "should_evaluate": false,
3004
+ "should_log": false,
3005
+ "should_save": true,
3006
+ "should_training_stop": true
3007
+ },
3008
+ "attributes": {}
3009
+ }
3010
+ },
3011
+ "total_flos": 0.0,
3012
+ "train_batch_size": 1,
3013
+ "trial_name": null,
3014
+ "trial_params": null
3015
+ }
checkpoint-853/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:347d3fe43a026d8e0d3dc886116e49e992d313a7046135a66cde752c9308dfd5
3
+ size 6776
checkpoint-853/zero_to_fp32.py ADDED
@@ -0,0 +1,604 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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, exclude_frozen_parameters):
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
+ exclude_frozen_parameters)
216
+ elif zero_stage == 3:
217
+ return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
218
+ exclude_frozen_parameters)
219
+
220
+
221
+ def _zero2_merge_frozen_params(state_dict, zero_model_states):
222
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
223
+ return
224
+
225
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
226
+ frozen_param_fragments = zero_model_states[0].frozen_param_fragments
227
+
228
+ if debug:
229
+ num_elem = sum(s.numel() for s in frozen_param_shapes.values())
230
+ print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
231
+
232
+ wanted_params = len(frozen_param_shapes)
233
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
234
+ avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
235
+ print(f'Frozen params: Have {avail_numel} numels to process.')
236
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
237
+
238
+ total_params = 0
239
+ total_numel = 0
240
+ for name, shape in frozen_param_shapes.items():
241
+ total_params += 1
242
+ unpartitioned_numel = shape.numel()
243
+ total_numel += unpartitioned_numel
244
+
245
+ state_dict[name] = frozen_param_fragments[name]
246
+
247
+ if debug:
248
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
249
+
250
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
251
+
252
+
253
+ def _has_callable(obj, fn):
254
+ attr = getattr(obj, fn, None)
255
+ return callable(attr)
256
+
257
+
258
+ def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
259
+ param_shapes = zero_model_states[0].param_shapes
260
+
261
+ # Reconstruction protocol:
262
+ #
263
+ # XXX: document this
264
+
265
+ if debug:
266
+ for i in range(world_size):
267
+ for j in range(len(fp32_flat_groups[0])):
268
+ print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
269
+
270
+ # XXX: memory usage doubles here (zero2)
271
+ num_param_groups = len(fp32_flat_groups[0])
272
+ merged_single_partition_of_fp32_groups = []
273
+ for i in range(num_param_groups):
274
+ merged_partitions = [sd[i] for sd in fp32_flat_groups]
275
+ full_single_fp32_vector = torch.cat(merged_partitions, 0)
276
+ merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
277
+ avail_numel = sum(
278
+ [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
279
+
280
+ if debug:
281
+ wanted_params = sum([len(shapes) for shapes in param_shapes])
282
+ wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
283
+ # not asserting if there is a mismatch due to possible padding
284
+ print(f"Have {avail_numel} numels to process.")
285
+ print(f"Need {wanted_numel} numels in {wanted_params} params.")
286
+
287
+ # params
288
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
289
+ # out-of-core computing solution
290
+ total_numel = 0
291
+ total_params = 0
292
+ for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
293
+ offset = 0
294
+ avail_numel = full_single_fp32_vector.numel()
295
+ for name, shape in shapes.items():
296
+
297
+ unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape)
298
+ total_numel += unpartitioned_numel
299
+ total_params += 1
300
+
301
+ if debug:
302
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
303
+ state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
304
+ offset += unpartitioned_numel
305
+
306
+ # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
307
+ # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
308
+ # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
309
+ # live optimizer object, so we are checking that the numbers are within the right range
310
+ align_to = 2 * world_size
311
+
312
+ def zero2_align(x):
313
+ return align_to * math.ceil(x / align_to)
314
+
315
+ if debug:
316
+ print(f"original offset={offset}, avail_numel={avail_numel}")
317
+
318
+ offset = zero2_align(offset)
319
+ avail_numel = zero2_align(avail_numel)
320
+
321
+ if debug:
322
+ print(f"aligned offset={offset}, avail_numel={avail_numel}")
323
+
324
+ # Sanity check
325
+ if offset != avail_numel:
326
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
327
+
328
+ print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
329
+
330
+
331
+ def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
332
+ exclude_frozen_parameters):
333
+ state_dict = OrderedDict()
334
+
335
+ # buffers
336
+ buffers = zero_model_states[0].buffers
337
+ state_dict.update(buffers)
338
+ if debug:
339
+ print(f"added {len(buffers)} buffers")
340
+
341
+ if not exclude_frozen_parameters:
342
+ _zero2_merge_frozen_params(state_dict, zero_model_states)
343
+
344
+ _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
345
+
346
+ # recover shared parameters
347
+ for pair in zero_model_states[0].shared_params:
348
+ if pair[1] in state_dict:
349
+ state_dict[pair[0]] = state_dict[pair[1]]
350
+
351
+ return state_dict
352
+
353
+
354
+ def zero3_partitioned_param_info(unpartitioned_numel, world_size):
355
+ remainder = unpartitioned_numel % world_size
356
+ padding_numel = (world_size - remainder) if remainder else 0
357
+ partitioned_numel = math.ceil(unpartitioned_numel / world_size)
358
+ return partitioned_numel, padding_numel
359
+
360
+
361
+ def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
362
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
363
+ return
364
+
365
+ if debug:
366
+ for i in range(world_size):
367
+ num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
368
+ print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
369
+
370
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
371
+ wanted_params = len(frozen_param_shapes)
372
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
373
+ avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
374
+ print(f'Frozen params: Have {avail_numel} numels to process.')
375
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
376
+
377
+ total_params = 0
378
+ total_numel = 0
379
+ for name, shape in zero_model_states[0].frozen_param_shapes.items():
380
+ total_params += 1
381
+ unpartitioned_numel = shape.numel()
382
+ total_numel += unpartitioned_numel
383
+
384
+ param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
385
+ state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
386
+
387
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
388
+
389
+ if debug:
390
+ print(
391
+ f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
392
+ )
393
+
394
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
395
+
396
+
397
+ def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
398
+ param_shapes = zero_model_states[0].param_shapes
399
+ avail_numel = fp32_flat_groups[0].numel() * world_size
400
+ # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
401
+ # param, re-consolidating each param, while dealing with padding if any
402
+
403
+ # merge list of dicts, preserving order
404
+ param_shapes = {k: v for d in param_shapes for k, v in d.items()}
405
+
406
+ if debug:
407
+ for i in range(world_size):
408
+ print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
409
+
410
+ wanted_params = len(param_shapes)
411
+ wanted_numel = sum(shape.numel() for shape in param_shapes.values())
412
+ # not asserting if there is a mismatch due to possible padding
413
+ avail_numel = fp32_flat_groups[0].numel() * world_size
414
+ print(f"Trainable params: Have {avail_numel} numels to process.")
415
+ print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
416
+
417
+ # params
418
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
419
+ # out-of-core computing solution
420
+ offset = 0
421
+ total_numel = 0
422
+ total_params = 0
423
+ for name, shape in param_shapes.items():
424
+
425
+ unpartitioned_numel = shape.numel()
426
+ total_numel += unpartitioned_numel
427
+ total_params += 1
428
+
429
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
430
+
431
+ if debug:
432
+ print(
433
+ f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
434
+ )
435
+
436
+ # XXX: memory usage doubles here
437
+ state_dict[name] = torch.cat(
438
+ tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
439
+ 0).narrow(0, 0, unpartitioned_numel).view(shape)
440
+ offset += partitioned_numel
441
+
442
+ offset *= world_size
443
+
444
+ # Sanity check
445
+ if offset != avail_numel:
446
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
447
+
448
+ print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
449
+
450
+
451
+ def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
452
+ exclude_frozen_parameters):
453
+ state_dict = OrderedDict()
454
+
455
+ # buffers
456
+ buffers = zero_model_states[0].buffers
457
+ state_dict.update(buffers)
458
+ if debug:
459
+ print(f"added {len(buffers)} buffers")
460
+
461
+ if not exclude_frozen_parameters:
462
+ _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
463
+
464
+ _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
465
+
466
+ # recover shared parameters
467
+ for pair in zero_model_states[0].shared_params:
468
+ if pair[1] in state_dict:
469
+ state_dict[pair[0]] = state_dict[pair[1]]
470
+
471
+ return state_dict
472
+
473
+
474
+ def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None, exclude_frozen_parameters=False):
475
+ """
476
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
477
+ ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
478
+ via a model hub.
479
+
480
+ Args:
481
+ - ``checkpoint_dir``: path to the desired checkpoint folder
482
+ - ``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``
483
+ - ``exclude_frozen_parameters``: exclude frozen parameters
484
+
485
+ Returns:
486
+ - pytorch ``state_dict``
487
+
488
+ Note: this approach may not work if your application doesn't have sufficient free CPU memory and
489
+ you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
490
+ the checkpoint.
491
+
492
+ A typical usage might be ::
493
+
494
+ from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
495
+ # do the training and checkpoint saving
496
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
497
+ model = model.cpu() # move to cpu
498
+ model.load_state_dict(state_dict)
499
+ # submit to model hub or save the model to share with others
500
+
501
+ In this example the ``model`` will no longer be usable in the deepspeed context of the same
502
+ application. i.e. you will need to re-initialize the deepspeed engine, since
503
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
504
+
505
+ If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
506
+
507
+ """
508
+ if tag is None:
509
+ latest_path = os.path.join(checkpoint_dir, 'latest')
510
+ if os.path.isfile(latest_path):
511
+ with open(latest_path, 'r') as fd:
512
+ tag = fd.read().strip()
513
+ else:
514
+ raise ValueError(f"Unable to find 'latest' file at {latest_path}")
515
+
516
+ ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
517
+
518
+ if not os.path.isdir(ds_checkpoint_dir):
519
+ raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
520
+
521
+ return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters)
522
+
523
+
524
+ def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None, exclude_frozen_parameters=False):
525
+ """
526
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
527
+ loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
528
+
529
+ Args:
530
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
531
+ - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin)
532
+ - ``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``
533
+ - ``exclude_frozen_parameters``: exclude frozen parameters
534
+ """
535
+
536
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag, exclude_frozen_parameters)
537
+ print(f"Saving fp32 state dict to {output_file}")
538
+ torch.save(state_dict, output_file)
539
+
540
+
541
+ def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
542
+ """
543
+ 1. Put the provided model to cpu
544
+ 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
545
+ 3. Load it into the provided model
546
+
547
+ Args:
548
+ - ``model``: the model object to update
549
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
550
+ - ``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``
551
+
552
+ Returns:
553
+ - ``model`: modified model
554
+
555
+ Make sure you have plenty of CPU memory available before you call this function. If you don't
556
+ have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
557
+ conveniently placed for you in the checkpoint folder.
558
+
559
+ A typical usage might be ::
560
+
561
+ from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
562
+ model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
563
+ # submit to model hub or save the model to share with others
564
+
565
+ Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
566
+ of the same application. i.e. you will need to re-initialize the deepspeed engine, since
567
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
568
+
569
+ """
570
+ logger.info(f"Extracting fp32 weights")
571
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
572
+
573
+ logger.info(f"Overwriting model with fp32 weights")
574
+ model = model.cpu()
575
+ model.load_state_dict(state_dict, strict=False)
576
+
577
+ return model
578
+
579
+
580
+ if __name__ == "__main__":
581
+
582
+ parser = argparse.ArgumentParser()
583
+ parser.add_argument("checkpoint_dir",
584
+ type=str,
585
+ help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
586
+ parser.add_argument(
587
+ "output_file",
588
+ type=str,
589
+ help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)")
590
+ parser.add_argument("-t",
591
+ "--tag",
592
+ type=str,
593
+ default=None,
594
+ help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
595
+ parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters")
596
+ parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
597
+ args = parser.parse_args()
598
+
599
+ debug = args.debug
600
+
601
+ convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir,
602
+ args.output_file,
603
+ tag=args.tag,
604
+ exclude_frozen_parameters=args.exclude_frozen_parameters)
config.json ADDED
@@ -0,0 +1,28 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "/root/autodl-tmp/bge-m3_r4",
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": "4.42.1",
25
+ "type_vocab_size": 1,
26
+ "use_cache": true,
27
+ "vocab_size": 250002
28
+ }
config_sentence_transformers.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "__version__": {
3
+ "sentence_transformers": "3.0.1",
4
+ "transformers": "4.42.1",
5
+ "pytorch": "2.3.0+cu121"
6
+ },
7
+ "prompts": {},
8
+ "default_prompt_name": null,
9
+ "similarity_fn_name": null
10
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5ff2b3be09c7552fc58248f097a32771e376f56eb50737f93e0f41cef389d71d
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
+ ]
runs/Aug22_17-17-24_autodl-container-c024408f5d-9bcd732d/events.out.tfevents.1724318254.autodl-container-c024408f5d-9bcd732d.5345.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:04011eacde85360baeb929a4b75d2544f7b0d570ca1f0fed03e9b6168747a9cb
3
+ size 5560
runs/Aug22_17-18-40_autodl-container-c024408f5d-9bcd732d/events.out.tfevents.1724318333.autodl-container-c024408f5d-9bcd732d.6318.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:14fbf8c125536ee0ff2735de382a9ec6685c0ddf465533d2cb0127204a6e3f67
3
+ size 95134
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
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:b74659c780d49afad7a7b9799868f75cbd3014fb6c34956e85a793028d38094a
3
+ size 17098251
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
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:347d3fe43a026d8e0d3dc886116e49e992d313a7046135a66cde752c9308dfd5
3
+ size 6776