Upload folder using huggingface_hub
Browse filesThis view is limited to 50 files because it contains too many changes.
See raw diff
- .gitattributes +4 -0
- checkpoint-1506/1_Pooling/config.json +10 -0
- checkpoint-1506/README.md +57 -0
- checkpoint-1506/colbert_linear.pt +3 -0
- checkpoint-1506/config.json +28 -0
- checkpoint-1506/config_sentence_transformers.json +9 -0
- checkpoint-1506/global_step1506/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt +3 -0
- checkpoint-1506/global_step1506/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt +3 -0
- checkpoint-1506/global_step1506/mp_rank_00_model_states.pt +3 -0
- checkpoint-1506/latest +1 -0
- checkpoint-1506/model.safetensors +3 -0
- checkpoint-1506/modules.json +20 -0
- checkpoint-1506/rng_state_0.pth +3 -0
- checkpoint-1506/rng_state_1.pth +3 -0
- checkpoint-1506/sentence_bert_config.json +4 -0
- checkpoint-1506/sentencepiece.bpe.model +3 -0
- checkpoint-1506/sparse_linear.pt +3 -0
- checkpoint-1506/special_tokens_map.json +51 -0
- checkpoint-1506/tokenizer.json +3 -0
- checkpoint-1506/tokenizer_config.json +55 -0
- checkpoint-1506/trainer_state.json +1071 -0
- checkpoint-1506/training_args.bin +3 -0
- checkpoint-1506/zero_to_fp32.py +587 -0
- checkpoint-3012/1_Pooling/config.json +10 -0
- checkpoint-3012/README.md +57 -0
- checkpoint-3012/colbert_linear.pt +3 -0
- checkpoint-3012/config.json +28 -0
- checkpoint-3012/config_sentence_transformers.json +9 -0
- checkpoint-3012/global_step3012/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt +3 -0
- checkpoint-3012/global_step3012/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt +3 -0
- checkpoint-3012/global_step3012/mp_rank_00_model_states.pt +3 -0
- checkpoint-3012/latest +1 -0
- checkpoint-3012/model.safetensors +3 -0
- checkpoint-3012/modules.json +20 -0
- checkpoint-3012/rng_state_0.pth +3 -0
- checkpoint-3012/rng_state_1.pth +3 -0
- checkpoint-3012/sentence_bert_config.json +4 -0
- checkpoint-3012/sentencepiece.bpe.model +3 -0
- checkpoint-3012/sparse_linear.pt +3 -0
- checkpoint-3012/special_tokens_map.json +51 -0
- checkpoint-3012/tokenizer.json +3 -0
- checkpoint-3012/tokenizer_config.json +55 -0
- checkpoint-3012/trainer_state.json +2128 -0
- checkpoint-3012/training_args.bin +3 -0
- checkpoint-3012/zero_to_fp32.py +587 -0
- checkpoint-4518/1_Pooling/config.json +10 -0
- checkpoint-4518/README.md +57 -0
- checkpoint-4518/colbert_linear.pt +3 -0
- checkpoint-4518/config.json +28 -0
- checkpoint-4518/config_sentence_transformers.json +9 -0
.gitattributes
CHANGED
@@ -33,3 +33,7 @@ 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-1506/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
37 |
+
checkpoint-3012/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
38 |
+
checkpoint-4518/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
39 |
+
checkpoint-6024/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
checkpoint-1506/1_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-1506/README.md
ADDED
@@ -0,0 +1,57 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
library_name: sentence-transformers
|
3 |
+
pipeline_tag: sentence-similarity
|
4 |
+
tags:
|
5 |
+
- sentence-transformers
|
6 |
+
- feature-extraction
|
7 |
+
- sentence-similarity
|
8 |
+
|
9 |
+
---
|
10 |
+
|
11 |
+
# {MODEL_NAME}
|
12 |
+
|
13 |
+
This is a [sentence-transformers](https://www.SBERT.net) model: It maps sentences & paragraphs to a 1024 dimensional dense vector space and can be used for tasks like clustering or semantic search.
|
14 |
+
|
15 |
+
<!--- Describe your model here -->
|
16 |
+
|
17 |
+
## Usage (Sentence-Transformers)
|
18 |
+
|
19 |
+
Using this model becomes easy when you have [sentence-transformers](https://www.SBERT.net) installed:
|
20 |
+
|
21 |
+
```
|
22 |
+
pip install -U sentence-transformers
|
23 |
+
```
|
24 |
+
|
25 |
+
Then you can use the model like this:
|
26 |
+
|
27 |
+
```python
|
28 |
+
from sentence_transformers import SentenceTransformer
|
29 |
+
sentences = ["This is an example sentence", "Each sentence is converted"]
|
30 |
+
|
31 |
+
model = SentenceTransformer('{MODEL_NAME}')
|
32 |
+
embeddings = model.encode(sentences)
|
33 |
+
print(embeddings)
|
34 |
+
```
|
35 |
+
|
36 |
+
|
37 |
+
|
38 |
+
## Evaluation Results
|
39 |
+
|
40 |
+
<!--- Describe how your model was evaluated -->
|
41 |
+
|
42 |
+
For an automated evaluation of this model, see the *Sentence Embeddings Benchmark*: [https://seb.sbert.net](https://seb.sbert.net?model_name={MODEL_NAME})
|
43 |
+
|
44 |
+
|
45 |
+
|
46 |
+
## Full Model Architecture
|
47 |
+
```
|
48 |
+
SentenceTransformer(
|
49 |
+
(0): Transformer({'max_seq_length': 8192, 'do_lower_case': False}) with Transformer model: XLMRobertaModel
|
50 |
+
(1): Pooling({'word_embedding_dimension': 1024, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})
|
51 |
+
(2): Normalize()
|
52 |
+
)
|
53 |
+
```
|
54 |
+
|
55 |
+
## Citing & Authors
|
56 |
+
|
57 |
+
<!--- Describe where people can find more information -->
|
checkpoint-1506/colbert_linear.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:19d2655409ead458a148bb728c124e4411e6b8e54701c3fcb44f3b1e3fc09703
|
3 |
+
size 2100227
|
checkpoint-1506/config.json
ADDED
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "saved_models/bgem3_unified_finetune_20240330/checkpoint-1506",
|
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": "4.39.1",
|
25 |
+
"type_vocab_size": 1,
|
26 |
+
"use_cache": true,
|
27 |
+
"vocab_size": 250002
|
28 |
+
}
|
checkpoint-1506/config_sentence_transformers.json
ADDED
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"__version__": {
|
3 |
+
"sentence_transformers": "2.6.0",
|
4 |
+
"transformers": "4.39.1",
|
5 |
+
"pytorch": "2.0.1+cu117"
|
6 |
+
},
|
7 |
+
"prompts": {},
|
8 |
+
"default_prompt_name": null
|
9 |
+
}
|
checkpoint-1506/global_step1506/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 sha256:64c1cc5993e46cc6f8f416f3bbee15229eb758d605d89c4705e9ffdb716dd51f
|
3 |
+
size 3412837975
|
checkpoint-1506/global_step1506/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:b8da5a1160c4762af2bf151a32acbe0c23503dc975f79b71281879829651ccaf
|
3 |
+
size 3412867031
|
checkpoint-1506/global_step1506/mp_rank_00_model_states.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:00d6dc99de1451e07b73edb6abc2505516017db237dba5408a64952f44fb1e3a
|
3 |
+
size 1137729563
|
checkpoint-1506/latest
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
global_step1506
|
checkpoint-1506/model.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:57c93df581f333085a70042b9fe06a278de8dfa1fd1577c743561365cd9af130
|
3 |
+
size 2271064456
|
checkpoint-1506/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-1506/rng_state_0.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6fa54f27d4dcee0fd1801d909c8335e58360dc3e06b24da249b72492669278d9
|
3 |
+
size 15607
|
checkpoint-1506/rng_state_1.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a96d6edb046b989f646f8c2e335c4d6d773d850f827e65e27af34b97c6cb96ce
|
3 |
+
size 15607
|
checkpoint-1506/sentence_bert_config.json
ADDED
@@ -0,0 +1,4 @@
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"max_seq_length": 8192,
|
3 |
+
"do_lower_case": false
|
4 |
+
}
|
checkpoint-1506/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-1506/sparse_linear.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1162364150f8f40f9204f24d0a6bb3f5772a948cd37e14aa4d7c4aaad8175eaa
|
3 |
+
size 3071
|
checkpoint-1506/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-1506/tokenizer.json
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:69564b696052886ed0ac63fa393e928384e0f8caada38c1f4864a9bfbf379c15
|
3 |
+
size 17098273
|
checkpoint-1506/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-1506/trainer_state.json
ADDED
@@ -0,0 +1,1071 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": null,
|
3 |
+
"best_model_checkpoint": null,
|
4 |
+
"epoch": 1.0,
|
5 |
+
"eval_steps": 500,
|
6 |
+
"global_step": 1506,
|
7 |
+
"is_hyper_param_search": false,
|
8 |
+
"is_local_process_zero": true,
|
9 |
+
"is_world_process_zero": true,
|
10 |
+
"log_history": [
|
11 |
+
{
|
12 |
+
"epoch": 0.01,
|
13 |
+
"grad_norm": 41.35739895402257,
|
14 |
+
"learning_rate": 7.193423539345941e-06,
|
15 |
+
"loss": 0.5141,
|
16 |
+
"step": 10
|
17 |
+
},
|
18 |
+
{
|
19 |
+
"epoch": 0.01,
|
20 |
+
"grad_norm": 10.443694874625423,
|
21 |
+
"learning_rate": 9.358859796204429e-06,
|
22 |
+
"loss": 0.4195,
|
23 |
+
"step": 20
|
24 |
+
},
|
25 |
+
{
|
26 |
+
"epoch": 0.02,
|
27 |
+
"grad_norm": 5.720021591145773,
|
28 |
+
"learning_rate": 1.0625558804168632e-05,
|
29 |
+
"loss": 0.2851,
|
30 |
+
"step": 30
|
31 |
+
},
|
32 |
+
{
|
33 |
+
"epoch": 0.03,
|
34 |
+
"grad_norm": 7.59250257002781,
|
35 |
+
"learning_rate": 1.1524296053062918e-05,
|
36 |
+
"loss": 0.2174,
|
37 |
+
"step": 40
|
38 |
+
},
|
39 |
+
{
|
40 |
+
"epoch": 0.03,
|
41 |
+
"grad_norm": 11.521550428114184,
|
42 |
+
"learning_rate": 1.2221410821833392e-05,
|
43 |
+
"loss": 0.1817,
|
44 |
+
"step": 50
|
45 |
+
},
|
46 |
+
{
|
47 |
+
"epoch": 0.04,
|
48 |
+
"grad_norm": 6.027998448017936,
|
49 |
+
"learning_rate": 1.2790995061027121e-05,
|
50 |
+
"loss": 0.1886,
|
51 |
+
"step": 60
|
52 |
+
},
|
53 |
+
{
|
54 |
+
"epoch": 0.05,
|
55 |
+
"grad_norm": 4.929816146115093,
|
56 |
+
"learning_rate": 1.3272571673439616e-05,
|
57 |
+
"loss": 0.1553,
|
58 |
+
"step": 70
|
59 |
+
},
|
60 |
+
{
|
61 |
+
"epoch": 0.05,
|
62 |
+
"grad_norm": 2.3881515971871345,
|
63 |
+
"learning_rate": 1.3689732309921406e-05,
|
64 |
+
"loss": 0.129,
|
65 |
+
"step": 80
|
66 |
+
},
|
67 |
+
{
|
68 |
+
"epoch": 0.06,
|
69 |
+
"grad_norm": 5.441118636680569,
|
70 |
+
"learning_rate": 1.4057694068991321e-05,
|
71 |
+
"loss": 0.1433,
|
72 |
+
"step": 90
|
73 |
+
},
|
74 |
+
{
|
75 |
+
"epoch": 0.07,
|
76 |
+
"grad_norm": 6.085213496930297,
|
77 |
+
"learning_rate": 1.4386847078691883e-05,
|
78 |
+
"loss": 0.1092,
|
79 |
+
"step": 100
|
80 |
+
},
|
81 |
+
{
|
82 |
+
"epoch": 0.07,
|
83 |
+
"grad_norm": 4.713361075579101,
|
84 |
+
"learning_rate": 1.4684602194465794e-05,
|
85 |
+
"loss": 0.1231,
|
86 |
+
"step": 110
|
87 |
+
},
|
88 |
+
{
|
89 |
+
"epoch": 0.08,
|
90 |
+
"grad_norm": 3.8259500358417924,
|
91 |
+
"learning_rate": 1.495643131788561e-05,
|
92 |
+
"loss": 0.0697,
|
93 |
+
"step": 120
|
94 |
+
},
|
95 |
+
{
|
96 |
+
"epoch": 0.09,
|
97 |
+
"grad_norm": 2.67920682727699,
|
98 |
+
"learning_rate": 1.5206489871327869e-05,
|
99 |
+
"loss": 0.084,
|
100 |
+
"step": 130
|
101 |
+
},
|
102 |
+
{
|
103 |
+
"epoch": 0.09,
|
104 |
+
"grad_norm": 8.189491553913532,
|
105 |
+
"learning_rate": 1.54380079302981e-05,
|
106 |
+
"loss": 0.1023,
|
107 |
+
"step": 140
|
108 |
+
},
|
109 |
+
{
|
110 |
+
"epoch": 0.1,
|
111 |
+
"grad_norm": 12.000369384166694,
|
112 |
+
"learning_rate": 1.5653546086656083e-05,
|
113 |
+
"loss": 0.0972,
|
114 |
+
"step": 150
|
115 |
+
},
|
116 |
+
{
|
117 |
+
"epoch": 0.11,
|
118 |
+
"grad_norm": 2.3726942012239953,
|
119 |
+
"learning_rate": 1.5855168566779895e-05,
|
120 |
+
"loss": 0.1036,
|
121 |
+
"step": 160
|
122 |
+
},
|
123 |
+
{
|
124 |
+
"epoch": 0.11,
|
125 |
+
"grad_norm": 2.0242117206977044,
|
126 |
+
"learning_rate": 1.604456377435124e-05,
|
127 |
+
"loss": 0.1081,
|
128 |
+
"step": 170
|
129 |
+
},
|
130 |
+
{
|
131 |
+
"epoch": 0.12,
|
132 |
+
"grad_norm": 2.2261934744219967,
|
133 |
+
"learning_rate": 1.6223130325849813e-05,
|
134 |
+
"loss": 0.082,
|
135 |
+
"step": 180
|
136 |
+
},
|
137 |
+
{
|
138 |
+
"epoch": 0.13,
|
139 |
+
"grad_norm": 4.70421073268978,
|
140 |
+
"learning_rate": 1.6392039793463407e-05,
|
141 |
+
"loss": 0.0744,
|
142 |
+
"step": 190
|
143 |
+
},
|
144 |
+
{
|
145 |
+
"epoch": 0.13,
|
146 |
+
"grad_norm": 12.064244277626111,
|
147 |
+
"learning_rate": 1.6552283335550368e-05,
|
148 |
+
"loss": 0.0934,
|
149 |
+
"step": 200
|
150 |
+
},
|
151 |
+
{
|
152 |
+
"epoch": 0.14,
|
153 |
+
"grad_norm": 2.150990470911233,
|
154 |
+
"learning_rate": 1.67047069382623e-05,
|
155 |
+
"loss": 0.0737,
|
156 |
+
"step": 210
|
157 |
+
},
|
158 |
+
{
|
159 |
+
"epoch": 0.15,
|
160 |
+
"grad_norm": 4.074258308547766,
|
161 |
+
"learning_rate": 1.6850038451324284e-05,
|
162 |
+
"loss": 0.0841,
|
163 |
+
"step": 220
|
164 |
+
},
|
165 |
+
{
|
166 |
+
"epoch": 0.15,
|
167 |
+
"grad_norm": 2.6764310841274237,
|
168 |
+
"learning_rate": 1.6988908609137504e-05,
|
169 |
+
"loss": 0.0821,
|
170 |
+
"step": 230
|
171 |
+
},
|
172 |
+
{
|
173 |
+
"epoch": 0.16,
|
174 |
+
"grad_norm": 3.1391283710987063,
|
175 |
+
"learning_rate": 1.71218675747441e-05,
|
176 |
+
"loss": 0.0747,
|
177 |
+
"step": 240
|
178 |
+
},
|
179 |
+
{
|
180 |
+
"epoch": 0.17,
|
181 |
+
"grad_norm": 3.1103728720325754,
|
182 |
+
"learning_rate": 1.7249398104320845e-05,
|
183 |
+
"loss": 0.0907,
|
184 |
+
"step": 250
|
185 |
+
},
|
186 |
+
{
|
187 |
+
"epoch": 0.17,
|
188 |
+
"grad_norm": 5.354173276825422,
|
189 |
+
"learning_rate": 1.7371926128186358e-05,
|
190 |
+
"loss": 0.0765,
|
191 |
+
"step": 260
|
192 |
+
},
|
193 |
+
{
|
194 |
+
"epoch": 0.18,
|
195 |
+
"grad_norm": 2.8677808267225005,
|
196 |
+
"learning_rate": 1.7489829333814013e-05,
|
197 |
+
"loss": 0.0777,
|
198 |
+
"step": 270
|
199 |
+
},
|
200 |
+
{
|
201 |
+
"epoch": 0.19,
|
202 |
+
"grad_norm": 2.6124301773304874,
|
203 |
+
"learning_rate": 1.760344418715659e-05,
|
204 |
+
"loss": 0.0897,
|
205 |
+
"step": 280
|
206 |
+
},
|
207 |
+
{
|
208 |
+
"epoch": 0.19,
|
209 |
+
"grad_norm": 2.915238130503912,
|
210 |
+
"learning_rate": 1.7713071721324668e-05,
|
211 |
+
"loss": 0.0866,
|
212 |
+
"step": 290
|
213 |
+
},
|
214 |
+
{
|
215 |
+
"epoch": 0.2,
|
216 |
+
"grad_norm": 4.061775729833908,
|
217 |
+
"learning_rate": 1.781898234351457e-05,
|
218 |
+
"loss": 0.0804,
|
219 |
+
"step": 300
|
220 |
+
},
|
221 |
+
{
|
222 |
+
"epoch": 0.21,
|
223 |
+
"grad_norm": 1.5648704857906242,
|
224 |
+
"learning_rate": 1.7921419853452233e-05,
|
225 |
+
"loss": 0.0834,
|
226 |
+
"step": 310
|
227 |
+
},
|
228 |
+
{
|
229 |
+
"epoch": 0.21,
|
230 |
+
"grad_norm": 2.225837991191409,
|
231 |
+
"learning_rate": 1.8020604823638384e-05,
|
232 |
+
"loss": 0.0646,
|
233 |
+
"step": 320
|
234 |
+
},
|
235 |
+
{
|
236 |
+
"epoch": 0.22,
|
237 |
+
"grad_norm": 4.041622096074851,
|
238 |
+
"learning_rate": 1.811673745928848e-05,
|
239 |
+
"loss": 0.0636,
|
240 |
+
"step": 330
|
241 |
+
},
|
242 |
+
{
|
243 |
+
"epoch": 0.23,
|
244 |
+
"grad_norm": 0.21100008000919218,
|
245 |
+
"learning_rate": 1.821000003120973e-05,
|
246 |
+
"loss": 0.0656,
|
247 |
+
"step": 340
|
248 |
+
},
|
249 |
+
{
|
250 |
+
"epoch": 0.23,
|
251 |
+
"grad_norm": 2.2923304434569443,
|
252 |
+
"learning_rate": 1.8300558955927067e-05,
|
253 |
+
"loss": 0.0756,
|
254 |
+
"step": 350
|
255 |
+
},
|
256 |
+
{
|
257 |
+
"epoch": 0.24,
|
258 |
+
"grad_norm": 3.6524729986322013,
|
259 |
+
"learning_rate": 1.83885665827083e-05,
|
260 |
+
"loss": 0.0466,
|
261 |
+
"step": 360
|
262 |
+
},
|
263 |
+
{
|
264 |
+
"epoch": 0.25,
|
265 |
+
"grad_norm": 2.1793660025907284,
|
266 |
+
"learning_rate": 1.847416273569235e-05,
|
267 |
+
"loss": 0.0704,
|
268 |
+
"step": 370
|
269 |
+
},
|
270 |
+
{
|
271 |
+
"epoch": 0.25,
|
272 |
+
"grad_norm": 4.363503091034068,
|
273 |
+
"learning_rate": 1.8557476050321896e-05,
|
274 |
+
"loss": 0.0712,
|
275 |
+
"step": 380
|
276 |
+
},
|
277 |
+
{
|
278 |
+
"epoch": 0.26,
|
279 |
+
"grad_norm": 3.0942807753734476,
|
280 |
+
"learning_rate": 1.863862513615056e-05,
|
281 |
+
"loss": 0.0831,
|
282 |
+
"step": 390
|
283 |
+
},
|
284 |
+
{
|
285 |
+
"epoch": 0.27,
|
286 |
+
"grad_norm": 1.789230871814937,
|
287 |
+
"learning_rate": 1.8717719592408857e-05,
|
288 |
+
"loss": 0.078,
|
289 |
+
"step": 400
|
290 |
+
},
|
291 |
+
{
|
292 |
+
"epoch": 0.27,
|
293 |
+
"grad_norm": 3.4361028432853677,
|
294 |
+
"learning_rate": 1.879486089815082e-05,
|
295 |
+
"loss": 0.0663,
|
296 |
+
"step": 410
|
297 |
+
},
|
298 |
+
{
|
299 |
+
"epoch": 0.28,
|
300 |
+
"grad_norm": 3.690439378170989,
|
301 |
+
"learning_rate": 1.8870143195120794e-05,
|
302 |
+
"loss": 0.0738,
|
303 |
+
"step": 420
|
304 |
+
},
|
305 |
+
{
|
306 |
+
"epoch": 0.29,
|
307 |
+
"grad_norm": 2.159404934400987,
|
308 |
+
"learning_rate": 1.8943653978491198e-05,
|
309 |
+
"loss": 0.0768,
|
310 |
+
"step": 430
|
311 |
+
},
|
312 |
+
{
|
313 |
+
"epoch": 0.29,
|
314 |
+
"grad_norm": 0.8042070009727055,
|
315 |
+
"learning_rate": 1.901547470818277e-05,
|
316 |
+
"loss": 0.0777,
|
317 |
+
"step": 440
|
318 |
+
},
|
319 |
+
{
|
320 |
+
"epoch": 0.3,
|
321 |
+
"grad_norm": 1.70758347995601,
|
322 |
+
"learning_rate": 1.9085681351478775e-05,
|
323 |
+
"loss": 0.05,
|
324 |
+
"step": 450
|
325 |
+
},
|
326 |
+
{
|
327 |
+
"epoch": 0.31,
|
328 |
+
"grad_norm": 1.2113560599237037,
|
329 |
+
"learning_rate": 1.9154344865995993e-05,
|
330 |
+
"loss": 0.0715,
|
331 |
+
"step": 460
|
332 |
+
},
|
333 |
+
{
|
334 |
+
"epoch": 0.31,
|
335 |
+
"grad_norm": 2.5816071085486527,
|
336 |
+
"learning_rate": 1.9221531630710657e-05,
|
337 |
+
"loss": 0.0688,
|
338 |
+
"step": 470
|
339 |
+
},
|
340 |
+
{
|
341 |
+
"epoch": 0.32,
|
342 |
+
"grad_norm": 1.4447841193953872,
|
343 |
+
"learning_rate": 1.9287303831602588e-05,
|
344 |
+
"loss": 0.0659,
|
345 |
+
"step": 480
|
346 |
+
},
|
347 |
+
{
|
348 |
+
"epoch": 0.33,
|
349 |
+
"grad_norm": 1.2668557011358101,
|
350 |
+
"learning_rate": 1.9351719807533285e-05,
|
351 |
+
"loss": 0.0515,
|
352 |
+
"step": 490
|
353 |
+
},
|
354 |
+
{
|
355 |
+
"epoch": 0.33,
|
356 |
+
"grad_norm": 2.0353893651728865,
|
357 |
+
"learning_rate": 1.9414834361179333e-05,
|
358 |
+
"loss": 0.0687,
|
359 |
+
"step": 500
|
360 |
+
},
|
361 |
+
{
|
362 |
+
"epoch": 0.34,
|
363 |
+
"grad_norm": 2.005863891715338,
|
364 |
+
"learning_rate": 1.947669903917393e-05,
|
365 |
+
"loss": 0.0459,
|
366 |
+
"step": 510
|
367 |
+
},
|
368 |
+
{
|
369 |
+
"epoch": 0.35,
|
370 |
+
"grad_norm": 1.4330303529877615,
|
371 |
+
"learning_rate": 1.9537362385044847e-05,
|
372 |
+
"loss": 0.0557,
|
373 |
+
"step": 520
|
374 |
+
},
|
375 |
+
{
|
376 |
+
"epoch": 0.35,
|
377 |
+
"grad_norm": 1.8683959370360874,
|
378 |
+
"learning_rate": 1.959687016805845e-05,
|
379 |
+
"loss": 0.0656,
|
380 |
+
"step": 530
|
381 |
+
},
|
382 |
+
{
|
383 |
+
"epoch": 0.36,
|
384 |
+
"grad_norm": 1.5963373312962135,
|
385 |
+
"learning_rate": 1.9655265590672502e-05,
|
386 |
+
"loss": 0.0519,
|
387 |
+
"step": 540
|
388 |
+
},
|
389 |
+
{
|
390 |
+
"epoch": 0.37,
|
391 |
+
"grad_norm": 1.824110595633705,
|
392 |
+
"learning_rate": 1.9712589476953243e-05,
|
393 |
+
"loss": 0.0557,
|
394 |
+
"step": 550
|
395 |
+
},
|
396 |
+
{
|
397 |
+
"epoch": 0.37,
|
398 |
+
"grad_norm": 2.024299349665376,
|
399 |
+
"learning_rate": 1.976888044401508e-05,
|
400 |
+
"loss": 0.0617,
|
401 |
+
"step": 560
|
402 |
+
},
|
403 |
+
{
|
404 |
+
"epoch": 0.38,
|
405 |
+
"grad_norm": 1.0490238284158413,
|
406 |
+
"learning_rate": 1.98241750582861e-05,
|
407 |
+
"loss": 0.0529,
|
408 |
+
"step": 570
|
409 |
+
},
|
410 |
+
{
|
411 |
+
"epoch": 0.39,
|
412 |
+
"grad_norm": 1.2602675746455845,
|
413 |
+
"learning_rate": 1.9878507978183157e-05,
|
414 |
+
"loss": 0.0617,
|
415 |
+
"step": 580
|
416 |
+
},
|
417 |
+
{
|
418 |
+
"epoch": 0.39,
|
419 |
+
"grad_norm": 1.7353304044327778,
|
420 |
+
"learning_rate": 1.9931912084590654e-05,
|
421 |
+
"loss": 0.0464,
|
422 |
+
"step": 590
|
423 |
+
},
|
424 |
+
{
|
425 |
+
"epoch": 0.4,
|
426 |
+
"grad_norm": 2.043608410331547,
|
427 |
+
"learning_rate": 1.998441860037306e-05,
|
428 |
+
"loss": 0.0716,
|
429 |
+
"step": 600
|
430 |
+
},
|
431 |
+
{
|
432 |
+
"epoch": 0.41,
|
433 |
+
"grad_norm": 0.9522735357022469,
|
434 |
+
"learning_rate": 1.997786386275595e-05,
|
435 |
+
"loss": 0.0496,
|
436 |
+
"step": 610
|
437 |
+
},
|
438 |
+
{
|
439 |
+
"epoch": 0.41,
|
440 |
+
"grad_norm": 1.3740017908881452,
|
441 |
+
"learning_rate": 1.9940970300682533e-05,
|
442 |
+
"loss": 0.0686,
|
443 |
+
"step": 620
|
444 |
+
},
|
445 |
+
{
|
446 |
+
"epoch": 0.42,
|
447 |
+
"grad_norm": 1.1533525092583308,
|
448 |
+
"learning_rate": 1.9904076738609114e-05,
|
449 |
+
"loss": 0.0562,
|
450 |
+
"step": 630
|
451 |
+
},
|
452 |
+
{
|
453 |
+
"epoch": 0.42,
|
454 |
+
"grad_norm": 1.564671647601224,
|
455 |
+
"learning_rate": 1.9867183176535695e-05,
|
456 |
+
"loss": 0.0553,
|
457 |
+
"step": 640
|
458 |
+
},
|
459 |
+
{
|
460 |
+
"epoch": 0.43,
|
461 |
+
"grad_norm": 0.8996693997352794,
|
462 |
+
"learning_rate": 1.9830289614462276e-05,
|
463 |
+
"loss": 0.0524,
|
464 |
+
"step": 650
|
465 |
+
},
|
466 |
+
{
|
467 |
+
"epoch": 0.44,
|
468 |
+
"grad_norm": 1.9340857910489522,
|
469 |
+
"learning_rate": 1.979339605238886e-05,
|
470 |
+
"loss": 0.0534,
|
471 |
+
"step": 660
|
472 |
+
},
|
473 |
+
{
|
474 |
+
"epoch": 0.44,
|
475 |
+
"grad_norm": 1.892682744806536,
|
476 |
+
"learning_rate": 1.9756502490315442e-05,
|
477 |
+
"loss": 0.0625,
|
478 |
+
"step": 670
|
479 |
+
},
|
480 |
+
{
|
481 |
+
"epoch": 0.45,
|
482 |
+
"grad_norm": 1.2063853192894494,
|
483 |
+
"learning_rate": 1.9719608928242023e-05,
|
484 |
+
"loss": 0.0519,
|
485 |
+
"step": 680
|
486 |
+
},
|
487 |
+
{
|
488 |
+
"epoch": 0.46,
|
489 |
+
"grad_norm": 1.984312822464147,
|
490 |
+
"learning_rate": 1.9682715366168604e-05,
|
491 |
+
"loss": 0.0494,
|
492 |
+
"step": 690
|
493 |
+
},
|
494 |
+
{
|
495 |
+
"epoch": 0.46,
|
496 |
+
"grad_norm": 1.6705446030280595,
|
497 |
+
"learning_rate": 1.9645821804095185e-05,
|
498 |
+
"loss": 0.059,
|
499 |
+
"step": 700
|
500 |
+
},
|
501 |
+
{
|
502 |
+
"epoch": 0.47,
|
503 |
+
"grad_norm": 1.7422620817223426,
|
504 |
+
"learning_rate": 1.960892824202177e-05,
|
505 |
+
"loss": 0.0454,
|
506 |
+
"step": 710
|
507 |
+
},
|
508 |
+
{
|
509 |
+
"epoch": 0.48,
|
510 |
+
"grad_norm": 1.3182961229762868,
|
511 |
+
"learning_rate": 1.957203467994835e-05,
|
512 |
+
"loss": 0.0581,
|
513 |
+
"step": 720
|
514 |
+
},
|
515 |
+
{
|
516 |
+
"epoch": 0.48,
|
517 |
+
"grad_norm": 1.330360819481426,
|
518 |
+
"learning_rate": 1.9535141117874932e-05,
|
519 |
+
"loss": 0.0575,
|
520 |
+
"step": 730
|
521 |
+
},
|
522 |
+
{
|
523 |
+
"epoch": 0.49,
|
524 |
+
"grad_norm": 1.9840741868184866,
|
525 |
+
"learning_rate": 1.9498247555801517e-05,
|
526 |
+
"loss": 0.063,
|
527 |
+
"step": 740
|
528 |
+
},
|
529 |
+
{
|
530 |
+
"epoch": 0.5,
|
531 |
+
"grad_norm": 1.6018064731760029,
|
532 |
+
"learning_rate": 1.9461353993728094e-05,
|
533 |
+
"loss": 0.0519,
|
534 |
+
"step": 750
|
535 |
+
},
|
536 |
+
{
|
537 |
+
"epoch": 0.5,
|
538 |
+
"grad_norm": 1.391561342963203,
|
539 |
+
"learning_rate": 1.9424460431654675e-05,
|
540 |
+
"loss": 0.0572,
|
541 |
+
"step": 760
|
542 |
+
},
|
543 |
+
{
|
544 |
+
"epoch": 0.51,
|
545 |
+
"grad_norm": 1.0319629863193043,
|
546 |
+
"learning_rate": 1.938756686958126e-05,
|
547 |
+
"loss": 0.0619,
|
548 |
+
"step": 770
|
549 |
+
},
|
550 |
+
{
|
551 |
+
"epoch": 0.52,
|
552 |
+
"grad_norm": 1.7398111190442345,
|
553 |
+
"learning_rate": 1.935067330750784e-05,
|
554 |
+
"loss": 0.05,
|
555 |
+
"step": 780
|
556 |
+
},
|
557 |
+
{
|
558 |
+
"epoch": 0.52,
|
559 |
+
"grad_norm": 1.675180749962208,
|
560 |
+
"learning_rate": 1.9313779745434422e-05,
|
561 |
+
"loss": 0.0654,
|
562 |
+
"step": 790
|
563 |
+
},
|
564 |
+
{
|
565 |
+
"epoch": 0.53,
|
566 |
+
"grad_norm": 1.486943722740635,
|
567 |
+
"learning_rate": 1.9276886183361007e-05,
|
568 |
+
"loss": 0.0463,
|
569 |
+
"step": 800
|
570 |
+
},
|
571 |
+
{
|
572 |
+
"epoch": 0.54,
|
573 |
+
"grad_norm": 1.1444156835752686,
|
574 |
+
"learning_rate": 1.9239992621287588e-05,
|
575 |
+
"loss": 0.0598,
|
576 |
+
"step": 810
|
577 |
+
},
|
578 |
+
{
|
579 |
+
"epoch": 0.54,
|
580 |
+
"grad_norm": 1.1260777714975718,
|
581 |
+
"learning_rate": 1.920309905921417e-05,
|
582 |
+
"loss": 0.039,
|
583 |
+
"step": 820
|
584 |
+
},
|
585 |
+
{
|
586 |
+
"epoch": 0.55,
|
587 |
+
"grad_norm": 1.328313971146618,
|
588 |
+
"learning_rate": 1.916620549714075e-05,
|
589 |
+
"loss": 0.0591,
|
590 |
+
"step": 830
|
591 |
+
},
|
592 |
+
{
|
593 |
+
"epoch": 0.56,
|
594 |
+
"grad_norm": 0.8805517771854091,
|
595 |
+
"learning_rate": 1.912931193506733e-05,
|
596 |
+
"loss": 0.0392,
|
597 |
+
"step": 840
|
598 |
+
},
|
599 |
+
{
|
600 |
+
"epoch": 0.56,
|
601 |
+
"grad_norm": 0.9704226543415952,
|
602 |
+
"learning_rate": 1.9092418372993916e-05,
|
603 |
+
"loss": 0.056,
|
604 |
+
"step": 850
|
605 |
+
},
|
606 |
+
{
|
607 |
+
"epoch": 0.57,
|
608 |
+
"grad_norm": 1.4328262810995938,
|
609 |
+
"learning_rate": 1.9055524810920497e-05,
|
610 |
+
"loss": 0.0602,
|
611 |
+
"step": 860
|
612 |
+
},
|
613 |
+
{
|
614 |
+
"epoch": 0.58,
|
615 |
+
"grad_norm": 2.6133255890647167,
|
616 |
+
"learning_rate": 1.9018631248847078e-05,
|
617 |
+
"loss": 0.0507,
|
618 |
+
"step": 870
|
619 |
+
},
|
620 |
+
{
|
621 |
+
"epoch": 0.58,
|
622 |
+
"grad_norm": 1.4574352859865667,
|
623 |
+
"learning_rate": 1.898173768677366e-05,
|
624 |
+
"loss": 0.0667,
|
625 |
+
"step": 880
|
626 |
+
},
|
627 |
+
{
|
628 |
+
"epoch": 0.59,
|
629 |
+
"grad_norm": 1.9116869451062006,
|
630 |
+
"learning_rate": 1.894484412470024e-05,
|
631 |
+
"loss": 0.0533,
|
632 |
+
"step": 890
|
633 |
+
},
|
634 |
+
{
|
635 |
+
"epoch": 0.6,
|
636 |
+
"grad_norm": 0.8152499611885836,
|
637 |
+
"learning_rate": 1.890795056262682e-05,
|
638 |
+
"loss": 0.0534,
|
639 |
+
"step": 900
|
640 |
+
},
|
641 |
+
{
|
642 |
+
"epoch": 0.6,
|
643 |
+
"grad_norm": 2.9440345318883696,
|
644 |
+
"learning_rate": 1.8871057000553406e-05,
|
645 |
+
"loss": 0.0419,
|
646 |
+
"step": 910
|
647 |
+
},
|
648 |
+
{
|
649 |
+
"epoch": 0.61,
|
650 |
+
"grad_norm": 2.7588896552457074,
|
651 |
+
"learning_rate": 1.8834163438479987e-05,
|
652 |
+
"loss": 0.0418,
|
653 |
+
"step": 920
|
654 |
+
},
|
655 |
+
{
|
656 |
+
"epoch": 0.62,
|
657 |
+
"grad_norm": 1.8886319208504987,
|
658 |
+
"learning_rate": 1.8797269876406568e-05,
|
659 |
+
"loss": 0.0397,
|
660 |
+
"step": 930
|
661 |
+
},
|
662 |
+
{
|
663 |
+
"epoch": 0.62,
|
664 |
+
"grad_norm": 1.8820615671771377,
|
665 |
+
"learning_rate": 1.876037631433315e-05,
|
666 |
+
"loss": 0.0454,
|
667 |
+
"step": 940
|
668 |
+
},
|
669 |
+
{
|
670 |
+
"epoch": 0.63,
|
671 |
+
"grad_norm": 1.2868214865586614,
|
672 |
+
"learning_rate": 1.872348275225973e-05,
|
673 |
+
"loss": 0.0373,
|
674 |
+
"step": 950
|
675 |
+
},
|
676 |
+
{
|
677 |
+
"epoch": 0.64,
|
678 |
+
"grad_norm": 2.016876778284122,
|
679 |
+
"learning_rate": 1.8686589190186315e-05,
|
680 |
+
"loss": 0.0619,
|
681 |
+
"step": 960
|
682 |
+
},
|
683 |
+
{
|
684 |
+
"epoch": 0.64,
|
685 |
+
"grad_norm": 0.8992764743601639,
|
686 |
+
"learning_rate": 1.8649695628112896e-05,
|
687 |
+
"loss": 0.0403,
|
688 |
+
"step": 970
|
689 |
+
},
|
690 |
+
{
|
691 |
+
"epoch": 0.65,
|
692 |
+
"grad_norm": 0.42416507318131186,
|
693 |
+
"learning_rate": 1.8612802066039477e-05,
|
694 |
+
"loss": 0.067,
|
695 |
+
"step": 980
|
696 |
+
},
|
697 |
+
{
|
698 |
+
"epoch": 0.66,
|
699 |
+
"grad_norm": 1.8601423331120188,
|
700 |
+
"learning_rate": 1.8575908503966062e-05,
|
701 |
+
"loss": 0.066,
|
702 |
+
"step": 990
|
703 |
+
},
|
704 |
+
{
|
705 |
+
"epoch": 0.66,
|
706 |
+
"grad_norm": 1.3893250379275068,
|
707 |
+
"learning_rate": 1.8539014941892643e-05,
|
708 |
+
"loss": 0.0437,
|
709 |
+
"step": 1000
|
710 |
+
},
|
711 |
+
{
|
712 |
+
"epoch": 0.67,
|
713 |
+
"grad_norm": 2.3240219930504376,
|
714 |
+
"learning_rate": 1.850212137981922e-05,
|
715 |
+
"loss": 0.0488,
|
716 |
+
"step": 1010
|
717 |
+
},
|
718 |
+
{
|
719 |
+
"epoch": 0.68,
|
720 |
+
"grad_norm": 0.8849896953543195,
|
721 |
+
"learning_rate": 1.8465227817745805e-05,
|
722 |
+
"loss": 0.0472,
|
723 |
+
"step": 1020
|
724 |
+
},
|
725 |
+
{
|
726 |
+
"epoch": 0.68,
|
727 |
+
"grad_norm": 0.8655212477880712,
|
728 |
+
"learning_rate": 1.8428334255672386e-05,
|
729 |
+
"loss": 0.0546,
|
730 |
+
"step": 1030
|
731 |
+
},
|
732 |
+
{
|
733 |
+
"epoch": 0.69,
|
734 |
+
"grad_norm": 1.2016665548799828,
|
735 |
+
"learning_rate": 1.8391440693598967e-05,
|
736 |
+
"loss": 0.0956,
|
737 |
+
"step": 1040
|
738 |
+
},
|
739 |
+
{
|
740 |
+
"epoch": 0.7,
|
741 |
+
"grad_norm": 1.492659569010543,
|
742 |
+
"learning_rate": 1.8354547131525552e-05,
|
743 |
+
"loss": 0.0449,
|
744 |
+
"step": 1050
|
745 |
+
},
|
746 |
+
{
|
747 |
+
"epoch": 0.7,
|
748 |
+
"grad_norm": 0.4224837643985645,
|
749 |
+
"learning_rate": 1.8317653569452133e-05,
|
750 |
+
"loss": 0.0545,
|
751 |
+
"step": 1060
|
752 |
+
},
|
753 |
+
{
|
754 |
+
"epoch": 0.71,
|
755 |
+
"grad_norm": 1.8405573730480247,
|
756 |
+
"learning_rate": 1.8280760007378714e-05,
|
757 |
+
"loss": 0.0328,
|
758 |
+
"step": 1070
|
759 |
+
},
|
760 |
+
{
|
761 |
+
"epoch": 0.72,
|
762 |
+
"grad_norm": 0.3988661237020132,
|
763 |
+
"learning_rate": 1.8243866445305295e-05,
|
764 |
+
"loss": 0.0451,
|
765 |
+
"step": 1080
|
766 |
+
},
|
767 |
+
{
|
768 |
+
"epoch": 0.72,
|
769 |
+
"grad_norm": 0.5573396543240564,
|
770 |
+
"learning_rate": 1.8206972883231876e-05,
|
771 |
+
"loss": 0.0502,
|
772 |
+
"step": 1090
|
773 |
+
},
|
774 |
+
{
|
775 |
+
"epoch": 0.73,
|
776 |
+
"grad_norm": 0.7855566721285819,
|
777 |
+
"learning_rate": 1.817007932115846e-05,
|
778 |
+
"loss": 0.0341,
|
779 |
+
"step": 1100
|
780 |
+
},
|
781 |
+
{
|
782 |
+
"epoch": 0.74,
|
783 |
+
"grad_norm": 1.497005175064917,
|
784 |
+
"learning_rate": 1.8133185759085042e-05,
|
785 |
+
"loss": 0.0513,
|
786 |
+
"step": 1110
|
787 |
+
},
|
788 |
+
{
|
789 |
+
"epoch": 0.74,
|
790 |
+
"grad_norm": 1.650975953086994,
|
791 |
+
"learning_rate": 1.8096292197011623e-05,
|
792 |
+
"loss": 0.0473,
|
793 |
+
"step": 1120
|
794 |
+
},
|
795 |
+
{
|
796 |
+
"epoch": 0.75,
|
797 |
+
"grad_norm": 0.8418054071334755,
|
798 |
+
"learning_rate": 1.8059398634938204e-05,
|
799 |
+
"loss": 0.0444,
|
800 |
+
"step": 1130
|
801 |
+
},
|
802 |
+
{
|
803 |
+
"epoch": 0.76,
|
804 |
+
"grad_norm": 1.33651471077315,
|
805 |
+
"learning_rate": 1.8022505072864785e-05,
|
806 |
+
"loss": 0.0476,
|
807 |
+
"step": 1140
|
808 |
+
},
|
809 |
+
{
|
810 |
+
"epoch": 0.76,
|
811 |
+
"grad_norm": 1.1529220423121023,
|
812 |
+
"learning_rate": 1.7985611510791367e-05,
|
813 |
+
"loss": 0.0379,
|
814 |
+
"step": 1150
|
815 |
+
},
|
816 |
+
{
|
817 |
+
"epoch": 0.77,
|
818 |
+
"grad_norm": 0.6912013035706749,
|
819 |
+
"learning_rate": 1.794871794871795e-05,
|
820 |
+
"loss": 0.0526,
|
821 |
+
"step": 1160
|
822 |
+
},
|
823 |
+
{
|
824 |
+
"epoch": 0.78,
|
825 |
+
"grad_norm": 2.806793867054328,
|
826 |
+
"learning_rate": 1.7911824386644532e-05,
|
827 |
+
"loss": 0.047,
|
828 |
+
"step": 1170
|
829 |
+
},
|
830 |
+
{
|
831 |
+
"epoch": 0.78,
|
832 |
+
"grad_norm": 3.323409487413899,
|
833 |
+
"learning_rate": 1.7874930824571113e-05,
|
834 |
+
"loss": 0.0455,
|
835 |
+
"step": 1180
|
836 |
+
},
|
837 |
+
{
|
838 |
+
"epoch": 0.79,
|
839 |
+
"grad_norm": 1.0877783561592873,
|
840 |
+
"learning_rate": 1.7838037262497695e-05,
|
841 |
+
"loss": 0.0555,
|
842 |
+
"step": 1190
|
843 |
+
},
|
844 |
+
{
|
845 |
+
"epoch": 0.8,
|
846 |
+
"grad_norm": 1.9158165193439103,
|
847 |
+
"learning_rate": 1.7801143700424276e-05,
|
848 |
+
"loss": 0.0458,
|
849 |
+
"step": 1200
|
850 |
+
},
|
851 |
+
{
|
852 |
+
"epoch": 0.8,
|
853 |
+
"grad_norm": 1.0863638939536142,
|
854 |
+
"learning_rate": 1.776425013835086e-05,
|
855 |
+
"loss": 0.0462,
|
856 |
+
"step": 1210
|
857 |
+
},
|
858 |
+
{
|
859 |
+
"epoch": 0.81,
|
860 |
+
"grad_norm": 2.3398031954044978,
|
861 |
+
"learning_rate": 1.772735657627744e-05,
|
862 |
+
"loss": 0.032,
|
863 |
+
"step": 1220
|
864 |
+
},
|
865 |
+
{
|
866 |
+
"epoch": 0.82,
|
867 |
+
"grad_norm": 0.9272917966743286,
|
868 |
+
"learning_rate": 1.7690463014204022e-05,
|
869 |
+
"loss": 0.0464,
|
870 |
+
"step": 1230
|
871 |
+
},
|
872 |
+
{
|
873 |
+
"epoch": 0.82,
|
874 |
+
"grad_norm": 1.4608044079217914,
|
875 |
+
"learning_rate": 1.7653569452130607e-05,
|
876 |
+
"loss": 0.0544,
|
877 |
+
"step": 1240
|
878 |
+
},
|
879 |
+
{
|
880 |
+
"epoch": 0.83,
|
881 |
+
"grad_norm": 2.0799767984416713,
|
882 |
+
"learning_rate": 1.7616675890057188e-05,
|
883 |
+
"loss": 0.0629,
|
884 |
+
"step": 1250
|
885 |
+
},
|
886 |
+
{
|
887 |
+
"epoch": 0.84,
|
888 |
+
"grad_norm": 2.438242481318271,
|
889 |
+
"learning_rate": 1.757978232798377e-05,
|
890 |
+
"loss": 0.0586,
|
891 |
+
"step": 1260
|
892 |
+
},
|
893 |
+
{
|
894 |
+
"epoch": 0.84,
|
895 |
+
"grad_norm": 0.22666963149041416,
|
896 |
+
"learning_rate": 1.754288876591035e-05,
|
897 |
+
"loss": 0.0419,
|
898 |
+
"step": 1270
|
899 |
+
},
|
900 |
+
{
|
901 |
+
"epoch": 0.85,
|
902 |
+
"grad_norm": 1.6906855856145777,
|
903 |
+
"learning_rate": 1.750599520383693e-05,
|
904 |
+
"loss": 0.0417,
|
905 |
+
"step": 1280
|
906 |
+
},
|
907 |
+
{
|
908 |
+
"epoch": 0.86,
|
909 |
+
"grad_norm": 0.8417294291584847,
|
910 |
+
"learning_rate": 1.7469101641763513e-05,
|
911 |
+
"loss": 0.0496,
|
912 |
+
"step": 1290
|
913 |
+
},
|
914 |
+
{
|
915 |
+
"epoch": 0.86,
|
916 |
+
"grad_norm": 2.384648750559231,
|
917 |
+
"learning_rate": 1.7432208079690097e-05,
|
918 |
+
"loss": 0.037,
|
919 |
+
"step": 1300
|
920 |
+
},
|
921 |
+
{
|
922 |
+
"epoch": 0.87,
|
923 |
+
"grad_norm": 1.0969098360012755,
|
924 |
+
"learning_rate": 1.7395314517616678e-05,
|
925 |
+
"loss": 0.0392,
|
926 |
+
"step": 1310
|
927 |
+
},
|
928 |
+
{
|
929 |
+
"epoch": 0.88,
|
930 |
+
"grad_norm": 4.265947825753472,
|
931 |
+
"learning_rate": 1.735842095554326e-05,
|
932 |
+
"loss": 0.0434,
|
933 |
+
"step": 1320
|
934 |
+
},
|
935 |
+
{
|
936 |
+
"epoch": 0.88,
|
937 |
+
"grad_norm": 1.0043877066833349,
|
938 |
+
"learning_rate": 1.732152739346984e-05,
|
939 |
+
"loss": 0.0423,
|
940 |
+
"step": 1330
|
941 |
+
},
|
942 |
+
{
|
943 |
+
"epoch": 0.89,
|
944 |
+
"grad_norm": 0.3834891081373628,
|
945 |
+
"learning_rate": 1.728463383139642e-05,
|
946 |
+
"loss": 0.0423,
|
947 |
+
"step": 1340
|
948 |
+
},
|
949 |
+
{
|
950 |
+
"epoch": 0.9,
|
951 |
+
"grad_norm": 0.8917610552257311,
|
952 |
+
"learning_rate": 1.7247740269323006e-05,
|
953 |
+
"loss": 0.0491,
|
954 |
+
"step": 1350
|
955 |
+
},
|
956 |
+
{
|
957 |
+
"epoch": 0.9,
|
958 |
+
"grad_norm": 1.7777935423085822,
|
959 |
+
"learning_rate": 1.7210846707249587e-05,
|
960 |
+
"loss": 0.0465,
|
961 |
+
"step": 1360
|
962 |
+
},
|
963 |
+
{
|
964 |
+
"epoch": 0.91,
|
965 |
+
"grad_norm": 2.882686457165806,
|
966 |
+
"learning_rate": 1.717395314517617e-05,
|
967 |
+
"loss": 0.0366,
|
968 |
+
"step": 1370
|
969 |
+
},
|
970 |
+
{
|
971 |
+
"epoch": 0.92,
|
972 |
+
"grad_norm": 1.4070944899475468,
|
973 |
+
"learning_rate": 1.713705958310275e-05,
|
974 |
+
"loss": 0.0364,
|
975 |
+
"step": 1380
|
976 |
+
},
|
977 |
+
{
|
978 |
+
"epoch": 0.92,
|
979 |
+
"grad_norm": 1.368387839772382,
|
980 |
+
"learning_rate": 1.710016602102933e-05,
|
981 |
+
"loss": 0.0302,
|
982 |
+
"step": 1390
|
983 |
+
},
|
984 |
+
{
|
985 |
+
"epoch": 0.93,
|
986 |
+
"grad_norm": 1.2202288141019488,
|
987 |
+
"learning_rate": 1.7063272458955912e-05,
|
988 |
+
"loss": 0.0455,
|
989 |
+
"step": 1400
|
990 |
+
},
|
991 |
+
{
|
992 |
+
"epoch": 0.94,
|
993 |
+
"grad_norm": 1.7419694841079256,
|
994 |
+
"learning_rate": 1.7026378896882496e-05,
|
995 |
+
"loss": 0.0402,
|
996 |
+
"step": 1410
|
997 |
+
},
|
998 |
+
{
|
999 |
+
"epoch": 0.94,
|
1000 |
+
"grad_norm": 1.5792094366611027,
|
1001 |
+
"learning_rate": 1.6989485334809077e-05,
|
1002 |
+
"loss": 0.0364,
|
1003 |
+
"step": 1420
|
1004 |
+
},
|
1005 |
+
{
|
1006 |
+
"epoch": 0.95,
|
1007 |
+
"grad_norm": 1.0092569099059323,
|
1008 |
+
"learning_rate": 1.695259177273566e-05,
|
1009 |
+
"loss": 0.0335,
|
1010 |
+
"step": 1430
|
1011 |
+
},
|
1012 |
+
{
|
1013 |
+
"epoch": 0.96,
|
1014 |
+
"grad_norm": 1.5374731387141227,
|
1015 |
+
"learning_rate": 1.691569821066224e-05,
|
1016 |
+
"loss": 0.0428,
|
1017 |
+
"step": 1440
|
1018 |
+
},
|
1019 |
+
{
|
1020 |
+
"epoch": 0.96,
|
1021 |
+
"grad_norm": 1.6268956058587998,
|
1022 |
+
"learning_rate": 1.687880464858882e-05,
|
1023 |
+
"loss": 0.0483,
|
1024 |
+
"step": 1450
|
1025 |
+
},
|
1026 |
+
{
|
1027 |
+
"epoch": 0.97,
|
1028 |
+
"grad_norm": 1.6628348966888566,
|
1029 |
+
"learning_rate": 1.6841911086515402e-05,
|
1030 |
+
"loss": 0.0487,
|
1031 |
+
"step": 1460
|
1032 |
+
},
|
1033 |
+
{
|
1034 |
+
"epoch": 0.98,
|
1035 |
+
"grad_norm": 8.037103442054311,
|
1036 |
+
"learning_rate": 1.6805017524441987e-05,
|
1037 |
+
"loss": 0.0679,
|
1038 |
+
"step": 1470
|
1039 |
+
},
|
1040 |
+
{
|
1041 |
+
"epoch": 0.98,
|
1042 |
+
"grad_norm": 1.620270154467724,
|
1043 |
+
"learning_rate": 1.6768123962368568e-05,
|
1044 |
+
"loss": 0.0437,
|
1045 |
+
"step": 1480
|
1046 |
+
},
|
1047 |
+
{
|
1048 |
+
"epoch": 0.99,
|
1049 |
+
"grad_norm": 1.0436875513091504,
|
1050 |
+
"learning_rate": 1.6731230400295152e-05,
|
1051 |
+
"loss": 0.0346,
|
1052 |
+
"step": 1490
|
1053 |
+
},
|
1054 |
+
{
|
1055 |
+
"epoch": 1.0,
|
1056 |
+
"grad_norm": 0.8374706362800531,
|
1057 |
+
"learning_rate": 1.6694336838221733e-05,
|
1058 |
+
"loss": 0.0517,
|
1059 |
+
"step": 1500
|
1060 |
+
}
|
1061 |
+
],
|
1062 |
+
"logging_steps": 10,
|
1063 |
+
"max_steps": 6024,
|
1064 |
+
"num_input_tokens_seen": 0,
|
1065 |
+
"num_train_epochs": 4,
|
1066 |
+
"save_steps": 500,
|
1067 |
+
"total_flos": 0.0,
|
1068 |
+
"train_batch_size": 1,
|
1069 |
+
"trial_name": null,
|
1070 |
+
"trial_params": null
|
1071 |
+
}
|
checkpoint-1506/training_args.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d0640503f1433ff703cc07d0997c18edbdba68efdd139a495f5d107bbd939a06
|
3 |
+
size 6395
|
checkpoint-1506/zero_to_fp32.py
ADDED
@@ -0,0 +1,587 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
#!/usr/bin/env python
|
2 |
+
|
3 |
+
# Copyright (c) Microsoft Corporation.
|
4 |
+
# SPDX-License-Identifier: Apache-2.0
|
5 |
+
|
6 |
+
# DeepSpeed Team
|
7 |
+
|
8 |
+
# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
|
9 |
+
# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
|
10 |
+
# the future. Once extracted, the weights don't require DeepSpeed and can be used in any
|
11 |
+
# application.
|
12 |
+
#
|
13 |
+
# example: python zero_to_fp32.py . pytorch_model.bin
|
14 |
+
|
15 |
+
import argparse
|
16 |
+
import torch
|
17 |
+
import glob
|
18 |
+
import math
|
19 |
+
import os
|
20 |
+
import re
|
21 |
+
from collections import OrderedDict
|
22 |
+
from dataclasses import dataclass
|
23 |
+
|
24 |
+
# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
|
25 |
+
# DeepSpeed data structures it has to be available in the current python environment.
|
26 |
+
from deepspeed.utils import logger
|
27 |
+
from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
|
28 |
+
FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
|
29 |
+
FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
|
30 |
+
|
31 |
+
|
32 |
+
@dataclass
|
33 |
+
class zero_model_state:
|
34 |
+
buffers: dict()
|
35 |
+
param_shapes: dict()
|
36 |
+
shared_params: list
|
37 |
+
ds_version: int
|
38 |
+
frozen_param_shapes: dict()
|
39 |
+
frozen_param_fragments: dict()
|
40 |
+
|
41 |
+
|
42 |
+
debug = 0
|
43 |
+
|
44 |
+
# load to cpu
|
45 |
+
device = torch.device('cpu')
|
46 |
+
|
47 |
+
|
48 |
+
def atoi(text):
|
49 |
+
return int(text) if text.isdigit() else text
|
50 |
+
|
51 |
+
|
52 |
+
def natural_keys(text):
|
53 |
+
'''
|
54 |
+
alist.sort(key=natural_keys) sorts in human order
|
55 |
+
http://nedbatchelder.com/blog/200712/human_sorting.html
|
56 |
+
(See Toothy's implementation in the comments)
|
57 |
+
'''
|
58 |
+
return [atoi(c) for c in re.split(r'(\d+)', text)]
|
59 |
+
|
60 |
+
|
61 |
+
def get_model_state_file(checkpoint_dir, zero_stage):
|
62 |
+
if not os.path.isdir(checkpoint_dir):
|
63 |
+
raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
|
64 |
+
|
65 |
+
# there should be only one file
|
66 |
+
if zero_stage <= 2:
|
67 |
+
file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
|
68 |
+
elif zero_stage == 3:
|
69 |
+
file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
|
70 |
+
|
71 |
+
if not os.path.exists(file):
|
72 |
+
raise FileNotFoundError(f"can't find model states file at '{file}'")
|
73 |
+
|
74 |
+
return file
|
75 |
+
|
76 |
+
|
77 |
+
def get_checkpoint_files(checkpoint_dir, glob_pattern):
|
78 |
+
# XXX: need to test that this simple glob rule works for multi-node setup too
|
79 |
+
ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
|
80 |
+
|
81 |
+
if len(ckpt_files) == 0:
|
82 |
+
raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
|
83 |
+
|
84 |
+
return ckpt_files
|
85 |
+
|
86 |
+
|
87 |
+
def get_optim_files(checkpoint_dir):
|
88 |
+
return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
|
89 |
+
|
90 |
+
|
91 |
+
def get_model_state_files(checkpoint_dir):
|
92 |
+
return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
|
93 |
+
|
94 |
+
|
95 |
+
def parse_model_states(files):
|
96 |
+
zero_model_states = []
|
97 |
+
for file in files:
|
98 |
+
state_dict = torch.load(file, map_location=device)
|
99 |
+
|
100 |
+
if BUFFER_NAMES not in state_dict:
|
101 |
+
raise ValueError(f"{file} is not a model state checkpoint")
|
102 |
+
buffer_names = state_dict[BUFFER_NAMES]
|
103 |
+
if debug:
|
104 |
+
print("Found buffers:", buffer_names)
|
105 |
+
|
106 |
+
# recover just the buffers while restoring them to fp32 if they were saved in fp16
|
107 |
+
buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
|
108 |
+
param_shapes = state_dict[PARAM_SHAPES]
|
109 |
+
|
110 |
+
# collect parameters that are included in param_shapes
|
111 |
+
param_names = []
|
112 |
+
for s in param_shapes:
|
113 |
+
for name in s.keys():
|
114 |
+
param_names.append(name)
|
115 |
+
|
116 |
+
# update with frozen parameters
|
117 |
+
frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
|
118 |
+
if frozen_param_shapes is not None:
|
119 |
+
if debug:
|
120 |
+
print(f"Found frozen_param_shapes: {frozen_param_shapes}")
|
121 |
+
param_names += list(frozen_param_shapes.keys())
|
122 |
+
|
123 |
+
# handle shared params
|
124 |
+
shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
|
125 |
+
|
126 |
+
ds_version = state_dict.get(DS_VERSION, None)
|
127 |
+
|
128 |
+
frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
|
129 |
+
|
130 |
+
z_model_state = zero_model_state(buffers=buffers,
|
131 |
+
param_shapes=param_shapes,
|
132 |
+
shared_params=shared_params,
|
133 |
+
ds_version=ds_version,
|
134 |
+
frozen_param_shapes=frozen_param_shapes,
|
135 |
+
frozen_param_fragments=frozen_param_fragments)
|
136 |
+
zero_model_states.append(z_model_state)
|
137 |
+
|
138 |
+
return zero_model_states
|
139 |
+
|
140 |
+
|
141 |
+
def parse_optim_states(files, ds_checkpoint_dir):
|
142 |
+
|
143 |
+
total_files = len(files)
|
144 |
+
state_dicts = []
|
145 |
+
for f in files:
|
146 |
+
state_dict = torch.load(f, map_location=device)
|
147 |
+
# immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
|
148 |
+
# and also handle the case where it was already removed by another helper script
|
149 |
+
state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
|
150 |
+
state_dicts.append(state_dict)
|
151 |
+
|
152 |
+
if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
|
153 |
+
raise ValueError(f"{files[0]} is not a zero checkpoint")
|
154 |
+
zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
|
155 |
+
world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
|
156 |
+
|
157 |
+
# For ZeRO-2 each param group can have different partition_count as data parallelism for expert
|
158 |
+
# parameters can be different from data parallelism for non-expert parameters. So we can just
|
159 |
+
# use the max of the partition_count to get the dp world_size.
|
160 |
+
|
161 |
+
if type(world_size) is list:
|
162 |
+
world_size = max(world_size)
|
163 |
+
|
164 |
+
if world_size != total_files:
|
165 |
+
raise ValueError(
|
166 |
+
f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
|
167 |
+
"Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
|
168 |
+
)
|
169 |
+
|
170 |
+
# the groups are named differently in each stage
|
171 |
+
if zero_stage <= 2:
|
172 |
+
fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
|
173 |
+
elif zero_stage == 3:
|
174 |
+
fp32_groups_key = FP32_FLAT_GROUPS
|
175 |
+
else:
|
176 |
+
raise ValueError(f"unknown zero stage {zero_stage}")
|
177 |
+
|
178 |
+
if zero_stage <= 2:
|
179 |
+
fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
|
180 |
+
elif zero_stage == 3:
|
181 |
+
# if there is more than one param group, there will be multiple flattened tensors - one
|
182 |
+
# flattened tensor per group - for simplicity merge them into a single tensor
|
183 |
+
#
|
184 |
+
# XXX: could make the script more memory efficient for when there are multiple groups - it
|
185 |
+
# will require matching the sub-lists of param_shapes for each param group flattened tensor
|
186 |
+
|
187 |
+
fp32_flat_groups = [
|
188 |
+
torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
|
189 |
+
]
|
190 |
+
|
191 |
+
return zero_stage, world_size, fp32_flat_groups
|
192 |
+
|
193 |
+
|
194 |
+
def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir):
|
195 |
+
"""
|
196 |
+
Returns fp32 state_dict reconstructed from ds checkpoint
|
197 |
+
|
198 |
+
Args:
|
199 |
+
- ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
|
200 |
+
|
201 |
+
"""
|
202 |
+
print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
|
203 |
+
|
204 |
+
optim_files = get_optim_files(ds_checkpoint_dir)
|
205 |
+
zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
|
206 |
+
print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
|
207 |
+
|
208 |
+
model_files = get_model_state_files(ds_checkpoint_dir)
|
209 |
+
|
210 |
+
zero_model_states = parse_model_states(model_files)
|
211 |
+
print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
|
212 |
+
|
213 |
+
if zero_stage <= 2:
|
214 |
+
return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states)
|
215 |
+
elif zero_stage == 3:
|
216 |
+
return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states)
|
217 |
+
|
218 |
+
|
219 |
+
def _zero2_merge_frozen_params(state_dict, zero_model_states):
|
220 |
+
if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
|
221 |
+
return
|
222 |
+
|
223 |
+
frozen_param_shapes = zero_model_states[0].frozen_param_shapes
|
224 |
+
frozen_param_fragments = zero_model_states[0].frozen_param_fragments
|
225 |
+
|
226 |
+
if debug:
|
227 |
+
num_elem = sum(s.numel() for s in frozen_param_shapes.values())
|
228 |
+
print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
|
229 |
+
|
230 |
+
wanted_params = len(frozen_param_shapes)
|
231 |
+
wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
|
232 |
+
avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
|
233 |
+
print(f'Frozen params: Have {avail_numel} numels to process.')
|
234 |
+
print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
|
235 |
+
|
236 |
+
total_params = 0
|
237 |
+
total_numel = 0
|
238 |
+
for name, shape in frozen_param_shapes.items():
|
239 |
+
total_params += 1
|
240 |
+
unpartitioned_numel = shape.numel()
|
241 |
+
total_numel += unpartitioned_numel
|
242 |
+
|
243 |
+
state_dict[name] = frozen_param_fragments[name]
|
244 |
+
|
245 |
+
if debug:
|
246 |
+
print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
|
247 |
+
|
248 |
+
print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
|
249 |
+
|
250 |
+
|
251 |
+
def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
|
252 |
+
param_shapes = zero_model_states[0].param_shapes
|
253 |
+
|
254 |
+
# Reconstruction protocol:
|
255 |
+
#
|
256 |
+
# XXX: document this
|
257 |
+
|
258 |
+
if debug:
|
259 |
+
for i in range(world_size):
|
260 |
+
for j in range(len(fp32_flat_groups[0])):
|
261 |
+
print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
|
262 |
+
|
263 |
+
# XXX: memory usage doubles here (zero2)
|
264 |
+
num_param_groups = len(fp32_flat_groups[0])
|
265 |
+
merged_single_partition_of_fp32_groups = []
|
266 |
+
for i in range(num_param_groups):
|
267 |
+
merged_partitions = [sd[i] for sd in fp32_flat_groups]
|
268 |
+
full_single_fp32_vector = torch.cat(merged_partitions, 0)
|
269 |
+
merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
|
270 |
+
avail_numel = sum(
|
271 |
+
[full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
|
272 |
+
|
273 |
+
if debug:
|
274 |
+
wanted_params = sum([len(shapes) for shapes in param_shapes])
|
275 |
+
wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
|
276 |
+
# not asserting if there is a mismatch due to possible padding
|
277 |
+
print(f"Have {avail_numel} numels to process.")
|
278 |
+
print(f"Need {wanted_numel} numels in {wanted_params} params.")
|
279 |
+
|
280 |
+
# params
|
281 |
+
# XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
|
282 |
+
# out-of-core computing solution
|
283 |
+
total_numel = 0
|
284 |
+
total_params = 0
|
285 |
+
for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
|
286 |
+
offset = 0
|
287 |
+
avail_numel = full_single_fp32_vector.numel()
|
288 |
+
for name, shape in shapes.items():
|
289 |
+
|
290 |
+
unpartitioned_numel = shape.numel()
|
291 |
+
total_numel += unpartitioned_numel
|
292 |
+
total_params += 1
|
293 |
+
|
294 |
+
if debug:
|
295 |
+
print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
|
296 |
+
state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
|
297 |
+
offset += unpartitioned_numel
|
298 |
+
|
299 |
+
# Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
|
300 |
+
# avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
|
301 |
+
# paddings performed in the code it's almost impossible to predict the exact numbers w/o the
|
302 |
+
# live optimizer object, so we are checking that the numbers are within the right range
|
303 |
+
align_to = 2 * world_size
|
304 |
+
|
305 |
+
def zero2_align(x):
|
306 |
+
return align_to * math.ceil(x / align_to)
|
307 |
+
|
308 |
+
if debug:
|
309 |
+
print(f"original offset={offset}, avail_numel={avail_numel}")
|
310 |
+
|
311 |
+
offset = zero2_align(offset)
|
312 |
+
avail_numel = zero2_align(avail_numel)
|
313 |
+
|
314 |
+
if debug:
|
315 |
+
print(f"aligned offset={offset}, avail_numel={avail_numel}")
|
316 |
+
|
317 |
+
# Sanity check
|
318 |
+
if offset != avail_numel:
|
319 |
+
raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
|
320 |
+
|
321 |
+
print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
|
322 |
+
|
323 |
+
|
324 |
+
def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states):
|
325 |
+
state_dict = OrderedDict()
|
326 |
+
|
327 |
+
# buffers
|
328 |
+
buffers = zero_model_states[0].buffers
|
329 |
+
state_dict.update(buffers)
|
330 |
+
if debug:
|
331 |
+
print(f"added {len(buffers)} buffers")
|
332 |
+
|
333 |
+
_zero2_merge_frozen_params(state_dict, zero_model_states)
|
334 |
+
|
335 |
+
_zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
|
336 |
+
|
337 |
+
# recover shared parameters
|
338 |
+
for pair in zero_model_states[0].shared_params:
|
339 |
+
if pair[1] in state_dict:
|
340 |
+
state_dict[pair[0]] = state_dict[pair[1]]
|
341 |
+
|
342 |
+
return state_dict
|
343 |
+
|
344 |
+
|
345 |
+
def zero3_partitioned_param_info(unpartitioned_numel, world_size):
|
346 |
+
remainder = unpartitioned_numel % world_size
|
347 |
+
padding_numel = (world_size - remainder) if remainder else 0
|
348 |
+
partitioned_numel = math.ceil(unpartitioned_numel / world_size)
|
349 |
+
return partitioned_numel, padding_numel
|
350 |
+
|
351 |
+
|
352 |
+
def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
|
353 |
+
if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
|
354 |
+
return
|
355 |
+
|
356 |
+
if debug:
|
357 |
+
for i in range(world_size):
|
358 |
+
num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
|
359 |
+
print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
|
360 |
+
|
361 |
+
frozen_param_shapes = zero_model_states[0].frozen_param_shapes
|
362 |
+
wanted_params = len(frozen_param_shapes)
|
363 |
+
wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
|
364 |
+
avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
|
365 |
+
print(f'Frozen params: Have {avail_numel} numels to process.')
|
366 |
+
print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
|
367 |
+
|
368 |
+
total_params = 0
|
369 |
+
total_numel = 0
|
370 |
+
for name, shape in zero_model_states[0].frozen_param_shapes.items():
|
371 |
+
total_params += 1
|
372 |
+
unpartitioned_numel = shape.numel()
|
373 |
+
total_numel += unpartitioned_numel
|
374 |
+
|
375 |
+
param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
|
376 |
+
state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
|
377 |
+
|
378 |
+
partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
|
379 |
+
|
380 |
+
if debug:
|
381 |
+
print(
|
382 |
+
f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
|
383 |
+
)
|
384 |
+
|
385 |
+
print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
|
386 |
+
|
387 |
+
|
388 |
+
def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
|
389 |
+
param_shapes = zero_model_states[0].param_shapes
|
390 |
+
avail_numel = fp32_flat_groups[0].numel() * world_size
|
391 |
+
# Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
|
392 |
+
# param, re-consolidating each param, while dealing with padding if any
|
393 |
+
|
394 |
+
# merge list of dicts, preserving order
|
395 |
+
param_shapes = {k: v for d in param_shapes for k, v in d.items()}
|
396 |
+
|
397 |
+
if debug:
|
398 |
+
for i in range(world_size):
|
399 |
+
print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
|
400 |
+
|
401 |
+
wanted_params = len(param_shapes)
|
402 |
+
wanted_numel = sum(shape.numel() for shape in param_shapes.values())
|
403 |
+
# not asserting if there is a mismatch due to possible padding
|
404 |
+
avail_numel = fp32_flat_groups[0].numel() * world_size
|
405 |
+
print(f"Trainable params: Have {avail_numel} numels to process.")
|
406 |
+
print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
|
407 |
+
|
408 |
+
# params
|
409 |
+
# XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
|
410 |
+
# out-of-core computing solution
|
411 |
+
offset = 0
|
412 |
+
total_numel = 0
|
413 |
+
total_params = 0
|
414 |
+
for name, shape in param_shapes.items():
|
415 |
+
|
416 |
+
unpartitioned_numel = shape.numel()
|
417 |
+
total_numel += unpartitioned_numel
|
418 |
+
total_params += 1
|
419 |
+
|
420 |
+
partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
|
421 |
+
|
422 |
+
if debug:
|
423 |
+
print(
|
424 |
+
f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
|
425 |
+
)
|
426 |
+
|
427 |
+
# XXX: memory usage doubles here
|
428 |
+
state_dict[name] = torch.cat(
|
429 |
+
tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
|
430 |
+
0).narrow(0, 0, unpartitioned_numel).view(shape)
|
431 |
+
offset += partitioned_numel
|
432 |
+
|
433 |
+
offset *= world_size
|
434 |
+
|
435 |
+
# Sanity check
|
436 |
+
if offset != avail_numel:
|
437 |
+
raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
|
438 |
+
|
439 |
+
print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
|
440 |
+
|
441 |
+
|
442 |
+
def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states):
|
443 |
+
state_dict = OrderedDict()
|
444 |
+
|
445 |
+
# buffers
|
446 |
+
buffers = zero_model_states[0].buffers
|
447 |
+
state_dict.update(buffers)
|
448 |
+
if debug:
|
449 |
+
print(f"added {len(buffers)} buffers")
|
450 |
+
|
451 |
+
_zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
|
452 |
+
|
453 |
+
_zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
|
454 |
+
|
455 |
+
# recover shared parameters
|
456 |
+
for pair in zero_model_states[0].shared_params:
|
457 |
+
if pair[1] in state_dict:
|
458 |
+
state_dict[pair[0]] = state_dict[pair[1]]
|
459 |
+
|
460 |
+
return state_dict
|
461 |
+
|
462 |
+
|
463 |
+
def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None):
|
464 |
+
"""
|
465 |
+
Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
|
466 |
+
``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
|
467 |
+
via a model hub.
|
468 |
+
|
469 |
+
Args:
|
470 |
+
- ``checkpoint_dir``: path to the desired checkpoint folder
|
471 |
+
- ``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``
|
472 |
+
|
473 |
+
Returns:
|
474 |
+
- pytorch ``state_dict``
|
475 |
+
|
476 |
+
Note: this approach may not work if your application doesn't have sufficient free CPU memory and
|
477 |
+
you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
|
478 |
+
the checkpoint.
|
479 |
+
|
480 |
+
A typical usage might be ::
|
481 |
+
|
482 |
+
from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
|
483 |
+
# do the training and checkpoint saving
|
484 |
+
state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
|
485 |
+
model = model.cpu() # move to cpu
|
486 |
+
model.load_state_dict(state_dict)
|
487 |
+
# submit to model hub or save the model to share with others
|
488 |
+
|
489 |
+
In this example the ``model`` will no longer be usable in the deepspeed context of the same
|
490 |
+
application. i.e. you will need to re-initialize the deepspeed engine, since
|
491 |
+
``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
|
492 |
+
|
493 |
+
If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
|
494 |
+
|
495 |
+
"""
|
496 |
+
if tag is None:
|
497 |
+
latest_path = os.path.join(checkpoint_dir, 'latest')
|
498 |
+
if os.path.isfile(latest_path):
|
499 |
+
with open(latest_path, 'r') as fd:
|
500 |
+
tag = fd.read().strip()
|
501 |
+
else:
|
502 |
+
raise ValueError(f"Unable to find 'latest' file at {latest_path}")
|
503 |
+
|
504 |
+
ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
|
505 |
+
|
506 |
+
if not os.path.isdir(ds_checkpoint_dir):
|
507 |
+
raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
|
508 |
+
|
509 |
+
return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir)
|
510 |
+
|
511 |
+
|
512 |
+
def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None):
|
513 |
+
"""
|
514 |
+
Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
|
515 |
+
loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
|
516 |
+
|
517 |
+
Args:
|
518 |
+
- ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
|
519 |
+
- ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin)
|
520 |
+
- ``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``
|
521 |
+
"""
|
522 |
+
|
523 |
+
state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
|
524 |
+
print(f"Saving fp32 state dict to {output_file}")
|
525 |
+
torch.save(state_dict, output_file)
|
526 |
+
|
527 |
+
|
528 |
+
def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
|
529 |
+
"""
|
530 |
+
1. Put the provided model to cpu
|
531 |
+
2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
|
532 |
+
3. Load it into the provided model
|
533 |
+
|
534 |
+
Args:
|
535 |
+
- ``model``: the model object to update
|
536 |
+
- ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
|
537 |
+
- ``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``
|
538 |
+
|
539 |
+
Returns:
|
540 |
+
- ``model`: modified model
|
541 |
+
|
542 |
+
Make sure you have plenty of CPU memory available before you call this function. If you don't
|
543 |
+
have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
|
544 |
+
conveniently placed for you in the checkpoint folder.
|
545 |
+
|
546 |
+
A typical usage might be ::
|
547 |
+
|
548 |
+
from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
|
549 |
+
model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
|
550 |
+
# submit to model hub or save the model to share with others
|
551 |
+
|
552 |
+
Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
|
553 |
+
of the same application. i.e. you will need to re-initialize the deepspeed engine, since
|
554 |
+
``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
|
555 |
+
|
556 |
+
"""
|
557 |
+
logger.info(f"Extracting fp32 weights")
|
558 |
+
state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
|
559 |
+
|
560 |
+
logger.info(f"Overwriting model with fp32 weights")
|
561 |
+
model = model.cpu()
|
562 |
+
model.load_state_dict(state_dict, strict=False)
|
563 |
+
|
564 |
+
return model
|
565 |
+
|
566 |
+
|
567 |
+
if __name__ == "__main__":
|
568 |
+
|
569 |
+
parser = argparse.ArgumentParser()
|
570 |
+
parser.add_argument("checkpoint_dir",
|
571 |
+
type=str,
|
572 |
+
help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
|
573 |
+
parser.add_argument(
|
574 |
+
"output_file",
|
575 |
+
type=str,
|
576 |
+
help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)")
|
577 |
+
parser.add_argument("-t",
|
578 |
+
"--tag",
|
579 |
+
type=str,
|
580 |
+
default=None,
|
581 |
+
help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
|
582 |
+
parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
|
583 |
+
args = parser.parse_args()
|
584 |
+
|
585 |
+
debug = args.debug
|
586 |
+
|
587 |
+
convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, args.output_file, tag=args.tag)
|
checkpoint-3012/1_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-3012/README.md
ADDED
@@ -0,0 +1,57 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
library_name: sentence-transformers
|
3 |
+
pipeline_tag: sentence-similarity
|
4 |
+
tags:
|
5 |
+
- sentence-transformers
|
6 |
+
- feature-extraction
|
7 |
+
- sentence-similarity
|
8 |
+
|
9 |
+
---
|
10 |
+
|
11 |
+
# {MODEL_NAME}
|
12 |
+
|
13 |
+
This is a [sentence-transformers](https://www.SBERT.net) model: It maps sentences & paragraphs to a 1024 dimensional dense vector space and can be used for tasks like clustering or semantic search.
|
14 |
+
|
15 |
+
<!--- Describe your model here -->
|
16 |
+
|
17 |
+
## Usage (Sentence-Transformers)
|
18 |
+
|
19 |
+
Using this model becomes easy when you have [sentence-transformers](https://www.SBERT.net) installed:
|
20 |
+
|
21 |
+
```
|
22 |
+
pip install -U sentence-transformers
|
23 |
+
```
|
24 |
+
|
25 |
+
Then you can use the model like this:
|
26 |
+
|
27 |
+
```python
|
28 |
+
from sentence_transformers import SentenceTransformer
|
29 |
+
sentences = ["This is an example sentence", "Each sentence is converted"]
|
30 |
+
|
31 |
+
model = SentenceTransformer('{MODEL_NAME}')
|
32 |
+
embeddings = model.encode(sentences)
|
33 |
+
print(embeddings)
|
34 |
+
```
|
35 |
+
|
36 |
+
|
37 |
+
|
38 |
+
## Evaluation Results
|
39 |
+
|
40 |
+
<!--- Describe how your model was evaluated -->
|
41 |
+
|
42 |
+
For an automated evaluation of this model, see the *Sentence Embeddings Benchmark*: [https://seb.sbert.net](https://seb.sbert.net?model_name={MODEL_NAME})
|
43 |
+
|
44 |
+
|
45 |
+
|
46 |
+
## Full Model Architecture
|
47 |
+
```
|
48 |
+
SentenceTransformer(
|
49 |
+
(0): Transformer({'max_seq_length': 8192, 'do_lower_case': False}) with Transformer model: XLMRobertaModel
|
50 |
+
(1): Pooling({'word_embedding_dimension': 1024, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})
|
51 |
+
(2): Normalize()
|
52 |
+
)
|
53 |
+
```
|
54 |
+
|
55 |
+
## Citing & Authors
|
56 |
+
|
57 |
+
<!--- Describe where people can find more information -->
|
checkpoint-3012/colbert_linear.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:0128bf0287a90a15b7eaec7c5f69e5f613ed96ea00aaaf9aac377e850e408150
|
3 |
+
size 2100227
|
checkpoint-3012/config.json
ADDED
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "saved_models/bgem3_unified_finetune_20240330/checkpoint-3012",
|
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": "4.39.1",
|
25 |
+
"type_vocab_size": 1,
|
26 |
+
"use_cache": true,
|
27 |
+
"vocab_size": 250002
|
28 |
+
}
|
checkpoint-3012/config_sentence_transformers.json
ADDED
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"__version__": {
|
3 |
+
"sentence_transformers": "2.6.0",
|
4 |
+
"transformers": "4.39.1",
|
5 |
+
"pytorch": "2.0.1+cu117"
|
6 |
+
},
|
7 |
+
"prompts": {},
|
8 |
+
"default_prompt_name": null
|
9 |
+
}
|
checkpoint-3012/global_step3012/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 sha256:9e307452471b3c898e9d737e82e226c9acaaebccae707863ebe0b17de0d49084
|
3 |
+
size 3412837975
|
checkpoint-3012/global_step3012/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 sha256:03b03e3850eb0cff1c13b757f5235cb3a3d3223776a58f781cffe8ad17186242
|
3 |
+
size 3412867031
|
checkpoint-3012/global_step3012/mp_rank_00_model_states.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5c11af2f4f6cc7501daf6f08f76c026344eb99602ad070d97afa940e3044afdf
|
3 |
+
size 1137729563
|
checkpoint-3012/latest
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
global_step3012
|
checkpoint-3012/model.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:f6e88b6ba28dae9a4fe44d8cd57f2ec5c4e28410d20e6741e936a9a5ec015291
|
3 |
+
size 2271064456
|
checkpoint-3012/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-3012/rng_state_0.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e75d231787a0d08e46d1f2c54ebb26d6c17903d0bd81a0b3c81c4e4da4d61e1b
|
3 |
+
size 15607
|
checkpoint-3012/rng_state_1.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:640768ea784ed4e73d22ca543abeda8a883aebcc8f2e2fedcec8e78da8876a12
|
3 |
+
size 15607
|
checkpoint-3012/sentence_bert_config.json
ADDED
@@ -0,0 +1,4 @@
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"max_seq_length": 8192,
|
3 |
+
"do_lower_case": false
|
4 |
+
}
|
checkpoint-3012/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-3012/sparse_linear.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:9e06286144912ddf5d97f4f6227e5f8ce29b1d8b02f4fb8605e02e63f85849fb
|
3 |
+
size 3071
|
checkpoint-3012/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-3012/tokenizer.json
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:69564b696052886ed0ac63fa393e928384e0f8caada38c1f4864a9bfbf379c15
|
3 |
+
size 17098273
|
checkpoint-3012/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-3012/trainer_state.json
ADDED
@@ -0,0 +1,2128 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": null,
|
3 |
+
"best_model_checkpoint": null,
|
4 |
+
"epoch": 2.0,
|
5 |
+
"eval_steps": 500,
|
6 |
+
"global_step": 3012,
|
7 |
+
"is_hyper_param_search": false,
|
8 |
+
"is_local_process_zero": true,
|
9 |
+
"is_world_process_zero": true,
|
10 |
+
"log_history": [
|
11 |
+
{
|
12 |
+
"epoch": 0.01,
|
13 |
+
"grad_norm": 41.35739895402257,
|
14 |
+
"learning_rate": 7.193423539345941e-06,
|
15 |
+
"loss": 0.5141,
|
16 |
+
"step": 10
|
17 |
+
},
|
18 |
+
{
|
19 |
+
"epoch": 0.01,
|
20 |
+
"grad_norm": 10.443694874625423,
|
21 |
+
"learning_rate": 9.358859796204429e-06,
|
22 |
+
"loss": 0.4195,
|
23 |
+
"step": 20
|
24 |
+
},
|
25 |
+
{
|
26 |
+
"epoch": 0.02,
|
27 |
+
"grad_norm": 5.720021591145773,
|
28 |
+
"learning_rate": 1.0625558804168632e-05,
|
29 |
+
"loss": 0.2851,
|
30 |
+
"step": 30
|
31 |
+
},
|
32 |
+
{
|
33 |
+
"epoch": 0.03,
|
34 |
+
"grad_norm": 7.59250257002781,
|
35 |
+
"learning_rate": 1.1524296053062918e-05,
|
36 |
+
"loss": 0.2174,
|
37 |
+
"step": 40
|
38 |
+
},
|
39 |
+
{
|
40 |
+
"epoch": 0.03,
|
41 |
+
"grad_norm": 11.521550428114184,
|
42 |
+
"learning_rate": 1.2221410821833392e-05,
|
43 |
+
"loss": 0.1817,
|
44 |
+
"step": 50
|
45 |
+
},
|
46 |
+
{
|
47 |
+
"epoch": 0.04,
|
48 |
+
"grad_norm": 6.027998448017936,
|
49 |
+
"learning_rate": 1.2790995061027121e-05,
|
50 |
+
"loss": 0.1886,
|
51 |
+
"step": 60
|
52 |
+
},
|
53 |
+
{
|
54 |
+
"epoch": 0.05,
|
55 |
+
"grad_norm": 4.929816146115093,
|
56 |
+
"learning_rate": 1.3272571673439616e-05,
|
57 |
+
"loss": 0.1553,
|
58 |
+
"step": 70
|
59 |
+
},
|
60 |
+
{
|
61 |
+
"epoch": 0.05,
|
62 |
+
"grad_norm": 2.3881515971871345,
|
63 |
+
"learning_rate": 1.3689732309921406e-05,
|
64 |
+
"loss": 0.129,
|
65 |
+
"step": 80
|
66 |
+
},
|
67 |
+
{
|
68 |
+
"epoch": 0.06,
|
69 |
+
"grad_norm": 5.441118636680569,
|
70 |
+
"learning_rate": 1.4057694068991321e-05,
|
71 |
+
"loss": 0.1433,
|
72 |
+
"step": 90
|
73 |
+
},
|
74 |
+
{
|
75 |
+
"epoch": 0.07,
|
76 |
+
"grad_norm": 6.085213496930297,
|
77 |
+
"learning_rate": 1.4386847078691883e-05,
|
78 |
+
"loss": 0.1092,
|
79 |
+
"step": 100
|
80 |
+
},
|
81 |
+
{
|
82 |
+
"epoch": 0.07,
|
83 |
+
"grad_norm": 4.713361075579101,
|
84 |
+
"learning_rate": 1.4684602194465794e-05,
|
85 |
+
"loss": 0.1231,
|
86 |
+
"step": 110
|
87 |
+
},
|
88 |
+
{
|
89 |
+
"epoch": 0.08,
|
90 |
+
"grad_norm": 3.8259500358417924,
|
91 |
+
"learning_rate": 1.495643131788561e-05,
|
92 |
+
"loss": 0.0697,
|
93 |
+
"step": 120
|
94 |
+
},
|
95 |
+
{
|
96 |
+
"epoch": 0.09,
|
97 |
+
"grad_norm": 2.67920682727699,
|
98 |
+
"learning_rate": 1.5206489871327869e-05,
|
99 |
+
"loss": 0.084,
|
100 |
+
"step": 130
|
101 |
+
},
|
102 |
+
{
|
103 |
+
"epoch": 0.09,
|
104 |
+
"grad_norm": 8.189491553913532,
|
105 |
+
"learning_rate": 1.54380079302981e-05,
|
106 |
+
"loss": 0.1023,
|
107 |
+
"step": 140
|
108 |
+
},
|
109 |
+
{
|
110 |
+
"epoch": 0.1,
|
111 |
+
"grad_norm": 12.000369384166694,
|
112 |
+
"learning_rate": 1.5653546086656083e-05,
|
113 |
+
"loss": 0.0972,
|
114 |
+
"step": 150
|
115 |
+
},
|
116 |
+
{
|
117 |
+
"epoch": 0.11,
|
118 |
+
"grad_norm": 2.3726942012239953,
|
119 |
+
"learning_rate": 1.5855168566779895e-05,
|
120 |
+
"loss": 0.1036,
|
121 |
+
"step": 160
|
122 |
+
},
|
123 |
+
{
|
124 |
+
"epoch": 0.11,
|
125 |
+
"grad_norm": 2.0242117206977044,
|
126 |
+
"learning_rate": 1.604456377435124e-05,
|
127 |
+
"loss": 0.1081,
|
128 |
+
"step": 170
|
129 |
+
},
|
130 |
+
{
|
131 |
+
"epoch": 0.12,
|
132 |
+
"grad_norm": 2.2261934744219967,
|
133 |
+
"learning_rate": 1.6223130325849813e-05,
|
134 |
+
"loss": 0.082,
|
135 |
+
"step": 180
|
136 |
+
},
|
137 |
+
{
|
138 |
+
"epoch": 0.13,
|
139 |
+
"grad_norm": 4.70421073268978,
|
140 |
+
"learning_rate": 1.6392039793463407e-05,
|
141 |
+
"loss": 0.0744,
|
142 |
+
"step": 190
|
143 |
+
},
|
144 |
+
{
|
145 |
+
"epoch": 0.13,
|
146 |
+
"grad_norm": 12.064244277626111,
|
147 |
+
"learning_rate": 1.6552283335550368e-05,
|
148 |
+
"loss": 0.0934,
|
149 |
+
"step": 200
|
150 |
+
},
|
151 |
+
{
|
152 |
+
"epoch": 0.14,
|
153 |
+
"grad_norm": 2.150990470911233,
|
154 |
+
"learning_rate": 1.67047069382623e-05,
|
155 |
+
"loss": 0.0737,
|
156 |
+
"step": 210
|
157 |
+
},
|
158 |
+
{
|
159 |
+
"epoch": 0.15,
|
160 |
+
"grad_norm": 4.074258308547766,
|
161 |
+
"learning_rate": 1.6850038451324284e-05,
|
162 |
+
"loss": 0.0841,
|
163 |
+
"step": 220
|
164 |
+
},
|
165 |
+
{
|
166 |
+
"epoch": 0.15,
|
167 |
+
"grad_norm": 2.6764310841274237,
|
168 |
+
"learning_rate": 1.6988908609137504e-05,
|
169 |
+
"loss": 0.0821,
|
170 |
+
"step": 230
|
171 |
+
},
|
172 |
+
{
|
173 |
+
"epoch": 0.16,
|
174 |
+
"grad_norm": 3.1391283710987063,
|
175 |
+
"learning_rate": 1.71218675747441e-05,
|
176 |
+
"loss": 0.0747,
|
177 |
+
"step": 240
|
178 |
+
},
|
179 |
+
{
|
180 |
+
"epoch": 0.17,
|
181 |
+
"grad_norm": 3.1103728720325754,
|
182 |
+
"learning_rate": 1.7249398104320845e-05,
|
183 |
+
"loss": 0.0907,
|
184 |
+
"step": 250
|
185 |
+
},
|
186 |
+
{
|
187 |
+
"epoch": 0.17,
|
188 |
+
"grad_norm": 5.354173276825422,
|
189 |
+
"learning_rate": 1.7371926128186358e-05,
|
190 |
+
"loss": 0.0765,
|
191 |
+
"step": 260
|
192 |
+
},
|
193 |
+
{
|
194 |
+
"epoch": 0.18,
|
195 |
+
"grad_norm": 2.8677808267225005,
|
196 |
+
"learning_rate": 1.7489829333814013e-05,
|
197 |
+
"loss": 0.0777,
|
198 |
+
"step": 270
|
199 |
+
},
|
200 |
+
{
|
201 |
+
"epoch": 0.19,
|
202 |
+
"grad_norm": 2.6124301773304874,
|
203 |
+
"learning_rate": 1.760344418715659e-05,
|
204 |
+
"loss": 0.0897,
|
205 |
+
"step": 280
|
206 |
+
},
|
207 |
+
{
|
208 |
+
"epoch": 0.19,
|
209 |
+
"grad_norm": 2.915238130503912,
|
210 |
+
"learning_rate": 1.7713071721324668e-05,
|
211 |
+
"loss": 0.0866,
|
212 |
+
"step": 290
|
213 |
+
},
|
214 |
+
{
|
215 |
+
"epoch": 0.2,
|
216 |
+
"grad_norm": 4.061775729833908,
|
217 |
+
"learning_rate": 1.781898234351457e-05,
|
218 |
+
"loss": 0.0804,
|
219 |
+
"step": 300
|
220 |
+
},
|
221 |
+
{
|
222 |
+
"epoch": 0.21,
|
223 |
+
"grad_norm": 1.5648704857906242,
|
224 |
+
"learning_rate": 1.7921419853452233e-05,
|
225 |
+
"loss": 0.0834,
|
226 |
+
"step": 310
|
227 |
+
},
|
228 |
+
{
|
229 |
+
"epoch": 0.21,
|
230 |
+
"grad_norm": 2.225837991191409,
|
231 |
+
"learning_rate": 1.8020604823638384e-05,
|
232 |
+
"loss": 0.0646,
|
233 |
+
"step": 320
|
234 |
+
},
|
235 |
+
{
|
236 |
+
"epoch": 0.22,
|
237 |
+
"grad_norm": 4.041622096074851,
|
238 |
+
"learning_rate": 1.811673745928848e-05,
|
239 |
+
"loss": 0.0636,
|
240 |
+
"step": 330
|
241 |
+
},
|
242 |
+
{
|
243 |
+
"epoch": 0.23,
|
244 |
+
"grad_norm": 0.21100008000919218,
|
245 |
+
"learning_rate": 1.821000003120973e-05,
|
246 |
+
"loss": 0.0656,
|
247 |
+
"step": 340
|
248 |
+
},
|
249 |
+
{
|
250 |
+
"epoch": 0.23,
|
251 |
+
"grad_norm": 2.2923304434569443,
|
252 |
+
"learning_rate": 1.8300558955927067e-05,
|
253 |
+
"loss": 0.0756,
|
254 |
+
"step": 350
|
255 |
+
},
|
256 |
+
{
|
257 |
+
"epoch": 0.24,
|
258 |
+
"grad_norm": 3.6524729986322013,
|
259 |
+
"learning_rate": 1.83885665827083e-05,
|
260 |
+
"loss": 0.0466,
|
261 |
+
"step": 360
|
262 |
+
},
|
263 |
+
{
|
264 |
+
"epoch": 0.25,
|
265 |
+
"grad_norm": 2.1793660025907284,
|
266 |
+
"learning_rate": 1.847416273569235e-05,
|
267 |
+
"loss": 0.0704,
|
268 |
+
"step": 370
|
269 |
+
},
|
270 |
+
{
|
271 |
+
"epoch": 0.25,
|
272 |
+
"grad_norm": 4.363503091034068,
|
273 |
+
"learning_rate": 1.8557476050321896e-05,
|
274 |
+
"loss": 0.0712,
|
275 |
+
"step": 380
|
276 |
+
},
|
277 |
+
{
|
278 |
+
"epoch": 0.26,
|
279 |
+
"grad_norm": 3.0942807753734476,
|
280 |
+
"learning_rate": 1.863862513615056e-05,
|
281 |
+
"loss": 0.0831,
|
282 |
+
"step": 390
|
283 |
+
},
|
284 |
+
{
|
285 |
+
"epoch": 0.27,
|
286 |
+
"grad_norm": 1.789230871814937,
|
287 |
+
"learning_rate": 1.8717719592408857e-05,
|
288 |
+
"loss": 0.078,
|
289 |
+
"step": 400
|
290 |
+
},
|
291 |
+
{
|
292 |
+
"epoch": 0.27,
|
293 |
+
"grad_norm": 3.4361028432853677,
|
294 |
+
"learning_rate": 1.879486089815082e-05,
|
295 |
+
"loss": 0.0663,
|
296 |
+
"step": 410
|
297 |
+
},
|
298 |
+
{
|
299 |
+
"epoch": 0.28,
|
300 |
+
"grad_norm": 3.690439378170989,
|
301 |
+
"learning_rate": 1.8870143195120794e-05,
|
302 |
+
"loss": 0.0738,
|
303 |
+
"step": 420
|
304 |
+
},
|
305 |
+
{
|
306 |
+
"epoch": 0.29,
|
307 |
+
"grad_norm": 2.159404934400987,
|
308 |
+
"learning_rate": 1.8943653978491198e-05,
|
309 |
+
"loss": 0.0768,
|
310 |
+
"step": 430
|
311 |
+
},
|
312 |
+
{
|
313 |
+
"epoch": 0.29,
|
314 |
+
"grad_norm": 0.8042070009727055,
|
315 |
+
"learning_rate": 1.901547470818277e-05,
|
316 |
+
"loss": 0.0777,
|
317 |
+
"step": 440
|
318 |
+
},
|
319 |
+
{
|
320 |
+
"epoch": 0.3,
|
321 |
+
"grad_norm": 1.70758347995601,
|
322 |
+
"learning_rate": 1.9085681351478775e-05,
|
323 |
+
"loss": 0.05,
|
324 |
+
"step": 450
|
325 |
+
},
|
326 |
+
{
|
327 |
+
"epoch": 0.31,
|
328 |
+
"grad_norm": 1.2113560599237037,
|
329 |
+
"learning_rate": 1.9154344865995993e-05,
|
330 |
+
"loss": 0.0715,
|
331 |
+
"step": 460
|
332 |
+
},
|
333 |
+
{
|
334 |
+
"epoch": 0.31,
|
335 |
+
"grad_norm": 2.5816071085486527,
|
336 |
+
"learning_rate": 1.9221531630710657e-05,
|
337 |
+
"loss": 0.0688,
|
338 |
+
"step": 470
|
339 |
+
},
|
340 |
+
{
|
341 |
+
"epoch": 0.32,
|
342 |
+
"grad_norm": 1.4447841193953872,
|
343 |
+
"learning_rate": 1.9287303831602588e-05,
|
344 |
+
"loss": 0.0659,
|
345 |
+
"step": 480
|
346 |
+
},
|
347 |
+
{
|
348 |
+
"epoch": 0.33,
|
349 |
+
"grad_norm": 1.2668557011358101,
|
350 |
+
"learning_rate": 1.9351719807533285e-05,
|
351 |
+
"loss": 0.0515,
|
352 |
+
"step": 490
|
353 |
+
},
|
354 |
+
{
|
355 |
+
"epoch": 0.33,
|
356 |
+
"grad_norm": 2.0353893651728865,
|
357 |
+
"learning_rate": 1.9414834361179333e-05,
|
358 |
+
"loss": 0.0687,
|
359 |
+
"step": 500
|
360 |
+
},
|
361 |
+
{
|
362 |
+
"epoch": 0.34,
|
363 |
+
"grad_norm": 2.005863891715338,
|
364 |
+
"learning_rate": 1.947669903917393e-05,
|
365 |
+
"loss": 0.0459,
|
366 |
+
"step": 510
|
367 |
+
},
|
368 |
+
{
|
369 |
+
"epoch": 0.35,
|
370 |
+
"grad_norm": 1.4330303529877615,
|
371 |
+
"learning_rate": 1.9537362385044847e-05,
|
372 |
+
"loss": 0.0557,
|
373 |
+
"step": 520
|
374 |
+
},
|
375 |
+
{
|
376 |
+
"epoch": 0.35,
|
377 |
+
"grad_norm": 1.8683959370360874,
|
378 |
+
"learning_rate": 1.959687016805845e-05,
|
379 |
+
"loss": 0.0656,
|
380 |
+
"step": 530
|
381 |
+
},
|
382 |
+
{
|
383 |
+
"epoch": 0.36,
|
384 |
+
"grad_norm": 1.5963373312962135,
|
385 |
+
"learning_rate": 1.9655265590672502e-05,
|
386 |
+
"loss": 0.0519,
|
387 |
+
"step": 540
|
388 |
+
},
|
389 |
+
{
|
390 |
+
"epoch": 0.37,
|
391 |
+
"grad_norm": 1.824110595633705,
|
392 |
+
"learning_rate": 1.9712589476953243e-05,
|
393 |
+
"loss": 0.0557,
|
394 |
+
"step": 550
|
395 |
+
},
|
396 |
+
{
|
397 |
+
"epoch": 0.37,
|
398 |
+
"grad_norm": 2.024299349665376,
|
399 |
+
"learning_rate": 1.976888044401508e-05,
|
400 |
+
"loss": 0.0617,
|
401 |
+
"step": 560
|
402 |
+
},
|
403 |
+
{
|
404 |
+
"epoch": 0.38,
|
405 |
+
"grad_norm": 1.0490238284158413,
|
406 |
+
"learning_rate": 1.98241750582861e-05,
|
407 |
+
"loss": 0.0529,
|
408 |
+
"step": 570
|
409 |
+
},
|
410 |
+
{
|
411 |
+
"epoch": 0.39,
|
412 |
+
"grad_norm": 1.2602675746455845,
|
413 |
+
"learning_rate": 1.9878507978183157e-05,
|
414 |
+
"loss": 0.0617,
|
415 |
+
"step": 580
|
416 |
+
},
|
417 |
+
{
|
418 |
+
"epoch": 0.39,
|
419 |
+
"grad_norm": 1.7353304044327778,
|
420 |
+
"learning_rate": 1.9931912084590654e-05,
|
421 |
+
"loss": 0.0464,
|
422 |
+
"step": 590
|
423 |
+
},
|
424 |
+
{
|
425 |
+
"epoch": 0.4,
|
426 |
+
"grad_norm": 2.043608410331547,
|
427 |
+
"learning_rate": 1.998441860037306e-05,
|
428 |
+
"loss": 0.0716,
|
429 |
+
"step": 600
|
430 |
+
},
|
431 |
+
{
|
432 |
+
"epoch": 0.41,
|
433 |
+
"grad_norm": 0.9522735357022469,
|
434 |
+
"learning_rate": 1.997786386275595e-05,
|
435 |
+
"loss": 0.0496,
|
436 |
+
"step": 610
|
437 |
+
},
|
438 |
+
{
|
439 |
+
"epoch": 0.41,
|
440 |
+
"grad_norm": 1.3740017908881452,
|
441 |
+
"learning_rate": 1.9940970300682533e-05,
|
442 |
+
"loss": 0.0686,
|
443 |
+
"step": 620
|
444 |
+
},
|
445 |
+
{
|
446 |
+
"epoch": 0.42,
|
447 |
+
"grad_norm": 1.1533525092583308,
|
448 |
+
"learning_rate": 1.9904076738609114e-05,
|
449 |
+
"loss": 0.0562,
|
450 |
+
"step": 630
|
451 |
+
},
|
452 |
+
{
|
453 |
+
"epoch": 0.42,
|
454 |
+
"grad_norm": 1.564671647601224,
|
455 |
+
"learning_rate": 1.9867183176535695e-05,
|
456 |
+
"loss": 0.0553,
|
457 |
+
"step": 640
|
458 |
+
},
|
459 |
+
{
|
460 |
+
"epoch": 0.43,
|
461 |
+
"grad_norm": 0.8996693997352794,
|
462 |
+
"learning_rate": 1.9830289614462276e-05,
|
463 |
+
"loss": 0.0524,
|
464 |
+
"step": 650
|
465 |
+
},
|
466 |
+
{
|
467 |
+
"epoch": 0.44,
|
468 |
+
"grad_norm": 1.9340857910489522,
|
469 |
+
"learning_rate": 1.979339605238886e-05,
|
470 |
+
"loss": 0.0534,
|
471 |
+
"step": 660
|
472 |
+
},
|
473 |
+
{
|
474 |
+
"epoch": 0.44,
|
475 |
+
"grad_norm": 1.892682744806536,
|
476 |
+
"learning_rate": 1.9756502490315442e-05,
|
477 |
+
"loss": 0.0625,
|
478 |
+
"step": 670
|
479 |
+
},
|
480 |
+
{
|
481 |
+
"epoch": 0.45,
|
482 |
+
"grad_norm": 1.2063853192894494,
|
483 |
+
"learning_rate": 1.9719608928242023e-05,
|
484 |
+
"loss": 0.0519,
|
485 |
+
"step": 680
|
486 |
+
},
|
487 |
+
{
|
488 |
+
"epoch": 0.46,
|
489 |
+
"grad_norm": 1.984312822464147,
|
490 |
+
"learning_rate": 1.9682715366168604e-05,
|
491 |
+
"loss": 0.0494,
|
492 |
+
"step": 690
|
493 |
+
},
|
494 |
+
{
|
495 |
+
"epoch": 0.46,
|
496 |
+
"grad_norm": 1.6705446030280595,
|
497 |
+
"learning_rate": 1.9645821804095185e-05,
|
498 |
+
"loss": 0.059,
|
499 |
+
"step": 700
|
500 |
+
},
|
501 |
+
{
|
502 |
+
"epoch": 0.47,
|
503 |
+
"grad_norm": 1.7422620817223426,
|
504 |
+
"learning_rate": 1.960892824202177e-05,
|
505 |
+
"loss": 0.0454,
|
506 |
+
"step": 710
|
507 |
+
},
|
508 |
+
{
|
509 |
+
"epoch": 0.48,
|
510 |
+
"grad_norm": 1.3182961229762868,
|
511 |
+
"learning_rate": 1.957203467994835e-05,
|
512 |
+
"loss": 0.0581,
|
513 |
+
"step": 720
|
514 |
+
},
|
515 |
+
{
|
516 |
+
"epoch": 0.48,
|
517 |
+
"grad_norm": 1.330360819481426,
|
518 |
+
"learning_rate": 1.9535141117874932e-05,
|
519 |
+
"loss": 0.0575,
|
520 |
+
"step": 730
|
521 |
+
},
|
522 |
+
{
|
523 |
+
"epoch": 0.49,
|
524 |
+
"grad_norm": 1.9840741868184866,
|
525 |
+
"learning_rate": 1.9498247555801517e-05,
|
526 |
+
"loss": 0.063,
|
527 |
+
"step": 740
|
528 |
+
},
|
529 |
+
{
|
530 |
+
"epoch": 0.5,
|
531 |
+
"grad_norm": 1.6018064731760029,
|
532 |
+
"learning_rate": 1.9461353993728094e-05,
|
533 |
+
"loss": 0.0519,
|
534 |
+
"step": 750
|
535 |
+
},
|
536 |
+
{
|
537 |
+
"epoch": 0.5,
|
538 |
+
"grad_norm": 1.391561342963203,
|
539 |
+
"learning_rate": 1.9424460431654675e-05,
|
540 |
+
"loss": 0.0572,
|
541 |
+
"step": 760
|
542 |
+
},
|
543 |
+
{
|
544 |
+
"epoch": 0.51,
|
545 |
+
"grad_norm": 1.0319629863193043,
|
546 |
+
"learning_rate": 1.938756686958126e-05,
|
547 |
+
"loss": 0.0619,
|
548 |
+
"step": 770
|
549 |
+
},
|
550 |
+
{
|
551 |
+
"epoch": 0.52,
|
552 |
+
"grad_norm": 1.7398111190442345,
|
553 |
+
"learning_rate": 1.935067330750784e-05,
|
554 |
+
"loss": 0.05,
|
555 |
+
"step": 780
|
556 |
+
},
|
557 |
+
{
|
558 |
+
"epoch": 0.52,
|
559 |
+
"grad_norm": 1.675180749962208,
|
560 |
+
"learning_rate": 1.9313779745434422e-05,
|
561 |
+
"loss": 0.0654,
|
562 |
+
"step": 790
|
563 |
+
},
|
564 |
+
{
|
565 |
+
"epoch": 0.53,
|
566 |
+
"grad_norm": 1.486943722740635,
|
567 |
+
"learning_rate": 1.9276886183361007e-05,
|
568 |
+
"loss": 0.0463,
|
569 |
+
"step": 800
|
570 |
+
},
|
571 |
+
{
|
572 |
+
"epoch": 0.54,
|
573 |
+
"grad_norm": 1.1444156835752686,
|
574 |
+
"learning_rate": 1.9239992621287588e-05,
|
575 |
+
"loss": 0.0598,
|
576 |
+
"step": 810
|
577 |
+
},
|
578 |
+
{
|
579 |
+
"epoch": 0.54,
|
580 |
+
"grad_norm": 1.1260777714975718,
|
581 |
+
"learning_rate": 1.920309905921417e-05,
|
582 |
+
"loss": 0.039,
|
583 |
+
"step": 820
|
584 |
+
},
|
585 |
+
{
|
586 |
+
"epoch": 0.55,
|
587 |
+
"grad_norm": 1.328313971146618,
|
588 |
+
"learning_rate": 1.916620549714075e-05,
|
589 |
+
"loss": 0.0591,
|
590 |
+
"step": 830
|
591 |
+
},
|
592 |
+
{
|
593 |
+
"epoch": 0.56,
|
594 |
+
"grad_norm": 0.8805517771854091,
|
595 |
+
"learning_rate": 1.912931193506733e-05,
|
596 |
+
"loss": 0.0392,
|
597 |
+
"step": 840
|
598 |
+
},
|
599 |
+
{
|
600 |
+
"epoch": 0.56,
|
601 |
+
"grad_norm": 0.9704226543415952,
|
602 |
+
"learning_rate": 1.9092418372993916e-05,
|
603 |
+
"loss": 0.056,
|
604 |
+
"step": 850
|
605 |
+
},
|
606 |
+
{
|
607 |
+
"epoch": 0.57,
|
608 |
+
"grad_norm": 1.4328262810995938,
|
609 |
+
"learning_rate": 1.9055524810920497e-05,
|
610 |
+
"loss": 0.0602,
|
611 |
+
"step": 860
|
612 |
+
},
|
613 |
+
{
|
614 |
+
"epoch": 0.58,
|
615 |
+
"grad_norm": 2.6133255890647167,
|
616 |
+
"learning_rate": 1.9018631248847078e-05,
|
617 |
+
"loss": 0.0507,
|
618 |
+
"step": 870
|
619 |
+
},
|
620 |
+
{
|
621 |
+
"epoch": 0.58,
|
622 |
+
"grad_norm": 1.4574352859865667,
|
623 |
+
"learning_rate": 1.898173768677366e-05,
|
624 |
+
"loss": 0.0667,
|
625 |
+
"step": 880
|
626 |
+
},
|
627 |
+
{
|
628 |
+
"epoch": 0.59,
|
629 |
+
"grad_norm": 1.9116869451062006,
|
630 |
+
"learning_rate": 1.894484412470024e-05,
|
631 |
+
"loss": 0.0533,
|
632 |
+
"step": 890
|
633 |
+
},
|
634 |
+
{
|
635 |
+
"epoch": 0.6,
|
636 |
+
"grad_norm": 0.8152499611885836,
|
637 |
+
"learning_rate": 1.890795056262682e-05,
|
638 |
+
"loss": 0.0534,
|
639 |
+
"step": 900
|
640 |
+
},
|
641 |
+
{
|
642 |
+
"epoch": 0.6,
|
643 |
+
"grad_norm": 2.9440345318883696,
|
644 |
+
"learning_rate": 1.8871057000553406e-05,
|
645 |
+
"loss": 0.0419,
|
646 |
+
"step": 910
|
647 |
+
},
|
648 |
+
{
|
649 |
+
"epoch": 0.61,
|
650 |
+
"grad_norm": 2.7588896552457074,
|
651 |
+
"learning_rate": 1.8834163438479987e-05,
|
652 |
+
"loss": 0.0418,
|
653 |
+
"step": 920
|
654 |
+
},
|
655 |
+
{
|
656 |
+
"epoch": 0.62,
|
657 |
+
"grad_norm": 1.8886319208504987,
|
658 |
+
"learning_rate": 1.8797269876406568e-05,
|
659 |
+
"loss": 0.0397,
|
660 |
+
"step": 930
|
661 |
+
},
|
662 |
+
{
|
663 |
+
"epoch": 0.62,
|
664 |
+
"grad_norm": 1.8820615671771377,
|
665 |
+
"learning_rate": 1.876037631433315e-05,
|
666 |
+
"loss": 0.0454,
|
667 |
+
"step": 940
|
668 |
+
},
|
669 |
+
{
|
670 |
+
"epoch": 0.63,
|
671 |
+
"grad_norm": 1.2868214865586614,
|
672 |
+
"learning_rate": 1.872348275225973e-05,
|
673 |
+
"loss": 0.0373,
|
674 |
+
"step": 950
|
675 |
+
},
|
676 |
+
{
|
677 |
+
"epoch": 0.64,
|
678 |
+
"grad_norm": 2.016876778284122,
|
679 |
+
"learning_rate": 1.8686589190186315e-05,
|
680 |
+
"loss": 0.0619,
|
681 |
+
"step": 960
|
682 |
+
},
|
683 |
+
{
|
684 |
+
"epoch": 0.64,
|
685 |
+
"grad_norm": 0.8992764743601639,
|
686 |
+
"learning_rate": 1.8649695628112896e-05,
|
687 |
+
"loss": 0.0403,
|
688 |
+
"step": 970
|
689 |
+
},
|
690 |
+
{
|
691 |
+
"epoch": 0.65,
|
692 |
+
"grad_norm": 0.42416507318131186,
|
693 |
+
"learning_rate": 1.8612802066039477e-05,
|
694 |
+
"loss": 0.067,
|
695 |
+
"step": 980
|
696 |
+
},
|
697 |
+
{
|
698 |
+
"epoch": 0.66,
|
699 |
+
"grad_norm": 1.8601423331120188,
|
700 |
+
"learning_rate": 1.8575908503966062e-05,
|
701 |
+
"loss": 0.066,
|
702 |
+
"step": 990
|
703 |
+
},
|
704 |
+
{
|
705 |
+
"epoch": 0.66,
|
706 |
+
"grad_norm": 1.3893250379275068,
|
707 |
+
"learning_rate": 1.8539014941892643e-05,
|
708 |
+
"loss": 0.0437,
|
709 |
+
"step": 1000
|
710 |
+
},
|
711 |
+
{
|
712 |
+
"epoch": 0.67,
|
713 |
+
"grad_norm": 2.3240219930504376,
|
714 |
+
"learning_rate": 1.850212137981922e-05,
|
715 |
+
"loss": 0.0488,
|
716 |
+
"step": 1010
|
717 |
+
},
|
718 |
+
{
|
719 |
+
"epoch": 0.68,
|
720 |
+
"grad_norm": 0.8849896953543195,
|
721 |
+
"learning_rate": 1.8465227817745805e-05,
|
722 |
+
"loss": 0.0472,
|
723 |
+
"step": 1020
|
724 |
+
},
|
725 |
+
{
|
726 |
+
"epoch": 0.68,
|
727 |
+
"grad_norm": 0.8655212477880712,
|
728 |
+
"learning_rate": 1.8428334255672386e-05,
|
729 |
+
"loss": 0.0546,
|
730 |
+
"step": 1030
|
731 |
+
},
|
732 |
+
{
|
733 |
+
"epoch": 0.69,
|
734 |
+
"grad_norm": 1.2016665548799828,
|
735 |
+
"learning_rate": 1.8391440693598967e-05,
|
736 |
+
"loss": 0.0956,
|
737 |
+
"step": 1040
|
738 |
+
},
|
739 |
+
{
|
740 |
+
"epoch": 0.7,
|
741 |
+
"grad_norm": 1.492659569010543,
|
742 |
+
"learning_rate": 1.8354547131525552e-05,
|
743 |
+
"loss": 0.0449,
|
744 |
+
"step": 1050
|
745 |
+
},
|
746 |
+
{
|
747 |
+
"epoch": 0.7,
|
748 |
+
"grad_norm": 0.4224837643985645,
|
749 |
+
"learning_rate": 1.8317653569452133e-05,
|
750 |
+
"loss": 0.0545,
|
751 |
+
"step": 1060
|
752 |
+
},
|
753 |
+
{
|
754 |
+
"epoch": 0.71,
|
755 |
+
"grad_norm": 1.8405573730480247,
|
756 |
+
"learning_rate": 1.8280760007378714e-05,
|
757 |
+
"loss": 0.0328,
|
758 |
+
"step": 1070
|
759 |
+
},
|
760 |
+
{
|
761 |
+
"epoch": 0.72,
|
762 |
+
"grad_norm": 0.3988661237020132,
|
763 |
+
"learning_rate": 1.8243866445305295e-05,
|
764 |
+
"loss": 0.0451,
|
765 |
+
"step": 1080
|
766 |
+
},
|
767 |
+
{
|
768 |
+
"epoch": 0.72,
|
769 |
+
"grad_norm": 0.5573396543240564,
|
770 |
+
"learning_rate": 1.8206972883231876e-05,
|
771 |
+
"loss": 0.0502,
|
772 |
+
"step": 1090
|
773 |
+
},
|
774 |
+
{
|
775 |
+
"epoch": 0.73,
|
776 |
+
"grad_norm": 0.7855566721285819,
|
777 |
+
"learning_rate": 1.817007932115846e-05,
|
778 |
+
"loss": 0.0341,
|
779 |
+
"step": 1100
|
780 |
+
},
|
781 |
+
{
|
782 |
+
"epoch": 0.74,
|
783 |
+
"grad_norm": 1.497005175064917,
|
784 |
+
"learning_rate": 1.8133185759085042e-05,
|
785 |
+
"loss": 0.0513,
|
786 |
+
"step": 1110
|
787 |
+
},
|
788 |
+
{
|
789 |
+
"epoch": 0.74,
|
790 |
+
"grad_norm": 1.650975953086994,
|
791 |
+
"learning_rate": 1.8096292197011623e-05,
|
792 |
+
"loss": 0.0473,
|
793 |
+
"step": 1120
|
794 |
+
},
|
795 |
+
{
|
796 |
+
"epoch": 0.75,
|
797 |
+
"grad_norm": 0.8418054071334755,
|
798 |
+
"learning_rate": 1.8059398634938204e-05,
|
799 |
+
"loss": 0.0444,
|
800 |
+
"step": 1130
|
801 |
+
},
|
802 |
+
{
|
803 |
+
"epoch": 0.76,
|
804 |
+
"grad_norm": 1.33651471077315,
|
805 |
+
"learning_rate": 1.8022505072864785e-05,
|
806 |
+
"loss": 0.0476,
|
807 |
+
"step": 1140
|
808 |
+
},
|
809 |
+
{
|
810 |
+
"epoch": 0.76,
|
811 |
+
"grad_norm": 1.1529220423121023,
|
812 |
+
"learning_rate": 1.7985611510791367e-05,
|
813 |
+
"loss": 0.0379,
|
814 |
+
"step": 1150
|
815 |
+
},
|
816 |
+
{
|
817 |
+
"epoch": 0.77,
|
818 |
+
"grad_norm": 0.6912013035706749,
|
819 |
+
"learning_rate": 1.794871794871795e-05,
|
820 |
+
"loss": 0.0526,
|
821 |
+
"step": 1160
|
822 |
+
},
|
823 |
+
{
|
824 |
+
"epoch": 0.78,
|
825 |
+
"grad_norm": 2.806793867054328,
|
826 |
+
"learning_rate": 1.7911824386644532e-05,
|
827 |
+
"loss": 0.047,
|
828 |
+
"step": 1170
|
829 |
+
},
|
830 |
+
{
|
831 |
+
"epoch": 0.78,
|
832 |
+
"grad_norm": 3.323409487413899,
|
833 |
+
"learning_rate": 1.7874930824571113e-05,
|
834 |
+
"loss": 0.0455,
|
835 |
+
"step": 1180
|
836 |
+
},
|
837 |
+
{
|
838 |
+
"epoch": 0.79,
|
839 |
+
"grad_norm": 1.0877783561592873,
|
840 |
+
"learning_rate": 1.7838037262497695e-05,
|
841 |
+
"loss": 0.0555,
|
842 |
+
"step": 1190
|
843 |
+
},
|
844 |
+
{
|
845 |
+
"epoch": 0.8,
|
846 |
+
"grad_norm": 1.9158165193439103,
|
847 |
+
"learning_rate": 1.7801143700424276e-05,
|
848 |
+
"loss": 0.0458,
|
849 |
+
"step": 1200
|
850 |
+
},
|
851 |
+
{
|
852 |
+
"epoch": 0.8,
|
853 |
+
"grad_norm": 1.0863638939536142,
|
854 |
+
"learning_rate": 1.776425013835086e-05,
|
855 |
+
"loss": 0.0462,
|
856 |
+
"step": 1210
|
857 |
+
},
|
858 |
+
{
|
859 |
+
"epoch": 0.81,
|
860 |
+
"grad_norm": 2.3398031954044978,
|
861 |
+
"learning_rate": 1.772735657627744e-05,
|
862 |
+
"loss": 0.032,
|
863 |
+
"step": 1220
|
864 |
+
},
|
865 |
+
{
|
866 |
+
"epoch": 0.82,
|
867 |
+
"grad_norm": 0.9272917966743286,
|
868 |
+
"learning_rate": 1.7690463014204022e-05,
|
869 |
+
"loss": 0.0464,
|
870 |
+
"step": 1230
|
871 |
+
},
|
872 |
+
{
|
873 |
+
"epoch": 0.82,
|
874 |
+
"grad_norm": 1.4608044079217914,
|
875 |
+
"learning_rate": 1.7653569452130607e-05,
|
876 |
+
"loss": 0.0544,
|
877 |
+
"step": 1240
|
878 |
+
},
|
879 |
+
{
|
880 |
+
"epoch": 0.83,
|
881 |
+
"grad_norm": 2.0799767984416713,
|
882 |
+
"learning_rate": 1.7616675890057188e-05,
|
883 |
+
"loss": 0.0629,
|
884 |
+
"step": 1250
|
885 |
+
},
|
886 |
+
{
|
887 |
+
"epoch": 0.84,
|
888 |
+
"grad_norm": 2.438242481318271,
|
889 |
+
"learning_rate": 1.757978232798377e-05,
|
890 |
+
"loss": 0.0586,
|
891 |
+
"step": 1260
|
892 |
+
},
|
893 |
+
{
|
894 |
+
"epoch": 0.84,
|
895 |
+
"grad_norm": 0.22666963149041416,
|
896 |
+
"learning_rate": 1.754288876591035e-05,
|
897 |
+
"loss": 0.0419,
|
898 |
+
"step": 1270
|
899 |
+
},
|
900 |
+
{
|
901 |
+
"epoch": 0.85,
|
902 |
+
"grad_norm": 1.6906855856145777,
|
903 |
+
"learning_rate": 1.750599520383693e-05,
|
904 |
+
"loss": 0.0417,
|
905 |
+
"step": 1280
|
906 |
+
},
|
907 |
+
{
|
908 |
+
"epoch": 0.86,
|
909 |
+
"grad_norm": 0.8417294291584847,
|
910 |
+
"learning_rate": 1.7469101641763513e-05,
|
911 |
+
"loss": 0.0496,
|
912 |
+
"step": 1290
|
913 |
+
},
|
914 |
+
{
|
915 |
+
"epoch": 0.86,
|
916 |
+
"grad_norm": 2.384648750559231,
|
917 |
+
"learning_rate": 1.7432208079690097e-05,
|
918 |
+
"loss": 0.037,
|
919 |
+
"step": 1300
|
920 |
+
},
|
921 |
+
{
|
922 |
+
"epoch": 0.87,
|
923 |
+
"grad_norm": 1.0969098360012755,
|
924 |
+
"learning_rate": 1.7395314517616678e-05,
|
925 |
+
"loss": 0.0392,
|
926 |
+
"step": 1310
|
927 |
+
},
|
928 |
+
{
|
929 |
+
"epoch": 0.88,
|
930 |
+
"grad_norm": 4.265947825753472,
|
931 |
+
"learning_rate": 1.735842095554326e-05,
|
932 |
+
"loss": 0.0434,
|
933 |
+
"step": 1320
|
934 |
+
},
|
935 |
+
{
|
936 |
+
"epoch": 0.88,
|
937 |
+
"grad_norm": 1.0043877066833349,
|
938 |
+
"learning_rate": 1.732152739346984e-05,
|
939 |
+
"loss": 0.0423,
|
940 |
+
"step": 1330
|
941 |
+
},
|
942 |
+
{
|
943 |
+
"epoch": 0.89,
|
944 |
+
"grad_norm": 0.3834891081373628,
|
945 |
+
"learning_rate": 1.728463383139642e-05,
|
946 |
+
"loss": 0.0423,
|
947 |
+
"step": 1340
|
948 |
+
},
|
949 |
+
{
|
950 |
+
"epoch": 0.9,
|
951 |
+
"grad_norm": 0.8917610552257311,
|
952 |
+
"learning_rate": 1.7247740269323006e-05,
|
953 |
+
"loss": 0.0491,
|
954 |
+
"step": 1350
|
955 |
+
},
|
956 |
+
{
|
957 |
+
"epoch": 0.9,
|
958 |
+
"grad_norm": 1.7777935423085822,
|
959 |
+
"learning_rate": 1.7210846707249587e-05,
|
960 |
+
"loss": 0.0465,
|
961 |
+
"step": 1360
|
962 |
+
},
|
963 |
+
{
|
964 |
+
"epoch": 0.91,
|
965 |
+
"grad_norm": 2.882686457165806,
|
966 |
+
"learning_rate": 1.717395314517617e-05,
|
967 |
+
"loss": 0.0366,
|
968 |
+
"step": 1370
|
969 |
+
},
|
970 |
+
{
|
971 |
+
"epoch": 0.92,
|
972 |
+
"grad_norm": 1.4070944899475468,
|
973 |
+
"learning_rate": 1.713705958310275e-05,
|
974 |
+
"loss": 0.0364,
|
975 |
+
"step": 1380
|
976 |
+
},
|
977 |
+
{
|
978 |
+
"epoch": 0.92,
|
979 |
+
"grad_norm": 1.368387839772382,
|
980 |
+
"learning_rate": 1.710016602102933e-05,
|
981 |
+
"loss": 0.0302,
|
982 |
+
"step": 1390
|
983 |
+
},
|
984 |
+
{
|
985 |
+
"epoch": 0.93,
|
986 |
+
"grad_norm": 1.2202288141019488,
|
987 |
+
"learning_rate": 1.7063272458955912e-05,
|
988 |
+
"loss": 0.0455,
|
989 |
+
"step": 1400
|
990 |
+
},
|
991 |
+
{
|
992 |
+
"epoch": 0.94,
|
993 |
+
"grad_norm": 1.7419694841079256,
|
994 |
+
"learning_rate": 1.7026378896882496e-05,
|
995 |
+
"loss": 0.0402,
|
996 |
+
"step": 1410
|
997 |
+
},
|
998 |
+
{
|
999 |
+
"epoch": 0.94,
|
1000 |
+
"grad_norm": 1.5792094366611027,
|
1001 |
+
"learning_rate": 1.6989485334809077e-05,
|
1002 |
+
"loss": 0.0364,
|
1003 |
+
"step": 1420
|
1004 |
+
},
|
1005 |
+
{
|
1006 |
+
"epoch": 0.95,
|
1007 |
+
"grad_norm": 1.0092569099059323,
|
1008 |
+
"learning_rate": 1.695259177273566e-05,
|
1009 |
+
"loss": 0.0335,
|
1010 |
+
"step": 1430
|
1011 |
+
},
|
1012 |
+
{
|
1013 |
+
"epoch": 0.96,
|
1014 |
+
"grad_norm": 1.5374731387141227,
|
1015 |
+
"learning_rate": 1.691569821066224e-05,
|
1016 |
+
"loss": 0.0428,
|
1017 |
+
"step": 1440
|
1018 |
+
},
|
1019 |
+
{
|
1020 |
+
"epoch": 0.96,
|
1021 |
+
"grad_norm": 1.6268956058587998,
|
1022 |
+
"learning_rate": 1.687880464858882e-05,
|
1023 |
+
"loss": 0.0483,
|
1024 |
+
"step": 1450
|
1025 |
+
},
|
1026 |
+
{
|
1027 |
+
"epoch": 0.97,
|
1028 |
+
"grad_norm": 1.6628348966888566,
|
1029 |
+
"learning_rate": 1.6841911086515402e-05,
|
1030 |
+
"loss": 0.0487,
|
1031 |
+
"step": 1460
|
1032 |
+
},
|
1033 |
+
{
|
1034 |
+
"epoch": 0.98,
|
1035 |
+
"grad_norm": 8.037103442054311,
|
1036 |
+
"learning_rate": 1.6805017524441987e-05,
|
1037 |
+
"loss": 0.0679,
|
1038 |
+
"step": 1470
|
1039 |
+
},
|
1040 |
+
{
|
1041 |
+
"epoch": 0.98,
|
1042 |
+
"grad_norm": 1.620270154467724,
|
1043 |
+
"learning_rate": 1.6768123962368568e-05,
|
1044 |
+
"loss": 0.0437,
|
1045 |
+
"step": 1480
|
1046 |
+
},
|
1047 |
+
{
|
1048 |
+
"epoch": 0.99,
|
1049 |
+
"grad_norm": 1.0436875513091504,
|
1050 |
+
"learning_rate": 1.6731230400295152e-05,
|
1051 |
+
"loss": 0.0346,
|
1052 |
+
"step": 1490
|
1053 |
+
},
|
1054 |
+
{
|
1055 |
+
"epoch": 1.0,
|
1056 |
+
"grad_norm": 0.8374706362800531,
|
1057 |
+
"learning_rate": 1.6694336838221733e-05,
|
1058 |
+
"loss": 0.0517,
|
1059 |
+
"step": 1500
|
1060 |
+
},
|
1061 |
+
{
|
1062 |
+
"epoch": 1.0,
|
1063 |
+
"grad_norm": 1.1939983249441695,
|
1064 |
+
"learning_rate": 1.6657443276148314e-05,
|
1065 |
+
"loss": 0.0393,
|
1066 |
+
"step": 1510
|
1067 |
+
},
|
1068 |
+
{
|
1069 |
+
"epoch": 1.01,
|
1070 |
+
"grad_norm": 1.0650702970882078,
|
1071 |
+
"learning_rate": 1.6620549714074896e-05,
|
1072 |
+
"loss": 0.0433,
|
1073 |
+
"step": 1520
|
1074 |
+
},
|
1075 |
+
{
|
1076 |
+
"epoch": 1.02,
|
1077 |
+
"grad_norm": 0.9588706576483126,
|
1078 |
+
"learning_rate": 1.6583656152001477e-05,
|
1079 |
+
"loss": 0.0381,
|
1080 |
+
"step": 1530
|
1081 |
+
},
|
1082 |
+
{
|
1083 |
+
"epoch": 1.02,
|
1084 |
+
"grad_norm": 1.4412233810414918,
|
1085 |
+
"learning_rate": 1.6546762589928058e-05,
|
1086 |
+
"loss": 0.0346,
|
1087 |
+
"step": 1540
|
1088 |
+
},
|
1089 |
+
{
|
1090 |
+
"epoch": 1.03,
|
1091 |
+
"grad_norm": 0.9533871774762268,
|
1092 |
+
"learning_rate": 1.6509869027854642e-05,
|
1093 |
+
"loss": 0.0239,
|
1094 |
+
"step": 1550
|
1095 |
+
},
|
1096 |
+
{
|
1097 |
+
"epoch": 1.04,
|
1098 |
+
"grad_norm": 0.44748765523495393,
|
1099 |
+
"learning_rate": 1.6472975465781223e-05,
|
1100 |
+
"loss": 0.0424,
|
1101 |
+
"step": 1560
|
1102 |
+
},
|
1103 |
+
{
|
1104 |
+
"epoch": 1.04,
|
1105 |
+
"grad_norm": 0.6542278196418172,
|
1106 |
+
"learning_rate": 1.6436081903707805e-05,
|
1107 |
+
"loss": 0.0322,
|
1108 |
+
"step": 1570
|
1109 |
+
},
|
1110 |
+
{
|
1111 |
+
"epoch": 1.05,
|
1112 |
+
"grad_norm": 0.774821420829752,
|
1113 |
+
"learning_rate": 1.6399188341634386e-05,
|
1114 |
+
"loss": 0.0337,
|
1115 |
+
"step": 1580
|
1116 |
+
},
|
1117 |
+
{
|
1118 |
+
"epoch": 1.06,
|
1119 |
+
"grad_norm": 3.4366984039627053,
|
1120 |
+
"learning_rate": 1.6362294779560967e-05,
|
1121 |
+
"loss": 0.0402,
|
1122 |
+
"step": 1590
|
1123 |
+
},
|
1124 |
+
{
|
1125 |
+
"epoch": 1.06,
|
1126 |
+
"grad_norm": 0.9409785268524657,
|
1127 |
+
"learning_rate": 1.6325401217487548e-05,
|
1128 |
+
"loss": 0.0351,
|
1129 |
+
"step": 1600
|
1130 |
+
},
|
1131 |
+
{
|
1132 |
+
"epoch": 1.07,
|
1133 |
+
"grad_norm": 1.3755255908395096,
|
1134 |
+
"learning_rate": 1.6288507655414133e-05,
|
1135 |
+
"loss": 0.0287,
|
1136 |
+
"step": 1610
|
1137 |
+
},
|
1138 |
+
{
|
1139 |
+
"epoch": 1.08,
|
1140 |
+
"grad_norm": 0.9105883051718862,
|
1141 |
+
"learning_rate": 1.6251614093340714e-05,
|
1142 |
+
"loss": 0.0377,
|
1143 |
+
"step": 1620
|
1144 |
+
},
|
1145 |
+
{
|
1146 |
+
"epoch": 1.08,
|
1147 |
+
"grad_norm": 0.669451044858242,
|
1148 |
+
"learning_rate": 1.6214720531267295e-05,
|
1149 |
+
"loss": 0.0397,
|
1150 |
+
"step": 1630
|
1151 |
+
},
|
1152 |
+
{
|
1153 |
+
"epoch": 1.09,
|
1154 |
+
"grad_norm": 1.0173498911812129,
|
1155 |
+
"learning_rate": 1.6177826969193876e-05,
|
1156 |
+
"loss": 0.0362,
|
1157 |
+
"step": 1640
|
1158 |
+
},
|
1159 |
+
{
|
1160 |
+
"epoch": 1.1,
|
1161 |
+
"grad_norm": 1.443726695479527,
|
1162 |
+
"learning_rate": 1.6140933407120457e-05,
|
1163 |
+
"loss": 0.0388,
|
1164 |
+
"step": 1650
|
1165 |
+
},
|
1166 |
+
{
|
1167 |
+
"epoch": 1.1,
|
1168 |
+
"grad_norm": 0.48614710000952543,
|
1169 |
+
"learning_rate": 1.610403984504704e-05,
|
1170 |
+
"loss": 0.044,
|
1171 |
+
"step": 1660
|
1172 |
+
},
|
1173 |
+
{
|
1174 |
+
"epoch": 1.11,
|
1175 |
+
"grad_norm": 0.8955875806318787,
|
1176 |
+
"learning_rate": 1.6067146282973623e-05,
|
1177 |
+
"loss": 0.0361,
|
1178 |
+
"step": 1670
|
1179 |
+
},
|
1180 |
+
{
|
1181 |
+
"epoch": 1.12,
|
1182 |
+
"grad_norm": 0.7136011214906065,
|
1183 |
+
"learning_rate": 1.6030252720900204e-05,
|
1184 |
+
"loss": 0.0399,
|
1185 |
+
"step": 1680
|
1186 |
+
},
|
1187 |
+
{
|
1188 |
+
"epoch": 1.12,
|
1189 |
+
"grad_norm": 1.342383821315822,
|
1190 |
+
"learning_rate": 1.599335915882679e-05,
|
1191 |
+
"loss": 0.0398,
|
1192 |
+
"step": 1690
|
1193 |
+
},
|
1194 |
+
{
|
1195 |
+
"epoch": 1.13,
|
1196 |
+
"grad_norm": 1.9154366056765233,
|
1197 |
+
"learning_rate": 1.5956465596753366e-05,
|
1198 |
+
"loss": 0.0351,
|
1199 |
+
"step": 1700
|
1200 |
+
},
|
1201 |
+
{
|
1202 |
+
"epoch": 1.14,
|
1203 |
+
"grad_norm": 3.380571358597728,
|
1204 |
+
"learning_rate": 1.5919572034679947e-05,
|
1205 |
+
"loss": 0.0355,
|
1206 |
+
"step": 1710
|
1207 |
+
},
|
1208 |
+
{
|
1209 |
+
"epoch": 1.14,
|
1210 |
+
"grad_norm": 1.4160947580259224,
|
1211 |
+
"learning_rate": 1.5882678472606532e-05,
|
1212 |
+
"loss": 0.0333,
|
1213 |
+
"step": 1720
|
1214 |
+
},
|
1215 |
+
{
|
1216 |
+
"epoch": 1.15,
|
1217 |
+
"grad_norm": 0.6737638581076882,
|
1218 |
+
"learning_rate": 1.5845784910533113e-05,
|
1219 |
+
"loss": 0.0296,
|
1220 |
+
"step": 1730
|
1221 |
+
},
|
1222 |
+
{
|
1223 |
+
"epoch": 1.16,
|
1224 |
+
"grad_norm": 1.206337590570484,
|
1225 |
+
"learning_rate": 1.5808891348459694e-05,
|
1226 |
+
"loss": 0.0308,
|
1227 |
+
"step": 1740
|
1228 |
+
},
|
1229 |
+
{
|
1230 |
+
"epoch": 1.16,
|
1231 |
+
"grad_norm": 0.7350933133142535,
|
1232 |
+
"learning_rate": 1.577199778638628e-05,
|
1233 |
+
"loss": 0.0335,
|
1234 |
+
"step": 1750
|
1235 |
+
},
|
1236 |
+
{
|
1237 |
+
"epoch": 1.17,
|
1238 |
+
"grad_norm": 1.3603619905771789,
|
1239 |
+
"learning_rate": 1.573510422431286e-05,
|
1240 |
+
"loss": 0.0254,
|
1241 |
+
"step": 1760
|
1242 |
+
},
|
1243 |
+
{
|
1244 |
+
"epoch": 1.18,
|
1245 |
+
"grad_norm": 0.6337146921122443,
|
1246 |
+
"learning_rate": 1.569821066223944e-05,
|
1247 |
+
"loss": 0.0381,
|
1248 |
+
"step": 1770
|
1249 |
+
},
|
1250 |
+
{
|
1251 |
+
"epoch": 1.18,
|
1252 |
+
"grad_norm": 1.3833442570741445,
|
1253 |
+
"learning_rate": 1.5661317100166022e-05,
|
1254 |
+
"loss": 0.0295,
|
1255 |
+
"step": 1780
|
1256 |
+
},
|
1257 |
+
{
|
1258 |
+
"epoch": 1.19,
|
1259 |
+
"grad_norm": 1.1660396503165016,
|
1260 |
+
"learning_rate": 1.5624423538092603e-05,
|
1261 |
+
"loss": 0.0313,
|
1262 |
+
"step": 1790
|
1263 |
+
},
|
1264 |
+
{
|
1265 |
+
"epoch": 1.2,
|
1266 |
+
"grad_norm": 1.085393536126589,
|
1267 |
+
"learning_rate": 1.5587529976019188e-05,
|
1268 |
+
"loss": 0.0294,
|
1269 |
+
"step": 1800
|
1270 |
+
},
|
1271 |
+
{
|
1272 |
+
"epoch": 1.2,
|
1273 |
+
"grad_norm": 0.9428955580680065,
|
1274 |
+
"learning_rate": 1.555063641394577e-05,
|
1275 |
+
"loss": 0.0403,
|
1276 |
+
"step": 1810
|
1277 |
+
},
|
1278 |
+
{
|
1279 |
+
"epoch": 1.21,
|
1280 |
+
"grad_norm": 1.2824389825772662,
|
1281 |
+
"learning_rate": 1.551374285187235e-05,
|
1282 |
+
"loss": 0.0326,
|
1283 |
+
"step": 1820
|
1284 |
+
},
|
1285 |
+
{
|
1286 |
+
"epoch": 1.22,
|
1287 |
+
"grad_norm": 1.071530839035684,
|
1288 |
+
"learning_rate": 1.547684928979893e-05,
|
1289 |
+
"loss": 0.0224,
|
1290 |
+
"step": 1830
|
1291 |
+
},
|
1292 |
+
{
|
1293 |
+
"epoch": 1.22,
|
1294 |
+
"grad_norm": 0.7250437295813942,
|
1295 |
+
"learning_rate": 1.5439955727725512e-05,
|
1296 |
+
"loss": 0.0253,
|
1297 |
+
"step": 1840
|
1298 |
+
},
|
1299 |
+
{
|
1300 |
+
"epoch": 1.23,
|
1301 |
+
"grad_norm": 0.4676245052117833,
|
1302 |
+
"learning_rate": 1.5403062165652093e-05,
|
1303 |
+
"loss": 0.0245,
|
1304 |
+
"step": 1850
|
1305 |
+
},
|
1306 |
+
{
|
1307 |
+
"epoch": 1.24,
|
1308 |
+
"grad_norm": 1.3654237978697314,
|
1309 |
+
"learning_rate": 1.5366168603578678e-05,
|
1310 |
+
"loss": 0.0274,
|
1311 |
+
"step": 1860
|
1312 |
+
},
|
1313 |
+
{
|
1314 |
+
"epoch": 1.24,
|
1315 |
+
"grad_norm": 0.8371654653693097,
|
1316 |
+
"learning_rate": 1.532927504150526e-05,
|
1317 |
+
"loss": 0.0408,
|
1318 |
+
"step": 1870
|
1319 |
+
},
|
1320 |
+
{
|
1321 |
+
"epoch": 1.25,
|
1322 |
+
"grad_norm": 0.918371136718616,
|
1323 |
+
"learning_rate": 1.529238147943184e-05,
|
1324 |
+
"loss": 0.0252,
|
1325 |
+
"step": 1880
|
1326 |
+
},
|
1327 |
+
{
|
1328 |
+
"epoch": 1.25,
|
1329 |
+
"grad_norm": 1.0100840319244568,
|
1330 |
+
"learning_rate": 1.5255487917358423e-05,
|
1331 |
+
"loss": 0.0282,
|
1332 |
+
"step": 1890
|
1333 |
+
},
|
1334 |
+
{
|
1335 |
+
"epoch": 1.26,
|
1336 |
+
"grad_norm": 1.2752226504227586,
|
1337 |
+
"learning_rate": 1.5218594355285004e-05,
|
1338 |
+
"loss": 0.0481,
|
1339 |
+
"step": 1900
|
1340 |
+
},
|
1341 |
+
{
|
1342 |
+
"epoch": 1.27,
|
1343 |
+
"grad_norm": 1.016628356639124,
|
1344 |
+
"learning_rate": 1.5181700793211587e-05,
|
1345 |
+
"loss": 0.0534,
|
1346 |
+
"step": 1910
|
1347 |
+
},
|
1348 |
+
{
|
1349 |
+
"epoch": 1.27,
|
1350 |
+
"grad_norm": 1.2018849311057271,
|
1351 |
+
"learning_rate": 1.5144807231138168e-05,
|
1352 |
+
"loss": 0.0357,
|
1353 |
+
"step": 1920
|
1354 |
+
},
|
1355 |
+
{
|
1356 |
+
"epoch": 1.28,
|
1357 |
+
"grad_norm": 1.3543149671289594,
|
1358 |
+
"learning_rate": 1.5107913669064749e-05,
|
1359 |
+
"loss": 0.0354,
|
1360 |
+
"step": 1930
|
1361 |
+
},
|
1362 |
+
{
|
1363 |
+
"epoch": 1.29,
|
1364 |
+
"grad_norm": 1.0460241475499934,
|
1365 |
+
"learning_rate": 1.5071020106991332e-05,
|
1366 |
+
"loss": 0.0328,
|
1367 |
+
"step": 1940
|
1368 |
+
},
|
1369 |
+
{
|
1370 |
+
"epoch": 1.29,
|
1371 |
+
"grad_norm": 0.7712993931217346,
|
1372 |
+
"learning_rate": 1.5034126544917913e-05,
|
1373 |
+
"loss": 0.0352,
|
1374 |
+
"step": 1950
|
1375 |
+
},
|
1376 |
+
{
|
1377 |
+
"epoch": 1.3,
|
1378 |
+
"grad_norm": 0.6403552627247137,
|
1379 |
+
"learning_rate": 1.4997232982844494e-05,
|
1380 |
+
"loss": 0.0374,
|
1381 |
+
"step": 1960
|
1382 |
+
},
|
1383 |
+
{
|
1384 |
+
"epoch": 1.31,
|
1385 |
+
"grad_norm": 0.6985534187744373,
|
1386 |
+
"learning_rate": 1.4960339420771077e-05,
|
1387 |
+
"loss": 0.0236,
|
1388 |
+
"step": 1970
|
1389 |
+
},
|
1390 |
+
{
|
1391 |
+
"epoch": 1.31,
|
1392 |
+
"grad_norm": 2.7973859948867315,
|
1393 |
+
"learning_rate": 1.4923445858697658e-05,
|
1394 |
+
"loss": 0.0333,
|
1395 |
+
"step": 1980
|
1396 |
+
},
|
1397 |
+
{
|
1398 |
+
"epoch": 1.32,
|
1399 |
+
"grad_norm": 0.7856751051336177,
|
1400 |
+
"learning_rate": 1.488655229662424e-05,
|
1401 |
+
"loss": 0.0283,
|
1402 |
+
"step": 1990
|
1403 |
+
},
|
1404 |
+
{
|
1405 |
+
"epoch": 1.33,
|
1406 |
+
"grad_norm": 1.1090831103953211,
|
1407 |
+
"learning_rate": 1.4849658734550822e-05,
|
1408 |
+
"loss": 0.0424,
|
1409 |
+
"step": 2000
|
1410 |
+
},
|
1411 |
+
{
|
1412 |
+
"epoch": 1.33,
|
1413 |
+
"grad_norm": 0.37705371680669364,
|
1414 |
+
"learning_rate": 1.4812765172477403e-05,
|
1415 |
+
"loss": 0.0264,
|
1416 |
+
"step": 2010
|
1417 |
+
},
|
1418 |
+
{
|
1419 |
+
"epoch": 1.34,
|
1420 |
+
"grad_norm": 2.011193423521568,
|
1421 |
+
"learning_rate": 1.4775871610403986e-05,
|
1422 |
+
"loss": 0.0325,
|
1423 |
+
"step": 2020
|
1424 |
+
},
|
1425 |
+
{
|
1426 |
+
"epoch": 1.35,
|
1427 |
+
"grad_norm": 1.00662492673962,
|
1428 |
+
"learning_rate": 1.4738978048330567e-05,
|
1429 |
+
"loss": 0.0325,
|
1430 |
+
"step": 2030
|
1431 |
+
},
|
1432 |
+
{
|
1433 |
+
"epoch": 1.35,
|
1434 |
+
"grad_norm": 1.0249189181939053,
|
1435 |
+
"learning_rate": 1.4702084486257148e-05,
|
1436 |
+
"loss": 0.0296,
|
1437 |
+
"step": 2040
|
1438 |
+
},
|
1439 |
+
{
|
1440 |
+
"epoch": 1.36,
|
1441 |
+
"grad_norm": 1.294162625472094,
|
1442 |
+
"learning_rate": 1.4665190924183733e-05,
|
1443 |
+
"loss": 0.0362,
|
1444 |
+
"step": 2050
|
1445 |
+
},
|
1446 |
+
{
|
1447 |
+
"epoch": 1.37,
|
1448 |
+
"grad_norm": 0.8753798886295676,
|
1449 |
+
"learning_rate": 1.4628297362110312e-05,
|
1450 |
+
"loss": 0.0352,
|
1451 |
+
"step": 2060
|
1452 |
+
},
|
1453 |
+
{
|
1454 |
+
"epoch": 1.37,
|
1455 |
+
"grad_norm": 0.32404034083576144,
|
1456 |
+
"learning_rate": 1.4591403800036893e-05,
|
1457 |
+
"loss": 0.0319,
|
1458 |
+
"step": 2070
|
1459 |
+
},
|
1460 |
+
{
|
1461 |
+
"epoch": 1.38,
|
1462 |
+
"grad_norm": 1.3881694244187606,
|
1463 |
+
"learning_rate": 1.4554510237963478e-05,
|
1464 |
+
"loss": 0.0383,
|
1465 |
+
"step": 2080
|
1466 |
+
},
|
1467 |
+
{
|
1468 |
+
"epoch": 1.39,
|
1469 |
+
"grad_norm": 0.9106958124280001,
|
1470 |
+
"learning_rate": 1.4517616675890059e-05,
|
1471 |
+
"loss": 0.0366,
|
1472 |
+
"step": 2090
|
1473 |
+
},
|
1474 |
+
{
|
1475 |
+
"epoch": 1.39,
|
1476 |
+
"grad_norm": 1.3920757998129725,
|
1477 |
+
"learning_rate": 1.448072311381664e-05,
|
1478 |
+
"loss": 0.0359,
|
1479 |
+
"step": 2100
|
1480 |
+
},
|
1481 |
+
{
|
1482 |
+
"epoch": 1.4,
|
1483 |
+
"grad_norm": 2.5378193320802556,
|
1484 |
+
"learning_rate": 1.4443829551743223e-05,
|
1485 |
+
"loss": 0.0375,
|
1486 |
+
"step": 2110
|
1487 |
+
},
|
1488 |
+
{
|
1489 |
+
"epoch": 1.41,
|
1490 |
+
"grad_norm": 1.6327275888367767,
|
1491 |
+
"learning_rate": 1.4406935989669804e-05,
|
1492 |
+
"loss": 0.0491,
|
1493 |
+
"step": 2120
|
1494 |
+
},
|
1495 |
+
{
|
1496 |
+
"epoch": 1.41,
|
1497 |
+
"grad_norm": 1.457863332305797,
|
1498 |
+
"learning_rate": 1.4370042427596385e-05,
|
1499 |
+
"loss": 0.0379,
|
1500 |
+
"step": 2130
|
1501 |
+
},
|
1502 |
+
{
|
1503 |
+
"epoch": 1.42,
|
1504 |
+
"grad_norm": 0.19454376571103232,
|
1505 |
+
"learning_rate": 1.4333148865522968e-05,
|
1506 |
+
"loss": 0.0357,
|
1507 |
+
"step": 2140
|
1508 |
+
},
|
1509 |
+
{
|
1510 |
+
"epoch": 1.43,
|
1511 |
+
"grad_norm": 0.7421971370673671,
|
1512 |
+
"learning_rate": 1.429625530344955e-05,
|
1513 |
+
"loss": 0.0418,
|
1514 |
+
"step": 2150
|
1515 |
+
},
|
1516 |
+
{
|
1517 |
+
"epoch": 1.43,
|
1518 |
+
"grad_norm": 0.0937417741981473,
|
1519 |
+
"learning_rate": 1.4259361741376132e-05,
|
1520 |
+
"loss": 0.0248,
|
1521 |
+
"step": 2160
|
1522 |
+
},
|
1523 |
+
{
|
1524 |
+
"epoch": 1.44,
|
1525 |
+
"grad_norm": 0.7544900086855553,
|
1526 |
+
"learning_rate": 1.4222468179302713e-05,
|
1527 |
+
"loss": 0.0242,
|
1528 |
+
"step": 2170
|
1529 |
+
},
|
1530 |
+
{
|
1531 |
+
"epoch": 1.45,
|
1532 |
+
"grad_norm": 0.10357082442667653,
|
1533 |
+
"learning_rate": 1.4185574617229294e-05,
|
1534 |
+
"loss": 0.0277,
|
1535 |
+
"step": 2180
|
1536 |
+
},
|
1537 |
+
{
|
1538 |
+
"epoch": 1.45,
|
1539 |
+
"grad_norm": 0.429204704240934,
|
1540 |
+
"learning_rate": 1.4148681055155877e-05,
|
1541 |
+
"loss": 0.0253,
|
1542 |
+
"step": 2190
|
1543 |
+
},
|
1544 |
+
{
|
1545 |
+
"epoch": 1.46,
|
1546 |
+
"grad_norm": 1.7367754983146269,
|
1547 |
+
"learning_rate": 1.4111787493082458e-05,
|
1548 |
+
"loss": 0.0337,
|
1549 |
+
"step": 2200
|
1550 |
+
},
|
1551 |
+
{
|
1552 |
+
"epoch": 1.47,
|
1553 |
+
"grad_norm": 0.6428519364176438,
|
1554 |
+
"learning_rate": 1.407489393100904e-05,
|
1555 |
+
"loss": 0.0331,
|
1556 |
+
"step": 2210
|
1557 |
+
},
|
1558 |
+
{
|
1559 |
+
"epoch": 1.47,
|
1560 |
+
"grad_norm": 0.7848514705508334,
|
1561 |
+
"learning_rate": 1.4038000368935622e-05,
|
1562 |
+
"loss": 0.031,
|
1563 |
+
"step": 2220
|
1564 |
+
},
|
1565 |
+
{
|
1566 |
+
"epoch": 1.48,
|
1567 |
+
"grad_norm": 1.2604412780544694,
|
1568 |
+
"learning_rate": 1.4001106806862203e-05,
|
1569 |
+
"loss": 0.0381,
|
1570 |
+
"step": 2230
|
1571 |
+
},
|
1572 |
+
{
|
1573 |
+
"epoch": 1.49,
|
1574 |
+
"grad_norm": 1.0154534415085918,
|
1575 |
+
"learning_rate": 1.3964213244788784e-05,
|
1576 |
+
"loss": 0.0294,
|
1577 |
+
"step": 2240
|
1578 |
+
},
|
1579 |
+
{
|
1580 |
+
"epoch": 1.49,
|
1581 |
+
"grad_norm": 1.2166332545112624,
|
1582 |
+
"learning_rate": 1.3927319682715367e-05,
|
1583 |
+
"loss": 0.0275,
|
1584 |
+
"step": 2250
|
1585 |
+
},
|
1586 |
+
{
|
1587 |
+
"epoch": 1.5,
|
1588 |
+
"grad_norm": 0.8758369938501386,
|
1589 |
+
"learning_rate": 1.3890426120641948e-05,
|
1590 |
+
"loss": 0.0359,
|
1591 |
+
"step": 2260
|
1592 |
+
},
|
1593 |
+
{
|
1594 |
+
"epoch": 1.51,
|
1595 |
+
"grad_norm": 0.9645035178583596,
|
1596 |
+
"learning_rate": 1.385353255856853e-05,
|
1597 |
+
"loss": 0.0257,
|
1598 |
+
"step": 2270
|
1599 |
+
},
|
1600 |
+
{
|
1601 |
+
"epoch": 1.51,
|
1602 |
+
"grad_norm": 0.24217577106008117,
|
1603 |
+
"learning_rate": 1.3816638996495112e-05,
|
1604 |
+
"loss": 0.0188,
|
1605 |
+
"step": 2280
|
1606 |
+
},
|
1607 |
+
{
|
1608 |
+
"epoch": 1.52,
|
1609 |
+
"grad_norm": 0.8656287403232453,
|
1610 |
+
"learning_rate": 1.3779745434421693e-05,
|
1611 |
+
"loss": 0.035,
|
1612 |
+
"step": 2290
|
1613 |
+
},
|
1614 |
+
{
|
1615 |
+
"epoch": 1.53,
|
1616 |
+
"grad_norm": 1.5535880950760363,
|
1617 |
+
"learning_rate": 1.3742851872348278e-05,
|
1618 |
+
"loss": 0.0325,
|
1619 |
+
"step": 2300
|
1620 |
+
},
|
1621 |
+
{
|
1622 |
+
"epoch": 1.53,
|
1623 |
+
"grad_norm": 0.6356266461691455,
|
1624 |
+
"learning_rate": 1.3705958310274859e-05,
|
1625 |
+
"loss": 0.0292,
|
1626 |
+
"step": 2310
|
1627 |
+
},
|
1628 |
+
{
|
1629 |
+
"epoch": 1.54,
|
1630 |
+
"grad_norm": 1.3067413428110854,
|
1631 |
+
"learning_rate": 1.3669064748201439e-05,
|
1632 |
+
"loss": 0.0206,
|
1633 |
+
"step": 2320
|
1634 |
+
},
|
1635 |
+
{
|
1636 |
+
"epoch": 1.55,
|
1637 |
+
"grad_norm": 1.3327319855425117,
|
1638 |
+
"learning_rate": 1.3632171186128023e-05,
|
1639 |
+
"loss": 0.023,
|
1640 |
+
"step": 2330
|
1641 |
+
},
|
1642 |
+
{
|
1643 |
+
"epoch": 1.55,
|
1644 |
+
"grad_norm": 2.308037550483557,
|
1645 |
+
"learning_rate": 1.3595277624054604e-05,
|
1646 |
+
"loss": 0.0283,
|
1647 |
+
"step": 2340
|
1648 |
+
},
|
1649 |
+
{
|
1650 |
+
"epoch": 1.56,
|
1651 |
+
"grad_norm": 1.447528110026965,
|
1652 |
+
"learning_rate": 1.3558384061981185e-05,
|
1653 |
+
"loss": 0.0344,
|
1654 |
+
"step": 2350
|
1655 |
+
},
|
1656 |
+
{
|
1657 |
+
"epoch": 1.57,
|
1658 |
+
"grad_norm": 1.4781377553893515,
|
1659 |
+
"learning_rate": 1.3521490499907768e-05,
|
1660 |
+
"loss": 0.0338,
|
1661 |
+
"step": 2360
|
1662 |
+
},
|
1663 |
+
{
|
1664 |
+
"epoch": 1.57,
|
1665 |
+
"grad_norm": 1.2213361639825704,
|
1666 |
+
"learning_rate": 1.348459693783435e-05,
|
1667 |
+
"loss": 0.0213,
|
1668 |
+
"step": 2370
|
1669 |
+
},
|
1670 |
+
{
|
1671 |
+
"epoch": 1.58,
|
1672 |
+
"grad_norm": 0.806579624483452,
|
1673 |
+
"learning_rate": 1.344770337576093e-05,
|
1674 |
+
"loss": 0.0308,
|
1675 |
+
"step": 2380
|
1676 |
+
},
|
1677 |
+
{
|
1678 |
+
"epoch": 1.59,
|
1679 |
+
"grad_norm": 0.36305855519021074,
|
1680 |
+
"learning_rate": 1.3410809813687513e-05,
|
1681 |
+
"loss": 0.0307,
|
1682 |
+
"step": 2390
|
1683 |
+
},
|
1684 |
+
{
|
1685 |
+
"epoch": 1.59,
|
1686 |
+
"grad_norm": 1.7963102033556697,
|
1687 |
+
"learning_rate": 1.3373916251614094e-05,
|
1688 |
+
"loss": 0.0438,
|
1689 |
+
"step": 2400
|
1690 |
+
},
|
1691 |
+
{
|
1692 |
+
"epoch": 1.6,
|
1693 |
+
"grad_norm": 0.8741626138535608,
|
1694 |
+
"learning_rate": 1.3337022689540676e-05,
|
1695 |
+
"loss": 0.0282,
|
1696 |
+
"step": 2410
|
1697 |
+
},
|
1698 |
+
{
|
1699 |
+
"epoch": 1.61,
|
1700 |
+
"grad_norm": 0.8183941890622122,
|
1701 |
+
"learning_rate": 1.3300129127467258e-05,
|
1702 |
+
"loss": 0.03,
|
1703 |
+
"step": 2420
|
1704 |
+
},
|
1705 |
+
{
|
1706 |
+
"epoch": 1.61,
|
1707 |
+
"grad_norm": 1.315171338874376,
|
1708 |
+
"learning_rate": 1.326323556539384e-05,
|
1709 |
+
"loss": 0.0347,
|
1710 |
+
"step": 2430
|
1711 |
+
},
|
1712 |
+
{
|
1713 |
+
"epoch": 1.62,
|
1714 |
+
"grad_norm": 1.268446896432008,
|
1715 |
+
"learning_rate": 1.3226342003320422e-05,
|
1716 |
+
"loss": 0.0376,
|
1717 |
+
"step": 2440
|
1718 |
+
},
|
1719 |
+
{
|
1720 |
+
"epoch": 1.63,
|
1721 |
+
"grad_norm": 0.913219454382149,
|
1722 |
+
"learning_rate": 1.3189448441247003e-05,
|
1723 |
+
"loss": 0.0358,
|
1724 |
+
"step": 2450
|
1725 |
+
},
|
1726 |
+
{
|
1727 |
+
"epoch": 1.63,
|
1728 |
+
"grad_norm": 1.1043754536185766,
|
1729 |
+
"learning_rate": 1.3152554879173585e-05,
|
1730 |
+
"loss": 0.0284,
|
1731 |
+
"step": 2460
|
1732 |
+
},
|
1733 |
+
{
|
1734 |
+
"epoch": 1.64,
|
1735 |
+
"grad_norm": 1.203565789266508,
|
1736 |
+
"learning_rate": 1.3115661317100167e-05,
|
1737 |
+
"loss": 0.0292,
|
1738 |
+
"step": 2470
|
1739 |
+
},
|
1740 |
+
{
|
1741 |
+
"epoch": 1.65,
|
1742 |
+
"grad_norm": 1.9568074416295906,
|
1743 |
+
"learning_rate": 1.3078767755026749e-05,
|
1744 |
+
"loss": 0.0262,
|
1745 |
+
"step": 2480
|
1746 |
+
},
|
1747 |
+
{
|
1748 |
+
"epoch": 1.65,
|
1749 |
+
"grad_norm": 0.2859744392048369,
|
1750 |
+
"learning_rate": 1.304187419295333e-05,
|
1751 |
+
"loss": 0.0295,
|
1752 |
+
"step": 2490
|
1753 |
+
},
|
1754 |
+
{
|
1755 |
+
"epoch": 1.66,
|
1756 |
+
"grad_norm": 1.077880610534017,
|
1757 |
+
"learning_rate": 1.3004980630879912e-05,
|
1758 |
+
"loss": 0.0288,
|
1759 |
+
"step": 2500
|
1760 |
+
},
|
1761 |
+
{
|
1762 |
+
"epoch": 1.67,
|
1763 |
+
"grad_norm": 0.9333475338900054,
|
1764 |
+
"learning_rate": 1.2968087068806494e-05,
|
1765 |
+
"loss": 0.0355,
|
1766 |
+
"step": 2510
|
1767 |
+
},
|
1768 |
+
{
|
1769 |
+
"epoch": 1.67,
|
1770 |
+
"grad_norm": 1.904061045593678,
|
1771 |
+
"learning_rate": 1.2931193506733075e-05,
|
1772 |
+
"loss": 0.0251,
|
1773 |
+
"step": 2520
|
1774 |
+
},
|
1775 |
+
{
|
1776 |
+
"epoch": 1.68,
|
1777 |
+
"grad_norm": 0.58158119774247,
|
1778 |
+
"learning_rate": 1.2894299944659658e-05,
|
1779 |
+
"loss": 0.0233,
|
1780 |
+
"step": 2530
|
1781 |
+
},
|
1782 |
+
{
|
1783 |
+
"epoch": 1.69,
|
1784 |
+
"grad_norm": 0.7277481905024581,
|
1785 |
+
"learning_rate": 1.2857406382586239e-05,
|
1786 |
+
"loss": 0.0298,
|
1787 |
+
"step": 2540
|
1788 |
+
},
|
1789 |
+
{
|
1790 |
+
"epoch": 1.69,
|
1791 |
+
"grad_norm": 1.8262592151991606,
|
1792 |
+
"learning_rate": 1.2820512820512823e-05,
|
1793 |
+
"loss": 0.0276,
|
1794 |
+
"step": 2550
|
1795 |
+
},
|
1796 |
+
{
|
1797 |
+
"epoch": 1.7,
|
1798 |
+
"grad_norm": 0.6465268769367779,
|
1799 |
+
"learning_rate": 1.2783619258439404e-05,
|
1800 |
+
"loss": 0.027,
|
1801 |
+
"step": 2560
|
1802 |
+
},
|
1803 |
+
{
|
1804 |
+
"epoch": 1.71,
|
1805 |
+
"grad_norm": 0.6081326854882737,
|
1806 |
+
"learning_rate": 1.2746725696365985e-05,
|
1807 |
+
"loss": 0.0337,
|
1808 |
+
"step": 2570
|
1809 |
+
},
|
1810 |
+
{
|
1811 |
+
"epoch": 1.71,
|
1812 |
+
"grad_norm": 0.7068771700231296,
|
1813 |
+
"learning_rate": 1.2709832134292568e-05,
|
1814 |
+
"loss": 0.0236,
|
1815 |
+
"step": 2580
|
1816 |
+
},
|
1817 |
+
{
|
1818 |
+
"epoch": 1.72,
|
1819 |
+
"grad_norm": 1.8049539476685674,
|
1820 |
+
"learning_rate": 1.267293857221915e-05,
|
1821 |
+
"loss": 0.0282,
|
1822 |
+
"step": 2590
|
1823 |
+
},
|
1824 |
+
{
|
1825 |
+
"epoch": 1.73,
|
1826 |
+
"grad_norm": 0.553245442957324,
|
1827 |
+
"learning_rate": 1.263604501014573e-05,
|
1828 |
+
"loss": 0.0226,
|
1829 |
+
"step": 2600
|
1830 |
+
},
|
1831 |
+
{
|
1832 |
+
"epoch": 1.73,
|
1833 |
+
"grad_norm": 0.8295872212818136,
|
1834 |
+
"learning_rate": 1.2599151448072313e-05,
|
1835 |
+
"loss": 0.0232,
|
1836 |
+
"step": 2610
|
1837 |
+
},
|
1838 |
+
{
|
1839 |
+
"epoch": 1.74,
|
1840 |
+
"grad_norm": 0.9798348798123743,
|
1841 |
+
"learning_rate": 1.2562257885998895e-05,
|
1842 |
+
"loss": 0.0456,
|
1843 |
+
"step": 2620
|
1844 |
+
},
|
1845 |
+
{
|
1846 |
+
"epoch": 1.75,
|
1847 |
+
"grad_norm": 1.19245304335107,
|
1848 |
+
"learning_rate": 1.2525364323925476e-05,
|
1849 |
+
"loss": 0.0284,
|
1850 |
+
"step": 2630
|
1851 |
+
},
|
1852 |
+
{
|
1853 |
+
"epoch": 1.75,
|
1854 |
+
"grad_norm": 1.5998767447680797,
|
1855 |
+
"learning_rate": 1.2488470761852058e-05,
|
1856 |
+
"loss": 0.0307,
|
1857 |
+
"step": 2640
|
1858 |
+
},
|
1859 |
+
{
|
1860 |
+
"epoch": 1.76,
|
1861 |
+
"grad_norm": 1.7431018978542536,
|
1862 |
+
"learning_rate": 1.245157719977864e-05,
|
1863 |
+
"loss": 0.0216,
|
1864 |
+
"step": 2650
|
1865 |
+
},
|
1866 |
+
{
|
1867 |
+
"epoch": 1.77,
|
1868 |
+
"grad_norm": 1.0864224894613053,
|
1869 |
+
"learning_rate": 1.241468363770522e-05,
|
1870 |
+
"loss": 0.0388,
|
1871 |
+
"step": 2660
|
1872 |
+
},
|
1873 |
+
{
|
1874 |
+
"epoch": 1.77,
|
1875 |
+
"grad_norm": 2.164074288346446,
|
1876 |
+
"learning_rate": 1.2377790075631804e-05,
|
1877 |
+
"loss": 0.0302,
|
1878 |
+
"step": 2670
|
1879 |
+
},
|
1880 |
+
{
|
1881 |
+
"epoch": 1.78,
|
1882 |
+
"grad_norm": 0.6890867086297026,
|
1883 |
+
"learning_rate": 1.2340896513558385e-05,
|
1884 |
+
"loss": 0.0258,
|
1885 |
+
"step": 2680
|
1886 |
+
},
|
1887 |
+
{
|
1888 |
+
"epoch": 1.79,
|
1889 |
+
"grad_norm": 1.1150440915546074,
|
1890 |
+
"learning_rate": 1.2304002951484968e-05,
|
1891 |
+
"loss": 0.0279,
|
1892 |
+
"step": 2690
|
1893 |
+
},
|
1894 |
+
{
|
1895 |
+
"epoch": 1.79,
|
1896 |
+
"grad_norm": 0.7515709240098276,
|
1897 |
+
"learning_rate": 1.2267109389411549e-05,
|
1898 |
+
"loss": 0.0419,
|
1899 |
+
"step": 2700
|
1900 |
+
},
|
1901 |
+
{
|
1902 |
+
"epoch": 1.8,
|
1903 |
+
"grad_norm": 0.7795946895425347,
|
1904 |
+
"learning_rate": 1.223021582733813e-05,
|
1905 |
+
"loss": 0.0152,
|
1906 |
+
"step": 2710
|
1907 |
+
},
|
1908 |
+
{
|
1909 |
+
"epoch": 1.81,
|
1910 |
+
"grad_norm": 0.7206757082802774,
|
1911 |
+
"learning_rate": 1.2193322265264713e-05,
|
1912 |
+
"loss": 0.039,
|
1913 |
+
"step": 2720
|
1914 |
+
},
|
1915 |
+
{
|
1916 |
+
"epoch": 1.81,
|
1917 |
+
"grad_norm": 0.9261417250470075,
|
1918 |
+
"learning_rate": 1.2156428703191294e-05,
|
1919 |
+
"loss": 0.031,
|
1920 |
+
"step": 2730
|
1921 |
+
},
|
1922 |
+
{
|
1923 |
+
"epoch": 1.82,
|
1924 |
+
"grad_norm": 0.9738836548814439,
|
1925 |
+
"learning_rate": 1.2119535141117875e-05,
|
1926 |
+
"loss": 0.0318,
|
1927 |
+
"step": 2740
|
1928 |
+
},
|
1929 |
+
{
|
1930 |
+
"epoch": 1.83,
|
1931 |
+
"grad_norm": 1.6517158198941853,
|
1932 |
+
"learning_rate": 1.2082641579044458e-05,
|
1933 |
+
"loss": 0.0346,
|
1934 |
+
"step": 2750
|
1935 |
+
},
|
1936 |
+
{
|
1937 |
+
"epoch": 1.83,
|
1938 |
+
"grad_norm": 0.8931816948527265,
|
1939 |
+
"learning_rate": 1.2045748016971039e-05,
|
1940 |
+
"loss": 0.0278,
|
1941 |
+
"step": 2760
|
1942 |
+
},
|
1943 |
+
{
|
1944 |
+
"epoch": 1.84,
|
1945 |
+
"grad_norm": 1.5388621088460506,
|
1946 |
+
"learning_rate": 1.200885445489762e-05,
|
1947 |
+
"loss": 0.0296,
|
1948 |
+
"step": 2770
|
1949 |
+
},
|
1950 |
+
{
|
1951 |
+
"epoch": 1.85,
|
1952 |
+
"grad_norm": 0.6080136375748658,
|
1953 |
+
"learning_rate": 1.1971960892824204e-05,
|
1954 |
+
"loss": 0.0314,
|
1955 |
+
"step": 2780
|
1956 |
+
},
|
1957 |
+
{
|
1958 |
+
"epoch": 1.85,
|
1959 |
+
"grad_norm": 0.7855564824393652,
|
1960 |
+
"learning_rate": 1.1935067330750784e-05,
|
1961 |
+
"loss": 0.0371,
|
1962 |
+
"step": 2790
|
1963 |
+
},
|
1964 |
+
{
|
1965 |
+
"epoch": 1.86,
|
1966 |
+
"grad_norm": 1.391898825501054,
|
1967 |
+
"learning_rate": 1.1898173768677365e-05,
|
1968 |
+
"loss": 0.0215,
|
1969 |
+
"step": 2800
|
1970 |
+
},
|
1971 |
+
{
|
1972 |
+
"epoch": 1.87,
|
1973 |
+
"grad_norm": 0.023622261403316365,
|
1974 |
+
"learning_rate": 1.186128020660395e-05,
|
1975 |
+
"loss": 0.0289,
|
1976 |
+
"step": 2810
|
1977 |
+
},
|
1978 |
+
{
|
1979 |
+
"epoch": 1.87,
|
1980 |
+
"grad_norm": 1.2844725300311655,
|
1981 |
+
"learning_rate": 1.182438664453053e-05,
|
1982 |
+
"loss": 0.0272,
|
1983 |
+
"step": 2820
|
1984 |
+
},
|
1985 |
+
{
|
1986 |
+
"epoch": 1.88,
|
1987 |
+
"grad_norm": 0.008098058731733223,
|
1988 |
+
"learning_rate": 1.1787493082457114e-05,
|
1989 |
+
"loss": 0.021,
|
1990 |
+
"step": 2830
|
1991 |
+
},
|
1992 |
+
{
|
1993 |
+
"epoch": 1.89,
|
1994 |
+
"grad_norm": 2.4455783687272565,
|
1995 |
+
"learning_rate": 1.1750599520383695e-05,
|
1996 |
+
"loss": 0.0387,
|
1997 |
+
"step": 2840
|
1998 |
+
},
|
1999 |
+
{
|
2000 |
+
"epoch": 1.89,
|
2001 |
+
"grad_norm": 0.5832035368747368,
|
2002 |
+
"learning_rate": 1.1713705958310276e-05,
|
2003 |
+
"loss": 0.0392,
|
2004 |
+
"step": 2850
|
2005 |
+
},
|
2006 |
+
{
|
2007 |
+
"epoch": 1.9,
|
2008 |
+
"grad_norm": 0.4288935348488189,
|
2009 |
+
"learning_rate": 1.1676812396236859e-05,
|
2010 |
+
"loss": 0.0243,
|
2011 |
+
"step": 2860
|
2012 |
+
},
|
2013 |
+
{
|
2014 |
+
"epoch": 1.91,
|
2015 |
+
"grad_norm": 0.85140142754813,
|
2016 |
+
"learning_rate": 1.163991883416344e-05,
|
2017 |
+
"loss": 0.0288,
|
2018 |
+
"step": 2870
|
2019 |
+
},
|
2020 |
+
{
|
2021 |
+
"epoch": 1.91,
|
2022 |
+
"grad_norm": 0.6208450731238502,
|
2023 |
+
"learning_rate": 1.1603025272090021e-05,
|
2024 |
+
"loss": 0.0257,
|
2025 |
+
"step": 2880
|
2026 |
+
},
|
2027 |
+
{
|
2028 |
+
"epoch": 1.92,
|
2029 |
+
"grad_norm": 0.5506545117162834,
|
2030 |
+
"learning_rate": 1.1566131710016604e-05,
|
2031 |
+
"loss": 0.0257,
|
2032 |
+
"step": 2890
|
2033 |
+
},
|
2034 |
+
{
|
2035 |
+
"epoch": 1.93,
|
2036 |
+
"grad_norm": 0.6983830876671564,
|
2037 |
+
"learning_rate": 1.1529238147943185e-05,
|
2038 |
+
"loss": 0.0153,
|
2039 |
+
"step": 2900
|
2040 |
+
},
|
2041 |
+
{
|
2042 |
+
"epoch": 1.93,
|
2043 |
+
"grad_norm": 0.5365569446878222,
|
2044 |
+
"learning_rate": 1.1492344585869766e-05,
|
2045 |
+
"loss": 0.0287,
|
2046 |
+
"step": 2910
|
2047 |
+
},
|
2048 |
+
{
|
2049 |
+
"epoch": 1.94,
|
2050 |
+
"grad_norm": 0.48404097576321364,
|
2051 |
+
"learning_rate": 1.1455451023796349e-05,
|
2052 |
+
"loss": 0.0189,
|
2053 |
+
"step": 2920
|
2054 |
+
},
|
2055 |
+
{
|
2056 |
+
"epoch": 1.95,
|
2057 |
+
"grad_norm": 0.2656124897423078,
|
2058 |
+
"learning_rate": 1.141855746172293e-05,
|
2059 |
+
"loss": 0.0321,
|
2060 |
+
"step": 2930
|
2061 |
+
},
|
2062 |
+
{
|
2063 |
+
"epoch": 1.95,
|
2064 |
+
"grad_norm": 0.9816015622085352,
|
2065 |
+
"learning_rate": 1.1381663899649511e-05,
|
2066 |
+
"loss": 0.0392,
|
2067 |
+
"step": 2940
|
2068 |
+
},
|
2069 |
+
{
|
2070 |
+
"epoch": 1.96,
|
2071 |
+
"grad_norm": 1.6810737744401563,
|
2072 |
+
"learning_rate": 1.1344770337576094e-05,
|
2073 |
+
"loss": 0.022,
|
2074 |
+
"step": 2950
|
2075 |
+
},
|
2076 |
+
{
|
2077 |
+
"epoch": 1.97,
|
2078 |
+
"grad_norm": 1.4543583055955254,
|
2079 |
+
"learning_rate": 1.1307876775502675e-05,
|
2080 |
+
"loss": 0.0331,
|
2081 |
+
"step": 2960
|
2082 |
+
},
|
2083 |
+
{
|
2084 |
+
"epoch": 1.97,
|
2085 |
+
"grad_norm": 1.4710753680279536,
|
2086 |
+
"learning_rate": 1.1270983213429258e-05,
|
2087 |
+
"loss": 0.0383,
|
2088 |
+
"step": 2970
|
2089 |
+
},
|
2090 |
+
{
|
2091 |
+
"epoch": 1.98,
|
2092 |
+
"grad_norm": 1.2478883072438642,
|
2093 |
+
"learning_rate": 1.1234089651355839e-05,
|
2094 |
+
"loss": 0.0316,
|
2095 |
+
"step": 2980
|
2096 |
+
},
|
2097 |
+
{
|
2098 |
+
"epoch": 1.99,
|
2099 |
+
"grad_norm": 2.2702061081470153,
|
2100 |
+
"learning_rate": 1.119719608928242e-05,
|
2101 |
+
"loss": 0.0282,
|
2102 |
+
"step": 2990
|
2103 |
+
},
|
2104 |
+
{
|
2105 |
+
"epoch": 1.99,
|
2106 |
+
"grad_norm": 0.6940800534656072,
|
2107 |
+
"learning_rate": 1.1160302527209005e-05,
|
2108 |
+
"loss": 0.0243,
|
2109 |
+
"step": 3000
|
2110 |
+
},
|
2111 |
+
{
|
2112 |
+
"epoch": 2.0,
|
2113 |
+
"grad_norm": 0.576849535693127,
|
2114 |
+
"learning_rate": 1.1123408965135584e-05,
|
2115 |
+
"loss": 0.0209,
|
2116 |
+
"step": 3010
|
2117 |
+
}
|
2118 |
+
],
|
2119 |
+
"logging_steps": 10,
|
2120 |
+
"max_steps": 6024,
|
2121 |
+
"num_input_tokens_seen": 0,
|
2122 |
+
"num_train_epochs": 4,
|
2123 |
+
"save_steps": 500,
|
2124 |
+
"total_flos": 0.0,
|
2125 |
+
"train_batch_size": 1,
|
2126 |
+
"trial_name": null,
|
2127 |
+
"trial_params": null
|
2128 |
+
}
|
checkpoint-3012/training_args.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d0640503f1433ff703cc07d0997c18edbdba68efdd139a495f5d107bbd939a06
|
3 |
+
size 6395
|
checkpoint-3012/zero_to_fp32.py
ADDED
@@ -0,0 +1,587 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
#!/usr/bin/env python
|
2 |
+
|
3 |
+
# Copyright (c) Microsoft Corporation.
|
4 |
+
# SPDX-License-Identifier: Apache-2.0
|
5 |
+
|
6 |
+
# DeepSpeed Team
|
7 |
+
|
8 |
+
# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
|
9 |
+
# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
|
10 |
+
# the future. Once extracted, the weights don't require DeepSpeed and can be used in any
|
11 |
+
# application.
|
12 |
+
#
|
13 |
+
# example: python zero_to_fp32.py . pytorch_model.bin
|
14 |
+
|
15 |
+
import argparse
|
16 |
+
import torch
|
17 |
+
import glob
|
18 |
+
import math
|
19 |
+
import os
|
20 |
+
import re
|
21 |
+
from collections import OrderedDict
|
22 |
+
from dataclasses import dataclass
|
23 |
+
|
24 |
+
# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
|
25 |
+
# DeepSpeed data structures it has to be available in the current python environment.
|
26 |
+
from deepspeed.utils import logger
|
27 |
+
from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
|
28 |
+
FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
|
29 |
+
FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
|
30 |
+
|
31 |
+
|
32 |
+
@dataclass
|
33 |
+
class zero_model_state:
|
34 |
+
buffers: dict()
|
35 |
+
param_shapes: dict()
|
36 |
+
shared_params: list
|
37 |
+
ds_version: int
|
38 |
+
frozen_param_shapes: dict()
|
39 |
+
frozen_param_fragments: dict()
|
40 |
+
|
41 |
+
|
42 |
+
debug = 0
|
43 |
+
|
44 |
+
# load to cpu
|
45 |
+
device = torch.device('cpu')
|
46 |
+
|
47 |
+
|
48 |
+
def atoi(text):
|
49 |
+
return int(text) if text.isdigit() else text
|
50 |
+
|
51 |
+
|
52 |
+
def natural_keys(text):
|
53 |
+
'''
|
54 |
+
alist.sort(key=natural_keys) sorts in human order
|
55 |
+
http://nedbatchelder.com/blog/200712/human_sorting.html
|
56 |
+
(See Toothy's implementation in the comments)
|
57 |
+
'''
|
58 |
+
return [atoi(c) for c in re.split(r'(\d+)', text)]
|
59 |
+
|
60 |
+
|
61 |
+
def get_model_state_file(checkpoint_dir, zero_stage):
|
62 |
+
if not os.path.isdir(checkpoint_dir):
|
63 |
+
raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
|
64 |
+
|
65 |
+
# there should be only one file
|
66 |
+
if zero_stage <= 2:
|
67 |
+
file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
|
68 |
+
elif zero_stage == 3:
|
69 |
+
file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
|
70 |
+
|
71 |
+
if not os.path.exists(file):
|
72 |
+
raise FileNotFoundError(f"can't find model states file at '{file}'")
|
73 |
+
|
74 |
+
return file
|
75 |
+
|
76 |
+
|
77 |
+
def get_checkpoint_files(checkpoint_dir, glob_pattern):
|
78 |
+
# XXX: need to test that this simple glob rule works for multi-node setup too
|
79 |
+
ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
|
80 |
+
|
81 |
+
if len(ckpt_files) == 0:
|
82 |
+
raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
|
83 |
+
|
84 |
+
return ckpt_files
|
85 |
+
|
86 |
+
|
87 |
+
def get_optim_files(checkpoint_dir):
|
88 |
+
return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
|
89 |
+
|
90 |
+
|
91 |
+
def get_model_state_files(checkpoint_dir):
|
92 |
+
return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
|
93 |
+
|
94 |
+
|
95 |
+
def parse_model_states(files):
|
96 |
+
zero_model_states = []
|
97 |
+
for file in files:
|
98 |
+
state_dict = torch.load(file, map_location=device)
|
99 |
+
|
100 |
+
if BUFFER_NAMES not in state_dict:
|
101 |
+
raise ValueError(f"{file} is not a model state checkpoint")
|
102 |
+
buffer_names = state_dict[BUFFER_NAMES]
|
103 |
+
if debug:
|
104 |
+
print("Found buffers:", buffer_names)
|
105 |
+
|
106 |
+
# recover just the buffers while restoring them to fp32 if they were saved in fp16
|
107 |
+
buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
|
108 |
+
param_shapes = state_dict[PARAM_SHAPES]
|
109 |
+
|
110 |
+
# collect parameters that are included in param_shapes
|
111 |
+
param_names = []
|
112 |
+
for s in param_shapes:
|
113 |
+
for name in s.keys():
|
114 |
+
param_names.append(name)
|
115 |
+
|
116 |
+
# update with frozen parameters
|
117 |
+
frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
|
118 |
+
if frozen_param_shapes is not None:
|
119 |
+
if debug:
|
120 |
+
print(f"Found frozen_param_shapes: {frozen_param_shapes}")
|
121 |
+
param_names += list(frozen_param_shapes.keys())
|
122 |
+
|
123 |
+
# handle shared params
|
124 |
+
shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
|
125 |
+
|
126 |
+
ds_version = state_dict.get(DS_VERSION, None)
|
127 |
+
|
128 |
+
frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
|
129 |
+
|
130 |
+
z_model_state = zero_model_state(buffers=buffers,
|
131 |
+
param_shapes=param_shapes,
|
132 |
+
shared_params=shared_params,
|
133 |
+
ds_version=ds_version,
|
134 |
+
frozen_param_shapes=frozen_param_shapes,
|
135 |
+
frozen_param_fragments=frozen_param_fragments)
|
136 |
+
zero_model_states.append(z_model_state)
|
137 |
+
|
138 |
+
return zero_model_states
|
139 |
+
|
140 |
+
|
141 |
+
def parse_optim_states(files, ds_checkpoint_dir):
|
142 |
+
|
143 |
+
total_files = len(files)
|
144 |
+
state_dicts = []
|
145 |
+
for f in files:
|
146 |
+
state_dict = torch.load(f, map_location=device)
|
147 |
+
# immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
|
148 |
+
# and also handle the case where it was already removed by another helper script
|
149 |
+
state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
|
150 |
+
state_dicts.append(state_dict)
|
151 |
+
|
152 |
+
if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
|
153 |
+
raise ValueError(f"{files[0]} is not a zero checkpoint")
|
154 |
+
zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
|
155 |
+
world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
|
156 |
+
|
157 |
+
# For ZeRO-2 each param group can have different partition_count as data parallelism for expert
|
158 |
+
# parameters can be different from data parallelism for non-expert parameters. So we can just
|
159 |
+
# use the max of the partition_count to get the dp world_size.
|
160 |
+
|
161 |
+
if type(world_size) is list:
|
162 |
+
world_size = max(world_size)
|
163 |
+
|
164 |
+
if world_size != total_files:
|
165 |
+
raise ValueError(
|
166 |
+
f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
|
167 |
+
"Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
|
168 |
+
)
|
169 |
+
|
170 |
+
# the groups are named differently in each stage
|
171 |
+
if zero_stage <= 2:
|
172 |
+
fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
|
173 |
+
elif zero_stage == 3:
|
174 |
+
fp32_groups_key = FP32_FLAT_GROUPS
|
175 |
+
else:
|
176 |
+
raise ValueError(f"unknown zero stage {zero_stage}")
|
177 |
+
|
178 |
+
if zero_stage <= 2:
|
179 |
+
fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
|
180 |
+
elif zero_stage == 3:
|
181 |
+
# if there is more than one param group, there will be multiple flattened tensors - one
|
182 |
+
# flattened tensor per group - for simplicity merge them into a single tensor
|
183 |
+
#
|
184 |
+
# XXX: could make the script more memory efficient for when there are multiple groups - it
|
185 |
+
# will require matching the sub-lists of param_shapes for each param group flattened tensor
|
186 |
+
|
187 |
+
fp32_flat_groups = [
|
188 |
+
torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
|
189 |
+
]
|
190 |
+
|
191 |
+
return zero_stage, world_size, fp32_flat_groups
|
192 |
+
|
193 |
+
|
194 |
+
def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir):
|
195 |
+
"""
|
196 |
+
Returns fp32 state_dict reconstructed from ds checkpoint
|
197 |
+
|
198 |
+
Args:
|
199 |
+
- ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
|
200 |
+
|
201 |
+
"""
|
202 |
+
print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
|
203 |
+
|
204 |
+
optim_files = get_optim_files(ds_checkpoint_dir)
|
205 |
+
zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
|
206 |
+
print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
|
207 |
+
|
208 |
+
model_files = get_model_state_files(ds_checkpoint_dir)
|
209 |
+
|
210 |
+
zero_model_states = parse_model_states(model_files)
|
211 |
+
print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
|
212 |
+
|
213 |
+
if zero_stage <= 2:
|
214 |
+
return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states)
|
215 |
+
elif zero_stage == 3:
|
216 |
+
return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states)
|
217 |
+
|
218 |
+
|
219 |
+
def _zero2_merge_frozen_params(state_dict, zero_model_states):
|
220 |
+
if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
|
221 |
+
return
|
222 |
+
|
223 |
+
frozen_param_shapes = zero_model_states[0].frozen_param_shapes
|
224 |
+
frozen_param_fragments = zero_model_states[0].frozen_param_fragments
|
225 |
+
|
226 |
+
if debug:
|
227 |
+
num_elem = sum(s.numel() for s in frozen_param_shapes.values())
|
228 |
+
print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
|
229 |
+
|
230 |
+
wanted_params = len(frozen_param_shapes)
|
231 |
+
wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
|
232 |
+
avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
|
233 |
+
print(f'Frozen params: Have {avail_numel} numels to process.')
|
234 |
+
print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
|
235 |
+
|
236 |
+
total_params = 0
|
237 |
+
total_numel = 0
|
238 |
+
for name, shape in frozen_param_shapes.items():
|
239 |
+
total_params += 1
|
240 |
+
unpartitioned_numel = shape.numel()
|
241 |
+
total_numel += unpartitioned_numel
|
242 |
+
|
243 |
+
state_dict[name] = frozen_param_fragments[name]
|
244 |
+
|
245 |
+
if debug:
|
246 |
+
print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
|
247 |
+
|
248 |
+
print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
|
249 |
+
|
250 |
+
|
251 |
+
def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
|
252 |
+
param_shapes = zero_model_states[0].param_shapes
|
253 |
+
|
254 |
+
# Reconstruction protocol:
|
255 |
+
#
|
256 |
+
# XXX: document this
|
257 |
+
|
258 |
+
if debug:
|
259 |
+
for i in range(world_size):
|
260 |
+
for j in range(len(fp32_flat_groups[0])):
|
261 |
+
print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
|
262 |
+
|
263 |
+
# XXX: memory usage doubles here (zero2)
|
264 |
+
num_param_groups = len(fp32_flat_groups[0])
|
265 |
+
merged_single_partition_of_fp32_groups = []
|
266 |
+
for i in range(num_param_groups):
|
267 |
+
merged_partitions = [sd[i] for sd in fp32_flat_groups]
|
268 |
+
full_single_fp32_vector = torch.cat(merged_partitions, 0)
|
269 |
+
merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
|
270 |
+
avail_numel = sum(
|
271 |
+
[full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
|
272 |
+
|
273 |
+
if debug:
|
274 |
+
wanted_params = sum([len(shapes) for shapes in param_shapes])
|
275 |
+
wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
|
276 |
+
# not asserting if there is a mismatch due to possible padding
|
277 |
+
print(f"Have {avail_numel} numels to process.")
|
278 |
+
print(f"Need {wanted_numel} numels in {wanted_params} params.")
|
279 |
+
|
280 |
+
# params
|
281 |
+
# XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
|
282 |
+
# out-of-core computing solution
|
283 |
+
total_numel = 0
|
284 |
+
total_params = 0
|
285 |
+
for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
|
286 |
+
offset = 0
|
287 |
+
avail_numel = full_single_fp32_vector.numel()
|
288 |
+
for name, shape in shapes.items():
|
289 |
+
|
290 |
+
unpartitioned_numel = shape.numel()
|
291 |
+
total_numel += unpartitioned_numel
|
292 |
+
total_params += 1
|
293 |
+
|
294 |
+
if debug:
|
295 |
+
print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
|
296 |
+
state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
|
297 |
+
offset += unpartitioned_numel
|
298 |
+
|
299 |
+
# Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
|
300 |
+
# avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
|
301 |
+
# paddings performed in the code it's almost impossible to predict the exact numbers w/o the
|
302 |
+
# live optimizer object, so we are checking that the numbers are within the right range
|
303 |
+
align_to = 2 * world_size
|
304 |
+
|
305 |
+
def zero2_align(x):
|
306 |
+
return align_to * math.ceil(x / align_to)
|
307 |
+
|
308 |
+
if debug:
|
309 |
+
print(f"original offset={offset}, avail_numel={avail_numel}")
|
310 |
+
|
311 |
+
offset = zero2_align(offset)
|
312 |
+
avail_numel = zero2_align(avail_numel)
|
313 |
+
|
314 |
+
if debug:
|
315 |
+
print(f"aligned offset={offset}, avail_numel={avail_numel}")
|
316 |
+
|
317 |
+
# Sanity check
|
318 |
+
if offset != avail_numel:
|
319 |
+
raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
|
320 |
+
|
321 |
+
print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
|
322 |
+
|
323 |
+
|
324 |
+
def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states):
|
325 |
+
state_dict = OrderedDict()
|
326 |
+
|
327 |
+
# buffers
|
328 |
+
buffers = zero_model_states[0].buffers
|
329 |
+
state_dict.update(buffers)
|
330 |
+
if debug:
|
331 |
+
print(f"added {len(buffers)} buffers")
|
332 |
+
|
333 |
+
_zero2_merge_frozen_params(state_dict, zero_model_states)
|
334 |
+
|
335 |
+
_zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
|
336 |
+
|
337 |
+
# recover shared parameters
|
338 |
+
for pair in zero_model_states[0].shared_params:
|
339 |
+
if pair[1] in state_dict:
|
340 |
+
state_dict[pair[0]] = state_dict[pair[1]]
|
341 |
+
|
342 |
+
return state_dict
|
343 |
+
|
344 |
+
|
345 |
+
def zero3_partitioned_param_info(unpartitioned_numel, world_size):
|
346 |
+
remainder = unpartitioned_numel % world_size
|
347 |
+
padding_numel = (world_size - remainder) if remainder else 0
|
348 |
+
partitioned_numel = math.ceil(unpartitioned_numel / world_size)
|
349 |
+
return partitioned_numel, padding_numel
|
350 |
+
|
351 |
+
|
352 |
+
def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
|
353 |
+
if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
|
354 |
+
return
|
355 |
+
|
356 |
+
if debug:
|
357 |
+
for i in range(world_size):
|
358 |
+
num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
|
359 |
+
print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
|
360 |
+
|
361 |
+
frozen_param_shapes = zero_model_states[0].frozen_param_shapes
|
362 |
+
wanted_params = len(frozen_param_shapes)
|
363 |
+
wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
|
364 |
+
avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
|
365 |
+
print(f'Frozen params: Have {avail_numel} numels to process.')
|
366 |
+
print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
|
367 |
+
|
368 |
+
total_params = 0
|
369 |
+
total_numel = 0
|
370 |
+
for name, shape in zero_model_states[0].frozen_param_shapes.items():
|
371 |
+
total_params += 1
|
372 |
+
unpartitioned_numel = shape.numel()
|
373 |
+
total_numel += unpartitioned_numel
|
374 |
+
|
375 |
+
param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
|
376 |
+
state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
|
377 |
+
|
378 |
+
partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
|
379 |
+
|
380 |
+
if debug:
|
381 |
+
print(
|
382 |
+
f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
|
383 |
+
)
|
384 |
+
|
385 |
+
print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
|
386 |
+
|
387 |
+
|
388 |
+
def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
|
389 |
+
param_shapes = zero_model_states[0].param_shapes
|
390 |
+
avail_numel = fp32_flat_groups[0].numel() * world_size
|
391 |
+
# Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
|
392 |
+
# param, re-consolidating each param, while dealing with padding if any
|
393 |
+
|
394 |
+
# merge list of dicts, preserving order
|
395 |
+
param_shapes = {k: v for d in param_shapes for k, v in d.items()}
|
396 |
+
|
397 |
+
if debug:
|
398 |
+
for i in range(world_size):
|
399 |
+
print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
|
400 |
+
|
401 |
+
wanted_params = len(param_shapes)
|
402 |
+
wanted_numel = sum(shape.numel() for shape in param_shapes.values())
|
403 |
+
# not asserting if there is a mismatch due to possible padding
|
404 |
+
avail_numel = fp32_flat_groups[0].numel() * world_size
|
405 |
+
print(f"Trainable params: Have {avail_numel} numels to process.")
|
406 |
+
print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
|
407 |
+
|
408 |
+
# params
|
409 |
+
# XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
|
410 |
+
# out-of-core computing solution
|
411 |
+
offset = 0
|
412 |
+
total_numel = 0
|
413 |
+
total_params = 0
|
414 |
+
for name, shape in param_shapes.items():
|
415 |
+
|
416 |
+
unpartitioned_numel = shape.numel()
|
417 |
+
total_numel += unpartitioned_numel
|
418 |
+
total_params += 1
|
419 |
+
|
420 |
+
partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
|
421 |
+
|
422 |
+
if debug:
|
423 |
+
print(
|
424 |
+
f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
|
425 |
+
)
|
426 |
+
|
427 |
+
# XXX: memory usage doubles here
|
428 |
+
state_dict[name] = torch.cat(
|
429 |
+
tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
|
430 |
+
0).narrow(0, 0, unpartitioned_numel).view(shape)
|
431 |
+
offset += partitioned_numel
|
432 |
+
|
433 |
+
offset *= world_size
|
434 |
+
|
435 |
+
# Sanity check
|
436 |
+
if offset != avail_numel:
|
437 |
+
raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
|
438 |
+
|
439 |
+
print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
|
440 |
+
|
441 |
+
|
442 |
+
def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states):
|
443 |
+
state_dict = OrderedDict()
|
444 |
+
|
445 |
+
# buffers
|
446 |
+
buffers = zero_model_states[0].buffers
|
447 |
+
state_dict.update(buffers)
|
448 |
+
if debug:
|
449 |
+
print(f"added {len(buffers)} buffers")
|
450 |
+
|
451 |
+
_zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
|
452 |
+
|
453 |
+
_zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
|
454 |
+
|
455 |
+
# recover shared parameters
|
456 |
+
for pair in zero_model_states[0].shared_params:
|
457 |
+
if pair[1] in state_dict:
|
458 |
+
state_dict[pair[0]] = state_dict[pair[1]]
|
459 |
+
|
460 |
+
return state_dict
|
461 |
+
|
462 |
+
|
463 |
+
def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None):
|
464 |
+
"""
|
465 |
+
Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
|
466 |
+
``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
|
467 |
+
via a model hub.
|
468 |
+
|
469 |
+
Args:
|
470 |
+
- ``checkpoint_dir``: path to the desired checkpoint folder
|
471 |
+
- ``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``
|
472 |
+
|
473 |
+
Returns:
|
474 |
+
- pytorch ``state_dict``
|
475 |
+
|
476 |
+
Note: this approach may not work if your application doesn't have sufficient free CPU memory and
|
477 |
+
you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
|
478 |
+
the checkpoint.
|
479 |
+
|
480 |
+
A typical usage might be ::
|
481 |
+
|
482 |
+
from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
|
483 |
+
# do the training and checkpoint saving
|
484 |
+
state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
|
485 |
+
model = model.cpu() # move to cpu
|
486 |
+
model.load_state_dict(state_dict)
|
487 |
+
# submit to model hub or save the model to share with others
|
488 |
+
|
489 |
+
In this example the ``model`` will no longer be usable in the deepspeed context of the same
|
490 |
+
application. i.e. you will need to re-initialize the deepspeed engine, since
|
491 |
+
``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
|
492 |
+
|
493 |
+
If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
|
494 |
+
|
495 |
+
"""
|
496 |
+
if tag is None:
|
497 |
+
latest_path = os.path.join(checkpoint_dir, 'latest')
|
498 |
+
if os.path.isfile(latest_path):
|
499 |
+
with open(latest_path, 'r') as fd:
|
500 |
+
tag = fd.read().strip()
|
501 |
+
else:
|
502 |
+
raise ValueError(f"Unable to find 'latest' file at {latest_path}")
|
503 |
+
|
504 |
+
ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
|
505 |
+
|
506 |
+
if not os.path.isdir(ds_checkpoint_dir):
|
507 |
+
raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
|
508 |
+
|
509 |
+
return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir)
|
510 |
+
|
511 |
+
|
512 |
+
def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None):
|
513 |
+
"""
|
514 |
+
Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
|
515 |
+
loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
|
516 |
+
|
517 |
+
Args:
|
518 |
+
- ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
|
519 |
+
- ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin)
|
520 |
+
- ``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``
|
521 |
+
"""
|
522 |
+
|
523 |
+
state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
|
524 |
+
print(f"Saving fp32 state dict to {output_file}")
|
525 |
+
torch.save(state_dict, output_file)
|
526 |
+
|
527 |
+
|
528 |
+
def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
|
529 |
+
"""
|
530 |
+
1. Put the provided model to cpu
|
531 |
+
2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
|
532 |
+
3. Load it into the provided model
|
533 |
+
|
534 |
+
Args:
|
535 |
+
- ``model``: the model object to update
|
536 |
+
- ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
|
537 |
+
- ``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``
|
538 |
+
|
539 |
+
Returns:
|
540 |
+
- ``model`: modified model
|
541 |
+
|
542 |
+
Make sure you have plenty of CPU memory available before you call this function. If you don't
|
543 |
+
have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
|
544 |
+
conveniently placed for you in the checkpoint folder.
|
545 |
+
|
546 |
+
A typical usage might be ::
|
547 |
+
|
548 |
+
from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
|
549 |
+
model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
|
550 |
+
# submit to model hub or save the model to share with others
|
551 |
+
|
552 |
+
Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
|
553 |
+
of the same application. i.e. you will need to re-initialize the deepspeed engine, since
|
554 |
+
``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
|
555 |
+
|
556 |
+
"""
|
557 |
+
logger.info(f"Extracting fp32 weights")
|
558 |
+
state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
|
559 |
+
|
560 |
+
logger.info(f"Overwriting model with fp32 weights")
|
561 |
+
model = model.cpu()
|
562 |
+
model.load_state_dict(state_dict, strict=False)
|
563 |
+
|
564 |
+
return model
|
565 |
+
|
566 |
+
|
567 |
+
if __name__ == "__main__":
|
568 |
+
|
569 |
+
parser = argparse.ArgumentParser()
|
570 |
+
parser.add_argument("checkpoint_dir",
|
571 |
+
type=str,
|
572 |
+
help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
|
573 |
+
parser.add_argument(
|
574 |
+
"output_file",
|
575 |
+
type=str,
|
576 |
+
help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)")
|
577 |
+
parser.add_argument("-t",
|
578 |
+
"--tag",
|
579 |
+
type=str,
|
580 |
+
default=None,
|
581 |
+
help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
|
582 |
+
parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
|
583 |
+
args = parser.parse_args()
|
584 |
+
|
585 |
+
debug = args.debug
|
586 |
+
|
587 |
+
convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, args.output_file, tag=args.tag)
|
checkpoint-4518/1_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-4518/README.md
ADDED
@@ -0,0 +1,57 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
library_name: sentence-transformers
|
3 |
+
pipeline_tag: sentence-similarity
|
4 |
+
tags:
|
5 |
+
- sentence-transformers
|
6 |
+
- feature-extraction
|
7 |
+
- sentence-similarity
|
8 |
+
|
9 |
+
---
|
10 |
+
|
11 |
+
# {MODEL_NAME}
|
12 |
+
|
13 |
+
This is a [sentence-transformers](https://www.SBERT.net) model: It maps sentences & paragraphs to a 1024 dimensional dense vector space and can be used for tasks like clustering or semantic search.
|
14 |
+
|
15 |
+
<!--- Describe your model here -->
|
16 |
+
|
17 |
+
## Usage (Sentence-Transformers)
|
18 |
+
|
19 |
+
Using this model becomes easy when you have [sentence-transformers](https://www.SBERT.net) installed:
|
20 |
+
|
21 |
+
```
|
22 |
+
pip install -U sentence-transformers
|
23 |
+
```
|
24 |
+
|
25 |
+
Then you can use the model like this:
|
26 |
+
|
27 |
+
```python
|
28 |
+
from sentence_transformers import SentenceTransformer
|
29 |
+
sentences = ["This is an example sentence", "Each sentence is converted"]
|
30 |
+
|
31 |
+
model = SentenceTransformer('{MODEL_NAME}')
|
32 |
+
embeddings = model.encode(sentences)
|
33 |
+
print(embeddings)
|
34 |
+
```
|
35 |
+
|
36 |
+
|
37 |
+
|
38 |
+
## Evaluation Results
|
39 |
+
|
40 |
+
<!--- Describe how your model was evaluated -->
|
41 |
+
|
42 |
+
For an automated evaluation of this model, see the *Sentence Embeddings Benchmark*: [https://seb.sbert.net](https://seb.sbert.net?model_name={MODEL_NAME})
|
43 |
+
|
44 |
+
|
45 |
+
|
46 |
+
## Full Model Architecture
|
47 |
+
```
|
48 |
+
SentenceTransformer(
|
49 |
+
(0): Transformer({'max_seq_length': 8192, 'do_lower_case': False}) with Transformer model: XLMRobertaModel
|
50 |
+
(1): Pooling({'word_embedding_dimension': 1024, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})
|
51 |
+
(2): Normalize()
|
52 |
+
)
|
53 |
+
```
|
54 |
+
|
55 |
+
## Citing & Authors
|
56 |
+
|
57 |
+
<!--- Describe where people can find more information -->
|
checkpoint-4518/colbert_linear.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:babc29a1ac86ee527a72a0b0fd28b100ef35d8c6e6742fea1809dab1030c6565
|
3 |
+
size 2100227
|
checkpoint-4518/config.json
ADDED
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "saved_models/bgem3_unified_finetune_20240330/checkpoint-4518",
|
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": "4.39.1",
|
25 |
+
"type_vocab_size": 1,
|
26 |
+
"use_cache": true,
|
27 |
+
"vocab_size": 250002
|
28 |
+
}
|
checkpoint-4518/config_sentence_transformers.json
ADDED
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"__version__": {
|
3 |
+
"sentence_transformers": "2.6.0",
|
4 |
+
"transformers": "4.39.1",
|
5 |
+
"pytorch": "2.0.1+cu117"
|
6 |
+
},
|
7 |
+
"prompts": {},
|
8 |
+
"default_prompt_name": null
|
9 |
+
}
|