nreimers commited on
Commit
2739dcf
1 Parent(s): 313db17

Add new SentenceTransformer model.

Browse files
1_Pooling/config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "word_embedding_dimension": 768,
3
+ "pooling_mode_cls_token": false,
4
+ "pooling_mode_mean_tokens": true,
5
+ "pooling_mode_max_tokens": false,
6
+ "pooling_mode_mean_sqrt_len_tokens": false
7
+ }
README.md CHANGED
@@ -1,20 +1,47 @@
1
  ---
2
- language: en
3
  tags:
4
- - exbert
5
- license: apache-2.0
6
- datasets:
7
- - snli
8
- - multi_nli
 
 
 
 
9
  ---
10
 
11
- # BERT base model (uncased) for Sentence Embeddings
12
- This is the `bert-base-nli-mean-tokens` model from the [sentence-transformers](https://github.com/UKPLab/sentence-transformers)-repository. The sentence-transformers repository allows to train and use Transformer models for generating sentence and text embeddings.
13
- The model is described in the paper [Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks](https://arxiv.org/abs/1908.10084)
14
 
15
- ## Usage (HuggingFace Models Repository)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
16
 
17
- You can use the model directly from the model repository to compute sentence embeddings:
18
  ```python
19
  from transformers import AutoTokenizer, AutoModel
20
  import torch
@@ -24,55 +51,54 @@ import torch
24
  def mean_pooling(model_output, attention_mask):
25
  token_embeddings = model_output[0] #First element of model_output contains all token embeddings
26
  input_mask_expanded = attention_mask.unsqueeze(-1).expand(token_embeddings.size()).float()
27
- sum_embeddings = torch.sum(token_embeddings * input_mask_expanded, 1)
28
- sum_mask = torch.clamp(input_mask_expanded.sum(1), min=1e-9)
29
- return sum_embeddings / sum_mask
30
 
31
 
 
 
32
 
33
- #Sentences we want sentence embeddings for
34
- sentences = ['This framework generates embeddings for each input sentence',
35
- 'Sentences are passed as a list of string.',
36
- 'The quick brown fox jumps over the lazy dog.']
37
 
38
- #Load AutoModel from huggingface model repository
39
- tokenizer = AutoTokenizer.from_pretrained("sentence-transformers/bert-base-nli-mean-tokens")
40
- model = AutoModel.from_pretrained("sentence-transformers/bert-base-nli-mean-tokens")
41
 
42
- #Tokenize sentences
43
- encoded_input = tokenizer(sentences, padding=True, truncation=True, max_length=128, return_tensors='pt')
44
-
45
- #Compute token embeddings
46
  with torch.no_grad():
47
  model_output = model(**encoded_input)
48
 
49
- #Perform pooling. In this case, mean pooling
50
  sentence_embeddings = mean_pooling(model_output, encoded_input['attention_mask'])
51
- ```
52
-
53
- ## Usage (Sentence-Transformers)
54
- Using this model becomes more convenient when you have [sentence-transformers](https://github.com/UKPLab/sentence-transformers) installed:
55
- ```
56
- pip install -U sentence-transformers
57
- ```
58
-
59
- Then you can use the model like this:
60
- ```python
61
- from sentence_transformers import SentenceTransformer
62
- model = SentenceTransformer('bert-base-nli-mean-tokens')
63
- sentences = ['This framework generates embeddings for each input sentence',
64
- 'Sentences are passed as a list of string.',
65
- 'The quick brown fox jumps over the lazy dog.']
66
- sentence_embeddings = model.encode(sentences)
67
 
68
  print("Sentence embeddings:")
69
  print(sentence_embeddings)
70
  ```
71
 
72
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
73
  ## Citing & Authors
 
 
 
74
  If you find this model helpful, feel free to cite our publication [Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks](https://arxiv.org/abs/1908.10084):
75
- ```
76
  @inproceedings{reimers-2019-sentence-bert,
77
  title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks",
78
  author = "Reimers, Nils and Gurevych, Iryna",
@@ -82,4 +108,4 @@ If you find this model helpful, feel free to cite our publication [Sentence-BERT
82
  publisher = "Association for Computational Linguistics",
83
  url = "http://arxiv.org/abs/1908.10084",
84
  }
85
- ```
 
1
  ---
2
+ pipeline_tag: sentence-similarity
3
  tags:
4
+ - sentence-transformers
5
+ - feature-extraction
6
+ - sentence-similarity
7
+ - transformers
8
+ - transformers
9
+ - transformers
10
+ - transformers
11
+ - transformers
12
+ - transformers
13
  ---
14
 
15
+ # sentence-transformers/bert-base-nli-mean-tokens
 
 
16
 
17
+ This is a [sentence-transformers](https://www.SBERT.net) model: It maps sentences & paragraphs to a 768 dimensional dense vector space and can be used for tasks like clustering or semantic search.
18
+
19
+
20
+
21
+ ## Usage (Sentence-Transformers)
22
+
23
+ Using this model becomes easy when you have [sentence-transformers](https://www.SBERT.net) installed:
24
+
25
+ ```
26
+ pip install -U sentence-transformers
27
+ ```
28
+
29
+ Then you can use the model like this:
30
+
31
+ ```python
32
+ from sentence_transformers import SentenceTransformer
33
+ sentences = ["This is an example sentence", "Each sentence is converted"]
34
+
35
+ model = SentenceTransformer('sentence-transformers/bert-base-nli-mean-tokens')
36
+ embeddings = model.encode(sentences)
37
+ print(embeddings)
38
+ ```
39
+
40
+
41
+
42
+ ## Usage (HuggingFace Transformers)
43
+ Without [sentence-transformers](https://www.SBERT.net), you can use the model like this: First, you pass your input through the transformer model, then you have to apply the right pooling-operation on-top of the contextualized word embeddings.
44
 
 
45
  ```python
46
  from transformers import AutoTokenizer, AutoModel
47
  import torch
 
51
  def mean_pooling(model_output, attention_mask):
52
  token_embeddings = model_output[0] #First element of model_output contains all token embeddings
53
  input_mask_expanded = attention_mask.unsqueeze(-1).expand(token_embeddings.size()).float()
54
+ return torch.sum(token_embeddings * input_mask_expanded, 1) / torch.clamp(input_mask_expanded.sum(1), min=1e-9)
 
 
55
 
56
 
57
+ # Sentences we want sentence embeddings for
58
+ sentences = ['This is an example sentence', 'Each sentence is converted']
59
 
60
+ # Load model from HuggingFace Hub
61
+ tokenizer = AutoTokenizer.from_pretrained('sentence-transformers/bert-base-nli-mean-tokens')
62
+ model = AutoModel.from_pretrained('sentence-transformers/bert-base-nli-mean-tokens')
 
63
 
64
+ # Tokenize sentences
65
+ encoded_input = tokenizer(sentences, padding=True, truncation=True, return_tensors='pt')
 
66
 
67
+ # Compute token embeddings
 
 
 
68
  with torch.no_grad():
69
  model_output = model(**encoded_input)
70
 
71
+ # Perform pooling. In this case, max pooling.
72
  sentence_embeddings = mean_pooling(model_output, encoded_input['attention_mask'])
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
73
 
74
  print("Sentence embeddings:")
75
  print(sentence_embeddings)
76
  ```
77
 
78
 
79
+
80
+ ## Evaluation Results
81
+
82
+
83
+
84
+ For an automated evaluation of this model, see the *Sentence Embeddings Benchmark*: [https://seb.sbert.net](https://seb.sbert.net?model_name=sentence-transformers/bert-base-nli-mean-tokens)
85
+
86
+
87
+
88
+ ## Full Model Architecture
89
+ ```
90
+ SentenceTransformer(
91
+ (0): Transformer({'max_seq_length': 128, 'do_lower_case': False}) with Transformer model: BertModel
92
+ (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False})
93
+ )
94
+ ```
95
+
96
  ## Citing & Authors
97
+
98
+ This model was trained by [sentence-transformers](https://www.sbert.net/).
99
+
100
  If you find this model helpful, feel free to cite our publication [Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks](https://arxiv.org/abs/1908.10084):
101
+ ```bibtex
102
  @inproceedings{reimers-2019-sentence-bert,
103
  title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks",
104
  author = "Reimers, Nils and Gurevych, Iryna",
 
108
  publisher = "Association for Computational Linguistics",
109
  url = "http://arxiv.org/abs/1908.10084",
110
  }
111
+ ```
config.json CHANGED
@@ -1,4 +1,5 @@
1
  {
 
2
  "architectures": [
3
  "BertModel"
4
  ],
@@ -15,6 +16,9 @@
15
  "num_attention_heads": 12,
16
  "num_hidden_layers": 12,
17
  "pad_token_id": 0,
 
 
18
  "type_vocab_size": 2,
 
19
  "vocab_size": 30522
20
- }
 
1
  {
2
+ "_name_or_path": "old_models/bert-base-nli-mean-tokens/0_BERT",
3
  "architectures": [
4
  "BertModel"
5
  ],
 
16
  "num_attention_heads": 12,
17
  "num_hidden_layers": 12,
18
  "pad_token_id": 0,
19
+ "position_embedding_type": "absolute",
20
+ "transformers_version": "4.7.0",
21
  "type_vocab_size": 2,
22
+ "use_cache": true,
23
  "vocab_size": 30522
24
+ }
config_sentence_transformers.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "__version__": {
3
+ "sentence_transformers": "2.0.0",
4
+ "transformers": "4.7.0",
5
+ "pytorch": "1.9.0+cu102"
6
+ }
7
+ }
modules.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
+ ]
pytorch_model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:38a04a21a955808f60c22ba92f4375394f22e84b18ba90094b00a04c9ac8c9eb
3
- size 438006864
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:47a5ed8c3b0ea722db61d4db2b3caaec3f751e72e0ebddf39f59a3c8d9784385
3
+ size 438007537
sentence_bert_config.json CHANGED
@@ -1,3 +1,4 @@
1
  {
2
- "max_seq_length": 128
 
3
  }
 
1
  {
2
+ "max_seq_length": 128,
3
+ "do_lower_case": false
4
  }
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"do_lower_case": true, "unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": "[MASK]", "tokenize_chinese_chars": true, "strip_accents": null, "special_tokens_map_file": "old_models/bert-base-nli-mean-tokens/0_BERT/special_tokens_map.json", "name_or_path": "old_models/bert-base-nli-mean-tokens/0_BERT", "do_basic_tokenize": true, "never_split": null}