Sentence Similarity
sentence-transformers
PyTorch
ONNX
bert
feature-extraction
Inference Endpoints
rawsh commited on
Commit
ac8480c
1 Parent(s): 4a3b479

Upload 15 files

Browse files
1_Pooling/config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "word_embedding_dimension": 384,
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 ADDED
@@ -0,0 +1,249 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ pipeline_tag: sentence-similarity
3
+ tags:
4
+ - sentence-transformers
5
+ - feature-extraction
6
+ - sentence-similarity
7
+ datasets:
8
+ - flax-sentence-embeddings/stackexchange_xml
9
+ - ms_marco
10
+ - gooaq
11
+ - yahoo_answers_topics
12
+ - search_qa
13
+ - eli5
14
+ - natural_questions
15
+ - trivia_qa
16
+ - embedding-data/QQP
17
+ - embedding-data/PAQ_pairs
18
+ - embedding-data/Amazon-QA
19
+ - embedding-data/WikiAnswers
20
+ ---
21
+
22
+ # multi-qa-MiniLM-distill-onnx-L6-cos-v1
23
+ This is a [sentence-transformers](https://www.SBERT.net) model: It maps sentences & paragraphs to a 384 dimensional dense vector space and was designed for **semantic search**. It has been trained on 215M (question, answer) pairs from diverse sources. For an introduction to semantic search, have a look at: [SBERT.net - Semantic Search](https://www.sbert.net/examples/applications/semantic-search/README.html)
24
+
25
+
26
+ ## Usage (Sentence-Transformers)
27
+ Using this model becomes easy when you have [sentence-transformers](https://www.SBERT.net) installed:
28
+
29
+ ```
30
+ pip install -U sentence-transformers
31
+ ```
32
+
33
+ Then you can use the model like this:
34
+ ```python
35
+ from sentence_transformers import SentenceTransformer, util
36
+
37
+ query = "How many people live in London?"
38
+ docs = ["Around 9 Million people live in London", "London is known for its financial district"]
39
+
40
+ #Load the model
41
+ model = SentenceTransformer('sentence-transformers/multi-qa-MiniLM-L6-cos-v1')
42
+
43
+ #Encode query and documents
44
+ query_emb = model.encode(query)
45
+ doc_emb = model.encode(docs)
46
+
47
+ #Compute dot score between query and all document embeddings
48
+ scores = util.dot_score(query_emb, doc_emb)[0].cpu().tolist()
49
+
50
+ #Combine docs & scores
51
+ doc_score_pairs = list(zip(docs, scores))
52
+
53
+ #Sort by decreasing score
54
+ doc_score_pairs = sorted(doc_score_pairs, key=lambda x: x[1], reverse=True)
55
+
56
+ #Output passages & scores
57
+ for doc, score in doc_score_pairs:
58
+ print(score, doc)
59
+ ```
60
+
61
+
62
+ ## PyTorch Usage (HuggingFace Transformers)
63
+ 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 correct pooling-operation on-top of the contextualized word embeddings.
64
+
65
+ ```python
66
+ from transformers import AutoTokenizer, AutoModel
67
+ import torch
68
+ import torch.nn.functional as F
69
+
70
+ #Mean Pooling - Take average of all tokens
71
+ def mean_pooling(model_output, attention_mask):
72
+ token_embeddings = model_output.last_hidden_state
73
+ input_mask_expanded = attention_mask.unsqueeze(-1).expand(token_embeddings.size()).float()
74
+ return torch.sum(token_embeddings * input_mask_expanded, 1) / torch.clamp(input_mask_expanded.sum(1), min=1e-9)
75
+
76
+
77
+ #Encode text
78
+ def encode(texts):
79
+ # Tokenize sentences
80
+ encoded_input = tokenizer(texts, padding=True, truncation=True, return_tensors='pt')
81
+
82
+ # Compute token embeddings
83
+ with torch.no_grad():
84
+ model_output = model(**encoded_input, return_dict=True)
85
+
86
+ # Perform pooling
87
+ embeddings = mean_pooling(model_output, encoded_input['attention_mask'])
88
+
89
+ # Normalize embeddings
90
+ embeddings = F.normalize(embeddings, p=2, dim=1)
91
+
92
+ return embeddings
93
+
94
+
95
+ # Sentences we want sentence embeddings for
96
+ query = "How many people live in London?"
97
+ docs = ["Around 9 Million people live in London", "London is known for its financial district"]
98
+
99
+ # Load model from HuggingFace Hub
100
+ tokenizer = AutoTokenizer.from_pretrained("sentence-transformers/multi-qa-MiniLM-L6-cos-v1")
101
+ model = AutoModel.from_pretrained("sentence-transformers/multi-qa-MiniLM-L6-cos-v1")
102
+
103
+ #Encode query and docs
104
+ query_emb = encode(query)
105
+ doc_emb = encode(docs)
106
+
107
+ #Compute dot score between query and all document embeddings
108
+ scores = torch.mm(query_emb, doc_emb.transpose(0, 1))[0].cpu().tolist()
109
+
110
+ #Combine docs & scores
111
+ doc_score_pairs = list(zip(docs, scores))
112
+
113
+ #Sort by decreasing score
114
+ doc_score_pairs = sorted(doc_score_pairs, key=lambda x: x[1], reverse=True)
115
+
116
+ #Output passages & scores
117
+ for doc, score in doc_score_pairs:
118
+ print(score, doc)
119
+ ```
120
+
121
+ ## TensorFlow Usage (HuggingFace Transformers)
122
+ Similarly to the PyTorch example above, to use the model with TensorFlow you pass your input through the transformer model, then you have to apply the correct pooling-operation on-top of the contextualized word embeddings.
123
+
124
+ ```python
125
+ from transformers import AutoTokenizer, TFAutoModel
126
+ import tensorflow as tf
127
+
128
+ #Mean Pooling - Take attention mask into account for correct averaging
129
+ def mean_pooling(model_output, attention_mask):
130
+ token_embeddings = model_output.last_hidden_state
131
+ input_mask_expanded = tf.cast(tf.tile(tf.expand_dims(attention_mask, -1), [1, 1, token_embeddings.shape[-1]]), tf.float32)
132
+ return tf.math.reduce_sum(token_embeddings * input_mask_expanded, 1) / tf.math.maximum(tf.math.reduce_sum(input_mask_expanded, 1), 1e-9)
133
+
134
+
135
+ #Encode text
136
+ def encode(texts):
137
+ # Tokenize sentences
138
+ encoded_input = tokenizer(texts, padding=True, truncation=True, return_tensors='tf')
139
+
140
+ # Compute token embeddings
141
+ model_output = model(**encoded_input, return_dict=True)
142
+
143
+ # Perform pooling
144
+ embeddings = mean_pooling(model_output, encoded_input['attention_mask'])
145
+
146
+ # Normalize embeddings
147
+ embeddings = tf.math.l2_normalize(embeddings, axis=1)
148
+
149
+ return embeddings
150
+
151
+
152
+ # Sentences we want sentence embeddings for
153
+ query = "How many people live in London?"
154
+ docs = ["Around 9 Million people live in London", "London is known for its financial district"]
155
+
156
+ # Load model from HuggingFace Hub
157
+ tokenizer = AutoTokenizer.from_pretrained("sentence-transformers/multi-qa-MiniLM-L6-cos-v1")
158
+ model = TFAutoModel.from_pretrained("sentence-transformers/multi-qa-MiniLM-L6-cos-v1")
159
+
160
+ #Encode query and docs
161
+ query_emb = encode(query)
162
+ doc_emb = encode(docs)
163
+
164
+ #Compute dot score between query and all document embeddings
165
+ scores = (query_emb @ tf.transpose(doc_emb))[0].numpy().tolist()
166
+
167
+ #Combine docs & scores
168
+ doc_score_pairs = list(zip(docs, scores))
169
+
170
+ #Sort by decreasing score
171
+ doc_score_pairs = sorted(doc_score_pairs, key=lambda x: x[1], reverse=True)
172
+
173
+ #Output passages & scores
174
+ for doc, score in doc_score_pairs:
175
+ print(score, doc)
176
+ ```
177
+
178
+ ## Technical Details
179
+
180
+ In the following some technical details how this model must be used:
181
+
182
+ | Setting | Value |
183
+ | --- | :---: |
184
+ | Dimensions | 384 |
185
+ | Produces normalized embeddings | Yes |
186
+ | Pooling-Method | Mean pooling |
187
+ | Suitable score functions | dot-product (`util.dot_score`), cosine-similarity (`util.cos_sim`), or euclidean distance |
188
+
189
+ Note: When loaded with `sentence-transformers`, this model produces normalized embeddings with length 1. In that case, dot-product and cosine-similarity are equivalent. dot-product is preferred as it is faster. Euclidean distance is proportional to dot-product and can also be used.
190
+
191
+ ----
192
+
193
+
194
+ ## Background
195
+
196
+ The project aims to train sentence embedding models on very large sentence level datasets using a self-supervised
197
+ contrastive learning objective. We use a contrastive learning objective: given a sentence from the pair, the model should predict which out of a set of randomly sampled other sentences, was actually paired with it in our dataset.
198
+
199
+ We developped this model during the
200
+ [Community week using JAX/Flax for NLP & CV](https://discuss.huggingface.co/t/open-to-the-community-community-week-using-jax-flax-for-nlp-cv/7104),
201
+ organized by Hugging Face. We developped this model as part of the project:
202
+ [Train the Best Sentence Embedding Model Ever with 1B Training Pairs](https://discuss.huggingface.co/t/train-the-best-sentence-embedding-model-ever-with-1b-training-pairs/7354). We benefited from efficient hardware infrastructure to run the project: 7 TPUs v3-8, as well as intervention from Googles Flax, JAX, and Cloud team member about efficient deep learning frameworks.
203
+
204
+ ## Intended uses
205
+
206
+ Our model is intented to be used for semantic search: It encodes queries / questions and text paragraphs in a dense vector space. It finds relevant documents for the given passages.
207
+
208
+ Note that there is a limit of 512 word pieces: Text longer than that will be truncated. Further note that the model was just trained on input text up to 250 word pieces. It might not work well for longer text.
209
+
210
+
211
+
212
+ ## Training procedure
213
+
214
+ The full training script is accessible in this current repository: `train_script.py`.
215
+
216
+ ### Pre-training
217
+
218
+ We use the pretrained [`nreimers/MiniLM-L6-H384-uncased`](https://huggingface.co/nreimers/MiniLM-L6-H384-uncased) model. Please refer to the model card for more detailed information about the pre-training procedure.
219
+
220
+ #### Training
221
+
222
+ We use the concatenation from multiple datasets to fine-tune our model. In total we have about 215M (question, answer) pairs.
223
+ We sampled each dataset given a weighted probability which configuration is detailed in the `data_config.json` file.
224
+
225
+ The model was trained with [MultipleNegativesRankingLoss](https://www.sbert.net/docs/package_reference/losses.html#multiplenegativesrankingloss) using Mean-pooling, cosine-similarity as similarity function, and a scale of 20.
226
+
227
+
228
+
229
+
230
+ | Dataset | Number of training tuples |
231
+ |--------------------------------------------------------|:--------------------------:|
232
+ | [WikiAnswers](https://github.com/afader/oqa#wikianswers-corpus) Duplicate question pairs from WikiAnswers | 77,427,422 |
233
+ | [PAQ](https://github.com/facebookresearch/PAQ) Automatically generated (Question, Paragraph) pairs for each paragraph in Wikipedia | 64,371,441 |
234
+ | [Stack Exchange](https://huggingface.co/datasets/flax-sentence-embeddings/stackexchange_xml) (Title, Body) pairs from all StackExchanges | 25,316,456 |
235
+ | [Stack Exchange](https://huggingface.co/datasets/flax-sentence-embeddings/stackexchange_xml) (Title, Answer) pairs from all StackExchanges | 21,396,559 |
236
+ | [MS MARCO](https://microsoft.github.io/msmarco/) Triplets (query, answer, hard_negative) for 500k queries from Bing search engine | 17,579,773 |
237
+ | [GOOAQ: Open Question Answering with Diverse Answer Types](https://github.com/allenai/gooaq) (query, answer) pairs for 3M Google queries and Google featured snippet | 3,012,496 |
238
+ | [Amazon-QA](http://jmcauley.ucsd.edu/data/amazon/qa/) (Question, Answer) pairs from Amazon product pages | 2,448,839
239
+ | [Yahoo Answers](https://www.kaggle.com/soumikrakshit/yahoo-answers-dataset) (Title, Answer) pairs from Yahoo Answers | 1,198,260 |
240
+ | [Yahoo Answers](https://www.kaggle.com/soumikrakshit/yahoo-answers-dataset) (Question, Answer) pairs from Yahoo Answers | 681,164 |
241
+ | [Yahoo Answers](https://www.kaggle.com/soumikrakshit/yahoo-answers-dataset) (Title, Question) pairs from Yahoo Answers | 659,896 |
242
+ | [SearchQA](https://huggingface.co/datasets/search_qa) (Question, Answer) pairs for 140k questions, each with Top5 Google snippets on that question | 582,261 |
243
+ | [ELI5](https://huggingface.co/datasets/eli5) (Question, Answer) pairs from Reddit ELI5 (explainlikeimfive) | 325,475 |
244
+ | [Stack Exchange](https://huggingface.co/datasets/flax-sentence-embeddings/stackexchange_xml) Duplicate questions pairs (titles) | 304,525 |
245
+ | [Quora Question Triplets](https://quoradata.quora.com/First-Quora-Dataset-Release-Question-Pairs) (Question, Duplicate_Question, Hard_Negative) triplets for Quora Questions Pairs dataset | 103,663 |
246
+ | [Natural Questions (NQ)](https://ai.google.com/research/NaturalQuestions) (Question, Paragraph) pairs for 100k real Google queries with relevant Wikipedia paragraph | 100,231 |
247
+ | [SQuAD2.0](https://rajpurkar.github.io/SQuAD-explorer/) (Question, Paragraph) pairs from SQuAD2.0 dataset | 87,599 |
248
+ | [TriviaQA](https://huggingface.co/datasets/trivia_qa) (Question, Evidence) pairs | 73,346 |
249
+ | **Total** | **214,988,242** |
config.json ADDED
@@ -0,0 +1,25 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "models/cos-v1-best",
3
+ "architectures": [
4
+ "BertModel"
5
+ ],
6
+ "attention_probs_dropout_prob": 0.1,
7
+ "classifier_dropout": null,
8
+ "gradient_checkpointing": false,
9
+ "hidden_act": "gelu",
10
+ "hidden_dropout_prob": 0.1,
11
+ "hidden_size": 384,
12
+ "initializer_range": 0.02,
13
+ "intermediate_size": 1536,
14
+ "layer_norm_eps": 1e-12,
15
+ "max_position_embeddings": 512,
16
+ "model_type": "bert",
17
+ "num_attention_heads": 12,
18
+ "num_hidden_layers": 4,
19
+ "pad_token_id": 0,
20
+ "position_embedding_type": "absolute",
21
+ "transformers_version": "4.29.2",
22
+ "type_vocab_size": 2,
23
+ "use_cache": true,
24
+ "vocab_size": 30522
25
+ }
config_sentence_transformers.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "__version__": {
3
+ "sentence_transformers": "2.0.0",
4
+ "transformers": "4.6.1",
5
+ "pytorch": "1.8.1"
6
+ }
7
+ }
eval/mse_evaluation__results.csv ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ epoch,steps,MSE
2
+ 0,5000,0.08164194878190756
3
+ 0,10000,0.06260087830014527
4
+ 0,15000,0.053438334725797176
5
+ 0,20000,0.04801056638825685
6
+ 0,25000,0.0445966434199363
7
+ 0,30000,0.04211385385133326
8
+ 0,35000,0.040131498826667666
9
+ 0,40000,0.03866856568492949
10
+ 0,45000,0.037445276393555105
11
+ 0,50000,0.03652568848337978
12
+ 0,55000,0.03562064084690064
13
+ 0,60000,0.03487043140921742
14
+ 0,65000,0.03426450421102345
15
+ 0,70000,0.03378996916580945
16
+ 0,75000,0.033300244831480086
17
+ 0,80000,0.032820276101119816
18
+ 0,85000,0.03252678434364498
19
+ 0,90000,0.032118355738930404
20
+ 0,95000,0.03190774004906416
21
+ 0,100000,0.031571456929668784
22
+ 0,105000,0.03135171136818826
23
+ 0,110000,0.031201684032566845
24
+ 0,115000,0.03108016971964389
25
+ 0,120000,0.030856009107083082
26
+ 0,125000,0.030771695310249925
27
+ 0,130000,0.03066081553697586
28
+ 0,135000,0.030605780193582177
29
+ 0,140000,0.030567956855520606
30
+ 0,-1,0.030567424255423248
eval/similarity_evaluation_sts-dev_results.csv ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ epoch,steps,cosine_pearson,cosine_spearman,euclidean_pearson,euclidean_spearman,manhattan_pearson,manhattan_spearman,dot_pearson,dot_spearman
2
+ 0,5000,0.7451538219966819,0.7672554362824703,0.7663367711499433,0.7672554362824703,0.7656859380863119,0.7665947792435013,0.7451538231648414,0.7672554362824703
3
+ 0,10000,0.7537744166831953,0.7727486754280707,0.7728111123629107,0.7727486754280707,0.772113563269406,0.7721335168433461,0.7537744171689775,0.7727495539481395
4
+ 0,15000,0.7660933742804406,0.7837588700618559,0.7836123904512655,0.7837588700618559,0.7829209061843033,0.7830875130305814,0.7660933801320919,0.7837588693651809
5
+ 0,20000,0.7684300833864042,0.7855982699077577,0.785532933626244,0.7855982692094479,0.7845338039609955,0.7845750033796356,0.7684300882853332,0.7855982699077577
6
+ 0,25000,0.7707353950224363,0.7876847428779042,0.7876465143907596,0.7876847421777396,0.7867830462833947,0.7869706516377712,0.7707354006643212,0.7876847428779042
7
+ 0,30000,0.77288987781661,0.7890679662524103,0.7890394555299924,0.7890679655510161,0.788137743070403,0.7882666447140309,0.7728898769077066,0.7890677146611634
8
+ 0,35000,0.7714964043132171,0.7879360079865678,0.7881483073210547,0.7879371284726863,0.7872797682613883,0.7873248066864478,0.7714964114116738,0.7879371284726863
9
+ 0,40000,0.7738038431607456,0.7900336982355982,0.7900015695492877,0.7900336975333457,0.789036404739011,0.7893954248514412,0.7738038443211506,0.7900336982355982
10
+ 0,45000,0.7806805368603946,0.7953170748402448,0.7953952611135355,0.7953170748402448,0.794482786961417,0.7946208003390048,0.7806805365188378,0.795316296004493
11
+ 0,50000,0.7758800035273927,0.7920287779042389,0.7917483214037102,0.792028777200213,0.7907080715200523,0.791222195497074,0.7758800064681861,0.7920287779042389
12
+ 0,55000,0.7748974633837462,0.7907568393040405,0.7905916308754704,0.7907568386011451,0.7897169211995143,0.7899495391123526,0.7748974644641963,0.7907568393040405
13
+ 0,60000,0.7784748672436804,0.7939635781718496,0.793964252540622,0.7939635774661037,0.7929685054517607,0.7930615590508188,0.7784748678309645,0.7939635774661037
14
+ 0,65000,0.7796288427258389,0.7951236643816901,0.7950697460353463,0.7951236636749132,0.7942962171276358,0.7945430404989343,0.7796288448617663,0.7951236643816901
15
+ 0,70000,0.7796424798426436,0.7952969563162274,0.7948492536277543,0.7952969556092964,0.7939768644705751,0.7944538553840851,0.7796424862799302,0.7952978166371057
16
+ 0,75000,0.7769598727458167,0.7927814277048575,0.7922166093080583,0.7927823631288534,0.7915981847762255,0.7921869463483725,0.7769598789365952,0.7927814277048575
17
+ 0,80000,0.7774742857996642,0.7934620778257588,0.7931720059341743,0.7934620771204588,0.792433728896597,0.7927788061869001,0.7774742937530655,0.7934620778257588
18
+ 0,85000,0.7778284283777634,0.7933801275912681,0.7931055150588028,0.793380126886041,0.7923659661509699,0.7928011869852436,0.7778284197659264,0.7933805724314327
19
+ 0,90000,0.7778190920123474,0.7932520598872288,0.7927980747236021,0.7932520591821155,0.7919765295341479,0.7923518432591761,0.7778190862553716,0.7932520591821155
20
+ 0,95000,0.7795342361348376,0.7951428547863926,0.7947477485866722,0.7951428540795986,0.793933820785105,0.794293699396111,0.7795342378034608,0.7951415618979867
21
+ 0,100000,0.7811558920026177,0.7964721083727638,0.7959974871261926,0.7964721083727638,0.7953036438517008,0.7956409736399953,0.7811558906487714,0.7964721083727638
22
+ 0,105000,0.7789333814592223,0.7941989626759279,0.7937732318420716,0.7941989619699729,0.793105607100401,0.7936110345184971,0.7789333777991321,0.7941989626759279
23
+ 0,110000,0.7790156686608669,0.7944607101875235,0.7939863009721635,0.7944607094813357,0.7932901632304183,0.7937474314096579,0.7790156623433281,0.7944607094813357
24
+ 0,115000,0.7773604697868309,0.7928823123011013,0.7925392387219641,0.7928823123011013,0.7918524555529456,0.7924313931360675,0.7773604674828332,0.7928823115963166
25
+ 0,120000,0.7797444066117021,0.7954454544567308,0.7948229695790241,0.7954453169215147,0.7941276931152423,0.7945565785810856,0.779744398705051,0.7954454537496678
26
+ 0,125000,0.7794772857230747,0.7950229169812024,0.79449104446514,0.7950229169812024,0.7937674344847226,0.79436565603504,0.7794772873884481,0.7950229169812024
27
+ 0,130000,0.7802911181903579,0.7956809903890825,0.7951756431049894,0.7956809903890825,0.794476798728313,0.7950676316078248,0.7802911189881387,0.7956809903890825
28
+ 0,135000,0.7798078498424627,0.7951594686768616,0.7947034191959423,0.7951594679700529,0.7939996232677503,0.7945372807084875,0.7798078491935548,0.7951594686768616
29
+ 0,140000,0.7795179285289633,0.7949521909537459,0.7944688530241,0.7949521902471214,0.7937473834870582,0.7943429424962535,0.7795179357332032,0.7949521902471214
30
+ 0,-1,0.7795472646980206,0.7949874080599001,0.7944970293639978,0.7949874073532442,0.7937810593147978,0.7944118411202525,0.7795472601861694,0.7949874080599001
model.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:897920e276fe3222f214b4b4e2c57cd0165b0e999d0d0b67e086ac5eac58522f
3
+ size 76059578
model_quantized.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:46453e4d5c32d5b0691bc353a62c60186f1db5dca3e89ff36b6e007519b8334a
3
+ size 19161868
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
+ ]
ort_config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "one_external_file": true,
3
+ "opset": null,
4
+ "optimization": {
5
+ "disable_attention": null,
6
+ "disable_attention_fusion": false,
7
+ "disable_bias_gelu": null,
8
+ "disable_bias_gelu_fusion": false,
9
+ "disable_bias_skip_layer_norm": null,
10
+ "disable_bias_skip_layer_norm_fusion": false,
11
+ "disable_embed_layer_norm": true,
12
+ "disable_embed_layer_norm_fusion": true,
13
+ "disable_gelu": null,
14
+ "disable_gelu_fusion": false,
15
+ "disable_group_norm_fusion": true,
16
+ "disable_layer_norm": null,
17
+ "disable_layer_norm_fusion": false,
18
+ "disable_packed_kv": true,
19
+ "disable_shape_inference": false,
20
+ "disable_skip_layer_norm": null,
21
+ "disable_skip_layer_norm_fusion": false,
22
+ "enable_gelu_approximation": true,
23
+ "enable_gemm_fast_gelu_fusion": false,
24
+ "enable_transformers_specific_optimizations": true,
25
+ "fp16": false,
26
+ "no_attention_mask": false,
27
+ "optimization_level": 2,
28
+ "optimize_for_gpu": false,
29
+ "optimize_with_onnxruntime_only": null,
30
+ "use_mask_index": false,
31
+ "use_multi_head_attention": false,
32
+ "use_raw_attention_mask": false
33
+ },
34
+ "optimum_version": "1.8.7.dev0",
35
+ "quantization": {},
36
+ "transformers_version": "4.29.2",
37
+ "use_external_data_format": false
38
+ }
sentence_bert_config.json ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ {
2
+ "max_seq_length": 512,
3
+ "do_lower_case": false
4
+ }
special_tokens_map.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "cls_token": "[CLS]",
3
+ "mask_token": "[MASK]",
4
+ "pad_token": "[PAD]",
5
+ "sep_token": "[SEP]",
6
+ "unk_token": "[UNK]"
7
+ }
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json ADDED
@@ -0,0 +1,15 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "clean_up_tokenization_spaces": true,
3
+ "cls_token": "[CLS]",
4
+ "do_basic_tokenize": true,
5
+ "do_lower_case": true,
6
+ "mask_token": "[MASK]",
7
+ "model_max_length": 512,
8
+ "never_split": null,
9
+ "pad_token": "[PAD]",
10
+ "sep_token": "[SEP]",
11
+ "strip_accents": null,
12
+ "tokenize_chinese_chars": true,
13
+ "tokenizer_class": "BertTokenizer",
14
+ "unk_token": "[UNK]"
15
+ }
vocab.txt ADDED
The diff for this file is too large to render. See raw diff