Hugging Face's logo

tomaarsen
/
NeoBERT-gooaq-8e-05

Sentence Similarity
sentence-transformers
Safetensors
English
neobert
feature-extraction
Generated from Trainer
dataset_size:3011496
loss:CachedMultipleNegativesRankingLoss
custom_code
Eval Results
Inference Endpoints
Model card Files Community
tomaarsen HF staff commited on
Commit
e34179d
·
verified ·
1 Parent(s): 16d0787

Add new SentenceTransformer model

Browse files
Files changed (11) hide show
  1. 1_Pooling/config.json +10 -0
  2. README.md +803 -0
  3. config.json +54 -0
  4. config_sentence_transformers.json +10 -0
  5. model.safetensors +3 -0
  6. modules.json +14 -0
  7. sentence_bert_config.json +4 -0
  8. special_tokens_map.json +37 -0
  9. tokenizer.json +0 -0
  10. tokenizer_config.json +61 -0
  11. vocab.txt +0 -0
1_Pooling/config.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
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
+ "pooling_mode_weightedmean_tokens": false,
8
+ "pooling_mode_lasttoken": false,
9
+ "include_prompt": true
10
+ }
README.md ADDED
@@ -0,0 +1,803 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ language:
3
+ - en
4
+ tags:
5
+ - sentence-transformers
6
+ - sentence-similarity
7
+ - feature-extraction
8
+ - generated_from_trainer
9
+ - dataset_size:3011496
10
+ - loss:CachedMultipleNegativesRankingLoss
11
+ base_model: chandar-lab/NeoBERT
12
+ widget:
13
+ - source_sentence: how much percent of alcohol is in scotch?
14
+ sentences:
15
+ - Our 24-hour day comes from the ancient Egyptians who divided day-time into 10
16
+ hours they measured with devices such as shadow clocks, and added a twilight hour
17
+ at the beginning and another one at the end of the day-time, says Lomb. "Night-time
18
+ was divided in 12 hours, based on the observations of stars.
19
+ - After distillation, a Scotch Whisky can be anywhere between 60-75% ABV, with American
20
+ Whiskey rocketing right into the 90% region. Before being placed in casks, Scotch
21
+ is usually diluted to around 63.5% ABV (68% for grain); welcome to the stage cask
22
+ strength Whisky.
23
+ - Money For Nothing. In season four Dominic West, the ostensible star of the series,
24
+ requested a reduced role so that he could spend more time with his family in London.
25
+ On the show it was explained that Jimmy McNulty had taken a patrol job which required
26
+ less strenuous work.
27
+ - source_sentence: what are the major causes of poor listening?
28
+ sentences:
29
+ - The four main causes of poor listening are due to not concentrating, listening
30
+ too hard, jumping to conclusions and focusing on delivery and personal appearance.
31
+ Sometimes we just don't feel attentive enough and hence don't concentrate.
32
+ - That's called being idle. “System Idle Process” is the software that runs when
33
+ the computer has absolutely nothing better to do. It has the lowest possible priority
34
+ and uses as few resources as possible, so that if anything at all comes along
35
+ for the CPU to work on, it can.
36
+ - 'No alcohol wine: how it''s made It''s not easy. There are three main methods
37
+ currently in use. Vacuum distillation sees alcohol and other volatiles removed
38
+ at a relatively low temperature (25°C-30°C), with aromatics blended back in afterwards.'
39
+ - source_sentence: are jess and justin still together?
40
+ sentences:
41
+ - Download photos and videos to your device On your iPhone, iPad, or iPod touch,
42
+ tap Settings > [your name] > iCloud > Photos. Then select Download and Keep Originals
43
+ and import the photos to your computer. On your Mac, open the Photos app. Select
44
+ the photos and videos you want to copy.
45
+ - Later, Justin reunites with Jessica at prom and the two get back together. ...
46
+ After a tearful goodbye to Jessica, the Jensens, and his friends, Justin dies
47
+ just before graduation.
48
+ - Incumbent president Muhammadu Buhari won his reelection bid, defeating his closest
49
+ rival Atiku Abubakar by over 3 million votes. He was issued a Certificate of Return,
50
+ and was sworn in on May 29, 2019, the former date of Democracy Day (Nigeria).
51
+ - source_sentence: when humans are depicted in hindu art?
52
+ sentences:
53
+ - 'Answer: Humans are depicted in Hindu art often in sensuous and erotic postures.'
54
+ - Bettas are carnivores. They require foods high in animal protein. Their preferred
55
+ diet in nature includes insects and insect larvae. In captivity, they thrive on
56
+ a varied diet of pellets or flakes made from fish meal, as well as frozen or freeze-dried
57
+ bloodworms.
58
+ - An active continental margin is found on the leading edge of the continent where
59
+ it is crashing into an oceanic plate. ... Passive continental margins are found
60
+ along the remaining coastlines.
61
+ - source_sentence: what is the difference between 18 and 20 inch tires?
62
+ sentences:
63
+ - '[''Alienware m17 R3. The best gaming laptop overall offers big power in slim,
64
+ redesigned chassis. ... '', ''Dell G3 15. ... '', ''Asus ROG Zephyrus G14. ...
65
+ '', ''Lenovo Legion Y545. ... '', ''Alienware Area 51m. ... '', ''Asus ROG Mothership.
66
+ ... '', ''Asus ROG Strix Scar III. ... '', ''HP Omen 17 (2019)'']'
67
+ - So extracurricular activities are just activities that you do outside of class.
68
+ The Common App says that extracurricular activities "include arts, athletics,
69
+ clubs, employment, personal commitments, and other pursuits."
70
+ - The only real difference is a 20" rim would be more likely to be damaged, as you
71
+ pointed out. Beyond looks, there is zero benefit for the 20" rim. Also, just the
72
+ availability of tires will likely be much more limited for the larger rim. ...
73
+ Tire selection is better for 18" wheels than 20" wheels.
74
+ datasets:
75
+ - sentence-transformers/gooaq
76
+ pipeline_tag: sentence-similarity
77
+ library_name: sentence-transformers
78
+ metrics:
79
+ - cosine_accuracy@1
80
+ - cosine_accuracy@3
81
+ - cosine_accuracy@5
82
+ - cosine_accuracy@10
83
+ - cosine_precision@1
84
+ - cosine_precision@3
85
+ - cosine_precision@5
86
+ - cosine_precision@10
87
+ - cosine_recall@1
88
+ - cosine_recall@3
89
+ - cosine_recall@5
90
+ - cosine_recall@10
91
+ - cosine_ndcg@10
92
+ - cosine_mrr@10
93
+ - cosine_map@100
94
+ model-index:
95
+ - name: SentenceTransformer based on chandar-lab/NeoBERT
96
+ results:
97
+ - task:
98
+ type: information-retrieval
99
+ name: Information Retrieval
100
+ dataset:
101
+ name: NanoNQ
102
+ type: NanoNQ
103
+ metrics:
104
+ - type: cosine_accuracy@1
105
+ value: 0.46
106
+ name: Cosine Accuracy@1
107
+ - type: cosine_accuracy@3
108
+ value: 0.64
109
+ name: Cosine Accuracy@3
110
+ - type: cosine_accuracy@5
111
+ value: 0.7
112
+ name: Cosine Accuracy@5
113
+ - type: cosine_accuracy@10
114
+ value: 0.76
115
+ name: Cosine Accuracy@10
116
+ - type: cosine_precision@1
117
+ value: 0.46
118
+ name: Cosine Precision@1
119
+ - type: cosine_precision@3
120
+ value: 0.22
121
+ name: Cosine Precision@3
122
+ - type: cosine_precision@5
123
+ value: 0.14400000000000002
124
+ name: Cosine Precision@5
125
+ - type: cosine_precision@10
126
+ value: 0.08
127
+ name: Cosine Precision@10
128
+ - type: cosine_recall@1
129
+ value: 0.43
130
+ name: Cosine Recall@1
131
+ - type: cosine_recall@3
132
+ value: 0.62
133
+ name: Cosine Recall@3
134
+ - type: cosine_recall@5
135
+ value: 0.68
136
+ name: Cosine Recall@5
137
+ - type: cosine_recall@10
138
+ value: 0.73
139
+ name: Cosine Recall@10
140
+ - type: cosine_ndcg@10
141
+ value: 0.592134936685869
142
+ name: Cosine Ndcg@10
143
+ - type: cosine_mrr@10
144
+ value: 0.5606666666666666
145
+ name: Cosine Mrr@10
146
+ - type: cosine_map@100
147
+ value: 0.5501347879979241
148
+ name: Cosine Map@100
149
+ - task:
150
+ type: information-retrieval
151
+ name: Information Retrieval
152
+ dataset:
153
+ name: NanoMSMARCO
154
+ type: NanoMSMARCO
155
+ metrics:
156
+ - type: cosine_accuracy@1
157
+ value: 0.32
158
+ name: Cosine Accuracy@1
159
+ - type: cosine_accuracy@3
160
+ value: 0.58
161
+ name: Cosine Accuracy@3
162
+ - type: cosine_accuracy@5
163
+ value: 0.68
164
+ name: Cosine Accuracy@5
165
+ - type: cosine_accuracy@10
166
+ value: 0.74
167
+ name: Cosine Accuracy@10
168
+ - type: cosine_precision@1
169
+ value: 0.32
170
+ name: Cosine Precision@1
171
+ - type: cosine_precision@3
172
+ value: 0.19333333333333333
173
+ name: Cosine Precision@3
174
+ - type: cosine_precision@5
175
+ value: 0.136
176
+ name: Cosine Precision@5
177
+ - type: cosine_precision@10
178
+ value: 0.07400000000000001
179
+ name: Cosine Precision@10
180
+ - type: cosine_recall@1
181
+ value: 0.32
182
+ name: Cosine Recall@1
183
+ - type: cosine_recall@3
184
+ value: 0.58
185
+ name: Cosine Recall@3
186
+ - type: cosine_recall@5
187
+ value: 0.68
188
+ name: Cosine Recall@5
189
+ - type: cosine_recall@10
190
+ value: 0.74
191
+ name: Cosine Recall@10
192
+ - type: cosine_ndcg@10
193
+ value: 0.5415424816174165
194
+ name: Cosine Ndcg@10
195
+ - type: cosine_mrr@10
196
+ value: 0.4768333333333334
197
+ name: Cosine Mrr@10
198
+ - type: cosine_map@100
199
+ value: 0.49019229786708785
200
+ name: Cosine Map@100
201
+ - task:
202
+ type: nano-beir
203
+ name: Nano BEIR
204
+ dataset:
205
+ name: NanoBEIR mean
206
+ type: NanoBEIR_mean
207
+ metrics:
208
+ - type: cosine_accuracy@1
209
+ value: 0.39
210
+ name: Cosine Accuracy@1
211
+ - type: cosine_accuracy@3
212
+ value: 0.61
213
+ name: Cosine Accuracy@3
214
+ - type: cosine_accuracy@5
215
+ value: 0.69
216
+ name: Cosine Accuracy@5
217
+ - type: cosine_accuracy@10
218
+ value: 0.75
219
+ name: Cosine Accuracy@10
220
+ - type: cosine_precision@1
221
+ value: 0.39
222
+ name: Cosine Precision@1
223
+ - type: cosine_precision@3
224
+ value: 0.20666666666666667
225
+ name: Cosine Precision@3
226
+ - type: cosine_precision@5
227
+ value: 0.14
228
+ name: Cosine Precision@5
229
+ - type: cosine_precision@10
230
+ value: 0.07700000000000001
231
+ name: Cosine Precision@10
232
+ - type: cosine_recall@1
233
+ value: 0.375
234
+ name: Cosine Recall@1
235
+ - type: cosine_recall@3
236
+ value: 0.6
237
+ name: Cosine Recall@3
238
+ - type: cosine_recall@5
239
+ value: 0.68
240
+ name: Cosine Recall@5
241
+ - type: cosine_recall@10
242
+ value: 0.735
243
+ name: Cosine Recall@10
244
+ - type: cosine_ndcg@10
245
+ value: 0.5668387091516427
246
+ name: Cosine Ndcg@10
247
+ - type: cosine_mrr@10
248
+ value: 0.51875
249
+ name: Cosine Mrr@10
250
+ - type: cosine_map@100
251
+ value: 0.520163542932506
252
+ name: Cosine Map@100
253
+ ---
254
+
255
+ # SentenceTransformer based on chandar-lab/NeoBERT
256
+
257
+ This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [chandar-lab/NeoBERT](https://huggingface.co/chandar-lab/NeoBERT) on the [gooaq](https://huggingface.co/datasets/sentence-transformers/gooaq) dataset. It maps sentences & paragraphs to a 768-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more.
258
+
259
+ ## Model Details
260
+
261
+ ### Model Description
262
+ - **Model Type:** Sentence Transformer
263
+ - **Base model:** [chandar-lab/NeoBERT](https://huggingface.co/chandar-lab/NeoBERT) <!-- at revision d97a4acdc851efed665d0550ea5704f00ad3ef76 -->
264
+ - **Maximum Sequence Length:** 8192 tokens
265
+ - **Output Dimensionality:** 768 dimensions
266
+ - **Similarity Function:** Cosine Similarity
267
+ - **Training Dataset:**
268
+ - [gooaq](https://huggingface.co/datasets/sentence-transformers/gooaq)
269
+ - **Language:** en
270
+ <!-- - **License:** Unknown -->
271
+
272
+ ### Model Sources
273
+
274
+ - **Documentation:** [Sentence Transformers Documentation](https://sbert.net)
275
+ - **Repository:** [Sentence Transformers on GitHub](https://github.com/UKPLab/sentence-transformers)
276
+ - **Hugging Face:** [Sentence Transformers on Hugging Face](https://huggingface.co/models?library=sentence-transformers)
277
+
278
+ ### Full Model Architecture
279
+
280
+ ```
281
+ SentenceTransformer(
282
+ (0): Transformer({'max_seq_length': 8192, 'do_lower_case': False}) with Transformer model: NeoBERT
283
+ (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, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})
284
+ )
285
+ ```
286
+
287
+ ## Usage
288
+
289
+ ### Direct Usage (Sentence Transformers)
290
+
291
+ First install the Sentence Transformers library:
292
+
293
+ ```bash
294
+ pip install -U sentence-transformers
295
+ ```
296
+
297
+ Then you can load this model and run inference.
298
+ ```python
299
+ from sentence_transformers import SentenceTransformer
300
+
301
+ # Download from the 🤗 Hub
302
+ model = SentenceTransformer("tomaarsen/NeoBERT-gooaq-8e-05")
303
+ # Run inference
304
+ sentences = [
305
+ 'what is the difference between 18 and 20 inch tires?',
306
+ 'The only real difference is a 20" rim would be more likely to be damaged, as you pointed out. Beyond looks, there is zero benefit for the 20" rim. Also, just the availability of tires will likely be much more limited for the larger rim. ... Tire selection is better for 18" wheels than 20" wheels.',
307
+ 'So extracurricular activities are just activities that you do outside of class. The Common App says that extracurricular activities "include arts, athletics, clubs, employment, personal commitments, and other pursuits."',
308
+ ]
309
+ embeddings = model.encode(sentences)
310
+ print(embeddings.shape)
311
+ # [3, 768]
312
+
313
+ # Get the similarity scores for the embeddings
314
+ similarities = model.similarity(embeddings, embeddings)
315
+ print(similarities.shape)
316
+ # [3, 3]
317
+ ```
318
+
319
+ <!--
320
+ ### Direct Usage (Transformers)
321
+
322
+ <details><summary>Click to see the direct usage in Transformers</summary>
323
+
324
+ </details>
325
+ -->
326
+
327
+ <!--
328
+ ### Downstream Usage (Sentence Transformers)
329
+
330
+ You can finetune this model on your own dataset.
331
+
332
+ <details><summary>Click to expand</summary>
333
+
334
+ </details>
335
+ -->
336
+
337
+ <!--
338
+ ### Out-of-Scope Use
339
+
340
+ *List how the model may foreseeably be misused and address what users ought not to do with the model.*
341
+ -->
342
+
343
+ ## Evaluation
344
+
345
+ ### Metrics
346
+
347
+ #### Information Retrieval
348
+
349
+ * Datasets: `NanoNQ` and `NanoMSMARCO`
350
+ * Evaluated with [<code>InformationRetrievalEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.InformationRetrievalEvaluator)
351
+
352
+ | Metric | NanoNQ | NanoMSMARCO |
353
+ |:--------------------|:-----------|:------------|
354
+ | cosine_accuracy@1 | 0.46 | 0.32 |
355
+ | cosine_accuracy@3 | 0.64 | 0.58 |
356
+ | cosine_accuracy@5 | 0.7 | 0.68 |
357
+ | cosine_accuracy@10 | 0.76 | 0.74 |
358
+ | cosine_precision@1 | 0.46 | 0.32 |
359
+ | cosine_precision@3 | 0.22 | 0.1933 |
360
+ | cosine_precision@5 | 0.144 | 0.136 |
361
+ | cosine_precision@10 | 0.08 | 0.074 |
362
+ | cosine_recall@1 | 0.43 | 0.32 |
363
+ | cosine_recall@3 | 0.62 | 0.58 |
364
+ | cosine_recall@5 | 0.68 | 0.68 |
365
+ | cosine_recall@10 | 0.73 | 0.74 |
366
+ | **cosine_ndcg@10** | **0.5921** | **0.5415** |
367
+ | cosine_mrr@10 | 0.5607 | 0.4768 |
368
+ | cosine_map@100 | 0.5501 | 0.4902 |
369
+
370
+ #### Nano BEIR
371
+
372
+ * Dataset: `NanoBEIR_mean`
373
+ * Evaluated with [<code>NanoBEIREvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.NanoBEIREvaluator)
374
+
375
+ | Metric | Value |
376
+ |:--------------------|:-----------|
377
+ | cosine_accuracy@1 | 0.39 |
378
+ | cosine_accuracy@3 | 0.61 |
379
+ | cosine_accuracy@5 | 0.69 |
380
+ | cosine_accuracy@10 | 0.75 |
381
+ | cosine_precision@1 | 0.39 |
382
+ | cosine_precision@3 | 0.2067 |
383
+ | cosine_precision@5 | 0.14 |
384
+ | cosine_precision@10 | 0.077 |
385
+ | cosine_recall@1 | 0.375 |
386
+ | cosine_recall@3 | 0.6 |
387
+ | cosine_recall@5 | 0.68 |
388
+ | cosine_recall@10 | 0.735 |
389
+ | **cosine_ndcg@10** | **0.5668** |
390
+ | cosine_mrr@10 | 0.5188 |
391
+ | cosine_map@100 | 0.5202 |
392
+
393
+ <!--
394
+ ## Bias, Risks and Limitations
395
+
396
+ *What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model.*
397
+ -->
398
+
399
+ <!--
400
+ ### Recommendations
401
+
402
+ *What are recommendations with respect to the foreseeable issues? For example, filtering explicit content.*
403
+ -->
404
+
405
+ ## Training Details
406
+
407
+ ### Training Dataset
408
+
409
+ #### gooaq
410
+
411
+ * Dataset: [gooaq](https://huggingface.co/datasets/sentence-transformers/gooaq) at [b089f72](https://huggingface.co/datasets/sentence-transformers/gooaq/tree/b089f728748a068b7bc5234e5bcf5b25e3c8279c)
412
+ * Size: 3,011,496 training samples
413
+ * Columns: <code>question</code> and <code>answer</code>
414
+ * Approximate statistics based on the first 1000 samples:
415
+ | | question | answer |
416
+ |:--------|:----------------------------------------------------------------------------------|:------------------------------------------------------------------------------------|
417
+ | type | string | string |
418
+ | details | <ul><li>min: 8 tokens</li><li>mean: 11.87 tokens</li><li>max: 23 tokens</li></ul> | <ul><li>min: 14 tokens</li><li>mean: 60.09 tokens</li><li>max: 201 tokens</li></ul> |
419
+ * Samples:
420
+ | question | answer |
421
+ |:-----------------------------------------------------------------------------|:---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
422
+ | <code>what is the difference between clay and mud mask?</code> | <code>The main difference between the two is that mud is a skin-healing agent, while clay is a cosmetic, drying agent. Clay masks are most useful for someone who has oily skin and is prone to breakouts of acne and blemishes.</code> |
423
+ | <code>myki how much on card?</code> | <code>A full fare myki card costs $6 and a concession, seniors or child myki costs $3. For more information about how to use your myki, visit ptv.vic.gov.au or call 1800 800 007.</code> |
424
+ | <code>how to find out if someone blocked your phone number on iphone?</code> | <code>If you get a notification like "Message Not Delivered" or you get no notification at all, that's a sign of a potential block. Next, you could try calling the person. If the call goes right to voicemail or rings once (or a half ring) then goes to voicemail, that's further evidence you may have been blocked.</code> |
425
+ * Loss: [<code>CachedMultipleNegativesRankingLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#cachedmultiplenegativesrankingloss) with these parameters:
426
+ ```json
427
+ {
428
+ "scale": 20.0,
429
+ "similarity_fct": "cos_sim"
430
+ }
431
+ ```
432
+
433
+ ### Evaluation Dataset
434
+
435
+ #### gooaq
436
+
437
+ * Dataset: [gooaq](https://huggingface.co/datasets/sentence-transformers/gooaq) at [b089f72](https://huggingface.co/datasets/sentence-transformers/gooaq/tree/b089f728748a068b7bc5234e5bcf5b25e3c8279c)
438
+ * Size: 1,000 evaluation samples
439
+ * Columns: <code>question</code> and <code>answer</code>
440
+ * Approximate statistics based on the first 1000 samples:
441
+ | | question | answer |
442
+ |:--------|:----------------------------------------------------------------------------------|:------------------------------------------------------------------------------------|
443
+ | type | string | string |
444
+ | details | <ul><li>min: 8 tokens</li><li>mean: 11.88 tokens</li><li>max: 22 tokens</li></ul> | <ul><li>min: 14 tokens</li><li>mean: 61.03 tokens</li><li>max: 127 tokens</li></ul> |
445
+ * Samples:
446
+ | question | answer |
447
+ |:-----------------------------------------------------------------------------|:-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
448
+ | <code>how do i program my directv remote with my tv?</code> | <code>['Press MENU on your remote.', 'Select Settings & Help > Settings > Remote Control > Program Remote.', 'Choose the device (TV, audio, DVD) you wish to program. ... ', 'Follow the on-screen prompts to complete programming.']</code> |
449
+ | <code>are rodrigues fruit bats nocturnal?</code> | <code>Before its numbers were threatened by habitat destruction, storms, and hunting, some of those groups could number 500 or more members. Sunrise, sunset. Rodrigues fruit bats are most active at dawn, at dusk, and at night.</code> |
450
+ | <code>why does your heart rate increase during exercise bbc bitesize?</code> | <code>During exercise there is an increase in physical activity and muscle cells respire more than they do when the body is at rest. The heart rate increases during exercise. The rate and depth of breathing increases - this makes sure that more oxygen is absorbed into the blood, and more carbon dioxide is removed from it.</code> |
451
+ * Loss: [<code>CachedMultipleNegativesRankingLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#cachedmultiplenegativesrankingloss) with these parameters:
452
+ ```json
453
+ {
454
+ "scale": 20.0,
455
+ "similarity_fct": "cos_sim"
456
+ }
457
+ ```
458
+
459
+ ### Training Hyperparameters
460
+ #### Non-Default Hyperparameters
461
+
462
+ - `eval_strategy`: steps
463
+ - `per_device_train_batch_size`: 2048
464
+ - `per_device_eval_batch_size`: 2048
465
+ - `learning_rate`: 8e-05
466
+ - `num_train_epochs`: 1
467
+ - `warmup_ratio`: 0.05
468
+ - `bf16`: True
469
+ - `batch_sampler`: no_duplicates
470
+
471
+ #### All Hyperparameters
472
+ <details><summary>Click to expand</summary>
473
+
474
+ - `overwrite_output_dir`: False
475
+ - `do_predict`: False
476
+ - `eval_strategy`: steps
477
+ - `prediction_loss_only`: True
478
+ - `per_device_train_batch_size`: 2048
479
+ - `per_device_eval_batch_size`: 2048
480
+ - `per_gpu_train_batch_size`: None
481
+ - `per_gpu_eval_batch_size`: None
482
+ - `gradient_accumulation_steps`: 1
483
+ - `eval_accumulation_steps`: None
484
+ - `torch_empty_cache_steps`: None
485
+ - `learning_rate`: 8e-05
486
+ - `weight_decay`: 0.0
487
+ - `adam_beta1`: 0.9
488
+ - `adam_beta2`: 0.999
489
+ - `adam_epsilon`: 1e-08
490
+ - `max_grad_norm`: 1.0
491
+ - `num_train_epochs`: 1
492
+ - `max_steps`: -1
493
+ - `lr_scheduler_type`: linear
494
+ - `lr_scheduler_kwargs`: {}
495
+ - `warmup_ratio`: 0.05
496
+ - `warmup_steps`: 0
497
+ - `log_level`: passive
498
+ - `log_level_replica`: warning
499
+ - `log_on_each_node`: True
500
+ - `logging_nan_inf_filter`: True
501
+ - `save_safetensors`: True
502
+ - `save_on_each_node`: False
503
+ - `save_only_model`: False
504
+ - `restore_callback_states_from_checkpoint`: False
505
+ - `no_cuda`: False
506
+ - `use_cpu`: False
507
+ - `use_mps_device`: False
508
+ - `seed`: 42
509
+ - `data_seed`: None
510
+ - `jit_mode_eval`: False
511
+ - `use_ipex`: False
512
+ - `bf16`: True
513
+ - `fp16`: False
514
+ - `fp16_opt_level`: O1
515
+ - `half_precision_backend`: auto
516
+ - `bf16_full_eval`: False
517
+ - `fp16_full_eval`: False
518
+ - `tf32`: None
519
+ - `local_rank`: 0
520
+ - `ddp_backend`: None
521
+ - `tpu_num_cores`: None
522
+ - `tpu_metrics_debug`: False
523
+ - `debug`: []
524
+ - `dataloader_drop_last`: False
525
+ - `dataloader_num_workers`: 0
526
+ - `dataloader_prefetch_factor`: None
527
+ - `past_index`: -1
528
+ - `disable_tqdm`: False
529
+ - `remove_unused_columns`: True
530
+ - `label_names`: None
531
+ - `load_best_model_at_end`: False
532
+ - `ignore_data_skip`: False
533
+ - `fsdp`: []
534
+ - `fsdp_min_num_params`: 0
535
+ - `fsdp_config`: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}
536
+ - `fsdp_transformer_layer_cls_to_wrap`: None
537
+ - `accelerator_config`: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
538
+ - `deepspeed`: None
539
+ - `label_smoothing_factor`: 0.0
540
+ - `optim`: adamw_torch
541
+ - `optim_args`: None
542
+ - `adafactor`: False
543
+ - `group_by_length`: False
544
+ - `length_column_name`: length
545
+ - `ddp_find_unused_parameters`: None
546
+ - `ddp_bucket_cap_mb`: None
547
+ - `ddp_broadcast_buffers`: False
548
+ - `dataloader_pin_memory`: True
549
+ - `dataloader_persistent_workers`: False
550
+ - `skip_memory_metrics`: True
551
+ - `use_legacy_prediction_loop`: False
552
+ - `push_to_hub`: False
553
+ - `resume_from_checkpoint`: None
554
+ - `hub_model_id`: None
555
+ - `hub_strategy`: every_save
556
+ - `hub_private_repo`: None
557
+ - `hub_always_push`: False
558
+ - `gradient_checkpointing`: False
559
+ - `gradient_checkpointing_kwargs`: None
560
+ - `include_inputs_for_metrics`: False
561
+ - `include_for_metrics`: []
562
+ - `eval_do_concat_batches`: True
563
+ - `fp16_backend`: auto
564
+ - `push_to_hub_model_id`: None
565
+ - `push_to_hub_organization`: None
566
+ - `mp_parameters`:
567
+ - `auto_find_batch_size`: False
568
+ - `full_determinism`: False
569
+ - `torchdynamo`: None
570
+ - `ray_scope`: last
571
+ - `ddp_timeout`: 1800
572
+ - `torch_compile`: False
573
+ - `torch_compile_backend`: None
574
+ - `torch_compile_mode`: None
575
+ - `dispatch_batches`: None
576
+ - `split_batches`: None
577
+ - `include_tokens_per_second`: False
578
+ - `include_num_input_tokens_seen`: False
579
+ - `neftune_noise_alpha`: None
580
+ - `optim_target_modules`: None
581
+ - `batch_eval_metrics`: False
582
+ - `eval_on_start`: False
583
+ - `use_liger_kernel`: False
584
+ - `eval_use_gather_object`: False
585
+ - `average_tokens_across_devices`: False
586
+ - `prompts`: None
587
+ - `batch_sampler`: no_duplicates
588
+ - `multi_dataset_batch_sampler`: proportional
589
+
590
+ </details>
591
+
592
+ ### Training Logs
593
+ <details><summary>Click to expand</summary>
594
+
595
+ | Epoch | Step | Training Loss | Validation Loss | NanoNQ_cosine_ndcg@10 | NanoMSMARCO_cosine_ndcg@10 | NanoBEIR_mean_cosine_ndcg@10 |
596
+ |:------:|:----:|:-------------:|:---------------:|:---------------------:|:--------------------------:|:----------------------------:|
597
+ | -1 | -1 | - | - | 0.0428 | 0.1127 | 0.0777 |
598
+ | 0.0068 | 10 | 4.2332 | - | - | - | - |
599
+ | 0.0136 | 20 | 1.5303 | - | - | - | - |
600
+ | 0.0204 | 30 | 0.887 | - | - | - | - |
601
+ | 0.0272 | 40 | 0.6286 | - | - | - | - |
602
+ | 0.0340 | 50 | 0.5193 | 0.2091 | 0.4434 | 0.4454 | 0.4444 |
603
+ | 0.0408 | 60 | 0.4423 | - | - | - | - |
604
+ | 0.0476 | 70 | 0.3842 | - | - | - | - |
605
+ | 0.0544 | 80 | 0.3576 | - | - | - | - |
606
+ | 0.0612 | 90 | 0.3301 | - | - | - | - |
607
+ | 0.0680 | 100 | 0.3135 | 0.1252 | 0.4606 | 0.5150 | 0.4878 |
608
+ | 0.0748 | 110 | 0.302 | - | - | - | - |
609
+ | 0.0816 | 120 | 0.277 | - | - | - | - |
610
+ | 0.0884 | 130 | 0.2694 | - | - | - | - |
611
+ | 0.0952 | 140 | 0.2628 | - | - | - | - |
612
+ | 0.1020 | 150 | 0.2471 | 0.0949 | 0.5135 | 0.5133 | 0.5134 |
613
+ | 0.1088 | 160 | 0.2343 | - | - | - | - |
614
+ | 0.1156 | 170 | 0.2386 | - | - | - | - |
615
+ | 0.1224 | 180 | 0.219 | - | - | - | - |
616
+ | 0.1292 | 190 | 0.217 | - | - | - | - |
617
+ | 0.1360 | 200 | 0.2073 | 0.0870 | 0.5281 | 0.4824 | 0.5052 |
618
+ | 0.1428 | 210 | 0.2208 | - | - | - | - |
619
+ | 0.1496 | 220 | 0.2046 | - | - | - | - |
620
+ | 0.1564 | 230 | 0.2045 | - | - | - | - |
621
+ | 0.1632 | 240 | 0.1987 | - | - | - | - |
622
+ | 0.1700 | 250 | 0.1949 | 0.0734 | 0.5781 | 0.4976 | 0.5378 |
623
+ | 0.1768 | 260 | 0.1888 | - | - | - | - |
624
+ | 0.1835 | 270 | 0.187 | - | - | - | - |
625
+ | 0.1903 | 280 | 0.1834 | - | - | - | - |
626
+ | 0.1971 | 290 | 0.1747 | - | - | - | - |
627
+ | 0.2039 | 300 | 0.1805 | 0.0663 | 0.5580 | 0.5453 | 0.5516 |
628
+ | 0.2107 | 310 | 0.1738 | - | - | - | - |
629
+ | 0.2175 | 320 | 0.1707 | - | - | - | - |
630
+ | 0.2243 | 330 | 0.1758 | - | - | - | - |
631
+ | 0.2311 | 340 | 0.1762 | - | - | - | - |
632
+ | 0.2379 | 350 | 0.1649 | 0.0624 | 0.5761 | 0.5310 | 0.5535 |
633
+ | 0.2447 | 360 | 0.1682 | - | - | - | - |
634
+ | 0.2515 | 370 | 0.1629 | - | - | - | - |
635
+ | 0.2583 | 380 | 0.1595 | - | - | - | - |
636
+ | 0.2651 | 390 | 0.1571 | - | - | - | - |
637
+ | 0.2719 | 400 | 0.1617 | 0.0592 | 0.5865 | 0.5193 | 0.5529 |
638
+ | 0.2787 | 410 | 0.1521 | - | - | - | - |
639
+ | 0.2855 | 420 | 0.1518 | - | - | - | - |
640
+ | 0.2923 | 430 | 0.1583 | - | - | - | - |
641
+ | 0.2991 | 440 | 0.1516 | - | - | - | - |
642
+ | 0.3059 | 450 | 0.1473 | 0.0570 | 0.5844 | 0.5181 | 0.5512 |
643
+ | 0.3127 | 460 | 0.1491 | - | - | - | - |
644
+ | 0.3195 | 470 | 0.1487 | - | - | - | - |
645
+ | 0.3263 | 480 | 0.1457 | - | - | - | - |
646
+ | 0.3331 | 490 | 0.1463 | - | - | - | - |
647
+ | 0.3399 | 500 | 0.141 | 0.0571 | 0.5652 | 0.5027 | 0.5340 |
648
+ | 0.3467 | 510 | 0.1438 | - | - | - | - |
649
+ | 0.3535 | 520 | 0.148 | - | - | - | - |
650
+ | 0.3603 | 530 | 0.136 | - | - | - | - |
651
+ | 0.3671 | 540 | 0.1359 | - | - | - | - |
652
+ | 0.3739 | 550 | 0.1388 | 0.0507 | 0.5457 | 0.4660 | 0.5058 |
653
+ | 0.3807 | 560 | 0.1358 | - | - | - | - |
654
+ | 0.3875 | 570 | 0.1365 | - | - | - | - |
655
+ | 0.3943 | 580 | 0.1328 | - | - | - | - |
656
+ | 0.4011 | 590 | 0.1404 | - | - | - | - |
657
+ | 0.4079 | 600 | 0.1304 | 0.0524 | 0.5477 | 0.5259 | 0.5368 |
658
+ | 0.4147 | 610 | 0.1321 | - | - | - | - |
659
+ | 0.4215 | 620 | 0.1322 | - | - | - | - |
660
+ | 0.4283 | 630 | 0.1262 | - | - | - | - |
661
+ | 0.4351 | 640 | 0.1339 | - | - | - | - |
662
+ | 0.4419 | 650 | 0.1257 | 0.0494 | 0.5564 | 0.4920 | 0.5242 |
663
+ | 0.4487 | 660 | 0.1247 | - | - | - | - |
664
+ | 0.4555 | 670 | 0.1316 | - | - | - | - |
665
+ | 0.4623 | 680 | 0.124 | - | - | - | - |
666
+ | 0.4691 | 690 | 0.1247 | - | - | - | - |
667
+ | 0.4759 | 700 | 0.1212 | 0.0480 | 0.5663 | 0.5040 | 0.5351 |
668
+ | 0.4827 | 710 | 0.1194 | - | - | - | - |
669
+ | 0.4895 | 720 | 0.1224 | - | - | - | - |
670
+ | 0.4963 | 730 | 0.1225 | - | - | - | - |
671
+ | 0.5031 | 740 | 0.1209 | - | - | - | - |
672
+ | 0.5099 | 750 | 0.1197 | 0.0447 | 0.5535 | 0.5127 | 0.5331 |
673
+ | 0.5167 | 760 | 0.1196 | - | - | - | - |
674
+ | 0.5235 | 770 | 0.1129 | - | - | - | - |
675
+ | 0.5303 | 780 | 0.1223 | - | - | - | - |
676
+ | 0.5370 | 790 | 0.1159 | - | - | - | - |
677
+ | 0.5438 | 800 | 0.1178 | 0.0412 | 0.5558 | 0.5275 | 0.5416 |
678
+ | 0.5506 | 810 | 0.1186 | - | - | - | - |
679
+ | 0.5574 | 820 | 0.1153 | - | - | - | - |
680
+ | 0.5642 | 830 | 0.1178 | - | - | - | - |
681
+ | 0.5710 | 840 | 0.1155 | - | - | - | - |
682
+ | 0.5778 | 850 | 0.1152 | 0.0432 | 0.5738 | 0.5243 | 0.5490 |
683
+ | 0.5846 | 860 | 0.1101 | - | - | - | - |
684
+ | 0.5914 | 870 | 0.1057 | - | - | - | - |
685
+ | 0.5982 | 880 | 0.1141 | - | - | - | - |
686
+ | 0.6050 | 890 | 0.1172 | - | - | - | - |
687
+ | 0.6118 | 900 | 0.1146 | 0.0414 | 0.5641 | 0.4805 | 0.5223 |
688
+ | 0.6186 | 910 | 0.1094 | - | - | - | - |
689
+ | 0.6254 | 920 | 0.1116 | - | - | - | - |
690
+ | 0.6322 | 930 | 0.111 | - | - | - | - |
691
+ | 0.6390 | 940 | 0.1078 | - | - | - | - |
692
+ | 0.6458 | 950 | 0.1041 | 0.0424 | 0.5883 | 0.5412 | 0.5647 |
693
+ | 0.6526 | 960 | 0.1068 | - | - | - | - |
694
+ | 0.6594 | 970 | 0.1076 | - | - | - | - |
695
+ | 0.6662 | 980 | 0.1068 | - | - | - | - |
696
+ | 0.6730 | 990 | 0.1038 | - | - | - | - |
697
+ | 0.6798 | 1000 | 0.1017 | 0.0409 | 0.5850 | 0.5117 | 0.5483 |
698
+ | 0.6866 | 1010 | 0.1079 | - | - | - | - |
699
+ | 0.6934 | 1020 | 0.1067 | - | - | - | - |
700
+ | 0.7002 | 1030 | 0.1079 | - | - | - | - |
701
+ | 0.7070 | 1040 | 0.1039 | - | - | - | - |
702
+ | 0.7138 | 1050 | 0.1016 | 0.0356 | 0.5927 | 0.5344 | 0.5636 |
703
+ | 0.7206 | 1060 | 0.1017 | - | - | - | - |
704
+ | 0.7274 | 1070 | 0.1029 | - | - | - | - |
705
+ | 0.7342 | 1080 | 0.1038 | - | - | - | - |
706
+ | 0.7410 | 1090 | 0.0994 | - | - | - | - |
707
+ | 0.7478 | 1100 | 0.0984 | 0.0376 | 0.5618 | 0.5321 | 0.5470 |
708
+ | 0.7546 | 1110 | 0.0966 | - | - | - | - |
709
+ | 0.7614 | 1120 | 0.1024 | - | - | - | - |
710
+ | 0.7682 | 1130 | 0.099 | - | - | - | - |
711
+ | 0.7750 | 1140 | 0.1017 | - | - | - | - |
712
+ | 0.7818 | 1150 | 0.0951 | 0.0368 | 0.5832 | 0.5073 | 0.5453 |
713
+ | 0.7886 | 1160 | 0.1008 | - | - | - | - |
714
+ | 0.7954 | 1170 | 0.096 | - | - | - | - |
715
+ | 0.8022 | 1180 | 0.0962 | - | - | - | - |
716
+ | 0.8090 | 1190 | 0.1004 | - | - | - | - |
717
+ | 0.8158 | 1200 | 0.0986 | 0.0321 | 0.5895 | 0.5242 | 0.5568 |
718
+ | 0.8226 | 1210 | 0.0966 | - | - | - | - |
719
+ | 0.8294 | 1220 | 0.096 | - | - | - | - |
720
+ | 0.8362 | 1230 | 0.0962 | - | - | - | - |
721
+ | 0.8430 | 1240 | 0.0987 | - | - | - | - |
722
+ | 0.8498 | 1250 | 0.096 | 0.0316 | 0.5801 | 0.5434 | 0.5617 |
723
+ | 0.8566 | 1260 | 0.097 | - | - | - | - |
724
+ | 0.8634 | 1270 | 0.0929 | - | - | - | - |
725
+ | 0.8702 | 1280 | 0.0973 | - | - | - | - |
726
+ | 0.8770 | 1290 | 0.0973 | - | - | - | - |
727
+ | 0.8838 | 1300 | 0.0939 | 0.0330 | 0.5916 | 0.5478 | 0.5697 |
728
+ | 0.8906 | 1310 | 0.0968 | - | - | - | - |
729
+ | 0.8973 | 1320 | 0.0969 | - | - | - | - |
730
+ | 0.9041 | 1330 | 0.0931 | - | - | - | - |
731
+ | 0.9109 | 1340 | 0.0919 | - | - | - | - |
732
+ | 0.9177 | 1350 | 0.0916 | 0.0324 | 0.5908 | 0.5308 | 0.5608 |
733
+ | 0.9245 | 1360 | 0.0903 | - | - | - | - |
734
+ | 0.9313 | 1370 | 0.0957 | - | - | - | - |
735
+ | 0.9381 | 1380 | 0.0891 | - | - | - | - |
736
+ | 0.9449 | 1390 | 0.0909 | - | - | - | - |
737
+ | 0.9517 | 1400 | 0.0924 | 0.0318 | 0.5823 | 0.5388 | 0.5605 |
738
+ | 0.9585 | 1410 | 0.0932 | - | - | - | - |
739
+ | 0.9653 | 1420 | 0.0916 | - | - | - | - |
740
+ | 0.9721 | 1430 | 0.0966 | - | - | - | - |
741
+ | 0.9789 | 1440 | 0.0864 | - | - | - | - |
742
+ | 0.9857 | 1450 | 0.0872 | 0.0311 | 0.5895 | 0.5442 | 0.5668 |
743
+ | 0.9925 | 1460 | 0.0897 | - | - | - | - |
744
+ | 0.9993 | 1470 | 0.086 | - | - | - | - |
745
+ | -1 | -1 | - | - | 0.5921 | 0.5415 | 0.5668 |
746
+
747
+ </details>
748
+
749
+ ### Framework Versions
750
+ - Python: 3.11.10
751
+ - Sentence Transformers: 3.5.0.dev0
752
+ - Transformers: 4.49.0
753
+ - PyTorch: 2.5.1+cu124
754
+ - Accelerate: 1.2.0
755
+ - Datasets: 2.21.0
756
+ - Tokenizers: 0.21.0
757
+
758
+ ## Citation
759
+
760
+ ### BibTeX
761
+
762
+ #### Sentence Transformers
763
+ ```bibtex
764
+ @inproceedings{reimers-2019-sentence-bert,
765
+ title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks",
766
+ author = "Reimers, Nils and Gurevych, Iryna",
767
+ booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing",
768
+ month = "11",
769
+ year = "2019",
770
+ publisher = "Association for Computational Linguistics",
771
+ url = "https://arxiv.org/abs/1908.10084",
772
+ }
773
+ ```
774
+
775
+ #### CachedMultipleNegativesRankingLoss
776
+ ```bibtex
777
+ @misc{gao2021scaling,
778
+ title={Scaling Deep Contrastive Learning Batch Size under Memory Limited Setup},
779
+ author={Luyu Gao and Yunyi Zhang and Jiawei Han and Jamie Callan},
780
+ year={2021},
781
+ eprint={2101.06983},
782
+ archivePrefix={arXiv},
783
+ primaryClass={cs.LG}
784
+ }
785
+ ```
786
+
787
+ <!--
788
+ ## Glossary
789
+
790
+ *Clearly define terms in order to be accessible across audiences.*
791
+ -->
792
+
793
+ <!--
794
+ ## Model Card Authors
795
+
796
+ *Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction.*
797
+ -->
798
+
799
+ <!--
800
+ ## Model Card Contact
801
+
802
+ *Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors.*
803
+ -->
config.json ADDED
@@ -0,0 +1,54 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "chandar-lab/NeoBERT",
3
+ "architectures": [
4
+ "NeoBERT"
5
+ ],
6
+ "auto_map": {
7
+ "AutoConfig": "chandar-lab/NeoBERT--model.NeoBERTConfig",
8
+ "AutoModel": "chandar-lab/NeoBERT--model.NeoBERT",
9
+ "AutoModelForMaskedLM": "chandar-lab/NeoBERT--model.NeoBERTLMHead",
10
+ "AutoModelForSequenceClassification": "chandar-lab/NeoBERT--model.NeoBERTForSequenceClassification"
11
+ },
12
+ "classifier_init_range": 0.02,
13
+ "decoder_init_range": 0.02,
14
+ "dim_head": 64,
15
+ "embedding_init_range": 0.02,
16
+ "hidden_size": 768,
17
+ "intermediate_size": 3072,
18
+ "kwargs": {
19
+ "_commit_hash": "d97a4acdc851efed665d0550ea5704f00ad3ef76",
20
+ "architectures": [
21
+ "NeoBERTLMHead"
22
+ ],
23
+ "attn_implementation": null,
24
+ "auto_map": {
25
+ "AutoConfig": "chandar-lab/NeoBERT--model.NeoBERTConfig",
26
+ "AutoModel": "chandar-lab/NeoBERT--model.NeoBERT",
27
+ "AutoModelForMaskedLM": "chandar-lab/NeoBERT--model.NeoBERTLMHead",
28
+ "AutoModelForSequenceClassification": "chandar-lab/NeoBERT--model.NeoBERTForSequenceClassification"
29
+ },
30
+ "classifier_init_range": 0.02,
31
+ "dim_head": 64,
32
+ "kwargs": {
33
+ "classifier_init_range": 0.02,
34
+ "pretrained_model_name_or_path": "google-bert/bert-base-uncased",
35
+ "trust_remote_code": true
36
+ },
37
+ "model_type": "neobert",
38
+ "pretrained_model_name_or_path": "google-bert/bert-base-uncased",
39
+ "torch_dtype": "float32",
40
+ "transformers_version": "4.48.2",
41
+ "trust_remote_code": true
42
+ },
43
+ "max_length": 4096,
44
+ "model_type": "neobert",
45
+ "norm_eps": 1e-05,
46
+ "num_attention_heads": 12,
47
+ "num_hidden_layers": 28,
48
+ "pad_token_id": 0,
49
+ "pretrained_model_name_or_path": "google-bert/bert-base-uncased",
50
+ "torch_dtype": "float32",
51
+ "transformers_version": "4.49.0",
52
+ "trust_remote_code": true,
53
+ "vocab_size": 30522
54
+ }
config_sentence_transformers.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "__version__": {
3
+ "sentence_transformers": "3.5.0.dev0",
4
+ "transformers": "4.49.0",
5
+ "pytorch": "2.5.1+cu124"
6
+ },
7
+ "prompts": {},
8
+ "default_prompt_name": null,
9
+ "similarity_fn_name": "cosine"
10
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1a4601209bc2749942169d43e934b9859efe03119252eb1fa7f696aa8d220cf1
3
+ size 886680744
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
+ ]
sentence_bert_config.json ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ {
2
+ "max_seq_length": 8192,
3
+ "do_lower_case": false
4
+ }
special_tokens_map.json ADDED
@@ -0,0 +1,37 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "cls_token": {
3
+ "content": "[CLS]",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "mask_token": {
10
+ "content": "[MASK]",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": {
17
+ "content": "[PAD]",
18
+ "lstrip": false,
19
+ "normalized": false,
20
+ "rstrip": false,
21
+ "single_word": false
22
+ },
23
+ "sep_token": {
24
+ "content": "[SEP]",
25
+ "lstrip": false,
26
+ "normalized": false,
27
+ "rstrip": false,
28
+ "single_word": false
29
+ },
30
+ "unk_token": {
31
+ "content": "[UNK]",
32
+ "lstrip": false,
33
+ "normalized": false,
34
+ "rstrip": false,
35
+ "single_word": false
36
+ }
37
+ }
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json ADDED
@@ -0,0 +1,61 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "added_tokens_decoder": {
3
+ "0": {
4
+ "content": "[PAD]",
5
+ "lstrip": false,
6
+ "normalized": false,
7
+ "rstrip": false,
8
+ "single_word": false,
9
+ "special": true
10
+ },
11
+ "100": {
12
+ "content": "[UNK]",
13
+ "lstrip": false,
14
+ "normalized": false,
15
+ "rstrip": false,
16
+ "single_word": false,
17
+ "special": true
18
+ },
19
+ "101": {
20
+ "content": "[CLS]",
21
+ "lstrip": false,
22
+ "normalized": false,
23
+ "rstrip": false,
24
+ "single_word": false,
25
+ "special": true
26
+ },
27
+ "102": {
28
+ "content": "[SEP]",
29
+ "lstrip": false,
30
+ "normalized": false,
31
+ "rstrip": false,
32
+ "single_word": false,
33
+ "special": true
34
+ },
35
+ "103": {
36
+ "content": "[MASK]",
37
+ "lstrip": false,
38
+ "normalized": false,
39
+ "rstrip": false,
40
+ "single_word": false,
41
+ "special": true
42
+ }
43
+ },
44
+ "clean_up_tokenization_spaces": false,
45
+ "cls_token": "[CLS]",
46
+ "do_lower_case": true,
47
+ "extra_special_tokens": {},
48
+ "mask_token": "[MASK]",
49
+ "model_input_names": [
50
+ "input_ids",
51
+ "attention_mask"
52
+ ],
53
+ "model_max_length": 4096,
54
+ "pad_token": "[PAD]",
55
+ "sep_token": "[SEP]",
56
+ "strip_accents": null,
57
+ "tokenize_chinese_chars": true,
58
+ "tokenizer_class": "BertTokenizer",
59
+ "unk_token": "[UNK]",
60
+ "vocab_size": 30522
61
+ }
vocab.txt ADDED
The diff for this file is too large to render. See raw diff