akot commited on
Commit
5c33e6c
1 Parent(s): 400a8d4

Add new SentenceTransformer model.

Browse files
.gitattributes CHANGED
@@ -33,3 +33,4 @@ 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
+ tokenizer.json filter=lfs diff=lfs merge=lfs -text
1_Pooling/config.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "word_embedding_dimension": 1024,
3
+ "pooling_mode_cls_token": true,
4
+ "pooling_mode_mean_tokens": false,
5
+ "pooling_mode_max_tokens": false,
6
+ "pooling_mode_mean_sqrt_len_tokens": false,
7
+ "pooling_mode_weightedmean_tokens": false,
8
+ "pooling_mode_lasttoken": false,
9
+ "include_prompt": true
10
+ }
README.md ADDED
@@ -0,0 +1,839 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: BAAI/bge-m3
3
+ datasets: []
4
+ language: []
5
+ library_name: sentence-transformers
6
+ metrics:
7
+ - cosine_accuracy@1
8
+ - cosine_accuracy@3
9
+ - cosine_accuracy@5
10
+ - cosine_accuracy@10
11
+ - cosine_precision@1
12
+ - cosine_precision@3
13
+ - cosine_precision@5
14
+ - cosine_precision@10
15
+ - cosine_recall@1
16
+ - cosine_recall@3
17
+ - cosine_recall@5
18
+ - cosine_recall@10
19
+ - cosine_ndcg@10
20
+ - cosine_mrr@10
21
+ - cosine_map@100
22
+ pipeline_tag: sentence-similarity
23
+ tags:
24
+ - sentence-transformers
25
+ - sentence-similarity
26
+ - feature-extraction
27
+ - generated_from_trainer
28
+ - dataset_size:4957
29
+ - loss:MatryoshkaLoss
30
+ - loss:MultipleNegativesRankingLoss
31
+ widget:
32
+ - source_sentence: 312 Aus steuerlicher Sicht ist es möglich, mehrere Versorgungszusagen
33
+ nebeneinander, also neben einer Altzusage auch eine Neuzusage zu erteilen (z.
34
+ B. „alte“ Direktversicherung und „neuer“ Pensionsfonds).
35
+ sentences:
36
+ - Wann liegt bei der betrieblichen Altersversorgung eine schädliche Verwendung vor?
37
+ - Welche steuerliche Behandlung erfahren Auszahlungen aus Altersvorsorgeverträgen
38
+ nach § 22 Nr. 5 EStG?
39
+ - Können verschiedene Versorgungszusagen wie Direktversicherung und Pensionsfonds
40
+ gleichzeitig bestehen?
41
+ - source_sentence: 5 Pflichtversicherte nach dem Gesetz über die Alterssicherung der
42
+ Landwirte gehören, soweit sie nicht als Pflichtversicherte der gesetzlichen Rentenversicherung
43
+ ohnehin bereits anspruchsberechtigt sind, in dieser Eigenschaft ebenfalls zum
44
+ begünstigten Personenkreis. Darunter fallen insbesondere die in Anlage 1 Abschnitt
45
+ B aufgeführten Personen.
46
+ sentences:
47
+ - Wann wird das Anrecht der ausgleichsberechtigten Person bei intern geteilter Altersvorsorge
48
+ als abgeschlossen betrachtet?
49
+ - Welche Personen sind in der Anlage 1 Abschnitt B bezüglich der Alterssicherung
50
+ der Landwirte aufgeführt?
51
+ - In welchen Fällen führt die Möglichkeit einer Beitragserstattung nicht zur Versagung
52
+ der Anerkennung als betriebliche Altersversorgung?
53
+ - source_sentence: 233 Voraussetzung für die Förderung durch Sonderausgabenabzug nach
54
+ § 10a EStG und Zulage nach Abschnitt XI EStG ist in den Fällen der Rz. 231 f.,
55
+ dass der Steuerpflichtige zum begünstigten Personenkreis gehört. Die zeitliche
56
+ Zuordnung dieser Altersvorsorgebeiträge richtet sich grundsätzlich nach § 11 Abs.
57
+ 2 EStG.
58
+ sentences:
59
+ - Wer gehört zum begünstigten Personenkreis für die Altersvorsorgeförderung?
60
+ - Wie werden erstattete Kosten eines Altersvorsorgevertrags besteuert, wenn sie
61
+ dem Steuerpflichtigen ausgezahlt werden?
62
+ - Ist der Übertragungswert einer betrieblichen Altersversorgung bei einem Arbeitgeberwechsel
63
+ steuerfrei?
64
+ - source_sentence: 127 Die Entnahme des Teilkapitalbetrags von bis zu 30 % des zur
65
+ Verfügung stehenden Kapitals aus dem Vertrag hat zu Beginn der Auszahlungsphase
66
+ zu erfolgen. Eine Verteilung über mehrere Auszahlungszeitpunkte ist nicht möglich.
67
+ sentences:
68
+ - Kann ich den Teilkapitalbetrag aus meiner Altersvorsorge zu verschiedenen Zeitpunkten
69
+ entnehmen?
70
+ - Welche Einkunftsarten können Leistungen aus einer Versorgungszusage des Arbeitgebers
71
+ sein?
72
+ - Was ist im Todesfall des Zulageberechtigten bezüglich der Förderbeiträge zu tun?
73
+ - source_sentence: '67 Abwandlung des Beispiels 1 in Rn. 66: A erhält zudem zwei Kinderzulagen
74
+ für seine in den Jahren 2004 und 2005 geborenen Kinder. Beitragspflichtige Einnahmen
75
+ 53.000 € 4 % 2.120 € höchstens 2.100 € anzusetzen 2.100 € abzüglich Zulage 175
76
+ € Mindesteigenbeitrag (§ 86 Abs. 1 Satz 2 EStG) 1.925 € Sockelbetrag (§ 86 Abs.
77
+ 1 Satz 4 EStG) 60 € maßgebend (§ 86 Abs. 1 Satz 5 EStG) 1.925 € Die von A geleisteten
78
+ Beiträge übersteigen den Mindesteigenbeitrag. Die Zulage wird nicht gekürzt.'
79
+ sentences:
80
+ - Wird die Zulage für A gekürzt, wenn die Beiträge den Mindesteigenbeitrag übersteigen?
81
+ - Was versteht man unter Sonderzahlungen des Arbeitgebers?
82
+ - Wie erfolgt die Besteuerung bei der ausgleichsberechtigten Person nach einer externen
83
+ Teilung?
84
+ model-index:
85
+ - name: SentenceTransformer based on BAAI/bge-m3
86
+ results:
87
+ - task:
88
+ type: information-retrieval
89
+ name: Information Retrieval
90
+ dataset:
91
+ name: dim 768
92
+ type: dim_768
93
+ metrics:
94
+ - type: cosine_accuracy@1
95
+ value: 0.019963702359346643
96
+ name: Cosine Accuracy@1
97
+ - type: cosine_accuracy@3
98
+ value: 0.24863883847549909
99
+ name: Cosine Accuracy@3
100
+ - type: cosine_accuracy@5
101
+ value: 0.3992740471869328
102
+ name: Cosine Accuracy@5
103
+ - type: cosine_accuracy@10
104
+ value: 0.6588021778584392
105
+ name: Cosine Accuracy@10
106
+ - type: cosine_precision@1
107
+ value: 0.019963702359346643
108
+ name: Cosine Precision@1
109
+ - type: cosine_precision@3
110
+ value: 0.08287961282516636
111
+ name: Cosine Precision@3
112
+ - type: cosine_precision@5
113
+ value: 0.07985480943738657
114
+ name: Cosine Precision@5
115
+ - type: cosine_precision@10
116
+ value: 0.06588021778584391
117
+ name: Cosine Precision@10
118
+ - type: cosine_recall@1
119
+ value: 0.019963702359346643
120
+ name: Cosine Recall@1
121
+ - type: cosine_recall@3
122
+ value: 0.24863883847549909
123
+ name: Cosine Recall@3
124
+ - type: cosine_recall@5
125
+ value: 0.3992740471869328
126
+ name: Cosine Recall@5
127
+ - type: cosine_recall@10
128
+ value: 0.6588021778584392
129
+ name: Cosine Recall@10
130
+ - type: cosine_ndcg@10
131
+ value: 0.3072766574898433
132
+ name: Cosine Ndcg@10
133
+ - type: cosine_mrr@10
134
+ value: 0.19910408204419097
135
+ name: Cosine Mrr@10
136
+ - type: cosine_map@100
137
+ value: 0.21588812107724084
138
+ name: Cosine Map@100
139
+ - task:
140
+ type: information-retrieval
141
+ name: Information Retrieval
142
+ dataset:
143
+ name: dim 512
144
+ type: dim_512
145
+ metrics:
146
+ - type: cosine_accuracy@1
147
+ value: 0.029038112522686024
148
+ name: Cosine Accuracy@1
149
+ - type: cosine_accuracy@3
150
+ value: 0.25226860254083483
151
+ name: Cosine Accuracy@3
152
+ - type: cosine_accuracy@5
153
+ value: 0.4029038112522686
154
+ name: Cosine Accuracy@5
155
+ - type: cosine_accuracy@10
156
+ value: 0.647912885662432
157
+ name: Cosine Accuracy@10
158
+ - type: cosine_precision@1
159
+ value: 0.029038112522686024
160
+ name: Cosine Precision@1
161
+ - type: cosine_precision@3
162
+ value: 0.08408953418027827
163
+ name: Cosine Precision@3
164
+ - type: cosine_precision@5
165
+ value: 0.08058076225045373
166
+ name: Cosine Precision@5
167
+ - type: cosine_precision@10
168
+ value: 0.0647912885662432
169
+ name: Cosine Precision@10
170
+ - type: cosine_recall@1
171
+ value: 0.029038112522686024
172
+ name: Cosine Recall@1
173
+ - type: cosine_recall@3
174
+ value: 0.25226860254083483
175
+ name: Cosine Recall@3
176
+ - type: cosine_recall@5
177
+ value: 0.4029038112522686
178
+ name: Cosine Recall@5
179
+ - type: cosine_recall@10
180
+ value: 0.647912885662432
181
+ name: Cosine Recall@10
182
+ - type: cosine_ndcg@10
183
+ value: 0.30835796317387987
184
+ name: Cosine Ndcg@10
185
+ - type: cosine_mrr@10
186
+ value: 0.20380765131218861
187
+ name: Cosine Mrr@10
188
+ - type: cosine_map@100
189
+ value: 0.22078173489998865
190
+ name: Cosine Map@100
191
+ - task:
192
+ type: information-retrieval
193
+ name: Information Retrieval
194
+ dataset:
195
+ name: dim 256
196
+ type: dim_256
197
+ metrics:
198
+ - type: cosine_accuracy@1
199
+ value: 0.034482758620689655
200
+ name: Cosine Accuracy@1
201
+ - type: cosine_accuracy@3
202
+ value: 0.23049001814882034
203
+ name: Cosine Accuracy@3
204
+ - type: cosine_accuracy@5
205
+ value: 0.38656987295825773
206
+ name: Cosine Accuracy@5
207
+ - type: cosine_accuracy@10
208
+ value: 0.647912885662432
209
+ name: Cosine Accuracy@10
210
+ - type: cosine_precision@1
211
+ value: 0.034482758620689655
212
+ name: Cosine Precision@1
213
+ - type: cosine_precision@3
214
+ value: 0.07683000604960677
215
+ name: Cosine Precision@3
216
+ - type: cosine_precision@5
217
+ value: 0.07731397459165154
218
+ name: Cosine Precision@5
219
+ - type: cosine_precision@10
220
+ value: 0.0647912885662432
221
+ name: Cosine Precision@10
222
+ - type: cosine_recall@1
223
+ value: 0.034482758620689655
224
+ name: Cosine Recall@1
225
+ - type: cosine_recall@3
226
+ value: 0.23049001814882034
227
+ name: Cosine Recall@3
228
+ - type: cosine_recall@5
229
+ value: 0.38656987295825773
230
+ name: Cosine Recall@5
231
+ - type: cosine_recall@10
232
+ value: 0.647912885662432
233
+ name: Cosine Recall@10
234
+ - type: cosine_ndcg@10
235
+ value: 0.3040645102017685
236
+ name: Cosine Ndcg@10
237
+ - type: cosine_mrr@10
238
+ value: 0.1990666320974852
239
+ name: Cosine Mrr@10
240
+ - type: cosine_map@100
241
+ value: 0.21577058637681837
242
+ name: Cosine Map@100
243
+ - task:
244
+ type: information-retrieval
245
+ name: Information Retrieval
246
+ dataset:
247
+ name: dim 128
248
+ type: dim_128
249
+ metrics:
250
+ - type: cosine_accuracy@1
251
+ value: 0.02722323049001815
252
+ name: Cosine Accuracy@1
253
+ - type: cosine_accuracy@3
254
+ value: 0.24319419237749546
255
+ name: Cosine Accuracy@3
256
+ - type: cosine_accuracy@5
257
+ value: 0.38656987295825773
258
+ name: Cosine Accuracy@5
259
+ - type: cosine_accuracy@10
260
+ value: 0.6442831215970962
261
+ name: Cosine Accuracy@10
262
+ - type: cosine_precision@1
263
+ value: 0.02722323049001815
264
+ name: Cosine Precision@1
265
+ - type: cosine_precision@3
266
+ value: 0.08106473079249849
267
+ name: Cosine Precision@3
268
+ - type: cosine_precision@5
269
+ value: 0.07731397459165155
270
+ name: Cosine Precision@5
271
+ - type: cosine_precision@10
272
+ value: 0.06442831215970962
273
+ name: Cosine Precision@10
274
+ - type: cosine_recall@1
275
+ value: 0.02722323049001815
276
+ name: Cosine Recall@1
277
+ - type: cosine_recall@3
278
+ value: 0.24319419237749546
279
+ name: Cosine Recall@3
280
+ - type: cosine_recall@5
281
+ value: 0.38656987295825773
282
+ name: Cosine Recall@5
283
+ - type: cosine_recall@10
284
+ value: 0.6442831215970962
285
+ name: Cosine Recall@10
286
+ - type: cosine_ndcg@10
287
+ value: 0.3030488823891233
288
+ name: Cosine Ndcg@10
289
+ - type: cosine_mrr@10
290
+ value: 0.19836804655316465
291
+ name: Cosine Mrr@10
292
+ - type: cosine_map@100
293
+ value: 0.21511274800304536
294
+ name: Cosine Map@100
295
+ - task:
296
+ type: information-retrieval
297
+ name: Information Retrieval
298
+ dataset:
299
+ name: dim 64
300
+ type: dim_64
301
+ metrics:
302
+ - type: cosine_accuracy@1
303
+ value: 0.016333938294010888
304
+ name: Cosine Accuracy@1
305
+ - type: cosine_accuracy@3
306
+ value: 0.2250453720508167
307
+ name: Cosine Accuracy@3
308
+ - type: cosine_accuracy@5
309
+ value: 0.37749546279491836
310
+ name: Cosine Accuracy@5
311
+ - type: cosine_accuracy@10
312
+ value: 0.617059891107078
313
+ name: Cosine Accuracy@10
314
+ - type: cosine_precision@1
315
+ value: 0.016333938294010888
316
+ name: Cosine Precision@1
317
+ - type: cosine_precision@3
318
+ value: 0.0750151240169389
319
+ name: Cosine Precision@3
320
+ - type: cosine_precision@5
321
+ value: 0.07549909255898368
322
+ name: Cosine Precision@5
323
+ - type: cosine_precision@10
324
+ value: 0.06170598911070781
325
+ name: Cosine Precision@10
326
+ - type: cosine_recall@1
327
+ value: 0.016333938294010888
328
+ name: Cosine Recall@1
329
+ - type: cosine_recall@3
330
+ value: 0.2250453720508167
331
+ name: Cosine Recall@3
332
+ - type: cosine_recall@5
333
+ value: 0.37749546279491836
334
+ name: Cosine Recall@5
335
+ - type: cosine_recall@10
336
+ value: 0.617059891107078
337
+ name: Cosine Recall@10
338
+ - type: cosine_ndcg@10
339
+ value: 0.28661971571769745
340
+ name: Cosine Ndcg@10
341
+ - type: cosine_mrr@10
342
+ value: 0.18480756488923475
343
+ name: Cosine Mrr@10
344
+ - type: cosine_map@100
345
+ value: 0.20148744214489955
346
+ name: Cosine Map@100
347
+ ---
348
+
349
+ # SentenceTransformer based on BAAI/bge-m3
350
+
351
+ This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [BAAI/bge-m3](https://huggingface.co/BAAI/bge-m3). It maps sentences & paragraphs to a 1024-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more.
352
+
353
+ ## Model Details
354
+
355
+ ### Model Description
356
+ - **Model Type:** Sentence Transformer
357
+ - **Base model:** [BAAI/bge-m3](https://huggingface.co/BAAI/bge-m3) <!-- at revision 5617a9f61b028005a4858fdac845db406aefb181 -->
358
+ - **Maximum Sequence Length:** 1024 tokens
359
+ - **Output Dimensionality:** 1024 tokens
360
+ - **Similarity Function:** Cosine Similarity
361
+ <!-- - **Training Dataset:** Unknown -->
362
+ <!-- - **Language:** Unknown -->
363
+ <!-- - **License:** Unknown -->
364
+
365
+ ### Model Sources
366
+
367
+ - **Documentation:** [Sentence Transformers Documentation](https://sbert.net)
368
+ - **Repository:** [Sentence Transformers on GitHub](https://github.com/UKPLab/sentence-transformers)
369
+ - **Hugging Face:** [Sentence Transformers on Hugging Face](https://huggingface.co/models?library=sentence-transformers)
370
+
371
+ ### Full Model Architecture
372
+
373
+ ```
374
+ SentenceTransformer(
375
+ (0): Transformer({'max_seq_length': 1024, 'do_lower_case': False}) with Transformer model: XLMRobertaModel
376
+ (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})
377
+ (2): Normalize()
378
+ )
379
+ ```
380
+
381
+ ## Usage
382
+
383
+ ### Direct Usage (Sentence Transformers)
384
+
385
+ First install the Sentence Transformers library:
386
+
387
+ ```bash
388
+ pip install -U sentence-transformers
389
+ ```
390
+
391
+ Then you can load this model and run inference.
392
+ ```python
393
+ from sentence_transformers import SentenceTransformer
394
+
395
+ # Download from the 🤗 Hub
396
+ model = SentenceTransformer("akot/bge-semantic-bmf-matryoshka")
397
+ # Run inference
398
+ sentences = [
399
+ '67 Abwandlung des Beispiels 1 in Rn. 66: A erhält zudem zwei Kinderzulagen für seine in den Jahren 2004 und 2005 geborenen Kinder. Beitragspflichtige Einnahmen 53.000 € 4 % 2.120 € höchstens 2.100 € anzusetzen 2.100 € abzüglich Zulage 175 € Mindesteigenbeitrag (§ 86 Abs. 1 Satz 2 EStG) 1.925 € Sockelbetrag (§ 86 Abs. 1 Satz 4 EStG) 60 € maßgebend (§ 86 Abs. 1 Satz 5 EStG) 1.925 € Die von A geleisteten Beiträge übersteigen den Mindesteigenbeitrag. Die Zulage wird nicht gekürzt.',
400
+ 'Wird die Zulage für A gekürzt, wenn die Beiträge den Mindesteigenbeitrag übersteigen?',
401
+ 'Wie erfolgt die Besteuerung bei der ausgleichsberechtigten Person nach einer externen Teilung?',
402
+ ]
403
+ embeddings = model.encode(sentences)
404
+ print(embeddings.shape)
405
+ # [3, 1024]
406
+
407
+ # Get the similarity scores for the embeddings
408
+ similarities = model.similarity(embeddings, embeddings)
409
+ print(similarities.shape)
410
+ # [3, 3]
411
+ ```
412
+
413
+ <!--
414
+ ### Direct Usage (Transformers)
415
+
416
+ <details><summary>Click to see the direct usage in Transformers</summary>
417
+
418
+ </details>
419
+ -->
420
+
421
+ <!--
422
+ ### Downstream Usage (Sentence Transformers)
423
+
424
+ You can finetune this model on your own dataset.
425
+
426
+ <details><summary>Click to expand</summary>
427
+
428
+ </details>
429
+ -->
430
+
431
+ <!--
432
+ ### Out-of-Scope Use
433
+
434
+ *List how the model may foreseeably be misused and address what users ought not to do with the model.*
435
+ -->
436
+
437
+ ## Evaluation
438
+
439
+ ### Metrics
440
+
441
+ #### Information Retrieval
442
+ * Dataset: `dim_768`
443
+ * Evaluated with [<code>InformationRetrievalEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.InformationRetrievalEvaluator)
444
+
445
+ | Metric | Value |
446
+ |:--------------------|:-----------|
447
+ | cosine_accuracy@1 | 0.02 |
448
+ | cosine_accuracy@3 | 0.2486 |
449
+ | cosine_accuracy@5 | 0.3993 |
450
+ | cosine_accuracy@10 | 0.6588 |
451
+ | cosine_precision@1 | 0.02 |
452
+ | cosine_precision@3 | 0.0829 |
453
+ | cosine_precision@5 | 0.0799 |
454
+ | cosine_precision@10 | 0.0659 |
455
+ | cosine_recall@1 | 0.02 |
456
+ | cosine_recall@3 | 0.2486 |
457
+ | cosine_recall@5 | 0.3993 |
458
+ | cosine_recall@10 | 0.6588 |
459
+ | cosine_ndcg@10 | 0.3073 |
460
+ | cosine_mrr@10 | 0.1991 |
461
+ | **cosine_map@100** | **0.2159** |
462
+
463
+ #### Information Retrieval
464
+ * Dataset: `dim_512`
465
+ * Evaluated with [<code>InformationRetrievalEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.InformationRetrievalEvaluator)
466
+
467
+ | Metric | Value |
468
+ |:--------------------|:-----------|
469
+ | cosine_accuracy@1 | 0.029 |
470
+ | cosine_accuracy@3 | 0.2523 |
471
+ | cosine_accuracy@5 | 0.4029 |
472
+ | cosine_accuracy@10 | 0.6479 |
473
+ | cosine_precision@1 | 0.029 |
474
+ | cosine_precision@3 | 0.0841 |
475
+ | cosine_precision@5 | 0.0806 |
476
+ | cosine_precision@10 | 0.0648 |
477
+ | cosine_recall@1 | 0.029 |
478
+ | cosine_recall@3 | 0.2523 |
479
+ | cosine_recall@5 | 0.4029 |
480
+ | cosine_recall@10 | 0.6479 |
481
+ | cosine_ndcg@10 | 0.3084 |
482
+ | cosine_mrr@10 | 0.2038 |
483
+ | **cosine_map@100** | **0.2208** |
484
+
485
+ #### Information Retrieval
486
+ * Dataset: `dim_256`
487
+ * Evaluated with [<code>InformationRetrievalEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.InformationRetrievalEvaluator)
488
+
489
+ | Metric | Value |
490
+ |:--------------------|:-----------|
491
+ | cosine_accuracy@1 | 0.0345 |
492
+ | cosine_accuracy@3 | 0.2305 |
493
+ | cosine_accuracy@5 | 0.3866 |
494
+ | cosine_accuracy@10 | 0.6479 |
495
+ | cosine_precision@1 | 0.0345 |
496
+ | cosine_precision@3 | 0.0768 |
497
+ | cosine_precision@5 | 0.0773 |
498
+ | cosine_precision@10 | 0.0648 |
499
+ | cosine_recall@1 | 0.0345 |
500
+ | cosine_recall@3 | 0.2305 |
501
+ | cosine_recall@5 | 0.3866 |
502
+ | cosine_recall@10 | 0.6479 |
503
+ | cosine_ndcg@10 | 0.3041 |
504
+ | cosine_mrr@10 | 0.1991 |
505
+ | **cosine_map@100** | **0.2158** |
506
+
507
+ #### Information Retrieval
508
+ * Dataset: `dim_128`
509
+ * Evaluated with [<code>InformationRetrievalEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.InformationRetrievalEvaluator)
510
+
511
+ | Metric | Value |
512
+ |:--------------------|:-----------|
513
+ | cosine_accuracy@1 | 0.0272 |
514
+ | cosine_accuracy@3 | 0.2432 |
515
+ | cosine_accuracy@5 | 0.3866 |
516
+ | cosine_accuracy@10 | 0.6443 |
517
+ | cosine_precision@1 | 0.0272 |
518
+ | cosine_precision@3 | 0.0811 |
519
+ | cosine_precision@5 | 0.0773 |
520
+ | cosine_precision@10 | 0.0644 |
521
+ | cosine_recall@1 | 0.0272 |
522
+ | cosine_recall@3 | 0.2432 |
523
+ | cosine_recall@5 | 0.3866 |
524
+ | cosine_recall@10 | 0.6443 |
525
+ | cosine_ndcg@10 | 0.303 |
526
+ | cosine_mrr@10 | 0.1984 |
527
+ | **cosine_map@100** | **0.2151** |
528
+
529
+ #### Information Retrieval
530
+ * Dataset: `dim_64`
531
+ * Evaluated with [<code>InformationRetrievalEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.InformationRetrievalEvaluator)
532
+
533
+ | Metric | Value |
534
+ |:--------------------|:-----------|
535
+ | cosine_accuracy@1 | 0.0163 |
536
+ | cosine_accuracy@3 | 0.225 |
537
+ | cosine_accuracy@5 | 0.3775 |
538
+ | cosine_accuracy@10 | 0.6171 |
539
+ | cosine_precision@1 | 0.0163 |
540
+ | cosine_precision@3 | 0.075 |
541
+ | cosine_precision@5 | 0.0755 |
542
+ | cosine_precision@10 | 0.0617 |
543
+ | cosine_recall@1 | 0.0163 |
544
+ | cosine_recall@3 | 0.225 |
545
+ | cosine_recall@5 | 0.3775 |
546
+ | cosine_recall@10 | 0.6171 |
547
+ | cosine_ndcg@10 | 0.2866 |
548
+ | cosine_mrr@10 | 0.1848 |
549
+ | **cosine_map@100** | **0.2015** |
550
+
551
+ <!--
552
+ ## Bias, Risks and Limitations
553
+
554
+ *What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model.*
555
+ -->
556
+
557
+ <!--
558
+ ### Recommendations
559
+
560
+ *What are recommendations with respect to the foreseeable issues? For example, filtering explicit content.*
561
+ -->
562
+
563
+ ## Training Details
564
+
565
+ ### Training Dataset
566
+
567
+ #### Unnamed Dataset
568
+
569
+
570
+ * Size: 4,957 training samples
571
+ * Columns: <code>positive</code> and <code>anchor</code>
572
+ * Approximate statistics based on the first 1000 samples:
573
+ | | positive | anchor |
574
+ |:--------|:------------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------|
575
+ | type | string | string |
576
+ | details | <ul><li>min: 6 tokens</li><li>mean: 176.7 tokens</li><li>max: 1024 tokens</li></ul> | <ul><li>min: 10 tokens</li><li>mean: 24.39 tokens</li><li>max: 51 tokens</li></ul> |
577
+ * Samples:
578
+ | positive | anchor |
579
+ |:-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:---------------------------------------------------------------------------------------------------------------------------------------------------------------------|
580
+ | <code>134 Eine Rückzahlungsverpflichtung besteht nicht für den Teil der Zulagen, der auf nach § 1 Abs. 1 Nr. 2 AltZertG angespartes gefördertes Altersvorsorgevermögen entfällt, wenn es in Form einer Hinterbliebenenrente an die dort genannten Hinterbliebenen ausgezahlt wird. Dies gilt auch für den entsprechenden Teil der Steuerermäßigung.</code> | <code>Muss man Zulagen zurückzahlen, wenn das Altersvorsorgevermögen als Hinterbliebenenrente ausgezahlt wird?</code> |
581
+ | <code>140 Beendet der Zulageberechtigte vor der vollständigen Rückzahlung des AltersvorsorgeEigenheimbetrags die Nutzung zu eigenen Wohnzwecken, wird er so behandelt, als habe er den noch nicht zurückgezahlten Betrag schädlich verwendet. Die auf den noch ausstehenden Rückzahlungsbetrag entfallenden Zulagen sowie die nach § 10a Abs. 4 EStG gesondert festgestellten Steuerermäßigungen sind zurückzuzahlen (§ 92a Abs. 3 EStG). Die im noch ausstehenden Rückzahlungsbetrag enthaltenen Zuwächse (z.B. Zinserträge und Kursgewinne) Seite 41 sind als sonstige Einkünfte zu versteuern (§ 22 Nr. 5 Satz 5 Halbsatz 1 EStG). Außerdem hat der Zulageberechtigte den Vorteil zu versteuern, der sich aus der zinslosen Nutzung des noch nicht zurückgezahlten Betrags ergibt. Zugrunde gelegt wird hierbei eine Verzinsung von 5 % (Zins und Zinseszins) für jedes volle Kalenderjahr der Nutzung (§ 22 Nr. 5 Satz 5 Halbsatz 2 EStG). Diese Folgen treten nicht ein, wenn er den noch nicht zurückgezahlten Betrag in ein Folgeobjekt investiert (§ 92a Abs. 4 Satz 3 Nr. 1 EStG) oder zugunsten eines auf seinen Namen lautenden zertifizierten Altersvorsorgevertrags einzahlt (§ 92a Abs. 4 Satz 3 Nr. 2 EStG).</code> | <code>Was geschieht steuerlich, wenn der AltersvorsorgeEigenheimbetrag nicht vollständig zurückgezahlt wird und die Immobilie nicht mehr selbst genutzt wird?</code> |
582
+ | <code>144 Die als Einkünfte nach § 22 Nr. 5 Satz 3 EStG i.V.m. § 22 Nr. 5 Satz 2 EStG zu besteuernden Beträge muss der Anbieter gem. § 94 Abs. 1 Satz 4 EStG dem Zulageberechtigten bescheinigen und im Wege des Rentenbezugsmitteilungsverfahrens (§ 22a EStG) mitteilen. Ergeben sich insoweit steuerpflichtige Einkünfte nach § 22 Nr. 5 Satz 3 EStG für einen anderen Leistungsempfänger (z. B. Erben), ist für diesen eine entsprechende Rentenbezugsmitteilung der ZfA zu übermitteln.</code> | <code>Was muss im Falle eines anderen Leistungsempfängers, wie Erben, hinsichtlich der Rentenbezugsmitteilung getan werden?</code> |
583
+ * Loss: [<code>MatryoshkaLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#matryoshkaloss) with these parameters:
584
+ ```json
585
+ {
586
+ "loss": "MultipleNegativesRankingLoss",
587
+ "matryoshka_dims": [
588
+ 768,
589
+ 512,
590
+ 256,
591
+ 128,
592
+ 64
593
+ ],
594
+ "matryoshka_weights": [
595
+ 1,
596
+ 1,
597
+ 1,
598
+ 1,
599
+ 1
600
+ ],
601
+ "n_dims_per_step": -1
602
+ }
603
+ ```
604
+
605
+ ### Training Hyperparameters
606
+ #### Non-Default Hyperparameters
607
+
608
+ - `eval_strategy`: epoch
609
+ - `per_device_train_batch_size`: 16
610
+ - `per_device_eval_batch_size`: 16
611
+ - `gradient_accumulation_steps`: 16
612
+ - `learning_rate`: 2e-05
613
+ - `num_train_epochs`: 10
614
+ - `lr_scheduler_type`: cosine
615
+ - `warmup_ratio`: 0.1
616
+ - `bf16`: True
617
+ - `tf32`: True
618
+ - `load_best_model_at_end`: True
619
+ - `optim`: adamw_torch_fused
620
+ - `batch_sampler`: no_duplicates
621
+
622
+ #### All Hyperparameters
623
+ <details><summary>Click to expand</summary>
624
+
625
+ - `overwrite_output_dir`: False
626
+ - `do_predict`: False
627
+ - `eval_strategy`: epoch
628
+ - `prediction_loss_only`: True
629
+ - `per_device_train_batch_size`: 16
630
+ - `per_device_eval_batch_size`: 16
631
+ - `per_gpu_train_batch_size`: None
632
+ - `per_gpu_eval_batch_size`: None
633
+ - `gradient_accumulation_steps`: 16
634
+ - `eval_accumulation_steps`: None
635
+ - `torch_empty_cache_steps`: None
636
+ - `learning_rate`: 2e-05
637
+ - `weight_decay`: 0.0
638
+ - `adam_beta1`: 0.9
639
+ - `adam_beta2`: 0.999
640
+ - `adam_epsilon`: 1e-08
641
+ - `max_grad_norm`: 1.0
642
+ - `num_train_epochs`: 10
643
+ - `max_steps`: -1
644
+ - `lr_scheduler_type`: cosine
645
+ - `lr_scheduler_kwargs`: {}
646
+ - `warmup_ratio`: 0.1
647
+ - `warmup_steps`: 0
648
+ - `log_level`: passive
649
+ - `log_level_replica`: warning
650
+ - `log_on_each_node`: True
651
+ - `logging_nan_inf_filter`: True
652
+ - `save_safetensors`: True
653
+ - `save_on_each_node`: False
654
+ - `save_only_model`: False
655
+ - `restore_callback_states_from_checkpoint`: False
656
+ - `no_cuda`: False
657
+ - `use_cpu`: False
658
+ - `use_mps_device`: False
659
+ - `seed`: 42
660
+ - `data_seed`: None
661
+ - `jit_mode_eval`: False
662
+ - `use_ipex`: False
663
+ - `bf16`: True
664
+ - `fp16`: False
665
+ - `fp16_opt_level`: O1
666
+ - `half_precision_backend`: auto
667
+ - `bf16_full_eval`: False
668
+ - `fp16_full_eval`: False
669
+ - `tf32`: True
670
+ - `local_rank`: 0
671
+ - `ddp_backend`: None
672
+ - `tpu_num_cores`: None
673
+ - `tpu_metrics_debug`: False
674
+ - `debug`: []
675
+ - `dataloader_drop_last`: False
676
+ - `dataloader_num_workers`: 0
677
+ - `dataloader_prefetch_factor`: None
678
+ - `past_index`: -1
679
+ - `disable_tqdm`: False
680
+ - `remove_unused_columns`: True
681
+ - `label_names`: None
682
+ - `load_best_model_at_end`: True
683
+ - `ignore_data_skip`: False
684
+ - `fsdp`: []
685
+ - `fsdp_min_num_params`: 0
686
+ - `fsdp_config`: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}
687
+ - `fsdp_transformer_layer_cls_to_wrap`: None
688
+ - `accelerator_config`: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
689
+ - `deepspeed`: None
690
+ - `label_smoothing_factor`: 0.0
691
+ - `optim`: adamw_torch_fused
692
+ - `optim_args`: None
693
+ - `adafactor`: False
694
+ - `group_by_length`: False
695
+ - `length_column_name`: length
696
+ - `ddp_find_unused_parameters`: None
697
+ - `ddp_bucket_cap_mb`: None
698
+ - `ddp_broadcast_buffers`: False
699
+ - `dataloader_pin_memory`: True
700
+ - `dataloader_persistent_workers`: False
701
+ - `skip_memory_metrics`: True
702
+ - `use_legacy_prediction_loop`: False
703
+ - `push_to_hub`: False
704
+ - `resume_from_checkpoint`: None
705
+ - `hub_model_id`: None
706
+ - `hub_strategy`: every_save
707
+ - `hub_private_repo`: False
708
+ - `hub_always_push`: False
709
+ - `gradient_checkpointing`: False
710
+ - `gradient_checkpointing_kwargs`: None
711
+ - `include_inputs_for_metrics`: False
712
+ - `eval_do_concat_batches`: True
713
+ - `fp16_backend`: auto
714
+ - `push_to_hub_model_id`: None
715
+ - `push_to_hub_organization`: None
716
+ - `mp_parameters`:
717
+ - `auto_find_batch_size`: False
718
+ - `full_determinism`: False
719
+ - `torchdynamo`: None
720
+ - `ray_scope`: last
721
+ - `ddp_timeout`: 1800
722
+ - `torch_compile`: False
723
+ - `torch_compile_backend`: None
724
+ - `torch_compile_mode`: None
725
+ - `dispatch_batches`: None
726
+ - `split_batches`: None
727
+ - `include_tokens_per_second`: False
728
+ - `include_num_input_tokens_seen`: False
729
+ - `neftune_noise_alpha`: None
730
+ - `optim_target_modules`: None
731
+ - `batch_eval_metrics`: False
732
+ - `eval_on_start`: False
733
+ - `eval_use_gather_object`: False
734
+ - `batch_sampler`: no_duplicates
735
+ - `multi_dataset_batch_sampler`: proportional
736
+
737
+ </details>
738
+
739
+ ### Training Logs
740
+ | Epoch | Step | Training Loss | dim_128_cosine_map@100 | dim_256_cosine_map@100 | dim_512_cosine_map@100 | dim_64_cosine_map@100 | dim_768_cosine_map@100 |
741
+ |:----------:|:------:|:-------------:|:----------------------:|:----------------------:|:----------------------:|:---------------------:|:----------------------:|
742
+ | 0.5161 | 10 | 2.8906 | - | - | - | - | - |
743
+ | 0.9806 | 19 | - | 0.1981 | 0.2156 | 0.2153 | 0.1849 | 0.2183 |
744
+ | 1.0323 | 20 | 1.6649 | - | - | - | - | - |
745
+ | 1.5484 | 30 | 0.992 | - | - | - | - | - |
746
+ | 1.9613 | 38 | - | 0.2141 | 0.2158 | 0.2206 | 0.1972 | 0.2196 |
747
+ | 2.0645 | 40 | 0.6799 | - | - | - | - | - |
748
+ | 2.5806 | 50 | 0.4886 | - | - | - | - | - |
749
+ | 2.9935 | 58 | - | 0.2145 | 0.2080 | 0.2205 | 0.2072 | 0.2177 |
750
+ | 3.0968 | 60 | 0.3464 | - | - | - | - | - |
751
+ | 3.6129 | 70 | 0.29 | - | - | - | - | - |
752
+ | 3.9742 | 77 | - | 0.2114 | 0.2161 | 0.2220 | 0.2074 | 0.2209 |
753
+ | 4.1290 | 80 | 0.2217 | - | - | - | - | - |
754
+ | 4.6452 | 90 | 0.2296 | - | - | - | - | - |
755
+ | **4.9548** | **96** | **-** | **0.2325** | **0.22** | **0.2283** | **0.2104** | **0.2231** |
756
+ | 5.1613 | 100 | 0.1665 | - | - | - | - | - |
757
+ | 5.6774 | 110 | 0.18 | - | - | - | - | - |
758
+ | 5.9871 | 116 | - | 0.2177 | 0.2152 | 0.2234 | 0.2061 | 0.2241 |
759
+ | 6.1935 | 120 | 0.131 | - | - | - | - | - |
760
+ | 6.7097 | 130 | 0.1502 | - | - | - | - | - |
761
+ | 6.9677 | 135 | - | 0.2127 | 0.2161 | 0.2248 | 0.2037 | 0.2226 |
762
+ | 7.2258 | 140 | 0.116 | - | - | - | - | - |
763
+ | 7.7419 | 150 | 0.1363 | - | - | - | - | - |
764
+ | 8.0 | 155 | - | 0.2196 | 0.2159 | 0.2239 | 0.2077 | 0.2233 |
765
+ | 8.2581 | 160 | 0.0976 | - | - | - | - | - |
766
+ | 8.7742 | 170 | 0.1242 | - | - | - | - | - |
767
+ | 8.9806 | 174 | - | 0.2153 | 0.2203 | 0.2293 | 0.2042 | 0.2192 |
768
+ | 9.2903 | 180 | 0.109 | - | - | - | - | - |
769
+ | 9.8065 | 190 | 0.1132 | 0.2151 | 0.2158 | 0.2208 | 0.2015 | 0.2159 |
770
+
771
+ * The bold row denotes the saved checkpoint.
772
+
773
+ ### Framework Versions
774
+ - Python: 3.11.4
775
+ - Sentence Transformers: 3.0.1
776
+ - Transformers: 4.44.0
777
+ - PyTorch: 2.4.0+cu121
778
+ - Accelerate: 0.33.0
779
+ - Datasets: 2.19.1
780
+ - Tokenizers: 0.19.1
781
+
782
+ ## Citation
783
+
784
+ ### BibTeX
785
+
786
+ #### Sentence Transformers
787
+ ```bibtex
788
+ @inproceedings{reimers-2019-sentence-bert,
789
+ title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks",
790
+ author = "Reimers, Nils and Gurevych, Iryna",
791
+ booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing",
792
+ month = "11",
793
+ year = "2019",
794
+ publisher = "Association for Computational Linguistics",
795
+ url = "https://arxiv.org/abs/1908.10084",
796
+ }
797
+ ```
798
+
799
+ #### MatryoshkaLoss
800
+ ```bibtex
801
+ @misc{kusupati2024matryoshka,
802
+ title={Matryoshka Representation Learning},
803
+ author={Aditya Kusupati and Gantavya Bhatt and Aniket Rege and Matthew Wallingford and Aditya Sinha and Vivek Ramanujan and William Howard-Snyder and Kaifeng Chen and Sham Kakade and Prateek Jain and Ali Farhadi},
804
+ year={2024},
805
+ eprint={2205.13147},
806
+ archivePrefix={arXiv},
807
+ primaryClass={cs.LG}
808
+ }
809
+ ```
810
+
811
+ #### MultipleNegativesRankingLoss
812
+ ```bibtex
813
+ @misc{henderson2017efficient,
814
+ title={Efficient Natural Language Response Suggestion for Smart Reply},
815
+ author={Matthew Henderson and Rami Al-Rfou and Brian Strope and Yun-hsuan Sung and Laszlo Lukacs and Ruiqi Guo and Sanjiv Kumar and Balint Miklos and Ray Kurzweil},
816
+ year={2017},
817
+ eprint={1705.00652},
818
+ archivePrefix={arXiv},
819
+ primaryClass={cs.CL}
820
+ }
821
+ ```
822
+
823
+ <!--
824
+ ## Glossary
825
+
826
+ *Clearly define terms in order to be accessible across audiences.*
827
+ -->
828
+
829
+ <!--
830
+ ## Model Card Authors
831
+
832
+ *Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction.*
833
+ -->
834
+
835
+ <!--
836
+ ## Model Card Contact
837
+
838
+ *Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors.*
839
+ -->
config.json ADDED
@@ -0,0 +1,28 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "BAAI/bge-m3",
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": "4.44.0",
25
+ "type_vocab_size": 1,
26
+ "use_cache": true,
27
+ "vocab_size": 250002
28
+ }
config_sentence_transformers.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "__version__": {
3
+ "sentence_transformers": "3.0.1",
4
+ "transformers": "4.44.0",
5
+ "pytorch": "2.4.0+cu121"
6
+ },
7
+ "prompts": {},
8
+ "default_prompt_name": null,
9
+ "similarity_fn_name": null
10
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0574a7dc1b87017509bdeccb8323fd13b48764752ecf105eb2dbcde58f5cb320
3
+ size 2271064456
modules.json ADDED
@@ -0,0 +1,20 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ [
2
+ {
3
+ "idx": 0,
4
+ "name": "0",
5
+ "path": "",
6
+ "type": "sentence_transformers.models.Transformer"
7
+ },
8
+ {
9
+ "idx": 1,
10
+ "name": "1",
11
+ "path": "1_Pooling",
12
+ "type": "sentence_transformers.models.Pooling"
13
+ },
14
+ {
15
+ "idx": 2,
16
+ "name": "2",
17
+ "path": "2_Normalize",
18
+ "type": "sentence_transformers.models.Normalize"
19
+ }
20
+ ]
sentence_bert_config.json ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ {
2
+ "max_seq_length": 1024,
3
+ "do_lower_case": false
4
+ }
special_tokens_map.json ADDED
@@ -0,0 +1,51 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "cls_token": {
10
+ "content": "<s>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "eos_token": {
17
+ "content": "</s>",
18
+ "lstrip": false,
19
+ "normalized": false,
20
+ "rstrip": false,
21
+ "single_word": false
22
+ },
23
+ "mask_token": {
24
+ "content": "<mask>",
25
+ "lstrip": true,
26
+ "normalized": false,
27
+ "rstrip": false,
28
+ "single_word": false
29
+ },
30
+ "pad_token": {
31
+ "content": "<pad>",
32
+ "lstrip": false,
33
+ "normalized": false,
34
+ "rstrip": false,
35
+ "single_word": false
36
+ },
37
+ "sep_token": {
38
+ "content": "</s>",
39
+ "lstrip": false,
40
+ "normalized": false,
41
+ "rstrip": false,
42
+ "single_word": false
43
+ },
44
+ "unk_token": {
45
+ "content": "<unk>",
46
+ "lstrip": false,
47
+ "normalized": false,
48
+ "rstrip": false,
49
+ "single_word": false
50
+ }
51
+ }
tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6e3b8957de04e3a4ed42b1a11381556f9adad8d0d502b9dd071c75f626b28f40
3
+ size 17083053
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
+ }