much1na commited on
Commit
5c3777b
·
verified ·
1 Parent(s): 09aa980

Upload folder using huggingface_hub

Browse files
checkpoint-368000/1_Pooling/config.json ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ {
2
+ "embedding_dimension": 512,
3
+ "pooling_mode": "mean",
4
+ "include_prompt": true
5
+ }
checkpoint-368000/2_Dense/config.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "in_features": 512,
3
+ "out_features": 1024,
4
+ "bias": false,
5
+ "activation_function": "torch.nn.modules.linear.Identity",
6
+ "module_input_name": "sentence_embedding",
7
+ "module_output_name": "sentence_embedding"
8
+ }
checkpoint-368000/2_Dense/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4a3840900f56c563e57a39255f848a4c4d5591f91597ed9d14e474cdc485f7ed
3
+ size 2097240
checkpoint-368000/3_Normalize/config.json ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ {
2
+ "module_input_name": "sentence_embedding",
3
+ "module_output_name": "sentence_embedding"
4
+ }
checkpoint-368000/README.md ADDED
@@ -0,0 +1,1448 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ tags:
3
+ - sentence-transformers
4
+ - sentence-similarity
5
+ - feature-extraction
6
+ - dense
7
+ - generated_from_trainer
8
+ - loss:EmbedDistillLoss
9
+ base_model: much1na/distil-cohere-v1-ckpt
10
+ widget:
11
+ - source_sentence: tty number
12
+ sentences:
13
+ - how old jojo
14
+ - what is nrc
15
+ - boulle work
16
+ - source_sentence: rbbb causes
17
+ sentences:
18
+ - what is gpr?
19
+ - ireland food
20
+ - define adobe
21
+ - source_sentence: 'List of HBO Films films This is a list of films produced and distributed
22
+ by HBO Films.
23
+
24
+ 1980s
25
+
26
+ 1990s
27
+
28
+ 2000s
29
+
30
+ 2010s
31
+
32
+ 2020s
33
+
34
+ Upcoming
35
+
36
+ HBO Documentary Films
37
+
38
+ HBO Comedy Specials'
39
+ sentences:
40
+ - 'The Seven Basic Plots The Seven Basic Plots: Why We Tell Stories is a 2004 book
41
+ by Christopher Booker containing a Jung-influenced analysis of stories and their
42
+ psychological meaning. Booker worked on the book for 34 years.
43
+
44
+ Summary'
45
+ - Empire of the Sun (band) Empire of the Sun is an Australian electronic music duo
46
+ formed in 2007. The duo is a collaboration between Luke Steele, of alternative
47
+ rock band The Sleepy Jackson, and Nick Littlemore, of electronic dance band Pnau.
48
+ - 'Going for Gold Going for Gold is a British television game show that originally
49
+ aired on BBC1 between 12 October 1987 and 9 July 1996. It was revived for Channel
50
+ 5 from 13 October 2008 to 20 March 2009.
51
+
52
+ Background'
53
+ - source_sentence: "Tex Stackexchange\n\nQ: LYx 2.2. (Cannot view the file pdf) I\
54
+ \ get this error when trying to view the file:\nThe external program pdflatex\
55
+ \ finished with an error ... check the logs.\n\n(I am using LyX 2.2.0 Windows\
56
+ \ 7) \n"
57
+ sentences:
58
+ - "Apple Stackexchange\n\nQ: Turn Wifi Off via command line I'm sharing my internet\
59
+ \ in work and forgot to turn it off! Does anyone know how I can do it via the\
60
+ \ command line? I have SSH access but thats it. D'oh. \n\nA: Try \nnetworksetup\
61
+ \ -setairportpower en0 off\n\n"
62
+ - 'Stackoverflow Stackexchange
63
+
64
+
65
+ Q: Convert Bitmap to File I understand that using BitmapFactory can convert a
66
+ File to a Bitmap, but is there any way to convert a Bitmap image to a File?
67
+
68
+
69
+ A: Try this:
70
+
71
+ bitmap.compress(Bitmap.CompressFormat.PNG, quality, outStream);
72
+
73
+
74
+ See this
75
+
76
+ '
77
+ - 'Stackoverflow Stackexchange
78
+
79
+
80
+ Q: How to run 16 bit code on 32 bit Linux? I have written a small 16-bit assembly
81
+ program that writes some values in some memory locations.Is there a way I can
82
+ test it in 32-bit protected mode on Linux?
83
+
84
+
85
+ A: qemu, dosbox, bochs
86
+
87
+ '
88
+ - source_sentence: 'Stackoverflow Stackexchange
89
+
90
+
91
+ Q: What is a good gitignore to use with Codeblocks C++ Projects? This is what
92
+ I have so far:
93
+
94
+ obj/
95
+
96
+ bin/
97
+
98
+
99
+ Anything else?
100
+
101
+
102
+ A: You can use this:
103
+
104
+ *.layout
105
+
106
+ *.depend
107
+
108
+ bin/
109
+
110
+ obj/
111
+
112
+
113
+ See:
114
+
115
+ https://www.gitignore.io/api/codeblocks
116
+
117
+ '
118
+ sentences:
119
+ - 'Stackoverflow Stackexchange
120
+
121
+
122
+ Q: Listen on a network port and save data to a text file Is there some easy way
123
+ in linux to listen for data on network and save them to a text file?
124
+
125
+ Thank you.
126
+
127
+
128
+ A: Also available is TCPDump.
129
+
130
+ sudo tcpdump -i eth0 -vvvvtttAXns 1500 ''port 10000''
131
+
132
+
133
+ '
134
+ - 'Stackoverflow Stackexchange
135
+
136
+
137
+ Q: Get the current date and time I want to get the current date and time.
138
+
139
+ For example:
140
+
141
+ 2012/11/13 06:30:38
142
+
143
+
144
+ What I have tried:
145
+
146
+ Dim d As System.DateTime
147
+
148
+ MsgBox(d.Year) ''Return 1
149
+
150
+
151
+
152
+ A: use DateTime.Now
153
+
154
+ try this:
155
+
156
+ DateTime.Now.ToString("yyyy/MM/dd HH:mm:ss")
157
+
158
+
159
+ '
160
+ - 'Stackoverflow Stackexchange
161
+
162
+
163
+ Q: Can azure CLI get a list of all resources running in a tenant? Is there a single
164
+ command way to get all the resources in a tenant dumped out in JSON format, either
165
+ in az cli or powershell?
166
+
167
+
168
+ A: Azure CLI
169
+
170
+ az resource list
171
+
172
+
173
+ Powershell
174
+
175
+ Get-AzureRmResource
176
+
177
+
178
+ '
179
+ pipeline_tag: sentence-similarity
180
+ library_name: sentence-transformers
181
+ metrics:
182
+ - cosine_accuracy@1
183
+ - cosine_accuracy@3
184
+ - cosine_accuracy@5
185
+ - cosine_accuracy@10
186
+ - cosine_precision@1
187
+ - cosine_precision@3
188
+ - cosine_precision@5
189
+ - cosine_precision@10
190
+ - cosine_recall@1
191
+ - cosine_recall@3
192
+ - cosine_recall@5
193
+ - cosine_recall@10
194
+ - cosine_ndcg@10
195
+ - cosine_mrr@10
196
+ - cosine_map@100
197
+ model-index:
198
+ - name: SentenceTransformer based on much1na/distil-cohere-v1-ckpt
199
+ results:
200
+ - task:
201
+ type: information-retrieval
202
+ name: Information Retrieval
203
+ dataset:
204
+ name: NanoClimateFEVER
205
+ type: NanoClimateFEVER
206
+ metrics:
207
+ - type: cosine_accuracy@1
208
+ value: 0.3
209
+ name: Cosine Accuracy@1
210
+ - type: cosine_accuracy@3
211
+ value: 0.48
212
+ name: Cosine Accuracy@3
213
+ - type: cosine_accuracy@5
214
+ value: 0.52
215
+ name: Cosine Accuracy@5
216
+ - type: cosine_accuracy@10
217
+ value: 0.62
218
+ name: Cosine Accuracy@10
219
+ - type: cosine_precision@1
220
+ value: 0.3
221
+ name: Cosine Precision@1
222
+ - type: cosine_precision@3
223
+ value: 0.18666666666666665
224
+ name: Cosine Precision@3
225
+ - type: cosine_precision@5
226
+ value: 0.124
227
+ name: Cosine Precision@5
228
+ - type: cosine_precision@10
229
+ value: 0.084
230
+ name: Cosine Precision@10
231
+ - type: cosine_recall@1
232
+ value: 0.15333333333333332
233
+ name: Cosine Recall@1
234
+ - type: cosine_recall@3
235
+ value: 0.23399999999999999
236
+ name: Cosine Recall@3
237
+ - type: cosine_recall@5
238
+ value: 0.25233333333333335
239
+ name: Cosine Recall@5
240
+ - type: cosine_recall@10
241
+ value: 0.32299999999999995
242
+ name: Cosine Recall@10
243
+ - type: cosine_ndcg@10
244
+ value: 0.29189003667470415
245
+ name: Cosine Ndcg@10
246
+ - type: cosine_mrr@10
247
+ value: 0.3999126984126985
248
+ name: Cosine Mrr@10
249
+ - type: cosine_map@100
250
+ value: 0.2516840415391831
251
+ name: Cosine Map@100
252
+ - task:
253
+ type: information-retrieval
254
+ name: Information Retrieval
255
+ dataset:
256
+ name: NanoDBPedia
257
+ type: NanoDBPedia
258
+ metrics:
259
+ - type: cosine_accuracy@1
260
+ value: 0.74
261
+ name: Cosine Accuracy@1
262
+ - type: cosine_accuracy@3
263
+ value: 0.84
264
+ name: Cosine Accuracy@3
265
+ - type: cosine_accuracy@5
266
+ value: 0.88
267
+ name: Cosine Accuracy@5
268
+ - type: cosine_accuracy@10
269
+ value: 0.94
270
+ name: Cosine Accuracy@10
271
+ - type: cosine_precision@1
272
+ value: 0.74
273
+ name: Cosine Precision@1
274
+ - type: cosine_precision@3
275
+ value: 0.5466666666666666
276
+ name: Cosine Precision@3
277
+ - type: cosine_precision@5
278
+ value: 0.48399999999999993
279
+ name: Cosine Precision@5
280
+ - type: cosine_precision@10
281
+ value: 0.446
282
+ name: Cosine Precision@10
283
+ - type: cosine_recall@1
284
+ value: 0.07789454341315716
285
+ name: Cosine Recall@1
286
+ - type: cosine_recall@3
287
+ value: 0.14709388513196536
288
+ name: Cosine Recall@3
289
+ - type: cosine_recall@5
290
+ value: 0.1870910227526472
291
+ name: Cosine Recall@5
292
+ - type: cosine_recall@10
293
+ value: 0.2928987864758593
294
+ name: Cosine Recall@10
295
+ - type: cosine_ndcg@10
296
+ value: 0.5529033661694585
297
+ name: Cosine Ndcg@10
298
+ - type: cosine_mrr@10
299
+ value: 0.8068888888888889
300
+ name: Cosine Mrr@10
301
+ - type: cosine_map@100
302
+ value: 0.3997469829295745
303
+ name: Cosine Map@100
304
+ - task:
305
+ type: information-retrieval
306
+ name: Information Retrieval
307
+ dataset:
308
+ name: NanoFEVER
309
+ type: NanoFEVER
310
+ metrics:
311
+ - type: cosine_accuracy@1
312
+ value: 0.56
313
+ name: Cosine Accuracy@1
314
+ - type: cosine_accuracy@3
315
+ value: 0.72
316
+ name: Cosine Accuracy@3
317
+ - type: cosine_accuracy@5
318
+ value: 0.82
319
+ name: Cosine Accuracy@5
320
+ - type: cosine_accuracy@10
321
+ value: 0.96
322
+ name: Cosine Accuracy@10
323
+ - type: cosine_precision@1
324
+ value: 0.56
325
+ name: Cosine Precision@1
326
+ - type: cosine_precision@3
327
+ value: 0.24666666666666667
328
+ name: Cosine Precision@3
329
+ - type: cosine_precision@5
330
+ value: 0.16799999999999998
331
+ name: Cosine Precision@5
332
+ - type: cosine_precision@10
333
+ value: 0.09999999999999998
334
+ name: Cosine Precision@10
335
+ - type: cosine_recall@1
336
+ value: 0.53
337
+ name: Cosine Recall@1
338
+ - type: cosine_recall@3
339
+ value: 0.6766666666666665
340
+ name: Cosine Recall@3
341
+ - type: cosine_recall@5
342
+ value: 0.7766666666666666
343
+ name: Cosine Recall@5
344
+ - type: cosine_recall@10
345
+ value: 0.9266666666666667
346
+ name: Cosine Recall@10
347
+ - type: cosine_ndcg@10
348
+ value: 0.7212617467191452
349
+ name: Cosine Ndcg@10
350
+ - type: cosine_mrr@10
351
+ value: 0.6735555555555557
352
+ name: Cosine Mrr@10
353
+ - type: cosine_map@100
354
+ value: 0.651959758620174
355
+ name: Cosine Map@100
356
+ - task:
357
+ type: information-retrieval
358
+ name: Information Retrieval
359
+ dataset:
360
+ name: NanoFiQA2018
361
+ type: NanoFiQA2018
362
+ metrics:
363
+ - type: cosine_accuracy@1
364
+ value: 0.34
365
+ name: Cosine Accuracy@1
366
+ - type: cosine_accuracy@3
367
+ value: 0.56
368
+ name: Cosine Accuracy@3
369
+ - type: cosine_accuracy@5
370
+ value: 0.62
371
+ name: Cosine Accuracy@5
372
+ - type: cosine_accuracy@10
373
+ value: 0.72
374
+ name: Cosine Accuracy@10
375
+ - type: cosine_precision@1
376
+ value: 0.34
377
+ name: Cosine Precision@1
378
+ - type: cosine_precision@3
379
+ value: 0.24
380
+ name: Cosine Precision@3
381
+ - type: cosine_precision@5
382
+ value: 0.184
383
+ name: Cosine Precision@5
384
+ - type: cosine_precision@10
385
+ value: 0.11399999999999999
386
+ name: Cosine Precision@10
387
+ - type: cosine_recall@1
388
+ value: 0.19374603174603172
389
+ name: Cosine Recall@1
390
+ - type: cosine_recall@3
391
+ value: 0.3669365079365079
392
+ name: Cosine Recall@3
393
+ - type: cosine_recall@5
394
+ value: 0.464015873015873
395
+ name: Cosine Recall@5
396
+ - type: cosine_recall@10
397
+ value: 0.5427619047619047
398
+ name: Cosine Recall@10
399
+ - type: cosine_ndcg@10
400
+ value: 0.4234107861413032
401
+ name: Cosine Ndcg@10
402
+ - type: cosine_mrr@10
403
+ value: 0.4598809523809524
404
+ name: Cosine Mrr@10
405
+ - type: cosine_map@100
406
+ value: 0.3572454750315232
407
+ name: Cosine Map@100
408
+ - task:
409
+ type: information-retrieval
410
+ name: Information Retrieval
411
+ dataset:
412
+ name: NanoHotpotQA
413
+ type: NanoHotpotQA
414
+ metrics:
415
+ - type: cosine_accuracy@1
416
+ value: 0.78
417
+ name: Cosine Accuracy@1
418
+ - type: cosine_accuracy@3
419
+ value: 0.9
420
+ name: Cosine Accuracy@3
421
+ - type: cosine_accuracy@5
422
+ value: 0.92
423
+ name: Cosine Accuracy@5
424
+ - type: cosine_accuracy@10
425
+ value: 1.0
426
+ name: Cosine Accuracy@10
427
+ - type: cosine_precision@1
428
+ value: 0.78
429
+ name: Cosine Precision@1
430
+ - type: cosine_precision@3
431
+ value: 0.43999999999999995
432
+ name: Cosine Precision@3
433
+ - type: cosine_precision@5
434
+ value: 0.284
435
+ name: Cosine Precision@5
436
+ - type: cosine_precision@10
437
+ value: 0.15999999999999998
438
+ name: Cosine Precision@10
439
+ - type: cosine_recall@1
440
+ value: 0.39
441
+ name: Cosine Recall@1
442
+ - type: cosine_recall@3
443
+ value: 0.66
444
+ name: Cosine Recall@3
445
+ - type: cosine_recall@5
446
+ value: 0.71
447
+ name: Cosine Recall@5
448
+ - type: cosine_recall@10
449
+ value: 0.8
450
+ name: Cosine Recall@10
451
+ - type: cosine_ndcg@10
452
+ value: 0.7354996123049614
453
+ name: Cosine Ndcg@10
454
+ - type: cosine_mrr@10
455
+ value: 0.8485238095238096
456
+ name: Cosine Mrr@10
457
+ - type: cosine_map@100
458
+ value: 0.6593936674010843
459
+ name: Cosine Map@100
460
+ - task:
461
+ type: information-retrieval
462
+ name: Information Retrieval
463
+ dataset:
464
+ name: NanoMSMARCO
465
+ type: NanoMSMARCO
466
+ metrics:
467
+ - type: cosine_accuracy@1
468
+ value: 0.44
469
+ name: Cosine Accuracy@1
470
+ - type: cosine_accuracy@3
471
+ value: 0.64
472
+ name: Cosine Accuracy@3
473
+ - type: cosine_accuracy@5
474
+ value: 0.7
475
+ name: Cosine Accuracy@5
476
+ - type: cosine_accuracy@10
477
+ value: 0.76
478
+ name: Cosine Accuracy@10
479
+ - type: cosine_precision@1
480
+ value: 0.44
481
+ name: Cosine Precision@1
482
+ - type: cosine_precision@3
483
+ value: 0.21333333333333332
484
+ name: Cosine Precision@3
485
+ - type: cosine_precision@5
486
+ value: 0.14
487
+ name: Cosine Precision@5
488
+ - type: cosine_precision@10
489
+ value: 0.07600000000000001
490
+ name: Cosine Precision@10
491
+ - type: cosine_recall@1
492
+ value: 0.44
493
+ name: Cosine Recall@1
494
+ - type: cosine_recall@3
495
+ value: 0.64
496
+ name: Cosine Recall@3
497
+ - type: cosine_recall@5
498
+ value: 0.7
499
+ name: Cosine Recall@5
500
+ - type: cosine_recall@10
501
+ value: 0.76
502
+ name: Cosine Recall@10
503
+ - type: cosine_ndcg@10
504
+ value: 0.6043814421453638
505
+ name: Cosine Ndcg@10
506
+ - type: cosine_mrr@10
507
+ value: 0.5544126984126984
508
+ name: Cosine Mrr@10
509
+ - type: cosine_map@100
510
+ value: 0.5676617874135179
511
+ name: Cosine Map@100
512
+ - task:
513
+ type: information-retrieval
514
+ name: Information Retrieval
515
+ dataset:
516
+ name: NanoNFCorpus
517
+ type: NanoNFCorpus
518
+ metrics:
519
+ - type: cosine_accuracy@1
520
+ value: 0.38
521
+ name: Cosine Accuracy@1
522
+ - type: cosine_accuracy@3
523
+ value: 0.52
524
+ name: Cosine Accuracy@3
525
+ - type: cosine_accuracy@5
526
+ value: 0.62
527
+ name: Cosine Accuracy@5
528
+ - type: cosine_accuracy@10
529
+ value: 0.72
530
+ name: Cosine Accuracy@10
531
+ - type: cosine_precision@1
532
+ value: 0.38
533
+ name: Cosine Precision@1
534
+ - type: cosine_precision@3
535
+ value: 0.3466666666666666
536
+ name: Cosine Precision@3
537
+ - type: cosine_precision@5
538
+ value: 0.336
539
+ name: Cosine Precision@5
540
+ - type: cosine_precision@10
541
+ value: 0.28800000000000003
542
+ name: Cosine Precision@10
543
+ - type: cosine_recall@1
544
+ value: 0.02361993221285506
545
+ name: Cosine Recall@1
546
+ - type: cosine_recall@3
547
+ value: 0.0550874334390053
548
+ name: Cosine Recall@3
549
+ - type: cosine_recall@5
550
+ value: 0.0994974298703705
551
+ name: Cosine Recall@5
552
+ - type: cosine_recall@10
553
+ value: 0.16071869880999604
554
+ name: Cosine Recall@10
555
+ - type: cosine_ndcg@10
556
+ value: 0.3357604511492006
557
+ name: Cosine Ndcg@10
558
+ - type: cosine_mrr@10
559
+ value: 0.4767698412698412
560
+ name: Cosine Mrr@10
561
+ - type: cosine_map@100
562
+ value: 0.14291785177024427
563
+ name: Cosine Map@100
564
+ - task:
565
+ type: information-retrieval
566
+ name: Information Retrieval
567
+ dataset:
568
+ name: NanoNQ
569
+ type: NanoNQ
570
+ metrics:
571
+ - type: cosine_accuracy@1
572
+ value: 0.56
573
+ name: Cosine Accuracy@1
574
+ - type: cosine_accuracy@3
575
+ value: 0.7
576
+ name: Cosine Accuracy@3
577
+ - type: cosine_accuracy@5
578
+ value: 0.78
579
+ name: Cosine Accuracy@5
580
+ - type: cosine_accuracy@10
581
+ value: 0.84
582
+ name: Cosine Accuracy@10
583
+ - type: cosine_precision@1
584
+ value: 0.56
585
+ name: Cosine Precision@1
586
+ - type: cosine_precision@3
587
+ value: 0.24
588
+ name: Cosine Precision@3
589
+ - type: cosine_precision@5
590
+ value: 0.16
591
+ name: Cosine Precision@5
592
+ - type: cosine_precision@10
593
+ value: 0.09
594
+ name: Cosine Precision@10
595
+ - type: cosine_recall@1
596
+ value: 0.53
597
+ name: Cosine Recall@1
598
+ - type: cosine_recall@3
599
+ value: 0.66
600
+ name: Cosine Recall@3
601
+ - type: cosine_recall@5
602
+ value: 0.73
603
+ name: Cosine Recall@5
604
+ - type: cosine_recall@10
605
+ value: 0.8
606
+ name: Cosine Recall@10
607
+ - type: cosine_ndcg@10
608
+ value: 0.6782676967719391
609
+ name: Cosine Ndcg@10
610
+ - type: cosine_mrr@10
611
+ value: 0.6575238095238095
612
+ name: Cosine Mrr@10
613
+ - type: cosine_map@100
614
+ value: 0.6356999690257737
615
+ name: Cosine Map@100
616
+ - task:
617
+ type: information-retrieval
618
+ name: Information Retrieval
619
+ dataset:
620
+ name: NanoQuoraRetrieval
621
+ type: NanoQuoraRetrieval
622
+ metrics:
623
+ - type: cosine_accuracy@1
624
+ value: 0.92
625
+ name: Cosine Accuracy@1
626
+ - type: cosine_accuracy@3
627
+ value: 0.98
628
+ name: Cosine Accuracy@3
629
+ - type: cosine_accuracy@5
630
+ value: 0.98
631
+ name: Cosine Accuracy@5
632
+ - type: cosine_accuracy@10
633
+ value: 1.0
634
+ name: Cosine Accuracy@10
635
+ - type: cosine_precision@1
636
+ value: 0.92
637
+ name: Cosine Precision@1
638
+ - type: cosine_precision@3
639
+ value: 0.3999999999999999
640
+ name: Cosine Precision@3
641
+ - type: cosine_precision@5
642
+ value: 0.24799999999999997
643
+ name: Cosine Precision@5
644
+ - type: cosine_precision@10
645
+ value: 0.13399999999999998
646
+ name: Cosine Precision@10
647
+ - type: cosine_recall@1
648
+ value: 0.8073333333333332
649
+ name: Cosine Recall@1
650
+ - type: cosine_recall@3
651
+ value: 0.9386666666666668
652
+ name: Cosine Recall@3
653
+ - type: cosine_recall@5
654
+ value: 0.9493333333333334
655
+ name: Cosine Recall@5
656
+ - type: cosine_recall@10
657
+ value: 0.9893333333333334
658
+ name: Cosine Recall@10
659
+ - type: cosine_ndcg@10
660
+ value: 0.9492763453627913
661
+ name: Cosine Ndcg@10
662
+ - type: cosine_mrr@10
663
+ value: 0.9533333333333333
664
+ name: Cosine Mrr@10
665
+ - type: cosine_map@100
666
+ value: 0.9267995189995191
667
+ name: Cosine Map@100
668
+ - task:
669
+ type: information-retrieval
670
+ name: Information Retrieval
671
+ dataset:
672
+ name: NanoSCIDOCS
673
+ type: NanoSCIDOCS
674
+ metrics:
675
+ - type: cosine_accuracy@1
676
+ value: 0.36
677
+ name: Cosine Accuracy@1
678
+ - type: cosine_accuracy@3
679
+ value: 0.6
680
+ name: Cosine Accuracy@3
681
+ - type: cosine_accuracy@5
682
+ value: 0.68
683
+ name: Cosine Accuracy@5
684
+ - type: cosine_accuracy@10
685
+ value: 0.8
686
+ name: Cosine Accuracy@10
687
+ - type: cosine_precision@1
688
+ value: 0.36
689
+ name: Cosine Precision@1
690
+ - type: cosine_precision@3
691
+ value: 0.3
692
+ name: Cosine Precision@3
693
+ - type: cosine_precision@5
694
+ value: 0.236
695
+ name: Cosine Precision@5
696
+ - type: cosine_precision@10
697
+ value: 0.16
698
+ name: Cosine Precision@10
699
+ - type: cosine_recall@1
700
+ value: 0.07566666666666669
701
+ name: Cosine Recall@1
702
+ - type: cosine_recall@3
703
+ value: 0.18566666666666662
704
+ name: Cosine Recall@3
705
+ - type: cosine_recall@5
706
+ value: 0.24366666666666664
707
+ name: Cosine Recall@5
708
+ - type: cosine_recall@10
709
+ value: 0.3306666666666666
710
+ name: Cosine Recall@10
711
+ - type: cosine_ndcg@10
712
+ value: 0.3193043286623152
713
+ name: Cosine Ndcg@10
714
+ - type: cosine_mrr@10
715
+ value: 0.4949999999999999
716
+ name: Cosine Mrr@10
717
+ - type: cosine_map@100
718
+ value: 0.25689128981497417
719
+ name: Cosine Map@100
720
+ - task:
721
+ type: information-retrieval
722
+ name: Information Retrieval
723
+ dataset:
724
+ name: NanoArguAna
725
+ type: NanoArguAna
726
+ metrics:
727
+ - type: cosine_accuracy@1
728
+ value: 0.16
729
+ name: Cosine Accuracy@1
730
+ - type: cosine_accuracy@3
731
+ value: 0.64
732
+ name: Cosine Accuracy@3
733
+ - type: cosine_accuracy@5
734
+ value: 0.7
735
+ name: Cosine Accuracy@5
736
+ - type: cosine_accuracy@10
737
+ value: 0.86
738
+ name: Cosine Accuracy@10
739
+ - type: cosine_precision@1
740
+ value: 0.16
741
+ name: Cosine Precision@1
742
+ - type: cosine_precision@3
743
+ value: 0.21333333333333335
744
+ name: Cosine Precision@3
745
+ - type: cosine_precision@5
746
+ value: 0.14
747
+ name: Cosine Precision@5
748
+ - type: cosine_precision@10
749
+ value: 0.08599999999999998
750
+ name: Cosine Precision@10
751
+ - type: cosine_recall@1
752
+ value: 0.16
753
+ name: Cosine Recall@1
754
+ - type: cosine_recall@3
755
+ value: 0.64
756
+ name: Cosine Recall@3
757
+ - type: cosine_recall@5
758
+ value: 0.7
759
+ name: Cosine Recall@5
760
+ - type: cosine_recall@10
761
+ value: 0.86
762
+ name: Cosine Recall@10
763
+ - type: cosine_ndcg@10
764
+ value: 0.5219432486598236
765
+ name: Cosine Ndcg@10
766
+ - type: cosine_mrr@10
767
+ value: 0.41343650793650794
768
+ name: Cosine Mrr@10
769
+ - type: cosine_map@100
770
+ value: 0.4212263014763014
771
+ name: Cosine Map@100
772
+ - task:
773
+ type: information-retrieval
774
+ name: Information Retrieval
775
+ dataset:
776
+ name: NanoSciFact
777
+ type: NanoSciFact
778
+ metrics:
779
+ - type: cosine_accuracy@1
780
+ value: 0.62
781
+ name: Cosine Accuracy@1
782
+ - type: cosine_accuracy@3
783
+ value: 0.74
784
+ name: Cosine Accuracy@3
785
+ - type: cosine_accuracy@5
786
+ value: 0.78
787
+ name: Cosine Accuracy@5
788
+ - type: cosine_accuracy@10
789
+ value: 0.84
790
+ name: Cosine Accuracy@10
791
+ - type: cosine_precision@1
792
+ value: 0.62
793
+ name: Cosine Precision@1
794
+ - type: cosine_precision@3
795
+ value: 0.26
796
+ name: Cosine Precision@3
797
+ - type: cosine_precision@5
798
+ value: 0.17599999999999993
799
+ name: Cosine Precision@5
800
+ - type: cosine_precision@10
801
+ value: 0.09399999999999999
802
+ name: Cosine Precision@10
803
+ - type: cosine_recall@1
804
+ value: 0.585
805
+ name: Cosine Recall@1
806
+ - type: cosine_recall@3
807
+ value: 0.72
808
+ name: Cosine Recall@3
809
+ - type: cosine_recall@5
810
+ value: 0.77
811
+ name: Cosine Recall@5
812
+ - type: cosine_recall@10
813
+ value: 0.83
814
+ name: Cosine Recall@10
815
+ - type: cosine_ndcg@10
816
+ value: 0.7216537280780712
817
+ name: Cosine Ndcg@10
818
+ - type: cosine_mrr@10
819
+ value: 0.6944444444444444
820
+ name: Cosine Mrr@10
821
+ - type: cosine_map@100
822
+ value: 0.6870715058750773
823
+ name: Cosine Map@100
824
+ - task:
825
+ type: information-retrieval
826
+ name: Information Retrieval
827
+ dataset:
828
+ name: NanoTouche2020
829
+ type: NanoTouche2020
830
+ metrics:
831
+ - type: cosine_accuracy@1
832
+ value: 0.6122448979591837
833
+ name: Cosine Accuracy@1
834
+ - type: cosine_accuracy@3
835
+ value: 0.8367346938775511
836
+ name: Cosine Accuracy@3
837
+ - type: cosine_accuracy@5
838
+ value: 0.9183673469387755
839
+ name: Cosine Accuracy@5
840
+ - type: cosine_accuracy@10
841
+ value: 0.9795918367346939
842
+ name: Cosine Accuracy@10
843
+ - type: cosine_precision@1
844
+ value: 0.6122448979591837
845
+ name: Cosine Precision@1
846
+ - type: cosine_precision@3
847
+ value: 0.5782312925170068
848
+ name: Cosine Precision@3
849
+ - type: cosine_precision@5
850
+ value: 0.5387755102040817
851
+ name: Cosine Precision@5
852
+ - type: cosine_precision@10
853
+ value: 0.47959183673469385
854
+ name: Cosine Precision@10
855
+ - type: cosine_recall@1
856
+ value: 0.04350189813068328
857
+ name: Cosine Recall@1
858
+ - type: cosine_recall@3
859
+ value: 0.12421061419636209
860
+ name: Cosine Recall@3
861
+ - type: cosine_recall@5
862
+ value: 0.18611320264650894
863
+ name: Cosine Recall@5
864
+ - type: cosine_recall@10
865
+ value: 0.3138456220969939
866
+ name: Cosine Recall@10
867
+ - type: cosine_ndcg@10
868
+ value: 0.5284984258757148
869
+ name: Cosine Ndcg@10
870
+ - type: cosine_mrr@10
871
+ value: 0.7362244897959183
872
+ name: Cosine Mrr@10
873
+ - type: cosine_map@100
874
+ value: 0.3950978450315826
875
+ name: Cosine Map@100
876
+ - task:
877
+ type: nano-beir
878
+ name: Nano BEIR
879
+ dataset:
880
+ name: NanoBEIR mean
881
+ type: NanoBEIR_mean
882
+ metrics:
883
+ - type: cosine_accuracy@1
884
+ value: 0.5209419152276296
885
+ name: Cosine Accuracy@1
886
+ - type: cosine_accuracy@3
887
+ value: 0.70436420722135
888
+ name: Cosine Accuracy@3
889
+ - type: cosine_accuracy@5
890
+ value: 0.7629513343799057
891
+ name: Cosine Accuracy@5
892
+ - type: cosine_accuracy@10
893
+ value: 0.8491993720565149
894
+ name: Cosine Accuracy@10
895
+ - type: cosine_precision@1
896
+ value: 0.5209419152276296
897
+ name: Cosine Precision@1
898
+ - type: cosine_precision@3
899
+ value: 0.3239665096807954
900
+ name: Cosine Precision@3
901
+ - type: cosine_precision@5
902
+ value: 0.24759811616954475
903
+ name: Cosine Precision@5
904
+ - type: cosine_precision@10
905
+ value: 0.17781475667189953
906
+ name: Cosine Precision@10
907
+ - type: cosine_recall@1
908
+ value: 0.3084689029873893
909
+ name: Cosine Recall@1
910
+ - type: cosine_recall@3
911
+ value: 0.4652560339002954
912
+ name: Cosine Recall@3
913
+ - type: cosine_recall@5
914
+ value: 0.520670579098877
915
+ name: Cosine Recall@5
916
+ - type: cosine_recall@10
917
+ value: 0.6099916676008785
918
+ name: Cosine Recall@10
919
+ - type: cosine_ndcg@10
920
+ value: 0.5680039395934455
921
+ name: Cosine Ndcg@10
922
+ - type: cosine_mrr@10
923
+ value: 0.6284543868829584
924
+ name: Cosine Mrr@10
925
+ - type: cosine_map@100
926
+ value: 0.48872276884065613
927
+ name: Cosine Map@100
928
+ ---
929
+
930
+ # SentenceTransformer based on much1na/distil-cohere-v1-ckpt
931
+
932
+ This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [much1na/distil-cohere-v1-ckpt](https://huggingface.co/much1na/distil-cohere-v1-ckpt) on the generator dataset. It maps inputs to a 1024-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, classification, clustering, and more.
933
+
934
+ ## Model Details
935
+
936
+ ### Model Description
937
+ - **Model Type:** Sentence Transformer
938
+ - **Base model:** [much1na/distil-cohere-v1-ckpt](https://huggingface.co/much1na/distil-cohere-v1-ckpt) <!-- at revision 487bc2f59d62211641241d4ae071987194523811 -->
939
+ - **Maximum Sequence Length:** 512 tokens
940
+ - **Output Dimensionality:** 1024 dimensions
941
+ - **Similarity Function:** Cosine Similarity
942
+ - **Supported Modality:** Text
943
+ - **Training Dataset:**
944
+ - generator
945
+ <!-- - **Language:** Unknown -->
946
+ <!-- - **License:** Unknown -->
947
+
948
+ ### Model Sources
949
+
950
+ - **Documentation:** [Sentence Transformers Documentation](https://sbert.net)
951
+ - **Repository:** [Sentence Transformers on GitHub](https://github.com/huggingface/sentence-transformers)
952
+ - **Hugging Face:** [Sentence Transformers on Hugging Face](https://huggingface.co/models?library=sentence-transformers)
953
+
954
+ ### Full Model Architecture
955
+
956
+ ```
957
+ SentenceTransformer(
958
+ (0): Transformer({'transformer_task': 'feature-extraction', 'modality_config': {'text': {'method': 'forward', 'method_output_name': 'last_hidden_state'}}, 'module_output_name': 'token_embeddings', 'architecture': 'BertModel'})
959
+ (1): Pooling({'embedding_dimension': 512, 'pooling_mode': 'mean', 'include_prompt': True})
960
+ (2): Dense({'in_features': 512, 'out_features': 1024, 'bias': False, 'activation_function': 'torch.nn.modules.linear.Identity', 'module_input_name': 'sentence_embedding', 'module_output_name': 'sentence_embedding'})
961
+ (3): Normalize({'module_input_name': 'sentence_embedding', 'module_output_name': 'sentence_embedding'})
962
+ )
963
+ ```
964
+
965
+ ## Usage
966
+
967
+ ### Direct Usage (Sentence Transformers)
968
+
969
+ First install the Sentence Transformers library:
970
+
971
+ ```bash
972
+ pip install -U sentence-transformers
973
+ ```
974
+ Then you can load this model and run inference.
975
+ ```python
976
+ from sentence_transformers import SentenceTransformer
977
+
978
+ # Download from the 🤗 Hub
979
+ model = SentenceTransformer("sentence_transformers_model_id")
980
+ # Run inference
981
+ queries = [
982
+ 'Stackoverflow Stackexchange\n\nQ: What is a good gitignore to use with Codeblocks C++ Projects? This is what I have so far:\nobj/\nbin/\n\nAnything else?\n\nA: You can use this:\n*.layout\n*.depend\nbin/\nobj/\n\nSee:\nhttps://www.gitignore.io/api/codeblocks\n',
983
+ ]
984
+ documents = [
985
+ "Stackoverflow Stackexchange\n\nQ: Listen on a network port and save data to a text file Is there some easy way in linux to listen for data on network and save them to a text file?\nThank you.\n\nA: Also available is TCPDump.\nsudo tcpdump -i eth0 -vvvvtttAXns 1500 'port 10000'\n\n",
986
+ 'Stackoverflow Stackexchange\n\nQ: Get the current date and time I want to get the current date and time.\nFor example:\n2012/11/13 06:30:38\n\nWhat I have tried:\nDim d As System.DateTime\nMsgBox(d.Year) \'Return 1\n\n\nA: use DateTime.Now\ntry this:\nDateTime.Now.ToString("yyyy/MM/dd HH:mm:ss")\n\n',
987
+ ]
988
+ query_embeddings = model.encode_query(queries)
989
+ document_embeddings = model.encode_document(documents)
990
+ print(query_embeddings.shape, document_embeddings.shape)
991
+ # [1, 1024] [2, 1024]
992
+
993
+ # Get the similarity scores for the embeddings
994
+ similarities = model.similarity(query_embeddings, document_embeddings)
995
+ print(similarities)
996
+ # tensor([[0.3207, 0.2160]])
997
+ ```
998
+ <!--
999
+ ### Direct Usage (Transformers)
1000
+
1001
+ <details><summary>Click to see the direct usage in Transformers</summary>
1002
+
1003
+ </details>
1004
+ -->
1005
+
1006
+ <!--
1007
+ ### Downstream Usage (Sentence Transformers)
1008
+
1009
+ You can finetune this model on your own dataset.
1010
+
1011
+ <details><summary>Click to expand</summary>
1012
+
1013
+ </details>
1014
+ -->
1015
+
1016
+ <!--
1017
+ ### Out-of-Scope Use
1018
+
1019
+ *List how the model may foreseeably be misused and address what users ought not to do with the model.*
1020
+ -->
1021
+
1022
+ ## Evaluation
1023
+
1024
+ ### Metrics
1025
+
1026
+ #### Information Retrieval
1027
+
1028
+ * Datasets: `NanoClimateFEVER`, `NanoDBPedia`, `NanoFEVER`, `NanoFiQA2018`, `NanoHotpotQA`, `NanoMSMARCO`, `NanoNFCorpus`, `NanoNQ`, `NanoQuoraRetrieval`, `NanoSCIDOCS`, `NanoArguAna`, `NanoSciFact` and `NanoTouche2020`
1029
+ * Evaluated with [<code>InformationRetrievalEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.sentence_transformer.evaluation.InformationRetrievalEvaluator)
1030
+
1031
+ | Metric | NanoClimateFEVER | NanoDBPedia | NanoFEVER | NanoFiQA2018 | NanoHotpotQA | NanoMSMARCO | NanoNFCorpus | NanoNQ | NanoQuoraRetrieval | NanoSCIDOCS | NanoArguAna | NanoSciFact | NanoTouche2020 |
1032
+ |:--------------------|:-----------------|:------------|:-----------|:-------------|:-------------|:------------|:-------------|:-----------|:-------------------|:------------|:------------|:------------|:---------------|
1033
+ | cosine_accuracy@1 | 0.3 | 0.74 | 0.56 | 0.34 | 0.78 | 0.44 | 0.38 | 0.56 | 0.92 | 0.36 | 0.16 | 0.62 | 0.6122 |
1034
+ | cosine_accuracy@3 | 0.48 | 0.84 | 0.72 | 0.56 | 0.9 | 0.64 | 0.52 | 0.7 | 0.98 | 0.6 | 0.64 | 0.74 | 0.8367 |
1035
+ | cosine_accuracy@5 | 0.52 | 0.88 | 0.82 | 0.62 | 0.92 | 0.7 | 0.62 | 0.78 | 0.98 | 0.68 | 0.7 | 0.78 | 0.9184 |
1036
+ | cosine_accuracy@10 | 0.62 | 0.94 | 0.96 | 0.72 | 1.0 | 0.76 | 0.72 | 0.84 | 1.0 | 0.8 | 0.86 | 0.84 | 0.9796 |
1037
+ | cosine_precision@1 | 0.3 | 0.74 | 0.56 | 0.34 | 0.78 | 0.44 | 0.38 | 0.56 | 0.92 | 0.36 | 0.16 | 0.62 | 0.6122 |
1038
+ | cosine_precision@3 | 0.1867 | 0.5467 | 0.2467 | 0.24 | 0.44 | 0.2133 | 0.3467 | 0.24 | 0.4 | 0.3 | 0.2133 | 0.26 | 0.5782 |
1039
+ | cosine_precision@5 | 0.124 | 0.484 | 0.168 | 0.184 | 0.284 | 0.14 | 0.336 | 0.16 | 0.248 | 0.236 | 0.14 | 0.176 | 0.5388 |
1040
+ | cosine_precision@10 | 0.084 | 0.446 | 0.1 | 0.114 | 0.16 | 0.076 | 0.288 | 0.09 | 0.134 | 0.16 | 0.086 | 0.094 | 0.4796 |
1041
+ | cosine_recall@1 | 0.1533 | 0.0779 | 0.53 | 0.1937 | 0.39 | 0.44 | 0.0236 | 0.53 | 0.8073 | 0.0757 | 0.16 | 0.585 | 0.0435 |
1042
+ | cosine_recall@3 | 0.234 | 0.1471 | 0.6767 | 0.3669 | 0.66 | 0.64 | 0.0551 | 0.66 | 0.9387 | 0.1857 | 0.64 | 0.72 | 0.1242 |
1043
+ | cosine_recall@5 | 0.2523 | 0.1871 | 0.7767 | 0.464 | 0.71 | 0.7 | 0.0995 | 0.73 | 0.9493 | 0.2437 | 0.7 | 0.77 | 0.1861 |
1044
+ | cosine_recall@10 | 0.323 | 0.2929 | 0.9267 | 0.5428 | 0.8 | 0.76 | 0.1607 | 0.8 | 0.9893 | 0.3307 | 0.86 | 0.83 | 0.3138 |
1045
+ | **cosine_ndcg@10** | **0.2919** | **0.5529** | **0.7213** | **0.4234** | **0.7355** | **0.6044** | **0.3358** | **0.6783** | **0.9493** | **0.3193** | **0.5219** | **0.7217** | **0.5285** |
1046
+ | cosine_mrr@10 | 0.3999 | 0.8069 | 0.6736 | 0.4599 | 0.8485 | 0.5544 | 0.4768 | 0.6575 | 0.9533 | 0.495 | 0.4134 | 0.6944 | 0.7362 |
1047
+ | cosine_map@100 | 0.2517 | 0.3997 | 0.652 | 0.3572 | 0.6594 | 0.5677 | 0.1429 | 0.6357 | 0.9268 | 0.2569 | 0.4212 | 0.6871 | 0.3951 |
1048
+
1049
+ #### Nano BEIR
1050
+
1051
+ * Dataset: `NanoBEIR_mean`
1052
+ * Evaluated with [<code>NanoBEIREvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.sentence_transformer.evaluation.NanoBEIREvaluator) with these parameters:
1053
+ ```json
1054
+ {
1055
+ "dataset_names": [
1056
+ "climatefever",
1057
+ "dbpedia",
1058
+ "fever",
1059
+ "fiqa2018",
1060
+ "hotpotqa",
1061
+ "msmarco",
1062
+ "nfcorpus",
1063
+ "nq",
1064
+ "quoraretrieval",
1065
+ "scidocs",
1066
+ "arguana",
1067
+ "scifact",
1068
+ "touche2020"
1069
+ ],
1070
+ "dataset_id": "sentence-transformers/NanoBEIR-en"
1071
+ }
1072
+ ```
1073
+
1074
+ | Metric | Value |
1075
+ |:--------------------|:----------|
1076
+ | cosine_accuracy@1 | 0.5209 |
1077
+ | cosine_accuracy@3 | 0.7044 |
1078
+ | cosine_accuracy@5 | 0.763 |
1079
+ | cosine_accuracy@10 | 0.8492 |
1080
+ | cosine_precision@1 | 0.5209 |
1081
+ | cosine_precision@3 | 0.324 |
1082
+ | cosine_precision@5 | 0.2476 |
1083
+ | cosine_precision@10 | 0.1778 |
1084
+ | cosine_recall@1 | 0.3085 |
1085
+ | cosine_recall@3 | 0.4653 |
1086
+ | cosine_recall@5 | 0.5207 |
1087
+ | cosine_recall@10 | 0.61 |
1088
+ | **cosine_ndcg@10** | **0.568** |
1089
+ | cosine_mrr@10 | 0.6285 |
1090
+ | cosine_map@100 | 0.4887 |
1091
+
1092
+ <!--
1093
+ ## Bias, Risks and Limitations
1094
+
1095
+ *What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model.*
1096
+ -->
1097
+
1098
+ <!--
1099
+ ### Recommendations
1100
+
1101
+ *What are recommendations with respect to the foreseeable issues? For example, filtering explicit content.*
1102
+ -->
1103
+
1104
+ ## Training Details
1105
+
1106
+ ### Training Dataset
1107
+
1108
+ #### generator
1109
+
1110
+ * Dataset: generator
1111
+ * Columns: <code>sentence</code> and <code>label</code>
1112
+ * Loss: [<code>EmbedDistillLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#embeddistillloss) with these parameters:
1113
+ ```json
1114
+ {
1115
+ "distance_metric": "cosine",
1116
+ "projection_dim": null
1117
+ }
1118
+ ```
1119
+
1120
+ ### Evaluation Datasets
1121
+ <details><summary>wiki</summary>
1122
+
1123
+ #### wiki
1124
+
1125
+ * Dataset: wiki
1126
+ * Size: 1,000 evaluation samples
1127
+ * Columns: <code>sentence</code> and <code>label</code>
1128
+ * Approximate statistics based on the first 100 samples:
1129
+ | | sentence | label |
1130
+ |:---------|:--------------------------------------------------------------------------------------|:--------------------------------------|
1131
+ | type | string | list |
1132
+ | modality | text | |
1133
+ | details | <ul><li>min: 103 tokens</li><li>mean: 319.54 tokens</li><li>max: 512 tokens</li></ul> | <ul><li>size: 1024 elements</li></ul> |
1134
+ * Samples:
1135
+ | sentence | label |
1136
+ |:------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:------------------------------------------------------------------------------------------------------------------------|
1137
+ | <code>Panthera spelaea Paleobiology P. spelaea inhabited colder environments such as mammoth steppe and boreal forest. It was one of the keystone species of the mammoth steppe, being one of the main apex predators alongside the gray wolf, cave hyena and brown bear. Large amounts of bones belonging to P. spelaea were excavated in caves, where bones of cave hyena, cave bear and Paleolithic artefacts were also found. Some of these accumulations of cave lion bones have been attributed to hoarding of meat from cave lion carcasses by cave hyenas in caves occupied by the latter. It is unclear whether P. spelaea was social like the modern lion; some evidence indicates that it may have been solitary.<br>Isotopic analyses of bone collagen samples extracted from fossils indicate that cave bear cubs, reindeer and other cervids (deer) were prominent in the diet of cave lions. Later cave lions seem to have preyed foremost on reindeer, up to the brink of local extinction or extirpation of both species. Other ...</code> | <code>[-0.008819580078125, -0.02508544921875, 0.012176513671875, 0.07574462890625, -0.0217742919921875, ...]</code> |
1138
+ | <code>Elizabeth David As a teenager David enjoyed painting, and her mother thought her talent worth developing. In 1930 she was sent to Paris, where she studied painting privately and enrolled at the Sorbonne for a course in French civilisation which covered history, literature and architecture. She found her Sorbonne studies arduous and in many ways uninspiring, but they left her with a love of French literature and a fluency in the language that remained with her throughout her life. She lodged with a Parisian family, whose fanatical devotion to the pleasures of the table she portrayed to comic effect in her French Provincial Cooking (1960). Nevertheless, she acknowledged in retrospect that the experience had been the most valuable part of her time in Paris: "I realized in what way the family had fulfilled their task of instilling French culture into at least one of their British charges. Forgotten were the Sorbonne professors. ... What had stuck was the taste for a kind of food quite idea...</code> | <code>[0.0191650390625, -0.01065826416015625, 0.034210205078125, 0.039459228515625, -0.0008692741394042969, ...]</code> |
1139
+ | <code>Guinea-Bissau Wars were increasingly waged for the sole purpose of capturing slaves to sell to the Europeans in exchange for imported goods. They resembled man-hunts more than conflicts over territory or political power. The nobles and kings benefited, while the common people bore the brunt of the raiding and insecurity. If a noble was captured, they were likely to be released, as the captors, whoever they were, would generally accept a ransom in exchange for freeing them. The relationship between kings and European traders was a partnership, with the two regularly making deals on how the trade was to be conducted, defining who could be enslaved and who could not, and the prices of the slaves. Contemporary chroniclers questioned multiple kings on their part in the slave trade, and noted that they recognised the trade as evil but participated because otherwise the Europeans would not buy a ny other goods from them.<br>Beginning in the late 18th century, European countries gradually began s...</code> | <code>[0.0003838539123535156, -0.027008056640625, 0.032684326171875, 0.036956787109375, -0.039581298828125, ...]</code> |
1140
+ * Loss: [<code>EmbedDistillLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#embeddistillloss) with these parameters:
1141
+ ```json
1142
+ {
1143
+ "distance_metric": "cosine",
1144
+ "projection_dim": null
1145
+ }
1146
+ ```
1147
+ </details>
1148
+ <details><summary>stackexchange</summary>
1149
+
1150
+ #### stackexchange
1151
+
1152
+ * Dataset: stackexchange
1153
+ * Size: 1,000 evaluation samples
1154
+ * Columns: <code>sentence</code> and <code>label</code>
1155
+ * Approximate statistics based on the first 100 samples:
1156
+ | | sentence | label |
1157
+ |:---------|:-------------------------------------------------------------------------------------|:--------------------------------------|
1158
+ | type | string | list |
1159
+ | modality | text | |
1160
+ | details | <ul><li>min: 72 tokens</li><li>mean: 178.89 tokens</li><li>max: 443 tokens</li></ul> | <ul><li>size: 1024 elements</li></ul> |
1161
+ * Samples:
1162
+ | sentence | label |
1163
+ |:-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:----------------------------------------------------------------------------------------------------------------------|
1164
+ | <code>Apple Stackexchange<br><br>Q: Why does my MacBook Pro take so long to go to sleep? My MacBook Pro takes an awful long time to go to sleep when I close the lid. Considering that I generally want to close the lid and then carry it somewhere, I want it to be fully asleep so I don't have any hard drive issues. Is there anything I can do to speed up the process?<br><br>A: I recommend a nifty little utility called SmartSleep that allows you to control when the saving of your RAM contents to disk for the Safe Sleep / Hibernate feature Mac OS X has.<br>Alternatively you can disable Safe Sleep altogether by running: sudo pmset -a hibernatemode 0. To change it back to the defaults, change the value to 3.<br>0 - Old style - just goes to sleep.<br>1 - Only Hibernate<br>3 - Default - goes to sleep but writes RAM contents to disk just in case.<br>5 - Only Hibernate mode but if you use secure virtual memory.<br>7 - The Default but if you use secure virtual memory.<br><br></code> | <code>[-0.06976318359375, -0.0247039794921875, -0.0095367431640625, 0.09149169921875, -0.053619384765625, ...]</code> |
1165
+ | <code>Apple Stackexchange<br><br>Q: iPhone backlight blinks on then goes out. What are my options? This morning I woke up to find my iPhone dark. If I pressed the home button to turn on the screen, the backlight would flash on briefly and then die. I've turned it on and off, done a hard reboot (home+lock), and I'm in the process of trying a restore.<br>Has anyone ever seen a similar issue?<br>The phone was purchased last June with a contract from ATT from an Apple store. If I take the phone in, will they fix it or replace it?<br>What should I do next?<br><br>A: You should take your phone in. They will fix it if they can and it's cheaper for them than giving you a new.<br>If you try to fix it yourself they can put the blame on you, so don't touch the hardware. <br></code> | <code>[-0.037384033203125, 0.020355224609375, 0.0137786865234375, 0.056610107421875, -0.0280609130859375, ...]</code> |
1166
+ | <code>Apple Stackexchange<br><br>Q: Can I use tethering to let an iPad use an iPhone's 3G? Given an iPad that doesn't have 3G but only wifi, can I make this use the 3G connection of an iPhone 4? <br>Does this require a cable between the devices, or does it work over (bluetooth\|wifi?) ?<br><br>A: I don't believe you can do this with the stock phone, however if you decide to jailbreakyour iPhone, you could use an application like MyWi to share your iPhone's 3G connection over WiFi<br></code> | <code>[-0.01351165771484375, -0.010284423828125, -0.0264892578125, 0.046356201171875, -0.03228759765625, ...]</code> |
1167
+ * Loss: [<code>EmbedDistillLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#embeddistillloss) with these parameters:
1168
+ ```json
1169
+ {
1170
+ "distance_metric": "cosine",
1171
+ "projection_dim": null
1172
+ }
1173
+ ```
1174
+ </details>
1175
+ <details><summary>msmarco_docs</summary>
1176
+
1177
+ #### msmarco_docs
1178
+
1179
+ * Dataset: msmarco_docs
1180
+ * Size: 1,000 evaluation samples
1181
+ * Columns: <code>sentence</code> and <code>label</code>
1182
+ * Approximate statistics based on the first 100 samples:
1183
+ | | sentence | label |
1184
+ |:---------|:------------------------------------------------------------------------------------|:--------------------------------------|
1185
+ | type | string | list |
1186
+ | modality | text | |
1187
+ | details | <ul><li>min: 23 tokens</li><li>mean: 77.07 tokens</li><li>max: 227 tokens</li></ul> | <ul><li>size: 1024 elements</li></ul> |
1188
+ * Samples:
1189
+ | sentence | label |
1190
+ |:--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:---------------------------------------------------------------------------------------------------------------------------|
1191
+ | <code>There are 25 fountains. The Stanford campus also encompasses the 70-acre, 140-retail-store Stanford Shopping Center, built in 1955, and the 700-acre, 150-company Stanford Research Park, created in 1951.</code> | <code>[-0.0478515625, 0.0002257823944091797, -0.08514404296875, -0.0233001708984375, -0.0210418701171875, ...]</code> |
1192
+ | <code>First, take the probiotic on an empty stomach. When the stomach is empty, it is not making much acid. You can open a capsule and mix it with a small amount of water. The stomach will empty out liquid in 15-20 minutes, but will keep food in it for 60-90 minutes.</code> | <code>[-0.033294677734375, -0.005718231201171875, -0.0265960693359375, 0.0204925537109375, -0.025390625, ...]</code> |
1193
+ | <code>Most of the magnesium in the body is found in the bones and inside the cells. Only a tiny amount of magnesium is normally present in the blood. Tests for other electrolytes, such as calcium, potassium, sodium, and phosphorus, may be done along with a test for magnesium.</code> | <code>[0.00592041015625, -0.0158843994140625, -0.0017023086547851562, -0.004730224609375, -0.0122222900390625, ...]</code> |
1194
+ * Loss: [<code>EmbedDistillLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#embeddistillloss) with these parameters:
1195
+ ```json
1196
+ {
1197
+ "distance_metric": "cosine",
1198
+ "projection_dim": null
1199
+ }
1200
+ ```
1201
+ </details>
1202
+ <details><summary>msmarco_queries</summary>
1203
+
1204
+ #### msmarco_queries
1205
+
1206
+ * Dataset: msmarco_queries
1207
+ * Size: 1,000 evaluation samples
1208
+ * Columns: <code>sentence</code> and <code>label</code>
1209
+ * Approximate statistics based on the first 100 samples:
1210
+ | | sentence | label |
1211
+ |:---------|:---------------------------------------------------------------------------------|:--------------------------------------|
1212
+ | type | string | list |
1213
+ | modality | text | |
1214
+ | details | <ul><li>min: 4 tokens</li><li>mean: 8.91 tokens</li><li>max: 21 tokens</li></ul> | <ul><li>size: 1024 elements</li></ul> |
1215
+ * Samples:
1216
+ | sentence | label |
1217
+ |:-------------------------------------------------------|:-----------------------------------------------------------------------------------------------------------------------|
1218
+ | <code>how much per month is a lambo</code> | <code>[-0.0255889892578125, 0.018035888671875, -0.047149658203125, -0.02044677734375, 0.01209259033203125, ...]</code> |
1219
+ | <code>what is basic price of lincoln mkz model?</code> | <code>[-0.05865478515625, -0.0253448486328125, -0.01363372802734375, 0.037811279296875, -0.0633544921875, ...]</code> |
1220
+ | <code>wendco of new hampshire, llc</code> | <code>[-0.01483917236328125, -0.047821044921875, -0.0298309326171875, -0.0416259765625, -0.04754638671875, ...]</code> |
1221
+ * Loss: [<code>EmbedDistillLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#embeddistillloss) with these parameters:
1222
+ ```json
1223
+ {
1224
+ "distance_metric": "cosine",
1225
+ "projection_dim": null
1226
+ }
1227
+ ```
1228
+ </details>
1229
+ <details><summary>sphere</summary>
1230
+
1231
+ #### sphere
1232
+
1233
+ * Dataset: sphere
1234
+ * Size: 1,000 evaluation samples
1235
+ * Columns: <code>sentence</code> and <code>label</code>
1236
+ * Approximate statistics based on the first 100 samples:
1237
+ | | sentence | label |
1238
+ |:---------|:-------------------------------------------------------------------------------------|:--------------------------------------|
1239
+ | type | string | list |
1240
+ | modality | text | |
1241
+ | details | <ul><li>min: 44 tokens</li><li>mean: 140.06 tokens</li><li>max: 321 tokens</li></ul> | <ul><li>size: 1024 elements</li></ul> |
1242
+ * Samples:
1243
+ | sentence | label |
1244
+ |:---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:---------------------------------------------------------------------------------------------------------------------------|
1245
+ | <code>Ed Schultz \| Project Gutenberg Self-Publishing - eBooks \| Read eBooks online political commentator and a former sports broadcaster. He was the host of The Ed Show, a weekday news talk program on MSNBC, and The Ed Schultz Show, a talk radio show, nationally syndicated by Dial Global.[2] The radio show ended on May 23, 2014, and was replaced by a one-hour podcast.[3] Broadcasting career 2 Sportscasting 2.1 Talk radio 2.2 Television show 2.3 Ed Shultz News and Commentary 3 Schultz was born in American players of Canadian football American talk radio hosts Canadian football quarterbacks Minnesota State University Moorhead alumni North Dakota Democrats People from Detroit Lakes, Minnesota People from Norfolk,</code> | <code>[0.030853271484375, -0.039520263671875, 0.0535888671875, -0.0031871795654296875, -0.0404052734375, ...]</code> |
1246
+ | <code>Tattoo Criminals — Elie Wiesel Should Be Investigated for Holocaust Fraud \| National Vanguard Jew, and Jews are never investigated for anything in Germany, let alone found guilty of something. Have you ever heard of it? I did an internet search and came up with nothing. State prosecutors had demanded that the accused [Zech] be sentenced to 10 months in jail without parole because, according to DW, he seemed to be “a politically motivated criminal, who had trampled on the core values of the German constitution.” What are the “core values” of the German constitution? Consulting the German Basic Law (constitutional law), I come up with a few: Art. 3: No person shall be</code> | <code>[-0.005931854248046875, -0.01629638671875, 0.036712646484375, 0.07537841796875, -0.033294677734375, ...]</code> |
1247
+ | <code>J.D. v. Atlanta Public Schools: A Lesser Spirit Would Have Been Crushed Long Ago" by Pamela Wright & Peter Wright the first time since 1998 . . . The school psychologist reported that J.D. was performing at the 2nd or 3rd grade level despite being in 9th grade . . . significant variations of his test scores showed that the I.Q. score may not be an accurate reflection of J.D.'s true ability." "A different school psychologist performed an additional psychological evaluation on J.D.. . . The July 2003 evaluation confirmed that J.D. was not MID but had a specific learning disability. The tests showed that J.D. had a full scale I.Q. of 82, which was in the low average range</code> | <code>[-0.0009756088256835938, -0.0179443359375, 0.0031375885009765625, 0.041839599609375, -0.034820556640625, ...]</code> |
1248
+ * Loss: [<code>EmbedDistillLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#embeddistillloss) with these parameters:
1249
+ ```json
1250
+ {
1251
+ "distance_metric": "cosine",
1252
+ "projection_dim": null
1253
+ }
1254
+ ```
1255
+ </details>
1256
+
1257
+ ### Training Hyperparameters
1258
+ #### Non-Default Hyperparameters
1259
+
1260
+ - `per_device_train_batch_size`: 64
1261
+ - `max_steps`: 452659
1262
+ - `learning_rate`: 3e-05
1263
+ - `lr_scheduler_type`: cosine_with_min_lr
1264
+ - `lr_scheduler_kwargs`: {'min_lr_rate': 0.1}
1265
+ - `warmup_steps`: 500
1266
+ - `gradient_accumulation_steps`: 2
1267
+ - `fp16`: True
1268
+ - `per_device_eval_batch_size`: 32
1269
+ - `ignore_data_skip`: True
1270
+ - `accelerator_config`: {'split_batches': False, 'dispatch_batches': False, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
1271
+ - `ddp_find_unused_parameters`: True
1272
+
1273
+ #### All Hyperparameters
1274
+ <details><summary>Click to expand</summary>
1275
+
1276
+ - `per_device_train_batch_size`: 64
1277
+ - `num_train_epochs`: 3.0
1278
+ - `max_steps`: 452659
1279
+ - `learning_rate`: 3e-05
1280
+ - `lr_scheduler_type`: cosine_with_min_lr
1281
+ - `lr_scheduler_kwargs`: {'min_lr_rate': 0.1}
1282
+ - `warmup_steps`: 500
1283
+ - `optim`: adamw_torch_fused
1284
+ - `optim_args`: None
1285
+ - `weight_decay`: 0.0
1286
+ - `adam_beta1`: 0.9
1287
+ - `adam_beta2`: 0.999
1288
+ - `adam_epsilon`: 1e-08
1289
+ - `optim_target_modules`: None
1290
+ - `gradient_accumulation_steps`: 2
1291
+ - `average_tokens_across_devices`: True
1292
+ - `max_grad_norm`: 1.0
1293
+ - `label_smoothing_factor`: 0.0
1294
+ - `bf16`: False
1295
+ - `fp16`: True
1296
+ - `bf16_full_eval`: False
1297
+ - `fp16_full_eval`: False
1298
+ - `tf32`: None
1299
+ - `gradient_checkpointing`: False
1300
+ - `gradient_checkpointing_kwargs`: None
1301
+ - `torch_compile`: False
1302
+ - `torch_compile_backend`: None
1303
+ - `torch_compile_mode`: None
1304
+ - `use_liger_kernel`: False
1305
+ - `liger_kernel_config`: None
1306
+ - `use_cache`: False
1307
+ - `neftune_noise_alpha`: None
1308
+ - `torch_empty_cache_steps`: None
1309
+ - `auto_find_batch_size`: False
1310
+ - `log_on_each_node`: True
1311
+ - `logging_nan_inf_filter`: True
1312
+ - `include_num_input_tokens_seen`: no
1313
+ - `log_level`: passive
1314
+ - `log_level_replica`: warning
1315
+ - `disable_tqdm`: False
1316
+ - `project`: huggingface
1317
+ - `trackio_space_id`: None
1318
+ - `trackio_bucket_id`: None
1319
+ - `trackio_static_space_id`: None
1320
+ - `per_device_eval_batch_size`: 32
1321
+ - `prediction_loss_only`: True
1322
+ - `eval_on_start`: False
1323
+ - `eval_do_concat_batches`: True
1324
+ - `eval_use_gather_object`: False
1325
+ - `eval_accumulation_steps`: None
1326
+ - `include_for_metrics`: []
1327
+ - `batch_eval_metrics`: False
1328
+ - `save_only_model`: False
1329
+ - `save_on_each_node`: False
1330
+ - `enable_jit_checkpoint`: False
1331
+ - `push_to_hub`: False
1332
+ - `hub_private_repo`: None
1333
+ - `hub_model_id`: None
1334
+ - `hub_strategy`: every_save
1335
+ - `hub_always_push`: False
1336
+ - `hub_revision`: None
1337
+ - `load_best_model_at_end`: False
1338
+ - `ignore_data_skip`: True
1339
+ - `restore_callback_states_from_checkpoint`: False
1340
+ - `full_determinism`: False
1341
+ - `seed`: 42
1342
+ - `data_seed`: None
1343
+ - `use_cpu`: False
1344
+ - `accelerator_config`: {'split_batches': False, 'dispatch_batches': False, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
1345
+ - `parallelism_config`: None
1346
+ - `dataloader_drop_last`: False
1347
+ - `dataloader_num_workers`: 0
1348
+ - `dataloader_pin_memory`: True
1349
+ - `dataloader_persistent_workers`: False
1350
+ - `dataloader_prefetch_factor`: None
1351
+ - `dataloader_multiprocessing_context`: None
1352
+ - `dataloader_in_order`: True
1353
+ - `remove_unused_columns`: True
1354
+ - `label_names`: None
1355
+ - `train_sampling_strategy`: random
1356
+ - `length_column_name`: length
1357
+ - `ddp_find_unused_parameters`: True
1358
+ - `ddp_bucket_cap_mb`: None
1359
+ - `ddp_broadcast_buffers`: False
1360
+ - `ddp_static_graph`: None
1361
+ - `ddp_backend`: None
1362
+ - `ddp_timeout`: 1800
1363
+ - `fsdp`: None
1364
+ - `fsdp_config`: None
1365
+ - `deepspeed`: None
1366
+ - `debug`: []
1367
+ - `skip_memory_metrics`: True
1368
+ - `do_predict`: False
1369
+ - `resume_from_checkpoint`: None
1370
+ - `local_rank`: -1
1371
+ - `prompts`: None
1372
+ - `batch_sampler`: batch_sampler
1373
+ - `multi_dataset_batch_sampler`: proportional
1374
+ - `router_mapping`: {}
1375
+ - `learning_rate_mapping`: {}
1376
+ - `warmup_ratio`: None
1377
+
1378
+ </details>
1379
+
1380
+ ### Training Logs
1381
+
1382
+
1383
+ ### Training Time
1384
+ - **Training**: 2.7 hours
1385
+
1386
+ ### Framework Versions
1387
+ - Python: 3.13.15
1388
+ - Sentence Transformers: 6.2.0
1389
+ - Transformers: 5.20.1
1390
+ - PyTorch: 2.11.0+cu130
1391
+ - Accelerate: 1.15.0
1392
+ - Datasets: 4.8.5
1393
+ - Tokenizers: 0.23.3
1394
+
1395
+ ## Additional Resources
1396
+
1397
+ - [Training and Finetuning Embedding Models with Sentence Transformers](https://huggingface.co/blog/train-sentence-transformers): the end-to-end guide for training or finetuning Sentence Transformer models.
1398
+ - [Introduction to Matryoshka Embedding Models](https://huggingface.co/blog/matryoshka): variable-size embeddings that can be truncated with minimal quality loss.
1399
+ - [Binary and Scalar Embedding Quantization for Significantly Faster & Cheaper Retrieval](https://huggingface.co/blog/embedding-quantization): post-training compression of embedding vectors.
1400
+ - [Multimodal Embedding & Reranker Models with Sentence Transformers](https://huggingface.co/blog/multimodal-sentence-transformers): use text, image, audio, and video models through the same API.
1401
+ - [Training and Finetuning Multimodal Embedding & Reranker Models with Sentence Transformers](https://huggingface.co/blog/train-multimodal-sentence-transformers): train multimodal embedding models, with a Visual Document Retrieval walkthrough.
1402
+
1403
+ ## Citation
1404
+
1405
+ ### BibTeX
1406
+
1407
+ #### Sentence Transformers
1408
+ ```bibtex
1409
+ @inproceedings{reimers-2019-sentence-bert,
1410
+ title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks",
1411
+ author = "Reimers, Nils and Gurevych, Iryna",
1412
+ booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing",
1413
+ month = "11",
1414
+ year = "2019",
1415
+ publisher = "Association for Computational Linguistics",
1416
+ url = "https://arxiv.org/abs/1908.10084",
1417
+ }
1418
+ ```
1419
+
1420
+ #### EmbedDistillLoss
1421
+ ```bibtex
1422
+ @article{kim2023embeddistill,
1423
+ title={EmbedDistill: A Geometric Knowledge Distillation for Information Retrieval},
1424
+ author={Kim, Seungyeon and Rawat, Ankit Singh and Zaheer, Manzil and Jayasumana, Sadeep and Sadhanala, Veeranjaneyulu and Jitkrittum, Wittawat and Menon, Aditya Krishna and Fergus, Rob and Kumar, Sanjiv},
1425
+ year={2023},
1426
+ eprint={2301.12005},
1427
+ archivePrefix={arXiv},
1428
+ primaryClass={cs.IR}
1429
+ }
1430
+ ```
1431
+
1432
+ <!--
1433
+ ## Glossary
1434
+
1435
+ *Clearly define terms in order to be accessible across audiences.*
1436
+ -->
1437
+
1438
+ <!--
1439
+ ## Model Card Authors
1440
+
1441
+ *Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction.*
1442
+ -->
1443
+
1444
+ <!--
1445
+ ## Model Card Contact
1446
+
1447
+ *Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors.*
1448
+ -->
checkpoint-368000/config.json ADDED
@@ -0,0 +1,28 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_cross_attention": false,
3
+ "architectures": [
4
+ "BertModel"
5
+ ],
6
+ "attention_probs_dropout_prob": 0.1,
7
+ "bos_token_id": null,
8
+ "classifier_dropout": null,
9
+ "dtype": "float32",
10
+ "eos_token_id": null,
11
+ "hidden_act": "gelu",
12
+ "hidden_dropout_prob": 0.1,
13
+ "hidden_size": 512,
14
+ "initializer_range": 0.02,
15
+ "intermediate_size": 2048,
16
+ "is_decoder": false,
17
+ "layer_norm_eps": 1e-12,
18
+ "max_position_embeddings": 512,
19
+ "model_type": "bert",
20
+ "num_attention_heads": 8,
21
+ "num_hidden_layers": 6,
22
+ "pad_token_id": 0,
23
+ "tie_word_embeddings": true,
24
+ "transformers_version": "5.20.1",
25
+ "type_vocab_size": 2,
26
+ "use_cache": false,
27
+ "vocab_size": 30522
28
+ }
checkpoint-368000/config_sentence_transformers.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "__version__": {
3
+ "pytorch": "2.11.0+cu130",
4
+ "sentence_transformers": "6.2.0",
5
+ "transformers": "5.20.1"
6
+ },
7
+ "default_prompt_name": null,
8
+ "model_type": "SentenceTransformer",
9
+ "prompts": {
10
+ "document": "",
11
+ "query": ""
12
+ },
13
+ "similarity_fn_name": "cosine"
14
+ }
checkpoint-368000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:672470c36dfdbb766d333ee39291ae96837d355a09987658cc9e7e22c614df01
3
+ size 140285096
checkpoint-368000/modules.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ [
2
+ {
3
+ "idx": 0,
4
+ "name": "0",
5
+ "path": "",
6
+ "type": "sentence_transformers.base.modules.transformer.Transformer"
7
+ },
8
+ {
9
+ "idx": 1,
10
+ "name": "1",
11
+ "path": "1_Pooling",
12
+ "type": "sentence_transformers.sentence_transformer.modules.pooling.Pooling"
13
+ },
14
+ {
15
+ "idx": 2,
16
+ "name": "2",
17
+ "path": "2_Dense",
18
+ "type": "sentence_transformers.base.modules.dense.Dense"
19
+ },
20
+ {
21
+ "idx": 3,
22
+ "name": "3",
23
+ "path": "3_Normalize",
24
+ "type": "sentence_transformers.base.modules.normalize.Normalize"
25
+ }
26
+ ]
checkpoint-368000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:661547e3fb2f188b4998bbd3cf6723ab131ed32420be316f80809751628731b3
3
+ size 282727883
checkpoint-368000/resume_check.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"plan_sha256": "8e27a3ab0e919c63ea2020b870b6d51a0b7e8947e8ecf234ba8164a11e60e8ff", "batches_per_step": 2}
checkpoint-368000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:67c49a3e00cf605c2fe20d295881b44c67e1b97eab96f03fd4a069a36ea5a0c8
3
+ size 14645
checkpoint-368000/scaler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9fa4a31c25601f32ae2765d0e837c884e153f35bd7833edc49aef49f9446c06d
3
+ size 1383
checkpoint-368000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d966a1d62443dd37c698b71455607992b1d4dd2acad3aab2bdc3de93d3c5aff4
3
+ size 1465
checkpoint-368000/sentence_bert_config.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "transformer_task": "feature-extraction",
3
+ "modality_config": {
4
+ "text": {
5
+ "method": "forward",
6
+ "method_output_name": "last_hidden_state"
7
+ }
8
+ },
9
+ "module_output_name": "token_embeddings"
10
+ }
checkpoint-368000/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-368000/tokenizer_config.json ADDED
@@ -0,0 +1,22 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "backend": "tokenizers",
3
+ "cls_token": "[CLS]",
4
+ "do_lower_case": true,
5
+ "is_local": false,
6
+ "local_files_only": false,
7
+ "mask_token": "[MASK]",
8
+ "max_length": 512,
9
+ "model_max_length": 512,
10
+ "pad_to_multiple_of": null,
11
+ "pad_token": "[PAD]",
12
+ "pad_token_type_id": 0,
13
+ "padding_side": "right",
14
+ "sep_token": "[SEP]",
15
+ "stride": 0,
16
+ "strip_accents": null,
17
+ "tokenize_chinese_chars": true,
18
+ "tokenizer_class": "BertTokenizer",
19
+ "truncation_side": "right",
20
+ "truncation_strategy": "longest_first",
21
+ "unk_token": "[UNK]"
22
+ }
checkpoint-368000/trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-368000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3e3b4a8172882851e46d1688ac8c7d6a88036a57aef91b7f79204654be4a616d
3
+ size 5585