README.md
83.2 KB · 3081 lines · markdown Raw
1 ---
2 pipeline_tag: sentence-similarity
3 tags:
4 - sentence-transformers
5 - feature-extraction
6 - sentence-similarity
7 - mteb
8 - arctic
9 - snowflake-arctic-embed
10 - transformers.js
11 model-index:
12 - name: snowflake-snowflake-arctic-embed-xs
13 results:
14 - task:
15 type: Classification
16 dataset:
17 type: mteb/amazon_counterfactual
18 name: MTEB AmazonCounterfactualClassification (en)
19 config: en
20 split: test
21 revision: e8379541af4e31359cca9fbcf4b00f2671dba205
22 metrics:
23 - type: accuracy
24 value: 65.08955223880598
25 - type: ap
26 value: 28.514291209445364
27 - type: f1
28 value: 59.2604580112738
29 - task:
30 type: Classification
31 dataset:
32 type: mteb/amazon_polarity
33 name: MTEB AmazonPolarityClassification
34 config: default
35 split: test
36 revision: e2d317d38cd51312af73b3d32a06d1a08b442046
37 metrics:
38 - type: accuracy
39 value: 70.035375
40 - type: ap
41 value: 64.29444264250405
42 - type: f1
43 value: 69.78382333907138
44 - task:
45 type: Classification
46 dataset:
47 type: mteb/amazon_reviews_multi
48 name: MTEB AmazonReviewsClassification (en)
49 config: en
50 split: test
51 revision: 1399c76144fd37290681b995c656ef9b2e06e26d
52 metrics:
53 - type: accuracy
54 value: 35.343999999999994
55 - type: f1
56 value: 34.69618251902858
57 - task:
58 type: Retrieval
59 dataset:
60 type: mteb/arguana
61 name: MTEB ArguAna
62 config: default
63 split: test
64 revision: c22ab2a51041ffd869aaddef7af8d8215647e41a
65 metrics:
66 - type: map_at_1
67 value: 28.592000000000002
68 - type: map_at_10
69 value: 43.597
70 - type: map_at_100
71 value: 44.614
72 - type: map_at_1000
73 value: 44.624
74 - type: map_at_3
75 value: 38.928000000000004
76 - type: map_at_5
77 value: 41.453
78 - type: mrr_at_1
79 value: 29.232000000000003
80 - type: mrr_at_10
81 value: 43.829
82 - type: mrr_at_100
83 value: 44.852
84 - type: mrr_at_1000
85 value: 44.862
86 - type: mrr_at_3
87 value: 39.118
88 - type: mrr_at_5
89 value: 41.703
90 - type: ndcg_at_1
91 value: 28.592000000000002
92 - type: ndcg_at_10
93 value: 52.081
94 - type: ndcg_at_100
95 value: 56.37
96 - type: ndcg_at_1000
97 value: 56.598000000000006
98 - type: ndcg_at_3
99 value: 42.42
100 - type: ndcg_at_5
101 value: 46.965
102 - type: precision_at_1
103 value: 28.592000000000002
104 - type: precision_at_10
105 value: 7.922999999999999
106 - type: precision_at_100
107 value: 0.979
108 - type: precision_at_1000
109 value: 0.1
110 - type: precision_at_3
111 value: 17.52
112 - type: precision_at_5
113 value: 12.717
114 - type: recall_at_1
115 value: 28.592000000000002
116 - type: recall_at_10
117 value: 79.232
118 - type: recall_at_100
119 value: 97.866
120 - type: recall_at_1000
121 value: 99.57300000000001
122 - type: recall_at_3
123 value: 52.559999999999995
124 - type: recall_at_5
125 value: 63.585
126 - task:
127 type: Clustering
128 dataset:
129 type: mteb/arxiv-clustering-p2p
130 name: MTEB ArxivClusteringP2P
131 config: default
132 split: test
133 revision: a122ad7f3f0291bf49cc6f4d32aa80929df69d5d
134 metrics:
135 - type: v_measure
136 value: 43.50220588953974
137 - task:
138 type: Clustering
139 dataset:
140 type: mteb/arxiv-clustering-s2s
141 name: MTEB ArxivClusteringS2S
142 config: default
143 split: test
144 revision: f910caf1a6075f7329cdf8c1a6135696f37dbd53
145 metrics:
146 - type: v_measure
147 value: 32.08725826118282
148 - task:
149 type: Reranking
150 dataset:
151 type: mteb/askubuntudupquestions-reranking
152 name: MTEB AskUbuntuDupQuestions
153 config: default
154 split: test
155 revision: 2000358ca161889fa9c082cb41daa8dcfb161a54
156 metrics:
157 - type: map
158 value: 60.25381587694928
159 - type: mrr
160 value: 73.79776194873148
161 - task:
162 type: STS
163 dataset:
164 type: mteb/biosses-sts
165 name: MTEB BIOSSES
166 config: default
167 split: test
168 revision: d3fb88f8f02e40887cd149695127462bbcf29b4a
169 metrics:
170 - type: cos_sim_pearson
171 value: 85.47489332445278
172 - type: cos_sim_spearman
173 value: 84.05432487336698
174 - type: euclidean_pearson
175 value: 84.5108222177219
176 - type: euclidean_spearman
177 value: 84.05432487336698
178 - type: manhattan_pearson
179 value: 84.20440618321464
180 - type: manhattan_spearman
181 value: 83.9290208134097
182 - task:
183 type: Classification
184 dataset:
185 type: mteb/banking77
186 name: MTEB Banking77Classification
187 config: default
188 split: test
189 revision: 0fd18e25b25c072e09e0d92ab615fda904d66300
190 metrics:
191 - type: accuracy
192 value: 76.37337662337663
193 - type: f1
194 value: 75.33296834885043
195 - task:
196 type: Clustering
197 dataset:
198 type: jinaai/big-patent-clustering
199 name: MTEB BigPatentClustering
200 config: default
201 split: test
202 revision: 62d5330920bca426ce9d3c76ea914f15fc83e891
203 metrics:
204 - type: v_measure
205 value: 21.31174373264835
206 - task:
207 type: Clustering
208 dataset:
209 type: mteb/biorxiv-clustering-p2p
210 name: MTEB BiorxivClusteringP2P
211 config: default
212 split: test
213 revision: 65b79d1d13f80053f67aca9498d9402c2d9f1f40
214 metrics:
215 - type: v_measure
216 value: 34.481973521597844
217 - task:
218 type: Clustering
219 dataset:
220 type: mteb/biorxiv-clustering-s2s
221 name: MTEB BiorxivClusteringS2S
222 config: default
223 split: test
224 revision: 258694dd0231531bc1fd9de6ceb52a0853c6d908
225 metrics:
226 - type: v_measure
227 value: 26.14094256567341
228 - task:
229 type: Retrieval
230 dataset:
231 type: mteb/cqadupstack-android
232 name: MTEB CQADupstackAndroidRetrieval
233 config: default
234 split: test
235 revision: f46a197baaae43b4f621051089b82a364682dfeb
236 metrics:
237 - type: map_at_1
238 value: 32.527
239 - type: map_at_10
240 value: 43.699
241 - type: map_at_100
242 value: 45.03
243 - type: map_at_1000
244 value: 45.157000000000004
245 - type: map_at_3
246 value: 39.943
247 - type: map_at_5
248 value: 42.324
249 - type: mrr_at_1
250 value: 39.771
251 - type: mrr_at_10
252 value: 49.277
253 - type: mrr_at_100
254 value: 49.956
255 - type: mrr_at_1000
256 value: 50.005
257 - type: mrr_at_3
258 value: 46.304
259 - type: mrr_at_5
260 value: 48.493
261 - type: ndcg_at_1
262 value: 39.771
263 - type: ndcg_at_10
264 value: 49.957
265 - type: ndcg_at_100
266 value: 54.678000000000004
267 - type: ndcg_at_1000
268 value: 56.751
269 - type: ndcg_at_3
270 value: 44.608
271 - type: ndcg_at_5
272 value: 47.687000000000005
273 - type: precision_at_1
274 value: 39.771
275 - type: precision_at_10
276 value: 9.557
277 - type: precision_at_100
278 value: 1.5010000000000001
279 - type: precision_at_1000
280 value: 0.194
281 - type: precision_at_3
282 value: 21.173000000000002
283 - type: precision_at_5
284 value: 15.794
285 - type: recall_at_1
286 value: 32.527
287 - type: recall_at_10
288 value: 61.791
289 - type: recall_at_100
290 value: 81.49300000000001
291 - type: recall_at_1000
292 value: 95.014
293 - type: recall_at_3
294 value: 46.605000000000004
295 - type: recall_at_5
296 value: 54.83
297 - task:
298 type: Retrieval
299 dataset:
300 type: mteb/cqadupstack-english
301 name: MTEB CQADupstackEnglishRetrieval
302 config: default
303 split: test
304 revision: ad9991cb51e31e31e430383c75ffb2885547b5f0
305 metrics:
306 - type: map_at_1
307 value: 29.424
308 - type: map_at_10
309 value: 38.667
310 - type: map_at_100
311 value: 39.771
312 - type: map_at_1000
313 value: 39.899
314 - type: map_at_3
315 value: 35.91
316 - type: map_at_5
317 value: 37.45
318 - type: mrr_at_1
319 value: 36.687999999999995
320 - type: mrr_at_10
321 value: 44.673
322 - type: mrr_at_100
323 value: 45.289
324 - type: mrr_at_1000
325 value: 45.338
326 - type: mrr_at_3
327 value: 42.601
328 - type: mrr_at_5
329 value: 43.875
330 - type: ndcg_at_1
331 value: 36.687999999999995
332 - type: ndcg_at_10
333 value: 44.013000000000005
334 - type: ndcg_at_100
335 value: 48.13
336 - type: ndcg_at_1000
337 value: 50.294000000000004
338 - type: ndcg_at_3
339 value: 40.056999999999995
340 - type: ndcg_at_5
341 value: 41.902
342 - type: precision_at_1
343 value: 36.687999999999995
344 - type: precision_at_10
345 value: 8.158999999999999
346 - type: precision_at_100
347 value: 1.321
348 - type: precision_at_1000
349 value: 0.179
350 - type: precision_at_3
351 value: 19.045
352 - type: precision_at_5
353 value: 13.427
354 - type: recall_at_1
355 value: 29.424
356 - type: recall_at_10
357 value: 53.08500000000001
358 - type: recall_at_100
359 value: 70.679
360 - type: recall_at_1000
361 value: 84.66
362 - type: recall_at_3
363 value: 41.399
364 - type: recall_at_5
365 value: 46.632
366 - task:
367 type: Retrieval
368 dataset:
369 type: mteb/cqadupstack-gaming
370 name: MTEB CQADupstackGamingRetrieval
371 config: default
372 split: test
373 revision: 4885aa143210c98657558c04aaf3dc47cfb54340
374 metrics:
375 - type: map_at_1
376 value: 39.747
377 - type: map_at_10
378 value: 51.452
379 - type: map_at_100
380 value: 52.384
381 - type: map_at_1000
382 value: 52.437
383 - type: map_at_3
384 value: 48.213
385 - type: map_at_5
386 value: 50.195
387 - type: mrr_at_1
388 value: 45.391999999999996
389 - type: mrr_at_10
390 value: 54.928
391 - type: mrr_at_100
392 value: 55.532000000000004
393 - type: mrr_at_1000
394 value: 55.565
395 - type: mrr_at_3
396 value: 52.456
397 - type: mrr_at_5
398 value: 54.054
399 - type: ndcg_at_1
400 value: 45.391999999999996
401 - type: ndcg_at_10
402 value: 57.055
403 - type: ndcg_at_100
404 value: 60.751999999999995
405 - type: ndcg_at_1000
406 value: 61.864
407 - type: ndcg_at_3
408 value: 51.662
409 - type: ndcg_at_5
410 value: 54.613
411 - type: precision_at_1
412 value: 45.391999999999996
413 - type: precision_at_10
414 value: 9.103
415 - type: precision_at_100
416 value: 1.1780000000000002
417 - type: precision_at_1000
418 value: 0.132
419 - type: precision_at_3
420 value: 22.717000000000002
421 - type: precision_at_5
422 value: 15.812000000000001
423 - type: recall_at_1
424 value: 39.747
425 - type: recall_at_10
426 value: 70.10499999999999
427 - type: recall_at_100
428 value: 86.23100000000001
429 - type: recall_at_1000
430 value: 94.025
431 - type: recall_at_3
432 value: 55.899
433 - type: recall_at_5
434 value: 63.05500000000001
435 - task:
436 type: Retrieval
437 dataset:
438 type: mteb/cqadupstack-gis
439 name: MTEB CQADupstackGisRetrieval
440 config: default
441 split: test
442 revision: 5003b3064772da1887988e05400cf3806fe491f2
443 metrics:
444 - type: map_at_1
445 value: 27.168999999999997
446 - type: map_at_10
447 value: 34.975
448 - type: map_at_100
449 value: 35.94
450 - type: map_at_1000
451 value: 36.021
452 - type: map_at_3
453 value: 32.35
454 - type: map_at_5
455 value: 33.831
456 - type: mrr_at_1
457 value: 28.701
458 - type: mrr_at_10
459 value: 36.698
460 - type: mrr_at_100
461 value: 37.546
462 - type: mrr_at_1000
463 value: 37.613
464 - type: mrr_at_3
465 value: 34.256
466 - type: mrr_at_5
467 value: 35.685
468 - type: ndcg_at_1
469 value: 28.701
470 - type: ndcg_at_10
471 value: 39.639
472 - type: ndcg_at_100
473 value: 44.389
474 - type: ndcg_at_1000
475 value: 46.46
476 - type: ndcg_at_3
477 value: 34.52
478 - type: ndcg_at_5
479 value: 37.076
480 - type: precision_at_1
481 value: 28.701
482 - type: precision_at_10
483 value: 5.955
484 - type: precision_at_100
485 value: 0.8880000000000001
486 - type: precision_at_1000
487 value: 0.109
488 - type: precision_at_3
489 value: 14.274999999999999
490 - type: precision_at_5
491 value: 10.011000000000001
492 - type: recall_at_1
493 value: 27.168999999999997
494 - type: recall_at_10
495 value: 52.347
496 - type: recall_at_100
497 value: 74.1
498 - type: recall_at_1000
499 value: 89.739
500 - type: recall_at_3
501 value: 38.567
502 - type: recall_at_5
503 value: 44.767
504 - task:
505 type: Retrieval
506 dataset:
507 type: mteb/cqadupstack-mathematica
508 name: MTEB CQADupstackMathematicaRetrieval
509 config: default
510 split: test
511 revision: 90fceea13679c63fe563ded68f3b6f06e50061de
512 metrics:
513 - type: map_at_1
514 value: 15.872
515 - type: map_at_10
516 value: 23.153000000000002
517 - type: map_at_100
518 value: 24.311
519 - type: map_at_1000
520 value: 24.432000000000002
521 - type: map_at_3
522 value: 20.707
523 - type: map_at_5
524 value: 21.921
525 - type: mrr_at_1
526 value: 19.776
527 - type: mrr_at_10
528 value: 27.755999999999997
529 - type: mrr_at_100
530 value: 28.709
531 - type: mrr_at_1000
532 value: 28.778
533 - type: mrr_at_3
534 value: 25.186999999999998
535 - type: mrr_at_5
536 value: 26.43
537 - type: ndcg_at_1
538 value: 19.776
539 - type: ndcg_at_10
540 value: 28.288999999999998
541 - type: ndcg_at_100
542 value: 34.011
543 - type: ndcg_at_1000
544 value: 36.916
545 - type: ndcg_at_3
546 value: 23.551
547 - type: ndcg_at_5
548 value: 25.429000000000002
549 - type: precision_at_1
550 value: 19.776
551 - type: precision_at_10
552 value: 5.311
553 - type: precision_at_100
554 value: 0.9440000000000001
555 - type: precision_at_1000
556 value: 0.132
557 - type: precision_at_3
558 value: 11.360000000000001
559 - type: precision_at_5
560 value: 8.209
561 - type: recall_at_1
562 value: 15.872
563 - type: recall_at_10
564 value: 39.726
565 - type: recall_at_100
566 value: 65.035
567 - type: recall_at_1000
568 value: 85.846
569 - type: recall_at_3
570 value: 26.432
571 - type: recall_at_5
572 value: 31.22
573 - task:
574 type: Retrieval
575 dataset:
576 type: mteb/cqadupstack-physics
577 name: MTEB CQADupstackPhysicsRetrieval
578 config: default
579 split: test
580 revision: 79531abbd1fb92d06c6d6315a0cbbbf5bb247ea4
581 metrics:
582 - type: map_at_1
583 value: 28.126
584 - type: map_at_10
585 value: 37.537
586 - type: map_at_100
587 value: 38.807
588 - type: map_at_1000
589 value: 38.923
590 - type: map_at_3
591 value: 34.65
592 - type: map_at_5
593 value: 36.248000000000005
594 - type: mrr_at_1
595 value: 34.649
596 - type: mrr_at_10
597 value: 42.893
598 - type: mrr_at_100
599 value: 43.721
600 - type: mrr_at_1000
601 value: 43.775999999999996
602 - type: mrr_at_3
603 value: 40.488
604 - type: mrr_at_5
605 value: 41.729
606 - type: ndcg_at_1
607 value: 34.649
608 - type: ndcg_at_10
609 value: 43.072
610 - type: ndcg_at_100
611 value: 48.464
612 - type: ndcg_at_1000
613 value: 50.724000000000004
614 - type: ndcg_at_3
615 value: 38.506
616 - type: ndcg_at_5
617 value: 40.522000000000006
618 - type: precision_at_1
619 value: 34.649
620 - type: precision_at_10
621 value: 7.68
622 - type: precision_at_100
623 value: 1.214
624 - type: precision_at_1000
625 value: 0.16
626 - type: precision_at_3
627 value: 18.029999999999998
628 - type: precision_at_5
629 value: 12.666
630 - type: recall_at_1
631 value: 28.126
632 - type: recall_at_10
633 value: 54.396
634 - type: recall_at_100
635 value: 76.988
636 - type: recall_at_1000
637 value: 91.85799999999999
638 - type: recall_at_3
639 value: 41.169
640 - type: recall_at_5
641 value: 46.658
642 - task:
643 type: Retrieval
644 dataset:
645 type: mteb/cqadupstack-programmers
646 name: MTEB CQADupstackProgrammersRetrieval
647 config: default
648 split: test
649 revision: 6184bc1440d2dbc7612be22b50686b8826d22b32
650 metrics:
651 - type: map_at_1
652 value: 26.68
653 - type: map_at_10
654 value: 35.702
655 - type: map_at_100
656 value: 36.864999999999995
657 - type: map_at_1000
658 value: 36.977
659 - type: map_at_3
660 value: 32.828
661 - type: map_at_5
662 value: 34.481
663 - type: mrr_at_1
664 value: 32.991
665 - type: mrr_at_10
666 value: 40.993
667 - type: mrr_at_100
668 value: 41.827
669 - type: mrr_at_1000
670 value: 41.887
671 - type: mrr_at_3
672 value: 38.623000000000005
673 - type: mrr_at_5
674 value: 40.021
675 - type: ndcg_at_1
676 value: 32.991
677 - type: ndcg_at_10
678 value: 41.036
679 - type: ndcg_at_100
680 value: 46.294000000000004
681 - type: ndcg_at_1000
682 value: 48.644
683 - type: ndcg_at_3
684 value: 36.419000000000004
685 - type: ndcg_at_5
686 value: 38.618
687 - type: precision_at_1
688 value: 32.991
689 - type: precision_at_10
690 value: 7.385999999999999
691 - type: precision_at_100
692 value: 1.176
693 - type: precision_at_1000
694 value: 0.151
695 - type: precision_at_3
696 value: 17.122999999999998
697 - type: precision_at_5
698 value: 12.215
699 - type: recall_at_1
700 value: 26.68
701 - type: recall_at_10
702 value: 51.644
703 - type: recall_at_100
704 value: 74.55000000000001
705 - type: recall_at_1000
706 value: 90.825
707 - type: recall_at_3
708 value: 38.579
709 - type: recall_at_5
710 value: 44.512
711 - task:
712 type: Retrieval
713 dataset:
714 type: mteb/cqadupstack
715 name: MTEB CQADupstackRetrieval
716 config: default
717 split: test
718 revision: 4ffe81d471b1924886b33c7567bfb200e9eec5c4
719 metrics:
720 - type: map_at_1
721 value: 26.30825
722 - type: map_at_10
723 value: 34.97866666666666
724 - type: map_at_100
725 value: 36.109249999999996
726 - type: map_at_1000
727 value: 36.22508333333333
728 - type: map_at_3
729 value: 32.239083333333326
730 - type: map_at_5
731 value: 33.75933333333334
732 - type: mrr_at_1
733 value: 31.05308333333333
734 - type: mrr_at_10
735 value: 39.099833333333336
736 - type: mrr_at_100
737 value: 39.92008333333334
738 - type: mrr_at_1000
739 value: 39.980000000000004
740 - type: mrr_at_3
741 value: 36.75958333333333
742 - type: mrr_at_5
743 value: 38.086416666666665
744 - type: ndcg_at_1
745 value: 31.05308333333333
746 - type: ndcg_at_10
747 value: 40.11558333333334
748 - type: ndcg_at_100
749 value: 45.05966666666667
750 - type: ndcg_at_1000
751 value: 47.36516666666667
752 - type: ndcg_at_3
753 value: 35.490833333333335
754 - type: ndcg_at_5
755 value: 37.64541666666666
756 - type: precision_at_1
757 value: 31.05308333333333
758 - type: precision_at_10
759 value: 6.968416666666666
760 - type: precision_at_100
761 value: 1.1156666666666666
762 - type: precision_at_1000
763 value: 0.14950000000000002
764 - type: precision_at_3
765 value: 16.123
766 - type: precision_at_5
767 value: 11.451166666666666
768 - type: recall_at_1
769 value: 26.30825
770 - type: recall_at_10
771 value: 51.19283333333333
772 - type: recall_at_100
773 value: 73.0285
774 - type: recall_at_1000
775 value: 89.11133333333333
776 - type: recall_at_3
777 value: 38.26208333333333
778 - type: recall_at_5
779 value: 43.855916666666666
780 - task:
781 type: Retrieval
782 dataset:
783 type: mteb/cqadupstack-stats
784 name: MTEB CQADupstackStatsRetrieval
785 config: default
786 split: test
787 revision: 65ac3a16b8e91f9cee4c9828cc7c335575432a2a
788 metrics:
789 - type: map_at_1
790 value: 23.363999999999997
791 - type: map_at_10
792 value: 30.606
793 - type: map_at_100
794 value: 31.491999999999997
795 - type: map_at_1000
796 value: 31.578
797 - type: map_at_3
798 value: 28.610000000000003
799 - type: map_at_5
800 value: 29.602
801 - type: mrr_at_1
802 value: 26.38
803 - type: mrr_at_10
804 value: 33.472
805 - type: mrr_at_100
806 value: 34.299
807 - type: mrr_at_1000
808 value: 34.361999999999995
809 - type: mrr_at_3
810 value: 31.696999999999996
811 - type: mrr_at_5
812 value: 32.503
813 - type: ndcg_at_1
814 value: 26.38
815 - type: ndcg_at_10
816 value: 34.772999999999996
817 - type: ndcg_at_100
818 value: 39.334
819 - type: ndcg_at_1000
820 value: 41.676
821 - type: ndcg_at_3
822 value: 31.097
823 - type: ndcg_at_5
824 value: 32.561
825 - type: precision_at_1
826 value: 26.38
827 - type: precision_at_10
828 value: 5.475
829 - type: precision_at_100
830 value: 0.84
831 - type: precision_at_1000
832 value: 0.11100000000000002
833 - type: precision_at_3
834 value: 13.395000000000001
835 - type: precision_at_5
836 value: 9.11
837 - type: recall_at_1
838 value: 23.363999999999997
839 - type: recall_at_10
840 value: 44.656
841 - type: recall_at_100
842 value: 65.77199999999999
843 - type: recall_at_1000
844 value: 83.462
845 - type: recall_at_3
846 value: 34.213
847 - type: recall_at_5
848 value: 38.091
849 - task:
850 type: Retrieval
851 dataset:
852 type: mteb/cqadupstack-tex
853 name: MTEB CQADupstackTexRetrieval
854 config: default
855 split: test
856 revision: 46989137a86843e03a6195de44b09deda022eec7
857 metrics:
858 - type: map_at_1
859 value: 17.971999999999998
860 - type: map_at_10
861 value: 24.913
862 - type: map_at_100
863 value: 25.916
864 - type: map_at_1000
865 value: 26.049
866 - type: map_at_3
867 value: 22.569
868 - type: map_at_5
869 value: 23.858999999999998
870 - type: mrr_at_1
871 value: 21.748
872 - type: mrr_at_10
873 value: 28.711
874 - type: mrr_at_100
875 value: 29.535
876 - type: mrr_at_1000
877 value: 29.621
878 - type: mrr_at_3
879 value: 26.484999999999996
880 - type: mrr_at_5
881 value: 27.701999999999998
882 - type: ndcg_at_1
883 value: 21.748
884 - type: ndcg_at_10
885 value: 29.412
886 - type: ndcg_at_100
887 value: 34.204
888 - type: ndcg_at_1000
889 value: 37.358000000000004
890 - type: ndcg_at_3
891 value: 25.202
892 - type: ndcg_at_5
893 value: 27.128000000000004
894 - type: precision_at_1
895 value: 21.748
896 - type: precision_at_10
897 value: 5.279
898 - type: precision_at_100
899 value: 0.902
900 - type: precision_at_1000
901 value: 0.135
902 - type: precision_at_3
903 value: 11.551
904 - type: precision_at_5
905 value: 8.437999999999999
906 - type: recall_at_1
907 value: 17.971999999999998
908 - type: recall_at_10
909 value: 39.186
910 - type: recall_at_100
911 value: 60.785999999999994
912 - type: recall_at_1000
913 value: 83.372
914 - type: recall_at_3
915 value: 27.584999999999997
916 - type: recall_at_5
917 value: 32.448
918 - task:
919 type: Retrieval
920 dataset:
921 type: mteb/cqadupstack-unix
922 name: MTEB CQADupstackUnixRetrieval
923 config: default
924 split: test
925 revision: 6c6430d3a6d36f8d2a829195bc5dc94d7e063e53
926 metrics:
927 - type: map_at_1
928 value: 26.684
929 - type: map_at_10
930 value: 35.188
931 - type: map_at_100
932 value: 36.379
933 - type: map_at_1000
934 value: 36.481
935 - type: map_at_3
936 value: 32.401
937 - type: map_at_5
938 value: 34.132
939 - type: mrr_at_1
940 value: 31.063000000000002
941 - type: mrr_at_10
942 value: 39.104
943 - type: mrr_at_100
944 value: 40.062999999999995
945 - type: mrr_at_1000
946 value: 40.119
947 - type: mrr_at_3
948 value: 36.692
949 - type: mrr_at_5
950 value: 38.161
951 - type: ndcg_at_1
952 value: 31.063000000000002
953 - type: ndcg_at_10
954 value: 40.096
955 - type: ndcg_at_100
956 value: 45.616
957 - type: ndcg_at_1000
958 value: 47.869
959 - type: ndcg_at_3
960 value: 35.256
961 - type: ndcg_at_5
962 value: 37.826
963 - type: precision_at_1
964 value: 31.063000000000002
965 - type: precision_at_10
966 value: 6.622999999999999
967 - type: precision_at_100
968 value: 1.046
969 - type: precision_at_1000
970 value: 0.135
971 - type: precision_at_3
972 value: 15.641
973 - type: precision_at_5
974 value: 11.231
975 - type: recall_at_1
976 value: 26.684
977 - type: recall_at_10
978 value: 51.092999999999996
979 - type: recall_at_100
980 value: 75.099
981 - type: recall_at_1000
982 value: 90.644
983 - type: recall_at_3
984 value: 38.063
985 - type: recall_at_5
986 value: 44.518
987 - task:
988 type: Retrieval
989 dataset:
990 type: mteb/cqadupstack-webmasters
991 name: MTEB CQADupstackWebmastersRetrieval
992 config: default
993 split: test
994 revision: 160c094312a0e1facb97e55eeddb698c0abe3571
995 metrics:
996 - type: map_at_1
997 value: 26.249
998 - type: map_at_10
999 value: 34.694
1000 - type: map_at_100
1001 value: 36.208
1002 - type: map_at_1000
1003 value: 36.443
1004 - type: map_at_3
1005 value: 31.868000000000002
1006 - type: map_at_5
1007 value: 33.018
1008 - type: mrr_at_1
1009 value: 31.818
1010 - type: mrr_at_10
1011 value: 39.416000000000004
1012 - type: mrr_at_100
1013 value: 40.327
1014 - type: mrr_at_1000
1015 value: 40.388000000000005
1016 - type: mrr_at_3
1017 value: 37.120999999999995
1018 - type: mrr_at_5
1019 value: 38.07
1020 - type: ndcg_at_1
1021 value: 31.818
1022 - type: ndcg_at_10
1023 value: 40.405
1024 - type: ndcg_at_100
1025 value: 45.816
1026 - type: ndcg_at_1000
1027 value: 48.403
1028 - type: ndcg_at_3
1029 value: 35.823
1030 - type: ndcg_at_5
1031 value: 37.191
1032 - type: precision_at_1
1033 value: 31.818
1034 - type: precision_at_10
1035 value: 7.806
1036 - type: precision_at_100
1037 value: 1.518
1038 - type: precision_at_1000
1039 value: 0.241
1040 - type: precision_at_3
1041 value: 16.535
1042 - type: precision_at_5
1043 value: 11.738999999999999
1044 - type: recall_at_1
1045 value: 26.249
1046 - type: recall_at_10
1047 value: 50.928
1048 - type: recall_at_100
1049 value: 75.271
1050 - type: recall_at_1000
1051 value: 91.535
1052 - type: recall_at_3
1053 value: 37.322
1054 - type: recall_at_5
1055 value: 41.318
1056 - task:
1057 type: Retrieval
1058 dataset:
1059 type: mteb/cqadupstack-wordpress
1060 name: MTEB CQADupstackWordpressRetrieval
1061 config: default
1062 split: test
1063 revision: 4ffe81d471b1924886b33c7567bfb200e9eec5c4
1064 metrics:
1065 - type: map_at_1
1066 value: 21.884999999999998
1067 - type: map_at_10
1068 value: 29.158
1069 - type: map_at_100
1070 value: 30.208000000000002
1071 - type: map_at_1000
1072 value: 30.304
1073 - type: map_at_3
1074 value: 26.82
1075 - type: map_at_5
1076 value: 28.051
1077 - type: mrr_at_1
1078 value: 23.66
1079 - type: mrr_at_10
1080 value: 31.277
1081 - type: mrr_at_100
1082 value: 32.237
1083 - type: mrr_at_1000
1084 value: 32.308
1085 - type: mrr_at_3
1086 value: 29.205
1087 - type: mrr_at_5
1088 value: 30.314000000000004
1089 - type: ndcg_at_1
1090 value: 23.66
1091 - type: ndcg_at_10
1092 value: 33.64
1093 - type: ndcg_at_100
1094 value: 39.028
1095 - type: ndcg_at_1000
1096 value: 41.423
1097 - type: ndcg_at_3
1098 value: 29.189
1099 - type: ndcg_at_5
1100 value: 31.191999999999997
1101 - type: precision_at_1
1102 value: 23.66
1103 - type: precision_at_10
1104 value: 5.287
1105 - type: precision_at_100
1106 value: 0.86
1107 - type: precision_at_1000
1108 value: 0.11499999999999999
1109 - type: precision_at_3
1110 value: 12.631
1111 - type: precision_at_5
1112 value: 8.762
1113 - type: recall_at_1
1114 value: 21.884999999999998
1115 - type: recall_at_10
1116 value: 45.357
1117 - type: recall_at_100
1118 value: 70.338
1119 - type: recall_at_1000
1120 value: 88.356
1121 - type: recall_at_3
1122 value: 33.312000000000005
1123 - type: recall_at_5
1124 value: 38.222
1125 - task:
1126 type: Retrieval
1127 dataset:
1128 type: mteb/climate-fever
1129 name: MTEB ClimateFEVER
1130 config: default
1131 split: test
1132 revision: 47f2ac6acb640fc46020b02a5b59fdda04d39380
1133 metrics:
1134 - type: map_at_1
1135 value: 13.058
1136 - type: map_at_10
1137 value: 21.549
1138 - type: map_at_100
1139 value: 23.287
1140 - type: map_at_1000
1141 value: 23.444000000000003
1142 - type: map_at_3
1143 value: 18.18
1144 - type: map_at_5
1145 value: 19.886
1146 - type: mrr_at_1
1147 value: 28.73
1148 - type: mrr_at_10
1149 value: 40.014
1150 - type: mrr_at_100
1151 value: 40.827000000000005
1152 - type: mrr_at_1000
1153 value: 40.866
1154 - type: mrr_at_3
1155 value: 36.602000000000004
1156 - type: mrr_at_5
1157 value: 38.702
1158 - type: ndcg_at_1
1159 value: 28.73
1160 - type: ndcg_at_10
1161 value: 29.881
1162 - type: ndcg_at_100
1163 value: 36.662
1164 - type: ndcg_at_1000
1165 value: 39.641999999999996
1166 - type: ndcg_at_3
1167 value: 24.661
1168 - type: ndcg_at_5
1169 value: 26.548
1170 - type: precision_at_1
1171 value: 28.73
1172 - type: precision_at_10
1173 value: 9.094
1174 - type: precision_at_100
1175 value: 1.6480000000000001
1176 - type: precision_at_1000
1177 value: 0.22100000000000003
1178 - type: precision_at_3
1179 value: 17.98
1180 - type: precision_at_5
1181 value: 13.811000000000002
1182 - type: recall_at_1
1183 value: 13.058
1184 - type: recall_at_10
1185 value: 35.458
1186 - type: recall_at_100
1187 value: 58.719
1188 - type: recall_at_1000
1189 value: 75.495
1190 - type: recall_at_3
1191 value: 22.607
1192 - type: recall_at_5
1193 value: 28.067999999999998
1194 - task:
1195 type: Retrieval
1196 dataset:
1197 type: mteb/dbpedia
1198 name: MTEB DBPedia
1199 config: default
1200 split: test
1201 revision: c0f706b76e590d620bd6618b3ca8efdd34e2d659
1202 metrics:
1203 - type: map_at_1
1204 value: 8.811
1205 - type: map_at_10
1206 value: 19.134999999999998
1207 - type: map_at_100
1208 value: 26.905
1209 - type: map_at_1000
1210 value: 28.503
1211 - type: map_at_3
1212 value: 13.863
1213 - type: map_at_5
1214 value: 16.062
1215 - type: mrr_at_1
1216 value: 67
1217 - type: mrr_at_10
1218 value: 74.607
1219 - type: mrr_at_100
1220 value: 74.941
1221 - type: mrr_at_1000
1222 value: 74.954
1223 - type: mrr_at_3
1224 value: 73.042
1225 - type: mrr_at_5
1226 value: 73.992
1227 - type: ndcg_at_1
1228 value: 52.87500000000001
1229 - type: ndcg_at_10
1230 value: 40.199
1231 - type: ndcg_at_100
1232 value: 44.901
1233 - type: ndcg_at_1000
1234 value: 52.239999999999995
1235 - type: ndcg_at_3
1236 value: 44.983000000000004
1237 - type: ndcg_at_5
1238 value: 42.137
1239 - type: precision_at_1
1240 value: 67
1241 - type: precision_at_10
1242 value: 31.8
1243 - type: precision_at_100
1244 value: 10.315000000000001
1245 - type: precision_at_1000
1246 value: 2.0420000000000003
1247 - type: precision_at_3
1248 value: 48.667
1249 - type: precision_at_5
1250 value: 40.9
1251 - type: recall_at_1
1252 value: 8.811
1253 - type: recall_at_10
1254 value: 24.503
1255 - type: recall_at_100
1256 value: 51.288999999999994
1257 - type: recall_at_1000
1258 value: 74.827
1259 - type: recall_at_3
1260 value: 15.254999999999999
1261 - type: recall_at_5
1262 value: 18.698999999999998
1263 - task:
1264 type: Classification
1265 dataset:
1266 type: mteb/emotion
1267 name: MTEB EmotionClassification
1268 config: default
1269 split: test
1270 revision: 4f58c6b202a23cf9a4da393831edf4f9183cad37
1271 metrics:
1272 - type: accuracy
1273 value: 41.839999999999996
1274 - type: f1
1275 value: 37.78718146306379
1276 - task:
1277 type: Retrieval
1278 dataset:
1279 type: mteb/fever
1280 name: MTEB FEVER
1281 config: default
1282 split: test
1283 revision: bea83ef9e8fb933d90a2f1d5515737465d613e12
1284 metrics:
1285 - type: map_at_1
1286 value: 68.47999999999999
1287 - type: map_at_10
1288 value: 78.782
1289 - type: map_at_100
1290 value: 79.021
1291 - type: map_at_1000
1292 value: 79.035
1293 - type: map_at_3
1294 value: 77.389
1295 - type: map_at_5
1296 value: 78.347
1297 - type: mrr_at_1
1298 value: 73.837
1299 - type: mrr_at_10
1300 value: 83.41499999999999
1301 - type: mrr_at_100
1302 value: 83.53399999999999
1303 - type: mrr_at_1000
1304 value: 83.535
1305 - type: mrr_at_3
1306 value: 82.32300000000001
1307 - type: mrr_at_5
1308 value: 83.13000000000001
1309 - type: ndcg_at_1
1310 value: 73.837
1311 - type: ndcg_at_10
1312 value: 83.404
1313 - type: ndcg_at_100
1314 value: 84.287
1315 - type: ndcg_at_1000
1316 value: 84.52199999999999
1317 - type: ndcg_at_3
1318 value: 81.072
1319 - type: ndcg_at_5
1320 value: 82.537
1321 - type: precision_at_1
1322 value: 73.837
1323 - type: precision_at_10
1324 value: 10.254000000000001
1325 - type: precision_at_100
1326 value: 1.088
1327 - type: precision_at_1000
1328 value: 0.11299999999999999
1329 - type: precision_at_3
1330 value: 31.538
1331 - type: precision_at_5
1332 value: 19.811
1333 - type: recall_at_1
1334 value: 68.47999999999999
1335 - type: recall_at_10
1336 value: 92.98100000000001
1337 - type: recall_at_100
1338 value: 96.50800000000001
1339 - type: recall_at_1000
1340 value: 97.925
1341 - type: recall_at_3
1342 value: 86.764
1343 - type: recall_at_5
1344 value: 90.39
1345 - task:
1346 type: Retrieval
1347 dataset:
1348 type: mteb/fiqa
1349 name: MTEB FiQA2018
1350 config: default
1351 split: test
1352 revision: 27a168819829fe9bcd655c2df245fb19452e8e06
1353 metrics:
1354 - type: map_at_1
1355 value: 16.786
1356 - type: map_at_10
1357 value: 26.97
1358 - type: map_at_100
1359 value: 28.488000000000003
1360 - type: map_at_1000
1361 value: 28.665000000000003
1362 - type: map_at_3
1363 value: 23.3
1364 - type: map_at_5
1365 value: 25.249
1366 - type: mrr_at_1
1367 value: 33.025
1368 - type: mrr_at_10
1369 value: 41.86
1370 - type: mrr_at_100
1371 value: 42.673
1372 - type: mrr_at_1000
1373 value: 42.714
1374 - type: mrr_at_3
1375 value: 39.403
1376 - type: mrr_at_5
1377 value: 40.723
1378 - type: ndcg_at_1
1379 value: 33.025
1380 - type: ndcg_at_10
1381 value: 34.522999999999996
1382 - type: ndcg_at_100
1383 value: 40.831
1384 - type: ndcg_at_1000
1385 value: 44.01
1386 - type: ndcg_at_3
1387 value: 30.698999999999998
1388 - type: ndcg_at_5
1389 value: 31.832
1390 - type: precision_at_1
1391 value: 33.025
1392 - type: precision_at_10
1393 value: 9.583
1394 - type: precision_at_100
1395 value: 1.619
1396 - type: precision_at_1000
1397 value: 0.22100000000000003
1398 - type: precision_at_3
1399 value: 20.216
1400 - type: precision_at_5
1401 value: 15.031
1402 - type: recall_at_1
1403 value: 16.786
1404 - type: recall_at_10
1405 value: 41.969
1406 - type: recall_at_100
1407 value: 66.353
1408 - type: recall_at_1000
1409 value: 85.299
1410 - type: recall_at_3
1411 value: 28.111000000000004
1412 - type: recall_at_5
1413 value: 33.645
1414 - task:
1415 type: Retrieval
1416 dataset:
1417 type: mteb/hotpotqa
1418 name: MTEB HotpotQA
1419 config: default
1420 split: test
1421 revision: ab518f4d6fcca38d87c25209f94beba119d02014
1422 metrics:
1423 - type: map_at_1
1424 value: 37.346000000000004
1425 - type: map_at_10
1426 value: 56.184999999999995
1427 - type: map_at_100
1428 value: 57.062000000000005
1429 - type: map_at_1000
1430 value: 57.126999999999995
1431 - type: map_at_3
1432 value: 52.815
1433 - type: map_at_5
1434 value: 54.893
1435 - type: mrr_at_1
1436 value: 74.693
1437 - type: mrr_at_10
1438 value: 81.128
1439 - type: mrr_at_100
1440 value: 81.356
1441 - type: mrr_at_1000
1442 value: 81.363
1443 - type: mrr_at_3
1444 value: 80.05600000000001
1445 - type: mrr_at_5
1446 value: 80.74
1447 - type: ndcg_at_1
1448 value: 74.693
1449 - type: ndcg_at_10
1450 value: 65.249
1451 - type: ndcg_at_100
1452 value: 68.357
1453 - type: ndcg_at_1000
1454 value: 69.64200000000001
1455 - type: ndcg_at_3
1456 value: 60.377
1457 - type: ndcg_at_5
1458 value: 63.044
1459 - type: precision_at_1
1460 value: 74.693
1461 - type: precision_at_10
1462 value: 13.630999999999998
1463 - type: precision_at_100
1464 value: 1.606
1465 - type: precision_at_1000
1466 value: 0.178
1467 - type: precision_at_3
1468 value: 38.222
1469 - type: precision_at_5
1470 value: 25.040000000000003
1471 - type: recall_at_1
1472 value: 37.346000000000004
1473 - type: recall_at_10
1474 value: 68.157
1475 - type: recall_at_100
1476 value: 80.297
1477 - type: recall_at_1000
1478 value: 88.832
1479 - type: recall_at_3
1480 value: 57.333
1481 - type: recall_at_5
1482 value: 62.6
1483 - task:
1484 type: Classification
1485 dataset:
1486 type: mteb/imdb
1487 name: MTEB ImdbClassification
1488 config: default
1489 split: test
1490 revision: 3d86128a09e091d6018b6d26cad27f2739fc2db7
1491 metrics:
1492 - type: accuracy
1493 value: 62.80240000000001
1494 - type: ap
1495 value: 58.22949464075975
1496 - type: f1
1497 value: 62.55694937343487
1498 - task:
1499 type: Retrieval
1500 dataset:
1501 type: mteb/msmarco
1502 name: MTEB MSMARCO
1503 config: default
1504 split: dev
1505 revision: c5a29a104738b98a9e76336939199e264163d4a0
1506 metrics:
1507 - type: map_at_1
1508 value: 20.918
1509 - type: map_at_10
1510 value: 32.732
1511 - type: map_at_100
1512 value: 33.922000000000004
1513 - type: map_at_1000
1514 value: 33.976
1515 - type: map_at_3
1516 value: 29.051
1517 - type: map_at_5
1518 value: 31.101
1519 - type: mrr_at_1
1520 value: 21.418
1521 - type: mrr_at_10
1522 value: 33.284000000000006
1523 - type: mrr_at_100
1524 value: 34.426
1525 - type: mrr_at_1000
1526 value: 34.473
1527 - type: mrr_at_3
1528 value: 29.644
1529 - type: mrr_at_5
1530 value: 31.691000000000003
1531 - type: ndcg_at_1
1532 value: 21.418
1533 - type: ndcg_at_10
1534 value: 39.427
1535 - type: ndcg_at_100
1536 value: 45.190999999999995
1537 - type: ndcg_at_1000
1538 value: 46.544000000000004
1539 - type: ndcg_at_3
1540 value: 31.885
1541 - type: ndcg_at_5
1542 value: 35.555
1543 - type: precision_at_1
1544 value: 21.418
1545 - type: precision_at_10
1546 value: 6.254999999999999
1547 - type: precision_at_100
1548 value: 0.915
1549 - type: precision_at_1000
1550 value: 0.10300000000000001
1551 - type: precision_at_3
1552 value: 13.591000000000001
1553 - type: precision_at_5
1554 value: 10.011000000000001
1555 - type: recall_at_1
1556 value: 20.918
1557 - type: recall_at_10
1558 value: 60.074000000000005
1559 - type: recall_at_100
1560 value: 86.726
1561 - type: recall_at_1000
1562 value: 97.116
1563 - type: recall_at_3
1564 value: 39.506
1565 - type: recall_at_5
1566 value: 48.319
1567 - task:
1568 type: Classification
1569 dataset:
1570 type: mteb/mtop_domain
1571 name: MTEB MTOPDomainClassification (en)
1572 config: en
1573 split: test
1574 revision: d80d48c1eb48d3562165c59d59d0034df9fff0bf
1575 metrics:
1576 - type: accuracy
1577 value: 90.79799361605106
1578 - type: f1
1579 value: 90.0757957511057
1580 - task:
1581 type: Classification
1582 dataset:
1583 type: mteb/mtop_intent
1584 name: MTEB MTOPIntentClassification (en)
1585 config: en
1586 split: test
1587 revision: ae001d0e6b1228650b7bd1c2c65fb50ad11a8aba
1588 metrics:
1589 - type: accuracy
1590 value: 58.00501595987233
1591 - type: f1
1592 value: 39.85731569133947
1593 - task:
1594 type: Classification
1595 dataset:
1596 type: masakhane/masakhanews
1597 name: MTEB MasakhaNEWSClassification (eng)
1598 config: eng
1599 split: test
1600 revision: 8ccc72e69e65f40c70e117d8b3c08306bb788b60
1601 metrics:
1602 - type: accuracy
1603 value: 77.10970464135022
1604 - type: f1
1605 value: 76.12037616356896
1606 - task:
1607 type: Clustering
1608 dataset:
1609 type: masakhane/masakhanews
1610 name: MTEB MasakhaNEWSClusteringP2P (eng)
1611 config: eng
1612 split: test
1613 revision: 8ccc72e69e65f40c70e117d8b3c08306bb788b60
1614 metrics:
1615 - type: v_measure
1616 value: 69.81323966287493
1617 - task:
1618 type: Clustering
1619 dataset:
1620 type: masakhane/masakhanews
1621 name: MTEB MasakhaNEWSClusteringS2S (eng)
1622 config: eng
1623 split: test
1624 revision: 8ccc72e69e65f40c70e117d8b3c08306bb788b60
1625 metrics:
1626 - type: v_measure
1627 value: 33.112774215788455
1628 - task:
1629 type: Classification
1630 dataset:
1631 type: mteb/amazon_massive_intent
1632 name: MTEB MassiveIntentClassification (en)
1633 config: en
1634 split: test
1635 revision: 31efe3c427b0bae9c22cbb560b8f15491cc6bed7
1636 metrics:
1637 - type: accuracy
1638 value: 63.51042367182246
1639 - type: f1
1640 value: 60.99310361578824
1641 - task:
1642 type: Classification
1643 dataset:
1644 type: mteb/amazon_massive_scenario
1645 name: MTEB MassiveScenarioClassification (en)
1646 config: en
1647 split: test
1648 revision: 7d571f92784cd94a019292a1f45445077d0ef634
1649 metrics:
1650 - type: accuracy
1651 value: 71.0053799596503
1652 - type: f1
1653 value: 69.7794673003686
1654 - task:
1655 type: Clustering
1656 dataset:
1657 type: mteb/medrxiv-clustering-p2p
1658 name: MTEB MedrxivClusteringP2P
1659 config: default
1660 split: test
1661 revision: e7a26af6f3ae46b30dde8737f02c07b1505bcc73
1662 metrics:
1663 - type: v_measure
1664 value: 30.56899174856954
1665 - task:
1666 type: Clustering
1667 dataset:
1668 type: mteb/medrxiv-clustering-s2s
1669 name: MTEB MedrxivClusteringS2S
1670 config: default
1671 split: test
1672 revision: 35191c8c0dca72d8ff3efcd72aa802307d469663
1673 metrics:
1674 - type: v_measure
1675 value: 26.21848014733929
1676 - task:
1677 type: Reranking
1678 dataset:
1679 type: mteb/mind_small
1680 name: MTEB MindSmallReranking
1681 config: default
1682 split: test
1683 revision: 3bdac13927fdc888b903db93b2ffdbd90b295a69
1684 metrics:
1685 - type: map
1686 value: 30.256308756916646
1687 - type: mrr
1688 value: 31.123872086825656
1689 - task:
1690 type: Retrieval
1691 dataset:
1692 type: mteb/nfcorpus
1693 name: MTEB NFCorpus
1694 config: default
1695 split: test
1696 revision: ec0fa4fe99da2ff19ca1214b7966684033a58814
1697 metrics:
1698 - type: map_at_1
1699 value: 5.07
1700 - type: map_at_10
1701 value: 11.286999999999999
1702 - type: map_at_100
1703 value: 13.630999999999998
1704 - type: map_at_1000
1705 value: 14.844
1706 - type: map_at_3
1707 value: 8.395
1708 - type: map_at_5
1709 value: 9.721
1710 - type: mrr_at_1
1711 value: 41.486000000000004
1712 - type: mrr_at_10
1713 value: 51.041000000000004
1714 - type: mrr_at_100
1715 value: 51.661
1716 - type: mrr_at_1000
1717 value: 51.7
1718 - type: mrr_at_3
1719 value: 49.226
1720 - type: mrr_at_5
1721 value: 50.526
1722 - type: ndcg_at_1
1723 value: 39.783
1724 - type: ndcg_at_10
1725 value: 30.885
1726 - type: ndcg_at_100
1727 value: 27.459
1728 - type: ndcg_at_1000
1729 value: 35.988
1730 - type: ndcg_at_3
1731 value: 36.705
1732 - type: ndcg_at_5
1733 value: 34.156
1734 - type: precision_at_1
1735 value: 41.486000000000004
1736 - type: precision_at_10
1737 value: 22.415
1738 - type: precision_at_100
1739 value: 6.819999999999999
1740 - type: precision_at_1000
1741 value: 1.8980000000000001
1742 - type: precision_at_3
1743 value: 34.572
1744 - type: precision_at_5
1745 value: 29.287999999999997
1746 - type: recall_at_1
1747 value: 5.07
1748 - type: recall_at_10
1749 value: 14.576
1750 - type: recall_at_100
1751 value: 27.112000000000002
1752 - type: recall_at_1000
1753 value: 57.995
1754 - type: recall_at_3
1755 value: 9.242
1756 - type: recall_at_5
1757 value: 11.668000000000001
1758 - task:
1759 type: Retrieval
1760 dataset:
1761 type: mteb/nq
1762 name: MTEB NQ
1763 config: default
1764 split: test
1765 revision: b774495ed302d8c44a3a7ea25c90dbce03968f31
1766 metrics:
1767 - type: map_at_1
1768 value: 32.263999999999996
1769 - type: map_at_10
1770 value: 47.219
1771 - type: map_at_100
1772 value: 48.209999999999994
1773 - type: map_at_1000
1774 value: 48.24
1775 - type: map_at_3
1776 value: 42.905
1777 - type: map_at_5
1778 value: 45.501000000000005
1779 - type: mrr_at_1
1780 value: 36.153
1781 - type: mrr_at_10
1782 value: 49.636
1783 - type: mrr_at_100
1784 value: 50.357
1785 - type: mrr_at_1000
1786 value: 50.378
1787 - type: mrr_at_3
1788 value: 46.094
1789 - type: mrr_at_5
1790 value: 48.233
1791 - type: ndcg_at_1
1792 value: 36.124
1793 - type: ndcg_at_10
1794 value: 54.764
1795 - type: ndcg_at_100
1796 value: 58.867999999999995
1797 - type: ndcg_at_1000
1798 value: 59.548
1799 - type: ndcg_at_3
1800 value: 46.717999999999996
1801 - type: ndcg_at_5
1802 value: 50.981
1803 - type: precision_at_1
1804 value: 36.124
1805 - type: precision_at_10
1806 value: 8.931000000000001
1807 - type: precision_at_100
1808 value: 1.126
1809 - type: precision_at_1000
1810 value: 0.11900000000000001
1811 - type: precision_at_3
1812 value: 21.051000000000002
1813 - type: precision_at_5
1814 value: 15.104000000000001
1815 - type: recall_at_1
1816 value: 32.263999999999996
1817 - type: recall_at_10
1818 value: 75.39099999999999
1819 - type: recall_at_100
1820 value: 93.038
1821 - type: recall_at_1000
1822 value: 98.006
1823 - type: recall_at_3
1824 value: 54.562999999999995
1825 - type: recall_at_5
1826 value: 64.352
1827 - task:
1828 type: Classification
1829 dataset:
1830 type: ag_news
1831 name: MTEB NewsClassification
1832 config: default
1833 split: test
1834 revision: eb185aade064a813bc0b7f42de02595523103ca4
1835 metrics:
1836 - type: accuracy
1837 value: 77.75
1838 - type: f1
1839 value: 77.504243291547
1840 - task:
1841 type: PairClassification
1842 dataset:
1843 type: GEM/opusparcus
1844 name: MTEB OpusparcusPC (en)
1845 config: en
1846 split: test
1847 revision: 9e9b1f8ef51616073f47f306f7f47dd91663f86a
1848 metrics:
1849 - type: cos_sim_accuracy
1850 value: 99.89816700610999
1851 - type: cos_sim_ap
1852 value: 100
1853 - type: cos_sim_f1
1854 value: 99.9490575649516
1855 - type: cos_sim_precision
1856 value: 100
1857 - type: cos_sim_recall
1858 value: 99.89816700610999
1859 - type: dot_accuracy
1860 value: 99.89816700610999
1861 - type: dot_ap
1862 value: 100
1863 - type: dot_f1
1864 value: 99.9490575649516
1865 - type: dot_precision
1866 value: 100
1867 - type: dot_recall
1868 value: 99.89816700610999
1869 - type: euclidean_accuracy
1870 value: 99.89816700610999
1871 - type: euclidean_ap
1872 value: 100
1873 - type: euclidean_f1
1874 value: 99.9490575649516
1875 - type: euclidean_precision
1876 value: 100
1877 - type: euclidean_recall
1878 value: 99.89816700610999
1879 - type: manhattan_accuracy
1880 value: 99.89816700610999
1881 - type: manhattan_ap
1882 value: 100
1883 - type: manhattan_f1
1884 value: 99.9490575649516
1885 - type: manhattan_precision
1886 value: 100
1887 - type: manhattan_recall
1888 value: 99.89816700610999
1889 - type: max_accuracy
1890 value: 99.89816700610999
1891 - type: max_ap
1892 value: 100
1893 - type: max_f1
1894 value: 99.9490575649516
1895 - task:
1896 type: PairClassification
1897 dataset:
1898 type: paws-x
1899 name: MTEB PawsX (en)
1900 config: en
1901 split: test
1902 revision: 8a04d940a42cd40658986fdd8e3da561533a3646
1903 metrics:
1904 - type: cos_sim_accuracy
1905 value: 61.75000000000001
1906 - type: cos_sim_ap
1907 value: 57.9482264289061
1908 - type: cos_sim_f1
1909 value: 62.444061962134256
1910 - type: cos_sim_precision
1911 value: 45.3953953953954
1912 - type: cos_sim_recall
1913 value: 100
1914 - type: dot_accuracy
1915 value: 61.75000000000001
1916 - type: dot_ap
1917 value: 57.94808038610475
1918 - type: dot_f1
1919 value: 62.444061962134256
1920 - type: dot_precision
1921 value: 45.3953953953954
1922 - type: dot_recall
1923 value: 100
1924 - type: euclidean_accuracy
1925 value: 61.75000000000001
1926 - type: euclidean_ap
1927 value: 57.94808038610475
1928 - type: euclidean_f1
1929 value: 62.444061962134256
1930 - type: euclidean_precision
1931 value: 45.3953953953954
1932 - type: euclidean_recall
1933 value: 100
1934 - type: manhattan_accuracy
1935 value: 61.7
1936 - type: manhattan_ap
1937 value: 57.996119308184966
1938 - type: manhattan_f1
1939 value: 62.46078773091669
1940 - type: manhattan_precision
1941 value: 45.66768603465851
1942 - type: manhattan_recall
1943 value: 98.78721058434398
1944 - type: max_accuracy
1945 value: 61.75000000000001
1946 - type: max_ap
1947 value: 57.996119308184966
1948 - type: max_f1
1949 value: 62.46078773091669
1950 - task:
1951 type: Retrieval
1952 dataset:
1953 type: mteb/quora
1954 name: MTEB QuoraRetrieval
1955 config: default
1956 split: test
1957 revision: e4e08e0b7dbe3c8700f0daef558ff32256715259
1958 metrics:
1959 - type: map_at_1
1960 value: 69.001
1961 - type: map_at_10
1962 value: 82.573
1963 - type: map_at_100
1964 value: 83.226
1965 - type: map_at_1000
1966 value: 83.246
1967 - type: map_at_3
1968 value: 79.625
1969 - type: map_at_5
1970 value: 81.491
1971 - type: mrr_at_1
1972 value: 79.44
1973 - type: mrr_at_10
1974 value: 85.928
1975 - type: mrr_at_100
1976 value: 86.05199999999999
1977 - type: mrr_at_1000
1978 value: 86.054
1979 - type: mrr_at_3
1980 value: 84.847
1981 - type: mrr_at_5
1982 value: 85.596
1983 - type: ndcg_at_1
1984 value: 79.41
1985 - type: ndcg_at_10
1986 value: 86.568
1987 - type: ndcg_at_100
1988 value: 87.965
1989 - type: ndcg_at_1000
1990 value: 88.134
1991 - type: ndcg_at_3
1992 value: 83.55900000000001
1993 - type: ndcg_at_5
1994 value: 85.244
1995 - type: precision_at_1
1996 value: 79.41
1997 - type: precision_at_10
1998 value: 13.108
1999 - type: precision_at_100
2000 value: 1.509
2001 - type: precision_at_1000
2002 value: 0.156
2003 - type: precision_at_3
2004 value: 36.443
2005 - type: precision_at_5
2006 value: 24.03
2007 - type: recall_at_1
2008 value: 69.001
2009 - type: recall_at_10
2010 value: 94.132
2011 - type: recall_at_100
2012 value: 99.043
2013 - type: recall_at_1000
2014 value: 99.878
2015 - type: recall_at_3
2016 value: 85.492
2017 - type: recall_at_5
2018 value: 90.226
2019 - task:
2020 type: Clustering
2021 dataset:
2022 type: mteb/reddit-clustering
2023 name: MTEB RedditClustering
2024 config: default
2025 split: test
2026 revision: 24640382cdbf8abc73003fb0fa6d111a705499eb
2027 metrics:
2028 - type: v_measure
2029 value: 48.3161352736264
2030 - task:
2031 type: Clustering
2032 dataset:
2033 type: mteb/reddit-clustering-p2p
2034 name: MTEB RedditClusteringP2P
2035 config: default
2036 split: test
2037 revision: 385e3cb46b4cfa89021f56c4380204149d0efe33
2038 metrics:
2039 - type: v_measure
2040 value: 57.83784484156747
2041 - task:
2042 type: Retrieval
2043 dataset:
2044 type: mteb/scidocs
2045 name: MTEB SCIDOCS
2046 config: default
2047 split: test
2048 revision: f8c2fcf00f625baaa80f62ec5bd9e1fff3b8ae88
2049 metrics:
2050 - type: map_at_1
2051 value: 4.403
2052 - type: map_at_10
2053 value: 10.922
2054 - type: map_at_100
2055 value: 12.626000000000001
2056 - type: map_at_1000
2057 value: 12.883
2058 - type: map_at_3
2059 value: 7.982
2060 - type: map_at_5
2061 value: 9.442
2062 - type: mrr_at_1
2063 value: 21.7
2064 - type: mrr_at_10
2065 value: 31.653
2066 - type: mrr_at_100
2067 value: 32.757999999999996
2068 - type: mrr_at_1000
2069 value: 32.824999999999996
2070 - type: mrr_at_3
2071 value: 28.266999999999996
2072 - type: mrr_at_5
2073 value: 30.127
2074 - type: ndcg_at_1
2075 value: 21.7
2076 - type: ndcg_at_10
2077 value: 18.355
2078 - type: ndcg_at_100
2079 value: 25.228
2080 - type: ndcg_at_1000
2081 value: 30.164
2082 - type: ndcg_at_3
2083 value: 17.549
2084 - type: ndcg_at_5
2085 value: 15.260000000000002
2086 - type: precision_at_1
2087 value: 21.7
2088 - type: precision_at_10
2089 value: 9.47
2090 - type: precision_at_100
2091 value: 1.9290000000000003
2092 - type: precision_at_1000
2093 value: 0.312
2094 - type: precision_at_3
2095 value: 16.3
2096 - type: precision_at_5
2097 value: 13.28
2098 - type: recall_at_1
2099 value: 4.403
2100 - type: recall_at_10
2101 value: 19.18
2102 - type: recall_at_100
2103 value: 39.182
2104 - type: recall_at_1000
2105 value: 63.378
2106 - type: recall_at_3
2107 value: 9.934999999999999
2108 - type: recall_at_5
2109 value: 13.459999999999999
2110 - task:
2111 type: STS
2112 dataset:
2113 type: mteb/sickr-sts
2114 name: MTEB SICK-R
2115 config: default
2116 split: test
2117 revision: 20a6d6f312dd54037fe07a32d58e5e168867909d
2118 metrics:
2119 - type: cos_sim_pearson
2120 value: 76.90841073432534
2121 - type: cos_sim_spearman
2122 value: 69.2566375434526
2123 - type: euclidean_pearson
2124 value: 73.00183878559413
2125 - type: euclidean_spearman
2126 value: 69.25664656235413
2127 - type: manhattan_pearson
2128 value: 72.89594756197533
2129 - type: manhattan_spearman
2130 value: 69.23247111043545
2131 - task:
2132 type: STS
2133 dataset:
2134 type: mteb/sts12-sts
2135 name: MTEB STS12
2136 config: default
2137 split: test
2138 revision: a0d554a64d88156834ff5ae9920b964011b16384
2139 metrics:
2140 - type: cos_sim_pearson
2141 value: 69.60878511794063
2142 - type: cos_sim_spearman
2143 value: 65.89916377105551
2144 - type: euclidean_pearson
2145 value: 66.90761876557181
2146 - type: euclidean_spearman
2147 value: 65.89915018368384
2148 - type: manhattan_pearson
2149 value: 66.78502575257721
2150 - type: manhattan_spearman
2151 value: 65.79977053467938
2152 - task:
2153 type: STS
2154 dataset:
2155 type: mteb/sts13-sts
2156 name: MTEB STS13
2157 config: default
2158 split: test
2159 revision: 7e90230a92c190f1bf69ae9002b8cea547a64cca
2160 metrics:
2161 - type: cos_sim_pearson
2162 value: 77.2869334987418
2163 - type: cos_sim_spearman
2164 value: 77.86961921643416
2165 - type: euclidean_pearson
2166 value: 77.43179820479914
2167 - type: euclidean_spearman
2168 value: 77.86961921643416
2169 - type: manhattan_pearson
2170 value: 77.18900647348373
2171 - type: manhattan_spearman
2172 value: 77.61209060062608
2173 - task:
2174 type: STS
2175 dataset:
2176 type: mteb/sts14-sts
2177 name: MTEB STS14
2178 config: default
2179 split: test
2180 revision: 6031580fec1f6af667f0bd2da0a551cf4f0b2375
2181 metrics:
2182 - type: cos_sim_pearson
2183 value: 76.26453932960364
2184 - type: cos_sim_spearman
2185 value: 72.81574657995401
2186 - type: euclidean_pearson
2187 value: 75.0708953437423
2188 - type: euclidean_spearman
2189 value: 72.81574657995401
2190 - type: manhattan_pearson
2191 value: 74.88396609999512
2192 - type: manhattan_spearman
2193 value: 72.65437562156805
2194 - task:
2195 type: STS
2196 dataset:
2197 type: mteb/sts15-sts
2198 name: MTEB STS15
2199 config: default
2200 split: test
2201 revision: ae752c7c21bf194d8b67fd573edf7ae58183cbe3
2202 metrics:
2203 - type: cos_sim_pearson
2204 value: 82.37827653919395
2205 - type: cos_sim_spearman
2206 value: 83.4885552472602
2207 - type: euclidean_pearson
2208 value: 82.89377087926749
2209 - type: euclidean_spearman
2210 value: 83.4885552472602
2211 - type: manhattan_pearson
2212 value: 82.82440771787735
2213 - type: manhattan_spearman
2214 value: 83.41449537888975
2215 - task:
2216 type: STS
2217 dataset:
2218 type: mteb/sts16-sts
2219 name: MTEB STS16
2220 config: default
2221 split: test
2222 revision: 4d8694f8f0e0100860b497b999b3dbed754a0513
2223 metrics:
2224 - type: cos_sim_pearson
2225 value: 78.7995043673964
2226 - type: cos_sim_spearman
2227 value: 80.57804447517638
2228 - type: euclidean_pearson
2229 value: 80.03013884278195
2230 - type: euclidean_spearman
2231 value: 80.57804447517638
2232 - type: manhattan_pearson
2233 value: 80.13406111544424
2234 - type: manhattan_spearman
2235 value: 80.65354602648962
2236 - task:
2237 type: STS
2238 dataset:
2239 type: mteb/sts17-crosslingual-sts
2240 name: MTEB STS17 (en-en)
2241 config: en-en
2242 split: test
2243 revision: af5e6fb845001ecf41f4c1e033ce921939a2a68d
2244 metrics:
2245 - type: cos_sim_pearson
2246 value: 83.63565989937278
2247 - type: cos_sim_spearman
2248 value: 84.4948593656943
2249 - type: euclidean_pearson
2250 value: 84.68743074820951
2251 - type: euclidean_spearman
2252 value: 84.4948593656943
2253 - type: manhattan_pearson
2254 value: 84.43639397781811
2255 - type: manhattan_spearman
2256 value: 84.32595552115242
2257 - task:
2258 type: STS
2259 dataset:
2260 type: mteb/sts22-crosslingual-sts
2261 name: MTEB STS22 (en)
2262 config: en
2263 split: test
2264 revision: eea2b4fe26a775864c896887d910b76a8098ad3f
2265 metrics:
2266 - type: cos_sim_pearson
2267 value: 65.06382649277246
2268 - type: cos_sim_spearman
2269 value: 66.28447782018655
2270 - type: euclidean_pearson
2271 value: 67.09895930908392
2272 - type: euclidean_spearman
2273 value: 66.28447782018655
2274 - type: manhattan_pearson
2275 value: 66.96342453888376
2276 - type: manhattan_spearman
2277 value: 66.33876259551842
2278 - task:
2279 type: STS
2280 dataset:
2281 type: mteb/stsbenchmark-sts
2282 name: MTEB STSBenchmark
2283 config: default
2284 split: test
2285 revision: b0fddb56ed78048fa8b90373c8a3cfc37b684831
2286 metrics:
2287 - type: cos_sim_pearson
2288 value: 78.43883428940346
2289 - type: cos_sim_spearman
2290 value: 79.18395553127085
2291 - type: euclidean_pearson
2292 value: 79.22986635457109
2293 - type: euclidean_spearman
2294 value: 79.18395553127085
2295 - type: manhattan_pearson
2296 value: 79.10921229934691
2297 - type: manhattan_spearman
2298 value: 79.02283553930171
2299 - task:
2300 type: STS
2301 dataset:
2302 type: PhilipMay/stsb_multi_mt
2303 name: MTEB STSBenchmarkMultilingualSTS (en)
2304 config: en
2305 split: test
2306 revision: 93d57ef91790589e3ce9c365164337a8a78b7632
2307 metrics:
2308 - type: cos_sim_pearson
2309 value: 78.43883433444418
2310 - type: cos_sim_spearman
2311 value: 79.18395553127085
2312 - type: euclidean_pearson
2313 value: 79.22986642351681
2314 - type: euclidean_spearman
2315 value: 79.18395553127085
2316 - type: manhattan_pearson
2317 value: 79.10921236746302
2318 - type: manhattan_spearman
2319 value: 79.02283553930171
2320 - task:
2321 type: Reranking
2322 dataset:
2323 type: mteb/scidocs-reranking
2324 name: MTEB SciDocsRR
2325 config: default
2326 split: test
2327 revision: d3c5e1fc0b855ab6097bf1cda04dd73947d7caab
2328 metrics:
2329 - type: map
2330 value: 76.9361627171417
2331 - type: mrr
2332 value: 93.06577046773126
2333 - task:
2334 type: Retrieval
2335 dataset:
2336 type: mteb/scifact
2337 name: MTEB SciFact
2338 config: default
2339 split: test
2340 revision: 0228b52cf27578f30900b9e5271d331663a030d7
2341 metrics:
2342 - type: map_at_1
2343 value: 50.693999999999996
2344 - type: map_at_10
2345 value: 59.784000000000006
2346 - type: map_at_100
2347 value: 60.443000000000005
2348 - type: map_at_1000
2349 value: 60.480000000000004
2350 - type: map_at_3
2351 value: 57.028
2352 - type: map_at_5
2353 value: 58.306999999999995
2354 - type: mrr_at_1
2355 value: 53.333
2356 - type: mrr_at_10
2357 value: 61.565000000000005
2358 - type: mrr_at_100
2359 value: 62.095
2360 - type: mrr_at_1000
2361 value: 62.131
2362 - type: mrr_at_3
2363 value: 59.721999999999994
2364 - type: mrr_at_5
2365 value: 60.589000000000006
2366 - type: ndcg_at_1
2367 value: 53.333
2368 - type: ndcg_at_10
2369 value: 64.512
2370 - type: ndcg_at_100
2371 value: 67.366
2372 - type: ndcg_at_1000
2373 value: 68.46799999999999
2374 - type: ndcg_at_3
2375 value: 59.748999999999995
2376 - type: ndcg_at_5
2377 value: 61.526
2378 - type: precision_at_1
2379 value: 53.333
2380 - type: precision_at_10
2381 value: 8.733
2382 - type: precision_at_100
2383 value: 1.027
2384 - type: precision_at_1000
2385 value: 0.11199999999999999
2386 - type: precision_at_3
2387 value: 23.222
2388 - type: precision_at_5
2389 value: 15.2
2390 - type: recall_at_1
2391 value: 50.693999999999996
2392 - type: recall_at_10
2393 value: 77.333
2394 - type: recall_at_100
2395 value: 90.10000000000001
2396 - type: recall_at_1000
2397 value: 99
2398 - type: recall_at_3
2399 value: 64.39399999999999
2400 - type: recall_at_5
2401 value: 68.7
2402 - task:
2403 type: PairClassification
2404 dataset:
2405 type: mteb/sprintduplicatequestions-pairclassification
2406 name: MTEB SprintDuplicateQuestions
2407 config: default
2408 split: test
2409 revision: d66bd1f72af766a5cc4b0ca5e00c162f89e8cc46
2410 metrics:
2411 - type: cos_sim_accuracy
2412 value: 99.81386138613861
2413 - type: cos_sim_ap
2414 value: 94.96375600031361
2415 - type: cos_sim_f1
2416 value: 90.36885245901641
2417 - type: cos_sim_precision
2418 value: 92.64705882352942
2419 - type: cos_sim_recall
2420 value: 88.2
2421 - type: dot_accuracy
2422 value: 99.81386138613861
2423 - type: dot_ap
2424 value: 94.96375600031361
2425 - type: dot_f1
2426 value: 90.36885245901641
2427 - type: dot_precision
2428 value: 92.64705882352942
2429 - type: dot_recall
2430 value: 88.2
2431 - type: euclidean_accuracy
2432 value: 99.81386138613861
2433 - type: euclidean_ap
2434 value: 94.96375600031361
2435 - type: euclidean_f1
2436 value: 90.36885245901641
2437 - type: euclidean_precision
2438 value: 92.64705882352942
2439 - type: euclidean_recall
2440 value: 88.2
2441 - type: manhattan_accuracy
2442 value: 99.81287128712871
2443 - type: manhattan_ap
2444 value: 94.92563500640084
2445 - type: manhattan_f1
2446 value: 90.27277406073082
2447 - type: manhattan_precision
2448 value: 93.00106044538707
2449 - type: manhattan_recall
2450 value: 87.7
2451 - type: max_accuracy
2452 value: 99.81386138613861
2453 - type: max_ap
2454 value: 94.96375600031361
2455 - type: max_f1
2456 value: 90.36885245901641
2457 - task:
2458 type: Clustering
2459 dataset:
2460 type: mteb/stackexchange-clustering
2461 name: MTEB StackExchangeClustering
2462 config: default
2463 split: test
2464 revision: 6cbc1f7b2bc0622f2e39d2c77fa502909748c259
2465 metrics:
2466 - type: v_measure
2467 value: 57.486984956276274
2468 - task:
2469 type: Clustering
2470 dataset:
2471 type: mteb/stackexchange-clustering-p2p
2472 name: MTEB StackExchangeClusteringP2P
2473 config: default
2474 split: test
2475 revision: 815ca46b2622cec33ccafc3735d572c266efdb44
2476 metrics:
2477 - type: v_measure
2478 value: 34.58453023612073
2479 - task:
2480 type: Reranking
2481 dataset:
2482 type: mteb/stackoverflowdupquestions-reranking
2483 name: MTEB StackOverflowDupQuestions
2484 config: default
2485 split: test
2486 revision: e185fbe320c72810689fc5848eb6114e1ef5ec69
2487 metrics:
2488 - type: map
2489 value: 50.16317315282306
2490 - type: mrr
2491 value: 50.82617137764197
2492 - task:
2493 type: Summarization
2494 dataset:
2495 type: mteb/summeval
2496 name: MTEB SummEval
2497 config: default
2498 split: test
2499 revision: cda12ad7615edc362dbf25a00fdd61d3b1eaf93c
2500 metrics:
2501 - type: cos_sim_pearson
2502 value: 30.2927995133324
2503 - type: cos_sim_spearman
2504 value: 30.09648622523191
2505 - type: dot_pearson
2506 value: 30.29279853541771
2507 - type: dot_spearman
2508 value: 30.09648622523191
2509 - task:
2510 type: Retrieval
2511 dataset:
2512 type: mteb/trec-covid
2513 name: MTEB TRECCOVID
2514 config: default
2515 split: test
2516 revision: bb9466bac8153a0349341eb1b22e06409e78ef4e
2517 metrics:
2518 - type: map_at_1
2519 value: 0.23500000000000001
2520 - type: map_at_10
2521 value: 2.01
2522 - type: map_at_100
2523 value: 12.064
2524 - type: map_at_1000
2525 value: 27.437
2526 - type: map_at_3
2527 value: 0.6649999999999999
2528 - type: map_at_5
2529 value: 1.0959999999999999
2530 - type: mrr_at_1
2531 value: 88
2532 - type: mrr_at_10
2533 value: 92.667
2534 - type: mrr_at_100
2535 value: 92.667
2536 - type: mrr_at_1000
2537 value: 92.667
2538 - type: mrr_at_3
2539 value: 91.667
2540 - type: mrr_at_5
2541 value: 92.667
2542 - type: ndcg_at_1
2543 value: 84
2544 - type: ndcg_at_10
2545 value: 79.431
2546 - type: ndcg_at_100
2547 value: 60.914
2548 - type: ndcg_at_1000
2549 value: 52.005
2550 - type: ndcg_at_3
2551 value: 82.285
2552 - type: ndcg_at_5
2553 value: 81.565
2554 - type: precision_at_1
2555 value: 88
2556 - type: precision_at_10
2557 value: 84.8
2558 - type: precision_at_100
2559 value: 62.32
2560 - type: precision_at_1000
2561 value: 23.014000000000003
2562 - type: precision_at_3
2563 value: 86.667
2564 - type: precision_at_5
2565 value: 87.2
2566 - type: recall_at_1
2567 value: 0.23500000000000001
2568 - type: recall_at_10
2569 value: 2.19
2570 - type: recall_at_100
2571 value: 14.904
2572 - type: recall_at_1000
2573 value: 47.875
2574 - type: recall_at_3
2575 value: 0.695
2576 - type: recall_at_5
2577 value: 1.165
2578 - task:
2579 type: Retrieval
2580 dataset:
2581 type: mteb/touche2020
2582 name: MTEB Touche2020
2583 config: default
2584 split: test
2585 revision: a34f9a33db75fa0cbb21bb5cfc3dae8dc8bec93f
2586 metrics:
2587 - type: map_at_1
2588 value: 3.639
2589 - type: map_at_10
2590 value: 14.184
2591 - type: map_at_100
2592 value: 20.61
2593 - type: map_at_1000
2594 value: 22.377
2595 - type: map_at_3
2596 value: 9.163
2597 - type: map_at_5
2598 value: 10.773000000000001
2599 - type: mrr_at_1
2600 value: 46.939
2601 - type: mrr_at_10
2602 value: 59.345000000000006
2603 - type: mrr_at_100
2604 value: 60.07599999999999
2605 - type: mrr_at_1000
2606 value: 60.07599999999999
2607 - type: mrr_at_3
2608 value: 55.782
2609 - type: mrr_at_5
2610 value: 58.231
2611 - type: ndcg_at_1
2612 value: 41.837
2613 - type: ndcg_at_10
2614 value: 32.789
2615 - type: ndcg_at_100
2616 value: 42.232
2617 - type: ndcg_at_1000
2618 value: 53.900999999999996
2619 - type: ndcg_at_3
2620 value: 41.963
2621 - type: ndcg_at_5
2622 value: 35.983
2623 - type: precision_at_1
2624 value: 46.939
2625 - type: precision_at_10
2626 value: 28.163
2627 - type: precision_at_100
2628 value: 8.102
2629 - type: precision_at_1000
2630 value: 1.59
2631 - type: precision_at_3
2632 value: 44.897999999999996
2633 - type: precision_at_5
2634 value: 34.694
2635 - type: recall_at_1
2636 value: 3.639
2637 - type: recall_at_10
2638 value: 19.308
2639 - type: recall_at_100
2640 value: 48.992000000000004
2641 - type: recall_at_1000
2642 value: 84.59400000000001
2643 - type: recall_at_3
2644 value: 9.956
2645 - type: recall_at_5
2646 value: 12.33
2647 - task:
2648 type: Classification
2649 dataset:
2650 type: mteb/toxic_conversations_50k
2651 name: MTEB ToxicConversationsClassification
2652 config: default
2653 split: test
2654 revision: edfaf9da55d3dd50d43143d90c1ac476895ae6de
2655 metrics:
2656 - type: accuracy
2657 value: 64.305
2658 - type: ap
2659 value: 11.330746746072599
2660 - type: f1
2661 value: 49.290704382387865
2662 - task:
2663 type: Classification
2664 dataset:
2665 type: mteb/tweet_sentiment_extraction
2666 name: MTEB TweetSentimentExtractionClassification
2667 config: default
2668 split: test
2669 revision: d604517c81ca91fe16a244d1248fc021f9ecee7a
2670 metrics:
2671 - type: accuracy
2672 value: 56.1941143180532
2673 - type: f1
2674 value: 56.40189765095578
2675 - task:
2676 type: Clustering
2677 dataset:
2678 type: mteb/twentynewsgroups-clustering
2679 name: MTEB TwentyNewsgroupsClustering
2680 config: default
2681 split: test
2682 revision: 6125ec4e24fa026cec8a478383ee943acfbd5449
2683 metrics:
2684 - type: v_measure
2685 value: 36.28189332526842
2686 - task:
2687 type: PairClassification
2688 dataset:
2689 type: mteb/twittersemeval2015-pairclassification
2690 name: MTEB TwitterSemEval2015
2691 config: default
2692 split: test
2693 revision: 70970daeab8776df92f5ea462b6173c0b46fd2d1
2694 metrics:
2695 - type: cos_sim_accuracy
2696 value: 83.1912737676581
2697 - type: cos_sim_ap
2698 value: 64.31536990146257
2699 - type: cos_sim_f1
2700 value: 61.095167030191696
2701 - type: cos_sim_precision
2702 value: 54.074375127006704
2703 - type: cos_sim_recall
2704 value: 70.21108179419525
2705 - type: dot_accuracy
2706 value: 83.1912737676581
2707 - type: dot_ap
2708 value: 64.31539216162541
2709 - type: dot_f1
2710 value: 61.095167030191696
2711 - type: dot_precision
2712 value: 54.074375127006704
2713 - type: dot_recall
2714 value: 70.21108179419525
2715 - type: euclidean_accuracy
2716 value: 83.1912737676581
2717 - type: euclidean_ap
2718 value: 64.31538391358727
2719 - type: euclidean_f1
2720 value: 61.095167030191696
2721 - type: euclidean_precision
2722 value: 54.074375127006704
2723 - type: euclidean_recall
2724 value: 70.21108179419525
2725 - type: manhattan_accuracy
2726 value: 83.07206294331525
2727 - type: manhattan_ap
2728 value: 64.14646315556838
2729 - type: manhattan_f1
2730 value: 61.194029850746254
2731 - type: manhattan_precision
2732 value: 54.166666666666664
2733 - type: manhattan_recall
2734 value: 70.31662269129288
2735 - type: max_accuracy
2736 value: 83.1912737676581
2737 - type: max_ap
2738 value: 64.31539216162541
2739 - type: max_f1
2740 value: 61.194029850746254
2741 - task:
2742 type: PairClassification
2743 dataset:
2744 type: mteb/twitterurlcorpus-pairclassification
2745 name: MTEB TwitterURLCorpus
2746 config: default
2747 split: test
2748 revision: 8b6510b0b1fa4e4c4f879467980e9be563ec1cdf
2749 metrics:
2750 - type: cos_sim_accuracy
2751 value: 88.38242713548337
2752 - type: cos_sim_ap
2753 value: 84.70041255196017
2754 - type: cos_sim_f1
2755 value: 77.13222561986515
2756 - type: cos_sim_precision
2757 value: 73.95266690215472
2758 - type: cos_sim_recall
2759 value: 80.59747459193102
2760 - type: dot_accuracy
2761 value: 88.38242713548337
2762 - type: dot_ap
2763 value: 84.7004118720222
2764 - type: dot_f1
2765 value: 77.13222561986515
2766 - type: dot_precision
2767 value: 73.95266690215472
2768 - type: dot_recall
2769 value: 80.59747459193102
2770 - type: euclidean_accuracy
2771 value: 88.38242713548337
2772 - type: euclidean_ap
2773 value: 84.70041593996575
2774 - type: euclidean_f1
2775 value: 77.13222561986515
2776 - type: euclidean_precision
2777 value: 73.95266690215472
2778 - type: euclidean_recall
2779 value: 80.59747459193102
2780 - type: manhattan_accuracy
2781 value: 88.36108200411378
2782 - type: manhattan_ap
2783 value: 84.66897701572054
2784 - type: manhattan_f1
2785 value: 77.00707640360645
2786 - type: manhattan_precision
2787 value: 72.17695778062082
2788 - type: manhattan_recall
2789 value: 82.53002771789343
2790 - type: max_accuracy
2791 value: 88.38242713548337
2792 - type: max_ap
2793 value: 84.70041593996575
2794 - type: max_f1
2795 value: 77.13222561986515
2796 - task:
2797 type: Clustering
2798 dataset:
2799 type: jinaai/cities_wiki_clustering
2800 name: MTEB WikiCitiesClustering
2801 config: default
2802 split: test
2803 revision: ddc9ee9242fa65332597f70e967ecc38b9d734fa
2804 metrics:
2805 - type: v_measure
2806 value: 81.46426354153643
2807 ---
2808 <h1 align="center">Snowflake's Arctic-embed-xs</h1>
2809 <h4 align="center">
2810 <p>
2811 <a href=#news>News</a> |
2812 <a href=#models>Models</a> |
2813 <a href=#usage>Usage</a> |
2814 <a href="#evaluation">Evaluation</a> |
2815 <a href="#contact">Contact</a> |
2816 <a href="#faq">FAQ</a>
2817 <a href="#license">License</a> |
2818 <a href="#acknowledgement">Acknowledgement</a>
2819 <p>
2820 </h4>
2821
2822
2823 ## News
2824
2825 12/04/2024: Release of [snowflake-arctic-embed-l-v2.0](https://huggingface.co/Snowflake/snowflake-arctic-embed-l-v2.0) and [snowflake-arctic-embed-m-v2.0](https://huggingface.co/Snowflake/snowflake-arctic-embed-m-v2.0) our newest models with multilingual workloads in mind. These models outperform prior versions of Arctic Embed and we suggest these replace prior versions!
2826
2827 07/26/2024: Release preprint [[2407.18887] Embedding And Clustering Your Data Can Improve Contrastive Pretraining](https://arxiv.org/abs/2407.18887) on arXiv.
2828
2829 07/18/2024: Release of `snowflake-arctic-embed-m-v1.5`, capable of producing highly compressible embedding vectors that preserve quality even when squished as small as 128 bytes per vector. Details about the development of this model are available in the [launch post on the Snowflake engineering blog](https://www.snowflake.com/engineering-blog/arctic-embed-m-v1-5-enterprise-retrieval/).
2830
2831 05/10/2024: Release the [technical report on Arctic Embed](https://arxiv.org/abs/2405.05374)
2832
2833 04/16/2024: Release the ** snowflake-arctic-embed ** family of text embedding models. The releases are state-of-the-art for Retrieval quality at each of their representative size profiles. [Technical Report]() is coming shortly. For more details, please refer to our Github: [Arctic-Text-Embed](https://github.com/Snowflake-Labs/arctic-embed).
2834
2835
2836 ## Models
2837
2838
2839 snowflake-arctic-embed is a suite of text embedding models that focuses on creating high-quality retrieval models optimized for performance.
2840
2841
2842 The `snowflake-arctic-embedding` models achieve **state-of-the-art performance on the MTEB/BEIR leaderboard** for each of their size variants. Evaluation is performed using these [scripts](https://github.com/Snowflake-Labs/snowflake-arctic-embed/tree/main/src). As shown below, each class of model size achieves SOTA retrieval accuracy compared to other top models.
2843
2844
2845 The models are trained by leveraging existing open-source text representation models, such as bert-base-uncased, and are trained in a multi-stage pipeline to optimize their retrieval performance. First, the models are trained with large batches of query-document pairs where negatives are derived in-batch—pretraining leverages about 400m samples of a mix of public datasets and proprietary web search data. Following pretraining models are further optimized with long training on a smaller dataset (about 1m samples) of triplets of query, positive document, and negative document derived from hard harmful mining. Mining of the negatives and data curation is crucial to retrieval accuracy. A detailed technical report can be found [here](https://arxiv.org/abs/2405.05374).
2846
2847
2848 | Name | MTEB Retrieval Score (NDCG @ 10) | Parameters (Millions) | Embedding Dimension |
2849 | ----------------------------------------------------------------------- | -------------------------------- | --------------------- | ------------------- |
2850 | [snowflake-arctic-embed-xs](https://huggingface.co/Snowflake/snowflake-arctic-embed-xs/) | 50.15 | 22 | 384 |
2851 | [snowflake-arctic-embed-s](https://huggingface.co/Snowflake/snowflake-arctic-embed-s/) | 51.98 | 33 | 384 |
2852 | [snowflake-arctic-embed-m](https://huggingface.co/Snowflake/snowflake-arctic-embed-m/) | 54.90 | 110 | 768 |
2853 | [snowflake-arctic-embed-m-long](https://huggingface.co/Snowflake/snowflake-arctic-embed-m-long/) | 54.83 | 137 | 768 |
2854 | [snowflake-arctic-embed-l](https://huggingface.co/Snowflake/snowflake-arctic-embed-l/) | 55.98 | 335 | 1024 |
2855
2856
2857 Aside from being great open-source models, the largest model, [snowflake-arctic-embed-l](https://huggingface.co/Snowflake/snowflake-arctic-embed-l/), can serve as a natural replacement for closed-source embedding, as shown below.
2858
2859
2860 | Model Name | MTEB Retrieval Score (NDCG @ 10) |
2861 | ------------------------------------------------------------------ | -------------------------------- |
2862 | [snowflake-arctic-embed-l](https://huggingface.co/Snowflake/snowflake-arctic-embed-l/) | 55.98 |
2863 | Google-gecko-text-embedding | 55.7 |
2864 | text-embedding-3-large | 55.44 |
2865 | Cohere-embed-english-v3.0 | 55.00 |
2866 | bge-large-en-v1.5 | 54.29 |
2867
2868
2869 ### [snowflake-arctic-embed-xs](https://huggingface.co/Snowflake/snowflake-arctic-embed-xs)
2870
2871
2872 This tiny model packs quite the punch. Based on the [all-MiniLM-L6-v2](https://huggingface.co/sentence-transformers/all-MiniLM-L6-v2) model with only 22m parameters and 384 dimensions, this model should meet even the strictest latency/TCO budgets. Despite its size, its retrieval accuracy is closer to that of models with 100m paramers.
2873
2874
2875 | Model Name | MTEB Retrieval Score (NDCG @ 10) |
2876 | ------------------------------------------------------------------- | -------------------------------- |
2877 | [snowflake-arctic-embed-xs](https://huggingface.co/Snowflake/snowflake-arctic-embed-xs/) | 50.15 |
2878 | GIST-all-MiniLM-L6-v2 | 45.12 |
2879 | gte-tiny | 44.92 |
2880 | all-MiniLM-L6-v2 | 41.95 |
2881 | bge-micro-v2 | 42.56 |
2882
2883
2884 ### [snowflake-arctic-embed-s](https://huggingface.co/Snowflake/snowflake-arctic-embed-s)
2885
2886
2887 Based on the [infloat/e5-small-unsupervised](https://huggingface.co/intfloat/e5-small-unsupervised) model, this small model does not trade off retrieval accuracy for its small size. With only 33m parameters and 384 dimensions, this model should easily allow scaling to large datasets.
2888
2889
2890 | Model Name | MTEB Retrieval Score (NDCG @ 10) |
2891 | ------------------------------------------------------------------ | -------------------------------- |
2892 | [snowflake-arctic-embed-s](https://huggingface.co/Snowflake/snowflake-arctic-embed-s/) | 51.98 |
2893 | bge-small-en-v1.5 | 51.68 |
2894 | Cohere-embed-english-light-v3.0 | 51.34 |
2895 | text-embedding-3-small | 51.08 |
2896 | e5-small-v2 | 49.04 |
2897
2898
2899 ### [snowflake-arctic-embed-m](https://huggingface.co/Snowflake/snowflake-arctic-embed-m/)
2900
2901
2902 Based on the [intfloat/e5-base-unsupervised](https://huggingface.co/intfloat/e5-base-unsupervised) model, this medium model is the workhorse that provides the best retrieval performance without slowing down inference.
2903
2904
2905 | Model Name | MTEB Retrieval Score (NDCG @ 10) |
2906 | ------------------------------------------------------------------ | -------------------------------- |
2907 | [snowflake-arctic-embed-m](https://huggingface.co/Snowflake/snowflake-arctic-embed-m/) | 54.90 |
2908 | bge-base-en-v1.5 | 53.25 |
2909 | nomic-embed-text-v1.5 | 53.25 |
2910 | GIST-Embedding-v0 | 52.31 |
2911 | gte-base | 52.31 |
2912
2913 ### [snowflake-arctic-embed-m-long](https://huggingface.co/Snowflake/snowflake-arctic-embed-m-long/)
2914
2915
2916 Based on the [nomic-embed-text-v1-unsupervised](https://huggingface.co/nomic-ai/nomic-embed-text-v1-unsupervised) model, this long-context variant of our medium-sized model is perfect for workloads that can be constrained by the regular 512 token context of our other models. Without the use of RPE, this model supports up to 2048 tokens. With RPE, it can scale to 8192!
2917
2918
2919 | Model Name | MTEB Retrieval Score (NDCG @ 10) |
2920 | ------------------------------------------------------------------ | -------------------------------- |
2921 | [snowflake-arctic-embed-m-long](https://huggingface.co/Snowflake/snowflake-arctic-embed-m-long/) | 54.83 |
2922 | nomic-embed-text-v1.5 | 53.01 |
2923 | nomic-embed-text-v1 | 52.81 |
2924
2925
2926
2927
2928 ### [snowflake-arctic-embed-l](https://huggingface.co/Snowflake/snowflake-arctic-embed-l/)
2929
2930
2931 Based on the [intfloat/e5-large-unsupervised](https://huggingface.co/intfloat/e5-large-unsupervised) model, this large model is a direct drop-in for closed APIs and delivers the most accurate retrieval experience.
2932
2933
2934 | Model Name | MTEB Retrieval Score (NDCG @ 10) |
2935 | ------------------------------------------------------------------ | -------------------------------- |
2936 | [snowflake-arctic-embed-l](https://huggingface.co/Snowflake/snowflake-arctic-embed-l/) | 55.98 |
2937 | UAE-Large-V1 | 54.66 |
2938 | bge-large-en-v1.5 | 54.29 |
2939 | mxbai-embed-large-v1 | 54.39 |
2940 | e5-Large-v2 | 50.56 |
2941
2942
2943 ## Usage
2944
2945 ### Using Sentence Transformers
2946
2947 You can use the sentence-transformers package to use an snowflake-arctic-embed model, as shown below.
2948
2949 ```python
2950 from sentence_transformers import SentenceTransformer
2951
2952 model = SentenceTransformer("Snowflake/snowflake-arctic-embed-xs")
2953
2954 queries = ['what is snowflake?', 'Where can I get the best tacos?']
2955 documents = ['The Data Cloud!', 'Mexico City of Course!']
2956
2957 query_embeddings = model.encode(queries, prompt_name="query")
2958 document_embeddings = model.encode(documents)
2959
2960 scores = query_embeddings @ document_embeddings.T
2961 for query, query_scores in zip(queries, scores):
2962 doc_score_pairs = list(zip(documents, query_scores))
2963 doc_score_pairs = sorted(doc_score_pairs, key=lambda x: x[1], reverse=True)
2964 # Output passages & scores
2965 print("Query:", query)
2966 for document, score in doc_score_pairs:
2967 print(score, document)
2968 ```
2969 ```
2970 Query: what is snowflake?
2971 0.57515126 The Data Cloud!
2972 0.45798576 Mexico City of Course!
2973 Query: Where can I get the best tacos?
2974 0.5636022 Mexico City of Course!
2975 0.5044898 The Data Cloud!
2976 ```
2977
2978 ### Using Huggingface transformers
2979
2980
2981 You can use the transformers package for a snowflake-arctic-embed model, as shown below. For optimal retrieval quality, use the CLS token to embed each text portion and use the query prefix below (just on the query).
2982
2983
2984
2985 ```python
2986 import torch
2987 from transformers import AutoModel, AutoTokenizer
2988
2989 tokenizer = AutoTokenizer.from_pretrained('Snowflake/snowflake-arctic-embed-xs')
2990 model = AutoModel.from_pretrained('Snowflake/snowflake-arctic-embed-xs', add_pooling_layer=False)
2991 model.eval()
2992
2993 query_prefix = 'Represent this sentence for searching relevant passages: '
2994 queries = ['what is snowflake?', 'Where can I get the best tacos?']
2995 queries_with_prefix = ["{}{}".format(query_prefix, i) for i in queries]
2996 query_tokens = tokenizer(queries_with_prefix, padding=True, truncation=True, return_tensors='pt', max_length=512)
2997
2998 documents = ['The Data Cloud!', 'Mexico City of Course!']
2999 document_tokens = tokenizer(documents, padding=True, truncation=True, return_tensors='pt', max_length=512)
3000
3001 # Compute token embeddings
3002 with torch.no_grad():
3003 query_embeddings = model(**query_tokens)[0][:, 0]
3004 document_embeddings = model(**document_tokens)[0][:, 0]
3005
3006
3007 # normalize embeddings
3008 query_embeddings = torch.nn.functional.normalize(query_embeddings, p=2, dim=1)
3009 document_embeddings = torch.nn.functional.normalize(document_embeddings, p=2, dim=1)
3010
3011 scores = torch.mm(query_embeddings, document_embeddings.transpose(0, 1))
3012 for query, query_scores in zip(queries, scores):
3013 doc_score_pairs = list(zip(documents, query_scores))
3014 doc_score_pairs = sorted(doc_score_pairs, key=lambda x: x[1], reverse=True)
3015 #Output passages & scores
3016 print("Query:", query)
3017 for document, score in doc_score_pairs:
3018 print(score, document)
3019 ```
3020
3021 ### Using Transformers.js
3022
3023 If you haven't already, you can install the [Transformers.js](https://huggingface.co/docs/transformers.js) JavaScript library from [NPM](https://www.npmjs.com/package/@xenova/transformers) by running:
3024 ```bash
3025 npm i @xenova/transformers
3026 ```
3027
3028 You can then use the model to compute embeddings as follows:
3029
3030 ```js
3031 import { pipeline, dot } from '@xenova/transformers';
3032
3033 // Create feature extraction pipeline
3034 const extractor = await pipeline('feature-extraction', 'Snowflake/snowflake-arctic-embed-xs', {
3035 quantized: false, // Comment out this line to use the quantized version
3036 });
3037
3038 // Generate sentence embeddings
3039 const sentences = [
3040 'Represent this sentence for searching relevant passages: Where can I get the best tacos?',
3041 'The Data Cloud!',
3042 'Mexico City of Course!',
3043 ]
3044 const output = await extractor(sentences, { normalize: true, pooling: 'cls' });
3045
3046 // Compute similarity scores
3047 const [source_embeddings, ...document_embeddings ] = output.tolist();
3048 const similarities = document_embeddings.map(x => dot(source_embeddings, x));
3049 console.log(similarities); // [0.5044895661144148, 0.5636021124426508]
3050 ```
3051
3052 ## FAQ
3053
3054
3055 TBD
3056
3057
3058 ## Contact
3059
3060
3061 Feel free to open an issue or pull request if you have any questions or suggestions about this project.
3062 You also can email Daniel Campos(daniel.campos@snowflake.com).
3063
3064
3065 ## License
3066
3067
3068 Arctic is licensed under the [Apache-2](https://www.apache.org/licenses/LICENSE-2.0). The released models can be used for commercial purposes free of charge.
3069
3070
3071 ## Acknowledgement
3072
3073
3074 We want to thank the open-source community, which has provided the great building blocks upon which we could make our models.
3075 We thank our modeling engineers, Danmei Xu, Luke Merrick, Gaurav Nuti, and Daniel Campos, for making these great models possible.
3076 We thank our leadership, Himabindu Pucha, Kelvin So, Vivek Raghunathan, and Sridhar Ramaswamy, for supporting this work.
3077 We also thank the open-source community for producing the great models we could build on top of and making these releases possible.
3078 Finally, we thank the researchers who created BEIR and MTEB benchmarks.
3079 It is largely thanks to their tireless work to define what better looks like that we could improve model performance.
3080
3081 <img referrerpolicy="no-referrer-when-downgrade" src="https://static.scarf.sh/a.png?x-pxid=15cd6ef8-397b-4e85-9d74-27ebdc7e9765" />