bartowski commited on
Commit
a37edc2
1 Parent(s): d45986a

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +5 -672
README.md CHANGED
@@ -1,682 +1,11 @@
1
  ---
2
  quantized_by: bartowski
3
  pipeline_tag: text-generation
4
- language:
5
- - en
6
- - ar
7
- - cs
8
- - de
9
- - es
10
- - fr
11
- - it
12
- - ja
13
- - ko
14
- - nl
15
- - pt
16
- - zh
17
- tags:
18
- - language
19
- - granite
20
- - embeddings
21
- - multilingual
22
- license: apache-2.0
23
- base_model: ibm-granite/granite-embedding-107m-multilingual
24
- model-index:
25
- - name: ibm-granite/granite-embedding-107m-multilingual
26
- results:
27
- - task:
28
- type: Retrieval
29
- dataset:
30
- name: Miracl (en)
31
- type: miracl/mmteb-miracl
32
- config: en
33
- split: dev
34
- metrics:
35
- - type: ndcg_at_1
36
- value: 0.41176
37
- - type: ndcg_at_10
38
- value: 0.46682
39
- - type: ndcg_at_100
40
- value: 0.54326
41
- - type: ndcg_at_1000
42
- value: 0.56567
43
- - type: ndcg_at_20
44
- value: 0.50157
45
- - type: ndcg_at_3
46
- value: 0.41197
47
- - type: ndcg_at_5
48
- value: 0.42086
49
- - type: recall_at_1
50
- value: 0.19322
51
- - type: recall_at_10
52
- value: 0.57721
53
- - type: recall_at_100
54
- value: 0.83256
55
- - type: recall_at_1000
56
- value: 0.95511
57
- - type: recall_at_20
58
- value: 0.6757
59
- - type: recall_at_3
60
- value: 0.37171
61
- - type: recall_at_5
62
- value: 0.44695
63
- - task:
64
- type: Retrieval
65
- dataset:
66
- name: Miracl (ar)
67
- type: miracl/mmteb-miracl
68
- config: ar
69
- split: dev
70
- metrics:
71
- - type: ndcg_at_1
72
- value: 0.55559
73
- - type: ndcg_at_10
74
- value: 0.62541
75
- - type: ndcg_at_100
76
- value: 0.67101
77
- - type: ndcg_at_1000
78
- value: 0.6805
79
- - type: ndcg_at_20
80
- value: 0.64739
81
- - type: ndcg_at_3
82
- value: 0.56439
83
- - type: ndcg_at_5
84
- value: 0.59347
85
- - type: recall_at_1
86
- value: 0.37009
87
- - type: recall_at_10
88
- value: 0.73317
89
- - type: recall_at_100
90
- value: 0.90066
91
- - type: recall_at_1000
92
- value: 0.96272
93
- - type: recall_at_20
94
- value: 0.80205
95
- - type: recall_at_3
96
- value: 0.56903
97
- - type: recall_at_5
98
- value: 0.6518
99
- - task:
100
- type: Retrieval
101
- dataset:
102
- name: Miracl (bn)
103
- type: miracl/mmteb-miracl
104
- config: bn
105
- split: dev
106
- metrics:
107
- - type: ndcg_at_1
108
- value: 0.56691
109
- - type: ndcg_at_10
110
- value: 0.65484
111
- - type: ndcg_at_100
112
- value: 0.70142
113
- - type: ndcg_at_1000
114
- value: 0.70994
115
- - type: ndcg_at_20
116
- value: 0.67838
117
- - type: ndcg_at_3
118
- value: 0.5988
119
- - type: ndcg_at_5
120
- value: 0.62718
121
- - type: recall_at_1
122
- value: 0.3605
123
- - type: recall_at_10
124
- value: 0.76854
125
- - type: recall_at_100
126
- value: 0.9285
127
- - type: recall_at_1000
128
- value: 0.97928
129
- - type: recall_at_20
130
- value: 0.83667
131
- - type: recall_at_3
132
- value: 0.61596
133
- - type: recall_at_5
134
- value: 0.69766
135
- - task:
136
- type: Retrieval
137
- dataset:
138
- name: Miracl (de)
139
- type: miracl/mmteb-miracl
140
- config: de
141
- split: dev
142
- metrics:
143
- - type: ndcg_at_1
144
- value: 0.41967
145
- - type: ndcg_at_10
146
- value: 0.45141
147
- - type: ndcg_at_100
148
- value: 0.53461
149
- - type: ndcg_at_1000
150
- value: 0.55463
151
- - type: ndcg_at_20
152
- value: 0.49012
153
- - type: ndcg_at_3
154
- value: 0.39486
155
- - type: ndcg_at_5
156
- value: 0.41496
157
- - type: recall_at_1
158
- value: 0.19494
159
- - type: recall_at_10
160
- value: 0.53774
161
- - type: recall_at_100
162
- value: 0.83314
163
- - type: recall_at_1000
164
- value: 0.95045
165
- - type: recall_at_20
166
- value: 0.65659
167
- - type: recall_at_3
168
- value: 0.3556
169
- - type: recall_at_5
170
- value: 0.44448
171
- - task:
172
- type: Retrieval
173
- dataset:
174
- name: Miracl (es)
175
- type: miracl/mmteb-miracl
176
- config: es
177
- split: dev
178
- metrics:
179
- - type: ndcg_at_1
180
- value: 0.54475
181
- - type: ndcg_at_10
182
- value: 0.46593
183
- - type: ndcg_at_100
184
- value: 0.58079
185
- - type: ndcg_at_1000
186
- value: 0.60656
187
- - type: ndcg_at_20
188
- value: 0.51858
189
- - type: ndcg_at_3
190
- value: 0.4578
191
- - type: ndcg_at_5
192
- value: 0.44321
193
- - type: recall_at_1
194
- value: 0.15966
195
- - type: recall_at_10
196
- value: 0.49343
197
- - type: recall_at_100
198
- value: 0.82684
199
- - type: recall_at_1000
200
- value: 0.95299
201
- - type: recall_at_20
202
- value: 0.62367
203
- - type: recall_at_3
204
- value: 0.2949
205
- - type: recall_at_5
206
- value: 0.37983
207
- - task:
208
- type: Retrieval
209
- dataset:
210
- name: Miracl (fa)
211
- type: miracl/mmteb-miracl
212
- config: fa
213
- split: dev
214
- metrics:
215
- - type: ndcg_at_1
216
- value: 0.36709
217
- - type: ndcg_at_10
218
- value: 0.46961
219
- - type: ndcg_at_100
220
- value: 0.53262
221
- - type: ndcg_at_1000
222
- value: 0.55024
223
- - type: ndcg_at_20
224
- value: 0.49892
225
- - type: ndcg_at_3
226
- value: 0.40235
227
- - type: ndcg_at_5
228
- value: 0.42866
229
- - type: recall_at_1
230
- value: 0.22735
231
- - type: recall_at_10
232
- value: 0.59949
233
- - type: recall_at_100
234
- value: 0.83867
235
- - type: recall_at_1000
236
- value: 0.95007
237
- - type: recall_at_20
238
- value: 0.68947
239
- - type: recall_at_3
240
- value: 0.41781
241
- - type: recall_at_5
242
- value: 0.49374
243
- - task:
244
- type: Retrieval
245
- dataset:
246
- name: Miracl (fi)
247
- type: miracl/mmteb-miracl
248
- config: fi
249
- split: dev
250
- metrics:
251
- - type: ndcg_at_1
252
- value: 0.59245
253
- - type: ndcg_at_10
254
- value: 0.65551
255
- - type: ndcg_at_100
256
- value: 0.6967
257
- - type: ndcg_at_1000
258
- value: 0.70521
259
- - type: ndcg_at_20
260
- value: 0.67552
261
- - type: ndcg_at_3
262
- value: 0.58876
263
- - type: ndcg_at_5
264
- value: 0.61779
265
- - type: recall_at_1
266
- value: 0.37669
267
- - type: recall_at_10
268
- value: 0.76529
269
- - type: recall_at_100
270
- value: 0.9156
271
- - type: recall_at_1000
272
- value: 0.96977
273
- - type: recall_at_20
274
- value: 0.82685
275
- - type: recall_at_3
276
- value: 0.60234
277
- - type: recall_at_5
278
- value: 0.67135
279
- - task:
280
- type: Retrieval
281
- dataset:
282
- name: Miracl (fr)
283
- type: miracl/mmteb-miracl
284
- config: fr
285
- split: dev
286
- metrics:
287
- - type: ndcg_at_1
288
- value: 0.38776
289
- - type: ndcg_at_10
290
- value: 0.47589
291
- - type: ndcg_at_100
292
- value: 0.54641
293
- - type: ndcg_at_1000
294
- value: 0.5629
295
- - type: ndcg_at_20
296
- value: 0.51203
297
- - type: ndcg_at_3
298
- value: 0.38924
299
- - type: ndcg_at_5
300
- value: 0.42572
301
- - type: recall_at_1
302
- value: 0.22082
303
- - type: recall_at_10
304
- value: 0.61619
305
- - type: recall_at_100
306
- value: 0.87237
307
- - type: recall_at_1000
308
- value: 0.97449
309
- - type: recall_at_20
310
- value: 0.72689
311
- - type: recall_at_3
312
- value: 0.39527
313
- - type: recall_at_5
314
- value: 0.48983
315
- - task:
316
- type: Retrieval
317
- dataset:
318
- name: Miracl (hi)
319
- type: miracl/mmteb-miracl
320
- config: hi
321
- split: dev
322
- metrics:
323
- - type: ndcg_at_1
324
- value: 0.33143
325
- - type: ndcg_at_10
326
- value: 0.42084
327
- - type: ndcg_at_100
328
- value: 0.48647
329
- - type: ndcg_at_1000
330
- value: 0.50712
331
- - type: ndcg_at_20
332
- value: 0.45399
333
- - type: ndcg_at_3
334
- value: 0.34988
335
- - type: ndcg_at_5
336
- value: 0.37938
337
- - type: recall_at_1
338
- value: 0.17852
339
- - type: recall_at_10
340
- value: 0.55217
341
- - type: recall_at_100
342
- value: 0.79929
343
- - type: recall_at_1000
344
- value: 0.93434
345
- - type: recall_at_20
346
- value: 0.65231
347
- - type: recall_at_3
348
- value: 0.33765
349
- - type: recall_at_5
350
- value: 0.43828
351
- - task:
352
- type: Retrieval
353
- dataset:
354
- name: Miracl (id)
355
- type: miracl/mmteb-miracl
356
- config: id
357
- split: dev
358
- metrics:
359
- - type: ndcg_at_1
360
- value: 0.43854
361
- - type: ndcg_at_10
362
- value: 0.45459
363
- - type: ndcg_at_100
364
- value: 0.53643
365
- - type: ndcg_at_1000
366
- value: 0.56052
367
- - type: ndcg_at_20
368
- value: 0.48795
369
- - type: ndcg_at_3
370
- value: 0.41041
371
- - type: ndcg_at_5
372
- value: 0.42235
373
- - type: recall_at_1
374
- value: 0.19193
375
- - type: recall_at_10
376
- value: 0.5289
377
- - type: recall_at_100
378
- value: 0.79649
379
- - type: recall_at_1000
380
- value: 0.92937
381
- - type: recall_at_20
382
- value: 0.61813
383
- - type: recall_at_3
384
- value: 0.35431
385
- - type: recall_at_5
386
- value: 0.43348
387
- - task:
388
- type: Retrieval
389
- dataset:
390
- name: Miracl (ja)
391
- type: miracl/mmteb-miracl
392
- config: ja
393
- split: dev
394
- metrics:
395
- - type: ndcg_at_1
396
- value: 0.53256
397
- - type: ndcg_at_10
398
- value: 0.59922
399
- - type: ndcg_at_100
400
- value: 0.65407
401
- - type: ndcg_at_1000
402
- value: 0.66484
403
- - type: ndcg_at_20
404
- value: 0.62596
405
- - type: ndcg_at_3
406
- value: 0.53717
407
- - type: ndcg_at_5
408
- value: 0.56523
409
- - type: recall_at_1
410
- value: 0.34555
411
- - type: recall_at_10
412
- value: 0.71476
413
- - type: recall_at_100
414
- value: 0.91152
415
- - type: recall_at_1000
416
- value: 0.97728
417
- - type: recall_at_20
418
- value: 0.79811
419
- - type: recall_at_3
420
- value: 0.53482
421
- - type: recall_at_5
422
- value: 0.62327
423
- - task:
424
- type: Retrieval
425
- dataset:
426
- name: Miracl (ko)
427
- type: miracl/mmteb-miracl
428
- config: ko
429
- split: dev
430
- metrics:
431
- - type: ndcg_at_1
432
- value: 0.5493
433
- - type: ndcg_at_10
434
- value: 0.58413
435
- - type: ndcg_at_100
436
- value: 0.64374
437
- - type: ndcg_at_1000
438
- value: 0.65655
439
- - type: ndcg_at_20
440
- value: 0.61732
441
- - type: ndcg_at_3
442
- value: 0.53068
443
- - type: ndcg_at_5
444
- value: 0.55202
445
- - type: recall_at_1
446
- value: 0.32602
447
- - type: recall_at_10
448
- value: 0.68647
449
- - type: recall_at_100
450
- value: 0.87746
451
- - type: recall_at_1000
452
- value: 0.95524
453
- - type: recall_at_20
454
- value: 0.78089
455
- - type: recall_at_3
456
- value: 0.49173
457
- - type: recall_at_5
458
- value: 0.5827
459
- - task:
460
- type: Retrieval
461
- dataset:
462
- name: Miracl (ru)
463
- type: miracl/mmteb-miracl
464
- config: ru
465
- split: dev
466
- metrics:
467
- - type: ndcg_at_1
468
- value: 0.43131
469
- - type: ndcg_at_10
470
- value: 0.48262
471
- - type: ndcg_at_100
472
- value: 0.56158
473
- - type: ndcg_at_1000
474
- value: 0.57929
475
- - type: ndcg_at_20
476
- value: 0.52023
477
- - type: ndcg_at_3
478
- value: 0.42808
479
- - type: ndcg_at_5
480
- value: 0.44373
481
- - type: recall_at_1
482
- value: 0.22018
483
- - type: recall_at_10
484
- value: 0.58034
485
- - type: recall_at_100
486
- value: 0.84074
487
- - type: recall_at_1000
488
- value: 0.93938
489
- - type: recall_at_20
490
- value: 0.68603
491
- - type: recall_at_3
492
- value: 0.39307
493
- - type: recall_at_5
494
- value: 0.47077
495
- - task:
496
- type: Retrieval
497
- dataset:
498
- name: Miracl (sw)
499
- type: miracl/mmteb-miracl
500
- config: sw
501
- split: dev
502
- metrics:
503
- - type: ndcg_at_1
504
- value: 0.50415
505
- - type: ndcg_at_10
506
- value: 0.59111
507
- - type: ndcg_at_100
508
- value: 0.64312
509
- - type: ndcg_at_1000
510
- value: 0.65089
511
- - type: ndcg_at_20
512
- value: 0.61651
513
- - type: ndcg_at_3
514
- value: 0.5304
515
- - type: ndcg_at_5
516
- value: 0.56139
517
- - type: recall_at_1
518
- value: 0.33267
519
- - type: recall_at_10
520
- value: 0.72082
521
- - type: recall_at_100
522
- value: 0.91377
523
- - type: recall_at_1000
524
- value: 0.96152
525
- - type: recall_at_20
526
- value: 0.79943
527
- - type: recall_at_3
528
- value: 0.5548
529
- - type: recall_at_5
530
- value: 0.64302
531
- - task:
532
- type: Retrieval
533
- dataset:
534
- name: Miracl (te)
535
- type: miracl/mmteb-miracl
536
- config: te
537
- split: dev
538
- metrics:
539
- - type: ndcg_at_1
540
- value: 0.64372
541
- - type: ndcg_at_10
542
- value: 0.78175
543
- - type: ndcg_at_100
544
- value: 0.79523
545
- - type: ndcg_at_1000
546
- value: 0.79774
547
- - type: ndcg_at_20
548
- value: 0.78826
549
- - type: ndcg_at_3
550
- value: 0.74856
551
- - type: ndcg_at_5
552
- value: 0.77128
553
- - type: recall_at_1
554
- value: 0.63688
555
- - type: recall_at_10
556
- value: 0.90358
557
- - type: recall_at_100
558
- value: 0.96558
559
- - type: recall_at_1000
560
- value: 0.9847
561
- - type: recall_at_20
562
- value: 0.92834
563
- - type: recall_at_3
564
- value: 0.81804
565
- - type: recall_at_5
566
- value: 0.87198
567
- - task:
568
- type: Retrieval
569
- dataset:
570
- name: Miracl (th)
571
- type: miracl/mmteb-miracl
572
- config: th
573
- split: dev
574
- metrics:
575
- - type: ndcg_at_1
576
- value: 0.65484
577
- - type: ndcg_at_10
578
- value: 0.71774
579
- - type: ndcg_at_100
580
- value: 0.75362
581
- - type: ndcg_at_1000
582
- value: 0.75898
583
- - type: ndcg_at_20
584
- value: 0.73709
585
- - type: ndcg_at_3
586
- value: 0.66199
587
- - type: ndcg_at_5
588
- value: 0.68451
589
- - type: recall_at_1
590
- value: 0.45911
591
- - type: recall_at_10
592
- value: 0.82619
593
- - type: recall_at_100
594
- value: 0.95515
595
- - type: recall_at_1000
596
- value: 0.98854
597
- - type: recall_at_20
598
- value: 0.88447
599
- - type: recall_at_3
600
- value: 0.67437
601
- - type: recall_at_5
602
- value: 0.73786
603
- - task:
604
- type: Retrieval
605
- dataset:
606
- name: Miracl (yo)
607
- type: miracl/mmteb-miracl
608
- config: yo
609
- split: dev
610
- metrics:
611
- - type: ndcg_at_1
612
- value: 0.46218
613
- - type: ndcg_at_10
614
- value: 0.64685
615
- - type: ndcg_at_100
616
- value: 0.66941
617
- - type: ndcg_at_1000
618
- value: 0.67361
619
- - type: ndcg_at_20
620
- value: 0.65548
621
- - type: ndcg_at_3
622
- value: 0.57609
623
- - type: ndcg_at_5
624
- value: 0.62021
625
- - type: recall_at_1
626
- value: 0.42787
627
- - type: recall_at_10
628
- value: 0.82913
629
- - type: recall_at_100
630
- value: 0.93277
631
- - type: recall_at_1000
632
- value: 0.96499
633
- - type: recall_at_20
634
- value: 0.85994
635
- - type: recall_at_3
636
- value: 0.65406
637
- - type: recall_at_5
638
- value: 0.7542
639
- - task:
640
- type: Retrieval
641
- dataset:
642
- name: Miracl (zh)
643
- type: miracl/mmteb-miracl
644
- config: zh
645
- split: dev
646
- metrics:
647
- - type: ndcg_at_1
648
- value: 0.41985
649
- - type: ndcg_at_10
650
- value: 0.4837
651
- - type: ndcg_at_100
652
- value: 0.55961
653
- - type: ndcg_at_1000
654
- value: 0.5762
655
- - type: ndcg_at_20
656
- value: 0.51595
657
- - type: ndcg_at_3
658
- value: 0.42094
659
- - type: ndcg_at_5
660
- value: 0.44273
661
- - type: recall_at_1
662
- value: 0.21446
663
- - type: recall_at_10
664
- value: 0.59695
665
- - type: recall_at_100
666
- value: 0.87388
667
- - type: recall_at_1000
668
- value: 0.96833
669
- - type: recall_at_20
670
- value: 0.69252
671
- - type: recall_at_3
672
- value: 0.40377
673
- - type: recall_at_5
674
- value: 0.4903
675
  ---
676
 
677
  ## Llamacpp Static Quantizations of granite-embedding-107m-multilingual
678
 
679
- Using <a href="https://github.com/ggerganov/llama.cpp/">llama.cpp</a> release <a href="https://github.com/ggerganov/llama.cpp/releases/tag/b4341">b4341</a> for quantization.
680
 
681
  Original model: https://huggingface.co/ibm-granite/granite-embedding-107m-multilingual
682
 
@@ -686,6 +15,10 @@ Run them in [LM Studio](https://lmstudio.ai/)
686
 
687
  No prompt format found, check original model page
688
 
 
 
 
 
689
  ## Download a file (not the whole branch) from below:
690
 
691
  | Filename | Quant type | File Size | Split | Description |
 
1
  ---
2
  quantized_by: bartowski
3
  pipeline_tag: text-generation
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4
  ---
5
 
6
  ## Llamacpp Static Quantizations of granite-embedding-107m-multilingual
7
 
8
+ Using <a href="https://github.com/ggerganov/llama.cpp/">llama.cpp</a> release <a href="https://github.com/ggerganov/llama.cpp/releases/tag/b4381">b4381</a> for quantization.
9
 
10
  Original model: https://huggingface.co/ibm-granite/granite-embedding-107m-multilingual
11
 
 
15
 
16
  No prompt format found, check original model page
17
 
18
+ ## What's new:
19
+
20
+ Fix tokenizer
21
+
22
  ## Download a file (not the whole branch) from below:
23
 
24
  | Filename | Quant type | File Size | Split | Description |