Hugging Face's logo

antoinelouis
/
colbert-xm

Sentence Similarity
Safetensors
81 languages
colbert-ai
xmod
colbert
passage-retrieval
Eval Results
Model card Files Community
2
antoinelouis commited on
Commit
f406563
•
1 Parent(s): 50cc033

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +339 -0
README.md CHANGED
@@ -11,6 +11,345 @@ tags:
11
  library_name: colbert-ai
12
  base_model: facebook/xmod-base
13
  inference: false
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
14
  language:
15
  - multilingual
16
  - af
 
11
  library_name: colbert-ai
12
  base_model: facebook/xmod-base
13
  inference: false
14
+ model-index:
15
+ - name: colbert-xm
16
+ results:
17
+ - task:
18
+ type: sentence-similarity
19
+ name: Passage Retrieval
20
+ dataset:
21
+ type: unicamp-dl/mmarco
22
+ name: mMARCO-ar
23
+ config: arabic
24
+ split: validation
25
+ metrics:
26
+ - type: recall_at_1000
27
+ name: Recall@1000
28
+ value: 74.8
29
+ - type: recall_at_500
30
+ name: Recall@500
31
+ value: 72.1
32
+ - type: recall_at_100
33
+ name: Recall@100
34
+ value: 60.4
35
+ - type: recall_at_10
36
+ name: Recall@10
37
+ value: 36.5
38
+ - type: mrr_at_10
39
+ name: MRR@10
40
+ value: 19.5
41
+ - task:
42
+ type: sentence-similarity
43
+ name: Passage Retrieval
44
+ dataset:
45
+ type: unicamp-dl/mmarco
46
+ name: mMARCO-de
47
+ config: german
48
+ split: validation
49
+ metrics:
50
+ - type: recall_at_1000
51
+ name: Recall@1000
52
+ value: 86.0
53
+ - type: recall_at_500
54
+ name: Recall@500
55
+ value: 84.1
56
+ - type: recall_at_100
57
+ name: Recall@100
58
+ value: 73.9
59
+ - type: recall_at_10
60
+ name: Recall@10
61
+ value: 49.5
62
+ - type: mrr_at_10
63
+ name: MRR@10
64
+ value: 27.0
65
+ - task:
66
+ type: sentence-similarity
67
+ name: Passage Retrieval
68
+ dataset:
69
+ type: unicamp-dl/mmarco
70
+ name: mMARCO-en
71
+ config: english
72
+ split: validation
73
+ metrics:
74
+ - type: recall_at_1000
75
+ name: Recall@1000
76
+ value: 96.5
77
+ - type: recall_at_500
78
+ name: Recall@500
79
+ value: 95.9
80
+ - type: recall_at_100
81
+ name: Recall@100
82
+ value: 89.3
83
+ - type: recall_at_10
84
+ name: Recall@10
85
+ value: 65.7
86
+ - type: mrr_at_10
87
+ name: MRR@10
88
+ value: 37.2
89
+ - task:
90
+ type: sentence-similarity
91
+ name: Passage Retrieval
92
+ dataset:
93
+ type: unicamp-dl/mmarco
94
+ name: mMARCO-es
95
+ config: spanish
96
+ split: validation
97
+ metrics:
98
+ - type: recall_at_1000
99
+ name: Recall@1000
100
+ value: 88.4
101
+ - type: recall_at_500
102
+ name: Recall@500
103
+ value: 86.8
104
+ - type: recall_at_100
105
+ name: Recall@100
106
+ value: 77.5
107
+ - type: recall_at_10
108
+ name: Recall@10
109
+ value: 52.0
110
+ - type: mrr_at_10
111
+ name: MRR@10
112
+ value: 28.5
113
+ - task:
114
+ type: sentence-similarity
115
+ name: Passage Retrieval
116
+ dataset:
117
+ type: unicamp-dl/mmarco
118
+ name: mMARCO-fr
119
+ config: french
120
+ split: validation
121
+ metrics:
122
+ - type: recall_at_1000
123
+ name: Recall@1000
124
+ value: 87.3
125
+ - type: recall_at_500
126
+ name: Recall@500
127
+ value: 85.7
128
+ - type: recall_at_100
129
+ name: Recall@100
130
+ value: 75.2
131
+ - type: recall_at_10
132
+ name: Recall@10
133
+ value: 49.2
134
+ - type: mrr_at_10
135
+ name: MRR@10
136
+ value: 26.9
137
+ - task:
138
+ type: sentence-similarity
139
+ name: Passage Retrieval
140
+ dataset:
141
+ type: unicamp-dl/mmarco
142
+ name: mMARCO-hi
143
+ config: hindi
144
+ split: validation
145
+ metrics:
146
+ - type: recall_at_1000
147
+ name: Recall@1000
148
+ value: 82.2
149
+ - type: recall_at_500
150
+ name: Recall@500
151
+ value: 79.9
152
+ - type: recall_at_100
153
+ name: Recall@100
154
+ value: 69.8
155
+ - type: recall_at_10
156
+ name: Recall@10
157
+ value: 44.2
158
+ - type: mrr_at_10
159
+ name: MRR@10
160
+ value: 23.8
161
+ - task:
162
+ type: sentence-similarity
163
+ name: Passage Retrieval
164
+ dataset:
165
+ type: unicamp-dl/mmarco
166
+ name: mMARCO-id
167
+ config: indonesian
168
+ split: validation
169
+ metrics:
170
+ - type: recall_at_1000
171
+ name: Recall@1000
172
+ value: 86.7
173
+ - type: recall_at_500
174
+ name: Recall@500
175
+ value: 84.8
176
+ - type: recall_at_100
177
+ name: Recall@100
178
+ value: 74.5
179
+ - type: recall_at_10
180
+ name: Recall@10
181
+ value: 48.3
182
+ - type: mrr_at_10
183
+ name: MRR@10
184
+ value: 26.3
185
+ - task:
186
+ type: sentence-similarity
187
+ name: Passage Retrieval
188
+ dataset:
189
+ type: unicamp-dl/mmarco
190
+ name: mMARCO-it
191
+ config: italian
192
+ split: validation
193
+ metrics:
194
+ - type: recall_at_1000
195
+ name: Recall@1000
196
+ value: 86.1
197
+ - type: recall_at_500
198
+ name: Recall@500
199
+ value: 84.3
200
+ - type: recall_at_100
201
+ name: Recall@100
202
+ value: 74.1
203
+ - type: recall_at_10
204
+ name: Recall@10
205
+ value: 48.2
206
+ - type: mrr_at_10
207
+ name: MRR@10
208
+ value: 26.5
209
+ - task:
210
+ type: sentence-similarity
211
+ name: Passage Retrieval
212
+ dataset:
213
+ type: unicamp-dl/mmarco
214
+ name: mMARCO-ja
215
+ config: japanese
216
+ split: validation
217
+ metrics:
218
+ - type: recall_at_1000
219
+ name: Recall@1000
220
+ value: 83.6
221
+ - type: recall_at_500
222
+ name: Recall@500
223
+ value: 81.8
224
+ - type: recall_at_100
225
+ name: Recall@100
226
+ value: 71.4
227
+ - type: recall_at_10
228
+ name: Recall@10
229
+ value: 44.6
230
+ - type: mrr_at_10
231
+ name: MRR@10
232
+ value: 24.1
233
+ - task:
234
+ type: sentence-similarity
235
+ name: Passage Retrieval
236
+ dataset:
237
+ type: unicamp-dl/mmarco
238
+ name: mMARCO-nl
239
+ config: dutch
240
+ split: validation
241
+ metrics:
242
+ - type: recall_at_1000
243
+ name: Recall@1000
244
+ value: 86.8
245
+ - type: recall_at_500
246
+ name: Recall@500
247
+ value: 85.0
248
+ - type: recall_at_100
249
+ name: Recall@100
250
+ value: 75.2
251
+ - type: recall_at_10
252
+ name: Recall@10
253
+ value: 49.8
254
+ - type: mrr_at_10
255
+ name: MRR@10
256
+ value: 27.5
257
+ - task:
258
+ type: sentence-similarity
259
+ name: Passage Retrieval
260
+ dataset:
261
+ type: unicamp-dl/mmarco
262
+ name: mMARCO-pt
263
+ config: portuguese
264
+ split: validation
265
+ metrics:
266
+ - type: recall_at_1000
267
+ name: Recall@1000
268
+ value: 87.1
269
+ - type: recall_at_500
270
+ name: Recall@500
271
+ value: 85.3
272
+ - type: recall_at_100
273
+ name: Recall@100
274
+ value: 75.8
275
+ - type: recall_at_10
276
+ name: Recall@10
277
+ value: 50.5
278
+ - type: mrr_at_10
279
+ name: MRR@10
280
+ value: 27.6
281
+ - task:
282
+ type: sentence-similarity
283
+ name: Passage Retrieval
284
+ dataset:
285
+ type: unicamp-dl/mmarco
286
+ name: mMARCO-ru
287
+ config: russian
288
+ split: validation
289
+ metrics:
290
+ - type: recall_at_1000
291
+ name: Recall@1000
292
+ value: 85.7
293
+ - type: recall_at_500
294
+ name: Recall@500
295
+ value: 83.8
296
+ - type: recall_at_100
297
+ name: Recall@100
298
+ value: 73.6
299
+ - type: recall_at_10
300
+ name: Recall@10
301
+ value: 47.3
302
+ - type: mrr_at_10
303
+ name: MRR@10
304
+ value: 25.1
305
+ - task:
306
+ type: sentence-similarity
307
+ name: Passage Retrieval
308
+ dataset:
309
+ type: unicamp-dl/mmarco
310
+ name: mMARCO-vi
311
+ config: vietnamese
312
+ split: validation
313
+ metrics:
314
+ - type: recall_at_1000
315
+ name: Recall@1000
316
+ value: 81.6
317
+ - type: recall_at_500
318
+ name: Recall@500
319
+ value: 79.0
320
+ - type: recall_at_100
321
+ name: Recall@100
322
+ value: 67.5
323
+ - type: recall_at_10
324
+ name: Recall@10
325
+ value: 42.4
326
+ - type: mrr_at_10
327
+ name: MRR@10
328
+ value: 22.6
329
+ - task:
330
+ type: sentence-similarity
331
+ name: Passage Retrieval
332
+ dataset:
333
+ type: unicamp-dl/mmarco
334
+ name: mMARCO-zh
335
+ config: chinese
336
+ split: validation
337
+ metrics:
338
+ - type: recall_at_1000
339
+ name: Recall@1000
340
+ value: 84.8
341
+ - type: recall_at_500
342
+ name: Recall@500
343
+ value: 83.1
344
+ - type: recall_at_100
345
+ name: Recall@100
346
+ value: 72.2
347
+ - type: recall_at_10
348
+ name: Recall@10
349
+ value: 46.0
350
+ - type: mrr_at_10
351
+ name: MRR@10
352
+ value: 24.6
353
  language:
354
  - multilingual
355
  - af