Hugging Face's logo

Datasets:
mteb
/
multi-hatecheck

Tasks:
Text Classification
Modalities:
Text
Sub-tasks:
sentiment-analysis
sentiment-scoring
sentiment-classification
Languages:
Arabic
Mandarin Chinese
German
Size:
10K - 100K
ArXiv:
Tags:
mteb
text
License:
Dataset card Data Studio Files
xet
Community
Samoed commited on
Commit
c27092b
·
verified ·
1 Parent(s): 8f95949

Add dataset card

Browse files
Files changed (1) hide show
  1. README.md +576 -63
README.md CHANGED
@@ -1,20 +1,26 @@
1
  ---
2
- license: cc-by-4.0
3
- task_categories:
4
- - text-classification
5
  language:
6
  - ara
7
- - por
8
  - eng
 
9
  - fra
 
10
  - ita
11
- - cmn
12
- - spa
13
  - nld
14
- - hin
15
- - deu
 
 
 
16
  size_categories:
17
  - 10K<n<100K
 
 
 
 
18
  configs:
19
  - config_name: default
20
  data_files:
@@ -64,70 +70,577 @@ configs:
64
  data_files:
65
  - path: multi-hatecheck/test/cmn.jsonl.gz
66
  split: test
 
 
 
67
  ---
 
 
 
 
 
 
 
68
 
69
- #### Description
70
- Combines multilingual HateCheck datasets (10 languages, including English), by Paul Roettger and colleagues (2021, 2022).
 
 
71
 
72
- The original English dataset can be found under https://github.com/Paul/hatecheck.
73
- Datasets for other languages are found at:
74
- - https://github.com/Paul/hatecheck-arabic
75
- - https://github.com/Paul/hatecheck-mandarin
76
- - https://github.com/Paul/hatecheck-german
77
- - https://github.com/Paul/hatecheck-french
78
- - https://github.com/Paul/hatecheck-hindi
79
- - https://github.com/Paul/hatecheck-italian
80
- - https://github.com/Paul/hatecheck-dutch
81
- - https://github.com/Paul/hatecheck-portuguese
82
- - https://github.com/Paul/hatecheck-spanish
83
- Make sure to credit the authors and cite relevant papers (see citation below) if you use these datasets.
84
 
85
 
86
- #### Bibtex citation
 
 
 
 
 
 
 
 
 
 
 
87
  ```
 
 
 
 
 
 
 
 
 
 
88
  @inproceedings{rottger-etal-2021-hatecheck,
89
- title = "{H}ate{C}heck: Functional Tests for Hate Speech Detection Models",
90
- author = {R{\"o}ttger, Paul and
91
- Vidgen, Bertie and
92
- Nguyen, Dong and
93
- Waseem, Zeerak and
94
- Margetts, Helen and
95
- Pierrehumbert, Janet},
96
- editor = "Zong, Chengqing and
97
- Xia, Fei and
98
- Li, Wenjie and
99
- Navigli, Roberto",
100
- booktitle = "Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers)",
101
- month = aug,
102
- year = "2021",
103
- address = "Online",
104
- publisher = "Association for Computational Linguistics",
105
- url = "https://aclanthology.org/2021.acl-long.4",
106
- doi = "10.18653/v1/2021.acl-long.4",
107
- pages = "41--58",
108
- abstract = "Detecting online hate is a difficult task that even state-of-the-art models struggle with. Typically, hate speech detection models are evaluated by measuring their performance on held-out test data using metrics such as accuracy and F1 score. However, this approach makes it difficult to identify specific model weak points. It also risks overestimating generalisable model performance due to increasingly well-evidenced systematic gaps and biases in hate speech datasets. To enable more targeted diagnostic insights, we introduce HateCheck, a suite of functional tests for hate speech detection models. We specify 29 model functionalities motivated by a review of previous research and a series of interviews with civil society stakeholders. We craft test cases for each functionality and validate their quality through a structured annotation process. To illustrate HateCheck{'}s utility, we test near-state-of-the-art transformer models as well as two popular commercial models, revealing critical model weaknesses.",
109
  }
110
 
111
  @inproceedings{rottger-etal-2022-multilingual,
112
- title = "Multilingual {H}ate{C}heck: Functional Tests for Multilingual Hate Speech Detection Models",
113
- author = {R{\"o}ttger, Paul and
114
- Seelawi, Haitham and
115
- Nozza, Debora and
116
- Talat, Zeerak and
117
- Vidgen, Bertie},
118
- editor = "Narang, Kanika and
119
- Mostafazadeh Davani, Aida and
120
- Mathias, Lambert and
121
- Vidgen, Bertie and
122
- Talat, Zeerak",
123
- booktitle = "Proceedings of the Sixth Workshop on Online Abuse and Harms (WOAH)",
124
- month = jul,
125
- year = "2022",
126
- address = "Seattle, Washington (Hybrid)",
127
- publisher = "Association for Computational Linguistics",
128
- url = "https://aclanthology.org/2022.woah-1.15",
129
- doi = "10.18653/v1/2022.woah-1.15",
130
- pages = "154--169",
131
- abstract = "Hate speech detection models are typically evaluated on held-out test sets. However, this risks painting an incomplete and potentially misleading picture of model performance because of increasingly well-documented systematic gaps and biases in hate speech datasets. To enable more targeted diagnostic insights, recent research has thus introduced functional tests for hate speech detection models. However, these tests currently only exist for English-language content, which means that they cannot support the development of more effective models in other languages spoken by billions across the world. To help address this issue, we introduce Multilingual HateCheck (MHC), a suite of functional tests for multilingual hate speech detection models. MHC covers 34 functionalities across ten languages, which is more languages than any other hate speech dataset. To illustrate MHC{'}s utility, we train and test a high-performing multilingual hate speech detection model, and reveal critical model weaknesses for monolingual and cross-lingual applications.",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
132
  }
133
  ```
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
+ annotations_creators:
3
+ - expert-annotated
 
4
  language:
5
  - ara
6
+ - cmn
7
  - eng
8
+ - deu
9
  - fra
10
+ - hin
11
  - ita
 
 
12
  - nld
13
+ - pol
14
+ - por
15
+ - spa
16
+ license: cc-by-4.0
17
+ multilinguality: multilingual
18
  size_categories:
19
  - 10K<n<100K
20
+ task_categories:
21
+ - text-classification
22
+ task_ids:
23
+ - Sentiment/Hate speech
24
  configs:
25
  - config_name: default
26
  data_files:
 
70
  data_files:
71
  - path: multi-hatecheck/test/cmn.jsonl.gz
72
  split: test
73
+ tags:
74
+ - mteb
75
+ - text
76
  ---
77
+ <!-- adapted from https://github.com/huggingface/huggingface_hub/blob/v0.30.2/src/huggingface_hub/templates/datasetcard_template.md -->
78
+
79
+ <div align="center" style="padding: 40px 20px; background-color: white; border-radius: 12px; box-shadow: 0 2px 10px rgba(0, 0, 0, 0.05); max-width: 600px; margin: 0 auto;">
80
+ <h1 style="font-size: 3.5rem; color: #1a1a1a; margin: 0 0 20px 0; letter-spacing: 2px; font-weight: 700;">MultiHateClassification</h1>
81
+ <div style="font-size: 1.5rem; color: #4a4a4a; margin-bottom: 5px; font-weight: 300;">An <a href="https://github.com/embeddings-benchmark/mteb" style="color: #2c5282; font-weight: 600; text-decoration: none;" onmouseover="this.style.textDecoration='underline'" onmouseout="this.style.textDecoration='none'">MTEB</a> dataset</div>
82
+ <div style="font-size: 0.9rem; color: #2c5282; margin-top: 10px;">Massive Text Embedding Benchmark</div>
83
+ </div>
84
 
85
+ Hate speech detection dataset with binary
86
+ (hateful vs non-hateful) labels. Includes 25+ distinct types of hate
87
+ and challenging non-hate, and 11 languages.
88
+
89
 
90
+ | | |
91
+ |---------------|---------------------------------------------|
92
+ | Task category | t2c |
93
+ | Domains | Constructed, Written |
94
+ | Reference | https://aclanthology.org/2022.woah-1.15/ |
 
 
 
 
 
 
 
95
 
96
 
97
+ ## How to evaluate on this task
98
+
99
+ You can evaluate an embedding model on this dataset using the following code:
100
+
101
+ ```python
102
+ import mteb
103
+
104
+ task = mteb.get_tasks(["MultiHateClassification"])
105
+ evaluator = mteb.MTEB(task)
106
+
107
+ model = mteb.get_model(YOUR_MODEL)
108
+ evaluator.run(model)
109
  ```
110
+
111
+ <!-- Datasets want link to arxiv in readme to autolink dataset with paper -->
112
+ To learn more about how to run models on `mteb` task check out the [GitHub repitory](https://github.com/embeddings-benchmark/mteb).
113
+
114
+ ## Citation
115
+
116
+ If you use this dataset, please cite the dataset as well as [mteb](https://github.com/embeddings-benchmark/mteb), as this dataset likely includes additional processing as a part of the [MMTEB Contribution](https://github.com/embeddings-benchmark/mteb/tree/main/docs/mmteb).
117
+
118
+ ```bibtex
119
+
120
  @inproceedings{rottger-etal-2021-hatecheck,
121
+ abstract = {Detecting online hate is a difficult task that even state-of-the-art models struggle with. Typically, hate speech detection models are evaluated by measuring their performance on held-out test data using metrics such as accuracy and F1 score. However, this approach makes it difficult to identify specific model weak points. It also risks overestimating generalisable model performance due to increasingly well-evidenced systematic gaps and biases in hate speech datasets. To enable more targeted diagnostic insights, we introduce HateCheck, a suite of functional tests for hate speech detection models. We specify 29 model functionalities motivated by a review of previous research and a series of interviews with civil society stakeholders. We craft test cases for each functionality and validate their quality through a structured annotation process. To illustrate HateCheck{'}s utility, we test near-state-of-the-art transformer models as well as two popular commercial models, revealing critical model weaknesses.},
122
+ address = {Online},
123
+ author = {R{\"o}ttger, Paul and
124
+ Vidgen, Bertie and
125
+ Nguyen, Dong and
126
+ Waseem, Zeerak and
127
+ Margetts, Helen and
128
+ Pierrehumbert, Janet},
129
+ booktitle = {Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers)},
130
+ doi = {10.18653/v1/2021.acl-long.4},
131
+ editor = {Zong, Chengqing and
132
+ Xia, Fei and
133
+ Li, Wenjie and
134
+ Navigli, Roberto},
135
+ month = aug,
136
+ pages = {41--58},
137
+ publisher = {Association for Computational Linguistics},
138
+ title = {{H}ate{C}heck: Functional Tests for Hate Speech Detection Models},
139
+ url = {https://aclanthology.org/2021.acl-long.4},
140
+ year = {2021},
141
  }
142
 
143
  @inproceedings{rottger-etal-2022-multilingual,
144
+ abstract = {Hate speech detection models are typically evaluated on held-out test sets. However, this risks painting an incomplete and potentially misleading picture of model performance because of increasingly well-documented systematic gaps and biases in hate speech datasets. To enable more targeted diagnostic insights, recent research has thus introduced functional tests for hate speech detection models. However, these tests currently only exist for English-language content, which means that they cannot support the development of more effective models in other languages spoken by billions across the world. To help address this issue, we introduce Multilingual HateCheck (MHC), a suite of functional tests for multilingual hate speech detection models. MHC covers 34 functionalities across ten languages, which is more languages than any other hate speech dataset. To illustrate MHC{'}s utility, we train and test a high-performing multilingual hate speech detection model, and reveal critical model weaknesses for monolingual and cross-lingual applications.},
145
+ address = {Seattle, Washington (Hybrid)},
146
+ author = {R{\"o}ttger, Paul and
147
+ Seelawi, Haitham and
148
+ Nozza, Debora and
149
+ Talat, Zeerak and
150
+ Vidgen, Bertie},
151
+ booktitle = {Proceedings of the Sixth Workshop on Online Abuse and Harms (WOAH)},
152
+ doi = {10.18653/v1/2022.woah-1.15},
153
+ editor = {Narang, Kanika and
154
+ Mostafazadeh Davani, Aida and
155
+ Mathias, Lambert and
156
+ Vidgen, Bertie and
157
+ Talat, Zeerak},
158
+ month = jul,
159
+ pages = {154--169},
160
+ publisher = {Association for Computational Linguistics},
161
+ title = {Multilingual {H}ate{C}heck: Functional Tests for Multilingual Hate Speech Detection Models},
162
+ url = {https://aclanthology.org/2022.woah-1.15},
163
+ year = {2022},
164
+ }
165
+
166
+
167
+ @article{enevoldsen2025mmtebmassivemultilingualtext,
168
+ title={MMTEB: Massive Multilingual Text Embedding Benchmark},
169
+ author={Kenneth Enevoldsen and Isaac Chung and Imene Kerboua and Márton Kardos and Ashwin Mathur and David Stap and Jay Gala and Wissam Siblini and Dominik Krzemiński and Genta Indra Winata and Saba Sturua and Saiteja Utpala and Mathieu Ciancone and Marion Schaeffer and Gabriel Sequeira and Diganta Misra and Shreeya Dhakal and Jonathan Rystrøm and Roman Solomatin and Ömer Çağatan and Akash Kundu and Martin Bernstorff and Shitao Xiao and Akshita Sukhlecha and Bhavish Pahwa and Rafał Poświata and Kranthi Kiran GV and Shawon Ashraf and Daniel Auras and Björn Plüster and Jan Philipp Harries and Loïc Magne and Isabelle Mohr and Mariya Hendriksen and Dawei Zhu and Hippolyte Gisserot-Boukhlef and Tom Aarsen and Jan Kostkan and Konrad Wojtasik and Taemin Lee and Marek Šuppa and Crystina Zhang and Roberta Rocca and Mohammed Hamdy and Andrianos Michail and John Yang and Manuel Faysse and Aleksei Vatolin and Nandan Thakur and Manan Dey and Dipam Vasani and Pranjal Chitale and Simone Tedeschi and Nguyen Tai and Artem Snegirev and Michael Günther and Mengzhou Xia and Weijia Shi and Xing Han Lù and Jordan Clive and Gayatri Krishnakumar and Anna Maksimova and Silvan Wehrli and Maria Tikhonova and Henil Panchal and Aleksandr Abramov and Malte Ostendorff and Zheng Liu and Simon Clematide and Lester James Miranda and Alena Fenogenova and Guangyu Song and Ruqiya Bin Safi and Wen-Ding Li and Alessia Borghini and Federico Cassano and Hongjin Su and Jimmy Lin and Howard Yen and Lasse Hansen and Sara Hooker and Chenghao Xiao and Vaibhav Adlakha and Orion Weller and Siva Reddy and Niklas Muennighoff},
170
+ publisher = {arXiv},
171
+ journal={arXiv preprint arXiv:2502.13595},
172
+ year={2025},
173
+ url={https://arxiv.org/abs/2502.13595},
174
+ doi = {10.48550/arXiv.2502.13595},
175
+ }
176
+
177
+ @article{muennighoff2022mteb,
178
+ author = {Muennighoff, Niklas and Tazi, Nouamane and Magne, Lo{\"\i}c and Reimers, Nils},
179
+ title = {MTEB: Massive Text Embedding Benchmark},
180
+ publisher = {arXiv},
181
+ journal={arXiv preprint arXiv:2210.07316},
182
+ year = {2022}
183
+ url = {https://arxiv.org/abs/2210.07316},
184
+ doi = {10.48550/ARXIV.2210.07316},
185
  }
186
  ```
187
+
188
+ # Dataset Statistics
189
+ <details>
190
+ <summary> Dataset Statistics</summary>
191
+
192
+ The following code contains the descriptive statistics from the task. These can also be obtained using:
193
+
194
+ ```python
195
+ import mteb
196
+
197
+ task = mteb.get_task("MultiHateClassification")
198
+
199
+ desc_stats = task.metadata.descriptive_stats
200
+ ```
201
+
202
+ ```json
203
+ {
204
+ "test": {
205
+ "num_samples": 11000,
206
+ "number_of_characters": 502013,
207
+ "number_texts_intersect_with_train": 16,
208
+ "min_text_length": 1,
209
+ "average_text_length": 45.63754545454545,
210
+ "max_text_length": 135,
211
+ "unique_text": 10990,
212
+ "unique_labels": 2,
213
+ "labels": {
214
+ "0": {
215
+ "count": 7661
216
+ },
217
+ "1": {
218
+ "count": 3339
219
+ }
220
+ },
221
+ "hf_subset_descriptive_stats": {
222
+ "ara": {
223
+ "num_samples": 1000,
224
+ "number_of_characters": 33644,
225
+ "number_texts_intersect_with_train": 5,
226
+ "min_text_length": 6,
227
+ "average_text_length": 33.644,
228
+ "max_text_length": 83,
229
+ "unique_text": 994,
230
+ "unique_labels": 2,
231
+ "labels": {
232
+ "0": {
233
+ "count": 699
234
+ },
235
+ "1": {
236
+ "count": 301
237
+ }
238
+ }
239
+ },
240
+ "cmn": {
241
+ "num_samples": 1000,
242
+ "number_of_characters": 14940,
243
+ "number_texts_intersect_with_train": 6,
244
+ "min_text_length": 5,
245
+ "average_text_length": 14.94,
246
+ "max_text_length": 34,
247
+ "unique_text": 999,
248
+ "unique_labels": 2,
249
+ "labels": {
250
+ "1": {
251
+ "count": 327
252
+ },
253
+ "0": {
254
+ "count": 673
255
+ }
256
+ }
257
+ },
258
+ "eng": {
259
+ "num_samples": 1000,
260
+ "number_of_characters": 48378,
261
+ "number_texts_intersect_with_train": 0,
262
+ "min_text_length": 11,
263
+ "average_text_length": 48.378,
264
+ "max_text_length": 100,
265
+ "unique_text": 1000,
266
+ "unique_labels": 2,
267
+ "labels": {
268
+ "0": {
269
+ "count": 686
270
+ },
271
+ "1": {
272
+ "count": 314
273
+ }
274
+ }
275
+ },
276
+ "deu": {
277
+ "num_samples": 1000,
278
+ "number_of_characters": 53350,
279
+ "number_texts_intersect_with_train": 0,
280
+ "min_text_length": 13,
281
+ "average_text_length": 53.35,
282
+ "max_text_length": 118,
283
+ "unique_text": 1000,
284
+ "unique_labels": 2,
285
+ "labels": {
286
+ "1": {
287
+ "count": 300
288
+ },
289
+ "0": {
290
+ "count": 700
291
+ }
292
+ }
293
+ },
294
+ "fra": {
295
+ "num_samples": 1000,
296
+ "number_of_characters": 55169,
297
+ "number_texts_intersect_with_train": 0,
298
+ "min_text_length": 1,
299
+ "average_text_length": 55.169,
300
+ "max_text_length": 135,
301
+ "unique_text": 1000,
302
+ "unique_labels": 2,
303
+ "labels": {
304
+ "0": {
305
+ "count": 699
306
+ },
307
+ "1": {
308
+ "count": 301
309
+ }
310
+ }
311
+ },
312
+ "hin": {
313
+ "num_samples": 1000,
314
+ "number_of_characters": 47262,
315
+ "number_texts_intersect_with_train": 3,
316
+ "min_text_length": 13,
317
+ "average_text_length": 47.262,
318
+ "max_text_length": 130,
319
+ "unique_text": 999,
320
+ "unique_labels": 2,
321
+ "labels": {
322
+ "0": {
323
+ "count": 698
324
+ },
325
+ "1": {
326
+ "count": 302
327
+ }
328
+ }
329
+ },
330
+ "ita": {
331
+ "num_samples": 1000,
332
+ "number_of_characters": 50502,
333
+ "number_texts_intersect_with_train": 1,
334
+ "min_text_length": 8,
335
+ "average_text_length": 50.502,
336
+ "max_text_length": 114,
337
+ "unique_text": 999,
338
+ "unique_labels": 2,
339
+ "labels": {
340
+ "0": {
341
+ "count": 700
342
+ },
343
+ "1": {
344
+ "count": 300
345
+ }
346
+ }
347
+ },
348
+ "nld": {
349
+ "num_samples": 1000,
350
+ "number_of_characters": 53056,
351
+ "number_texts_intersect_with_train": 1,
352
+ "min_text_length": 15,
353
+ "average_text_length": 53.056,
354
+ "max_text_length": 121,
355
+ "unique_text": 999,
356
+ "unique_labels": 2,
357
+ "labels": {
358
+ "1": {
359
+ "count": 298
360
+ },
361
+ "0": {
362
+ "count": 702
363
+ }
364
+ }
365
+ },
366
+ "pol": {
367
+ "num_samples": 1000,
368
+ "number_of_characters": 48907,
369
+ "number_texts_intersect_with_train": 0,
370
+ "min_text_length": 9,
371
+ "average_text_length": 48.907,
372
+ "max_text_length": 109,
373
+ "unique_text": 1000,
374
+ "unique_labels": 2,
375
+ "labels": {
376
+ "1": {
377
+ "count": 298
378
+ },
379
+ "0": {
380
+ "count": 702
381
+ }
382
+ }
383
+ },
384
+ "por": {
385
+ "num_samples": 1000,
386
+ "number_of_characters": 48400,
387
+ "number_texts_intersect_with_train": 0,
388
+ "min_text_length": 9,
389
+ "average_text_length": 48.4,
390
+ "max_text_length": 109,
391
+ "unique_text": 1000,
392
+ "unique_labels": 2,
393
+ "labels": {
394
+ "0": {
395
+ "count": 698
396
+ },
397
+ "1": {
398
+ "count": 302
399
+ }
400
+ }
401
+ },
402
+ "spa": {
403
+ "num_samples": 1000,
404
+ "number_of_characters": 48405,
405
+ "number_texts_intersect_with_train": 0,
406
+ "min_text_length": 11,
407
+ "average_text_length": 48.405,
408
+ "max_text_length": 106,
409
+ "unique_text": 1000,
410
+ "unique_labels": 2,
411
+ "labels": {
412
+ "1": {
413
+ "count": 296
414
+ },
415
+ "0": {
416
+ "count": 704
417
+ }
418
+ }
419
+ }
420
+ }
421
+ },
422
+ "train": {
423
+ "num_samples": 11000,
424
+ "number_of_characters": 505993,
425
+ "number_texts_intersect_with_train": null,
426
+ "min_text_length": 4,
427
+ "average_text_length": 45.99936363636364,
428
+ "max_text_length": 131,
429
+ "unique_text": 10993,
430
+ "unique_labels": 2,
431
+ "labels": {
432
+ "0": {
433
+ "count": 7659
434
+ },
435
+ "1": {
436
+ "count": 3341
437
+ }
438
+ },
439
+ "hf_subset_descriptive_stats": {
440
+ "ara": {
441
+ "num_samples": 1000,
442
+ "number_of_characters": 34165,
443
+ "number_texts_intersect_with_train": null,
444
+ "min_text_length": 9,
445
+ "average_text_length": 34.165,
446
+ "max_text_length": 86,
447
+ "unique_text": 997,
448
+ "unique_labels": 2,
449
+ "labels": {
450
+ "0": {
451
+ "count": 699
452
+ },
453
+ "1": {
454
+ "count": 301
455
+ }
456
+ }
457
+ },
458
+ "cmn": {
459
+ "num_samples": 1000,
460
+ "number_of_characters": 14946,
461
+ "number_texts_intersect_with_train": null,
462
+ "min_text_length": 4,
463
+ "average_text_length": 14.946,
464
+ "max_text_length": 34,
465
+ "unique_text": 996,
466
+ "unique_labels": 2,
467
+ "labels": {
468
+ "1": {
469
+ "count": 331
470
+ },
471
+ "0": {
472
+ "count": 669
473
+ }
474
+ }
475
+ },
476
+ "eng": {
477
+ "num_samples": 1000,
478
+ "number_of_characters": 48134,
479
+ "number_texts_intersect_with_train": null,
480
+ "min_text_length": 13,
481
+ "average_text_length": 48.134,
482
+ "max_text_length": 101,
483
+ "unique_text": 1000,
484
+ "unique_labels": 2,
485
+ "labels": {
486
+ "0": {
487
+ "count": 687
488
+ },
489
+ "1": {
490
+ "count": 313
491
+ }
492
+ }
493
+ },
494
+ "deu": {
495
+ "num_samples": 1000,
496
+ "number_of_characters": 52717,
497
+ "number_texts_intersect_with_train": null,
498
+ "min_text_length": 13,
499
+ "average_text_length": 52.717,
500
+ "max_text_length": 121,
501
+ "unique_text": 1000,
502
+ "unique_labels": 2,
503
+ "labels": {
504
+ "1": {
505
+ "count": 302
506
+ },
507
+ "0": {
508
+ "count": 698
509
+ }
510
+ }
511
+ },
512
+ "fra": {
513
+ "num_samples": 1000,
514
+ "number_of_characters": 56635,
515
+ "number_texts_intersect_with_train": null,
516
+ "min_text_length": 11,
517
+ "average_text_length": 56.635,
518
+ "max_text_length": 131,
519
+ "unique_text": 1000,
520
+ "unique_labels": 2,
521
+ "labels": {
522
+ "0": {
523
+ "count": 700
524
+ },
525
+ "1": {
526
+ "count": 300
527
+ }
528
+ }
529
+ },
530
+ "hin": {
531
+ "num_samples": 1000,
532
+ "number_of_characters": 47185,
533
+ "number_texts_intersect_with_train": null,
534
+ "min_text_length": 15,
535
+ "average_text_length": 47.185,
536
+ "max_text_length": 128,
537
+ "unique_text": 1000,
538
+ "unique_labels": 2,
539
+ "labels": {
540
+ "1": {
541
+ "count": 301
542
+ },
543
+ "0": {
544
+ "count": 699
545
+ }
546
+ }
547
+ },
548
+ "ita": {
549
+ "num_samples": 1000,
550
+ "number_of_characters": 51270,
551
+ "number_texts_intersect_with_train": null,
552
+ "min_text_length": 10,
553
+ "average_text_length": 51.27,
554
+ "max_text_length": 125,
555
+ "unique_text": 1000,
556
+ "unique_labels": 2,
557
+ "labels": {
558
+ "0": {
559
+ "count": 701
560
+ },
561
+ "1": {
562
+ "count": 299
563
+ }
564
+ }
565
+ },
566
+ "nld": {
567
+ "num_samples": 1000,
568
+ "number_of_characters": 53534,
569
+ "number_texts_intersect_with_train": null,
570
+ "min_text_length": 14,
571
+ "average_text_length": 53.534,
572
+ "max_text_length": 122,
573
+ "unique_text": 1000,
574
+ "unique_labels": 2,
575
+ "labels": {
576
+ "0": {
577
+ "count": 699
578
+ },
579
+ "1": {
580
+ "count": 301
581
+ }
582
+ }
583
+ },
584
+ "pol": {
585
+ "num_samples": 1000,
586
+ "number_of_characters": 50010,
587
+ "number_texts_intersect_with_train": null,
588
+ "min_text_length": 13,
589
+ "average_text_length": 50.01,
590
+ "max_text_length": 111,
591
+ "unique_text": 1000,
592
+ "unique_labels": 2,
593
+ "labels": {
594
+ "0": {
595
+ "count": 703
596
+ },
597
+ "1": {
598
+ "count": 297
599
+ }
600
+ }
601
+ },
602
+ "por": {
603
+ "num_samples": 1000,
604
+ "number_of_characters": 49011,
605
+ "number_texts_intersect_with_train": null,
606
+ "min_text_length": 14,
607
+ "average_text_length": 49.011,
608
+ "max_text_length": 108,
609
+ "unique_text": 1000,
610
+ "unique_labels": 2,
611
+ "labels": {
612
+ "0": {
613
+ "count": 701
614
+ },
615
+ "1": {
616
+ "count": 299
617
+ }
618
+ }
619
+ },
620
+ "spa": {
621
+ "num_samples": 1000,
622
+ "number_of_characters": 48386,
623
+ "number_texts_intersect_with_train": null,
624
+ "min_text_length": 13,
625
+ "average_text_length": 48.386,
626
+ "max_text_length": 103,
627
+ "unique_text": 1000,
628
+ "unique_labels": 2,
629
+ "labels": {
630
+ "0": {
631
+ "count": 703
632
+ },
633
+ "1": {
634
+ "count": 297
635
+ }
636
+ }
637
+ }
638
+ }
639
+ }
640
+ }
641
+ ```
642
+
643
+ </details>
644
+
645
+ ---
646
+ *This dataset card was automatically generated using [MTEB](https://github.com/embeddings-benchmark/mteb)*