Hugging Face's logo

Datasets:
bigscience
/
xP3mt

Tasks:
Other
Languages:
Akan
Arabic
Assamese
Multilinguality:
multilingual
Size Categories:
100M<n<1B
Annotations Creators:
expert-generated
crowdsourced
ArXiv:
Tags:
License:
Dataset card Viewer Files Community
Muennighoff commited on
Commit
4df1619
•
1 Parent(s): 27a1238

Create README.md

Browse files
Files changed (1) hide show
  1. README.md +347 -0
README.md ADDED
@@ -0,0 +1,347 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ annotations_creators:
3
+ - expert-generated
4
+ - crowdsourced
5
+ language:
6
+ - ak
7
+ - ar
8
+ - as
9
+ - bm
10
+ - bn
11
+ - ca
12
+ - code
13
+ - en
14
+ - es
15
+ - eu
16
+ - fon
17
+ - fr
18
+ - gu
19
+ - hi
20
+ - id
21
+ - ig
22
+ - ki
23
+ - kn
24
+ - lg
25
+ - ln
26
+ - ml
27
+ - mr
28
+ - ne
29
+ - nso
30
+ - ny
31
+ - or
32
+ - pa
33
+ - pt
34
+ - rn
35
+ - rw
36
+ - sn
37
+ - st
38
+ - sw
39
+ - ta
40
+ - te
41
+ - tn
42
+ - ts
43
+ - tum
44
+ - tw
45
+ - ur
46
+ - vi
47
+ - wo
48
+ - xh
49
+ - yo
50
+ - zh
51
+ - zu
52
+ programming_language:
53
+ - C
54
+ - C++
55
+ - C#
56
+ - Go
57
+ - Java
58
+ - JavaScript
59
+ - Lua
60
+ - PHP
61
+ - Python
62
+ - Ruby
63
+ - Rust
64
+ - Scala
65
+ - TypeScript
66
+ license:
67
+ - apache-2.0
68
+ multilinguality:
69
+ - multilingual
70
+ pretty_name: xP3
71
+ size_categories:
72
+ - 100M<n<1B
73
+ task_categories:
74
+ - other
75
+ ---
76
+
77
+ # Dataset Card for xP3
78
+
79
+ ## Table of Contents
80
+ - [Table of Contents](#table-of-contents)
81
+ - [Dataset Description](#dataset-description)
82
+ - [Dataset Summary](#dataset-summary)
83
+ - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards)
84
+ - [Languages](#languages)
85
+ - [Dataset Structure](#dataset-structure)
86
+ - [Data Instances](#data-instances)
87
+ - [Data Fields](#data-fields)
88
+ - [Data Splits](#data-splits)
89
+ - [Dataset Creation](#dataset-creation)
90
+ - [Curation Rationale](#curation-rationale)
91
+ - [Source Data](#source-data)
92
+ - [Annotations](#annotations)
93
+ - [Additional Information](#additional-information)
94
+ - [Licensing Information](#licensing-information)
95
+ - [Citation Information](#citation-information)
96
+ - [Contributions](#contributions)
97
+
98
+ ## Dataset Description
99
+
100
+ - **Repository:** https://github.com/bigscience-workshop/xmtf
101
+ - **Paper:** [Crosslingual Generalization through Multitask Finetuning](https://arxiv.org/abs/2211.01786)
102
+ - **Point of Contact:** [Niklas Muennighoff](mailto:niklas@hf.co)
103
+
104
+ ### Dataset Summary
105
+
106
+ > xP3 (Crosslingual Public Pool of Prompts) is a collection of prompts & datasets across 46 of languages & 16 NLP tasks. It is used for the training of BLOOMZ and mT0, multilingual language models capable of following human instructions in dozens of languages zero-shot.
107
+
108
+ - **Creation:** The dataset can be recreated using instructions available [here](https://github.com/bigscience-workshop/xmtf#create-xp3). We provide this version to save processing time and ease reproducibility.
109
+ - **Languages:** 46 (Can be extended by [recreating with more splits](https://github.com/bigscience-workshop/xmtf#create-xp3))
110
+ - **xP3 Dataset Family:**
111
+
112
+ <table>
113
+ <tr>
114
+ <th>Name</th>
115
+ <th>Explanation</th>
116
+ <th>Example models</th>
117
+ </tr>
118
+ <tr>
119
+ <td><a href=https://huggingface.co/datasets/bigscience/xP3>xP3</a></t>
120
+ <td>Mixture of 13 training tasks in 46 languages with English prompts</td>
121
+ <td><a href=https://huggingface.co/bigscience/bloomz>bloomz</a> & <a href=https://huggingface.co/bigscience/mt0-xxl>mt0-xxl</a></td>
122
+ </tr>
123
+ <tr>
124
+ <td><a href=https://huggingface.co/datasets/bigscience/xP3mt>xP3mt</a></t>
125
+ <td>Mixture of 13 training tasks in 46 languages with prompts in 20 languages (machine-translated from English)</td>
126
+ <td><a href=https://huggingface.co/bigscience/bloomz-mt>bloomz-mt</a> & <a href=https://huggingface.co/bigscience/mt0-xxl-mt>mt0-xxl-mt</a></td>
127
+ </tr>
128
+ <tr>
129
+ <td><a href=https://huggingface.co/datasets/bigscience/xP3all>xP3all</a></t>
130
+ <td>xP3 + our evaluation datasets adding an additional 3 tasks for a total of 16 tasks in 46 languages with English prompts</td>
131
+ <td></td>
132
+ </tr>
133
+ <tr>
134
+ <td><a href=https://huggingface.co/datasets/bigscience/xP3megds>xP3megds</a></t>
135
+ <td><a href=https://github.com/bigscience-workshop/Megatron-DeepSpeed>Megatron-DeepSpeed</a> processed version of xP3</td>
136
+ <td><a href=https://huggingface.co/bigscience/bloomz>bloomz</a></td>
137
+ </tr>
138
+ <tr>
139
+ <td><a href=https://huggingface.co/datasets/Muennighoff/P3>P3</a></t>
140
+ <td>Repreprocessed version of the English-only <a href=https://huggingface.co/datasets/bigscience/P3>P3</a> with 8 training tasks</td>
141
+ <td><a href=https://huggingface.co/bigscience/bloomz-p3>bloomz-p3</a> & <a href=https://huggingface.co/bigscience/mt0-xxl-p3>mt0-xxl-p3</a></td>
142
+ </tr>
143
+ </table>
144
+
145
+ ## Dataset Structure
146
+
147
+ ### Data Instances
148
+
149
+ An example of "train" looks as follows:
150
+ ```json
151
+ {
152
+ "inputs": "Sentence 1: Fue académico en literatura metafísica, teología y ciencias clásicas.\nSentence 2: Fue académico en literatura metafísica, teología y ciencia clásica. Question: Can we rewrite Sentence 1 to Sentence 2? Yes or No?",
153
+ "targets": "Yes"
154
+ }
155
+ ```
156
+
157
+ ### Data Fields
158
+
159
+ The data fields are the same among all splits:
160
+ - `inputs`: the natural language input fed to the model
161
+ - `targets`: the natural language target that the model has to generate
162
+
163
+ ### Data Splits
164
+
165
+ The below table summarizes sizes per language (computed from the `merged_{lang}.jsonl` files). Due to languages like `tw` only being single sentence translation samples from Flores, their byte percentage is significantly lower than their sample percentage. We machine-translated prompts for monolingual datasets, thus languages with only crosslingual datasets (e.g. Translation) do not have non-English prompts. Languages without non-English prompts are equivalent to [xP3](https://huggingface.co/datasets/bigscience/xP3).
166
+
167
+ |Language|Size|%|Samples|%|Non-English prompts|
168
+ |tw|106288|0.11|265071|0.33||
169
+ |bm|107056|0.11|265180|0.33||
170
+ |ak|108096|0.11|265071|0.33||
171
+ |ca|110608|0.11|271191|0.34||
172
+ |eu|113008|0.12|281199|0.35||
173
+ |fon|113072|0.12|265063|0.33||
174
+ |st|114080|0.12|265063|0.33||
175
+ |ki|115040|0.12|265180|0.33||
176
+ |tum|116032|0.12|265063|0.33||
177
+ |wo|122560|0.13|365063|0.46||
178
+ |ln|126304|0.13|365060|0.46||
179
+ |as|156256|0.16|265063|0.33||
180
+ |or|161472|0.17|265063|0.33||
181
+ |kn|165456|0.17|265063|0.33||
182
+ |ml|175040|0.18|265864|0.33||
183
+ |rn|192992|0.2|318189|0.4||
184
+ |nso|229712|0.24|915051|1.14||
185
+ |tn|235536|0.24|915054|1.14||
186
+ |lg|235936|0.24|915021|1.14||
187
+ |rw|249360|0.26|915043|1.14||
188
+ |ts|250256|0.26|915044|1.14||
189
+ |sn|252496|0.26|865056|1.08||
190
+ |xh|254672|0.26|915058|1.14||
191
+ |zu|263712|0.27|915061|1.14||
192
+ |ny|272128|0.28|915063|1.14||
193
+ |ig|325440|0.33|950097|1.19|✅|
194
+ |yo|339664|0.35|913021|1.14|✅|
195
+ |ne|398144|0.41|315754|0.39|✅|
196
+ |pa|529632|0.55|339210|0.42|✅|
197
+ |sw|561392|0.58|1114439|1.39|✅|
198
+ |gu|566576|0.58|347499|0.43|✅|
199
+ |mr|674000|0.69|417269|0.52|✅|
200
+ |bn|854864|0.88|428725|0.54|✅|
201
+ |ta|943440|0.97|410633|0.51|✅|
202
+ |te|1384016|1.42|573354|0.72|✅|
203
+ |ur|1944416|2.0|855756|1.07|✅|
204
+ |vi|3113184|3.2|1667306|2.08|✅|
205
+ |code|4330752|4.46|2707724|3.38||
206
+ |hi|4469712|4.6|1543441|1.93|✅|
207
+ |id|4538768|4.67|2582272|3.22|✅|
208
+ |zh|4604112|4.74|3571636|4.46|✅|
209
+ |ar|4703968|4.84|2148970|2.68|✅|
210
+ |fr|5558912|5.72|5055942|6.31|✅|
211
+ |pt|6130016|6.31|3562772|4.45|✅|
212
+ |es|7579424|7.8|5151349|6.43|✅|
213
+ |en|39252528|40.4|32740750|40.87||
214
+ |total|97150128|100.0|80100816|100.0|
215
+
216
+ ## Dataset Creation
217
+
218
+ ### Source Data
219
+
220
+ #### Training datasets
221
+
222
+ - Code Miscellaneous
223
+ - [CodeComplex](https://huggingface.co/datasets/codeparrot/codecomplex)
224
+ - [Docstring Corpus](https://huggingface.co/datasets/teven/code_docstring_corpus)
225
+ - [GreatCode](https://huggingface.co/datasets/great_code)
226
+ - [State Changes](https://huggingface.co/datasets/Fraser/python-state-changes)
227
+ - Closed-book QA
228
+ - [Hotpot QA](https://huggingface.co/datasets/hotpot_qa)
229
+ - [Trivia QA](https://huggingface.co/datasets/trivia_qa)
230
+ - [Web Questions](https://huggingface.co/datasets/web_questions)
231
+ - [Wiki QA](https://huggingface.co/datasets/wiki_qa)
232
+ - Extractive QA
233
+ - [Adversarial QA](https://huggingface.co/datasets/adversarial_qa)
234
+ - [CMRC2018](https://huggingface.co/datasets/cmrc2018)
235
+ - [DRCD](https://huggingface.co/datasets/clue)
236
+ - [DuoRC](https://huggingface.co/datasets/duorc)
237
+ - [MLQA](https://huggingface.co/datasets/mlqa)
238
+ - [Quoref](https://huggingface.co/datasets/quoref)
239
+ - [ReCoRD](https://huggingface.co/datasets/super_glue)
240
+ - [ROPES](https://huggingface.co/datasets/ropes)
241
+ - [SQuAD v2](https://huggingface.co/datasets/squad_v2)
242
+ - [xQuAD](https://huggingface.co/datasets/xquad)
243
+ - TyDI QA
244
+ - [Primary](https://huggingface.co/datasets/khalidalt/tydiqa-primary)
245
+ - [Goldp](https://huggingface.co/datasets/khalidalt/tydiqa-goldp)
246
+ - Multiple-Choice QA
247
+ - [ARC](https://huggingface.co/datasets/ai2_arc)
248
+ - [C3](https://huggingface.co/datasets/c3)
249
+ - [CoS-E](https://huggingface.co/datasets/cos_e)
250
+ - [Cosmos](https://huggingface.co/datasets/cosmos)
251
+ - [DREAM](https://huggingface.co/datasets/dream)
252
+ - [MultiRC](https://huggingface.co/datasets/super_glue)
253
+ - [OpenBookQA](https://huggingface.co/datasets/openbookqa)
254
+ - [PiQA](https://huggingface.co/datasets/piqa)
255
+ - [QUAIL](https://huggingface.co/datasets/quail)
256
+ - [QuaRel](https://huggingface.co/datasets/quarel)
257
+ - [QuaRTz](https://huggingface.co/datasets/quartz)
258
+ - [QASC](https://huggingface.co/datasets/qasc)
259
+ - [RACE](https://huggingface.co/datasets/race)
260
+ - [SciQ](https://huggingface.co/datasets/sciq)
261
+ - [Social IQA](https://huggingface.co/datasets/social_i_qa)
262
+ - [Wiki Hop](https://huggingface.co/datasets/wiki_hop)
263
+ - [WiQA](https://huggingface.co/datasets/wiqa)
264
+ - Paraphrase Identification
265
+ - [MRPC](https://huggingface.co/datasets/super_glue)
266
+ - [PAWS](https://huggingface.co/datasets/paws)
267
+ - [PAWS-X](https://huggingface.co/datasets/paws-x)
268
+ - [QQP](https://huggingface.co/datasets/qqp)
269
+ - Program Synthesis
270
+ - [APPS](https://huggingface.co/datasets/codeparrot/apps)
271
+ - [CodeContests](https://huggingface.co/datasets/teven/code_contests)
272
+ - [JupyterCodePairs](https://huggingface.co/datasets/codeparrot/github-jupyter-text-code-pairs)
273
+ - [MBPP](https://huggingface.co/datasets/Muennighoff/mbpp)
274
+ - [NeuralCodeSearch](https://huggingface.co/datasets/neural_code_search)
275
+ - [XLCoST](https://huggingface.co/datasets/codeparrot/xlcost-text-to-code)
276
+ - Structure-to-text
277
+ - [Common Gen](https://huggingface.co/datasets/common_gen)
278
+ - [Wiki Bio](https://huggingface.co/datasets/wiki_bio)
279
+ - Sentiment
280
+ - [Amazon](https://huggingface.co/datasets/amazon_polarity)
281
+ - [App Reviews](https://huggingface.co/datasets/app_reviews)
282
+ - [IMDB](https://huggingface.co/datasets/imdb)
283
+ - [Rotten Tomatoes](https://huggingface.co/datasets/rotten_tomatoes)
284
+ - [Yelp](https://huggingface.co/datasets/yelp_review_full)
285
+ - Simplification
286
+ - [BiSECT](https://huggingface.co/datasets/GEM/BiSECT)
287
+ - Summarization
288
+ - [CNN Daily Mail](https://huggingface.co/datasets/cnn_dailymail)
289
+ - [Gigaword](https://huggingface.co/datasets/gigaword)
290
+ - [MultiNews](https://huggingface.co/datasets/multi_news)
291
+ - [SamSum](https://huggingface.co/datasets/samsum)
292
+ - [Wiki-Lingua](https://huggingface.co/datasets/GEM/wiki_lingua)
293
+ - [XLSum](https://huggingface.co/datasets/GEM/xlsum)
294
+ - [XSum](https://huggingface.co/datasets/xsum)
295
+ - Topic Classification
296
+ - [AG News](https://huggingface.co/datasets/ag_news)
297
+ - [DBPedia](https://huggingface.co/datasets/dbpedia_14)
298
+ - [TNEWS](https://huggingface.co/datasets/clue)
299
+ - [TREC](https://huggingface.co/datasets/trec)
300
+ - [CSL](https://huggingface.co/datasets/clue)
301
+ - Translation
302
+ - [Flores-200](https://huggingface.co/datasets/Muennighoff/flores200)
303
+ - [Tatoeba](https://huggingface.co/datasets/Helsinki-NLP/tatoeba_mt)
304
+ - Word Sense disambiguation
305
+ - [WiC](https://huggingface.co/datasets/super_glue)
306
+ - [XL-WiC](https://huggingface.co/datasets/pasinit/xlwic)
307
+
308
+ #### Evaluation datasets (included in [xP3all](https://huggingface.co/datasets/bigscience/xP3all) except for HumanEval)
309
+
310
+ - Natural Language Inference
311
+ - [ANLI](https://huggingface.co/datasets/anli)
312
+ - [CB](https://huggingface.co/datasets/super_glue)
313
+ - [RTE](https://huggingface.co/datasets/super_glue)
314
+ - [XNLI](https://huggingface.co/datasets/xnli)
315
+ - Coreference Resolution
316
+ - [Winogrande](https://huggingface.co/datasets/winogrande)
317
+ - [XWinograd](https://huggingface.co/datasets/Muennighoff/xwinograd)
318
+ - Program Synthesis
319
+ - [HumanEval](https://huggingface.co/datasets/openai_humaneval)
320
+ - Sentence Completion
321
+ - [COPA](https://huggingface.co/datasets/super_glue)
322
+ - [Story Cloze](https://huggingface.co/datasets/story_cloze)
323
+ - [XCOPA](https://huggingface.co/datasets/xcopa)
324
+ - [XStoryCloze](https://huggingface.co/datasets/Muennighoff/xstory_cloze)
325
+
326
+ ## Additional Information
327
+
328
+ ### Licensing Information
329
+
330
+ The dataset is released under Apache 2.0.
331
+
332
+ ### Citation Information
333
+
334
+ ```bibtex
335
+ @misc{muennighoff2022crosslingual,
336
+ title={Crosslingual Generalization through Multitask Finetuning},
337
+ author={Niklas Muennighoff and Thomas Wang and Lintang Sutawika and Adam Roberts and Stella Biderman and Teven Le Scao and M Saiful Bari and Sheng Shen and Zheng-Xin Yong and Hailey Schoelkopf and Xiangru Tang and Dragomir Radev and Alham Fikri Aji and Khalid Almubarak and Samuel Albanie and Zaid Alyafeai and Albert Webson and Edward Raff and Colin Raffel},
338
+ year={2022},
339
+ eprint={2211.01786},
340
+ archivePrefix={arXiv},
341
+ primaryClass={cs.CL}
342
+ }
343
+ ```
344
+
345
+ ### Contributions
346
+
347
+ Thanks to the contributors of [promptsource](https://github.com/bigscience-workshop/promptsource/graphs/contributors) for adding many prompts used in this dataset.