Datasets:
Convert dataset to Parquet (#3)
Browse files- Convert dataset to Parquet (c4b2c0b4fb719ce456f0d3435a52d522aa7177f4)
- Add bzd data files (fbf2b05544cd2371cda92976f7219c3ef317d4de)
- Add cni data files (005bb9e5c78fcdebd9d7eaa3de57dba7b08bbaf3)
- Add gn data files (6fd0ce310f42e765026c9355ffd84ce6fe99372b)
- Add hch data files (56acd4de2e9ddefb31cd6fd8ed2eae9188edd7df)
- Add nah data files (ed236eba71bc9cbbb2ff3a98b31c9a607983308b)
- Add oto data files (39bc0e04d6ec2f8c1e2ec1ea188e5fc55d40f63a)
- Add quy data files (914f324fdabeb4f6f78ad9cca47e3afb28f80b8c)
- Add shp data files (cf426f329996e3d6e7ce11d71b2988c4f7401f16)
- Add tar data files (5b7333aae06501207852617fdb52e058b03b1cf3)
- Delete loading script (24e96babbc9322448e657fdf63c86a81b3211d46)
- Add all_languages data files (668b53909c205ece201487394833a36ebab47245)
- Delete legacy dataset_infos.json (0b4bf179ffc3fcc8fff712e8fa1e4a9f89c8562a)
- README.md +130 -63
- all_languages/test-00000-of-00001.parquet +3 -0
- all_languages/validation-00000-of-00001.parquet +3 -0
- americas_nli.py +0 -177
- aym/test-00000-of-00001.parquet +3 -0
- aym/validation-00000-of-00001.parquet +3 -0
- bzd/test-00000-of-00001.parquet +3 -0
- bzd/validation-00000-of-00001.parquet +3 -0
- cni/test-00000-of-00001.parquet +3 -0
- cni/validation-00000-of-00001.parquet +3 -0
- dataset_infos.json +0 -1
- gn/test-00000-of-00001.parquet +3 -0
- gn/validation-00000-of-00001.parquet +3 -0
- hch/test-00000-of-00001.parquet +3 -0
- hch/validation-00000-of-00001.parquet +3 -0
- nah/test-00000-of-00001.parquet +3 -0
- nah/validation-00000-of-00001.parquet +3 -0
- oto/test-00000-of-00001.parquet +3 -0
- oto/validation-00000-of-00001.parquet +3 -0
- quy/test-00000-of-00001.parquet +3 -0
- quy/validation-00000-of-00001.parquet +3 -0
- shp/test-00000-of-00001.parquet +3 -0
- shp/validation-00000-of-00001.parquet +3 -0
- tar/test-00000-of-00001.parquet +3 -0
- tar/validation-00000-of-00001.parquet +3 -0
@@ -29,6 +29,30 @@ task_ids:
|
|
29 |
- natural-language-inference
|
30 |
pretty_name: 'AmericasNLI: A NLI Corpus of 10 Indigenous Low-Resource Languages.'
|
31 |
dataset_info:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
32 |
- config_name: aym
|
33 |
features:
|
34 |
- name: premise
|
@@ -44,13 +68,13 @@ dataset_info:
|
|
44 |
'2': contradiction
|
45 |
splits:
|
46 |
- name: validation
|
47 |
-
num_bytes:
|
48 |
num_examples: 743
|
49 |
- name: test
|
50 |
-
num_bytes:
|
51 |
num_examples: 750
|
52 |
-
download_size:
|
53 |
-
dataset_size:
|
54 |
- config_name: bzd
|
55 |
features:
|
56 |
- name: premise
|
@@ -66,13 +90,13 @@ dataset_info:
|
|
66 |
'2': contradiction
|
67 |
splits:
|
68 |
- name: validation
|
69 |
-
num_bytes:
|
70 |
num_examples: 743
|
71 |
- name: test
|
72 |
-
num_bytes:
|
73 |
num_examples: 750
|
74 |
-
download_size:
|
75 |
-
dataset_size:
|
76 |
- config_name: cni
|
77 |
features:
|
78 |
- name: premise
|
@@ -88,13 +112,13 @@ dataset_info:
|
|
88 |
'2': contradiction
|
89 |
splits:
|
90 |
- name: validation
|
91 |
-
num_bytes:
|
92 |
num_examples: 658
|
93 |
- name: test
|
94 |
-
num_bytes:
|
95 |
num_examples: 750
|
96 |
-
download_size:
|
97 |
-
dataset_size:
|
98 |
- config_name: gn
|
99 |
features:
|
100 |
- name: premise
|
@@ -110,13 +134,13 @@ dataset_info:
|
|
110 |
'2': contradiction
|
111 |
splits:
|
112 |
- name: validation
|
113 |
-
num_bytes:
|
114 |
num_examples: 743
|
115 |
- name: test
|
116 |
-
num_bytes:
|
117 |
num_examples: 750
|
118 |
-
download_size:
|
119 |
-
dataset_size:
|
120 |
- config_name: hch
|
121 |
features:
|
122 |
- name: premise
|
@@ -132,13 +156,13 @@ dataset_info:
|
|
132 |
'2': contradiction
|
133 |
splits:
|
134 |
- name: validation
|
135 |
-
num_bytes:
|
136 |
num_examples: 743
|
137 |
- name: test
|
138 |
-
num_bytes:
|
139 |
num_examples: 750
|
140 |
-
download_size:
|
141 |
-
dataset_size:
|
142 |
- config_name: nah
|
143 |
features:
|
144 |
- name: premise
|
@@ -154,13 +178,13 @@ dataset_info:
|
|
154 |
'2': contradiction
|
155 |
splits:
|
156 |
- name: validation
|
157 |
-
num_bytes:
|
158 |
num_examples: 376
|
159 |
- name: test
|
160 |
-
num_bytes:
|
161 |
num_examples: 738
|
162 |
-
download_size:
|
163 |
-
dataset_size:
|
164 |
- config_name: oto
|
165 |
features:
|
166 |
- name: premise
|
@@ -176,13 +200,13 @@ dataset_info:
|
|
176 |
'2': contradiction
|
177 |
splits:
|
178 |
- name: validation
|
179 |
-
num_bytes:
|
180 |
num_examples: 222
|
181 |
- name: test
|
182 |
-
num_bytes:
|
183 |
num_examples: 748
|
184 |
-
download_size:
|
185 |
-
dataset_size:
|
186 |
- config_name: quy
|
187 |
features:
|
188 |
- name: premise
|
@@ -198,13 +222,13 @@ dataset_info:
|
|
198 |
'2': contradiction
|
199 |
splits:
|
200 |
- name: validation
|
201 |
-
num_bytes:
|
202 |
num_examples: 743
|
203 |
- name: test
|
204 |
-
num_bytes:
|
205 |
num_examples: 750
|
206 |
-
download_size:
|
207 |
-
dataset_size:
|
208 |
- config_name: shp
|
209 |
features:
|
210 |
- name: premise
|
@@ -220,13 +244,13 @@ dataset_info:
|
|
220 |
'2': contradiction
|
221 |
splits:
|
222 |
- name: validation
|
223 |
-
num_bytes:
|
224 |
num_examples: 743
|
225 |
- name: test
|
226 |
-
num_bytes:
|
227 |
num_examples: 750
|
228 |
-
download_size:
|
229 |
-
dataset_size:
|
230 |
- config_name: tar
|
231 |
features:
|
232 |
- name: premise
|
@@ -242,37 +266,80 @@ dataset_info:
|
|
242 |
'2': contradiction
|
243 |
splits:
|
244 |
- name: validation
|
245 |
-
num_bytes:
|
246 |
num_examples: 743
|
247 |
- name: test
|
248 |
-
num_bytes:
|
249 |
num_examples: 750
|
250 |
-
download_size:
|
251 |
-
dataset_size:
|
|
|
252 |
- config_name: all_languages
|
253 |
-
|
254 |
-
-
|
255 |
-
|
256 |
-
-
|
257 |
-
|
258 |
-
|
259 |
-
|
260 |
-
-
|
261 |
-
|
262 |
-
|
263 |
-
|
264 |
-
|
265 |
-
|
266 |
-
|
267 |
-
|
268 |
-
-
|
269 |
-
|
270 |
-
|
271 |
-
|
272 |
-
|
273 |
-
|
274 |
-
|
275 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
276 |
---
|
277 |
|
278 |
# Dataset Card for AmericasNLI
|
|
|
29 |
- natural-language-inference
|
30 |
pretty_name: 'AmericasNLI: A NLI Corpus of 10 Indigenous Low-Resource Languages.'
|
31 |
dataset_info:
|
32 |
+
- config_name: all_languages
|
33 |
+
features:
|
34 |
+
- name: language
|
35 |
+
dtype: string
|
36 |
+
- name: premise
|
37 |
+
dtype: string
|
38 |
+
- name: hypothesis
|
39 |
+
dtype: string
|
40 |
+
- name: label
|
41 |
+
dtype:
|
42 |
+
class_label:
|
43 |
+
names:
|
44 |
+
'0': entailment
|
45 |
+
'1': neutral
|
46 |
+
'2': contradiction
|
47 |
+
splits:
|
48 |
+
- name: validation
|
49 |
+
num_bytes: 1129080
|
50 |
+
num_examples: 6457
|
51 |
+
- name: test
|
52 |
+
num_bytes: 1210579
|
53 |
+
num_examples: 7486
|
54 |
+
download_size: 791239
|
55 |
+
dataset_size: 2339659
|
56 |
- config_name: aym
|
57 |
features:
|
58 |
- name: premise
|
|
|
68 |
'2': contradiction
|
69 |
splits:
|
70 |
- name: validation
|
71 |
+
num_bytes: 117530
|
72 |
num_examples: 743
|
73 |
- name: test
|
74 |
+
num_bytes: 115251
|
75 |
num_examples: 750
|
76 |
+
download_size: 87882
|
77 |
+
dataset_size: 232781
|
78 |
- config_name: bzd
|
79 |
features:
|
80 |
- name: premise
|
|
|
90 |
'2': contradiction
|
91 |
splits:
|
92 |
- name: validation
|
93 |
+
num_bytes: 143354
|
94 |
num_examples: 743
|
95 |
- name: test
|
96 |
+
num_bytes: 127676
|
97 |
num_examples: 750
|
98 |
+
download_size: 91039
|
99 |
+
dataset_size: 271030
|
100 |
- config_name: cni
|
101 |
features:
|
102 |
- name: premise
|
|
|
112 |
'2': contradiction
|
113 |
splits:
|
114 |
- name: validation
|
115 |
+
num_bytes: 113256
|
116 |
num_examples: 658
|
117 |
- name: test
|
118 |
+
num_bytes: 116284
|
119 |
num_examples: 750
|
120 |
+
download_size: 78899
|
121 |
+
dataset_size: 229540
|
122 |
- config_name: gn
|
123 |
features:
|
124 |
- name: premise
|
|
|
134 |
'2': contradiction
|
135 |
splits:
|
136 |
- name: validation
|
137 |
+
num_bytes: 115135
|
138 |
num_examples: 743
|
139 |
- name: test
|
140 |
+
num_bytes: 101948
|
141 |
num_examples: 750
|
142 |
+
download_size: 80429
|
143 |
+
dataset_size: 217083
|
144 |
- config_name: hch
|
145 |
features:
|
146 |
- name: premise
|
|
|
156 |
'2': contradiction
|
157 |
splits:
|
158 |
- name: validation
|
159 |
+
num_bytes: 127966
|
160 |
num_examples: 743
|
161 |
- name: test
|
162 |
+
num_bytes: 120857
|
163 |
num_examples: 750
|
164 |
+
download_size: 90748
|
165 |
+
dataset_size: 248823
|
166 |
- config_name: nah
|
167 |
features:
|
168 |
- name: premise
|
|
|
178 |
'2': contradiction
|
179 |
splits:
|
180 |
- name: validation
|
181 |
+
num_bytes: 50741
|
182 |
num_examples: 376
|
183 |
- name: test
|
184 |
+
num_bytes: 102953
|
185 |
num_examples: 738
|
186 |
+
download_size: 56953
|
187 |
+
dataset_size: 153694
|
188 |
- config_name: oto
|
189 |
features:
|
190 |
- name: premise
|
|
|
200 |
'2': contradiction
|
201 |
splits:
|
202 |
- name: validation
|
203 |
+
num_bytes: 27010
|
204 |
num_examples: 222
|
205 |
- name: test
|
206 |
+
num_bytes: 119650
|
207 |
num_examples: 748
|
208 |
+
download_size: 57849
|
209 |
+
dataset_size: 146660
|
210 |
- config_name: quy
|
211 |
features:
|
212 |
- name: premise
|
|
|
222 |
'2': contradiction
|
223 |
splits:
|
224 |
- name: validation
|
225 |
+
num_bytes: 125636
|
226 |
num_examples: 743
|
227 |
- name: test
|
228 |
+
num_bytes: 112750
|
229 |
num_examples: 750
|
230 |
+
download_size: 85673
|
231 |
+
dataset_size: 238386
|
232 |
- config_name: shp
|
233 |
features:
|
234 |
- name: premise
|
|
|
244 |
'2': contradiction
|
245 |
splits:
|
246 |
- name: validation
|
247 |
+
num_bytes: 124500
|
248 |
num_examples: 743
|
249 |
- name: test
|
250 |
+
num_bytes: 118934
|
251 |
num_examples: 750
|
252 |
+
download_size: 85544
|
253 |
+
dataset_size: 243434
|
254 |
- config_name: tar
|
255 |
features:
|
256 |
- name: premise
|
|
|
266 |
'2': contradiction
|
267 |
splits:
|
268 |
- name: validation
|
269 |
+
num_bytes: 139496
|
270 |
num_examples: 743
|
271 |
- name: test
|
272 |
+
num_bytes: 122624
|
273 |
num_examples: 750
|
274 |
+
download_size: 89683
|
275 |
+
dataset_size: 262120
|
276 |
+
configs:
|
277 |
- config_name: all_languages
|
278 |
+
data_files:
|
279 |
+
- split: validation
|
280 |
+
path: all_languages/validation-*
|
281 |
+
- split: test
|
282 |
+
path: all_languages/test-*
|
283 |
+
- config_name: aym
|
284 |
+
data_files:
|
285 |
+
- split: validation
|
286 |
+
path: aym/validation-*
|
287 |
+
- split: test
|
288 |
+
path: aym/test-*
|
289 |
+
- config_name: bzd
|
290 |
+
data_files:
|
291 |
+
- split: validation
|
292 |
+
path: bzd/validation-*
|
293 |
+
- split: test
|
294 |
+
path: bzd/test-*
|
295 |
+
- config_name: cni
|
296 |
+
data_files:
|
297 |
+
- split: validation
|
298 |
+
path: cni/validation-*
|
299 |
+
- split: test
|
300 |
+
path: cni/test-*
|
301 |
+
- config_name: gn
|
302 |
+
data_files:
|
303 |
+
- split: validation
|
304 |
+
path: gn/validation-*
|
305 |
+
- split: test
|
306 |
+
path: gn/test-*
|
307 |
+
- config_name: hch
|
308 |
+
data_files:
|
309 |
+
- split: validation
|
310 |
+
path: hch/validation-*
|
311 |
+
- split: test
|
312 |
+
path: hch/test-*
|
313 |
+
- config_name: nah
|
314 |
+
data_files:
|
315 |
+
- split: validation
|
316 |
+
path: nah/validation-*
|
317 |
+
- split: test
|
318 |
+
path: nah/test-*
|
319 |
+
- config_name: oto
|
320 |
+
data_files:
|
321 |
+
- split: validation
|
322 |
+
path: oto/validation-*
|
323 |
+
- split: test
|
324 |
+
path: oto/test-*
|
325 |
+
- config_name: quy
|
326 |
+
data_files:
|
327 |
+
- split: validation
|
328 |
+
path: quy/validation-*
|
329 |
+
- split: test
|
330 |
+
path: quy/test-*
|
331 |
+
- config_name: shp
|
332 |
+
data_files:
|
333 |
+
- split: validation
|
334 |
+
path: shp/validation-*
|
335 |
+
- split: test
|
336 |
+
path: shp/test-*
|
337 |
+
- config_name: tar
|
338 |
+
data_files:
|
339 |
+
- split: validation
|
340 |
+
path: tar/validation-*
|
341 |
+
- split: test
|
342 |
+
path: tar/test-*
|
343 |
---
|
344 |
|
345 |
# Dataset Card for AmericasNLI
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6197779797f2b60bab6de7520d9bb0a80c1d62016b4401b508d6f3b9bd69b979
|
3 |
+
size 411132
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1dc0f6fa9663309e352ae0d9bbae42939c68edc0e32abe70d02f9184d74124e5
|
3 |
+
size 380107
|
@@ -1,177 +0,0 @@
|
|
1 |
-
# coding=utf-8
|
2 |
-
# Copyright 2020 The TensorFlow Datasets Authors and the HuggingFace Datasets Authors. Licensed under the Apache License, Version 2.0 (the "License");
|
3 |
-
# you may not use this file except in compliance with the License.
|
4 |
-
# You may obtain a copy of the License at
|
5 |
-
#
|
6 |
-
# http://www.apache.org/licenses/LICENSE-2.0
|
7 |
-
#
|
8 |
-
# Unless required by applicable law or agreed to in writing, software
|
9 |
-
# distributed under the License is distributed on an "AS IS" BASIS,
|
10 |
-
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
11 |
-
# See the License for the specific language governing permissions and
|
12 |
-
# limitations under the License.
|
13 |
-
|
14 |
-
# Lint as: python3
|
15 |
-
"""AmericasNLI: A NLI Corpus of 10 Indigenous Low-Resource Languages."""
|
16 |
-
|
17 |
-
|
18 |
-
import csv
|
19 |
-
|
20 |
-
import datasets
|
21 |
-
|
22 |
-
|
23 |
-
_CITATION = """
|
24 |
-
@article{DBLP:journals/corr/abs-2104-08726,
|
25 |
-
author = {Abteen Ebrahimi and
|
26 |
-
Manuel Mager and
|
27 |
-
Arturo Oncevay and
|
28 |
-
Vishrav Chaudhary and
|
29 |
-
Luis Chiruzzo and
|
30 |
-
Angela Fan and
|
31 |
-
John Ortega and
|
32 |
-
Ricardo Ramos and
|
33 |
-
Annette Rios and
|
34 |
-
Ivan Vladimir and
|
35 |
-
Gustavo A. Gim{\'{e}}nez{-}Lugo and
|
36 |
-
Elisabeth Mager and
|
37 |
-
Graham Neubig and
|
38 |
-
Alexis Palmer and
|
39 |
-
Rolando A. Coto Solano and
|
40 |
-
Ngoc Thang Vu and
|
41 |
-
Katharina Kann},
|
42 |
-
title = {AmericasNLI: Evaluating Zero-shot Natural Language Understanding of
|
43 |
-
Pretrained Multilingual Models in Truly Low-resource Languages},
|
44 |
-
journal = {CoRR},
|
45 |
-
volume = {abs/2104.08726},
|
46 |
-
year = {2021},
|
47 |
-
url = {https://arxiv.org/abs/2104.08726},
|
48 |
-
eprinttype = {arXiv},
|
49 |
-
eprint = {2104.08726},
|
50 |
-
timestamp = {Mon, 26 Apr 2021 17:25:10 +0200},
|
51 |
-
biburl = {https://dblp.org/rec/journals/corr/abs-2104-08726.bib},
|
52 |
-
bibsource = {dblp computer science bibliography, https://dblp.org}
|
53 |
-
}
|
54 |
-
"""
|
55 |
-
|
56 |
-
_DESCRIPTION = """\
|
57 |
-
AmericasNLI is an extension of XNLI (Conneau et al., 2018) – a natural language inference (NLI) dataset covering 15 high-resource languages – to 10 low-resource indigenous languages spoken in the Americas: Ashaninka, Aymara, Bribri, Guarani, Nahuatl, Otomi, Quechua, Raramuri, Shipibo-Konibo, and Wixarika. As with MNLI, the goal is to predict textual entailment (does sentence A imply/contradict/neither sentence B) and is a classification task (given two sentences, predict one of three labels).
|
58 |
-
"""
|
59 |
-
|
60 |
-
VERSION = datasets.Version("1.0.0", "")
|
61 |
-
_DEV_DATA_URL = "https://raw.githubusercontent.com/nala-cub/AmericasNLI/main/dev.tsv"
|
62 |
-
_TEST_DATA_URL = "https://raw.githubusercontent.com/nala-cub/AmericasNLI/main/test.tsv"
|
63 |
-
|
64 |
-
_LANGUAGES = ("aym", "bzd", "cni", "gn", "hch", "nah", "oto", "quy", "shp", "tar")
|
65 |
-
|
66 |
-
|
67 |
-
class AmericasNLIConfig(datasets.BuilderConfig):
|
68 |
-
"""BuilderConfig for AmericasNLI."""
|
69 |
-
|
70 |
-
def __init__(self, language: str, languages=None, **kwargs):
|
71 |
-
"""BuilderConfig for AmericasNLI.
|
72 |
-
|
73 |
-
Args:
|
74 |
-
language: One of aym, bzd, cni, gn, hch, nah, oto, quy, shp, tar or all_languages
|
75 |
-
**kwargs: keyword arguments forwarded to super.
|
76 |
-
"""
|
77 |
-
super(AmericasNLIConfig, self).__init__(**kwargs)
|
78 |
-
self.language = language
|
79 |
-
if language != "all_languages":
|
80 |
-
self.languages = [language]
|
81 |
-
else:
|
82 |
-
self.languages = languages if languages is not None else _LANGUAGES
|
83 |
-
|
84 |
-
|
85 |
-
class AmericasNLI(datasets.GeneratorBasedBuilder):
|
86 |
-
"""TODO"""
|
87 |
-
|
88 |
-
VERSION = VERSION
|
89 |
-
BUILDER_CONFIG_CLASS = AmericasNLIConfig
|
90 |
-
BUILDER_CONFIGS = [
|
91 |
-
AmericasNLIConfig(
|
92 |
-
name=lang,
|
93 |
-
language=lang,
|
94 |
-
version=VERSION,
|
95 |
-
description=f"Plain text import of AmericasNLI for the {lang} language",
|
96 |
-
)
|
97 |
-
for lang in _LANGUAGES
|
98 |
-
] + [
|
99 |
-
AmericasNLIConfig(
|
100 |
-
name="all_languages",
|
101 |
-
language="all_languages",
|
102 |
-
version=VERSION,
|
103 |
-
description="Plain text import of AmericasNLI for all languages",
|
104 |
-
)
|
105 |
-
]
|
106 |
-
|
107 |
-
def _info(self):
|
108 |
-
if self.config.language == "all_languages":
|
109 |
-
features = datasets.Features(
|
110 |
-
{
|
111 |
-
"language": datasets.Value("string"),
|
112 |
-
"premise": datasets.Value("string"),
|
113 |
-
"hypothesis": datasets.Value("string"),
|
114 |
-
"label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
|
115 |
-
}
|
116 |
-
)
|
117 |
-
else:
|
118 |
-
features = datasets.Features(
|
119 |
-
{
|
120 |
-
"premise": datasets.Value("string"),
|
121 |
-
"hypothesis": datasets.Value("string"),
|
122 |
-
"label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
|
123 |
-
}
|
124 |
-
)
|
125 |
-
return datasets.DatasetInfo(
|
126 |
-
description=_DESCRIPTION,
|
127 |
-
features=features,
|
128 |
-
# No default supervised_keys (as we have to pass both premise
|
129 |
-
# and hypothesis as input).
|
130 |
-
supervised_keys=None,
|
131 |
-
homepage="https://github.com/nala-cub/AmericasNLI",
|
132 |
-
citation=_CITATION,
|
133 |
-
)
|
134 |
-
|
135 |
-
def _split_generators(self, dl_manager):
|
136 |
-
dl_paths = dl_manager.download(
|
137 |
-
{
|
138 |
-
"dev_data": _DEV_DATA_URL,
|
139 |
-
"test_data": _TEST_DATA_URL,
|
140 |
-
}
|
141 |
-
)
|
142 |
-
return [
|
143 |
-
datasets.SplitGenerator(
|
144 |
-
name=datasets.Split.VALIDATION,
|
145 |
-
gen_kwargs={
|
146 |
-
"filepath": dl_paths["dev_data"],
|
147 |
-
},
|
148 |
-
),
|
149 |
-
datasets.SplitGenerator(
|
150 |
-
name=datasets.Split.TEST,
|
151 |
-
gen_kwargs={
|
152 |
-
"filepath": dl_paths["test_data"],
|
153 |
-
},
|
154 |
-
),
|
155 |
-
]
|
156 |
-
|
157 |
-
def _generate_examples(self, filepath: str):
|
158 |
-
"""This function returns the examples in the raw (text) form."""
|
159 |
-
idx = 0
|
160 |
-
with open(filepath, encoding="utf-8") as f:
|
161 |
-
reader = csv.DictReader(f, delimiter="\t", quoting=csv.QUOTE_NONE)
|
162 |
-
for row in reader:
|
163 |
-
if row["language"] == self.config.language:
|
164 |
-
yield idx, {
|
165 |
-
"premise": row["premise"],
|
166 |
-
"hypothesis": row["hypothesis"],
|
167 |
-
"label": row["label"],
|
168 |
-
}
|
169 |
-
idx += 1
|
170 |
-
elif self.config.language == "all_languages":
|
171 |
-
yield idx, {
|
172 |
-
"language": row["language"],
|
173 |
-
"premise": row["premise"],
|
174 |
-
"hypothesis": row["hypothesis"],
|
175 |
-
"label": row["label"],
|
176 |
-
}
|
177 |
-
idx += 1
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:896a0b3bb572957898fd3816c2784120f51f553d19556bfab4b6f5183de25a21
|
3 |
+
size 43340
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e013d265c28563d845fd8d459b186ab0eca8c3e8ca01d86be0169f5ff5362c9e
|
3 |
+
size 44542
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:dddd941efee902888917a817b0b10175f94fbebbe8be3bf4748100b5b9191c3a
|
3 |
+
size 42857
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:da1d32006763143587ae9696c9490da6c2878b8a517ace51556584c7328e92ee
|
3 |
+
size 48182
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ab5948cebe3927db3e19464ea0aa3bf8a53c87ea9a730526ccb875fa009922c4
|
3 |
+
size 40572
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:56630838e3349201cc9a7095da4144c1a974765c285aea09e4c6dc3e14001459
|
3 |
+
size 38327
|
@@ -1 +0,0 @@
|
|
1 |
-
{"aym": {"description": "AmericasNLI is an extension of XNLI (Conneau et al., 2018) \u2013 a natural language inference (NLI) dataset covering 15 high-resource languages \u2013 to 10 low-resource indigenous languages spoken in the Americas: Ashaninka, Aymara, Bribri, Guarani, Nahuatl, Otomi, Quechua, Raramuri, Shipibo-Konibo, and Wixarika. As with MNLI, the goal is to predict textual entailment (does sentence A imply/contradict/neither sentence B) and is a classification task (given two sentences, predict one of three labels).\n", "citation": "\n@article{DBLP:journals/corr/abs-2104-08726,\n author = {Abteen Ebrahimi and\n Manuel Mager and\n Arturo Oncevay and\n Vishrav Chaudhary and\n Luis Chiruzzo and\n Angela Fan and\n John Ortega and\n Ricardo Ramos and\n Annette Rios and\n Ivan Vladimir and\n Gustavo A. Gim{'{e}}nez{-}Lugo and\n Elisabeth Mager and\n Graham Neubig and\n Alexis Palmer and\n Rolando A. Coto Solano and\n Ngoc Thang Vu and\n Katharina Kann},\n title = {AmericasNLI: Evaluating Zero-shot Natural Language Understanding of\n Pretrained Multilingual Models in Truly Low-resource Languages},\n journal = {CoRR},\n volume = {abs/2104.08726},\n year = {2021},\n url = {https://arxiv.org/abs/2104.08726},\n eprinttype = {arXiv},\n eprint = {2104.08726},\n timestamp = {Mon, 26 Apr 2021 17:25:10 +0200},\n biburl = {https://dblp.org/rec/journals/corr/abs-2104-08726.bib},\n bibsource = {dblp computer science bibliography, https://dblp.org}\n}\n", "homepage": "TODO", "license": "", "features": {"premise": {"dtype": "string", "id": null, "_type": "Value"}, "hypothesis": {"dtype": "string", "id": null, "_type": "Value"}, "label": {"num_classes": 3, "names": ["entailment", "neutral", "contradiction"], "names_file": null, "id": null, "_type": "ClassLabel"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "americas_nli", "config_name": "aym", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 117538, "num_examples": 743, "dataset_name": "americas_nli"}, "test": {"name": "test", "num_bytes": 115259, "num_examples": 750, "dataset_name": "americas_nli"}}, "download_checksums": {"https://raw.githubusercontent.com/nala-cub/AmericasNLI/main/dev.tsv": {"num_bytes": 1090405, "checksum": "a2678f2820a2008547c5d993118979cc82a25d51a73570571566a1b74d8e8530"}, "https://raw.githubusercontent.com/nala-cub/AmericasNLI/main/test.tsv": {"num_bytes": 1165688, "checksum": "1e16c058de33ddaab4a037b1078a31ecab08afddfdc840140593b6da1439bcf8"}}, "download_size": 2256093, "post_processing_size": null, "dataset_size": 232797, "size_in_bytes": 2488890}, "bzd": {"description": "AmericasNLI is an extension of XNLI (Conneau et al., 2018) \u2013 a natural language inference (NLI) dataset covering 15 high-resource languages \u2013 to 10 low-resource indigenous languages spoken in the Americas: Ashaninka, Aymara, Bribri, Guarani, Nahuatl, Otomi, Quechua, Raramuri, Shipibo-Konibo, and Wixarika. As with MNLI, the goal is to predict textual entailment (does sentence A imply/contradict/neither sentence B) and is a classification task (given two sentences, predict one of three labels).\n", "citation": "\n@article{DBLP:journals/corr/abs-2104-08726,\n author = {Abteen Ebrahimi and\n Manuel Mager and\n Arturo Oncevay and\n Vishrav Chaudhary and\n Luis Chiruzzo and\n Angela Fan and\n John Ortega and\n Ricardo Ramos and\n Annette Rios and\n Ivan Vladimir and\n Gustavo A. Gim{'{e}}nez{-}Lugo and\n Elisabeth Mager and\n Graham Neubig and\n Alexis Palmer and\n Rolando A. Coto Solano and\n Ngoc Thang Vu and\n Katharina Kann},\n title = {AmericasNLI: Evaluating Zero-shot Natural Language Understanding of\n Pretrained Multilingual Models in Truly Low-resource Languages},\n journal = {CoRR},\n volume = {abs/2104.08726},\n year = {2021},\n url = {https://arxiv.org/abs/2104.08726},\n eprinttype = {arXiv},\n eprint = {2104.08726},\n timestamp = {Mon, 26 Apr 2021 17:25:10 +0200},\n biburl = {https://dblp.org/rec/journals/corr/abs-2104-08726.bib},\n bibsource = {dblp computer science bibliography, https://dblp.org}\n}\n", "homepage": "TODO", "license": "", "features": {"premise": {"dtype": "string", "id": null, "_type": "Value"}, "hypothesis": {"dtype": "string", "id": null, "_type": "Value"}, "label": {"num_classes": 3, "names": ["entailment", "neutral", "contradiction"], "names_file": null, "id": null, "_type": "ClassLabel"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "americas_nli", "config_name": "bzd", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 143362, "num_examples": 743, "dataset_name": "americas_nli"}, "test": {"name": "test", "num_bytes": 127684, "num_examples": 750, "dataset_name": "americas_nli"}}, "download_checksums": {"https://raw.githubusercontent.com/nala-cub/AmericasNLI/main/dev.tsv": {"num_bytes": 1090405, "checksum": "a2678f2820a2008547c5d993118979cc82a25d51a73570571566a1b74d8e8530"}, "https://raw.githubusercontent.com/nala-cub/AmericasNLI/main/test.tsv": {"num_bytes": 1165688, "checksum": "1e16c058de33ddaab4a037b1078a31ecab08afddfdc840140593b6da1439bcf8"}}, "download_size": 2256093, "post_processing_size": null, "dataset_size": 271046, "size_in_bytes": 2527139}, "cni": {"description": "AmericasNLI is an extension of XNLI (Conneau et al., 2018) \u2013 a natural language inference (NLI) dataset covering 15 high-resource languages \u2013 to 10 low-resource indigenous languages spoken in the Americas: Ashaninka, Aymara, Bribri, Guarani, Nahuatl, Otomi, Quechua, Raramuri, Shipibo-Konibo, and Wixarika. As with MNLI, the goal is to predict textual entailment (does sentence A imply/contradict/neither sentence B) and is a classification task (given two sentences, predict one of three labels).\n", "citation": "\n@article{DBLP:journals/corr/abs-2104-08726,\n author = {Abteen Ebrahimi and\n Manuel Mager and\n Arturo Oncevay and\n Vishrav Chaudhary and\n Luis Chiruzzo and\n Angela Fan and\n John Ortega and\n Ricardo Ramos and\n Annette Rios and\n Ivan Vladimir and\n Gustavo A. Gim{'{e}}nez{-}Lugo and\n Elisabeth Mager and\n Graham Neubig and\n Alexis Palmer and\n Rolando A. Coto Solano and\n Ngoc Thang Vu and\n Katharina Kann},\n title = {AmericasNLI: Evaluating Zero-shot Natural Language Understanding of\n Pretrained Multilingual Models in Truly Low-resource Languages},\n journal = {CoRR},\n volume = {abs/2104.08726},\n year = {2021},\n url = {https://arxiv.org/abs/2104.08726},\n eprinttype = {arXiv},\n eprint = {2104.08726},\n timestamp = {Mon, 26 Apr 2021 17:25:10 +0200},\n biburl = {https://dblp.org/rec/journals/corr/abs-2104-08726.bib},\n bibsource = {dblp computer science bibliography, https://dblp.org}\n}\n", "homepage": "TODO", "license": "", "features": {"premise": {"dtype": "string", "id": null, "_type": "Value"}, "hypothesis": {"dtype": "string", "id": null, "_type": "Value"}, "label": {"num_classes": 3, "names": ["entailment", "neutral", "contradiction"], "names_file": null, "id": null, "_type": "ClassLabel"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "americas_nli", "config_name": "cni", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 113264, "num_examples": 658, "dataset_name": "americas_nli"}, "test": {"name": "test", "num_bytes": 116292, "num_examples": 750, "dataset_name": "americas_nli"}}, "download_checksums": {"https://raw.githubusercontent.com/nala-cub/AmericasNLI/main/dev.tsv": {"num_bytes": 1090405, "checksum": "a2678f2820a2008547c5d993118979cc82a25d51a73570571566a1b74d8e8530"}, "https://raw.githubusercontent.com/nala-cub/AmericasNLI/main/test.tsv": {"num_bytes": 1165688, "checksum": "1e16c058de33ddaab4a037b1078a31ecab08afddfdc840140593b6da1439bcf8"}}, "download_size": 2256093, "post_processing_size": null, "dataset_size": 229556, "size_in_bytes": 2485649}, "gn": {"description": "AmericasNLI is an extension of XNLI (Conneau et al., 2018) \u2013 a natural language inference (NLI) dataset covering 15 high-resource languages \u2013 to 10 low-resource indigenous languages spoken in the Americas: Ashaninka, Aymara, Bribri, Guarani, Nahuatl, Otomi, Quechua, Raramuri, Shipibo-Konibo, and Wixarika. As with MNLI, the goal is to predict textual entailment (does sentence A imply/contradict/neither sentence B) and is a classification task (given two sentences, predict one of three labels).\n", "citation": "\n@article{DBLP:journals/corr/abs-2104-08726,\n author = {Abteen Ebrahimi and\n Manuel Mager and\n Arturo Oncevay and\n Vishrav Chaudhary and\n Luis Chiruzzo and\n Angela Fan and\n John Ortega and\n Ricardo Ramos and\n Annette Rios and\n Ivan Vladimir and\n Gustavo A. Gim{'{e}}nez{-}Lugo and\n Elisabeth Mager and\n Graham Neubig and\n Alexis Palmer and\n Rolando A. Coto Solano and\n Ngoc Thang Vu and\n Katharina Kann},\n title = {AmericasNLI: Evaluating Zero-shot Natural Language Understanding of\n Pretrained Multilingual Models in Truly Low-resource Languages},\n journal = {CoRR},\n volume = {abs/2104.08726},\n year = {2021},\n url = {https://arxiv.org/abs/2104.08726},\n eprinttype = {arXiv},\n eprint = {2104.08726},\n timestamp = {Mon, 26 Apr 2021 17:25:10 +0200},\n biburl = {https://dblp.org/rec/journals/corr/abs-2104-08726.bib},\n bibsource = {dblp computer science bibliography, https://dblp.org}\n}\n", "homepage": "TODO", "license": "", "features": {"premise": {"dtype": "string", "id": null, "_type": "Value"}, "hypothesis": {"dtype": "string", "id": null, "_type": "Value"}, "label": {"num_classes": 3, "names": ["entailment", "neutral", "contradiction"], "names_file": null, "id": null, "_type": "ClassLabel"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "americas_nli", "config_name": "gn", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 115143, "num_examples": 743, "dataset_name": "americas_nli"}, "test": {"name": "test", "num_bytes": 101956, "num_examples": 750, "dataset_name": "americas_nli"}}, "download_checksums": {"https://raw.githubusercontent.com/nala-cub/AmericasNLI/main/dev.tsv": {"num_bytes": 1090405, "checksum": "a2678f2820a2008547c5d993118979cc82a25d51a73570571566a1b74d8e8530"}, "https://raw.githubusercontent.com/nala-cub/AmericasNLI/main/test.tsv": {"num_bytes": 1165688, "checksum": "1e16c058de33ddaab4a037b1078a31ecab08afddfdc840140593b6da1439bcf8"}}, "download_size": 2256093, "post_processing_size": null, "dataset_size": 217099, "size_in_bytes": 2473192}, "hch": {"description": "AmericasNLI is an extension of XNLI (Conneau et al., 2018) \u2013 a natural language inference (NLI) dataset covering 15 high-resource languages \u2013 to 10 low-resource indigenous languages spoken in the Americas: Ashaninka, Aymara, Bribri, Guarani, Nahuatl, Otomi, Quechua, Raramuri, Shipibo-Konibo, and Wixarika. As with MNLI, the goal is to predict textual entailment (does sentence A imply/contradict/neither sentence B) and is a classification task (given two sentences, predict one of three labels).\n", "citation": "\n@article{DBLP:journals/corr/abs-2104-08726,\n author = {Abteen Ebrahimi and\n Manuel Mager and\n Arturo Oncevay and\n Vishrav Chaudhary and\n Luis Chiruzzo and\n Angela Fan and\n John Ortega and\n Ricardo Ramos and\n Annette Rios and\n Ivan Vladimir and\n Gustavo A. Gim{'{e}}nez{-}Lugo and\n Elisabeth Mager and\n Graham Neubig and\n Alexis Palmer and\n Rolando A. Coto Solano and\n Ngoc Thang Vu and\n Katharina Kann},\n title = {AmericasNLI: Evaluating Zero-shot Natural Language Understanding of\n Pretrained Multilingual Models in Truly Low-resource Languages},\n journal = {CoRR},\n volume = {abs/2104.08726},\n year = {2021},\n url = {https://arxiv.org/abs/2104.08726},\n eprinttype = {arXiv},\n eprint = {2104.08726},\n timestamp = {Mon, 26 Apr 2021 17:25:10 +0200},\n biburl = {https://dblp.org/rec/journals/corr/abs-2104-08726.bib},\n bibsource = {dblp computer science bibliography, https://dblp.org}\n}\n", "homepage": "TODO", "license": "", "features": {"premise": {"dtype": "string", "id": null, "_type": "Value"}, "hypothesis": {"dtype": "string", "id": null, "_type": "Value"}, "label": {"num_classes": 3, "names": ["entailment", "neutral", "contradiction"], "names_file": null, "id": null, "_type": "ClassLabel"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "americas_nli", "config_name": "hch", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 127974, "num_examples": 743, "dataset_name": "americas_nli"}, "test": {"name": "test", "num_bytes": 120865, "num_examples": 750, "dataset_name": "americas_nli"}}, "download_checksums": {"https://raw.githubusercontent.com/nala-cub/AmericasNLI/main/dev.tsv": {"num_bytes": 1090405, "checksum": "a2678f2820a2008547c5d993118979cc82a25d51a73570571566a1b74d8e8530"}, "https://raw.githubusercontent.com/nala-cub/AmericasNLI/main/test.tsv": {"num_bytes": 1165688, "checksum": "1e16c058de33ddaab4a037b1078a31ecab08afddfdc840140593b6da1439bcf8"}}, "download_size": 2256093, "post_processing_size": null, "dataset_size": 248839, "size_in_bytes": 2504932}, "nah": {"description": "AmericasNLI is an extension of XNLI (Conneau et al., 2018) \u2013 a natural language inference (NLI) dataset covering 15 high-resource languages \u2013 to 10 low-resource indigenous languages spoken in the Americas: Ashaninka, Aymara, Bribri, Guarani, Nahuatl, Otomi, Quechua, Raramuri, Shipibo-Konibo, and Wixarika. As with MNLI, the goal is to predict textual entailment (does sentence A imply/contradict/neither sentence B) and is a classification task (given two sentences, predict one of three labels).\n", "citation": "\n@article{DBLP:journals/corr/abs-2104-08726,\n author = {Abteen Ebrahimi and\n Manuel Mager and\n Arturo Oncevay and\n Vishrav Chaudhary and\n Luis Chiruzzo and\n Angela Fan and\n John Ortega and\n Ricardo Ramos and\n Annette Rios and\n Ivan Vladimir and\n Gustavo A. Gim{'{e}}nez{-}Lugo and\n Elisabeth Mager and\n Graham Neubig and\n Alexis Palmer and\n Rolando A. Coto Solano and\n Ngoc Thang Vu and\n Katharina Kann},\n title = {AmericasNLI: Evaluating Zero-shot Natural Language Understanding of\n Pretrained Multilingual Models in Truly Low-resource Languages},\n journal = {CoRR},\n volume = {abs/2104.08726},\n year = {2021},\n url = {https://arxiv.org/abs/2104.08726},\n eprinttype = {arXiv},\n eprint = {2104.08726},\n timestamp = {Mon, 26 Apr 2021 17:25:10 +0200},\n biburl = {https://dblp.org/rec/journals/corr/abs-2104-08726.bib},\n bibsource = {dblp computer science bibliography, https://dblp.org}\n}\n", "homepage": "TODO", "license": "", "features": {"premise": {"dtype": "string", "id": null, "_type": "Value"}, "hypothesis": {"dtype": "string", "id": null, "_type": "Value"}, "label": {"num_classes": 3, "names": ["entailment", "neutral", "contradiction"], "names_file": null, "id": null, "_type": "ClassLabel"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "americas_nli", "config_name": "nah", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 50749, "num_examples": 376, "dataset_name": "americas_nli"}, "test": {"name": "test", "num_bytes": 102961, "num_examples": 738, "dataset_name": "americas_nli"}}, "download_checksums": {"https://raw.githubusercontent.com/nala-cub/AmericasNLI/main/dev.tsv": {"num_bytes": 1090405, "checksum": "a2678f2820a2008547c5d993118979cc82a25d51a73570571566a1b74d8e8530"}, "https://raw.githubusercontent.com/nala-cub/AmericasNLI/main/test.tsv": {"num_bytes": 1165688, "checksum": "1e16c058de33ddaab4a037b1078a31ecab08afddfdc840140593b6da1439bcf8"}}, "download_size": 2256093, "post_processing_size": null, "dataset_size": 153710, "size_in_bytes": 2409803}, "oto": {"description": "AmericasNLI is an extension of XNLI (Conneau et al., 2018) \u2013 a natural language inference (NLI) dataset covering 15 high-resource languages \u2013 to 10 low-resource indigenous languages spoken in the Americas: Ashaninka, Aymara, Bribri, Guarani, Nahuatl, Otomi, Quechua, Raramuri, Shipibo-Konibo, and Wixarika. As with MNLI, the goal is to predict textual entailment (does sentence A imply/contradict/neither sentence B) and is a classification task (given two sentences, predict one of three labels).\n", "citation": "\n@article{DBLP:journals/corr/abs-2104-08726,\n author = {Abteen Ebrahimi and\n Manuel Mager and\n Arturo Oncevay and\n Vishrav Chaudhary and\n Luis Chiruzzo and\n Angela Fan and\n John Ortega and\n Ricardo Ramos and\n Annette Rios and\n Ivan Vladimir and\n Gustavo A. Gim{'{e}}nez{-}Lugo and\n Elisabeth Mager and\n Graham Neubig and\n Alexis Palmer and\n Rolando A. Coto Solano and\n Ngoc Thang Vu and\n Katharina Kann},\n title = {AmericasNLI: Evaluating Zero-shot Natural Language Understanding of\n Pretrained Multilingual Models in Truly Low-resource Languages},\n journal = {CoRR},\n volume = {abs/2104.08726},\n year = {2021},\n url = {https://arxiv.org/abs/2104.08726},\n eprinttype = {arXiv},\n eprint = {2104.08726},\n timestamp = {Mon, 26 Apr 2021 17:25:10 +0200},\n biburl = {https://dblp.org/rec/journals/corr/abs-2104-08726.bib},\n bibsource = {dblp computer science bibliography, https://dblp.org}\n}\n", "homepage": "TODO", "license": "", "features": {"premise": {"dtype": "string", "id": null, "_type": "Value"}, "hypothesis": {"dtype": "string", "id": null, "_type": "Value"}, "label": {"num_classes": 3, "names": ["entailment", "neutral", "contradiction"], "names_file": null, "id": null, "_type": "ClassLabel"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "americas_nli", "config_name": "oto", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 27018, "num_examples": 222, "dataset_name": "americas_nli"}, "test": {"name": "test", "num_bytes": 119658, "num_examples": 748, "dataset_name": "americas_nli"}}, "download_checksums": {"https://raw.githubusercontent.com/nala-cub/AmericasNLI/main/dev.tsv": {"num_bytes": 1090405, "checksum": "a2678f2820a2008547c5d993118979cc82a25d51a73570571566a1b74d8e8530"}, "https://raw.githubusercontent.com/nala-cub/AmericasNLI/main/test.tsv": {"num_bytes": 1165688, "checksum": "1e16c058de33ddaab4a037b1078a31ecab08afddfdc840140593b6da1439bcf8"}}, "download_size": 2256093, "post_processing_size": null, "dataset_size": 146676, "size_in_bytes": 2402769}, "quy": {"description": "AmericasNLI is an extension of XNLI (Conneau et al., 2018) \u2013 a natural language inference (NLI) dataset covering 15 high-resource languages \u2013 to 10 low-resource indigenous languages spoken in the Americas: Ashaninka, Aymara, Bribri, Guarani, Nahuatl, Otomi, Quechua, Raramuri, Shipibo-Konibo, and Wixarika. As with MNLI, the goal is to predict textual entailment (does sentence A imply/contradict/neither sentence B) and is a classification task (given two sentences, predict one of three labels).\n", "citation": "\n@article{DBLP:journals/corr/abs-2104-08726,\n author = {Abteen Ebrahimi and\n Manuel Mager and\n Arturo Oncevay and\n Vishrav Chaudhary and\n Luis Chiruzzo and\n Angela Fan and\n John Ortega and\n Ricardo Ramos and\n Annette Rios and\n Ivan Vladimir and\n Gustavo A. Gim{'{e}}nez{-}Lugo and\n Elisabeth Mager and\n Graham Neubig and\n Alexis Palmer and\n Rolando A. Coto Solano and\n Ngoc Thang Vu and\n Katharina Kann},\n title = {AmericasNLI: Evaluating Zero-shot Natural Language Understanding of\n Pretrained Multilingual Models in Truly Low-resource Languages},\n journal = {CoRR},\n volume = {abs/2104.08726},\n year = {2021},\n url = {https://arxiv.org/abs/2104.08726},\n eprinttype = {arXiv},\n eprint = {2104.08726},\n timestamp = {Mon, 26 Apr 2021 17:25:10 +0200},\n biburl = {https://dblp.org/rec/journals/corr/abs-2104-08726.bib},\n bibsource = {dblp computer science bibliography, https://dblp.org}\n}\n", "homepage": "TODO", "license": "", "features": {"premise": {"dtype": "string", "id": null, "_type": "Value"}, "hypothesis": {"dtype": "string", "id": null, "_type": "Value"}, "label": {"num_classes": 3, "names": ["entailment", "neutral", "contradiction"], "names_file": null, "id": null, "_type": "ClassLabel"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "americas_nli", "config_name": "quy", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 125644, "num_examples": 743, "dataset_name": "americas_nli"}, "test": {"name": "test", "num_bytes": 112758, "num_examples": 750, "dataset_name": "americas_nli"}}, "download_checksums": {"https://raw.githubusercontent.com/nala-cub/AmericasNLI/main/dev.tsv": {"num_bytes": 1090405, "checksum": "a2678f2820a2008547c5d993118979cc82a25d51a73570571566a1b74d8e8530"}, "https://raw.githubusercontent.com/nala-cub/AmericasNLI/main/test.tsv": {"num_bytes": 1165688, "checksum": "1e16c058de33ddaab4a037b1078a31ecab08afddfdc840140593b6da1439bcf8"}}, "download_size": 2256093, "post_processing_size": null, "dataset_size": 238402, "size_in_bytes": 2494495}, "shp": {"description": "AmericasNLI is an extension of XNLI (Conneau et al., 2018) \u2013 a natural language inference (NLI) dataset covering 15 high-resource languages \u2013 to 10 low-resource indigenous languages spoken in the Americas: Ashaninka, Aymara, Bribri, Guarani, Nahuatl, Otomi, Quechua, Raramuri, Shipibo-Konibo, and Wixarika. As with MNLI, the goal is to predict textual entailment (does sentence A imply/contradict/neither sentence B) and is a classification task (given two sentences, predict one of three labels).\n", "citation": "\n@article{DBLP:journals/corr/abs-2104-08726,\n author = {Abteen Ebrahimi and\n Manuel Mager and\n Arturo Oncevay and\n Vishrav Chaudhary and\n Luis Chiruzzo and\n Angela Fan and\n John Ortega and\n Ricardo Ramos and\n Annette Rios and\n Ivan Vladimir and\n Gustavo A. Gim{'{e}}nez{-}Lugo and\n Elisabeth Mager and\n Graham Neubig and\n Alexis Palmer and\n Rolando A. Coto Solano and\n Ngoc Thang Vu and\n Katharina Kann},\n title = {AmericasNLI: Evaluating Zero-shot Natural Language Understanding of\n Pretrained Multilingual Models in Truly Low-resource Languages},\n journal = {CoRR},\n volume = {abs/2104.08726},\n year = {2021},\n url = {https://arxiv.org/abs/2104.08726},\n eprinttype = {arXiv},\n eprint = {2104.08726},\n timestamp = {Mon, 26 Apr 2021 17:25:10 +0200},\n biburl = {https://dblp.org/rec/journals/corr/abs-2104-08726.bib},\n bibsource = {dblp computer science bibliography, https://dblp.org}\n}\n", "homepage": "TODO", "license": "", "features": {"premise": {"dtype": "string", "id": null, "_type": "Value"}, "hypothesis": {"dtype": "string", "id": null, "_type": "Value"}, "label": {"num_classes": 3, "names": ["entailment", "neutral", "contradiction"], "names_file": null, "id": null, "_type": "ClassLabel"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "americas_nli", "config_name": "shp", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 124508, "num_examples": 743, "dataset_name": "americas_nli"}, "test": {"name": "test", "num_bytes": 118942, "num_examples": 750, "dataset_name": "americas_nli"}}, "download_checksums": {"https://raw.githubusercontent.com/nala-cub/AmericasNLI/main/dev.tsv": {"num_bytes": 1090405, "checksum": "a2678f2820a2008547c5d993118979cc82a25d51a73570571566a1b74d8e8530"}, "https://raw.githubusercontent.com/nala-cub/AmericasNLI/main/test.tsv": {"num_bytes": 1165688, "checksum": "1e16c058de33ddaab4a037b1078a31ecab08afddfdc840140593b6da1439bcf8"}}, "download_size": 2256093, "post_processing_size": null, "dataset_size": 243450, "size_in_bytes": 2499543}, "tar": {"description": "AmericasNLI is an extension of XNLI (Conneau et al., 2018) \u2013 a natural language inference (NLI) dataset covering 15 high-resource languages \u2013 to 10 low-resource indigenous languages spoken in the Americas: Ashaninka, Aymara, Bribri, Guarani, Nahuatl, Otomi, Quechua, Raramuri, Shipibo-Konibo, and Wixarika. As with MNLI, the goal is to predict textual entailment (does sentence A imply/contradict/neither sentence B) and is a classification task (given two sentences, predict one of three labels).\n", "citation": "\n@article{DBLP:journals/corr/abs-2104-08726,\n author = {Abteen Ebrahimi and\n Manuel Mager and\n Arturo Oncevay and\n Vishrav Chaudhary and\n Luis Chiruzzo and\n Angela Fan and\n John Ortega and\n Ricardo Ramos and\n Annette Rios and\n Ivan Vladimir and\n Gustavo A. Gim{'{e}}nez{-}Lugo and\n Elisabeth Mager and\n Graham Neubig and\n Alexis Palmer and\n Rolando A. Coto Solano and\n Ngoc Thang Vu and\n Katharina Kann},\n title = {AmericasNLI: Evaluating Zero-shot Natural Language Understanding of\n Pretrained Multilingual Models in Truly Low-resource Languages},\n journal = {CoRR},\n volume = {abs/2104.08726},\n year = {2021},\n url = {https://arxiv.org/abs/2104.08726},\n eprinttype = {arXiv},\n eprint = {2104.08726},\n timestamp = {Mon, 26 Apr 2021 17:25:10 +0200},\n biburl = {https://dblp.org/rec/journals/corr/abs-2104-08726.bib},\n bibsource = {dblp computer science bibliography, https://dblp.org}\n}\n", "homepage": "TODO", "license": "", "features": {"premise": {"dtype": "string", "id": null, "_type": "Value"}, "hypothesis": {"dtype": "string", "id": null, "_type": "Value"}, "label": {"num_classes": 3, "names": ["entailment", "neutral", "contradiction"], "names_file": null, "id": null, "_type": "ClassLabel"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "americas_nli", "config_name": "tar", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 139504, "num_examples": 743, "dataset_name": "americas_nli"}, "test": {"name": "test", "num_bytes": 122632, "num_examples": 750, "dataset_name": "americas_nli"}}, "download_checksums": {"https://raw.githubusercontent.com/nala-cub/AmericasNLI/main/dev.tsv": {"num_bytes": 1090405, "checksum": "a2678f2820a2008547c5d993118979cc82a25d51a73570571566a1b74d8e8530"}, "https://raw.githubusercontent.com/nala-cub/AmericasNLI/main/test.tsv": {"num_bytes": 1165688, "checksum": "1e16c058de33ddaab4a037b1078a31ecab08afddfdc840140593b6da1439bcf8"}}, "download_size": 2256093, "post_processing_size": null, "dataset_size": 262136, "size_in_bytes": 2518229}, "all_languages": {"description": "AmericasNLI is an extension of XNLI (Conneau et al., 2018) \u2013 a natural language inference (NLI) dataset covering 15 high-resource languages \u2013 to 10 low-resource indigenous languages spoken in the Americas: Ashaninka, Aymara, Bribri, Guarani, Nahuatl, Otomi, Quechua, Raramuri, Shipibo-Konibo, and Wixarika. As with MNLI, the goal is to predict textual entailment (does sentence A imply/contradict/neither sentence B) and is a classification task (given two sentences, predict one of three labels).\n", "citation": "\n@article{DBLP:journals/corr/abs-2104-08726,\n author = {Abteen Ebrahimi and\n Manuel Mager and\n Arturo Oncevay and\n Vishrav Chaudhary and\n Luis Chiruzzo and\n Angela Fan and\n John Ortega and\n Ricardo Ramos and\n Annette Rios and\n Ivan Vladimir and\n Gustavo A. Gim{'{e}}nez{-}Lugo and\n Elisabeth Mager and\n Graham Neubig and\n Alexis Palmer and\n Rolando A. Coto Solano and\n Ngoc Thang Vu and\n Katharina Kann},\n title = {AmericasNLI: Evaluating Zero-shot Natural Language Understanding of\n Pretrained Multilingual Models in Truly Low-resource Languages},\n journal = {CoRR},\n volume = {abs/2104.08726},\n year = {2021},\n url = {https://arxiv.org/abs/2104.08726},\n eprinttype = {arXiv},\n eprint = {2104.08726},\n timestamp = {Mon, 26 Apr 2021 17:25:10 +0200},\n biburl = {https://dblp.org/rec/journals/corr/abs-2104-08726.bib},\n bibsource = {dblp computer science bibliography, https://dblp.org}\n}\n", "homepage": "TODO", "license": "", "features": {"language": {"dtype": "string", "id": null, "_type": "Value"}, "premise": {"dtype": "string", "id": null, "_type": "Value"}, "hypothesis": {"dtype": "string", "id": null, "_type": "Value"}, "label": {"num_classes": 3, "names": ["entailment", "neutral", "contradiction"], "names_file": null, "id": null, "_type": "ClassLabel"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "americas_nli", "config_name": "all_languages", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"validation": {"name": "validation", "num_bytes": 1129092, "num_examples": 6457, "dataset_name": "americas_nli"}, "test": {"name": "test", "num_bytes": 1210591, "num_examples": 7486, "dataset_name": "americas_nli"}}, "download_checksums": {"https://raw.githubusercontent.com/nala-cub/AmericasNLI/main/dev.tsv": {"num_bytes": 1090405, "checksum": "a2678f2820a2008547c5d993118979cc82a25d51a73570571566a1b74d8e8530"}, "https://raw.githubusercontent.com/nala-cub/AmericasNLI/main/test.tsv": {"num_bytes": 1165688, "checksum": "1e16c058de33ddaab4a037b1078a31ecab08afddfdc840140593b6da1439bcf8"}}, "download_size": 2256093, "post_processing_size": null, "dataset_size": 2339683, "size_in_bytes": 4595776}}
|
|
|
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8dbd99b1f8cb88fdac4872e2fd48dd2acde44f057d9296754c6777af2e96e127
|
3 |
+
size 38172
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:fd99e85bf961c6b142e1d8b78d540ad7e84d892b1792b5a4cf735c088564c684
|
3 |
+
size 42257
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:f63e8edc98f52fbc181f052125b3c00ad81dfe98664d853b017d84360cc244b1
|
3 |
+
size 44003
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:cab185baa7fd1fb7aec50a380d467b4a6a37df330de959ffec4dc31f067a92e6
|
3 |
+
size 46745
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:04fdf01933b4359b26bb8b6c112e4ffd8ea70b4677aa153fa48fc99538f88a06
|
3 |
+
size 36715
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5b25cc0fb743cb19acaec77bf54fa190c08c6bdd7c168021a568922919a56937
|
3 |
+
size 20238
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c719ef8ba5413006001aa03a0acc90d07d5a1333bd43739259152ffcdc209af1
|
3 |
+
size 44068
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:3b67c8b565fea52357d3bd1c264c738d0d04ed05110ee11e1879a04534e8221b
|
3 |
+
size 13781
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:741df884950981010ec64e4826fb6a990d336ca0b1ccf936d8789dc1fd5fc474
|
3 |
+
size 40858
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d8475ad23b027008dadba25a0b209a78e89994e8385e7eb1d9e76110f07b3c3f
|
3 |
+
size 44815
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8aff1cbdcd32edfcf41f7d6d454c0b1abf4c85e269ce94221e6c726785a56f59
|
3 |
+
size 42017
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6371ebaf3c800f9dabca3199bd8a62acfa32b91b29345422988553a205ae9e2f
|
3 |
+
size 43527
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:78b54f010e4918d73146e5f0136838a5340dda64b82d0f306f20a5b61155367c
|
3 |
+
size 42615
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1533f0b3f08499775cca3a167f78cf0ab75334655a293198555ddfb7c1df113b
|
3 |
+
size 47068
|