Commit
•
2f687f5
0
Parent(s):
Update files from the datasets library (from 1.4.0)
Browse filesRelease notes: https://github.com/huggingface/datasets/releases/tag/1.4.0
- .gitattributes +27 -0
- README.md +268 -0
- dataset_infos.json +1 -0
- dummy/all/1.1.0/dummy_data.zip +3 -0
- m_lama.py +241 -0
.gitattributes
ADDED
@@ -0,0 +1,27 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
*.7z filter=lfs diff=lfs merge=lfs -text
|
2 |
+
*.arrow filter=lfs diff=lfs merge=lfs -text
|
3 |
+
*.bin filter=lfs diff=lfs merge=lfs -text
|
4 |
+
*.bin.* filter=lfs diff=lfs merge=lfs -text
|
5 |
+
*.bz2 filter=lfs diff=lfs merge=lfs -text
|
6 |
+
*.ftz filter=lfs diff=lfs merge=lfs -text
|
7 |
+
*.gz filter=lfs diff=lfs merge=lfs -text
|
8 |
+
*.h5 filter=lfs diff=lfs merge=lfs -text
|
9 |
+
*.joblib filter=lfs diff=lfs merge=lfs -text
|
10 |
+
*.lfs.* filter=lfs diff=lfs merge=lfs -text
|
11 |
+
*.model filter=lfs diff=lfs merge=lfs -text
|
12 |
+
*.msgpack filter=lfs diff=lfs merge=lfs -text
|
13 |
+
*.onnx filter=lfs diff=lfs merge=lfs -text
|
14 |
+
*.ot filter=lfs diff=lfs merge=lfs -text
|
15 |
+
*.parquet filter=lfs diff=lfs merge=lfs -text
|
16 |
+
*.pb filter=lfs diff=lfs merge=lfs -text
|
17 |
+
*.pt filter=lfs diff=lfs merge=lfs -text
|
18 |
+
*.pth filter=lfs diff=lfs merge=lfs -text
|
19 |
+
*.rar filter=lfs diff=lfs merge=lfs -text
|
20 |
+
saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
21 |
+
*.tar.* filter=lfs diff=lfs merge=lfs -text
|
22 |
+
*.tflite filter=lfs diff=lfs merge=lfs -text
|
23 |
+
*.tgz filter=lfs diff=lfs merge=lfs -text
|
24 |
+
*.xz filter=lfs diff=lfs merge=lfs -text
|
25 |
+
*.zip filter=lfs diff=lfs merge=lfs -text
|
26 |
+
*.zstandard filter=lfs diff=lfs merge=lfs -text
|
27 |
+
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
README.md
ADDED
@@ -0,0 +1,268 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
annotations_creators:
|
3 |
+
- crowdsourced
|
4 |
+
- expert-generated
|
5 |
+
- machine-generated
|
6 |
+
language_creators:
|
7 |
+
- crowdsourced
|
8 |
+
- expert-generated
|
9 |
+
- machine-generated
|
10 |
+
languages:
|
11 |
+
- af
|
12 |
+
- ar
|
13 |
+
- az
|
14 |
+
- be
|
15 |
+
- bg
|
16 |
+
- bn
|
17 |
+
- ca
|
18 |
+
- ceb
|
19 |
+
- cs
|
20 |
+
- cy
|
21 |
+
- da
|
22 |
+
- de
|
23 |
+
- el
|
24 |
+
- en
|
25 |
+
- es
|
26 |
+
- et
|
27 |
+
- eu
|
28 |
+
- fa
|
29 |
+
- fi
|
30 |
+
- fr
|
31 |
+
- ga
|
32 |
+
- gl
|
33 |
+
- he
|
34 |
+
- hi
|
35 |
+
- hr
|
36 |
+
- hu
|
37 |
+
- hy
|
38 |
+
- id
|
39 |
+
- it
|
40 |
+
- ja
|
41 |
+
- ka
|
42 |
+
- ko
|
43 |
+
- la
|
44 |
+
- lt
|
45 |
+
- lv
|
46 |
+
- ms
|
47 |
+
- nl
|
48 |
+
- pl
|
49 |
+
- pt
|
50 |
+
- ro
|
51 |
+
- ru
|
52 |
+
- sk
|
53 |
+
- sl
|
54 |
+
- sq
|
55 |
+
- sr
|
56 |
+
- sv
|
57 |
+
- ta
|
58 |
+
- th
|
59 |
+
- tr
|
60 |
+
- uk
|
61 |
+
- ur
|
62 |
+
- vi
|
63 |
+
- zh
|
64 |
+
licenses:
|
65 |
+
- cc-by-nc-sa-4-0
|
66 |
+
multilinguality:
|
67 |
+
- translation
|
68 |
+
size_categories:
|
69 |
+
- 1M>n>100K
|
70 |
+
source_datasets:
|
71 |
+
- extended|lama
|
72 |
+
task_categories:
|
73 |
+
- question-answering
|
74 |
+
- text-scoring
|
75 |
+
task_ids:
|
76 |
+
- open-domain-qa
|
77 |
+
- text-scoring-other-probing
|
78 |
+
---
|
79 |
+
|
80 |
+
# Dataset Card for [Dataset Name]
|
81 |
+
|
82 |
+
## Table of Contents
|
83 |
+
- [Dataset Card for [Dataset Name]](#dataset-card-for-dataset-name)
|
84 |
+
- [Table of Contents](#table-of-contents)
|
85 |
+
- [Dataset Description](#dataset-description)
|
86 |
+
- [Dataset Summary](#dataset-summary)
|
87 |
+
- [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards)
|
88 |
+
- [Languages](#languages)
|
89 |
+
- [Dataset Structure](#dataset-structure)
|
90 |
+
- [Data Instances](#data-instances)
|
91 |
+
- [Data Fields](#data-fields)
|
92 |
+
- [Data Splits](#data-splits)
|
93 |
+
- [Dataset Creation](#dataset-creation)
|
94 |
+
- [Curation Rationale](#curation-rationale)
|
95 |
+
- [Source Data](#source-data)
|
96 |
+
- [Initial Data Collection and Normalization](#initial-data-collection-and-normalization)
|
97 |
+
- [Who are the source language producers?](#who-are-the-source-language-producers)
|
98 |
+
- [Annotations](#annotations)
|
99 |
+
- [Annotation process](#annotation-process)
|
100 |
+
- [Who are the annotators?](#who-are-the-annotators)
|
101 |
+
- [Personal and Sensitive Information](#personal-and-sensitive-information)
|
102 |
+
- [Considerations for Using the Data](#considerations-for-using-the-data)
|
103 |
+
- [Social Impact of Dataset](#social-impact-of-dataset)
|
104 |
+
- [Discussion of Biases](#discussion-of-biases)
|
105 |
+
- [Other Known Limitations](#other-known-limitations)
|
106 |
+
- [Additional Information](#additional-information)
|
107 |
+
- [Dataset Curators](#dataset-curators)
|
108 |
+
- [Licensing Information](#licensing-information)
|
109 |
+
- [Citation Information](#citation-information)
|
110 |
+
- [Contributions](#contributions)
|
111 |
+
|
112 |
+
## Dataset Description
|
113 |
+
|
114 |
+
- **Homepage:** [Multilingual LAMA](http://cistern.cis.lmu.de/mlama/)
|
115 |
+
- **Repository:** [Github](https://github.com/norakassner/mlama)
|
116 |
+
- **Paper:** [Arxiv](https://arxiv.org/abs/2102.00894)
|
117 |
+
- **Point of Contact:** [Contact section](http://cistern.cis.lmu.de/mlama/)
|
118 |
+
|
119 |
+
|
120 |
+
|
121 |
+
### Dataset Summary
|
122 |
+
|
123 |
+
This dataset provides the data for mLAMA, a multilingual version of LAMA.
|
124 |
+
Regarding LAMA see https://github.com/facebookresearch/LAMA. For mLAMA
|
125 |
+
the TREx and GoogleRE part of LAMA was considered and machine translated using
|
126 |
+
Google Translate, and the Wikidata and Google Knowledge Graph API. The machine
|
127 |
+
translated templates were checked for validity, i.e., whether they contain
|
128 |
+
exactly one '[X]' and one '[Y]'.
|
129 |
+
|
130 |
+
This data can be used for creating fill-in-the-blank queries like
|
131 |
+
"Paris is the capital of [MASK]" across 53 languages. For more details see
|
132 |
+
the website http://cistern.cis.lmu.de/mlama/ or the github repo https://github.com/norakassner/mlama.
|
133 |
+
|
134 |
+
### Supported Tasks and Leaderboards
|
135 |
+
|
136 |
+
Language model knowledge probing.
|
137 |
+
|
138 |
+
### Languages
|
139 |
+
|
140 |
+
This dataset contains data in 53 languages:
|
141 |
+
af,ar,az,be,bg,bn,ca,ceb,cs,cy,da,de,el,en,es,et,eu,fa,fi,fr,ga,gl,he,hi,hr,hu,hy,id,it,ja,ka,ko,la,lt,lv,ms,nl,pl,pt,ro,ru,sk,sl,sq,sr,sv,ta,th,tr,uk,ur,vi,zh
|
142 |
+
|
143 |
+
## Dataset Structure
|
144 |
+
For each of the 53 languages and each of the 43 relations/predicates there is a set of triples.
|
145 |
+
|
146 |
+
### Data Instances
|
147 |
+
For each language and relation there are triples, that consists of an object, a predicate and a subject. For each predicate there is a template available. An example for `dataset["test"][0]` is given here:
|
148 |
+
```python
|
149 |
+
{
|
150 |
+
'language': 'af',
|
151 |
+
'lineid': 0,
|
152 |
+
'obj_label': 'Frankryk',
|
153 |
+
'obj_uri': 'Q142',
|
154 |
+
'predicate_id': 'P1001',
|
155 |
+
'sub_label': 'President van Frankryk',
|
156 |
+
'sub_uri': 'Q191954',
|
157 |
+
'template': "[X] is 'n wettige term in [Y].",
|
158 |
+
'uuid': '3fe3d4da-9df9-45ba-8109-784ce5fba38a'
|
159 |
+
}
|
160 |
+
```
|
161 |
+
|
162 |
+
|
163 |
+
### Data Fields
|
164 |
+
|
165 |
+
Each instance has the following fields
|
166 |
+
* "uuid": a unique identifier
|
167 |
+
* "lineid": a identifier unique to mlama
|
168 |
+
* "obj_id": knowledge graph id of the object
|
169 |
+
* "obj_label": surface form of the object
|
170 |
+
* "sub_id": knowledge graph id of the subject
|
171 |
+
* "sub_label": surface form of the subject
|
172 |
+
* "template": template
|
173 |
+
* "language": language code
|
174 |
+
* "predicate_id": relation id
|
175 |
+
|
176 |
+
|
177 |
+
### Data Splits
|
178 |
+
|
179 |
+
There is only one partition that is labelled as 'test data'.
|
180 |
+
|
181 |
+
## Dataset Creation
|
182 |
+
|
183 |
+
### Curation Rationale
|
184 |
+
|
185 |
+
The dataset was translated into 53 languages to investigate knowledge in pretrained language models
|
186 |
+
multilingually.
|
187 |
+
|
188 |
+
### Source Data
|
189 |
+
|
190 |
+
#### Initial Data Collection and Normalization
|
191 |
+
|
192 |
+
The data has several sources:
|
193 |
+
|
194 |
+
LAMA (https://github.com/facebookresearch/LAMA) licensed under Attribution-NonCommercial 4.0 International (CC BY-NC 4.0)
|
195 |
+
T-REx (https://hadyelsahar.github.io/t-rex/) licensed under Creative Commons Attribution-ShareAlike 4.0 International License
|
196 |
+
Google-RE (https://github.com/google-research-datasets/relation-extraction-corpus)
|
197 |
+
Wikidata (https://www.wikidata.org/) licensed under Creative Commons CC0 License and Creative Commons Attribution-ShareAlike License
|
198 |
+
|
199 |
+
#### Who are the source language producers?
|
200 |
+
|
201 |
+
See links above.
|
202 |
+
|
203 |
+
### Annotations
|
204 |
+
|
205 |
+
#### Annotation process
|
206 |
+
|
207 |
+
Crowdsourced (wikidata) and machine translated.
|
208 |
+
|
209 |
+
#### Who are the annotators?
|
210 |
+
|
211 |
+
Unknown.
|
212 |
+
|
213 |
+
### Personal and Sensitive Information
|
214 |
+
|
215 |
+
Names of (most likely) famous people who have entries in Google Knowledge Graph or Wikidata.
|
216 |
+
|
217 |
+
## Considerations for Using the Data
|
218 |
+
|
219 |
+
Data was created through machine translation and automatic processes.
|
220 |
+
|
221 |
+
### Social Impact of Dataset
|
222 |
+
|
223 |
+
[More Information Needed]
|
224 |
+
|
225 |
+
### Discussion of Biases
|
226 |
+
|
227 |
+
[More Information Needed]
|
228 |
+
|
229 |
+
### Other Known Limitations
|
230 |
+
|
231 |
+
Not all triples are available in all languages.
|
232 |
+
|
233 |
+
|
234 |
+
## Additional Information
|
235 |
+
|
236 |
+
### Dataset Curators
|
237 |
+
|
238 |
+
The authors of the mLAMA paper and the authors of the original datasets.
|
239 |
+
|
240 |
+
### Licensing Information
|
241 |
+
|
242 |
+
The Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0). https://creativecommons.org/licenses/by-nc-sa/4.0/
|
243 |
+
|
244 |
+
### Citation Information
|
245 |
+
|
246 |
+
```
|
247 |
+
@article{kassner2021multilingual,
|
248 |
+
author = {Nora Kassner and
|
249 |
+
Philipp Dufter and
|
250 |
+
Hinrich Sch{\"{u}}tze},
|
251 |
+
title = {Multilingual {LAMA:} Investigating Knowledge in Multilingual Pretrained
|
252 |
+
Language Models},
|
253 |
+
journal = {CoRR},
|
254 |
+
volume = {abs/2102.00894},
|
255 |
+
year = {2021},
|
256 |
+
url = {https://arxiv.org/abs/2102.00894},
|
257 |
+
archivePrefix = {arXiv},
|
258 |
+
eprint = {2102.00894},
|
259 |
+
timestamp = {Tue, 09 Feb 2021 13:35:56 +0100},
|
260 |
+
biburl = {https://dblp.org/rec/journals/corr/abs-2102-00894.bib},
|
261 |
+
bibsource = {dblp computer science bibliography, https://dblp.org},
|
262 |
+
note = {to appear in EACL2021}
|
263 |
+
}
|
264 |
+
```
|
265 |
+
|
266 |
+
### Contributions
|
267 |
+
|
268 |
+
Thanks to [@pdufter](https://github.com/pdufter) for adding this dataset.
|
dataset_infos.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"all": {"description": "mLAMA: a multilingual version of the LAMA benchmark (T-REx and GoogleRE) covering 53 languages.", "citation": "\n@article{kassner2021multilingual,\n author = {Nora Kassner and\n Philipp Dufter and\n Hinrich Sch{\"{u}}tze},\n title = {Multilingual {LAMA:} Investigating Knowledge in Multilingual Pretrained\n Language Models},\n journal = {CoRR},\n volume = {abs/2102.00894},\n year = {2021},\n url = {https://arxiv.org/abs/2102.00894},\n archivePrefix = {arXiv},\n eprint = {2102.00894},\n timestamp = {Tue, 09 Feb 2021 13:35:56 +0100},\n biburl = {https://dblp.org/rec/journals/corr/abs-2102-00894.bib},\n bibsource = {dblp computer science bibliography, https://dblp.org},\n note = {to appear in EACL2021}\n}\n", "homepage": "http://cistern.cis.lmu.de/mlama/", "license": "The Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0). https://creativecommons.org/licenses/by-nc-sa/4.0/", "features": {"uuid": {"dtype": "string", "id": null, "_type": "Value"}, "lineid": {"dtype": "uint32", "id": null, "_type": "Value"}, "obj_uri": {"dtype": "string", "id": null, "_type": "Value"}, "obj_label": {"dtype": "string", "id": null, "_type": "Value"}, "sub_uri": {"dtype": "string", "id": null, "_type": "Value"}, "sub_label": {"dtype": "string", "id": null, "_type": "Value"}, "template": {"dtype": "string", "id": null, "_type": "Value"}, "language": {"dtype": "string", "id": null, "_type": "Value"}, "predicate_id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "builder_name": "m_lama", "config_name": "all", "version": {"version_str": "1.1.0", "description": null, "major": 1, "minor": 1, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 125919995, "num_examples": 843143, "dataset_name": "m_lama"}}, "download_checksums": {"http://cistern.cis.lmu.de/mlama/mlama1.1.zip": {"num_bytes": 40772287, "checksum": "043dc82b1b4b72de10ec98fb3a75341af13a1b439f6ee8e769398f42bd6d5883"}}, "download_size": 40772287, "post_processing_size": null, "dataset_size": 125919995, "size_in_bytes": 166692282}}
|
dummy/all/1.1.0/dummy_data.zip
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:88f979f6051f015349ff51138ae78f590791c3f5a981129ed45b996b494aa4c0
|
3 |
+
size 699202
|
m_lama.py
ADDED
@@ -0,0 +1,241 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# coding=utf-8
|
2 |
+
# Copyright 2020 The HuggingFace Datasets Authors and the current dataset script contributor.
|
3 |
+
#
|
4 |
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
5 |
+
# you may not use this file except in compliance with the License.
|
6 |
+
# You may obtain a copy of the License at
|
7 |
+
#
|
8 |
+
# http://www.apache.org/licenses/LICENSE-2.0
|
9 |
+
#
|
10 |
+
# Unless required by applicable law or agreed to in writing, software
|
11 |
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
12 |
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
13 |
+
# See the License for the specific language governing permissions and
|
14 |
+
# limitations under the License.
|
15 |
+
"""The mLAMA Dataset"""
|
16 |
+
|
17 |
+
from __future__ import absolute_import, division, print_function
|
18 |
+
|
19 |
+
import json
|
20 |
+
import os
|
21 |
+
|
22 |
+
import datasets
|
23 |
+
|
24 |
+
|
25 |
+
_CITATION = """
|
26 |
+
@article{kassner2021multilingual,
|
27 |
+
author = {Nora Kassner and
|
28 |
+
Philipp Dufter and
|
29 |
+
Hinrich Sch{\"{u}}tze},
|
30 |
+
title = {Multilingual {LAMA:} Investigating Knowledge in Multilingual Pretrained
|
31 |
+
Language Models},
|
32 |
+
journal = {CoRR},
|
33 |
+
volume = {abs/2102.00894},
|
34 |
+
year = {2021},
|
35 |
+
url = {https://arxiv.org/abs/2102.00894},
|
36 |
+
archivePrefix = {arXiv},
|
37 |
+
eprint = {2102.00894},
|
38 |
+
timestamp = {Tue, 09 Feb 2021 13:35:56 +0100},
|
39 |
+
biburl = {https://dblp.org/rec/journals/corr/abs-2102-00894.bib},
|
40 |
+
bibsource = {dblp computer science bibliography, https://dblp.org},
|
41 |
+
note = {to appear in EACL2021}
|
42 |
+
}
|
43 |
+
"""
|
44 |
+
|
45 |
+
|
46 |
+
_DESCRIPTION = """mLAMA: a multilingual version of the LAMA benchmark (T-REx and GoogleRE) covering 53 languages."""
|
47 |
+
|
48 |
+
_HOMEPAGE = "http://cistern.cis.lmu.de/mlama/"
|
49 |
+
|
50 |
+
_LICENSE = "The Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0). https://creativecommons.org/licenses/by-nc-sa/4.0/"
|
51 |
+
|
52 |
+
_URL = "http://cistern.cis.lmu.de/mlama/mlama1.1.zip"
|
53 |
+
|
54 |
+
_LANGUAGES = (
|
55 |
+
"af",
|
56 |
+
"ar",
|
57 |
+
"az",
|
58 |
+
"be",
|
59 |
+
"bg",
|
60 |
+
"bn",
|
61 |
+
"ca",
|
62 |
+
"ceb",
|
63 |
+
"cs",
|
64 |
+
"cy",
|
65 |
+
"da",
|
66 |
+
"de",
|
67 |
+
"el",
|
68 |
+
"en",
|
69 |
+
"es",
|
70 |
+
"et",
|
71 |
+
"eu",
|
72 |
+
"fa",
|
73 |
+
"fi",
|
74 |
+
"fr",
|
75 |
+
"ga",
|
76 |
+
"gl",
|
77 |
+
"he",
|
78 |
+
"hi",
|
79 |
+
"hr",
|
80 |
+
"hu",
|
81 |
+
"hy",
|
82 |
+
"id",
|
83 |
+
"it",
|
84 |
+
"ja",
|
85 |
+
"ka",
|
86 |
+
"ko",
|
87 |
+
"la",
|
88 |
+
"lt",
|
89 |
+
"lv",
|
90 |
+
"ms",
|
91 |
+
"nl",
|
92 |
+
"pl",
|
93 |
+
"pt",
|
94 |
+
"ro",
|
95 |
+
"ru",
|
96 |
+
"sk",
|
97 |
+
"sl",
|
98 |
+
"sq",
|
99 |
+
"sr",
|
100 |
+
"sv",
|
101 |
+
"ta",
|
102 |
+
"th",
|
103 |
+
"tr",
|
104 |
+
"uk",
|
105 |
+
"ur",
|
106 |
+
"vi",
|
107 |
+
"zh",
|
108 |
+
)
|
109 |
+
_RELATIONS = (
|
110 |
+
"place_of_birth",
|
111 |
+
"place_of_death",
|
112 |
+
"P1001",
|
113 |
+
"P101",
|
114 |
+
"P103",
|
115 |
+
"P106",
|
116 |
+
"P108",
|
117 |
+
"P127",
|
118 |
+
"P1303",
|
119 |
+
"P131",
|
120 |
+
"P136",
|
121 |
+
"P1376",
|
122 |
+
"P138",
|
123 |
+
"P140",
|
124 |
+
"P1412",
|
125 |
+
"P159",
|
126 |
+
"P17",
|
127 |
+
"P176",
|
128 |
+
"P178",
|
129 |
+
"P19",
|
130 |
+
"P190",
|
131 |
+
"P20",
|
132 |
+
"P264",
|
133 |
+
"P27",
|
134 |
+
"P276",
|
135 |
+
"P279",
|
136 |
+
"P30",
|
137 |
+
"P31",
|
138 |
+
"P36",
|
139 |
+
"P361",
|
140 |
+
"P364",
|
141 |
+
"P37",
|
142 |
+
"P39",
|
143 |
+
"P407",
|
144 |
+
"P413",
|
145 |
+
"P449",
|
146 |
+
"P463",
|
147 |
+
"P47",
|
148 |
+
"P495",
|
149 |
+
"P527",
|
150 |
+
"P530",
|
151 |
+
"P740",
|
152 |
+
"P937",
|
153 |
+
)
|
154 |
+
|
155 |
+
|
156 |
+
class MLamaConfig(datasets.BuilderConfig):
|
157 |
+
"""BuilderConfig for mLAMA."""
|
158 |
+
|
159 |
+
def __init__(self, languages=None, relations=None, **kwargs):
|
160 |
+
"""BuilderConfig for mLAMA.
|
161 |
+
Args:
|
162 |
+
languages: A subset of af,ar,az,be,bg,bn,ca,ceb,cs,cy,da,de,el,en,es,et,eu,fa,fi,fr,ga,gl,he,hi,hr,hu,hy,id,it,ja,ka,ko,la,lt,lv,ms,nl,pl,pt,ro,ru,sk,sl,sq,sr,sv,ta,th,tr,uk,ur,vi,zh
|
163 |
+
relations: A subset of place_of_birth,place_of_death,P1001,P101,P103,P106,P108,P127,P1303,P131,P136,P1376,P138,P140,P1412,P159,P17,P176,P178,P19,P190,P20,P264,P27,P276,P279,P30,P31,P36,P361,P364,P37,P39,P407,P413,P449,P463,P47,P495,P527,P530,P740,P937
|
164 |
+
**kwargs: keyword arguments forwarded to super.
|
165 |
+
"""
|
166 |
+
super(MLamaConfig, self).__init__(**kwargs)
|
167 |
+
self.languages = languages if languages is not None else _LANGUAGES
|
168 |
+
self.relations = relations if relations is not None else _RELATIONS
|
169 |
+
|
170 |
+
|
171 |
+
class MLama(datasets.GeneratorBasedBuilder):
|
172 |
+
"""multilingual LAMA Dataset (mLAMA)"""
|
173 |
+
|
174 |
+
VERSION = datasets.Version("1.1.0")
|
175 |
+
BUILDER_CONFIG_CLASS = MLamaConfig
|
176 |
+
BUILDER_CONFIGS = [
|
177 |
+
MLamaConfig(
|
178 |
+
name="all",
|
179 |
+
languages=None,
|
180 |
+
relations=None,
|
181 |
+
version=datasets.Version("1.1.0"),
|
182 |
+
description="Import of mLAMA for all languages and all relations.",
|
183 |
+
)
|
184 |
+
]
|
185 |
+
|
186 |
+
def _info(self):
|
187 |
+
features = datasets.Features(
|
188 |
+
{
|
189 |
+
"uuid": datasets.Value("string"),
|
190 |
+
"lineid": datasets.Value("uint32"),
|
191 |
+
"obj_uri": datasets.Value("string"),
|
192 |
+
"obj_label": datasets.Value("string"),
|
193 |
+
"sub_uri": datasets.Value("string"),
|
194 |
+
"sub_label": datasets.Value("string"),
|
195 |
+
"template": datasets.Value("string"),
|
196 |
+
"language": datasets.Value("string"),
|
197 |
+
"predicate_id": datasets.Value("string"),
|
198 |
+
}
|
199 |
+
)
|
200 |
+
return datasets.DatasetInfo(
|
201 |
+
description=_DESCRIPTION,
|
202 |
+
features=features,
|
203 |
+
supervised_keys=None,
|
204 |
+
homepage=_HOMEPAGE,
|
205 |
+
license=_LICENSE,
|
206 |
+
citation=_CITATION,
|
207 |
+
)
|
208 |
+
|
209 |
+
def _split_generators(self, dl_manager):
|
210 |
+
"""Returns SplitGenerators."""
|
211 |
+
data_dir = dl_manager.download_and_extract(_URL)
|
212 |
+
return [
|
213 |
+
datasets.SplitGenerator(
|
214 |
+
name=datasets.Split.TEST,
|
215 |
+
gen_kwargs={
|
216 |
+
"filepath": os.path.join(data_dir, "mlama1.1"),
|
217 |
+
"split": "test",
|
218 |
+
},
|
219 |
+
),
|
220 |
+
]
|
221 |
+
|
222 |
+
def _generate_examples(self, filepath, split):
|
223 |
+
""" Yields examples from the mLAMA dataset. """
|
224 |
+
id_ = -1
|
225 |
+
for language in self.config.languages:
|
226 |
+
# load templates
|
227 |
+
templates = {}
|
228 |
+
with open(os.path.join(filepath, language, "templates.jsonl"), encoding="utf-8") as fp:
|
229 |
+
for line in fp:
|
230 |
+
line = json.loads(line)
|
231 |
+
templates[line["relation"]] = line["template"]
|
232 |
+
for relation in self.config.relations:
|
233 |
+
# load triples
|
234 |
+
with open(os.path.join(filepath, language, f"{relation}.jsonl"), encoding="utf-8") as fp:
|
235 |
+
for line in fp:
|
236 |
+
triple = json.loads(line)
|
237 |
+
triple["language"] = language
|
238 |
+
triple["predicate_id"] = relation
|
239 |
+
triple["template"] = templates.get(relation, "")
|
240 |
+
id_ += 1
|
241 |
+
yield id_, triple
|