Datasets:

Sub-tasks:
extractive-qa
Multilinguality:
multilingual
Size Categories:
10K<n<100K
Language Creators:
crowdsourced
Annotations Creators:
crowdsourced
Source Datasets:
original
License:
system HF staff commited on
Commit
7ffe9b1
1 Parent(s): 29fac8c

Update files from the datasets library (from 1.0.2)

Browse files

Release notes: https://github.com/huggingface/datasets/releases/tag/1.0.2

Files changed (2) hide show
  1. dataset_infos.json +0 -0
  2. mlqa.py +6 -6
dataset_infos.json CHANGED
The diff for this file is too large to render. See raw diff
 
mlqa.py CHANGED
@@ -95,11 +95,11 @@ class Mlqa(datasets.GeneratorBasedBuilder):
95
  features=datasets.Features(
96
  {
97
  "context": datasets.Value("string"),
98
- "questions": datasets.Value("string"),
99
  "answers": datasets.features.Sequence(
100
- {"start": datasets.Value("int32"), "text": datasets.Value("string")}
101
  ),
102
- "ids": datasets.Value("string"),
103
  # These are the features of your dataset like images, labels ...
104
  }
105
  ),
@@ -202,7 +202,7 @@ class Mlqa(datasets.GeneratorBasedBuilder):
202
  answers_text = [answer["text"] for answer in answers]
203
  yield id_, {
204
  "context": context,
205
- "questions": question,
206
- "answers": {"start": answers_start, "text": answers_text},
207
- "ids": id_,
208
  }
 
95
  features=datasets.Features(
96
  {
97
  "context": datasets.Value("string"),
98
+ "question": datasets.Value("string"),
99
  "answers": datasets.features.Sequence(
100
+ {"answer_start": datasets.Value("int32"), "text": datasets.Value("string")}
101
  ),
102
+ "id": datasets.Value("string"),
103
  # These are the features of your dataset like images, labels ...
104
  }
105
  ),
 
202
  answers_text = [answer["text"] for answer in answers]
203
  yield id_, {
204
  "context": context,
205
+ "question": question,
206
+ "answers": {"answer_start": answers_start, "text": answers_text},
207
+ "id": id_,
208
  }