Commit
•
835b8af
0
Parent(s):
Update files from the datasets library (from 1.0.0)
Browse filesRelease notes: https://github.com/huggingface/datasets/releases/tag/1.0.0
- .gitattributes +27 -0
- dataset_infos.json +1 -0
- dummy/amazon/1.0.0/dummy_data.zip +3 -0
- dummy/new_wiki/1.0.0/dummy_data.zip +3 -0
- dummy/nyt/1.0.0/dummy_data.zip +3 -0
- dummy/reddit/1.0.0/dummy_data.zip +3 -0
- squadshifts.py +175 -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
|
dataset_infos.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"new_wiki": {"description": "SquadShifts consists of four new test sets for the Stanford Question Answering Dataset (SQuAD) from four different domains: Wikipedia articles, New York \\ \nTimes articles, Reddit comments, and Amazon product reviews. Each dataset was generated using the same data generating pipeline, Amazon Mechanical Turk interface, and data cleaning code as the original SQuAD v1.1 dataset. The \"new-wikipedia\" dataset measures overfitting on the original SQuAD v1.1 dataset. The \"new-york-times\", \"reddit\", and \"amazon\" datasets measure robustness to natural distribution shifts. We encourage SQuAD model developers to also evaluate their methods on these new datasets! ", "citation": "@inproceedings{miller2020effect,\n author = {J. Miller and K. Krauth and B. Recht and L. Schmidt},\n booktitle = {International Conference on Machine Learning (ICML)},\n title = {The Effect of Natural Distribution Shift on Question Answering Models},\n year = {2020},\n}\n", "homepage": "https://modestyachts.github.io/squadshifts-website/index.html", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"text": {"dtype": "string", "id": null, "_type": "Value"}, "answer_start": {"dtype": "int32", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "supervised_keys": null, "builder_name": "squad_shifts", "config_name": "new_wiki", "version": {"version_str": "1.0.0", "description": "", "datasets_version_to_prepare": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 7861360, "num_examples": 7938, "dataset_name": "squad_shifts"}}, "download_checksums": {"https://raw.githubusercontent.com/modestyachts/squadshifts-website/master/datasets/new_wiki_v1.0.json": {"num_bytes": 3607393, "checksum": "1d50a0ca4cdab88d038b52ea8a2d0510af6038d7f41a0a256b9659237769d0ee"}, "https://raw.githubusercontent.com/modestyachts/squadshifts-website/master/datasets/nyt_v1.0.json": {"num_bytes": 4550058, "checksum": "fd55ae252cc735a695fab4473d4cd605b154183ac3b3eb63068f53cfc86f36d6"}, "https://raw.githubusercontent.com/modestyachts/squadshifts-website/master/datasets/reddit_v1.0.json": {"num_bytes": 4170330, "checksum": "b4ac346dbf1816fdec80fdcf91d06a4b754fa51e47960fa8614a94f37ea42a4b"}, "https://raw.githubusercontent.com/modestyachts/squadshifts-website/master/datasets/amazon_reviews_v1.0.json": {"num_bytes": 4177842, "checksum": "9db7bdee84a91f0fb80b9e74f365042fbbb217cb8e0a7e73bb071245472589b6"}}, "download_size": 16505623, "dataset_size": 7861360, "size_in_bytes": 24366983}, "nyt": {"description": "SquadShifts consists of four new test sets for the Stanford Question Answering Dataset (SQuAD) from four different domains: Wikipedia articles, New York \\ \nTimes articles, Reddit comments, and Amazon product reviews. Each dataset was generated using the same data generating pipeline, Amazon Mechanical Turk interface, and data cleaning code as the original SQuAD v1.1 dataset. The \"new-wikipedia\" dataset measures overfitting on the original SQuAD v1.1 dataset. The \"new-york-times\", \"reddit\", and \"amazon\" datasets measure robustness to natural distribution shifts. We encourage SQuAD model developers to also evaluate their methods on these new datasets! ", "citation": "@inproceedings{miller2020effect,\n author = {J. Miller and K. Krauth and B. Recht and L. Schmidt},\n booktitle = {International Conference on Machine Learning (ICML)},\n title = {The Effect of Natural Distribution Shift on Question Answering Models},\n year = {2020},\n}\n", "homepage": "https://modestyachts.github.io/squadshifts-website/index.html", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"text": {"dtype": "string", "id": null, "_type": "Value"}, "answer_start": {"dtype": "int32", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "supervised_keys": null, "builder_name": "squad_shifts", "config_name": "nyt", "version": {"version_str": "1.0.0", "description": "", "datasets_version_to_prepare": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 10787650, "num_examples": 10065, "dataset_name": "squad_shifts"}}, "download_checksums": {"https://raw.githubusercontent.com/modestyachts/squadshifts-website/master/datasets/new_wiki_v1.0.json": {"num_bytes": 3607393, "checksum": "1d50a0ca4cdab88d038b52ea8a2d0510af6038d7f41a0a256b9659237769d0ee"}, "https://raw.githubusercontent.com/modestyachts/squadshifts-website/master/datasets/nyt_v1.0.json": {"num_bytes": 4550058, "checksum": "fd55ae252cc735a695fab4473d4cd605b154183ac3b3eb63068f53cfc86f36d6"}, "https://raw.githubusercontent.com/modestyachts/squadshifts-website/master/datasets/reddit_v1.0.json": {"num_bytes": 4170330, "checksum": "b4ac346dbf1816fdec80fdcf91d06a4b754fa51e47960fa8614a94f37ea42a4b"}, "https://raw.githubusercontent.com/modestyachts/squadshifts-website/master/datasets/amazon_reviews_v1.0.json": {"num_bytes": 4177842, "checksum": "9db7bdee84a91f0fb80b9e74f365042fbbb217cb8e0a7e73bb071245472589b6"}}, "download_size": 16505623, "dataset_size": 10787650, "size_in_bytes": 27293273}, "reddit": {"description": "SquadShifts consists of four new test sets for the Stanford Question Answering Dataset (SQuAD) from four different domains: Wikipedia articles, New York \\ \nTimes articles, Reddit comments, and Amazon product reviews. Each dataset was generated using the same data generating pipeline, Amazon Mechanical Turk interface, and data cleaning code as the original SQuAD v1.1 dataset. The \"new-wikipedia\" dataset measures overfitting on the original SQuAD v1.1 dataset. The \"new-york-times\", \"reddit\", and \"amazon\" datasets measure robustness to natural distribution shifts. We encourage SQuAD model developers to also evaluate their methods on these new datasets! ", "citation": "@inproceedings{miller2020effect,\n author = {J. Miller and K. Krauth and B. Recht and L. Schmidt},\n booktitle = {International Conference on Machine Learning (ICML)},\n title = {The Effect of Natural Distribution Shift on Question Answering Models},\n year = {2020},\n}\n", "homepage": "https://modestyachts.github.io/squadshifts-website/index.html", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"text": {"dtype": "string", "id": null, "_type": "Value"}, "answer_start": {"dtype": "int32", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "supervised_keys": null, "builder_name": "squad_shifts", "config_name": "reddit", "version": {"version_str": "1.0.0", "description": "", "datasets_version_to_prepare": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 9469258, "num_examples": 9803, "dataset_name": "squad_shifts"}}, "download_checksums": {"https://raw.githubusercontent.com/modestyachts/squadshifts-website/master/datasets/new_wiki_v1.0.json": {"num_bytes": 3607393, "checksum": "1d50a0ca4cdab88d038b52ea8a2d0510af6038d7f41a0a256b9659237769d0ee"}, "https://raw.githubusercontent.com/modestyachts/squadshifts-website/master/datasets/nyt_v1.0.json": {"num_bytes": 4550058, "checksum": "fd55ae252cc735a695fab4473d4cd605b154183ac3b3eb63068f53cfc86f36d6"}, "https://raw.githubusercontent.com/modestyachts/squadshifts-website/master/datasets/reddit_v1.0.json": {"num_bytes": 4170330, "checksum": "b4ac346dbf1816fdec80fdcf91d06a4b754fa51e47960fa8614a94f37ea42a4b"}, "https://raw.githubusercontent.com/modestyachts/squadshifts-website/master/datasets/amazon_reviews_v1.0.json": {"num_bytes": 4177842, "checksum": "9db7bdee84a91f0fb80b9e74f365042fbbb217cb8e0a7e73bb071245472589b6"}}, "download_size": 16505623, "dataset_size": 9469258, "size_in_bytes": 25974881}, "amazon": {"description": "SquadShifts consists of four new test sets for the Stanford Question Answering Dataset (SQuAD) from four different domains: Wikipedia articles, New York \\ \nTimes articles, Reddit comments, and Amazon product reviews. Each dataset was generated using the same data generating pipeline, Amazon Mechanical Turk interface, and data cleaning code as the original SQuAD v1.1 dataset. The \"new-wikipedia\" dataset measures overfitting on the original SQuAD v1.1 dataset. The \"new-york-times\", \"reddit\", and \"amazon\" datasets measure robustness to natural distribution shifts. We encourage SQuAD model developers to also evaluate their methods on these new datasets! ", "citation": "@inproceedings{miller2020effect,\n author = {J. Miller and K. Krauth and B. Recht and L. Schmidt},\n booktitle = {International Conference on Machine Learning (ICML)},\n title = {The Effect of Natural Distribution Shift on Question Answering Models},\n year = {2020},\n}\n", "homepage": "https://modestyachts.github.io/squadshifts-website/index.html", "license": "", "features": {"id": {"dtype": "string", "id": null, "_type": "Value"}, "title": {"dtype": "string", "id": null, "_type": "Value"}, "context": {"dtype": "string", "id": null, "_type": "Value"}, "question": {"dtype": "string", "id": null, "_type": "Value"}, "answers": {"feature": {"text": {"dtype": "string", "id": null, "_type": "Value"}, "answer_start": {"dtype": "int32", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "supervised_keys": null, "builder_name": "squad_shifts", "config_name": "amazon", "version": {"version_str": "1.0.0", "description": "", "datasets_version_to_prepare": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 9440353, "num_examples": 9885, "dataset_name": "squad_shifts"}}, "download_checksums": {"https://raw.githubusercontent.com/modestyachts/squadshifts-website/master/datasets/new_wiki_v1.0.json": {"num_bytes": 3607393, "checksum": "1d50a0ca4cdab88d038b52ea8a2d0510af6038d7f41a0a256b9659237769d0ee"}, "https://raw.githubusercontent.com/modestyachts/squadshifts-website/master/datasets/nyt_v1.0.json": {"num_bytes": 4550058, "checksum": "fd55ae252cc735a695fab4473d4cd605b154183ac3b3eb63068f53cfc86f36d6"}, "https://raw.githubusercontent.com/modestyachts/squadshifts-website/master/datasets/reddit_v1.0.json": {"num_bytes": 4170330, "checksum": "b4ac346dbf1816fdec80fdcf91d06a4b754fa51e47960fa8614a94f37ea42a4b"}, "https://raw.githubusercontent.com/modestyachts/squadshifts-website/master/datasets/amazon_reviews_v1.0.json": {"num_bytes": 4177842, "checksum": "9db7bdee84a91f0fb80b9e74f365042fbbb217cb8e0a7e73bb071245472589b6"}}, "download_size": 16505623, "dataset_size": 9440353, "size_in_bytes": 25945976}}
|
dummy/amazon/1.0.0/dummy_data.zip
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:96ce3f5e6a15091aeaeadbe2a23f59e5d0366ccdc92c9d9a5f033725e861a272
|
3 |
+
size 930
|
dummy/new_wiki/1.0.0/dummy_data.zip
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:910171f69848e74b07c1c10d99f3207b3f260b33689aa6c6469380644be3b2ef
|
3 |
+
size 928
|
dummy/nyt/1.0.0/dummy_data.zip
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5c0b7c381ff7c0f7ba88675cf8d26e0343aaeb97482bc3f83162b8b36b08fbde
|
3 |
+
size 918
|
dummy/reddit/1.0.0/dummy_data.zip
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:cdf9a9f31adb94d156385b984d74725c03708b19d8e4bfeaea9f3fa622218789
|
3 |
+
size 912
|
squadshifts.py
ADDED
@@ -0,0 +1,175 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# coding=utf-8
|
2 |
+
# Copyright 2020 The TensorFlow Datasets Authors and the HuggingFace Datasets Authors.
|
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 |
+
|
16 |
+
# Lint as: python3
|
17 |
+
"""SQUAD: The Stanford Question Answering Dataset."""
|
18 |
+
|
19 |
+
from __future__ import absolute_import, division, print_function
|
20 |
+
|
21 |
+
import json
|
22 |
+
import logging
|
23 |
+
import os
|
24 |
+
|
25 |
+
import datasets
|
26 |
+
|
27 |
+
|
28 |
+
_CITATION = """\
|
29 |
+
@inproceedings{miller2020effect,
|
30 |
+
author = {J. Miller and K. Krauth and B. Recht and L. Schmidt},
|
31 |
+
booktitle = {International Conference on Machine Learning (ICML)},
|
32 |
+
title = {The Effect of Natural Distribution Shift on Question Answering Models},
|
33 |
+
year = {2020},
|
34 |
+
}
|
35 |
+
"""
|
36 |
+
|
37 |
+
_DESCRIPTION = r"""\
|
38 |
+
SquadShifts consists of four new test sets for the Stanford Question Answering \
|
39 |
+
Dataset (SQuAD) from four different domains: Wikipedia articles, New York \
|
40 |
+
Times articles, Reddit comments, and Amazon product reviews. Each dataset \
|
41 |
+
was generated using the same data generating pipeline, Amazon Mechanical \
|
42 |
+
Turk interface, and data cleaning code as the original SQuAD v1.1 dataset. \
|
43 |
+
The "new-wikipedia" dataset measures overfitting on the original SQuAD v1.1 \
|
44 |
+
dataset. The "new-york-times", "reddit", and "amazon" datasets measure \
|
45 |
+
robustness to natural distribution shifts. We encourage SQuAD model developers \
|
46 |
+
to also evaluate their methods on these new datasets! \
|
47 |
+
"""
|
48 |
+
|
49 |
+
|
50 |
+
class SquadShiftsConfig(datasets.BuilderConfig):
|
51 |
+
"""BuilderConfig for SquadShifts."""
|
52 |
+
|
53 |
+
def __init__(self, **kwargs):
|
54 |
+
"""BuilderConfig for SQUAD.
|
55 |
+
|
56 |
+
Args:
|
57 |
+
**kwargs: keyword arguments forwarded to super.
|
58 |
+
"""
|
59 |
+
super(SquadShiftsConfig, self).__init__(**kwargs)
|
60 |
+
|
61 |
+
|
62 |
+
class SquadShifts(datasets.GeneratorBasedBuilder):
|
63 |
+
"""SquadShifts consists of four new test sets for the SQUAD dataset."""
|
64 |
+
|
65 |
+
_URL = "https://raw.githubusercontent.com/modestyachts/squadshifts-website/master/datasets"
|
66 |
+
_NEW_WIKI_FILE = "new_wiki_v1.0.json"
|
67 |
+
_NYT_FILE = "nyt_v1.0.json"
|
68 |
+
_REDDIT_FILE = "reddit_v1.0.json"
|
69 |
+
_AMAZON_FILE = "amazon_reviews_v1.0.json"
|
70 |
+
|
71 |
+
BUILDER_CONFIGS = [
|
72 |
+
SquadShiftsConfig(
|
73 |
+
name="new_wiki",
|
74 |
+
version=datasets.Version("1.0.0", ""),
|
75 |
+
description="SQuADShifts New Wikipedia article dataset",
|
76 |
+
),
|
77 |
+
SquadShiftsConfig(
|
78 |
+
name="nyt",
|
79 |
+
version=datasets.Version("1.0.0", ""),
|
80 |
+
description="SQuADShifts New York Times article dataset.",
|
81 |
+
),
|
82 |
+
SquadShiftsConfig(
|
83 |
+
name="reddit",
|
84 |
+
version=datasets.Version("1.0.0", ""),
|
85 |
+
description="SQuADShifts Reddit comment dataset.",
|
86 |
+
),
|
87 |
+
SquadShiftsConfig(
|
88 |
+
name="amazon",
|
89 |
+
version=datasets.Version("1.0.0", ""),
|
90 |
+
description="SQuADShifts Amazon product review dataset.",
|
91 |
+
),
|
92 |
+
]
|
93 |
+
|
94 |
+
def _info(self):
|
95 |
+
return datasets.DatasetInfo(
|
96 |
+
description=_DESCRIPTION,
|
97 |
+
features=datasets.Features(
|
98 |
+
{
|
99 |
+
"id": datasets.Value("string"),
|
100 |
+
"title": datasets.Value("string"),
|
101 |
+
"context": datasets.Value("string"),
|
102 |
+
"question": datasets.Value("string"),
|
103 |
+
"answers": datasets.features.Sequence(
|
104 |
+
{
|
105 |
+
"text": datasets.Value("string"),
|
106 |
+
"answer_start": datasets.Value("int32"),
|
107 |
+
}
|
108 |
+
),
|
109 |
+
}
|
110 |
+
),
|
111 |
+
# No default supervised_keys (as we have to pass both question
|
112 |
+
# and context as input).
|
113 |
+
supervised_keys=None,
|
114 |
+
homepage="https://modestyachts.github.io/squadshifts-website/index.html",
|
115 |
+
citation=_CITATION,
|
116 |
+
)
|
117 |
+
|
118 |
+
def _split_generators(self, dl_manager):
|
119 |
+
urls_to_download = {
|
120 |
+
"new_wiki": os.path.join(self._URL, self._NEW_WIKI_FILE),
|
121 |
+
"nyt": os.path.join(self._URL, self._NYT_FILE),
|
122 |
+
"reddit": os.path.join(self._URL, self._REDDIT_FILE),
|
123 |
+
"amazon": os.path.join(self._URL, self._AMAZON_FILE),
|
124 |
+
}
|
125 |
+
downloaded_files = dl_manager.download_and_extract(urls_to_download)
|
126 |
+
|
127 |
+
if self.config.name == "new_wiki" or self.config.name == "default":
|
128 |
+
return [
|
129 |
+
datasets.SplitGenerator(
|
130 |
+
name=datasets.Split.TEST, gen_kwargs={"filepath": downloaded_files["new_wiki"]}
|
131 |
+
),
|
132 |
+
]
|
133 |
+
elif self.config.name == "nyt":
|
134 |
+
return [
|
135 |
+
datasets.SplitGenerator(name=datasets.Split.TEST, gen_kwargs={"filepath": downloaded_files["nyt"]}),
|
136 |
+
]
|
137 |
+
elif self.config.name == "reddit":
|
138 |
+
return [
|
139 |
+
datasets.SplitGenerator(name=datasets.Split.TEST, gen_kwargs={"filepath": downloaded_files["reddit"]}),
|
140 |
+
]
|
141 |
+
elif self.config.name == "amazon":
|
142 |
+
return [
|
143 |
+
datasets.SplitGenerator(name=datasets.Split.TEST, gen_kwargs={"filepath": downloaded_files["amazon"]}),
|
144 |
+
]
|
145 |
+
else:
|
146 |
+
raise ValueError("SQuADShifts dataset name {} not found!".format(self.config.name))
|
147 |
+
|
148 |
+
def _generate_examples(self, filepath):
|
149 |
+
"""This function returns the examples in the raw (text) form."""
|
150 |
+
logging.info("generating examples from = %s", filepath)
|
151 |
+
with open(filepath, encoding="utf-8") as f:
|
152 |
+
squad = json.load(f)
|
153 |
+
for article in squad["data"]:
|
154 |
+
title = article.get("title", "").strip()
|
155 |
+
for paragraph in article["paragraphs"]:
|
156 |
+
context = paragraph["context"].strip()
|
157 |
+
for qa in paragraph["qas"]:
|
158 |
+
question = qa["question"].strip()
|
159 |
+
id_ = qa["id"]
|
160 |
+
|
161 |
+
answer_starts = [answer["answer_start"] for answer in qa["answers"]]
|
162 |
+
answers = [answer["text"].strip() for answer in qa["answers"]]
|
163 |
+
|
164 |
+
# Features currently used are "context", "question", and "answers".
|
165 |
+
# Others are extracted here for the ease of future expansions.
|
166 |
+
yield id_, {
|
167 |
+
"title": title,
|
168 |
+
"context": context,
|
169 |
+
"question": question,
|
170 |
+
"id": id_,
|
171 |
+
"answers": {
|
172 |
+
"answer_start": answer_starts,
|
173 |
+
"text": answers,
|
174 |
+
},
|
175 |
+
}
|