Delete the script
Browse files- newyorker_caption_contest.py +0 -402
newyorker_caption_contest.py
DELETED
@@ -1,402 +0,0 @@
|
|
1 |
-
# coding=utf-8
|
2 |
-
# Lint as: python3
|
3 |
-
"""The Caption Contest benchmark."""
|
4 |
-
|
5 |
-
|
6 |
-
import json
|
7 |
-
import os
|
8 |
-
import datasets
|
9 |
-
import base64
|
10 |
-
import pprint
|
11 |
-
|
12 |
-
|
13 |
-
_CAPTION_CONTEST_TASKS_CITATION = """\
|
14 |
-
@article{hessel2022androids,
|
15 |
-
title={Do Androids Laugh at Electric Sheep? Humor" Understanding" Benchmarks from The New Yorker Caption Contest},
|
16 |
-
author={Hessel, Jack and Marasovi{\'c}, Ana and Hwang, Jena D and Lee, Lillian and Da, Jeff and Zellers, Rowan and Mankoff, Robert and Choi, Yejin},
|
17 |
-
journal={arXiv preprint arXiv:2209.06293},
|
18 |
-
year={2022}
|
19 |
-
}
|
20 |
-
|
21 |
-
www.capcon.dev
|
22 |
-
|
23 |
-
Our data contributions are:
|
24 |
-
|
25 |
-
- The cartoon-level annotations;
|
26 |
-
- The joke explanations;
|
27 |
-
- and the framing of the tasks
|
28 |
-
We release these data we contribute under CC-BY (see DATASET_LICENSE).
|
29 |
-
|
30 |
-
If you find this data useful in your work, in addition to citing our contributions, please also cite the following, from which the cartoons/captions in our corpus are derived:
|
31 |
-
|
32 |
-
@misc{newyorkernextmldataset,
|
33 |
-
author={Jain, Lalit and Jamieson, Kevin and Mankoff, Robert and Nowak, Robert and Sievert, Scott},
|
34 |
-
title={The {N}ew {Y}orker Cartoon Caption Contest Dataset},
|
35 |
-
year={2020},
|
36 |
-
url={https://nextml.github.io/caption-contest-data/}
|
37 |
-
}
|
38 |
-
|
39 |
-
@inproceedings{radev-etal-2016-humor,
|
40 |
-
title = "Humor in Collective Discourse: Unsupervised Funniness Detection in The {New Yorker} Cartoon Caption Contest",
|
41 |
-
author = "Radev, Dragomir and
|
42 |
-
Stent, Amanda and
|
43 |
-
Tetreault, Joel and
|
44 |
-
Pappu, Aasish and
|
45 |
-
Iliakopoulou, Aikaterini and
|
46 |
-
Chanfreau, Agustin and
|
47 |
-
de Juan, Paloma and
|
48 |
-
Vallmitjana, Jordi and
|
49 |
-
Jaimes, Alejandro and
|
50 |
-
Jha, Rahul and
|
51 |
-
Mankoff, Robert",
|
52 |
-
booktitle = "LREC",
|
53 |
-
year = "2016",
|
54 |
-
}
|
55 |
-
|
56 |
-
@inproceedings{shahaf2015inside,
|
57 |
-
title={Inside jokes: Identifying humorous cartoon captions},
|
58 |
-
author={Shahaf, Dafna and Horvitz, Eric and Mankoff, Robert},
|
59 |
-
booktitle={KDD},
|
60 |
-
year={2015},
|
61 |
-
}
|
62 |
-
"""
|
63 |
-
|
64 |
-
|
65 |
-
_CAPTION_CONTEST_DESCRIPTION = """\
|
66 |
-
There are 3 caption contest tasks, described in the paper. In the Matching multiple choice task, models must recognize a caption written about a cartoon (vs. options that were not). In the Quality Ranking task, models must evaluate the quality
|
67 |
-
of that caption by scoring it more highly than a lower quality option from the same contest. In the Explanation Generation task, models must explain why the joke is funny.
|
68 |
-
"""
|
69 |
-
|
70 |
-
_MATCHING_DESCRIPTION = """\
|
71 |
-
You are given a cartoon and 5 captions. Only one of the captions was truly written about the cartoon. You must select it.
|
72 |
-
"""
|
73 |
-
|
74 |
-
_RANKING_DESCRIPTION = """\
|
75 |
-
You are given a cartoon and 2 captions. One of the captions was selected by crowd voting or New Yorker editors as high quality. You must select it.
|
76 |
-
"""
|
77 |
-
|
78 |
-
_EXPLANATION_DESCRIPTION = """\
|
79 |
-
You are given a cartoon and a caption that was written about it. You must autoregressively generate a joke explanation.
|
80 |
-
"""
|
81 |
-
|
82 |
-
|
83 |
-
_IMAGES_URL = "https://storage.googleapis.com/ai2-jack-public/caption_contest_data_public/all_contest_images.zip"
|
84 |
-
|
85 |
-
|
86 |
-
def _get_configs_crossvals():
|
87 |
-
cross_val_configs = []
|
88 |
-
for split_idx in [1,2,3,4]:
|
89 |
-
cur_split_configs = [
|
90 |
-
CaptionContestConfig(
|
91 |
-
name='matching_{}'.format(split_idx),
|
92 |
-
description=_MATCHING_DESCRIPTION,
|
93 |
-
features=[
|
94 |
-
'image',
|
95 |
-
'contest_number',
|
96 |
-
'image_location',
|
97 |
-
'image_description',
|
98 |
-
'image_uncanny_description',
|
99 |
-
'entities',
|
100 |
-
'questions',
|
101 |
-
'caption_choices',
|
102 |
-
'from_description',
|
103 |
-
],
|
104 |
-
label_classes=["A", "B", "C", "D", "E"],
|
105 |
-
data_url='https://storage.googleapis.com/ai2-jack-public/caption_contest_data_public/huggingface_hub/v1.0/matching_{}.zip'.format(split_idx),
|
106 |
-
url='www.capcon.dev',
|
107 |
-
citation=_CAPTION_CONTEST_TASKS_CITATION,
|
108 |
-
),
|
109 |
-
|
110 |
-
CaptionContestConfig(
|
111 |
-
name='matching_from_pixels_{}'.format(split_idx),
|
112 |
-
description=_MATCHING_DESCRIPTION,
|
113 |
-
features=[
|
114 |
-
'image',
|
115 |
-
'contest_number',
|
116 |
-
'caption_choices',
|
117 |
-
],
|
118 |
-
label_classes=["A", "B", "C", "D", "E"],
|
119 |
-
data_url='https://storage.googleapis.com/ai2-jack-public/caption_contest_data_public/huggingface_hub/v1.0/matching_from_pixels_{}.zip'.format(split_idx),
|
120 |
-
url='www.capcon.dev',
|
121 |
-
citation=_CAPTION_CONTEST_TASKS_CITATION,
|
122 |
-
),
|
123 |
-
|
124 |
-
CaptionContestConfig(
|
125 |
-
name='ranking_{}'.format(split_idx),
|
126 |
-
description=_RANKING_DESCRIPTION,
|
127 |
-
features=[
|
128 |
-
'image',
|
129 |
-
'contest_number',
|
130 |
-
'image_location',
|
131 |
-
'image_description',
|
132 |
-
'image_uncanny_description',
|
133 |
-
'entities',
|
134 |
-
'questions',
|
135 |
-
'caption_choices',
|
136 |
-
'from_description',
|
137 |
-
'winner_source',
|
138 |
-
],
|
139 |
-
|
140 |
-
label_classes=["A", "B"],
|
141 |
-
data_url='https://storage.googleapis.com/ai2-jack-public/caption_contest_data_public/huggingface_hub/v1.0/ranking_{}.zip'.format(split_idx),
|
142 |
-
url='www.capcon.dev',
|
143 |
-
citation=_CAPTION_CONTEST_TASKS_CITATION,
|
144 |
-
),
|
145 |
-
|
146 |
-
CaptionContestConfig(
|
147 |
-
name='ranking_from_pixels_{}'.format(split_idx),
|
148 |
-
description=_RANKING_DESCRIPTION,
|
149 |
-
features=[
|
150 |
-
'image',
|
151 |
-
'contest_number',
|
152 |
-
'caption_choices',
|
153 |
-
'winner_source',
|
154 |
-
],
|
155 |
-
label_classes=["A", "B"],
|
156 |
-
data_url='https://storage.googleapis.com/ai2-jack-public/caption_contest_data_public/huggingface_hub/v1.0/ranking_from_pixels_{}.zip'.format(split_idx),
|
157 |
-
url='www.capcon.dev',
|
158 |
-
citation=_CAPTION_CONTEST_TASKS_CITATION,
|
159 |
-
),
|
160 |
-
|
161 |
-
|
162 |
-
CaptionContestConfig(
|
163 |
-
name='explanation_{}'.format(split_idx),
|
164 |
-
description=_EXPLANATION_DESCRIPTION,
|
165 |
-
features=[
|
166 |
-
'image',
|
167 |
-
'contest_number',
|
168 |
-
'image_location',
|
169 |
-
'image_description',
|
170 |
-
'image_uncanny_description',
|
171 |
-
'entities',
|
172 |
-
'questions',
|
173 |
-
'caption_choices',
|
174 |
-
'from_description',
|
175 |
-
],
|
176 |
-
label_classes=None,
|
177 |
-
data_url='https://storage.googleapis.com/ai2-jack-public/caption_contest_data_public/huggingface_hub/v1.0/explanation_{}.zip'.format(split_idx),
|
178 |
-
url='www.capcon.dev',
|
179 |
-
citation=_CAPTION_CONTEST_TASKS_CITATION,
|
180 |
-
),
|
181 |
-
|
182 |
-
CaptionContestConfig(
|
183 |
-
name='explanation_from_pixels_{}'.format(split_idx),
|
184 |
-
description=_EXPLANATION_DESCRIPTION,
|
185 |
-
features=[
|
186 |
-
'image',
|
187 |
-
'contest_number',
|
188 |
-
'caption_choices',
|
189 |
-
],
|
190 |
-
label_classes=None,
|
191 |
-
data_url='https://storage.googleapis.com/ai2-jack-public/caption_contest_data_public/huggingface_hub/v1.0/explanation_from_pixels_{}.zip'.format(split_idx),
|
192 |
-
url='www.capcon.dev',
|
193 |
-
citation=_CAPTION_CONTEST_TASKS_CITATION,
|
194 |
-
),
|
195 |
-
]
|
196 |
-
cross_val_configs.extend(cur_split_configs)
|
197 |
-
return cross_val_configs
|
198 |
-
|
199 |
-
|
200 |
-
class CaptionContestConfig(datasets.BuilderConfig):
|
201 |
-
"""BuilderConfig for Caption Contest."""
|
202 |
-
|
203 |
-
def __init__(self, features, data_url, citation, url, label_classes=None, **kwargs):
|
204 |
-
"""BuilderConfig for Caption Contest.
|
205 |
-
Args:
|
206 |
-
features: `list[string]`, list of the features that will appear in the
|
207 |
-
feature dict. Should not include "label".
|
208 |
-
data_url: `string`, url to download the zip file from.
|
209 |
-
citation: `string`, citation for the data set.
|
210 |
-
url: `string`, url for information about the data set.
|
211 |
-
label_classes: `list[string]`, the list of classes for the label if the
|
212 |
-
label is present as a string. If not provided, there is no fixed label set.
|
213 |
-
**kwargs: keyword arguments forwarded to super.
|
214 |
-
"""
|
215 |
-
|
216 |
-
super(CaptionContestConfig, self).__init__(version=datasets.Version("1.0.0"), **kwargs)
|
217 |
-
self.features = features
|
218 |
-
self.data_url = data_url
|
219 |
-
self.citation = citation
|
220 |
-
self.url = url
|
221 |
-
self.label_classes = label_classes
|
222 |
-
|
223 |
-
|
224 |
-
class CaptionContest(datasets.GeneratorBasedBuilder):
|
225 |
-
"""The CaptionContest benchmark."""
|
226 |
-
|
227 |
-
BUILDER_CONFIGS = [
|
228 |
-
CaptionContestConfig(
|
229 |
-
name='matching',
|
230 |
-
description=_MATCHING_DESCRIPTION,
|
231 |
-
features=[
|
232 |
-
'image',
|
233 |
-
'contest_number',
|
234 |
-
'image_location',
|
235 |
-
'image_description',
|
236 |
-
'image_uncanny_description',
|
237 |
-
'entities',
|
238 |
-
'questions',
|
239 |
-
'caption_choices',
|
240 |
-
'from_description',
|
241 |
-
],
|
242 |
-
label_classes=["A", "B", "C", "D", "E"],
|
243 |
-
data_url='https://storage.googleapis.com/ai2-jack-public/caption_contest_data_public/huggingface_hub/v1.0/matching.zip',
|
244 |
-
url='www.capcon.dev',
|
245 |
-
citation=_CAPTION_CONTEST_TASKS_CITATION,
|
246 |
-
),
|
247 |
-
|
248 |
-
CaptionContestConfig(
|
249 |
-
name='matching_from_pixels',
|
250 |
-
description=_MATCHING_DESCRIPTION,
|
251 |
-
features=[
|
252 |
-
'image',
|
253 |
-
'contest_number',
|
254 |
-
'caption_choices',
|
255 |
-
],
|
256 |
-
label_classes=["A", "B", "C", "D", "E"],
|
257 |
-
data_url='https://storage.googleapis.com/ai2-jack-public/caption_contest_data_public/huggingface_hub/v1.0/matching_from_pixels.zip',
|
258 |
-
url='www.capcon.dev',
|
259 |
-
citation=_CAPTION_CONTEST_TASKS_CITATION,
|
260 |
-
),
|
261 |
-
|
262 |
-
CaptionContestConfig(
|
263 |
-
name='ranking',
|
264 |
-
description=_RANKING_DESCRIPTION,
|
265 |
-
features=[
|
266 |
-
'image',
|
267 |
-
'contest_number',
|
268 |
-
'image_location',
|
269 |
-
'image_description',
|
270 |
-
'image_uncanny_description',
|
271 |
-
'entities',
|
272 |
-
'questions',
|
273 |
-
'caption_choices',
|
274 |
-
'from_description',
|
275 |
-
'winner_source',
|
276 |
-
],
|
277 |
-
label_classes=["A", "B"],
|
278 |
-
data_url='https://storage.googleapis.com/ai2-jack-public/caption_contest_data_public/huggingface_hub/v1.0/ranking.zip',
|
279 |
-
url='www.capcon.dev',
|
280 |
-
citation=_CAPTION_CONTEST_TASKS_CITATION,
|
281 |
-
),
|
282 |
-
|
283 |
-
CaptionContestConfig(
|
284 |
-
name='ranking_from_pixels',
|
285 |
-
description=_RANKING_DESCRIPTION,
|
286 |
-
features=[
|
287 |
-
'image',
|
288 |
-
'contest_number',
|
289 |
-
'caption_choices',
|
290 |
-
'winner_source',
|
291 |
-
],
|
292 |
-
label_classes=["A", "B"],
|
293 |
-
data_url='https://storage.googleapis.com/ai2-jack-public/caption_contest_data_public/huggingface_hub/v1.0/ranking_from_pixels.zip',
|
294 |
-
url='www.capcon.dev',
|
295 |
-
citation=_CAPTION_CONTEST_TASKS_CITATION,
|
296 |
-
),
|
297 |
-
|
298 |
-
|
299 |
-
CaptionContestConfig(
|
300 |
-
name='explanation',
|
301 |
-
description=_EXPLANATION_DESCRIPTION,
|
302 |
-
features=[
|
303 |
-
'image',
|
304 |
-
'contest_number',
|
305 |
-
'image_location',
|
306 |
-
'image_description',
|
307 |
-
'image_uncanny_description',
|
308 |
-
'entities',
|
309 |
-
'questions',
|
310 |
-
'caption_choices',
|
311 |
-
'from_description',
|
312 |
-
],
|
313 |
-
label_classes=None,
|
314 |
-
data_url='https://storage.googleapis.com/ai2-jack-public/caption_contest_data_public/huggingface_hub/v1.0/explanation.zip',
|
315 |
-
url='www.capcon.dev',
|
316 |
-
citation=_CAPTION_CONTEST_TASKS_CITATION,
|
317 |
-
),
|
318 |
-
|
319 |
-
CaptionContestConfig(
|
320 |
-
name='explanation_from_pixels',
|
321 |
-
description=_EXPLANATION_DESCRIPTION,
|
322 |
-
features=[
|
323 |
-
'image',
|
324 |
-
'contest_number',
|
325 |
-
'caption_choices',
|
326 |
-
],
|
327 |
-
label_classes=None,
|
328 |
-
data_url='https://storage.googleapis.com/ai2-jack-public/caption_contest_data_public/huggingface_hub/v1.0/explanation_from_pixels.zip',
|
329 |
-
url='www.capcon.dev',
|
330 |
-
citation=_CAPTION_CONTEST_TASKS_CITATION,
|
331 |
-
),
|
332 |
-
] + _get_configs_crossvals()
|
333 |
-
|
334 |
-
|
335 |
-
def _info(self):
|
336 |
-
features = {feature: datasets.Value("string") for feature in self.config.features}
|
337 |
-
# things are strings except for contest_number, entities, questions, and caption choices (if not explanation)
|
338 |
-
features['contest_number'] = datasets.Value("int32")
|
339 |
-
if 'explanation' not in self.config.name:
|
340 |
-
features['caption_choices'] = datasets.features.Sequence(datasets.Value("string"))
|
341 |
-
|
342 |
-
if 'entities' in features:
|
343 |
-
features['entities'] = datasets.features.Sequence(datasets.Value("string"))
|
344 |
-
|
345 |
-
if 'questions' in features:
|
346 |
-
features['questions'] = datasets.features.Sequence(datasets.Value("string"))
|
347 |
-
|
348 |
-
if 'image' in features:
|
349 |
-
features['image'] = datasets.Image()
|
350 |
-
|
351 |
-
features['label'] = datasets.Value("string")
|
352 |
-
features['n_tokens_label'] = datasets.Value("int32")
|
353 |
-
features['instance_id'] = datasets.Value("string")
|
354 |
-
|
355 |
-
return datasets.DatasetInfo(
|
356 |
-
description=_CAPTION_CONTEST_DESCRIPTION + self.config.description,
|
357 |
-
features=datasets.Features(features),
|
358 |
-
homepage=self.config.url,
|
359 |
-
citation=self.config.citation
|
360 |
-
)
|
361 |
-
|
362 |
-
def _split_generators(self, dl_manager):
|
363 |
-
dl_dir = dl_manager.download_and_extract(self.config.data_url) or ""
|
364 |
-
self.images_dir = dl_manager.download_and_extract(_IMAGES_URL)
|
365 |
-
task_name = _get_task_name_from_data_url(self.config.data_url)
|
366 |
-
dl_dir = os.path.join(dl_dir, task_name)
|
367 |
-
|
368 |
-
return [
|
369 |
-
datasets.SplitGenerator(
|
370 |
-
name=datasets.Split.TRAIN,
|
371 |
-
gen_kwargs={
|
372 |
-
"data_file": os.path.join(dl_dir, "train.jsonl"),
|
373 |
-
"split": datasets.Split.TRAIN,
|
374 |
-
},
|
375 |
-
),
|
376 |
-
datasets.SplitGenerator(
|
377 |
-
name=datasets.Split.VALIDATION,
|
378 |
-
gen_kwargs={
|
379 |
-
"data_file": os.path.join(dl_dir, "val.jsonl"),
|
380 |
-
"split": datasets.Split.VALIDATION,
|
381 |
-
},
|
382 |
-
),
|
383 |
-
datasets.SplitGenerator(
|
384 |
-
name=datasets.Split.TEST,
|
385 |
-
gen_kwargs={
|
386 |
-
"data_file": os.path.join(dl_dir, "test.jsonl"),
|
387 |
-
"split": datasets.Split.TEST,
|
388 |
-
},
|
389 |
-
),
|
390 |
-
]
|
391 |
-
|
392 |
-
def _generate_examples(self, data_file, split):
|
393 |
-
with open(data_file, encoding="utf-8") as f:
|
394 |
-
for line in f:
|
395 |
-
row = json.loads(line)
|
396 |
-
with open(self.images_dir + "/all_contest_images/{}.jpeg".format(row['contest_number']), "rb") as image:
|
397 |
-
row['image'] = {"path": self.images_dir + "/all_contest_images/{}.jpeg".format(row['contest_number']),
|
398 |
-
"bytes": image.read()}
|
399 |
-
yield row['instance_id'], row
|
400 |
-
|
401 |
-
def _get_task_name_from_data_url(data_url):
|
402 |
-
return data_url.split("/")[-1].split(".")[0]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|