Commit
•
132a562
1
Parent(s):
3c4e300
Delete vasr.py
Browse files
vasr.py
DELETED
|
@@ -1,128 +0,0 @@
|
|
| 1 |
-
# Copyright 2020 The HuggingFace Datasets Authors and the current dataset script contributor.
|
| 2 |
-
#
|
| 3 |
-
# Licensed under the Apache License, Version 2.0 (the "License");
|
| 4 |
-
# you may not use this file except in compliance with the License.
|
| 5 |
-
# You may obtain a copy of the License at
|
| 6 |
-
#
|
| 7 |
-
# http://www.apache.org/licenses/LICENSE-2.0
|
| 8 |
-
#
|
| 9 |
-
# Unless required by applicable law or agreed to in writing, software
|
| 10 |
-
# distributed under the License is distributed on an "AS IS" BASIS,
|
| 11 |
-
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
| 12 |
-
# See the License for the specific language governing permissions and
|
| 13 |
-
# limitations under the License.
|
| 14 |
-
""" VASR Loading Script """
|
| 15 |
-
|
| 16 |
-
import json
|
| 17 |
-
import os
|
| 18 |
-
import pandas as pd
|
| 19 |
-
import datasets
|
| 20 |
-
from huggingface_hub import hf_hub_url
|
| 21 |
-
|
| 22 |
-
# Find for instance the citation on arxiv or on the dataset repo/website
|
| 23 |
-
_CITATION = """
|
| 24 |
-
"""
|
| 25 |
-
|
| 26 |
-
_DESCRIPTION = """\
|
| 27 |
-
VASR is a challenging dataset for evaluating computer vision commonsense reasoning abilities. Given a triplet of images, the task is to select an image candidate B' that completes the analogy (A to A' is like B to what?). Unlike previous work on visual analogy that focused on simple image transformations, we tackle complex analogies requiring understanding of scenes. Our experiments demonstrate that state-of-the-art models struggle with carefully chosen distractors (±53%, compared to 90% human accuracy).
|
| 28 |
-
"""
|
| 29 |
-
|
| 30 |
-
_HOMEPAGE = "https://vasr-dataset.github.io/"
|
| 31 |
-
|
| 32 |
-
_LICENSE = "https://creativecommons.org/licenses/by/4.0/"
|
| 33 |
-
|
| 34 |
-
_URL = "https://huggingface.co/datasets/nlphuji/vasr/blob/main"
|
| 35 |
-
_URLS = {
|
| 36 |
-
"train": os.path.join(_URL, "train_gold.csv"),
|
| 37 |
-
"dev": os.path.join(_URL, "dev_gold.csv"),
|
| 38 |
-
"test": os.path.join(_URL, "test_gold.csv"),
|
| 39 |
-
}
|
| 40 |
-
|
| 41 |
-
|
| 42 |
-
class Vasr(datasets.GeneratorBasedBuilder):
|
| 43 |
-
VERSION = datasets.Version("1.1.0")
|
| 44 |
-
|
| 45 |
-
# If you need to make complex sub-parts in the datasets with configurable options
|
| 46 |
-
# You can create your own builder configuration class to store attribute, inheriting from datasets.BuilderConfig
|
| 47 |
-
# BUILDER_CONFIG_CLASS = MyBuilderConfig
|
| 48 |
-
|
| 49 |
-
# You will be able to load one or the other configurations in the following list with
|
| 50 |
-
# data = datasets.load_dataset('vasr', 'test')
|
| 51 |
-
BUILDER_CONFIGS = [
|
| 52 |
-
datasets.BuilderConfig(name="plain_text", version=VERSION, description="vasr dataset")
|
| 53 |
-
]
|
| 54 |
-
IMAGE_EXTENSION = "jpg"
|
| 55 |
-
|
| 56 |
-
def _info(self):
|
| 57 |
-
features = datasets.Features(
|
| 58 |
-
{
|
| 59 |
-
"A_img": [datasets.Value("string")],
|
| 60 |
-
"B_img": [datasets.Value("string")],
|
| 61 |
-
"C_img": [datasets.Value("string")],
|
| 62 |
-
"candidates": [datasets.Value("string")],
|
| 63 |
-
"label": datasets.Value("int64"),
|
| 64 |
-
"D_img": [datasets.Value("string")],
|
| 65 |
-
"A_verb": [datasets.Value("string")],
|
| 66 |
-
"B_verb": [datasets.Value("string")],
|
| 67 |
-
"C_verb": [datasets.Value("string")],
|
| 68 |
-
"D_verb": [datasets.Value("string")],
|
| 69 |
-
"diff_item_A": [datasets.Value("string")],
|
| 70 |
-
"diff_item_A_str_first": [datasets.Value("string")],
|
| 71 |
-
"diff_item_B": [datasets.Value("string")],
|
| 72 |
-
"diff_item_B_str_first": [datasets.Value("string")],
|
| 73 |
-
}
|
| 74 |
-
)
|
| 75 |
-
return datasets.DatasetInfo(
|
| 76 |
-
# This is the description that will appear on the datasets page.
|
| 77 |
-
description=_DESCRIPTION,
|
| 78 |
-
# This defines the different columns of the dataset and their types
|
| 79 |
-
features=features, # Here we define them above because they are different between the two configurations
|
| 80 |
-
# If there's a common (input, target) tuple from the features, uncomment supervised_keys line below and
|
| 81 |
-
# specify them. They'll be used if as_supervised=True in builder.as_dataset.
|
| 82 |
-
# supervised_keys=("sentence", "label"),
|
| 83 |
-
# Homepage of the dataset for documentation
|
| 84 |
-
homepage=_HOMEPAGE,
|
| 85 |
-
# License for the dataset if available
|
| 86 |
-
license=_LICENSE,
|
| 87 |
-
# Citation for the dataset
|
| 88 |
-
citation=_CITATION,
|
| 89 |
-
)
|
| 90 |
-
|
| 91 |
-
def _split_generators(self, dl_manager):
|
| 92 |
-
# If several configurations are possible (listed in BUILDER_CONFIGS), the configuration selected by the user is in self.config.name
|
| 93 |
-
downloaded_files = dl_manager.download_and_extract(_URLS)
|
| 94 |
-
|
| 95 |
-
# dl_manager is a datasets.download.DownloadManager that can be used to download and extract URLS
|
| 96 |
-
# It can accept any type or nested list/dict and will give back the same structure with the url replaced with path to local files.
|
| 97 |
-
# By default the archives will be extracted and a path to a cached folder where they are extracted is returned instead of the archive
|
| 98 |
-
data_dir = dl_manager.download_and_extract({
|
| 99 |
-
"images_dir": hf_hub_url("datasets/nlphuji/vasr", filename="vasr_images.zip")
|
| 100 |
-
})
|
| 101 |
-
|
| 102 |
-
# return [datasets.SplitGenerator(name=datasets.Split.TEST, gen_kwargs=data_dir)]
|
| 103 |
-
return [
|
| 104 |
-
datasets.SplitGenerator(name=datasets.Split.TEST,
|
| 105 |
-
gen_kwargs={**data_dir, **{'filepath': downloaded_files["test"]}}),
|
| 106 |
-
datasets.SplitGenerator(name=datasets.Split.TRAIN,
|
| 107 |
-
gen_kwargs={**data_dir, **{'filepath': downloaded_files["train"]}}),
|
| 108 |
-
datasets.SplitGenerator(name=datasets.Split.VALIDATION,
|
| 109 |
-
gen_kwargs={**data_dir, **{'filepath': downloaded_files["dev"]}}),
|
| 110 |
-
]
|
| 111 |
-
|
| 112 |
-
# method parameters are unpacked from `gen_kwargs` as given in `_split_generators`
|
| 113 |
-
def _generate_examples(self, examples_csv, images_dir):
|
| 114 |
-
# The `key` is for legacy reasons (tfds) and is not important in itself, but must be unique for each example.
|
| 115 |
-
|
| 116 |
-
df = pd.read_csv(examples_csv)
|
| 117 |
-
|
| 118 |
-
# columns_to_serialize = ['candidates', 'associations']
|
| 119 |
-
# for c in columns_to_serialize:
|
| 120 |
-
# df[c] = df[c].apply(json.loads)
|
| 121 |
-
|
| 122 |
-
for r_idx, r in df.iterrows():
|
| 123 |
-
r_dict = r.to_dict()
|
| 124 |
-
r_dict['candidates'] = json.loads(r_dict['candidates'])
|
| 125 |
-
candidates_images = [os.path.join(images_dir, "vasr_images", f"{x}.{self.IMAGE_EXTENSION}") for x in
|
| 126 |
-
r_dict['candidates']]
|
| 127 |
-
r_dict['candidates_images'] = candidates_images
|
| 128 |
-
yield r_idx, r_dict
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|