File size: 15,425 Bytes
c2027a4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
# coding=utf-8
# Copyright 2022 The HuggingFace Datasets Authors and the current dataset script contributor.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

import json
import os
import re
import zipfile
from pathlib import Path
from typing import Dict, List, Tuple

import datasets

from seacrowd.utils import schemas
from seacrowd.utils.configs import SEACrowdConfig
from seacrowd.utils.constants import Licenses, Tasks

_CITATION = """\
@article{zhang2023m3exam,
      title={M3Exam: A Multilingual, Multimodal, Multilevel Benchmark for Examining Large Language Models},
      author={Wenxuan Zhang and Sharifah Mahani Aljunied and Chang Gao and Yew Ken Chia and Lidong Bing},
      year={2023},
      eprint={2306.05179},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
}
"""

_DATASETNAME = "m3exam"

_DESCRIPTION = """\
M3Exam is a novel benchmark sourced from real and official human exam questions for evaluating LLMs\
in a multilingual, multimodal, and multilevel context. In total, M3Exam contains 12,317 questions in 9\
diverse languages with three educational levels, where about 23% of the questions require processing images\
for successful solving. M3Exam dataset covers 3 languages spoken in Southeast Asia.
"""

_HOMEPAGE = "https://github.com/DAMO-NLP-SG/M3Exam"

_LANGUAGES = ["jav", "tha", "vie"]
_LANG_MAPPER = {"jav": "javanese", "tha": "thai", "vie": "vietnamese"}
_LICENSE = Licenses.CC_BY_NC_SA_4_0.value

_LOCAL = False
_PASSWORD = "12317".encode("utf-8")  # password to unzip dataset after downloading
_URLS = {
    _DATASETNAME: "https://drive.usercontent.google.com/download?id=1eREETRklmXJLXrNPTyHxQ3RFdPhq_Nes&authuser=0&confirm=t",
}

_SUPPORTED_TASKS = [Tasks.QUESTION_ANSWERING, Tasks.VISUAL_QUESTION_ANSWERING]

_SOURCE_VERSION = "1.0.0"

_SEACROWD_VERSION = "2024.06.20"


class M3ExamDataset(datasets.GeneratorBasedBuilder):
    """
    M3Exam is a novel benchmark sourced from real and official human exam questions for evaluating LLMs
    in a multilingual, multimodal, and multilevel context. In total, M3Exam contains 12,317 questions in 9
    diverse languages with three educational levels, where about 23% of the questions require processing images
    for successful solving. M3Exam dataset covers 3 languages spoken in Southeast Asia.
    """

    SOURCE_VERSION = datasets.Version(_SOURCE_VERSION)
    SEACROWD_VERSION = datasets.Version(_SEACROWD_VERSION)

    BUILDER_CONFIGS = (
        [SEACrowdConfig(name=f"{_DATASETNAME}_{lang}_source", version=datasets.Version(_SOURCE_VERSION), description=f"{_DATASETNAME} source schema", schema="source", subset_id=f"{_DATASETNAME}") for lang in _LANGUAGES]
        + [
            SEACrowdConfig(
                name=f"{_DATASETNAME}_{lang}_seacrowd_qa",
                version=datasets.Version(_SEACROWD_VERSION),
                description=f"{_DATASETNAME} SEACrowd schema",
                schema="seacrowd_qa",
                subset_id=f"{_DATASETNAME}",
            )
            for lang in _LANGUAGES
        ]
        + [
            SEACrowdConfig(
                name=f"{_DATASETNAME}_{lang}_seacrowd_imqa",
                version=datasets.Version(_SEACROWD_VERSION),
                description=f"{_DATASETNAME} SEACrowd schema",
                schema="seacrowd_imqa",
                subset_id=f"{_DATASETNAME}",
            )
            for lang in _LANGUAGES
        ]
    )

    DEFAULT_CONFIG_NAME = f"{_DATASETNAME}_jav_source"

    def _info(self) -> datasets.DatasetInfo:

        if self.config.schema == "source":
            features = datasets.Features(
                {
                    "question_text": datasets.Value("string"),
                    "background_description": datasets.Sequence(datasets.Value("string")),
                    "answer_text": datasets.Value("string"),
                    "options": datasets.Sequence(datasets.Value("string")),
                    "language": datasets.Value("string"),
                    "level": datasets.Value("string"),
                    "subject": datasets.Value("string"),
                    "subject_category": datasets.Value("string"),
                    "year": datasets.Value("string"),
                    "need_image": datasets.Value("string"),
                    "image_paths": datasets.Sequence(datasets.Value("string")),
                }
            )
        elif self.config.schema == "seacrowd_qa":
            features = schemas.qa_features
            features["meta"] = {
                "background_description": datasets.Sequence(datasets.Value("string")),
                "level": datasets.Value("string"),
                "subject": datasets.Value("string"),
                "subject_category": datasets.Value("string"),
                "year": datasets.Value("string"),
            }
        elif self.config.schema == "seacrowd_imqa":
            features = schemas.imqa_features
            features["meta"] = {
                "background_description": datasets.Sequence(datasets.Value("string")),
                "level": datasets.Value("string"),
                "subject": datasets.Value("string"),
                "subject_category": datasets.Value("string"),
                "year": datasets.Value("string"),
            }

        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=features,
            homepage=_HOMEPAGE,
            license=_LICENSE,
            citation=_CITATION,
        )

    def _split_generators(self, dl_manager: datasets.DownloadManager) -> List[datasets.SplitGenerator]:
        """Returns SplitGenerators."""
        urls = _URLS[_DATASETNAME]
        lang = self.config.name.split("_")[1]

        data_dir = dl_manager.download(urls)

        if not os.path.exists(data_dir + "_extracted"):
            if not os.path.exists(data_dir + ".zip"):
                os.rename(data_dir, data_dir + ".zip")
            with zipfile.ZipFile(data_dir + ".zip", "r") as zip_ref:
                zip_ref.extractall(data_dir + "_extracted", pwd=_PASSWORD)  # unzipping with password
        if not os.path.exists(data_dir):
            os.rename(data_dir + ".zip", data_dir)
        image_generator = [
            datasets.SplitGenerator(
                name=datasets.Split.TRAIN,
                gen_kwargs={
                    "filepath": os.path.join(data_dir + "_extracted", "data/multimodal-question"),
                    "split": "train",
                },
            ),
        ]

        text_generator = [
            datasets.SplitGenerator(
                name=datasets.Split.TEST,
                gen_kwargs={
                    "filepath": os.path.join(data_dir + "_extracted", f"data/text-question/{_LANG_MAPPER[lang]}-questions-test.json"),
                    "split": "test",
                },
            ),
            datasets.SplitGenerator(
                name=datasets.Split.VALIDATION,
                gen_kwargs={
                    "filepath": os.path.join(data_dir + "_extracted", f"data/text-question/{_LANG_MAPPER[lang]}-questions-dev.json"),
                    "split": "dev",
                },
            ),
        ]
        if "imqa" in self.config.name:
            return image_generator
        else:
            if "source" in self.config.name:
                image_generator.extend(text_generator)
                return image_generator
            else:
                return text_generator

    def _generate_examples(self, filepath: Path, split: str) -> Tuple[int, Dict]:
        """Yields examples as (key, example) tuples."""
        lang = self.config.name.split("_")[1]
        thai_answer_mapper = {"1": "1", "2": "2", "3": "3", "4": "4", "5": "5", "๑": "1", "๒": "2", "๓": "3", "๔": "4", "๕": "5"}
        if self.config.schema == "source":
            if split == "train":
                filepath_json = os.path.join(filepath, f"{_LANG_MAPPER[lang]}-questions-image.json")
                with open(filepath_json, "r") as file:
                    data = json.load(file)
                idx = 0
                for json_obj in data:
                    image_paths = []
                    for text in [json_obj["question_text"]] + json_obj["options"] + json_obj["background_description"]:
                        matches = re.findall(r"\[image-(\d+)\.(jpg|png)\]", text)
                        if matches:
                            image_path = [os.path.join(filepath, f"images-{_LANG_MAPPER[lang]}/image-{image_number[0]}.{image_number[1]}") for image_number in matches]
                            image_paths.extend(image_path)
                    example = {
                        "question_text": json_obj["question_text"],
                        "background_description": json_obj["background_description"] if "background_description" in json_obj.keys() else None,
                        "answer_text": json_obj["answer_text"],
                        "options": json_obj["options"],
                        "language": json_obj["language"] if "language" in json_obj.keys() else None,
                        "level": json_obj["level"] if "level" in json_obj.keys() else None,
                        "subject": json_obj["subject"] if "subject" in json_obj.keys() else None,
                        "subject_category": json_obj["subject_category"] if "subject_category" in json_obj.keys() else None,
                        "year": json_obj["year"] if "year" in json_obj.keys() else None,
                        "need_image": "yes",
                        "image_paths": image_paths,
                    }
                    yield idx, example
                    idx += 1
            else:
                with open(filepath, "r") as file:
                    data = json.load(file)
                idx = 0
                for json_obj in data:
                    example = {
                        "question_text": json_obj["question_text"],
                        "background_description": json_obj["background_description"] if "background_description" in json_obj.keys() else None,
                        "answer_text": json_obj["answer_text"],
                        "options": json_obj["options"],
                        "language": json_obj["language"] if "language" in json_obj.keys() else None,
                        "level": json_obj["level"] if "level" in json_obj.keys() else None,
                        "subject": json_obj["subject"] if "subject" in json_obj.keys() else None,
                        "subject_category": json_obj["subject_category"] if "subject_category" in json_obj.keys() else None,
                        "year": json_obj["year"] if "year" in json_obj.keys() else None,
                        "need_image": "no",
                        "image_paths": None,
                    }
                    yield idx, example
                    idx += 1

        elif self.config.schema == "seacrowd_qa":
            with open(filepath, "r") as file:
                data = json.load(file)
            idx = 0

            for json_obj in data:
                answer = [".".join(answer.split(".")[1:]).strip() for answer in json_obj["options"] if json_obj["answer_text"] == answer.split(".")[0]]
                if "_tha_" in self.config.name and len(answer) == 0:
                    answer = [".".join(answer.split(".")[1:]).strip() for answer in json_obj["options"] if thai_answer_mapper[json_obj["answer_text"]] == thai_answer_mapper[answer.split(".")[0]]]

                example = {
                    "id": idx,
                    "question_id": idx,
                    "document_id": idx,
                    "question": json_obj["question_text"],
                    "type": "multiple_choice",
                    "choices": [".".join(answer.split(".")[1:]).strip() for answer in json_obj["options"]],
                    "context": "",
                    "answer": answer,
                    "meta": {
                        "background_description": json_obj["background_description"] if "background_description" in json_obj.keys() else None,
                        "level": json_obj["level"] if "level" in json_obj.keys() else None,
                        "subject": json_obj["subject"] if "subject" in json_obj.keys() else None,
                        "subject_category": json_obj["subject_category"] if "subject_category" in json_obj.keys() else None,
                        "year": json_obj["year"] if "year" in json_obj.keys() else None,
                    },
                }
                yield idx, example
                idx += 1

        elif self.config.schema == "seacrowd_imqa":
            filepath_json = os.path.join(filepath, f"{_LANG_MAPPER[lang]}-questions-image.json")
            with open(filepath_json, "r") as file:
                data = json.load(file)
            idx = 0

            for json_obj in data:
                answer = [".".join(answer.split(".")[1:]).strip() for answer in json_obj["options"] if json_obj["answer_text"] == answer.split(".")[0]]
                if "_tha_" in self.config.name and len(answer) == 0:
                    answer = [".".join(answer.split(".")[1:]).strip() for answer in json_obj["options"] if thai_answer_mapper[json_obj["answer_text"]] == thai_answer_mapper[answer.split(".")[0]]]
                image_paths = []
                for text in [json_obj["question_text"]] + json_obj["options"] + json_obj["background_description"]:
                    matches = re.findall(r"\[image-(\d+)\.(jpg|png)\]", text)
                    if matches:
                        image_path = [os.path.join(filepath, f"images-{_LANG_MAPPER[lang]}/image-{image_number[0]}.{image_number[1]}") for image_number in matches]
                        image_paths.extend(image_path)

                example = {
                    "id": idx,
                    "question_id": idx,
                    "document_id": idx,
                    "questions": [json_obj["question_text"]],
                    "type": "multiple_choice",
                    "choices": [".".join(answer.split(".")[1:]).strip() for answer in json_obj["options"]],
                    "context": "",
                    "answer": answer,
                    "image_paths": image_paths,
                    "meta": {
                        "background_description": json_obj["background_description"] if "background_description" in json_obj.keys() else None,
                        "level": json_obj["level"] if "level" in json_obj.keys() else None,
                        "subject": json_obj["subject"] if "subject" in json_obj.keys() else None,
                        "subject_category": json_obj["subject_category"] if "subject_category" in json_obj.keys() else None,
                        "year": json_obj["year"] if "year" in json_obj.keys() else None,
                    },
                }
                yield idx, example
                idx += 1