ikala-ray commited on
Commit
079214d
1 Parent(s): 8a07398

Revert back to python script based loading, cause hf auto builder is broken

Browse files
Files changed (1) hide show
  1. tmmluplus.py +124 -0
tmmluplus.py ADDED
@@ -0,0 +1,124 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
+ import os
15
+
16
+ import datasets
17
+ import pandas as pd
18
+
19
+
20
+ _DESCRIPTION = """\
21
+ TMMLU2 data loader
22
+ """
23
+ _DATA_PATH = "data"
24
+
25
+ task_list = [
26
+ 'dentistry', 'traditional_chinese_medicine_clinical_medicine', 'clinical_psychology',
27
+ 'technical', 'culinary_skills', 'mechanical', 'logic_reasoning', 'real_estate',
28
+ 'general_principles_of_law', 'finance_banking', 'anti_money_laundering', 'ttqav2',
29
+ 'marketing_management', 'business_management', 'organic_chemistry', 'advance_chemistry',
30
+ 'physics', 'secondary_physics', 'human_behavior', 'national_protection', 'jce_humanities',
31
+ 'politic_science', 'agriculture', 'official_document_management',
32
+ 'financial_analysis', 'pharmacy', 'educational_psychology', 'statistics_and_machine_learning',
33
+ 'management_accounting', 'introduction_to_law', 'computer_science', 'veterinary_pathology',
34
+ 'accounting', 'fire_science', 'optometry', 'insurance_studies', 'pharmacology', 'taxation',
35
+ 'education_(profession_level)', 'economics',
36
+ 'veterinary_pharmacology', 'nautical_science', 'occupational_therapy_for_psychological_disorders',
37
+ 'trust_practice', 'geography_of_taiwan', 'physical_education', 'auditing', 'administrative_law',
38
+ 'basic_medical_science', 'macroeconomics', 'trade', 'chinese_language_and_literature',
39
+ 'tve_design', 'junior_science_exam', 'junior_math_exam', 'junior_chinese_exam',
40
+ 'junior_social_studies', 'tve_mathematics', 'tve_chinese_language',
41
+ 'tve_natural_sciences', 'junior_chemistry', 'music', 'education',
42
+ 'three_principles_of_people', 'taiwanese_hokkien',
43
+ 'engineering_math'
44
+ ]
45
+
46
+ _URLs = {
47
+ task_name: {
48
+ split_name: [
49
+ os.path.join(
50
+ _DATA_PATH, task_name+"_"+split_name+".csv"
51
+ ), # TODO -> handle multiple shards
52
+ ]
53
+ for split_name in ['dev', 'test', 'val']
54
+ }
55
+ for task_name in task_list
56
+ }
57
+
58
+
59
+ class TMMLU2Config(datasets.BuilderConfig):
60
+ def __init__(self, **kwargs):
61
+ super().__init__(version=datasets.Version("1.0.0"), **kwargs)
62
+
63
+
64
+ class TMMLU2(datasets.GeneratorBasedBuilder):
65
+ BUILDER_CONFIGS = [
66
+ TMMLU2Config(
67
+ name=task_name,
68
+ )
69
+ for task_name in task_list
70
+ ]
71
+
72
+ def _info(self):
73
+ features = datasets.Features(
74
+ {
75
+ "question": datasets.Value("string"),
76
+ "A": datasets.Value("string"),
77
+ "B": datasets.Value("string"),
78
+ "C": datasets.Value("string"),
79
+ "D": datasets.Value("string"),
80
+ "answer": datasets.Value("string"),
81
+ }
82
+ )
83
+ return datasets.DatasetInfo(
84
+ description=_DESCRIPTION,
85
+ features=features,
86
+ )
87
+
88
+ def _split_generators(self, dl_manager):
89
+ task_name = self.config.name
90
+ data_dir = dl_manager.download(_URLs[task_name])
91
+ return [
92
+ datasets.SplitGenerator(
93
+ name=datasets.Split.TEST,
94
+ gen_kwargs={
95
+ "filepath": data_dir['test'],
96
+ },
97
+ ),
98
+ datasets.SplitGenerator(
99
+ name=datasets.Split.VALIDATION,
100
+ gen_kwargs={
101
+ "filepath": data_dir['val'],
102
+ },
103
+ ),
104
+ datasets.SplitGenerator(
105
+ name=datasets.Split.TRAIN,
106
+ gen_kwargs={
107
+ "filepath": data_dir['dev'],
108
+ },
109
+ ),
110
+ ]
111
+
112
+ def _generate_examples(self, filepath):
113
+ if isinstance(filepath, list):
114
+ filepath = filepath[0]
115
+ df = pd.read_csv(filepath)
116
+
117
+ for i, instance in enumerate(df.to_dict(orient="records")):
118
+ yield i, {'question': instance['question'],
119
+ 'A': instance['A'],
120
+ 'B': instance['B'],
121
+ 'C': instance['C'],
122
+ 'D': instance['D'],
123
+ 'answer': instance['answer']
124
+ }