loubnabnl HF staff commited on
Commit
1da1e81
1 Parent(s): d108334

add readme and loading script

Browse files
Files changed (2) hide show
  1. README.md +137 -0
  2. conala.py +111 -0
README.md ADDED
@@ -0,0 +1,137 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ annotations_creators: []
3
+ language_creators:
4
+ - crowdsourced
5
+ - expert-generated
6
+ language: ["code"]
7
+ multilinguality:
8
+ - monolingual
9
+ pretty_name: CoNaLa
10
+ size_categories:
11
+ - unknown
12
+ source_datasets:
13
+ - original
14
+ task_categories:
15
+ - text2text-generation
16
+ task_ids:
17
+ - text2text-generation-other-code-generation
18
+ ---
19
+
20
+ # CoNaLa Dataset
21
+
22
+ ## Dataset Description
23
+
24
+ - **Repository:** https://conala-corpus.github.io/
25
+ - **Paper:** [Learning to Mine Aligned Code and
26
+ Natural Language Pairs from Stack Overflow](https://arxiv.org/pdf/1805.08949.pdf)
27
+
28
+ ### Dataset Summary
29
+ [CoNaLa](https://conala-corpus.github.io/) is a benchmark of code and natural language pairs, for the evaluation of code generation tasks. The dataset was crawled from Stack Overflow, automatically filtered, then curated by annotators, split into 2,379 training and 500 test examples, consisting of code and natural language pairs. The automatically mined dataset is also available with almost 600k examples.
30
+
31
+
32
+ ### Supported Tasks and Leaderboards
33
+ This dataset is used to evaluate code generations.
34
+
35
+ ### Languages
36
+ English - Python code.
37
+
38
+ ## Dataset Structure
39
+ ```python
40
+ dataset_curated = load_dataset("conala")
41
+ DatasetDict({
42
+ train: Dataset({
43
+ features: ['question_id', 'intent', 'rewritten_intent', 'snippet'],
44
+ num_rows: 2379
45
+ })
46
+ test: Dataset({
47
+ features: ['question_id', 'intent', 'rewritten_intent', 'snippet'],
48
+ num_rows: 500
49
+ })
50
+ })
51
+
52
+ dataset_mined = load_dataset("conala", "mined")
53
+ DatasetDict({
54
+ train: Dataset({
55
+ features: ['question_id', 'parent_answer_post_id', 'prob', 'snippet', 'intent', 'id'],
56
+ num_rows: 593891
57
+ })
58
+ })
59
+ ```
60
+ ### Data Instances
61
+
62
+ #### CoNaLa - curated
63
+ This is the curated dataset by annotators
64
+ ```
65
+ {
66
+ 'question_id': 41067960,
67
+ 'intent': 'How to convert a list of multiple integers into a single integer?',
68
+ 'rewritten_intent': "Concatenate elements of a list 'x' of multiple integers to a single integer",
69
+ 'snippet': 'sum(d * 10 ** i for i, d in enumerate(x[::-1]))'
70
+ }
71
+ ```
72
+
73
+ #### CoNaLa - mined
74
+ This is the automatically mined dataset before curation
75
+ ```
76
+ {
77
+ 'question_id': 34705205,
78
+ 'parent_answer_post_id': 34705233,
79
+ 'prob': 0.8690001442846342,
80
+ 'snippet': 'sorted(l, key=lambda x: (-int(x[1]), x[0]))',
81
+ 'intent': 'Sort a nested list by two elements',
82
+ 'id': '34705205_34705233_0'
83
+ }
84
+ ```
85
+
86
+ ### Data Fields
87
+ Curated:
88
+
89
+ |Field|Type|Description|
90
+ |---|---|---|
91
+ |question_id|int64|Id of the Stack Overflow question|
92
+ |intent|string|Natural Language intent (i.e., the title of a Stack Overflow question)|
93
+ |rewritten_intent|string|Crowdsourced revised intents that try to better reflect the full meaning of the code|
94
+ |snippet|string| Code snippet that implements the intent|
95
+
96
+ Mined:
97
+
98
+ |Field|Type|Description|
99
+ |---|---|---|
100
+ |question_id|int64|Id of the Stack Overflow question|
101
+ |parent_answer_post_id|int64|Id of the answer post from which the candidate snippet is extracted|
102
+ |intent|string|Natural Language intent (i.e., the title of a Stack Overflow question)|
103
+ |snippet|string| Code snippet that implements the intent|
104
+ |id|string|Unique id for this intent/snippet pair|
105
+ |prob|float64|Probability given by the mining model|
106
+
107
+ ### Data Splits
108
+ There are two version of the dataset (curated and mined), mined only has a train split and curated has two splits: train and test.
109
+
110
+ ### Dataset Creation
111
+ The dataset was crawled from Stack Overflow, automatically filtered, then curated by annotators. For more details, please refer to the original [paper](https://arxiv.org/pdf/1805.08949.pdf)
112
+
113
+ ### Data Fields
114
+
115
+ |Field|Type|Description|
116
+ |---|---|---|
117
+ |question_id|int64|id of the Stack Overflow question|
118
+ |intent|string|Natural Language intent (i.e., the title of a Stack Overflow question)|
119
+ |rewritten_intent|string|Crowdsourced revised intents that try to better reflect the full meaning of the code|
120
+ |snippet|string| Code snippet that implements the intent|
121
+
122
+ ### Data Splits
123
+
124
+ The dataset contains a train and test splits with 2379 and 500 samples respectively.
125
+
126
+ ### Citation Information
127
+
128
+ ```
129
+ @inproceedings{yin2018learning,
130
+ title={Learning to mine aligned code and natural language pairs from stack overflow},
131
+ author={Yin, Pengcheng and Deng, Bowen and Chen, Edgar and Vasilescu, Bogdan and Neubig, Graham},
132
+ booktitle={2018 IEEE/ACM 15th international conference on mining software repositories (MSR)},
133
+ pages={476--486},
134
+ year={2018},
135
+ organization={IEEE}
136
+ }
137
+ ```
conala.py ADDED
@@ -0,0 +1,111 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # coding=utf-8
2
+ # Copyright 2020 The HuggingFace Datasets Authors and the current dataset script contributor.
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
+ """CoNaLa dataset."""
16
+
17
+ import json
18
+ import datasets
19
+
20
+
21
+ _CITATION = """\
22
+ @inproceedings{yin2018learning,
23
+ title={Learning to mine aligned code and natural language pairs from stack overflow},
24
+ author={Yin, Pengcheng and Deng, Bowen and Chen, Edgar and Vasilescu, Bogdan and Neubig, Graham},
25
+ booktitle={2018 IEEE/ACM 15th international conference on mining software repositories (MSR)},
26
+ pages={476--486},
27
+ year={2018},
28
+ organization={IEEE}
29
+ }
30
+ """
31
+
32
+ _DESCRIPTION = """\
33
+ CoNaLa is a dataset of code and natural language pairs crawled from Stack Overflow, for more details please refer to this paper: https://arxiv.org/pdf/1805.08949.pdf or the dataset page https://conala-corpus.github.io/.
34
+ """
35
+
36
+ _HOMEPAGE = "https://conala-corpus.github.io/"
37
+ _URLs = {
38
+ "mined": "data/conala-mined.json",
39
+ "curated": {"train": "data/conala-paired-train.json", "test": "data/conala-paired-test.json" },
40
+ }
41
+
42
+ class Conala(datasets.GeneratorBasedBuilder):
43
+ """CoNaLa Code dataset."""
44
+
45
+ VERSION = datasets.Version("1.1.0")
46
+
47
+
48
+ BUILDER_CONFIGS = [
49
+ datasets.BuilderConfig(
50
+ name="curated",
51
+ version=datasets.Version("1.1.0"),
52
+ description=_DESCRIPTION,
53
+ ),
54
+ datasets.BuilderConfig(name="mined", version=datasets.Version("1.1.0"), description=_DESCRIPTION),
55
+ ]
56
+
57
+ DEFAULT_CONFIG_NAME = "curated"
58
+
59
+
60
+ def _info(self):
61
+ if self.config.name == "curated":
62
+ features=datasets.Features({"question_id": datasets.Value("int64"),
63
+ "intent": datasets.Value("string"),
64
+ "rewritten_intent": datasets.Value("string"),
65
+ "snippet": datasets.Value("string"),
66
+ })
67
+ else:
68
+ features=datasets.Features({"question_id": datasets.Value("int64"),
69
+ "parent_answer_post_id": datasets.Value("int64"),
70
+ "prob": datasets.Value("float64"),
71
+ "snippet": datasets.Value("string"),
72
+ "intent": datasets.Value("string"),
73
+ "id": datasets.Value("string"),
74
+ })
75
+ return datasets.DatasetInfo(
76
+ description=_DESCRIPTION,
77
+ features=features,
78
+ supervised_keys=None,
79
+ citation=_CITATION,
80
+ homepage=_HOMEPAGE)
81
+
82
+ def _split_generators(self, dl_manager):
83
+ """Returns SplitGenerators."""
84
+ config_urls = _URLs[self.config.name]
85
+ data_dir = dl_manager.download_and_extract(config_urls)
86
+ if self.config.name == "curated":
87
+ return [
88
+ datasets.SplitGenerator(
89
+ name=datasets.Split.TRAIN,
90
+ gen_kwargs={"filepath": data_dir["train"], "split": "train"},
91
+ ),
92
+ datasets.SplitGenerator(
93
+ name=datasets.Split.TEST,
94
+ gen_kwargs={"filepath": data_dir["test"], "split": "test"},
95
+ ),
96
+ ]
97
+ else:
98
+ return [
99
+ datasets.SplitGenerator(
100
+ name=datasets.Split.TRAIN,
101
+ gen_kwargs={"filepath": data_dir, "split": "train"},
102
+ ),
103
+ ]
104
+
105
+
106
+ def _generate_examples(self, filepath, split):
107
+ key = 0
108
+ for line in open(filepath, encoding="utf-8"):
109
+ line = json.loads(line)
110
+ yield key, line
111
+ key += 1