shuyanzh commited on
Commit
62e85e8
1 Parent(s): 68f25c6

Create README.md

Browse files
Files changed (1) hide show
  1. README.md +97 -0
README.md ADDED
@@ -0,0 +1,97 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ annotations_creators: []
3
+ language_creators:
4
+ - crowdsourced
5
+ - expert-generated
6
+ language:
7
+ - code
8
+ license:
9
+ - mit
10
+ multilinguality:
11
+ - monolingual
12
+ size_categories:
13
+ - unknown
14
+ source_datasets:
15
+ - original
16
+ task_categories:
17
+ - text2text-generation
18
+ task_ids: []
19
+ pretty_name: DocPrompting-CoNaLa
20
+ tags:
21
+ - code-generation
22
+ - doc retrieval
23
+ - retrieval augmented generation
24
+ ---
25
+
26
+ ## Dataset Description
27
+ - **Repository:** https://github.com/shuyanzhou/docprompting
28
+ - **Paper:** [DocPrompting: Generating Code by Retrieving the Docs](https://arxiv.org/pdf/2207.05987.pdf)
29
+
30
+ ### Dataset Summary
31
+ This is the natural language to bash generation dataset we harvested from the English subset of [`tldr`](https://github.com/tldr-pages/tldr)
32
+ We split the dataset by bash commands. Every command in the dev and test set is held out from the training set.
33
+
34
+ ### Supported Tasks and Leaderboards
35
+ This dataset is used to evaluate code generations.
36
+
37
+ ### Languages
38
+ English - Bash
39
+
40
+ ## Dataset Structure
41
+ ```python
42
+ dataset = load_dataset("neulab/tldr")
43
+ DatasetDict({
44
+ train: Dataset({
45
+ features: ['question_id', 'nl', 'cmd', 'oracle_man', 'cmd_name', 'tldr_cmd_name', 'manual_exist', 'matching_info'],
46
+ num_rows: 6414
47
+ })
48
+ test: Dataset({
49
+ features: ['question_id', 'nl', 'cmd', 'oracle_man', 'cmd_name', 'tldr_cmd_name', 'manual_exist', 'matching_info'],
50
+ num_rows: 928
51
+ })
52
+ validation: Dataset({
53
+ features: ['question_id', 'nl', 'cmd', 'oracle_man', 'cmd_name', 'tldr_cmd_name', 'manual_exist', 'matching_info'],
54
+ num_rows: 1845
55
+ })
56
+ })
57
+
58
+ code_docs = load_dataset("neulab/docprompting-conala", "docs")
59
+ DatasetDict({
60
+ train: Dataset({
61
+ features: ['doc_id', 'doc_content'],
62
+ num_rows: 439064
63
+ })
64
+ })
65
+ ```
66
+
67
+ ### Data Fields
68
+ train/dev/test:
69
+ - nl: The natural language intent
70
+ - cmd: The reference code snippet
71
+ - question_id: the unique id of a question
72
+ - oracle_man: The `doc_id` of the functions used in the reference code snippet. The corresponding contents are in `doc` split
73
+ - cmd_name: the bash command of this code snippet
74
+ - tldr_cmd_name: the bash command used in tldr github repo. The `cmd_name` and `tldr_cmd_name` can be different due to naming difference
75
+ - manual_exist: whether the manual exists in https://manned.org
76
+ - matching_info: each code snippets have multiple tokens, this is the detailed reference doc matching on each token.
77
+
78
+
79
+ docs:
80
+ - doc_id: the id of a doc
81
+ - doc_content: the content of the doc
82
+
83
+ ## Dataset Creation
84
+ The dataset was curated from [`tldr`](https://github.com/tldr-pages/tldr).
85
+ The project aims to provide frequent usage of bash commands with natural language intents.
86
+ For more details, please check the repo.
87
+
88
+ ### Citation Information
89
+
90
+ ```
91
+ @article{zhou2022doccoder,
92
+ title={DocCoder: Generating Code by Retrieving and Reading Docs},
93
+ author={Zhou, Shuyan and Alon, Uri and Xu, Frank F and Jiang, Zhengbao and Neubig, Graham},
94
+ journal={arXiv preprint arXiv:2207.05987},
95
+ year={2022}
96
+ }
97
+ ```