Datasets:

Modalities:
Text
ArXiv:
Libraries:
Datasets
License:
RaymondLi commited on
Commit
b06d231
·
1 Parent(s): b156158

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +20 -3
README.md CHANGED
@@ -15,6 +15,10 @@ The Recode benchmark proposes to apply code and natural language transformations
15
  This dataset contains the perturbed version of HumanEval that they released.
16
  It was automatically generated from the [HumanEval](https://huggingface.co/datasets/openai_humaneval) dataset.
17
 
 
 
 
 
18
  ### Languages
19
 
20
  The programming problems are written in Python and contains docstrings and comments in English.
@@ -27,11 +31,19 @@ The programming problems are written in Python and contains docstrings and comme
27
 
28
  ### Data Fields
29
 
30
- [More Information Needed]
 
 
 
 
 
 
 
 
31
 
32
  ### Data Splits
33
 
34
- [More Information Needed]
35
 
36
  ## Dataset Creation
37
 
@@ -89,7 +101,12 @@ The programming problems are written in Python and contains docstrings and comme
89
 
90
  ### Citation Information
91
 
92
- [More Information Needed]
 
 
 
 
 
93
 
94
  ### Contributions
95
 
 
15
  This dataset contains the perturbed version of HumanEval that they released.
16
  It was automatically generated from the [HumanEval](https://huggingface.co/datasets/openai_humaneval) dataset.
17
 
18
+ ### Subsets
19
+
20
+ There are four transformation categories that form the subsets of this dataset: `func_name`, `nlaugmenter`, `natgen` and `format`.
21
+
22
  ### Languages
23
 
24
  The programming problems are written in Python and contains docstrings and comments in English.
 
31
 
32
  ### Data Fields
33
 
34
+ - `task_id`: ID of the original HumanEval example
35
+ - `prompt`: the perturbed prompt
36
+ - `entry_point`: entry point for test
37
+ - `canonical_solution`: solution for the problem in the `prompt`
38
+ - `test`: contains function to test generated code for correctness
39
+ - `seed`: seed of the perturbed prompt
40
+ - `perturbation_name`: name of the perturbation
41
+ - `partial`: partial solution to the problem. This field is only present for transformation categories that affect a partial solution: `natgen` and `format`.
42
+
43
 
44
  ### Data Splits
45
 
46
+ The dataset only has a test split.
47
 
48
  ## Dataset Creation
49
 
 
101
 
102
  ### Citation Information
103
 
104
+ @article{wang2022recode,
105
+ title={ReCode: Robustness Evaluation of Code Generation Models},
106
+ author={Wang, Shiqi and Li, Zheng and Qian, Haifeng and Yang, Chenghao and Wang, Zijian and Shang, Mingyue and Kumar, Varun and Tan, Samson and Ray, Baishakhi and Bhatia, Parminder and others},
107
+ journal={arXiv preprint arXiv:2212.10264},
108
+ year={2022}
109
+ }
110
 
111
  ### Contributions
112