metadata
configs:
- config_name: default
data_files:
- split: dev
path: data/dev*
- split: test
path: data/test*
- split: valid
path: data/valid*
Languages
The text of the dataset is primarily English, but we make no effort to filter or otherwise clean based on language type.
Data Instances
An example of a BenchDyne_Seed datum is as follows:
instance_id: (str) - A formatted instance id
repo: (str) - The repository name
base_commit: (str) - The base commit hash
url: (str) - The URL of the repository
lang: (str) - The programming language of the repository
bug_description: (str) - The bug description
build_sh: (str) - The build script to build the project
dockerfile: (str) - The Dockerfile
work_dir: (str) - The working directory, usually the root of the repository
additional_files: (list) - A list of additional files that can be used for building a Docker image