Datasets:

Modalities:
Text
Formats:
json
Size:
< 1K
ArXiv:
Tags:
RES-Q / README.md
becklabash's picture
Update README.md
c0e850b verified
metadata
dataset_info:
  splits:
    - name: test
      num_bytes: 4645427
      num_examples: 100
configs:
  - config_name: default
    data_files:
      - split: test
        path: dataset.json

Dataset Summary

RES-Q is a codebase editing benchmark based on compact, natural language instructions. The task is to, given an edit instruction and a codebase, produce a patch file that makes the correct edit to the codebase.

The dataset was released as part of RES-Q: Evaluating Code-Editing Large Language Model Systems at the Repository Scale

Dataset Structure

An example of a RES-Q task instance is as follows:

id (str) - A unique identifier for the task instance.
repo_url (str) - The URL of the repository involved in the task.
instruction (str) - The repository edit instruction.
base_commit (str) - The commit hash of the repository representing the HEAD of the repository immediately before the instruction was carried out.
test_script (str) - The task's test suite as a python script to be run from the root of the repository. 
testbed_environment (str) - The python version used to run the test suite. 
requirements_txt (str) - The pip package dependencies required to run the test suite.
solution_commit (str) - The commit hash of the repository representing the HEAD of the repository immediately after the instruction was carried out.
solution_patch (str) - The patch in the unified diff format representing the difference between the base and solution commit.
modified_files (list): A list of dictionaries containing the relative paths and content of files modified by the solution commit.
language (str) - The primary programming language of the repository.