README.md

metadata

annotations_creators: []
language_creators:
  - found
languages:
  - Python, Java
licenses:
  - cc-by-nc-nd-4.0
multilinguality:
  - multilingual
pretty_name: RepoBench-Retrieval
size_categories:
  - unknown
source_datasets:
  - original
task_categories:
  - text-retrieval
task_ids:
  - document-retrieval

Dataset Card for RepoBench-R

Table of Contents

• Dataset Card for RepoBench-R
  • Table of Contents
  • Dataset Description
    • Dataset Summary
    • Supported Tasks
  • Dataset Structure
  • Dataset Creation
    • Curation Rationale
    • Source Data
      • Initial Data Collection and Normalization
      • Who are the source language producers?
    • Annotations
      • Annotation process
      • Who are the annotators?
    • Personal and Sensitive Information
  • Considerations for Using the Data
    • Social Impact of Dataset
    • Discussion of Biases
    • Other Known Limitations
  • Additional Information
    • Dataset Curators
    • Licensing Information
    • Citation Information
    • Contributions

Dataset Description

• Homepage: https://github.com/Leolty/repobench
• Paper: https://arxiv.org/abs/2306.03091
• Point of Contact: til040@ucsd.edu

Dataset Summary

RepoBench-R is a subtask of RepoBench, targeting the retrieval component of a repository-level auto-completion system, focusing on extracting the most relevant code snippet from a project repository for next-line code prediction.

Supported Tasks

The dataset supports two programming languages, Python and Java, and contains two settings:

• cff: short for cross_file_first, where the cross-file module in next line is first used in the current file.
• cfr: short for cross_file_random, where the cross-file module in next line is not first used in the current file.

For each setting, we provide two levels of difficulty: easy and hard. Suppose the number of code snippets in the context is $k$, then for the easy subset, we have $5 \leq k < 10$, while for the hard subset, we have $k \geq 10$.

So, the specific information of each split is as follows:

• python_cff:
  • easy: 8,000 samples
  • hard: 4,000 samples
• python_cfr:
  • easy: 8,000 samples
  • hard: 4,000 samples
• java_cff:
  • easy: 8,000 samples
  • hard: 4,000 samples
• java_cfr:
  • easy: 8,000 samples
  • hard: 4,000 samples

Dataset Structure

Dataset Creation

Curation Rationale

[More Information Needed]

Source Data

Initial Data Collection and Normalization

[More Information Needed]

Who are the source language producers?

[More Information Needed]

Annotations

Annotation process

[More Information Needed]

Who are the annotators?

[More Information Needed]

Personal and Sensitive Information

[More Information Needed]

Considerations for Using the Data

Social Impact of Dataset

[More Information Needed]

Discussion of Biases

[More Information Needed]

Other Known Limitations

[More Information Needed]

Additional Information

Dataset Curators

[More Information Needed]

Licensing Information

[More Information Needed]

Citation Information

[More Information Needed]

Contributions

Thanks to @github-username for adding this dataset.