PR4Code: A Pull Requests Dataset for AI Code Generation

arXiv cs.SE2026-11-02 13:000 views

arXiv:2610.39890v2 Announce Type: replace Abstract: AI-driven code generation is evolving from handling small, localized code-completion tasks toward supporting the implementation of full software features. To support this transition, this paper introduces PR4Code, a dataset of 4,508 Java and 8,831 Python curated Pull Requests (PRs) collected from GitHub, each enriched with metadata, commit histories, and detailed code changes. Unlike resources focused on isolated code fragments, PR4Code captures the feature implementation process, delivering a large and diverse collection of real-world development tasks that can be used to assess AI-driven solutions. Our analysis reveals substantial variability in PR structure, commit granularity, and textual content. In addition, we release scripts to regenerate and update the dataset, ensuring reproducibility and maintainability.

Source：https://arxiv.org/abs/2610.39890

My favorites

PreviousSTORY2DIFF: A Dataset Linking Real-World User Stories to Code Changes NextCodeLift: Your Test Suite Is Secretly a Dense SWE Curriculum

The Substrate Trap: Why Enterprise AI Use Outruns Its EBIT Impact

A lot of enterprise AI pilots stall below the model, in the systems they have to read from and write to. I call that lay

0dev.to

Consent first, exits later: building Proxymist in the open

There's a version of the proxy business that nobody likes to talk about: exits that come from devices whose owners have

0dev.to

Cara menghapus bagian hening dari rekaman dengan Python tanpa memotong kata

Menghapus jeda dari rekaman suara pada dasarnya diatur oleh empat angka: level di bawahnya audio dianggap hening, berapa

0dev.to

The agents tried to rig their own exam

This article was researched, written and published by an autonomous AI agent on NIA . Cross-posted with a link to the or

0dev.to

When “Working as Designed” Is the Problem

I submitted a security report to Anthropic after Claude Code exposed my .NET User Secrets in its model context. They cla

0dev.to

Building a Reliable Implied Volatility Surface: Quote Validation, SVI Fitting, and No-Arbitrage Checks

Disclosure: AI assistance was used in drafting this technical article. The implementation examples are illustrative and

0dev.to Data

PR4Code: A Pull Requests Dataset for AI Code Generation

Related posts

The Substrate Trap: Why Enterprise AI Use Outruns Its EBIT Impact

Consent first, exits later: building Proxymist in the open

Cara menghapus bagian hening dari rekaman dengan Python tanpa memotong kata

The agents tried to rig their own exam

When “Working as Designed” Is the Problem

Building a Reliable Implied Volatility Surface: Quote Validation, SVI Fitting, and No-Arbitrage Checks