Paper 2602.23866
SWE-rebench V2: Language-Agnostic SWE Task Collection at Scale
- Published
- Feb 2026
- Research lab
- Independent
- Citations
- 11
- GitHub
- 76 stars
01 In brief
Summary
SWE-rebench V2 introduces a language-agnostic automated pipeline for harvesting executable real-world software engineering (SWE) tasks at scale, addressing the scarcity of large-scale training data for reinforcement learning (RL).
The pipeline mines pull request histories, synthesizes repository-specific installation and test procedures via an interactive setup agent, validates environments through dual-pass execution, filters underspecified issues using an ensemble of LLM judges, and enriches tasks with diagnostic metadata.
The authors release 32,079 stable issue-linked tasks from 3,617 repositories across 20 languages, plus 120,000+ PR-derived tasks with generated problem statements.
Ablations show interactive setup agents outperform non-interactive pipelines, and ensembling improves issue clarity filtering.
Diagnostic analysis of seven frontier models across five languages identifies confounders like test suite coupling and implicit naming, enabling stratified filtering.
Limitations include lack of end-to-end RL training ablations and potential environment drift.
02 From the paper
Abstract
Software engineering agents (SWE) are improving rapidly, with recent gains largely driven by reinforcement learning (RL). However, RL training is constrained by the scarcity of large-scale task collections with reproducible execution environments and reliable test suites. Although a growing number of benchmarks have emerged, datasets suitable for training remain limited in scale and diversity or often target a limited set of high-resource language ecosystems. We introduce SWE-rebench V2, a language-agnostic automated pipeline for harvesting executable real-world SWE tasks and constructing RL training environments at scale. The pipeline synthesizes repository-specific installation and test procedures via an interactive setup agent, and filters unsound instances using an ensemble of LLM judges, validated against human-verified SWE-bench annotations. Using this pipeline, we construct a dataset of 32,079 tasks spanning 20 languages and 3,617 repositories, with pre-built images for reproducible execution. To further scale training data, we additionally release 120,000+ tasks with installation instructions, fail-to-pass tests and rich metadata, where the problem statement is generated based on the original pull request description. We validate the collected instances through a diagnostic study that covers a subset of tasks in five programming languages across seven popular models, and provide instance-level metadata that flags common confounders such as overly restrictive tests and underspecified descriptions. We release the datasets, the collection and execution code, and associated artifacts to enable large-scale training of SWE agents across diverse languages and repositories.