Sign in to view source links and access this dataset
Description
SWE-Explore-Bench is a dataset created for the SWE-Explore project, which benchmarks how coding agents explore software repositories. The dataset was authored by Shaoqiu Zhang, Yuhang Wang, and others, and was last updated on June 8, 2026. Its specific content and structure are detailed on its Hugging Face dataset page.
Use Cases
Benchmarking the exploration capabilities of AI coding agents based on repository navigation tasks.
Training agents to understand codebase structure and locate relevant code snippets.
Evaluating agent performance on software engineering workflows that require contextual search.
Strengths
Dataset is explicitly designed for benchmarking a specific task in AI for software engineering.
Associated with a cited research paper titled 'SWE-Explore: Benchmarking How Coding Agents Explore Repositories'.
Limitations
Column-level documentation is absent; field semantics must be inferred after download.
Row count is unknown, which may limit suitability assessment.
Description metadata is limited; actual data quality requires manual inspection after download.
Provenance
Source
huggingface
Freshness
Last updated 2026-06-08 15:41:51; freshness should be verified.
License is unknown; users should verify terms of use before application.