High-Coder-Reasoning-Multi-Turn
- Type
- dataset
- Venue
- Crownelius (individual)
- Year
- 2026
- Source
- huggingface
- Access
- free
- Language
- Multiple programming languages (C++, Go, Java, C#, Python, etc.)
- Added
- 2026-07-17T20:18:03.799152+00:00
- Verified
- 2026-07-17T20:18:03.799152+00:00
Summary
High-Coder-Reasoning-Multi-Turn is a synthetic dataset of 26.9k multi-turn coding conversations generated using the openrouter/hunter-alpha model. Each sample includes code in various programming languages with transformations (translate, fix, critique+analyze), provenance metadata, quality signals, and deduplication hashes. The dataset covers code review, translation between languages, and bug-fixing tasks with detailed reasoning traces.
Keywords
code reasoning multi-turn synthetic code-review translation bug-fixing hunter-alpha
Topics
Code
Research notes
- Generated by openrouter/hunter-alpha model. Includes provenance tracking (source_platform, model, pipeline, transform_type) and quality signals (has_imports, has_critique, has_analysis, synthetic flag). MIT license per license_spdx field.