0x404/ccs-code-llama-7b
The 0x404/ccs-code-llama-7b model is a 7 billion parameter Code Llama-based model, fine-tuned using LORA SFT for automated classification of Git commit messages. It specializes in categorizing commits into ten conventional types (e.g., feat, fix, docs, chore) by analyzing both the commit message and the corresponding Git diff. This model is optimized for developers needing to enforce conventional commit standards and streamline code review processes.
Loading preview...
Overview
The 0x404/ccs-code-llama-7b is a specialized 7 billion parameter model built upon the Code Llama architecture. It has been fine-tuned using LORA SFT to automatically classify Git commits into one of ten predefined conventional commit types. This model takes both the commit message and the Git diff as input to accurately determine the commit's category.
Key Capabilities
- Automated Commit Classification: Classifies commits into categories like
feat,fix,perf,style,refactor,docs,test,ci,build, andchore. - Contextual Analysis: Utilizes both the commit message and the associated Git diff for more accurate classification.
- Customizable Context Window: Supports adjusting the context window (default 1024 tokens) to include more of the Git diff, enhancing classification accuracy for larger changes.
- Integration Ready: Provides a straightforward Python pipeline for integration into development workflows, demonstrated with examples for fetching commit data from GitHub.
Good for
- Enforcing Conventional Commits: Automating the categorization of commits to maintain consistent commit history.
- Code Review Automation: Assisting in code review processes by pre-classifying changes.
- Developer Tooling: Integrating into CI/CD pipelines or developer tools to streamline commit management and analysis.