zeekay f926b2fa3d feat: Add comprehensive GRPO implementation with code quality focus ...

Major additions:
- Code quality reward function for evaluating programming style
- Programming style dataset with good/bad code examples
- Daily training pipeline for continuous improvement
- Code example collector from Hanzo Chat conversations
- Comprehensive CLI tool with multiple commands
- Docker configuration for training, development, and inference
- GitHub Actions workflow for automated training pipeline
- Makefile for easy command execution

Features:
- Collect code examples from chat interactions
- Analyze code quality with multiple metrics (syntax, documentation, type hints, etc.)
- Create training pairs from real-world code improvements
- Automated daily fine-tuning with deployment pipeline
- Support for multiple programming languages
- Integration with Hanzo AI cloud platform

2025-07-03 23:33:10 -04:00

..

collect_code_examples.py

feat: Add comprehensive GRPO implementation with code quality focus

2025-07-03 23:33:10 -04:00

daily_training.py

feat: Add comprehensive GRPO implementation with code quality focus

2025-07-03 23:33:10 -04:00