This research project implements vision-level token masking for PHI compliance in OCR systems.
Key components:
- Synthetic medical data generation pipeline using Synthea
- PHI annotation and preprocessing tools
- LoRA adapter implementation for PHI detection
- Configuration files for training and model setup
- Comprehensive documentation and setup instructions
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
