# Roadmap

1. Baseline runnable stack (this version).
2. Larger model GRPO loops with distributed training.
3. Richer real-world drug-label ingestion and calibration.
4. Extended clinical validation and publication artifacts.