Drop-in starter configurations. Copy whichever matches your use case to
the project root as config.yaml, edit, and run.
| Config | Use case |
|---|---|
configs/minimal-rag.yaml |
Text-only synthetic records grounded in retrieved literature; no imaging, no LLM keys required |
configs/minimal-imaging.yaml |
Chest-radiograph synthetic records via Hugging Face diffusers; needs HF_TOKEN for gated models |
configs/dpo-export.yaml |
Generate + export dpo_jsonl / rl_jsonl preference data ready for TRL / axolotl trainers |
All three pass AppConfig validation as written — CI exercises them.