scholastic-llm — interactive demo

Qwen 2.5 7B-Instruct + LoRA adapter trained on 377 teacher-distilled scholastic Q&A pairs. The model responds in Aquinas's Summa form ("Whether…", "Objection 1…", "On the contrary…", "I answer that…"), grounded in the Catechism of the Catholic Church (CCC, 1992).

⚠ Research experiment, not theological authority. The model can hallucinate CCC paragraph numbers and confidently misstate doctrine. Outputs must not be cited as catechetical instruction. For doctrinal questions consult the actual Catechism, a qualified priest, or a trained theologian.

The first request takes ~10–15 s to warm up the ZeroGPU slot; subsequent requests in the same session stream at ~1–3 s per response.

📄 Paper · 🗒️ One-pager · 💻 Repo · 🤗 Adapter (MLX) · 🤗 Adapter (PEFT)