Large Discovery Models combine LLM proposals with Bayesian scoring for open-ended search

A new paper proposes Large Discovery Models (LDMs), an iterative loop where LLMs generate candidate proposals and Bayesian nonparametric reward surrogates score them, guiding search under uncertainty. The method is applied to molecular, protein, and program discovery.

2026-08-18 ~ 2026-08-19 · 2 related posts

• Large Discovery Models: Empirically-grounded Model-Based Open-Ended Search — Yangtze-ailab · 2026-08-18
• Large Discovery Models: LLM Proposes, Bayesian Surrogate Scores, 2.4x Lower Error — _akhaliq · 2026-08-19