@JeffGuo__
Pinned post
RL causing diversity collapse in generative models (e.g. LLMs) is *not* a training failure.
It’s what KL / entropy-regularized RL is provably *designed* to do.
The good news: we have a simple, principled fix.
Accepted to #ICLR2026
🧵🔽 [1/n]View quoted post on X
Thrilled to share our new work MatInvent, a general and efficient reinforcement learning workflow that optimizes diffusion models for goal-directed crystal generation. Thanks to @JeffGuo__ , @efssh , @pschwllr , @SchwallerGroup , @NCCR_Catalysis .View quoted post on X
arxiv.org/abs/2511.03112
pubs.rsc.org Directly optimizing for synthesizability in generative molecular design using retrosynthesis models† Synthesizability in generative molecular design remains a pressing challenge. Existing methods to assess synthesizability include heuristics-based metrics Molecular generative models can *directly* optimize for synthesizability using retrosynthesis models!
Check out initial results which can be an alternative to synthesizability-constrained generation
Pre-print: arxiv.org/abs/2407.12186…
Code: github.com/schwallergroup…
(1/2)View quoted post on X
