Search papers, labs, and topics across Lattice.
This paper introduces a post-hoc method for estimating uncertainty in pretrained molecular diffusion models, addressing the lack of quality signals in generated molecules. By applying a Laplace approximation to the denoising network, the authors measure the variability of noise predictions throughout the generation process. The resulting uncertainty scores correlate negatively with established quality metrics, demonstrating their utility in filtering samples to enhance model performance during test-time scaling.
Uncertainty scores derived from molecular diffusion models can effectively predict sample quality, enabling targeted filtering that boosts model performance.
Diffusion models have seen wide adoption for 3D molecular generation, yet they offer no principled signal of when a generated molecule is likely to be of low quality. We propose a post-hoc method for estimating per-sample uncertainty in pretrained molecular diffusion models. Building on a Laplace approximation of the denoising network, we measure the variability of the noise prediction across the generation trajectory. Empirically, we show that the resulting uncertainty score is informative of sample quality, exhibiting a negative correlation with established sample-level quality metrics. We further study how the proposed uncertainty score can be used to filter generated samples, improving model performance via test-time scaling.
