Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper investigates the conditions under which local score models can successfully extrapolate from small to large systems, revealing that architectural locality alone is insufficient for stable size transfer. The authors demonstrate that the quasi-locality of the Gaussian-smoothed score, influenced by posterior covariance, is crucial for effective extrapolation, as it determines whether a model's receptive field can adequately cover the necessary response range. They introduce Finite-Depth Local Flow (FDLF) as a diagnostic benchmark and empirically validate the relationship between spatial mixing, smoothed-score quasi-locality, and model performance in size transfer tasks.
Local score models can fail to extrapolate across sizes when spatial mixing weakens, revealing a critical dependency on the quasi-locality of the Gaussian-smoothed score.
Scientific generative modeling often requires size transfer, where models trained on small systems are evaluated on larger ones. While translation-invariant architectures enable this evaluation, we show that architectural locality alone does not guarantee stable size extrapolation. Instead, stable extrapolation is governed by the quasi-locality of the Gaussian-smoothed score. Through Tweedie's formula, far-away perturbations can influence local score components via posterior covariance, meaning a local model succeeds only if its receptive field covers the smoothed score's response range. We formalize this mechanism, proving a size-uniform comparison theorem for local marginals under reverse diffusion. We also introduce Finite-Depth Local Flow (FDLF), a white-box diagnostic benchmark with exact scores, densities, and controllable response ranges. Empirically, we validate the interplay between spatial mixing, smoothed-score quasi-locality, and model receptive fields. Under spatial mixing, the smoothed score remains quasi-local relative to the receptive field, enabling stable extrapolation. Conversely, when spatial mixing weakens, the score's locality rapidly degrades, causing size transfer to fail.