PFGM++: Unlocking the Potential of Physics-Inspired Generative Models | Yilun Xu

Valence Labs is a research engine within Recursion committed to advancing the frontier of AI in drug discovery. Learn more about our open roles: https://www.valencelabs.com/careers

Join the Learning on Graphs and Geometry Reading Group on Slack: https://join.slack.com/t/logag/shared...

Abstract: We introduce a new family of physics-inspired generative models termed PFGM++ that unifies diffusion models and Poisson Flow Generative Models (PFGM). These models realize generative trajectories for N dimensional data by embedding paths in N+D dimensional space while still controlling the progression with a simple scalar norm of the D additional variables. The new models reduce to PFGM when D=1 and to diffusion models when D→∞. The flexibility of choosing D allows us to trade off robustness against rigidity as increasing D results in more concentrated coupling between the data and the additional variable norms. We dispense with the biased large batch field targets used in PFGM and instead provide an unbiased perturbation-based objective similar to diffusion models. To explore different choices of D, we provide a direct alignment method for transferring well-tuned hyperparameters from diffusion models (D→∞) to any finite D values. Our experiments show that models with finite D can be superior to previous state-of-the-art diffusion models on CIFAR-10/FFHQ 64×64 datasets, with FID scores of 1.91/2.43 when D=2048/128. In class-conditional setting, D=2048 yields current state-of-the-art FID of 1.74 on CIFAR-10. In addition, we demonstrate that models with smaller D exhibit improved robustness against modeling errors.

Speaker: Yilun Xu - https://yilun-xu.com/

Twitter Hannes:   / hannesstaerk  
Twitter Dominique:   / dom_beaini  

~

Chapters

00:00 - Intro
04:36 - Poisson Flow Generative Models
17:28 - Experiments: Generation Quality and Speed
23:13 - Next Generation Poisson Flow: PFGM++
36:20 - Summary and Q+A