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

1 code implementation • 8 Feb 2023 • Yilun Xu, Ziming Liu, Yonglong Tian, Shangyuan Tong, Max Tegmark, Tommi Jaakkola

The new models reduce to PFGM when $D{=}1$ and to diffusion models when $D{\to}\infty$.

Image Generation Playing the Game of 2048

Stable Target Field for Reduced Variance Score Estimation in Diffusion Models

1 code implementation • 1 Feb 2023 • Yilun Xu, Shangyuan Tong, Tommi Jaakkola

We show that the procedure indeed helps in the challenging intermediate regime by reducing (the trace of) the covariance of training targets.

Denoising Image Generation

Adversarial Support Alignment

1 code implementation • ICLR 2022 • Shangyuan Tong, Timur Garipov, Yang Zhang, Shiyu Chang, Tommi S. Jaakkola

Furthermore, we show that our approach can be viewed as a limit of existing notions of alignment by increasing transportation assignment tolerance.

Domain Adaptation

The Benefits of Pairwise Discriminators for Adversarial Training

1 code implementation • 20 Feb 2020 • Shangyuan Tong, Timur Garipov, Tommi Jaakkola

We provide sufficient conditions for local convergence; characterize the capacity balance that should guide the discriminator and generator choices; and construct examples of minimally sufficient discriminators.