ACDiT: Interpolating Autoregressive Conditional Modeling and Diffusion Transformer

Understand

Autoregressive and diffusion models have achieved remarkable progress in language models and visual generation, respectively.

• We present ACDiT, a novel Autoregressive blockwise Conditional Diffusion Transformer, that innovatively combines autoregressive and diffusion paradigms for continuous visual information.
• By introducing a block-wise autoregressive unit, ACDiT offers a flexible interpolation between token-wise autoregression and full-sequence diffusion, bypassing the limitations of discrete tokenization.
• The generation of each block is formulated as a conditional diffusion process, conditioned on prior blocks.