Scale-Wise VAR is Secretly Discrete Diffusion
This work provides a theoretical bridge between autoregressive and diffusion models for visual generation, offering incremental improvements in efficiency and performance for AI researchers and practitioners.
The paper shows that autoregressive transformers for visual generation (VAR) with a Markovian attention mask are mathematically equivalent to discrete diffusion models, enabling the integration of diffusion advantages like iterative refinement into VAR for faster convergence, lower inference cost, and improved zero-shot reconstruction across multiple datasets.
Autoregressive (AR) transformers have emerged as a powerful paradigm for visual generation, largely due to their scalability, computational efficiency and unified architecture with language and vision. Among them, next scale prediction Visual Autoregressive Generation (VAR) has recently demonstrated remarkable performance, even surpassing diffusion-based models. In this work, we revisit VAR and uncover a theoretical insight: when equipped with a Markovian attention mask, VAR is mathematically equivalent to a discrete diffusion. We term this reinterpretation as Scalable Visual Refinement with Discrete Diffusion (SRDD), establishing a principled bridge between AR transformers and diffusion models. Leveraging this new perspective, we show how one can directly import the advantages of diffusion such as iterative refinement and reduce architectural inefficiencies into VAR, yielding faster convergence, lower inference cost, and improved zero-shot reconstruction. Across multiple datasets, we show that the diffusion based perspective of VAR leads to consistent gains in efficiency and generation.