Research Unlocking Any-Order Generation in Pretrained Autoregressive Image Models
A method has been proposed to overcome the constraints of fixed raster-scan ordering in pretrained autoregressive image models, enabling inpainting and editing.
Autoregressive (AR) image generators are powerful in principle but inflexible in practice. Training with a fixed raster-scan order hardcodes a single factorization of the image distribution into the model, leaving it unable to inpaint, outpaint, or edit at inference time.
Existing solutions, including the modern discrete-diffusion family, train from scratch on a mixture of token orderings, requiring roughly 3x the compute of standard raster-scan training. Furthermore, discrete diffusion sacrifices compatibility with the AR inference ecosystem, such as KV caching and vLLM, by construction. The researchers argue that this cost is unnecessary.
This item is a summary. The full, checkable version is at the original source:
Read the original at Hugging Face Papers ↗