titleOrbitQuant: Data-Agnostic Quantization for Image and Video Diffusion Transformers
authorsDonghyun Lee, Jitesh Chavan, Duy Nguyen, Sam Huang, Liming Jiang, Priyadarshini Panda, Timo Mertens, Saurabh Shukla
date2026-07-02
venuearXiv (preprint, not peer reviewed)
idarXiv:2607.02461
affiliationCantina, Inc.
upvotes36 (Hugging Face Daily Papers, 2026-11-01)
linksabs pdf hf
abstractDiffusion transformers (DiTs) achieve state-of-the-art image and video generation, but their multi-step sampling and growing parameter count make inference expensive. Post-training quantization (PTQ) is the natural remedy, yet DiT activations shift across timesteps, prompts, and guidance branches, forcing prior methods to re-fit calibration data for every new checkpoint or modality. We present OrbitQuant, a data-agnostic weight-activation quantizer that bypasses range estimation by quantizing in a normalized, rotated basis. In this basis, a randomized permuted block-Hadamard (RPBH) rotation concentrates each coordinate around one fixed, known marginal regardless of the input, so a single Lloyd-Max codebook serves all timesteps, prompts, and layers of a given input dimension. We extend the same quantizer to weight rows offline, absorbing the rotation into the weights so that it cancels inside each linear layer and only a forward rotation on the activations remains at runtime. The same recipe transfers from image to video with no per-modality tuning. Across FLUX.1, Z-Image-Turbo, Wan 2.1, and CogVideoX, it sets the state of the art for PTQ at several low-bit settings. It also pushes PTQ of image diffusion transformers to W2A4 with usable generation quality.

cite

@misc{lee2026orbitquant,
  title={OrbitQuant: Data-Agnostic Quantization for Image and Video Diffusion Transformers},
  author={Donghyun Lee and Jitesh Chavan and Duy Nguyen and Sam Huang and Liming Jiang and Priyadarshini Panda and Timo Mertens and Saurabh Shukla},
  year={2026},
  eprint={2607.02461},
  archivePrefix={arXiv},
  url={https://arxiv.org/abs/2607.02461}
}