Work

Tail-Aware HiFloat4: W4A4 Post-Training Quantization for Wan2.2

Year: 2026

Type: preprint

Abstract: This report describes Tail-Aware HiFloat4, our submission to the low-bit text-to-video generation quantization challenge. Our method adapts the public ViDiT-Q post-training quantization pipeline to Wan2.2 under the HiFloat4 numerical format. We quantize the main linear layers in both Wan2.2 transformer modules with W4A4 HiFloat4 fake quantization, keep numerically sensitive boundary modules in high precision, and introduce an activation-tail-aware percentile calibration module for channel-mask construction. Together with compact PTQ-state restoration, this design reduces the influence of rare calibration outliers while keeping the runtime HiFloat4 arithmetic and sampling pipeline unchanged.

Source: arXiv (Cornell University)

DOI: 10.48550/arxiv.2605.26628

Authors: Feng, Zhanfeng, Guo, Shuai, Di, Xin, Peng, Long, Cao, Yang, Zha, Zhengjun

Institution: University of Science and Technology of China

Topic: Advanced Neural Network Applications

Subfield: Computer Vision and Pattern Recognition

Field: Computer Science

Domain: Physical Sciences

Open Access status: green

OpenAlex ID: W7162552865

Cites:

Cited by:

Related to: