Tail-Aware HiFloat4: W4A4 Post-Training Quantization for Wan2.2
Not on the map. Why?
Abstract (arXiv)
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.
Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 24 of its 27 references that are in the map. First posted on arXiv 26 May 2026.
Builds on
24 of 27 references.
References · 24
- 2020Denoising Diffusion Probabilistic Modelscited by 21,076
- 2021High-Resolution Image Synthesis with Latent Diffusion Modelscited by 18,863
- 2022Scalable Diffusion Models with Transformerscited by 6,054
- 2025Wan: Open and Advanced Large-Scale Video Generative Modelscited by 2,708
- 2022GPTQ: Accurate Post-Training Quantization for Generative Pre-trained Transformerscited by 1,919
- 2023VBench: Comprehensive Benchmark Suite for Video Generative Modelscited by 1,822
- 2022Imagen Video: High Definition Video Generation with Diffusion Modelscited by 1,778
- 2022Make-A-Video: Text-to-Video Generation without Text-Video Datacited by 1,742
- 2024HunyuanVideo: A Systematic Framework For Large Video Generative Modelscited by 1,660
- 2017Quantization and Training of Neural Networks for Efficient Integer-Arithmetic-Only Inferencecited by 1,593
- 2022SmoothQuant: Accurate and Efficient Post-Training Quantization for Large Language Modelscited by 1,386
- 2022CogVideo: Large-scale Pretraining for Text-to-Video Generation via Transformerscited by 985
12 more
- 2024QuaRot: Outlier-Free 4-Bit Inference in Rotated LLMscited by 634
- 2020Up or Down? Adaptive Rounding for Post-Training Quantizationcited by 621
- 2022ZeroQuant: Efficient and Affordable Post-Training Quantization for Large-Scale Transformerscited by 526
- 2023Q-Diffusion: Quantizing Diffusion Modelscited by 273
- 2022Post-training Quantization on Diffusion Modelscited by 267
- 2024ViDiT-Q: Efficient and Accurate Quantization of Diffusion Transformers for Image and Video Generationcited by 115
- 2024Q-DiT: Accurate Post-Training Quantization for Diffusion Transformerscited by 94
- 2024PTQ4DiT: Post-training Quantization for Diffusion Transformerscited by 79
- 2023EfficientDM: Efficient Quantization-Aware Fine-Tuning of Low-Bit Diffusion Modelscited by 78
- 2025OpenS2V-Nexus: A Detailed Benchmark and Million-Scale Dataset for Subject-to-Video Generationcited by 73
- 2024VQ4DiT: Efficient Post-Training Vector Quantization for Diffusion Transformerscited by 12
- 2026HiFloat4 Format for Language Model Inferencecited by 11
Built on by
Not cited yet.
Similar, not cited
Close in wording, no citation either way
- 2026Timestep-Aware SVDQuant-GPTQ for W4A4 Quantization of Wan2.2-I2V
- 2026TR-PTQ: High-Accuracy Integer-Only Transformer Post Training Quantization via Taylor Region Reformulation
- 2026W4A4 Quantization for Inference on Wan2.2-I2V-A14B
- 2026Boundary-Protection W8A8 HiFloat8 Quantization for Large-Scale Text-to-Video Diffusion Transformers
- 2025QuantCache: Adaptive Importance-Guided Quantization with Hierarchical Latent and Layer Caching for Video Generation
- 2026Activation Outliers Matter: Robust Recovery for Quantized Multimodal LLMs
- 2025Q-VDiT: Towards Accurate Quantization and Distillation of Video-Generation Diffusion Transformers
- 2025SignRoundV2: Toward Closing the Performance Gap in Extremely Low-Bit Post-Training Quantization for LLMs