mlsanityData to 14 Oct 2026

PulseQuant: Propagation-Guided Subspace Correction for 4-Bit Video Diffusion Transformers

arXiv 2609.33384 ↗First posted 27 Sep 2026cs.CVcited by 1

On the map: Diffusion models → Video generation also fits: Inverse problems and image restoration

Abstract (arXiv)

Quantization errors in video diffusion transformers can be amplified or attenuated by subsequent denoising updates, making local reconstruction error an incomplete predictor of final impact. We introduce PulseQuant, a 4-bit post-training quantization method that combines trajectory sensitivity with activation geometry to guide offline calibration. Isolated block--step interventions estimate propagation risk, which prioritizes sensitive trajectory states during row-radius selection. With these radii fixed, response-subspace correction uses neighboring-code edits to reduce residual components along dominant activation directions. Both stages preserve the original 4-bit weight representation. Controlled interventions show that short-horizon propagated error predicts final latent error more reliably than immediate block-output error, supporting calibration beyond local reconstruction objectives. Evaluations on Wan models, Self Forcing, and MiniMax-H3 demonstrate improvements in key consistency and dense-reference metrics while remaining competitive on other attributes across model scales and generation paradigms.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Placed in Video generation, one of 14 directions in Diffusion models. Builds on 31 of its 35 references that are in the map. First posted on arXiv 27 Sep 2026.

Builds on

31 of 35 references.

Closest ancestors · 2

Same direction, newest first

  1. 2026VC-Attention: Value Smoothing and Softmax Casting for Low-bit Attention14 Sep
  2. 2025Self Forcing: Bridging the Train-Test Gap in Autoregressive Video Diffusion9 Jun

Other references · 29

  1. 2018The Unreasonable Effectiveness of Deep Features as a Perceptual Metriccited by 11,177
  2. 2022Scalable Diffusion Models with Transformerscited by 6,054
  3. 2025Wan: Open and Advanced Large-Scale Video Generative Modelscited by 2,708
  4. 2022GPTQ: Accurate Post-Training Quantization for Generative Pre-trained Transformerscited by 1,919
  5. 2023VBench: Comprehensive Benchmark Suite for Video Generative Modelscited by 1,822
  6. 2022SmoothQuant: Accurate and Efficient Post-Training Quantization for Large Language Modelscited by 1,386
  7. 2023AWQ: Activation-aware Weight Quantization for LLM Compression and Accelerationcited by 1,298
  8. 2024QuaRot: Outlier-Free 4-Bit Inference in Rotated LLMscited by 634
  9. 2020Up or Down? Adaptive Rounding for Post-Training Quantizationcited by 621
  10. 2021BRECQ: Pushing the Limit of Post-Training Quantization by Block Reconstructioncited by 479
  11. 2024SpinQuant: LLM quantization with learned rotationscited by 471
  12. 2023OmniQuant: Omnidirectionally Calibrated Quantization for Large Language Modelscited by 424
17 more
  1. 2024QuIP#: Even Better LLM Quantization with Hadamard Incoherence and Lattice Codebookscited by 378
  2. 2023Q-Diffusion: Quantizing Diffusion Modelscited by 273
  3. 2022Post-training Quantization on Diffusion Modelscited by 267
  4. 2024Extreme Compression of Large Language Models via Additive Quantizationcited by 232
  5. 2023PTQD: Accurate Post-Training Quantization for Diffusion Modelscited by 170
  6. 2024SVDQuant: Absorbing Outliers by Low-Rank Components for 4-Bit Diffusion Modelscited by 170
  7. 2024ViDiT-Q: Efficient and Accurate Quantization of Diffusion Transformers for Image and Video Generationcited by 115
  8. 2024SageAttention2: Efficient Attention with Thorough Outlier Smoothing and Per-thread INT4 Quantizationcited by 114
  9. 2024Q-DiT: Accurate Post-Training Quantization for Diffusion Transformerscited by 94
  10. 2023Temporal Dynamic Quantization for Diffusion Modelscited by 81
  11. 2023TFMQ-DM: Temporal Feature Maintenance Quantization for Diffusion Modelscited by 69
  12. 2023Towards Accurate Post-training Quantization for Diffusion Modelscited by 47
  13. 2025Q-VDiT: Towards Accurate Quantization and Distillation of Video-Generation Diffusion Transformerscited by 23
  14. 2025QVGen: Pushing the Limit of Quantized Video Generative Modelscited by 22
  15. 2025DVD-Quant: Data-free Video Diffusion Transformers Quantizationcited by 15
  16. 2026OrbitQuant: Data-Agnostic Quantization for Image and Video Diffusion Transformerscited by 6
  17. 2025AccuQuant: Simulating Multiple Denoising Steps for Quantizing Diffusion Modelscited by 3

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026VDOT++: Unified Few-Step Video Generation via Unbalanced Optimal Transport Distillationcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 20266Bit-Diffusion: Inference-Time Mixed-Precision Quantization for Video Diffusion Models
  2. 2026Timestep-Aware SVDQuant-GPTQ for W4A4 Quantization of Wan2.2-I2V
  3. 2025S2Q-VDiT: Accurate Quantized Video Diffusion Transformer with Salient Data and Sparse Token Distillation
  4. 2024TaQ-DiT: Time-aware Quantization for Diffusion Transformers
  5. 2026DiRotQ: Rotation-Aware Quantization for 4-bit Diffusion Transformers
  6. 2026Deflating the Hessian: Rank-4 W4A4 Quantization for Multimodal Diffusion Transformers
  7. 2025Error Propagation Mechanisms and Compensation Strategies for Quantized Diffusion
  8. 2026LSGQuant: Layer-Sensitivity Guided Quantization for One-Step Diffusion Real-World Video Super-Resolution