mlsanityData to 14 Oct 2026

VisionMX: Unlocking Microscaling Post-Training Quantization for Vision Models

arXiv 2610.03218 ↗First posted 2 Oct 2026cs.CVcited by 0

Not on the map. Why?

Abstract (arXiv)

Microscaling (MX) formats are emerging as a hardware-supported approach to efficient training and inference. They combine low-precision elements with shared block scales, but their impact on vision models remains underexplored. We systematically investigate post-training MX quantization across vision models and tasks. An analysis of direct conversion identifies three sources of error: block-scale representation, the poor alignment of some small convolutional weight tensors with nonuniform element grids, and the underuse of signed codes by nonnegative activations. These findings motivate VisionMX, a post-training MX quantization method that optimizes bounded weight rounding and applies a foldable affine correction to activations. We evaluate VisionMX across image classification, object detection, semantic segmentation, and low-light image enhancement using several MX-style formats. It improves on direct conversion and the evaluated post-training quantization baselines, with the largest performance recoveries in architectures most sensitive to MX conversion

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 41 of its 49 references that are in the map. First posted on arXiv 2 Oct 2026.

Builds on

41 of 49 references.

References · 41

  1. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  2. 2014Microsoft COCO: Common Objects in Contextcited by 17,608
  3. 2015Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networkscited by 9,639
  4. 2014ImageNet Large Scale Visual Recognition Challengecited by 8,498
  5. 2018MobileNetV2: Inverted Residuals and Linear Bottleneckscited by 4,795
  6. 2022A ConvNet for the 2020scited by 3,579
  7. 2016Aggregated Residual Transformations for Deep Neural Networkscited by 2,546
  8. 2017Rethinking Atrous Convolution for Semantic Image Segmentationcited by 2,161
  9. 2017Quantization and Training of Neural Networks for Efficient Integer-Arithmetic-Only Inferencecited by 1,593
  10. 2019FCOS: Fully Convolutional One-Stage Object Detectioncited by 1,577
  11. 2021A Survey of Quantization Methods for Efficient Neural Network Inferencecited by 815
  12. 2021Going deeper with Image Transformerscited by 764
29 more
  1. 2020Up or Down? Adaptive Rounding for Post-Training Quantizationcited by 621
  2. 2019HAWQ: Hessian AWare Quantization of Neural Networks with Mixed-Precisioncited by 402
  3. 2019Data-Free Quantization Through Weight Equalization and Bias Correctioncited by 336
  4. 2021Post-Training Quantization for Vision Transformercited by 278
  5. 2023Rewarded soups: towards Pareto-optimal alignment by interpolating weights fine-tuned on diverse rewardscited by 243
  6. 2022EfficientFormer: Vision Transformers at MobileNet Speedcited by 241
  7. 2023Retinexformer: One-stage Retinex-based Transformer for Low-light Image Enhancementcited by 231
  8. 2020LSQ+: Improving low-bit quantization through learnable offsets and better initializationcited by 189
  9. 2022QDrop: Randomly Dropping Quantization for Extremely Low-bit Post-Training Quantizationcited by 180
  10. 2021PTQ4ViT: Post-training quantization for vision transformers with twin uniform quantizationcited by 179
  11. 2023Microscaling Data Formats for Deep Learningcited by 152
  12. 2022RepQ-ViT: Scale Reparameterization for Post-Training Quantization of Vision Transformerscited by 137
  13. 2022Overcoming Oscillations in Quantization-Aware Trainingcited by 119
  14. 2025HVI: A New Color Space for Low-light Image Enhancementcited by 63
  15. 2022NoisyQuant: Noisy Bias-Enhanced Post-Training Activation Quantization for Vision Transformerscited by 61
  16. 2021Convergence Rates of Stochastic Gradient Descent under Infinite Noise Variancecited by 35
  17. 2024MicroScopiQ: Accelerating Foundational Models through Outlier-Aware Microscaling Quantizationcited by 28
  18. 2024Instance-Aware Group Quantization for Vision Transformerscited by 25
  19. 2025Oscillation-Reduced MXFP4 Training for Vision Transformerscited by 21
  20. 2025FIMA-Q: Post-Training Quantization for Vision Transformers by Fisher Information Matrix Approximationcited by 19
  21. 2024Accelerating Sinkhorn Algorithm with Sparse Newton Iterationscited by 11
  22. 2025MicroMix: Efficient Mixed-Precision Quantization with Microscaling Formats for Large Language Modelscited by 11
  23. 2021HPTQ: Hardware-Friendly Post Training Quantizationcited by 10
  24. 2026Is Finer Better? The Limits of Microscaling Formats in Large Language Modelscited by 8
  25. 2019Acceleration of SVRG and Katyusha X by Inexact Preconditioningcited by 6
  26. 2023EPTQ: Enhanced Post-Training Quantization via Hessian-guided Network-wise Optimizationcited by 6
  27. 2026FAAR: Format-Aware Adaptive Rounding for NVFP4cited by 4
  28. 2026LATMiX: Learnable Affine Transformations for Microscaling Quantization of LLMscited by 2
  29. 2026Characterizing the Impact of NVFP4 Quantization for Low-Power Edge AI Deploymentcited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026MiX: Micro-Inverted-Scaling for End-to-End Low-Bit Vision-Language Model Acceleration
  2. 2026SCULPT: Training Edge Vision Models for Post-Training Quantization Readiness
  3. 2025Dissecting Bit-Level Scaling Laws in Quantizing Vision Generative Models
  4. 2026Fine-Grained Post-Training Quantization for Large Vision Language Models with Quantization-Aware Integrated Gradients
  5. 2026MicroQonv: Reshaping Convolution Tensors for Efficient Microscaling in Training and Inference
  6. 2026MXAttention: Data-Free Optimal Scaling and Pre-Normalization Quantization for MXFP4 Attention
  7. 2025Post-Training Quantization for Vision Mamba with k-Scaled Quantization and Reparameterization
  8. 2026RAMP: Robust Adaptive Mixed-Precision Quantization for Edge CPU Vision Models