mlsanityData to 14 Oct 2026

Precision As You Need: Stochastic Computing Is a Dense Adaptive Quantizer

arXiv 2609.32922 ↗First posted 26 Sep 2026cs.AIcited by 0

Not on the map. Why?

Abstract (arXiv)

Matrix multiplications dominate the inference cost of modern transformer-based vision models, yet existing efficiency techniques such as post-training quantization and mixed-precision inference are largely limited to the small set of fixed-width formats (INT4, INT8, BF16, and FP16) supported by conventional accelerators. We revisit stochastic computing (SC) as a way to lift this constraint: viewed as a dense adaptive quantizer, SC controls precision by bit-stream length L rather than a fixed datapath, while each multiplication reduces to a single AND/XNOR gate. We build a GPU library that emulates SC matrix multiplication at scale, exposes stream lengths as first-class kernel arguments, and evaluates SC end-to-end on image classification, object detection and instance segmentation, class-conditional image generation, and visual world-model planning. On top of this substrate, we develop a dynamic per-row mixed-precision policy that assigns stream length per token or group at matched average budget, requires no retraining, and uses the same SC hardware across schedules. Across tasks, SC remains competitive with fixed-format INT quantization at matched bit budgets, while per-row mixed precision helps maintain accuracy at lower average stream lengths. These results provide software-level feasibility evidence that SC can serve as a dense-precision substrate for fine-grained mixed-precision inference on modern vision transformers.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 25 of its 44 references that are in the map. First posted on arXiv 26 Sep 2026.

Builds on

25 of 44 references.

References · 25

  1. 2020Language Models are Few-Shot Learnerscited by 32,419
  2. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 30,186
  3. 2023DINOv2: Learning Robust Visual Features without Supervisioncited by 7,974
  4. 2017Mask R-CNNcited by 7,277
  5. 2022Scalable Diffusion Models with Transformerscited by 6,054
  6. 2018World Modelscited by 1,513
  7. 2022SmoothQuant: Accurate and Efficient Post-Training Quantization for Large Language Modelscited by 1,386
  8. 2023AWQ: Activation-aware Weight Quantization for LLM Compression and Accelerationcited by 1,298
  9. 2022LLM.int8(): 8-bit Matrix Multiplication for Transformers at Scalecited by 906
  10. 2022Exploring Plain Vision Transformer Backbones for Object Detectioncited by 724
  11. 2022EVA: Exploring the Limits of Masked Visual Representation Learning at Scalecited by 691
  12. 2021BRECQ: Pushing the Limit of Post-Training Quantization by Block Reconstructioncited by 479
13 more
  1. 2021A White Paper on Neural Network Quantizationcited by 472
  2. 2018HAQ: Hardware-Aware Automated Quantization with Mixed Precisioncited by 443
  3. 2019HAWQ: Hessian AWare Quantization of Neural Networks with Mixed-Precisioncited by 402
  4. 2023Q-Diffusion: Quantizing Diffusion Modelscited by 273
  5. 2021PTQ4ViT: Post-training quantization for vision transformers with twin uniform quantizationcited by 179
  6. 2022RepQ-ViT: Scale Reparameterization for Post-Training Quantization of Vision Transformerscited by 137
  7. 2024Q-DiT: Accurate Post-Training Quantization for Diffusion Transformerscited by 94
  8. 2024MixDQ: Memory-Efficient Few-Step Text-to-Image Diffusion Models with Metric-Decoupled Mixed Precision Quantizationcited by 43
  9. 2024Quantization without Tearscited by 25
  10. 2024MPQ-DM: Mixed Precision Quantization for Extremely Low Bit Diffusion Modelscited by 21
  11. 2024BitMoD: Bit-serial Mixture-of-Datatype LLM Accelerationcited by 18
  12. 2025MixDiT: Accelerating Image Diffusion Transformer Inference with Mixed-Precision MX Quantizationcited by 8
  13. 2016SC-DCNN: Highly-Scalable Deep Convolutional Neural Network using Stochastic Computingcited by 7

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026Amortized-Precision Quantization for Early-Exit Vision Transformers
  2. 2026QFlash: Bridging Quantization and Memory Efficiency in Vision Transformer Attention
  3. 2026STQuant: Spatio-Temporal Adaptive Framework for Optimizer Quantization in Large Multimodal Model Training
  4. 2026ViM-Q: Scalable Algorithm-Hardware Co-Design for Vision Mamba Model Inference on FPGA
  5. 2025FlexiQ: Adaptive Mixed-Precision Quantization for Latency/Accuracy Trade-Offs in Deep Neural Networks
  6. 2026RAMP: Robust Adaptive Mixed-Precision Quantization for Edge CPU Vision Models
  7. 2023SONIQ: System-Optimized Noise-Injected Ultra-Low-Precision Quantization with Full-Precision Parity
  8. 2026Achieving Near-Zero-Overhead Multi-Model Hierarchical Classification in Real-Time Detection Pipelines