mlsanityData to 14 Oct 2026

HoloQ-VLA: Uniform W4A4 Quantization of Vision-Language-Action Models

arXiv 2605.28803 ↗First posted 27 May 2026cs.CV cs.LGcited by 2

On the map: Vision-language-action models → Efficient VLA inference and adaptation

Abstract (arXiv)

Vision-Language-Action (VLA) models unify perception, reasoning, and control in a single policy, but their multi-billion-parameter backbones and diffusion-based action heads make on-device deployment prohibitively expensive. Low-bit post-training quantization (PTQ) is the natural remedy, yet the diffusion action head that emits continuous control signals is highly sensitive to it: a few weight and activation outliers are enough to destabilize the head, so prior work leaves it at full precision or falls back to mixed-precision schemes, and uniformly quantizing the whole model to low bit-width remains an open challenge. We present HoloQ-VLA, the first training-free PTQ framework that compresses both the language backbone and the entire diffusion action head to uniform W4A4 precision without mixed-precision allocation. Instead of trading weight quality against activation quality, HoloQ-VLA targets the two outlier sources with complementary transforms: a weight-adapted rotation composed with an activation-dispersing Hadamard transform, together with per-step scaling that absorbs the dynamic-range drift exhibited by the action head across denoising steps. On LIBERO, HoloQ-VLA compresses Pi-0.5 and GR00T-N1.5 to W4A4 with 98.0% and 87.8% task success rates, matching or exceeding their FP16 references of 97.1% and 87.0%, while reducing the static memory footprint by 74.2%. Real-world manipulation experiments further demonstrate that HoloQ-VLA maintains smooth and accurate control across diverse real-world scenarios.

Key facts. Cited by 2 papers in the map (data through 14 Oct 2026). Placed in Efficient VLA inference and adaptation, one of 13 directions in Vision-language-action models. Builds on 21 of its 23 references that are in the map. First posted on arXiv 27 May 2026.

Builds on

21 of 23 references.

Closest ancestors · 1

Same direction, newest first

  1. 2026QVLA: Not All Channels Are Equal in Vision-Language-Action Model's Quantization3 Feb

Other references · 20

  1. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 30,186
  2. 2023Llama 2: Open Foundation and Fine-Tuned Chat Modelscited by 13,028
  3. 2024OpenVLA: An Open-Source Vision-Language-Action Modelcited by 2,887
  4. 2023Diffusion Policy: Visuomotor Policy Learning via Action Diffusioncited by 2,838
  5. 2022GPTQ: Accurate Post-Training Quantization for Generative Pre-trained Transformerscited by 1,919
  6. 2025π0.5: a Vision-Language-Action Model with Open-World Generalizationcited by 1,761
  7. 2023LIBERO: Benchmarking Knowledge Transfer for Lifelong Robot Learningcited by 1,410
  8. 2023AWQ: Activation-aware Weight Quantization for LLM Compression and Accelerationcited by 1,298
  9. 2025GR00T N1: An Open Foundation Model for Generalist Humanoid Robotscited by 1,225
  10. 2024QuaRot: Outlier-Free 4-Bit Inference in Rotated LLMscited by 634
  11. 2023OmniQuant: Omnidirectionally Calibrated Quantization for Large Language Modelscited by 424
  12. 2024QServe: W4A8KV4 Quantization and System Co-design for Efficient LLM Servingcited by 208
8 more
  1. 2024SVDQuant: Absorbing Outliers by Low-Rank Components for 4-Bit Diffusion Modelscited by 170
  2. 2024DuQuant: Distributing Outliers via Dual Transformation Makes Stronger Quantized LLMscited by 167
  3. 2024FlatQuant: Flatness Matters for LLM Quantizationcited by 131
  4. 2025OstQuant: Refining Large Language Model Quantization with Orthogonal and Scaling Transformations for Better Distribution Fittingcited by 97
  5. 2024PTQ4DiT: Post-training Quantization for Diffusion Transformerscited by 79
  6. 2025The Geometry of LLM Quantization: GPTQ as Babai's Nearest Plane Algorithmcited by 20
  7. 2025PoTPTQ: A Two-step Power-of-Two Post-training for LLMscited by 3
  8. 2026OJBKQ: Objective-Joint Babai-Klein Quantizationcited by 1

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 2.

2 papers.

  1. 2026FoldQuantVLA: Native Low-Bit Quantization of Vision-Language-Action Models via Consistent Foldingcited by 0
  2. 2026Beyond In-Distribution Preservation: Recovering Generalization in Quantized VLAs via Vulnerability-Oriented Tuningcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026QuantVLA: Scale-Calibrated Post-Training Quantization for Vision-Language-Action Models
  2. 2026VLAQuantBench: Closed-Loop Evaluation of Post-Training Quantization for Vision-Language-Action Models
  3. 2026HBVLA: Pushing 1-Bit Post-Training Quantization for Vision-Language-Action Models
  4. 2026CHASE-VLA: Post-Training Quantization Framework for Vision-Language-Action Models with Chunk-Aware Scale Estimation
  5. 2026ActQuant: Sub-4-bit Action-Guided Quantization for Vision-Language-Action Models
  6. 2026DA-PTQ: Drift-Aware Post-Training Quantization for Efficient Vision-Language-Action Models
  7. 2025SQAP-VLA: A Synergistic Quantization-Aware Pruning Framework for High-Performance Vision-Language-Action Models
  8. 2026DyQ-VLA: Temporal-Dynamic-Aware Quantization for Embodied Vision-Language-Action Models