arXiv cs.AI· Arash Akbari, Arman Akbari, Masih Eskandar, Qitao Tan, Yixiao Chen, Jingwu Luo, Bertha Pangaribuan, Liyun Zhang, Jennifer Dy, Geng Yuan, Xue Lin, Gaowen Liu, Stratis Ioannidis, Yanzhi Wang·· 4 小时前AI 评分50
ActQuant：面向视觉-语言-动作模型的 4-bit 以下动作引导权重量化
ActQuant: Sub-4-bit Action-Guided Backbone Weight-Only Quantization for Vision-Language-Action Models
AI 导读
研究者提出动作引导混合精度训练后量化框架 ActQuant，针对 VLA 模型进行 4-bit 以下权重量化。
正文 · 原文
Authors:Arash Akbari, Arman Akbari, Masih Eskandar, Qitao Tan, Yixiao Chen, Jingwu Luo, Bertha Pangaribuan, Liyun Zhang, Jennifer Dy, Geng Yuan, Xue Lin, Gaowen Liu, Stratis Ioannidis, Yanzhi Wang
Abstract:Vision-Language-Action (VLA) models exhibit remarkable action generation for embodied intelligence, but their heavy compute make deployment on edge platforms impractical. Aggressive, sub-4-bit weight quantization is the natural solution, yet existing post-training quantization (PTQ) methods suffer severe performance degradation in this regime. To address this, we introduce ActQuant, an action-guided mixed-precision PTQ framework that operates in two stages: (1) an inter-tensor bit allocator that assigns each weight matrix a single bit-width based on how much it contributes to predicting the agent's actions; (2) an intra-tensor scale optimizer tunes per-block quantization scales using action-aware curvature, so that dynamic range is concentrated on the weights most influential for control. To deliver the on-device benefits of our aggressive quantization, we further introduce this http URL, an agentic conversion pipeline that ports architectures into a native C/C++ runtime with efficient low-bit kernels. We evaluate ActQuant both in simulation and on a real-world 6-DoF UR3 arm, with all models deployed through this http URL. On the LIBERO benchmark, ActQuant is the only method that operates at or below 3 bits-per-weight, retaining 95.0% on OpenVLA-OFT and 94.8% on $\pi_{0.5}$. Pushed further, ActQuant reaches 2.5 bpw at 90.1% on OpenVLA-OFT, compressing the backbone from 14.3 GB to 2.7 GB (5.3$\times$). On the physical UR3 arm, $\pi_{0.5}$ quantized with ActQuant retains the baseline's success rate while reducing the memory footprint by 2.5$\times$.
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2605.24011 [cs.CV] |
| (or arXiv:2605.24011v4 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2605.24011 arXiv-issued DOI via DataCite |
Submission history
From: Arash Akbari [view email]
[v1]
Tue, 19 May 2026 19:57:26 UTC (1,871 KB)
[v2]
Thu, 4 Jun 2026 20:07:22 UTC (1,871 KB)
[v3]
Mon, 3 Aug 2026 19:37:08 UTC (1,871 KB)
[v4]
Thu, 5 Nov 2026 06:15:12 UTC (1,781 KB)
来源：arXiv cs.AI · arxiv.org