arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2611.06075cs.AI
收藏

UniCompress:从多LoRA专家模型到面向边缘智能的单个紧凑模型

UniCompress: From Multi-LoRA Experts to One Compact Model for Edge Intelligence

Juhaokan Technology Co., Ltd.(聚好看科技股份有限公司)

Mingkang Liu, Peipei Jiao, Weiming Meng, Tianyuan Liu, Lichen Pang, Junyan Li

AI总结:

UniCompress是将多LoRA专家转化为无适配器紧凑模型的框架,通过教师引导联合量化压缩VLMs,在四项具身任务上实现76%存储缩减,性能优于GPTQ-INT2且接近BF16教师模型。

摘要:

具身与边缘系统越来越依赖视觉语言模型(VLMs)来支持异构技能,但为每项技能单独部署任务适配器同时保留高精度主干会产生巨大的内存和管理成本。本文提出UniCompress,这是一种面向部署的框架,可将多任务专业知识转化为单个无适配器的紧凑模型。从合并的BF16精度Qwen3-VL-4B教师模型出发,UniCompress使用GPTQ初始化和教师引导的量化恢复构建非对称INT2精度学生模型。该方法对视觉编码器和语言主干中的投影权重进行联合量化与恢复,在保留多任务行为的同时,实现了完整视觉语言模型的激进压缩。我们在四项异构具身任务上评估UniCompress:姿态识别、家庭安全理解、多模态对话路由以及移动GUI动作预测。UniCompress将近似权重存储从BF16精度下的8.9 GB减少到INT2精度下的2.1 GB,对应减少约76%。它的INT2精度宏任务得分比GPTQ-INT2提高了34.3个百分点,同时与合并的BF16精度教师模型的得分差距保持在0.9个百分点以内。这些结果证明了教师引导的联合视觉语言INT2恢复在资源受限的具身平台上部署单个多技能VLM的潜力。

英文摘要:

Embodied and edge systems increasingly rely on vision-language models (VLMs) to support heterogeneous skills, yet deploying a separate task adapter for each skill while retaining a high-precision backbone creates substantial memory and management costs. We present UniCompress, a deployment-oriented framework that converts multi-task expertise into one compact, adapter-free model. Starting from a consolidated BF16 Qwen3-VL-4B teacher, UniCompress constructs an asymmetric INT2 student using GPTQ initialization and teacher-guided quantization recovery. The approach jointly quantizes and recovers the projection weights in the vision encoder and language backbone, enabling aggressive compression of the complete vision-language model while preserving its multi-task behavior. We evaluate UniCompress on four heterogeneous embodied tasks: posture recognition, household-safety understanding, multimodal dialogue routing, and mobile-GUI action prediction. UniCompress reduces the approximate weight storage from 8.9 GB in BF16 to 2.1 GB in INT2, corresponding to a reduction of approximately 76%. It improves the INT2 macro task score by 34.3 percentage points over GPTQ-INT2 while remaining within 0.9 points of the consolidated BF16 teacher. These results demonstrate the potential of teacher-guided joint vision-language INT2 recovery for deploying a single multi-skill VLM on resource-constrained embodied platforms.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑