arXiv cs.AI· Mingkang Liu, Peipei Jiao, Weiming Meng, Tianyuan Liu, Lichen Pang, Junyan Li·· 2 小时前AI 评分36
UniCompress：将多任务多 LoRA 专家压缩为一个紧凑模型，面向边缘智能
UniCompress: From Multi-LoRA Experts to One Compact Model for Edge Intelligence
AI 导读
arXiv 论文 UniCompress 提出面向部署的压缩框架，把多任务能力转化为单个无适配器的紧凑模型。
正文 · 原文
Abstract: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.
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2611.06075 [cs.AI] |
| (or arXiv:2611.06075v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2611.06075 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Lichen Pang [view email]
[v1]
Wed, 4 Nov 2026 07:18:55 UTC (182 KB)
来源：arXiv cs.AI · arxiv.org