Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

Q-VLM: Post-training Quantization for Large Vision-Language Models

About

In this paper, we propose a post-training quantization framework of large vision-language models (LVLMs) for efficient multi-modal inference. Conventional quantization methods sequentially search the layer-wise rounding functions by minimizing activation discretization errors, which fails to acquire optimal quantization strategy without considering cross-layer dependency. On the contrary, we mine the cross-layer dependency that significantly influences discretization errors of the entire vision-language model, and embed this dependency into optimal quantization strategy searching with low search cost. Specifically, we observe the strong correlation between the activation entropy and the cross-layer dependency concerning output discretization errors. Therefore, we employ the entropy as the proxy to partition blocks optimally, which aims to achieve satisfying trade-offs between discretization errors and the search cost. Moreover, we optimize the visual encoder to disentangle the cross-layer dependency for fine-grained decomposition of search space, so that the search cost is further reduced without harming the quantization accuracy. Experimental results demonstrate that our method compresses the memory by 2.78x and increase generate speed by 1.44x about 13B LLaVA model without performance degradation on diverse multi-modal reasoning tasks. Code is available at https://github.com/ChangyuanWang17/QVLM.

Changyuan Wang, Ziwei Wang, Xiuwei Xu, Yansong Tang, Jie Zhou, Jiwen Lu• 2024

Related benchmarks

TaskDatasetResultRank
Visual Question AnsweringVQA v2
Accuracy79.65
1165
Visual Question AnsweringVizWiz
Accuracy53.69
1043
Visual Question AnsweringScienceQA (test)
Accuracy89.81
95
Visual Question AnsweringSQA
Accuracy72.27
23
Multimodal Question AnsweringScienceQA v1.3 (test)
NAT Score0.8954
21
Hateful Meme DetectionHateful Memes (val test)
Accuracy51.05
15
Visual Question AnsweringVizWiz (val test)
Accuracy37.6
15
Multi-modal ReasoningScience QA v1.3-13B (test)
Time8.9
7
Showing 8 of 8 rows

Other info

Code

Follow for update