arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2412.01487cs.AI
收藏

FastRM:一种面向多模态生成模型的高效自动可解释性框架

FastRM: An efficient and automatic explainability framework for multimodal generative models

Intel Labs(英特尔实验室);Hugging Face

Gabriela Ben-Melech Stan, Estelle Aflalo, Man Luo, Shachar Rosenman, Tiep Le, Sayak Paul, Shao-Yen Tseng, Vasudev Lal

AI总结:

本文提出FastRM框架,通过高效预测大型视觉语言模型的相关性图并评估模型置信度,显著降低计算时间与内存占用,推动可解释AI在实际应用中的部署。

摘要:

大型视觉语言模型(LVLMs)在文本和视觉输入上展现了卓越的推理能力。然而,这些模型仍然容易生成错误信息。识别并缓解无根据的响应对于开发可信赖的人工智能至关重要。传统的可解释性方法,如基于梯度的相关性图,能够揭示模型的决策过程,但通常计算开销大,不适合实时输出验证。在本工作中,我们提出了FastRM,一种用于预测LVLMs可解释相关性图的高效方法。此外,FastRM还提供了对模型置信度的定量和定性评估。实验结果表明,与传统相关性图生成方法相比,FastRM实现了99.8%的计算时间减少和44.4%的内存占用减少。FastRM使可解释人工智能更加实用和可扩展,从而促进其在现实应用中的部署,并使用户能够更有效地评估模型输出的可靠性。

英文摘要:

Large Vision Language Models (LVLMs) have demonstrated remarkable reasoning capabilities over textual and visual inputs. However, these models remain prone to generating misinformation. Identifying and mitigating ungrounded responses is crucial for developing trustworthy AI. Traditional explainability methods such as gradient-based relevancy maps, offer insight into the decision process of models, but are often computationally expensive and unsuitable for real-time output validation. In this work, we introduce FastRM, an efficient method for predicting explainable Relevancy Maps of LVLMs. Furthermore, FastRM provides both quantitative and qualitative assessment of model confidence. Experimental results demonstrate that FastRM achieves a 99.8% reduction in computation time and a 44.4% reduction in memory footprint compared to traditional relevancy map generation. FastRM allows explainable AI to be more practical and scalable, thereby promoting its deployment in real-world applications and enabling users to more effectively evaluate the reliability of model outputs.

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。

↑