[论文解读] Negation Blindness in Large Language Models: Unveiling the NO Syndrome in Image Generation

Mohammad Nadeem, Shahab Saquib Sohail+3 more|arXiv (Cornell University)|Aug 27, 2024

Speech and dialogue systems被引用 7

一句话总结

该论文在 LLMs 的图像生成中识别出一个根本性局限：否定盲点（NO syndrome），表明像 GPT-4、Gemini 和 Copilot 这样的模型在针对生成图像的否定提示时难以给出正确回答，并提出一个上下文感知的 RL 反馈回路作为解决方案。

ABSTRACT

Foundational Large Language Models (LLMs) have changed the way we perceive technology. They have been shown to excel in tasks ranging from poem writing and coding to essay generation and puzzle solving. With the incorporation of image generation capability, they have become more comprehensive and versatile AI tools. At the same time, researchers are striving to identify the limitations of these tools to improve them further. Currently identified flaws include hallucination, biases, and bypassing restricted commands to generate harmful content. In the present work, we have identified a fundamental limitation related to the image generation ability of LLMs, and termed it The NO Syndrome. This negation blindness refers to LLMs inability to correctly comprehend NO related natural language prompts to generate the desired images. Interestingly, all tested LLMs including GPT-4, Gemini, and Copilot were found to be suffering from this syndrome. To demonstrate the generalization of this limitation, we carried out simulation experiments and conducted entropy-based and benchmark statistical analysis tests on various LLMs in multiple languages, including English, Hindi, and French. We conclude that the NO syndrome is a significant flaw in current LLMs that needs to be addressed. A related finding of this study showed a consistent discrepancy between image and textual responses as a result of this NO syndrome. We posit that the introduction of a negation context-aware reinforcement learning based feedback loop between the LLMs textual response and generated image could help ensure the generated text is based on both the LLMs correct contextual understanding of the negation query and the generated visual output.

研究动机与目标

• 识别并描述与否定理解相关的基于 LLM 的图像生成中的基本局限性。
• 证明多种领先的 LLM 在不同语言中都存在否定盲点。
• 分析否定提示对文本输出与视觉输出的影响。
• 提出可能的机制以及用于缓解 NO syndrome 的基于反馈的方法。

提出的方法

• 在多种语言（英语、印地语、法语）对若干 LLM 进行仿真实验。
• 应用基于熵的和基准统计分析来评估否定处理。
• 将文本响应与相应生成的图像进行比较以量化 NO syndrome。
• 提出一个否定情境感知的强化学习反馈回路，将 LLM 的文本输出与生成的视觉内容联系起来。

实验结果

研究问题

• RQ1 在处理否定提示时，当前的 LLM 是否在图像生成中表现出否定盲点（NO syndrome）？
• RQ2 NO syndrome 是否在不同模型（例如 GPT-4、Gemini、Copilot）和语言之间保持一致？
• RQ3 在否定提示下，文本响应与生成的图像之间的差异是什么？
• RQ4 是否可以通过一个否定情境感知的 RL 反馈机制来改善文本输出与图像输出之间的一致性？

主要发现

• 包括 GPT-4、Gemini 和 Copilot 在内的 LLM 显示出在图像生成中的否定盲点。
• NO syndrome 在多种语言（英语、印地语、法语）中被观察到。
• 在否定提示下，图像输出与相应文本响应之间存在一贯的差异。