Skip to yearly menu bar Skip to main content

Workshop: Instruction Tuning and Instruction Following

Sight Beyond Text: Multi-Modal Training Enhances LLMs in Truthfulness and Ethics

Haoqin Tu · Bingchen Zhao · Chen Wei · Cihang Xie

Keywords: [ alignment ] [ visual instruction tuning ] [ large language model ] [ Multi-modality ]


Multi-modal large language models (MLLMs) are trained based on large language models (LLM), with an enhanced capability to comprehend multi-modal inputs and generate textual responses. While they excel in multi-modal tasks, the pure NLP abilities of MLLMs are often underestimated and left untested.In this study, we get out of the box and unveil an intriguing characteristic of MLLMs --- our preliminary results suggest that visual instruction tuning, a prevailing strategy for transitioning LLMs into MLLMs, unexpectedly and interestingly helps models attain both improved truthfulness and ethical alignment in the pure NLP context. For example, a visual-instruction-tuned LLaMA2 7B model surpasses the performance of the LLaMA2-chat 7B model, fine-tuned with over one million human annotations, on \texttt{TruthfulQA} and \texttt{Ethics} benchmarks. Further analysis reveals that the improved alignment can be attributed to the superior instruction quality inherent to visual-text data. In releasing our code at \url{}, we aspire to foster further exploration into the intrinsic value of visual-text synergies and, in a broader scope, multi-modal interactions in alignment research.

Chat is not available.