Counterfactual Debating with Preset Stances for Hallucination Elimination of LLMs
This addresses hallucination issues in LLMs, which is a critical problem for improving reliability in NLP applications, though it appears incremental as it builds on existing reasoning-based methods.
The paper tackles the hallucination problem in Large Language Models by proposing a CounterFactual Multi-Agent Debate framework that overrides inherent biases through preset stances and debate, achieving superior results on four datasets across three tasks.
Large Language Models (LLMs) excel in various natural language processing tasks but struggle with hallucination issues. Existing solutions have considered utilizing LLMs' inherent reasoning abilities to alleviate hallucination, such as self-correction and diverse sampling methods. However, these methods often overtrust LLMs' initial answers due to inherent biases. The key to alleviating this issue lies in overriding LLMs' inherent biases for answer inspection. To this end, we propose a CounterFactual Multi-Agent Debate (CFMAD) framework. CFMAD presets the stances of LLMs to override their inherent biases by compelling LLMs to generate justifications for a predetermined answer's correctness. The LLMs with different predetermined stances are engaged with a skeptical critic for counterfactual debate on the rationality of generated justifications. Finally, the debate process is evaluated by a third-party judge to determine the final answer. Extensive experiments on four datasets of three tasks demonstrate the superiority of CFMAD over existing methods.