Monitoring Decoding: Mitigating Hallucination via Evaluating the Factuality of Partial Response during Generation

2025-03-05

AI Summary

Large language models (LLMs) frequently generate factually inconsistent outputs—so-called “factual hallucinations.” Existing multi-sample self-consistency approaches suffer from high latency and fail to correct high-confidence errors. To address this, we propose an online factual monitoring and tree-based intervention mechanism operating during autoregressive decoding: at each token step, it dynamically assesses the factual plausibility of partial generations, identifies high-risk factual keywords, and triggers localized re-decoding to rectify inconsistencies. Crucially, our method abandons the costly full-sequence resampling paradigm, enabling real-time, fine-grained, low-overhead factual calibration *during* generation. Evaluated across multiple factual consistency benchmarks, our approach achieves significantly higher factual accuracy than self-consistency baselines while reducing inference latency by over 40% and substantially lowering computational overhead.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Fact-Checking / Misinformation Detection (NLP Focus)
• Reasoning under Uncertainty: Sequential Decision Making

Application Category

• Search and Retrieval-Augmented AI: Large language models for search
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

While large language models have demonstrated exceptional performance across a wide range of tasks, they remain susceptible to hallucinations -- generating plausible yet factually incorrect contents. Existing methods to mitigating such risk often rely on sampling multiple full-length generations, which introduces significant response latency and becomes ineffective when the model consistently produces hallucinated outputs with high confidence. To address these limitations, we introduce Monitoring Decoding (MD), a novel framework that dynamically monitors the generation process and selectively applies in-process interventions, focusing on revising crucial tokens responsible for hallucinations. Instead of waiting until completion of multiple full-length generations, we identify hallucination-prone tokens during generation using a monitor function, and further refine these tokens through a tree-based decoding strategy. This approach ensures an enhanced factual accuracy and coherence in the generated output while maintaining efficiency. Experimental results demonstrate that MD consistently outperforms self-consistency-based approaches in both effectiveness and efficiency, achieving higher factual accuracy while significantly reducing computational overhead.

Problem

• Mitigates hallucinations in large language models
• Dynamically monitors and revises hallucination-prone tokens
• Improves factual accuracy and reduces computational overhead

Innovation

• Dynamic monitoring of generation process
• Selective in-process intervention on tokens
• Tree-based decoding for factual accuracy