Expanding before Inferring: Enhancing Factuality in Large Language Models through Premature Layers Interpolation

TL;DR

This paper introduces PLI, a training-free, plug-and-play method that improves factual accuracy in large language models by interpolating premature layers, effectively reducing hallucinations without extensive retraining.

Contribution

The paper proposes PLI, a novel layer interpolation technique that enhances factuality in LLMs without additional training, addressing hallucinations more efficiently.

Findings

PLI reduces hallucinations across multiple datasets.

PLI outperforms existing baseline methods.

Layer interpolation aligns with LLMs' internal mechanisms.

Abstract

Large Language Models (LLMs) demonstrate remarkable capabilities in text understanding and generation. However, their tendency to produce factually inconsistent outputs, commonly referred to as ''hallucinations'', remains a critical challenge. Existing approaches, such as retrieval-based and inference-time correction methods, primarily address this issue at the input or output level, often overlooking the intrinsic information refinement process and the role of premature layers. Meanwhile, alignment- and fine-tuning-based methods are resource-intensive. In this paper, we propose PLI (Premature Layers Interpolation), a novel, training-free, and plug-and-play intervention designed to enhance factuality. PLI mitigates hallucinations by inserting premature layers formed through mathematical interpolation with adjacent layers. Inspired by stable diffusion and sampling steps, PLI extends the depth of information processing and transmission in LLMs, improving factual coherence. Experiments on four publicly available datasets demonstrate that PLI effectively reduces hallucinations while outperforming existing baselines in most cases. Further analysis suggests that the success of layer interpolation is closely linked to LLMs' internal mechanisms. Our dataset and code are available at https://github.com/CuSO4-Chen/PLI.