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

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Chen, Dingwei, Liu, Ziqiang, Fang, Feiteng, Leong, Chak Tou, Ni, Shiwen, Argha, Ahmadreza, Alinejad-Rokny, Hamid, Yang, Min, Li, Chengming
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866911364981719040
author Chen, Dingwei
Liu, Ziqiang
Fang, Feiteng
Leong, Chak Tou
Ni, Shiwen
Argha, Ahmadreza
Alinejad-Rokny, Hamid
Yang, Min
Li, Chengming
author_facet Chen, Dingwei
Liu, Ziqiang
Fang, Feiteng
Leong, Chak Tou
Ni, Shiwen
Argha, Ahmadreza
Alinejad-Rokny, Hamid
Yang, Min
Li, Chengming
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2506_02973
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Expanding before Inferring: Enhancing Factuality in Large Language Models through Premature Layers Interpolation
Chen, Dingwei
Liu, Ziqiang
Fang, Feiteng
Leong, Chak Tou
Ni, Shiwen
Argha, Ahmadreza
Alinejad-Rokny, Hamid
Yang, Min
Li, Chengming
Computation and Language
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.
title Expanding before Inferring: Enhancing Factuality in Large Language Models through Premature Layers Interpolation
topic Computation and Language
url https://arxiv.org/abs/2506.02973