Salvato in:
Dettagli Bibliografici
Autori principali: Cheng, Liang, Li, Tianyi, Wang, Zhaowei, Liu, Tianyang, Steedman, Mark
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:https://arxiv.org/abs/2503.11614
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866917957293047808
author Cheng, Liang
Li, Tianyi
Wang, Zhaowei
Liu, Tianyang
Steedman, Mark
author_facet Cheng, Liang
Li, Tianyi
Wang, Zhaowei
Liu, Tianyang
Steedman, Mark
contents LLMs are often claimed to be capable of Natural Language Inference (NLI), which is widely regarded as a cornerstone of more complex forms of reasoning. However, recent works show that LLMs still suffer from hallucinations in NLI due to attestation bias, where LLMs overly rely on propositional memory to build shortcuts. To solve the issue, we design an unsupervised framework to construct counterfactual reasoning data and fine-tune LLMs to reduce attestation bias. To measure bias reduction, we build bias-adversarial variants of NLI datasets with randomly replaced predicates in premises while keeping hypotheses unchanged. Extensive evaluations show that our framework can significantly reduce hallucinations from attestation bias. Then, we further evaluate LLMs fine-tuned with our framework on original NLI datasets and their bias-neutralized versions, where original entities are replaced with randomly sampled ones. Extensive results show that our framework consistently improves inferential performance on both original and bias-neutralized NLI datasets.
format Preprint
id arxiv_https___arxiv_org_abs_2503_11614
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Neutralizing Bias in LLM Reasoning using Entailment Graphs
Cheng, Liang
Li, Tianyi
Wang, Zhaowei
Liu, Tianyang
Steedman, Mark
Computation and Language
LLMs are often claimed to be capable of Natural Language Inference (NLI), which is widely regarded as a cornerstone of more complex forms of reasoning. However, recent works show that LLMs still suffer from hallucinations in NLI due to attestation bias, where LLMs overly rely on propositional memory to build shortcuts. To solve the issue, we design an unsupervised framework to construct counterfactual reasoning data and fine-tune LLMs to reduce attestation bias. To measure bias reduction, we build bias-adversarial variants of NLI datasets with randomly replaced predicates in premises while keeping hypotheses unchanged. Extensive evaluations show that our framework can significantly reduce hallucinations from attestation bias. Then, we further evaluate LLMs fine-tuned with our framework on original NLI datasets and their bias-neutralized versions, where original entities are replaced with randomly sampled ones. Extensive results show that our framework consistently improves inferential performance on both original and bias-neutralized NLI datasets.
title Neutralizing Bias in LLM Reasoning using Entailment Graphs
topic Computation and Language
url https://arxiv.org/abs/2503.11614