LLMs are Frequency Pattern Learners in Natural Language Inference

• URL: http://arxiv.org/abs/2505.21011v1
• Date: Tue, 27 May 2025 10:45:29 GMT
• Title: LLMs are Frequency Pattern Learners in Natural Language Inference
• Authors: Liang Cheng, Zhaowei Wang, Mark Steedman,
• Abstract summary: We identify a consistent frequency bias, where predicates in hypotheses occur more frequently than those in premises for positive instances.<n>We find that LLMs exploit frequency bias for inference and perform poorly on adversarial instances.<n>We compute the frequencies of hyponyms and their corresponding hypernyms from WordNet, revealing a correlation between frequency bias and textual entailment.
• Score: 10.931224681685194
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: While fine-tuning LLMs on NLI corpora improves their inferential performance, the underlying mechanisms driving this improvement remain largely opaque. In this work, we conduct a series of experiments to investigate what LLMs actually learn during fine-tuning. We begin by analyzing predicate frequencies in premises and hypotheses across NLI datasets and identify a consistent frequency bias, where predicates in hypotheses occur more frequently than those in premises for positive instances. To assess the impact of this bias, we evaluate both standard and NLI fine-tuned LLMs on bias-consistent and bias-adversarial cases. We find that LLMs exploit frequency bias for inference and perform poorly on adversarial instances. Furthermore, fine-tuned LLMs exhibit significantly increased reliance on this bias, suggesting that they are learning these frequency patterns from datasets. Finally, we compute the frequencies of hyponyms and their corresponding hypernyms from WordNet, revealing a correlation between frequency bias and textual entailment. These findings help explain why learning frequency patterns can enhance model performance on inference tasks.

Related papers

• Mitigating Spurious Correlations in LLMs via Causality-Aware Post-Training [57.03005244917803]
  Large language models (LLMs) often fail on out-of-distribution (OOD) samples due to spurious correlations acquired during pre-training.<n>Here, we aim to mitigate such spurious correlations through causality-aware post-training (CAPT)<n> Experiments on the formal causal inference benchmark CLadder and the logical reasoning dataset PrOntoQA show that 3B-scale language models fine-tuned with CAPT can outperform both traditional SFT and larger LLMs on in-distribution (ID) and OOD tasks.
  arXiv  Detail & Related papers  (2025-06-11T06:30:28Z)
• Benchmarking and Pushing the Multi-Bias Elimination Boundary of LLMs via Causal Effect Estimation-guided Debiasing [41.90664134611629]
  Current large language models (LLMs) may still utilize bias during inference, leading to the poor generalizability of LLMs.<n>We propose a multi-bias benchmark where each piece of data contains five types of biases.<n>We show that CMBE can effectively eliminate multiple types of bias simultaneously to enhance the generalizability of LLMs.
  arXiv  Detail & Related papers  (2025-05-22T11:04:09Z)
• Neutralizing Bias in LLM Reasoning using Entailment Graphs [13.5088417466172]
  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.<n>We design an unsupervised framework to construct counterfactual reasoning data and fine-tune LLMs to reduce attestation bias.<n>Our framework consistently improves inferential performance on both original and bias-neutralized NLI datasets.
  arXiv  Detail & Related papers  (2025-03-14T17:33:30Z)
• How far can bias go? -- Tracing bias from pretraining data to alignment [54.51310112013655]
  This study examines the correlation between gender-occupation bias in pre-training data and their manifestation in LLMs.<n>Our findings reveal that biases present in pre-training data are amplified in model outputs.
  arXiv  Detail & Related papers  (2024-11-28T16:20:25Z)
• Justice or Prejudice? Quantifying Biases in LLM-as-a-Judge [84.34545223897578]
  Despite their excellence in many domains, potential issues are under-explored, undermining their reliability and the scope of their utility. We identify 12 key potential biases and propose a new automated bias quantification framework-CALM- which quantifies and analyzes each type of bias in LLM-as-a-Judge. Our work highlights the need for stakeholders to address these issues and remind users to exercise caution in LLM-as-a-Judge applications.
  arXiv  Detail & Related papers  (2024-10-03T17:53:30Z)
• Tuning Frequency Bias of State Space Models [48.60241978021799]
  State space models (SSMs) leverage linear, time-invariant (LTI) systems to learn sequences with long-range dependencies. We find that SSMs exhibit an implicit bias toward capturing low-frequency components more effectively than high-frequency ones.
  arXiv  Detail & Related papers  (2024-10-02T21:04:22Z)
• A Theory of LLM Sampling: Part Descriptive and Part Prescriptive [53.08398658452411]
  Large Language Models (LLMs) are increasingly utilized in autonomous decision-making.<n>We show that this sampling behavior resembles that of human decision-making.<n>We show that this deviation of a sample from the statistical norm towards a prescriptive component consistently appears in concepts across diverse real-world domains.
  arXiv  Detail & Related papers  (2024-02-16T18:28:43Z)
• The Gaps between Pre-train and Downstream Settings in Bias Evaluation and Debiasing [74.7319697510621]
  In-Context Learning (ICL) induces smaller changes to PLMs compared to FT-based debiasing methods. ICL-based debiasing methods show a higher correlation between intrinsic and extrinsic bias scores compared to FT-based methods.
  arXiv  Detail & Related papers  (2024-01-16T17:15:08Z)
• An Empirical Study of Catastrophic Forgetting in Large Language Models During Continual Fine-tuning [70.48605869773814]
  Catastrophic forgetting (CF) is a phenomenon that occurs in machine learning when a model forgets previously learned information.<n>This study empirically evaluates the forgetting phenomenon in large language models during continual instruction tuning.
  arXiv  Detail & Related papers  (2023-08-17T02:53:23Z)
• Sources of Hallucination by Large Language Models on Inference Tasks [16.644096408742325]
  Large Language Models (LLMs) are claimed to be capable of Natural Language Inference (NLI) We present a series of behavioral studies on several LLM families which probe their behavior using controlled experiments.
  arXiv  Detail & Related papers  (2023-05-23T22:24:44Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.