# Likelihood-based Mitigation of Evaluation Bias in Large Language Models

arXiv: https://arxiv.org/abs/2402.15987 · HTML: https://mlsanity.com/paper/2402.15987/

## Key facts

- Cited by 9 papers in the map (data through 14 Oct 2026).
- Builds on 11 of its 18 references that are in the map.
- First posted on arXiv 25 Feb 2024.

## Builds on (most cited first, 11 of 11)

- [Llama 2: Open Foundation and Fine-Tuned Chat Models](https://mlsanity.com/paper/2307.09288/) (2023)
- [Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena](https://mlsanity.com/paper/2306.05685/) (2023)
- [G-Eval: NLG Evaluation using GPT-4 with Better Human Alignment](https://mlsanity.com/paper/2303.16634/) (2023)
- [Can Large Language Models Be an Alternative to Human Evaluations?](https://mlsanity.com/paper/2305.01937/) (2023)
- [Large Language Models Are State-of-the-Art Evaluators of Translation Quality](https://mlsanity.com/paper/2302.14520/) (2023)
- [Experts, Errors, and Context: A Large-Scale Study of Human Evaluation for Machine Translation](https://mlsanity.com/paper/2104.14478/) (2021)
- [To Ship or Not to Ship: An Extensive Evaluation of Automatic Metrics for Machine Translation](https://mlsanity.com/paper/2107.10821/) (2021)
- [OpenMEVA: A Benchmark for Evaluating Open-ended Story Generation Metrics](https://mlsanity.com/paper/2105.08920/) (2021)
- [Evaluating Gender Bias of Pre-trained Language Models in Natural Language Inference by Considering All Labels](https://mlsanity.com/paper/2309.09697/) (2023)
- [The Impact of Debiasing on the Performance of Language Models in Downstream Tasks is Underestimated](https://mlsanity.com/paper/2309.09092/) (2023)
- [Language Models as an Alternative Evaluator of Word Order Hypotheses: A Case Study in Japanese](https://mlsanity.com/paper/2005.00842/) (2020)

## Built on by (most cited first, 9 of 9)

- [What Makes a Good Story and How Can We Measure It? A Comprehensive Survey of Story Evaluation](https://mlsanity.com/paper/2408.14622/) (2024)
- [A Survey on Human-AI Collaboration with Large Foundation Models](https://mlsanity.com/paper/2403.04931/) (2024)
- [A Little Leak Will Sink a Great Ship: Survey of Transparency for Large Language Models from Start to Finish](https://mlsanity.com/paper/2403.16139/) (2024)
- [JailNewsBench: Multi-Lingual and Regional Benchmark for Fake News Generation under Jailbreak Attacks](https://mlsanity.com/paper/2603.01291/) (2026)
- [Perplexity Trap: PLM-Based Retrievers Overrate Low Perplexity Documents](https://mlsanity.com/paper/2503.08684/) (2025)
- [EQUATOR: A Deterministic Framework for Evaluating LLM Reasoning with Open-Ended Questions. # v1.0.0-beta](https://mlsanity.com/paper/2501.00257/) (2024)
- [MLLM-as-a-Judge Exhibits Model Preference Bias](https://mlsanity.com/paper/2604.11589/) (2026)
- [When Noise Lowers The Loss: Rethinking Likelihood-Based Evaluation in Music Large Language Models](https://mlsanity.com/paper/2602.02738/) (2026)
- [Exposure is not manifestation: measurement target and output resolution jointly determine which behavioural-faithfulness evaluator wins](https://mlsanity.com/paper/2607.09306/) (2026)

## Abstract (arXiv)

Large Language Models (LLMs) are widely used to evaluate natural language generation tasks as automated metrics. However, the likelihood, a measure of LLM's plausibility for a sentence, can vary due to superficial differences in sentences, such as word order and sentence structure. It is therefore possible that there might be a likelihood bias if LLMs are used for evaluation: they might overrate sentences with higher likelihoods while underrating those with lower likelihoods. In this paper, we investigate the presence and impact of likelihood bias in LLM-based evaluators. We also propose a method to mitigate the likelihood bias. Our method utilizes highly biased instances as few-shot examples for in-context learning. Our experiments in evaluating the data-to-text and grammatical error correction tasks reveal that several LLMs we test display a likelihood bias. Furthermore, our proposed method successfully mitigates this bias, also improving evaluation performance (in terms of correlation of models with human scores) significantly.