Pairwise or Pointwise? Evaluating Feedback Protocols for Bias in LLM-Based Evaluation

• URL: http://arxiv.org/abs/2504.14716v1
• Date: Sun, 20 Apr 2025 19:05:59 GMT
• Title: Pairwise or Pointwise? Evaluating Feedback Protocols for Bias in LLM-Based Evaluation
• Authors: Tuhina Tripathi, Manya Wadhwa, Greg Durrett, Scott Niekum,
• Abstract summary: We show that the choice of feedback protocol can significantly affect evaluation reliability and induce systematic biases.<n>In particular, we show that pairwise evaluation protocols are more vulnerable to distracted evaluation.
• Score: 57.380464382910375
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Large Language Models (LLMs) are widely used as proxies for human labelers in both training (Reinforcement Learning from AI Feedback) and large-scale response evaluation (LLM-as-a-judge). Alignment and evaluation are critical components in the development of reliable LLMs, and the choice of feedback protocol plays a central role in both but remains understudied. In this work, we show that the choice of feedback protocol (absolute scores versus relative preferences) can significantly affect evaluation reliability and induce systematic biases. In particular, we show that pairwise evaluation protocols are more vulnerable to distracted evaluation. Generator models can exploit spurious attributes (or distractor features) favored by the LLM judge, resulting in inflated scores for lower-quality outputs and misleading training signals. We find that absolute scoring is more robust to such manipulation, producing judgments that better reflect response quality and are less influenced by distractor features. Our results demonstrate that generator models can flip preferences by embedding distractor features, skewing LLM-as-a-judge comparisons and leading to inaccurate conclusions about model quality in benchmark evaluations. Pairwise preferences flip in about 35% of the cases, compared to only 9% for absolute scores. We offer recommendations for choosing feedback protocols based on dataset characteristics and evaluation objectives.

Related papers

• HREF: Human Response-Guided Evaluation of Instruction Following in Language Models [61.273153125847166]
  We develop a new evaluation benchmark, Human Response-Guided Evaluation of Instruction Following (HREF)<n>In addition to providing reliable evaluation, HREF emphasizes individual task performance and is free from contamination.<n>We study the impact of key design choices in HREF, including the size of the evaluation set, the judge model, the baseline model, and the prompt template.
  arXiv  Detail & Related papers  (2024-12-20T03:26:47Z)
• Diverging Preferences: When do Annotators Disagree and do Models Know? [92.24651142187989]
  We develop a taxonomy of disagreement sources spanning 10 categories across four high-level classes. We find that the majority of disagreements are in opposition with standard reward modeling approaches. We develop methods for identifying diverging preferences to mitigate their influence on evaluation and training.
  arXiv  Detail & Related papers  (2024-10-18T17:32:22Z)
• Direct Judgement Preference Optimization [66.83088028268318]
  We train large language models (LLMs) as generative judges to evaluate and critique other models' outputs. We employ three approaches to collect the preference pairs for different use cases, each aimed at improving our generative judge from a different perspective. Our model robustly counters inherent biases such as position and length bias, flexibly adapts to any evaluation protocol specified by practitioners, and provides helpful language feedback for improving downstream generator models.
  arXiv  Detail & Related papers  (2024-09-23T02:08:20Z)
• Aligning Model Evaluations with Human Preferences: Mitigating Token Count Bias in Language Model Assessments [2.1370543868467275]
  This follow-up paper explores methods to align Large Language Models evaluator preferences with human evaluations. We employed Bayesian statistics and a t-test to quantify this bias and developed a recalibration procedure to adjust the GPTScorer. Our findings significantly improve aligning the recalibrated LLM evaluator with human evaluations across multiple use cases.
  arXiv  Detail & Related papers  (2024-07-05T09:26:40Z)
• The Comparative Trap: Pairwise Comparisons Amplifies Biased Preferences of LLM Evaluators [31.520403357740317]
  Large language models (LLMs) are increasingly used as evaluators for natural language generation tasks.<n>LLMs display biased preferences, such as favoring verbosity and authoritative tones.<n>We introduce PRePair, which integrates pointwise reasoning within a pairwise framework.
  arXiv  Detail & Related papers  (2024-06-18T06:43:04Z)
• Fairer Preferences Elicit Improved Human-Aligned Large Language Model Judgments [41.25558612970942]
  We show that large language models (LLMs) exhibit preference biases and worrying sensitivity to prompt designs. Motivated by this phenomenon, we propose an automatic Zero-shot Evaluation-oriented Prompt Optimization framework, ZEPO.
  arXiv  Detail & Related papers  (2024-06-17T09:48:53Z)
• Aligning with Human Judgement: The Role of Pairwise Preference in Large Language Model Evaluators [48.54465599914978]
  Large Language Models (LLMs) have demonstrated promising capabilities as automatic evaluators in assessing the quality of generated natural language.<n>LLMs still exhibit biases in evaluation and often struggle to generate coherent evaluations that align with human assessments.<n>We introduce Pairwise-preference Search (PAIRS), an uncertainty-guided search-based rank aggregation method that employs LLMs to conduct pairwise comparisons locally and efficiently ranks candidate texts globally.
  arXiv  Detail & Related papers  (2024-03-25T17:11:28Z)
• Self-Evaluation Improves Selective Generation in Large Language Models [54.003992911447696]
  We reformulate open-ended generation tasks into token-level prediction tasks. We instruct an LLM to self-evaluate its answers. We benchmark a range of scoring methods based on self-evaluation.
  arXiv  Detail & Related papers  (2023-12-14T19:09:22Z)
• Peering Through Preferences: Unraveling Feedback Acquisition for Aligning Large Language Models [32.843361525236965]
  We analyze the effect of sparse feedback on the alignment and evaluation of large language models. We find that preferences from ratings and rankings significantly disagree 60% for both human and AI annotators. Our findings shed light on critical gaps in methods for evaluating the real-world utility of language models.
  arXiv  Detail & Related papers  (2023-08-30T07:35:32Z)

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.