Mitigating Selection Bias in Large Language Models via Permutation-Aware GRPO

• URL: http://arxiv.org/abs/2603.21016v1
• Date: Sun, 22 Mar 2026 02:29:40 GMT
• Title: Mitigating Selection Bias in Large Language Models via Permutation-Aware GRPO
• Abstract summary: Large language models often exhibit selection bias due to non-semantic factors like option positions and label symbols.<n>We propose Permutation-Aware Group Relative Policy Optimization (PA-GRPO), which mitigates selection bias by enforcing permutation-consistent semantic reasoning.
• Score: 5.442224574457385
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Large language models (LLMs) used for multiple-choice and pairwise evaluation tasks often exhibit selection bias due to non-semantic factors like option positions and label symbols. Existing inference-time debiasing is costly and may harm reasoning, while pointwise training ignores that the same question should yield consistent answers across permutations. To address this issue, we propose Permutation-Aware Group Relative Policy Optimization (PA-GRPO), which mitigates selection bias by enforcing permutation-consistent semantic reasoning. PA-GRPO constructs a permutation group for each instance by generating multiple candidate permutations, and optimizes the model using two complementary mechanisms: (1) cross-permutation advantage, which computes advantages relative to the mean reward over all permutations of the same instance, and (2) consistency-aware reward, which encourages the model to produce consistent decisions across different permutations. Experimental results demonstrate that PA-GRPO outperforms strong baselines across seven benchmarks, substantially reducing selection bias while maintaining high overall performance. The code will be made available on Github (https://github.com/ECNU-Text-Computing/PA-GRPO).

Related papers

• F-GRPO: Factorized Group-Relative Policy Optimization for Unified Candidate Generation and Ranking [79.49893545611779]
  Large Language Models (LLMs) can generate a subset and order it within a single autoregressive pass.<n>This flexibility introduces a new optimization challenge: the model must search an output space while receiving utility feedback only after the full ranked list is generated.<n>This credit assignment gap makes end-to-end optimization unstable and sample-inefficient.<n>We propose a unified framework that performs both within a single autoregressive rollout and optimize them end-to-end via factorized group-relative policy optimization.
  arXiv  Detail & Related papers  (2026-05-13T04:52:33Z)
• Efficient Test-Time Inference via Deterministic Exploration of Truncated Decoding Trees [68.04613115686509]
  Self-consistency boosts inference-time performance by sampling multiple reasoning traces in parallel and voting.<n>We propose Distinct Leafion (DLE), a deterministic decoding method that treats truncated sampling as a pruned tree and systematically enumerates distinct leaves instead of sampling with replacement.<n>DLE explores higher-quality reasoning traces, yielding better performance on math, coding, and general reasoning tasks.
  arXiv  Detail & Related papers  (2026-04-22T12:42:03Z)
• Balanced Aggregation: Understanding and Fixing Aggregation Bias in GRPO [70.38763678943648]
  Reinforcement learning with verifiable rewards (RLVR) has become a central paradigm for improving reasoning and code generation in large language models.<n>Standard GRPO uses sequence aggregation, while recent work has advocated token aggregation as a better alternative.<n>We show that these two rules induce different optimization biases; token aggregation introduces sign-length coupling, while sequence aggregation implicitly downweights longer responses.
  arXiv  Detail & Related papers  (2026-04-14T09:48:46Z)
• Succinct QUBO formulations for permutation problems by sorting networks [0.1590850178837849]
  We introduce a QUBO formulation for permutations using compare-exchange networks, with only $O(n log2 n)$ binary variables.<n>A central feature of our approach is each permutation corresponds to a unique variable assignment, enabling unbiased sampling.<n>We expect this method to be practically useful in areas where unbiased sampling of constrained permutations is important.
  arXiv  Detail & Related papers  (2026-03-08T10:37:52Z)
• Transform-Augmented GRPO Improves Pass@k [50.3707071191733]
  Group Relative Policy Optimization (GRPO) was designed to improve reasoning, but it worsens this situation through two failure modes.<n>We propose TA-GRPO (Transform-Augmented GRPO), which generates semantically equivalent transformed variants of each question.<n>This pooled computation ensures mixed rewards even when the original question is too easy or too hard, while training on diverse phrasings promotes multiple solution strategies.
  arXiv  Detail & Related papers  (2026-01-30T02:43:29Z)
• Quantifying and Mitigating Selection Bias in LLMs: A Transferable LoRA Fine-Tuning and Efficient Majority Voting Approach [13.829059542429876]
  Multiple Choice Question (MCQ) answering is a widely used method for evaluating the performance of Large Language Models (LLMs)<n>LLMs often exhibit selection bias in MCQ tasks, where their choices are influenced by factors like answer position or option symbols rather than the content.
  arXiv  Detail & Related papers  (2025-11-17T21:31:37Z)
• Reasoning Planning for Language Models [23.519351730129426]
  We introduce EPIC, an Ensemble Planning with Contrastive learning framework.<n>EPIC learns a shared representation space that captures both model reasoning abilities and query-method compatibility.<n> Experiments on diverse mathematical reasoning tasks show that EPIC consistently selects optimal reasoning methods.
  arXiv  Detail & Related papers  (2025-11-01T11:51:53Z)
• Learning Distributions over Permutations and Rankings with Factorized Representations [6.51628774380971]
  Learning distributions over permutations is a fundamental problem in machine learning.<n>We propose a novel approach that leverages alternative representations for permutations.<n>We show that our method learns nontrivial distributions even in the least expressive mode.
  arXiv  Detail & Related papers  (2025-05-30T14:53:40Z)
• Large Language Models Are Not Robust Multiple Choice Selectors [117.72712117510953]
  Multiple choice questions (MCQs) serve as a common yet important task format in the evaluation of large language models (LLMs) This work shows that modern LLMs are vulnerable to option position changes due to their inherent "selection bias" We propose a label-free, inference-time debiasing method, called PriDe, which separates the model's prior bias for option IDs from the overall prediction distribution.
  arXiv  Detail & Related papers  (2023-09-07T17:44:56Z)
• Compositional Generalization without Trees using Multiset Tagging and Latent Permutations [121.37328648951993]
  We phrase semantic parsing as a two-step process: we first tag each input token with a multiset of output tokens. Then we arrange the tokens into an output sequence using a new way of parameterizing and predicting permutations. Our model outperforms pretrained seq2seq models and prior work on realistic semantic parsing tasks.
  arXiv  Detail & Related papers  (2023-05-26T14:09:35Z)
• Probabilistic Permutation Graph Search: Black-Box Optimization for Fairness in Ranking [53.94413894017409]
  We present a novel way of representing permutation distributions, based on the notion of permutation graphs. Similar to PL, our distribution representation, called PPG, can be used for black-box optimization of fairness.
  arXiv  Detail & Related papers  (2022-04-28T20:38:34Z)

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.