Not All Negative Samples Are Equal: LLMs Learn Better from Plausible Reasoning

• URL: http://arxiv.org/abs/2602.03516v2
• Date: Wed, 04 Feb 2026 04:25:31 GMT
• Title: Not All Negative Samples Are Equal: LLMs Learn Better from Plausible Reasoning
• Authors: Zixiang Di, Jinyi Han, Shuo Zhang, Ying Liao, Zhi Li, Xiaofeng Ji, Yongqi Wang, Zheming Yang, Ming Gao, Bingdong Li, Jie Wang,
• Abstract summary: Plausible Negative Samples (PNS) is a method that synthesizes high-quality negative samples exhibiting expected format and structural coherence.<n>PNS trains a dedicated model via reverse reinforcement learning guided by a composite reward combining format compliance, accuracy inversion, and chain-of-thought evaluation.<n>PNS consistently outperforms other negative sample methods, achieving an average improvement of 2.03% over RL-trained models.
• Score: 23.352305094623123
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Learning from negative samples holds great promise for improving Large Language Model (LLM) reasoning capability, yet existing methods treat all incorrect responses as equally informative, overlooking the crucial role of sample quality. To address this, we propose Plausible Negative Samples (PNS), a method that synthesizes high-quality negative samples exhibiting expected format and structural coherence while ultimately yielding incorrect answers. PNS trains a dedicated model via reverse reinforcement learning (RL) guided by a composite reward combining format compliance, accuracy inversion, reward model assessment, and chain-of-thought evaluation, generating responses nearly indistinguishable from correct solutions. We further validate PNS as a plug-and-play data source for preference optimization across three backbone models on seven mathematical reasoning benchmarks. Results demonstrate that PNS consistently outperforms other negative sample synthesis methods, achieving an average improvement of 2.03% over RL-trained models.

Related papers

• Dissecting Long-Chain-of-Thought Reasoning Models: An Empirical Study [91.78803511141975]
  This work focuses on the roles of positive and negative samples in scaling reinforcement learning.<n>We identify substantial data inefficiency in group relative policy optimization, where over half of the samples yield zero advantage.<n>We investigate unstable performance across various reasoning models and benchmarks, attributing instability to uncertain problems with ambiguous outcomes.
  arXiv  Detail & Related papers  (2025-06-05T11:47:10Z)
• The Surprising Effectiveness of Negative Reinforcement in LLM Reasoning [37.13807960501503]
  Reinforcement learning with verifiable rewards (RLVR) is a promising approach for training language models (LMs)<n>We decompose the learning signal into reinforcing correct responses and penalizing incorrect ones, referred to as Positive and Negative Sample Reinforcement (PSR and NSR)<n>We show that NSR works by suppressing incorrect generations and redistributing probability mass toward other plausible candidates, guided by the model's prior beliefs.
  arXiv  Detail & Related papers  (2025-06-02T06:10:54Z)
• Unearthing Gems from Stones: Policy Optimization with Negative Sample Augmentation for LLM Reasoning [41.83677588934301]
  We propose Behavior Constrained Policy Gradient with Negative Sample Augmentation (BCPG-NSA)<n>BCPG-NSA is a fine-grained offline framework that encompasses three stages: 1) sample segmentation, 2) consensus-based step correctness assessment combining LLM and PRM judgers, and 3) policy optimization with NSA designed to effectively mine positive steps within negative samples.<n> Experimental results show that BCPG-NSA outperforms baselines on several challenging math/coding reasoning benchmarks using the same training dataset.
  arXiv  Detail & Related papers  (2025-05-20T14:16:49Z)
• RL on Incorrect Synthetic Data Scales the Efficiency of LLM Math Reasoning by Eight-Fold [41.28168368547099]
  Training on model-generated synthetic data is a promising approach for finetuning LLMs, but it remains unclear when it helps or hurts. We show that training on per-step negatives can help to unlearn spurious correlations in the positive data.
  arXiv  Detail & Related papers  (2024-06-20T17:45:54Z)
• Re-ReST: Reflection-Reinforced Self-Training for Language Agents [101.22559705696885]
  Self-training in language agents can generate supervision from the agent itself.<n>We present Reflection-Reinforced Self-Training (Re-ReST), which uses a textitreflector to refine low-quality generated samples.
  arXiv  Detail & Related papers  (2024-06-03T16:21:38Z)
• Secrets of RLHF in Large Language Models Part II: Reward Modeling [134.97964938009588]
  We introduce a series of novel methods to mitigate the influence of incorrect and ambiguous preferences in the dataset. We also introduce contrastive learning to enhance the ability of reward models to distinguish between chosen and rejected responses.
  arXiv  Detail & Related papers  (2024-01-11T17:56:59Z)
• Beyond Human Data: Scaling Self-Training for Problem-Solving with Language Models [115.501751261878]
  Fine-tuning language models(LMs) on human-generated data remains a prevalent practice. We investigate whether we can go beyond human data on tasks where we have access to scalar feedback. We find that ReST$EM$ scales favorably with model size and significantly surpasses fine-tuning only on human data.
  arXiv  Detail & Related papers  (2023-12-11T18:17:43Z)
• Towards Robust Visual Question Answering: Making the Most of Biased Samples via Contrastive Learning [54.61762276179205]
  We propose a novel contrastive learning approach, MMBS, for building robust VQA models by Making the Most of Biased Samples. Specifically, we construct positive samples for contrastive learning by eliminating the information related to spurious correlation from the original training samples. We validate our contributions by achieving competitive performance on the OOD dataset VQA-CP v2 while preserving robust performance on the ID dataset VQA v2.
  arXiv  Detail & Related papers  (2022-10-10T11:05:21Z)
• Rethinking InfoNCE: How Many Negative Samples Do You Need? [54.146208195806636]
  We study how many negative samples are optimal for InfoNCE in different scenarios via a semi-quantitative theoretical framework. We estimate the optimal negative sampling ratio using the $K$ value that maximizes the training effectiveness function.
  arXiv  Detail & Related papers  (2021-05-27T08:38:29Z)
• Positive-Congruent Training: Towards Regression-Free Model Updates [87.25247195148187]
  In image classification, sample-wise inconsistencies appear as "negative flips" A new model incorrectly predicts the output for a test sample that was correctly classified by the old (reference) model. We propose a simple approach for PC training, Focal Distillation, which enforces congruence with the reference model.
  arXiv  Detail & Related papers  (2020-11-18T09:00:44Z)

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.