paper-with-me

홈 › Papers

The Good, The Bad, and The Hybrid: A Reward Structure Showdown in Reasoning Models Training

2025-11-17 · Subramanyam Sahoo arxiv

Reward design is central to reinforcement learning from human feedback (RLHF) and alignment research. In this work, we propose a unified framework to study hard, continuous, and hybrid reward structures for fine-tuning large language models (LLMs) on mathematical reasoning tasks. Using Qwen3-4B with LoRA fine-tuning on the GSM8K dataset, we formalize and empirically evaluate reward formulations that incorporate correctness, perplexity, reasoning quality, and consistency. We introduce an adaptive hybrid reward scheduler that transitions between discrete and continuous signals, balancing exploration and stability. Our results show that hybrid reward structures improve convergence speed and training stability over purely hard or continuous approaches, offering insights for alignment via adaptive reward modeling.

📄 PDF Abstract BibTeX arXiv:2511.13016

Code (0)

등록된 구현이 없습니다.

Tasks

Reinforcement LearningMathematical Reasoning

Similar Papers 제목 키워드 기반

Feel-Good Thompson Sampling for Contextual Bandits: a Markov Chain Monte Carlo Showdown

2025-07-21 · Emile Anand, Sarah Liaw arxiv

Thompson Sampling (TS) is widely used to address the exploration/exploitation tradeoff in contextual bandits, yet recent theory shows that it does not explore aggressively enough in high-dimensional problems. Feel-Good T…

Results of the 2024 Video Browser Showdown

2024-12-13 · Luca Rossetto, Klaus Schoeffmann, Cathal Gurrin, Jakub Lokoč 외

This report presents the results of the 13th Video Browser Showdown, held at the 2024 International Conference on Multimedia Modeling on the 29th of January 2024 in Amsterdam, the Netherlands.

Hybrid Reinforcement: When Reward Is Sparse, It's Better to Be Dense

2025-10-08 · Leitian Tao, Ilia Kulikov, Swarnadeep Saha, Tianlu Wang 외 arxiv

Post-training for reasoning of large language models (LLMs) increasingly relies on verifiable rewards: deterministic checkers that provide 0-1 correctness signals. While reliable, such binary feedback is brittle--many ta…

Reinforcement LearningMathematical Reasoning

How Much Reasoning Do Retrieval-Augmented Models Add beyond LLMs? A Benchmarking Framework for Multi-Hop Inference over Hybrid Knowledge

2026-02-10 · Junhong Lin, Bing Zhang, Song Wang, Ziyan Liu 외 arxiv

Large language models (LLMs) continue to struggle with knowledge-intensive questions that require up-to-date information and multi-hop reasoning. Augmenting LLMs with hybrid external knowledge, such as unstructured text …

Continual PretrainingKnowledge Graphs

Reward Granularity in RLVR: Comparing Process and Outcome Reward Structures for Mathematical Reasoning in Small Language Models

2026-07-03 · Anagha Radhakrishna Palandye, Rebecca Glick, Osheen Kaul arxiv

Reinforcement Learning with Verifiable Rewards (RLVR) has emerged as a promising paradigm for improving mathematical reasoning in language models. Yet most RLVR work rewards only the final answer (outcome-based rewards),…

Reinforcement LearningMathematical Reasoning