S-GRPO: Early Exit via Reinforcement Learning in Reasoning Models

• URL: http://arxiv.org/abs/2505.07686v2
• Date: Sat, 17 May 2025 04:01:57 GMT
• Title: S-GRPO: Early Exit via Reinforcement Learning in Reasoning Models
• Authors: Muzhi Dai, Chenxu Yang, Qingyi Si,
• Abstract summary: Test-Time Scaling emerges as an active research focus in the large language model community.<n>Recent studies reveal that reasoning models (even Qwen3) consistently exhibit excessive thought redundancy.<n>This paper introduces Serial-Group Decaying-Reward Policy Optimization (S-GRPO), a novel reinforcement learning paradigm.
• Score: 2.9925837108958864
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: As Test-Time Scaling emerges as an active research focus in the large language model community, advanced post-training methods increasingly emphasize extending chain-of-thought (CoT) generation length, thereby enhancing reasoning capabilities to approach Deepseek R1-like reasoning models. However, recent studies reveal that reasoning models (even Qwen3) consistently exhibit excessive thought redundancy in CoT generation. This overthinking issue arises from the inherent limitations of conventional outcome-reward reinforcement learning, which systematically overlooks the regulation of intermediate reasoning processes. This paper introduces Serial-Group Decaying-Reward Policy Optimization (S-GRPO), a novel reinforcement learning paradigm that enables models to implicitly evaluate the sufficiency of intermediate reasoning steps, thereby facilitating early exit in CoT generation. Unlike GRPO, which samples multiple possible reasoning paths in parallel (parallel group), S-GRPO only samples one reasoning path and serially selects multiple temporal positions from the path to exit thinking and directly generate answers (serial group). For correct answers within a serial group, rewards gradually decrease based on the exit positions along the reasoning path from front to back. This design encourages the model to produce more accurate and concise thoughts, while also incentivizing early thinking termination when appropriate. Empirical evaluations demonstrate that S-GRPO is compatible with state-of-the-art reasoning models, including Qwen3 and Deepseek-distill. Across diverse benchmarks such as GSM8K, AIME 2024, AMC 2023, MATH-500, and GPQA Diamond, S-GRPO achieves a substantial reduction in sequence length (35.4% - 61.1%) while simultaneously improving accuracy (absolute 0.72% - 6.08%).

Related papers

• Accelerating LLM Reasoning via Early Rejection with Partial Reward Modeling [12.835376812101323]
  We introduce the hypothesis that PRMs are also Partial Reward Models.<n>This allows for principled early rejection based on intermediate token-level signals.<n>On math reasoning benchmarks, our method achieves up to 1.4$times$-9$times$ reduction in inference FLOPs without degrading final performance.
  arXiv  Detail & Related papers  (2025-08-04T00:58:56Z)
• Lost at the Beginning of Reasoning [82.18834329384514]
  We show that the first reasoning step exerts a disproportionately large influence on the final prediction.<n>We propose an efficient sampling strategy that leverages a reward model to identify and retain high-quality first reasoning steps.<n>We introduce a new benchmark specifically constructed with deliberately flawed first reasoning steps to systematically evaluate model self-correction capabilities.
  arXiv  Detail & Related papers  (2025-06-27T09:53:57Z)
• ConciseHint: Boosting Efficient Reasoning via Continuous Concise Hints during Generation [53.149817480019834]
  Recent advancements in large reasoning models (LRMs) have achieved notable performance enhancements on complex reasoning tasks by scaling up the generation length by Chain-of-Thought (CoT)<n>We propose a framework dubbed ConciseHint, which continuously encourages the reasoning model to speak concisely by injecting the textual hint during the token generation of the reasoning process.<n>Experiments on the state-of-the-art LRMs, including DeepSeek-R1 and Qwen-3 series, demonstrate that our method can effectively produce concise reasoning processes while maintaining performance well.
  arXiv  Detail & Related papers  (2025-06-23T16:20:44Z)
• GRPO-CARE: Consistency-Aware Reinforcement Learning for Multimodal Reasoning [53.894789613838654]
  We introduce SEED-Bench-R1, a benchmark with complex real-world videos requiring balanced perception and reasoning.<n>Using SEED-Bench-R1, we find that standard GRPO, while improving answer accuracy, often reduces logical coherence between reasoning steps and answers, with only a 57.9% consistency rate.<n>We propose GRPO-CARE, a consistency-aware RL framework optimizing both answer correctness and reasoning coherence without explicit supervision.
  arXiv  Detail & Related papers  (2025-06-19T08:49:13Z)
• Walk Before You Run! Concise LLM Reasoning via Reinforcement Learning [10.255235456427037]
  We propose a simple yet effective two-stage reinforcement learning framework for achieving concise reasoning in Large Language Models (LLMs)<n>The first stage, using more training steps, aims to incentivize the model's reasoning capabilities via Group Relative Policy Optimization.<n>The second stage, using fewer training steps, explicitly enforces conciseness and improves efficiency via Length-aware Group Relative Policy Optimization.
  arXiv  Detail & Related papers  (2025-05-27T13:29:51Z)
• Interleaved Reasoning for Large Language Models via Reinforcement Learning [22.403928213802036]
  Long chain-of-thought (CoT) enhances large language models' (LLM) reasoning capabilities.<n>We propose a novel training paradigm that uses reinforcement learning (RL) to guide reasoning LLMs to interleave thinking and answering for multi-hop questions.
  arXiv  Detail & Related papers  (2025-05-26T07:58:17Z)
• Let LLMs Break Free from Overthinking via Self-Braking Tuning [60.08396797526657]
  Large reasoning models (LRMs) have significantly enhanced their reasoning capabilities by generating longer chains of thought.<n>This performance gain comes at the cost of a substantial increase in redundant reasoning during the generation process.<n>We propose a novel framework, Self-Braking Tuning (SBT), which tackles overthinking from the perspective of allowing the model to regulate its own reasoning process.
  arXiv  Detail & Related papers  (2025-05-20T16:53:40Z)
• Dynamic Early Exit in Reasoning Models [14.508648537186989]
  Overthinking in long chain-of-thought (CoT) generation slows down the efficiency of problem solving.<n>We propose a method that allows LLMs to self-truncate CoT sequences by early exit during generation.<n>Our method requires no additional training and can be seamlessly integrated into existing o1-like reasoning LLMs.
  arXiv  Detail & Related papers  (2025-04-22T13:36:53Z)
• Think Deep, Think Fast: Investigating Efficiency of Verifier-free Inference-time-scaling Methods [39.89239733570008]
  This work conducts a comprehensive analysis of inference-time scaling methods for both reasoning and non-reasoning models.<n>We find that non-reasoning models, even with an extremely high inference budget, still fall substantially behind reasoning models.<n>For reasoning models, majority voting proves to be a robust inference strategy, generally competitive or outperforming other more sophisticated ITC methods.
  arXiv  Detail & Related papers  (2025-04-18T19:32:55Z)
• A Minimalist Approach to LLM Reasoning: from Rejection Sampling to Reinforce [68.99924691391048]
  We revisit GRPO from a reinforce-like algorithm perspective and analyze its core components.<n>We find that a simple rejection sampling baseline, RAFT, yields competitive performance than GRPO and PPO.<n>Motivated by this insight, we propose Reinforce-Rej, a minimal extension of policy gradient that filters both entirely incorrect and entirely correct samples.
  arXiv  Detail & Related papers  (2025-04-15T16:15:02Z)
• Revisiting the Test-Time Scaling of o1-like Models: Do they Truly Possess Test-Time Scaling Capabilities? [61.85289698610747]
  We study whether o1-like large language models (LLMs) truly possess test-time scaling capabilities.<n>We find that longer CoTs of these o1-like models do not consistently enhance accuracy.<n>We propose Shortest Majority Vote, a method that combines parallel scaling strategies with CoT length characteristics.
  arXiv  Detail & Related papers  (2025-02-17T07:21:11Z)
• Metastable Dynamics of Chain-of-Thought Reasoning: Provable Benefits of Search, RL and Distillation [40.861314212279474]
  We study inference-time compute by viewing chain-of-thought (CoT) generation as a metastable Markov process.<n>We prove that implementing a search protocol that rewards sparse edges improves CoT by decreasing the expected number of steps to reach different clusters.<n>We also show that the information gained by search can be utilized to obtain a better reasoning model.
  arXiv  Detail & Related papers  (2025-02-02T18:19:14Z)
• Chain-of-Retrieval Augmented Generation [72.06205327186069]
  This paper introduces an approach for training o1-like RAG models that retrieve and reason over relevant information step by step before generating the final answer.<n>Our proposed method, CoRAG, allows the model to dynamically reformulate the query based on the evolving state.
  arXiv  Detail & Related papers  (2025-01-24T09:12:52Z)
• Posterior Coreset Construction with Kernelized Stein Discrepancy for Model-Based Reinforcement Learning [78.30395044401321]
  We develop a novel model-based approach to reinforcement learning (MBRL) It relaxes the assumptions on the target transition model to belong to a generic family of mixture models. It can achieve up-to 50 percent reduction in wall clock time in some continuous control environments.
  arXiv  Detail & Related papers  (2022-06-02T17:27:49Z)
• Nested-Wasserstein Self-Imitation Learning for Sequence Generation [158.19606942252284]
  We propose the concept of nested-Wasserstein distance for distributional semantic matching. A novel nested-Wasserstein self-imitation learning framework is developed, encouraging the model to exploit historical high-rewarded sequences.
  arXiv  Detail & Related papers  (2020-01-20T02:19:13Z)

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.