Sample More to Think Less: Group Filtered Policy Optimization for Concise Reasoning
- URL: http://arxiv.org/abs/2508.09726v1
- Date: Wed, 13 Aug 2025 11:43:49 GMT
- Title: Sample More to Think Less: Group Filtered Policy Optimization for Concise Reasoning
- Authors: Vaishnavi Shrivastava, Ahmed Awadallah, Vidhisha Balachandran, Shivam Garg, Harkirat Behl, Dimitris Papailiopoulos,
- Abstract summary: Group Filtered Policy Optimization curbs this length explosion by sampling larger groups per problem during training.<n>GFPO cuts GRPO's length inflation by 46-71% across challenging STEM and coding benchmarks.<n> Adaptive Difficulty GFPO allocates more training resources to harder problems based on real-time difficulty estimates.
- Score: 7.260825775935882
- License: http://creativecommons.org/licenses/by/4.0/
- Abstract: Large language models trained with reinforcement learning with verifiable rewards tend to trade accuracy for length--inflating response lengths to achieve gains in accuracy. While longer answers may be warranted for harder problems, many tokens are merely "filler": repetitive, verbose text that makes no real progress. We introduce GFPO (Group Filtered Policy Optimization), which curbs this length explosion by sampling larger groups per problem during training and filtering responses to train on based on two key metrics: (1) response length and (2) token efficiency: reward per token ratio. By sampling more at training time, we teach models to think less at inference time. On the Phi-4-reasoning model, GFPO cuts GRPO's length inflation by 46-71% across challenging STEM and coding benchmarks (AIME 24/25, GPQA, Omni-MATH, LiveCodeBench) while maintaining accuracy. Optimizing for reward per token further increases reductions in length inflation to 71-85%. We also propose Adaptive Difficulty GFPO, which dynamically allocates more training resources to harder problems based on real-time difficulty estimates, improving the balance between computational efficiency and accuracy especially on difficult questions. GFPO demonstrates that increased training-time compute directly translates to reduced test-time compute--a simple yet effective trade-off for efficient reasoning.
Related papers
- On-Policy Supervised Fine-Tuning for Efficient Reasoning [27.67711115864118]
Large reasoning models (LRMs) are commonly trained with reinforcement learning (RL) to explore long chain-of-thought reasoning.<n>Recent methods add multi-reward objectives to jointly optimize correctness and brevity, but these complex extensions often destabilize training and yield suboptimal trade-offs.<n>We propose a simplified training strategy on-policy SFT, which reduces CoT length by up to 80 while maintaining original accuracy.
arXiv Detail & Related papers (2026-02-13T19:16:39Z) - Correct, Concise and Complete: Multi-stage Training For Adaptive Reasoning [11.179446105672461]
We propose a multi-stage efficient reasoning method that combines supervised fine-tuning and reinforcement learning.<n>Our approach reduces response length by an average of 28% for 8B models and 40% for 32B models.<n>It achieves a superior trade-off compared to more complex state-of-the-art efficient reasoning methods.
arXiv Detail & Related papers (2026-01-06T12:31:51Z) - DLER: Doing Length pEnalty Right - Incentivizing More Intelligence per Token via Reinforcement Learning [134.03095505580276]
Doing Length pEnalty Right (DLER) is a training recipe combining batch-wise reward normalization, higher clipping, dynamic sampling, and a simple truncation length penalty.<n>DLER achieves state-of-the-art accuracy--efficiency trade-offs, cutting output length by over 70 percent while surpassing all previous baseline accuracy.
arXiv Detail & Related papers (2025-10-16T20:05:57Z) - Think Right: Learning to Mitigate Under-Over Thinking via Adaptive, Attentive Compression [68.69801176669843]
We propose an online post-training RL method that prunes redundant steps and estimates difficulty.<n> TRAAC (Think Right with Adaptive, Attentive Compression) achieves an average absolute accuracy gain of 8.4%.<n>Although our models are trained on math datasets, they show accuracy and efficiency gains on out-of-distribution non-math datasets.
arXiv Detail & Related papers (2025-10-02T02:00:20Z) - Thinking Augmented Pre-training [88.04395622064708]
Thinking augmented Pre-Training is a universal methodology that augments text with automatically generated thinking trajectories.<n>This paper introduces a simple and scalable approach to improve the data efficiency of large language model (LLM) training by augmenting existing text data with thinking trajectories.
arXiv Detail & Related papers (2025-09-24T14:45:13Z) - Staying in the Sweet Spot: Responsive Reasoning Evolution via Capability-Adaptive Hint Scaffolding [59.60915947702282]
Reinforcement learning with verifiable rewards (RLVR) has achieved remarkable success in enhancing the reasoning capabilities of large language models (LLMs)<n>Existing RLVR methods often suffer from exploration inefficiency due to mismatches between the training data's difficulty and the model's capability.<n>We propose SEELE, a novel supervision-aided RLVR framework that dynamically adjusts problem difficulty to stay within the high-efficiency region.
arXiv Detail & Related papers (2025-09-08T17:36:21Z) - Train Long, Think Short: Curriculum Learning for Efficient Reasoning [51.506559652495476]
We propose a curriculum learning strategy for length-controlled reasoning.<n>Our method starts with generous token budgets and gradually tightens them over training.<n>Experiments on GSM8K, MATH500, SVAMP, College Math, and GSM+ demonstrate that curriculum-based training consistently outperforms fixed-budget baselines.
arXiv Detail & Related papers (2025-08-12T13:48:03Z) - Just Enough Thinking: Efficient Reasoning with Adaptive Length Penalties Reinforcement Learning [42.82825782517565]
Posttraining DeepScaleR-1.5B with ALP cuts average token usage by 50% without significantly dropping performance.<n>Relative to fixed-budget and uniform penalty baselines, ALP redistributes its reduced budget more intelligently by cutting compute on easy prompts and reallocating saved tokens to difficult ones.
arXiv Detail & Related papers (2025-06-05T17:17:05Z) - Reinforcing Video Reasoning with Focused Thinking [65.85683941058916]
We propose TW-GRPO, a novel framework that enhances visual reasoning with focused thinking and dense reward granularity.<n>Specifically, we employ a token weighting mechanism that prioritizes tokens with high informational density.<n>We also reformulate RL training by shifting from single-choice to multi-choice QA tasks.
arXiv Detail & Related papers (2025-05-30T15:42:19Z) - Making Small Language Models Efficient Reasoners: Intervention, Supervision, Reinforcement [22.801244105119025]
We propose new algorithms to improve token-efficient reasoning with small-scale models by effectively trading off accuracy and computation.<n>We first show that the post-SFT model fails to determine the optimal stopping point of the reasoning process, resulting in verbose and repetitive outputs.<n>Experiments on four reasoning benchmarks, MATH500, AMC, AIME24 and OlympiadBench, demonstrate that TS is highly effective compared to s1's budget forcing approach.
arXiv Detail & Related papers (2025-05-12T18:04:39Z) - Fast-Slow Thinking GRPO for Large Vision-Language Model Reasoning [5.774374541467044]
We present FAST-GRPO, a variant of GRPO that dynamically adapts reasoning depth based on question characteristics.<n>We show that FAST achieves state-of-the-art accuracy with over 10% relative improvement compared to the base model.
arXiv Detail & Related papers (2025-04-25T16:11:23Z) - CPPO: Accelerating the Training of Group Relative Policy Optimization-Based Reasoning Models [68.26281707780761]
This paper introduces Completion Pruning Policy Optimization (CPPO) to accelerate the training of reasoning models.<n>We show that CPPO achieves up to $8.32times$ speedup on GSM8K and $3.51times$ on Math while preserving or even enhancing the accuracy compared to the original GRPO.
arXiv Detail & Related papers (2025-03-28T11:30:05Z) - MC-BERT: Efficient Language Pre-Training via a Meta Controller [96.68140474547602]
Large-scale pre-training is computationally expensive.
ELECTRA, an early attempt to accelerate pre-training, trains a discriminative model that predicts whether each input token was replaced by a generator.
We propose a novel meta-learning framework, MC-BERT, to achieve better efficiency and effectiveness.
arXiv Detail & Related papers (2020-06-10T09:22:19Z)
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.