More Than One Teacher: Adaptive Multi-Guidance Policy Optimization for Diverse Exploration
- URL: http://arxiv.org/abs/2510.02227v2
- Date: Thu, 09 Oct 2025 17:18:49 GMT
- Title: More Than One Teacher: Adaptive Multi-Guidance Policy Optimization for Diverse Exploration
- Authors: Xiaoyang Yuan, Yujuan Ding, Yi Bin, Wenqi Shao, Jinyu Cai, Jingkuan Song, Yang Yang, Heng Tao Shen,
- Abstract summary: "guidance-on-demand" approach expands exploration while preserving the value of self-discovery.<n>Experiments show AMPO substantially outperforms a strong baseline.<n>Using four peer-sized teachers, our method achieves comparable results to approaches that leverage a single, more powerful teacher.
- Score: 103.1589018460702
- License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
- Abstract: Reinforcement Learning with Verifiable Rewards (RLVR) is a promising paradigm for enhancing the reasoning ability in Large Language Models (LLMs). However, prevailing methods primarily rely on self-exploration or a single off-policy teacher to elicit long chain-of-thought (LongCoT) reasoning, which may introduce intrinsic model biases and restrict exploration, ultimately limiting reasoning diversity and performance. Drawing inspiration from multi-teacher strategies in knowledge distillation, we introduce Adaptive Multi-Guidance Policy Optimization (AMPO), a novel framework that adaptively leverages guidance from multiple proficient teacher models, but only when the on-policy model fails to generate correct solutions. This "guidance-on-demand" approach expands exploration while preserving the value of self-discovery. Moreover, AMPO incorporates a comprehension-based selection mechanism, prompting the student to learn from the reasoning paths that it is most likely to comprehend, thus balancing broad exploration with effective exploitation. Extensive experiments show AMPO substantially outperforms a strong baseline (GRPO), with a 4.3% improvement on mathematical reasoning tasks and 12.2% on out-of-distribution tasks, while significantly boosting Pass@k performance and enabling more diverse exploration. Notably, using four peer-sized teachers, our method achieves comparable results to approaches that leverage a single, more powerful teacher (e.g., DeepSeek-R1) with more data. These results demonstrate a more efficient and scalable path to superior reasoning and generalizability. Our code is available at https://github.com/SII-Enigma/AMPO.
Related papers
- Learning beyond Teacher: Generalized On-Policy Distillation with Reward Extrapolation [57.524909883706556]
On-policy distillation (OPD) has demonstrated strong empirical gains in improving student performance.<n>This work introduces a flexible reference model and a reward scaling factor that controls the relative weight of the reward term against the KL regularization.<n>In particular, in the setting where we merge the knowledge from different domain experts, obtained by applying domain-specific RL to the same student model, ExOPD enables the student to even surpass the teacher's performance boundary.
arXiv Detail & Related papers (2026-02-12T16:14:29Z) - HINT: Helping Ineffective Rollouts Navigate Towards Effectiveness [49.72591739116668]
Reinforcement Learning (RL) has become a key driver for enhancing the long chain-of-thought (CoT) reasoning capabilities of Large Language Models (LLMs)<n>However, prevalent methods like GRPO often fail when task difficulty exceeds the model's capacity, leading to reward sparsity and inefficient training.<n>We propose HINT: Helping Ineffective rollouts Navigate Towards effectiveness, an adaptive hinting framework.
arXiv Detail & Related papers (2025-10-10T13:42:03Z) - Selective Expert Guidance for Effective and Diverse Exploration in Reinforcement Learning of LLMs [49.72591739116668]
Reinforcement Learning with Verifiable Rewards (RLVR) has become a widely adopted technique for enhancing the reasoning ability of Large Language Models (LLMs)<n>Existing methods address this issue by imitating expert trajectories, which improve effectiveness but neglect diversity.<n>We propose MENTOR: Mixed-policy Expert Navigation for Token-level Optimization of Reasoning.
arXiv Detail & Related papers (2025-10-05T10:38:55Z) - Perception-Aware Policy Optimization for Multimodal Reasoning [79.56070395437898]
A major source of error in current multimodal reasoning lies in the perception of visual inputs.<n>We propose PAPO, a novel policy gradient algorithm that encourages the model to learn to perceive while learning to reason.<n>We observe a substantial reduction of 30.5% in perception errors, indicating improved perceptual capabilities with PAPO.
arXiv Detail & Related papers (2025-07-08T23:22:34Z) - Exploring and Exploiting the Inherent Efficiency within Large Reasoning Models for Self-Guided Efficiency Enhancement [101.77467538102924]
Large reasoning models (LRMs) exhibit overthinking, which hinders efficiency and inflates inference cost.<n>We propose two lightweight methods to enhance LRM efficiency.<n>First, we introduce Efficiency Steering, a training-free activation steering technique that modulates reasoning behavior via a single direction.<n>Second, we develop Self-Rewarded Efficiency RL, a reinforcement learning framework that dynamically balances task accuracy and brevity.
arXiv Detail & Related papers (2025-06-18T17:18:12Z) - Consistent Paths Lead to Truth: Self-Rewarding Reinforcement Learning for LLM Reasoning [87.7836502955847]
We propose a novel self-rewarding reinforcement learning framework to enhance Large Language Model (LLM) reasoning.<n>Our key insight is that correct responses often exhibit consistent trajectory patterns in terms of model likelihood.<n>We introduce CoVo, an intrinsic reward mechanism that integrates Consistency and Volatility via a robust vector-space aggregation strategy.
arXiv Detail & Related papers (2025-06-10T12:40:39Z) - Diversity-Aware Policy Optimization for Large Language Model Reasoning [30.460540027658173]
We investigate the impact of diversity in RL-based training for large language models.<n>We propose a novel diversity-aware policy optimization method.<n>Our method achieves a 3.5 percent average improvement across four mathematical reasoning benchmarks.
arXiv Detail & Related papers (2025-05-29T13:27:44Z) - 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) - Offline Reinforcement Learning for LLM Multi-Step Reasoning [15.687002884103537]
OREO (Offline Reasoning Optimization) is an offline reinforcement learning method for enhancing multi-step reasoning.<n>It reduces the need to collect pairwise data and enables better credit assignment.<n>It surpasses existing offline learning methods on multi-step reasoning benchmarks.
arXiv Detail & Related papers (2024-12-20T18:49:45Z) - Scaling of Search and Learning: A Roadmap to Reproduce o1 from Reinforcement Learning Perspective [77.94874338927492]
OpenAI has claimed that the main techinique behinds o1 is the reinforcement learning.<n>This paper analyzes the roadmap to achieving o1 from the perspective of reinforcement learning.
arXiv Detail & Related papers (2024-12-18T18:24:47Z) - Ready Policy One: World Building Through Active Learning [35.358315617358976]
We introduce Ready Policy One (RP1), a framework that views Model-Based Reinforcement Learning as an active learning problem.
RP1 achieves this by utilizing a hybrid objective function, which crucially adapts during optimization.
We rigorously evaluate our method on a variety of continuous control tasks, and demonstrate statistically significant gains over existing approaches.
arXiv Detail & Related papers (2020-02-07T09:57:53Z)
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.