Controllable Exploration in Hybrid-Policy RLVR for Multi-Modal Reasoning

• URL: http://arxiv.org/abs/2602.20197v1
• Date: Sun, 22 Feb 2026 07:23:36 GMT
• Title: Controllable Exploration in Hybrid-Policy RLVR for Multi-Modal Reasoning
• Authors: Zhuoxu Huang, Mengxi Jia, Hao Sun, Xuelong Li, Jungong Han,
• Abstract summary: CalibRL is a hybrid-policy RLVR framework that supports controllable exploration with expert guidance.<n>CalibRL increases policy entropy in a guided manner and clarifies the target distribution.<n>Experiments across eight benchmarks, including both in-domain and out-of-domain settings, demonstrate consistent improvements.
• Score: 88.42566960813438
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Reinforcement Learning with verifiable rewards (RLVR) has emerged as a primary learning paradigm for enhancing the reasoning capabilities of multi-modal large language models (MLLMs). However, during RL training, the enormous state space of MLLM and sparse rewards often leads to entropy collapse, policy degradation, or over-exploitation of suboptimal behaviors. This necessitates an exploration strategy that maintains productive stochasticity while avoiding the drawbacks of uncontrolled random sampling, yielding inefficient exploration. In this paper, we propose CalibRL, a hybrid-policy RLVR framework that supports controllable exploration with expert guidance, enabled by two key mechanisms. First, a distribution-aware advantage weighting scales updates by group rareness to calibrate the distribution, therefore preserving exploration. Meanwhile, the asymmetric activation function (LeakyReLU) leverages the expert knowledge as a calibration baseline to moderate overconfident updates while preserving their corrective direction. CalibRL increases policy entropy in a guided manner and clarifies the target distribution by estimating the on-policy distribution through online sampling. Updates are driven by these informative behaviors, avoiding convergence to erroneous patterns. Importantly, these designs help alleviate the distributional mismatch between the model's policy and expert trajectories, thereby achieving a more stable balance between exploration and exploitation. Extensive experiments across eight benchmarks, including both in-domain and out-of-domain settings, demonstrate consistent improvements, validating the effectiveness of our controllable hybrid-policy RLVR training. Code is available at https://github.com/zhh6425/CalibRL.

Related papers

• Learning to Explore with Parameter-Space Noise: A Deep Dive into Parameter-Space Noise for Reinforcement Learning with Verifiable Rewards [16.22162269278471]
  PSN-RLVR perturbs policy parameters before rollout generation to induce temporally consistent, trajectory-level exploration.<n>We propose a computationally efficient real-time adaptive noise scheduler driven by a lightweight surrogate that combines semantic diversity with normalized self-certainty.
  arXiv  Detail & Related papers  (2026-01-30T13:10:30Z)
• When Sharpening Becomes Collapse: Sampling Bias and Semantic Coupling in RL with Verifiable Rewards [20.896576101848655]
  We study whetherReinforcement Learning with Verifiable Rewards elicits novel capabilities or merely sharpens the distribution over existing knowledge.<n>We propose inverse-success advantage calibration to prioritize difficult queries and distribution-level calibration to diversify sampling via a memory network.
  arXiv  Detail & Related papers  (2026-01-22T03:15:57Z)
• Steering Vision-Language-Action Models as Anti-Exploration: A Test-Time Scaling Approach [78.4812458793128]
  We propose textbfTACO, a test-time-scaling framework that applies a lightweight pseudo-count estimator as a high-fidelity verifier of action chunks.<n>Our method resembles the classical anti-exploration principle in offline reinforcement learning (RL), and being gradient-free, it incurs significant computational benefits.
  arXiv  Detail & Related papers  (2025-12-02T14:42:54Z)
• The Choice of Divergence: A Neglected Key to Mitigating Diversity Collapse in Reinforcement Learning with Verifiable Reward [57.56453588632619]
  A central paradox in fine-tuning Large Language Models (LLMs) with Reinforcement Learning with Verifiable Reward (RLVR) is the frequent degradation of multi-attempt performance.<n>This is often accompanied by catastrophic forgetting, where models lose previously acquired skills.<n>We argue that standard RLVR objectives lack a crucial mechanism for knowledge retention.
  arXiv  Detail & Related papers  (2025-09-09T06:34:32Z)
• RL-PLUS: Countering Capability Boundary Collapse of LLMs in Reinforcement Learning with Hybrid-policy Optimization [111.1749164063616]
  We propose RL-PLUS, a novel hybrid-policy optimization approach for Large Language Models (LLMs)<n> RL-PLUS synergizes internal exploitation with external data to achieve stronger reasoning capabilities and surpass the boundaries of base models.<n>We provide both theoretical analysis and extensive experiments to demonstrate the superiority and generalizability of our approach.
  arXiv  Detail & Related papers  (2025-07-31T23:55:29Z)
• GenPO: Generative Diffusion Models Meet On-Policy Reinforcement Learning [34.25769740497309]
  GenPO is a generative policy optimization framework that leverages exact diffusion inversion to construct invertible action mappings.<n>GenPO is the first method to successfully integrate diffusion policies into on-policy RL, unlocking their potential for large-scale parallelized training and real-world robotic deployment.
  arXiv  Detail & Related papers  (2025-05-24T15:57:07Z)
• Diffusion-based Reinforcement Learning via Q-weighted Variational Policy Optimization [55.97310586039358]
  Diffusion models have garnered widespread attention in Reinforcement Learning (RL) for their powerful expressiveness and multimodality.<n>We propose a novel model-free diffusion-based online RL algorithm, Q-weighted Variational Policy Optimization (QVPO)<n>Specifically, we introduce the Q-weighted variational loss, which can be proved to be a tight lower bound of the policy objective in online RL under certain conditions.<n>We also develop an efficient behavior policy to enhance sample efficiency by reducing the variance of the diffusion policy during online interactions.
  arXiv  Detail & Related papers  (2024-05-25T10:45:46Z)
• One-Step Distributional Reinforcement Learning [10.64435582017292]
  We present the simpler one-step distributional reinforcement learning (OS-DistrRL) framework. We show that our approach comes with a unified theory for both policy evaluation and control. We propose two OS-DistrRL algorithms for which we provide an almost sure convergence analysis.
  arXiv  Detail & Related papers  (2023-04-27T06:57:00Z)
• Reinforcement Learning from Diverse Human Preferences [68.4294547285359]
  This paper develops a method for crowd-sourcing preference labels and learning from diverse human preferences. The proposed method is tested on a variety of tasks in DMcontrol and Meta-world. It has shown consistent and significant improvements over existing preference-based RL algorithms when learning from diverse feedback.
  arXiv  Detail & Related papers  (2023-01-27T15:18:54Z)
• Normality-Guided Distributional Reinforcement Learning for Continuous Control [13.818149654692863]
  Learning a predictive model of the mean return, or value function, plays a critical role in many reinforcement learning algorithms.<n>We study the value distribution in several continuous control tasks and find that the learned value distribution is empirically quite close to normal.<n>We propose a policy update strategy based on the correctness as measured by structural characteristics of the value distribution not present in the standard value function.
  arXiv  Detail & Related papers  (2022-08-28T02:52:10Z)

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.