paper-with-me

홈 › Papers

OptPO: Optimal Rollout Allocation for Test-time Policy Optimization

2025-12-02 · Youkang Wang, Jian Wang, Rubing Chen, Tianyi Zeng, Xiao-Yong Wei, Qing Li arxiv

Test-time policy optimization enables large language models (LLMs) to adapt to distribution shifts by leveraging feedback from self-generated rollouts. However, existing methods rely on fixed-budget majority voting to estimate rewards, incurring substantial computational redundancy. We propose Optimal Rollout Allocation for Test-time Policy Optimization (OptPO), a principled framework that adaptively allocates inference budgets. By formulating the voting process as a Bayesian sequential probability ratio test, OptPO dynamically halts sampling once the posterior confidence in a consensus answer exceeds a specified threshold. Crucially, it utilizes the retained rollouts for on-policy updates, seamlessly integrating with algorithms like PPO or GRPO without requiring ground-truth labels. Across diverse reasoning benchmarks, OptPO significantly reduces rollout overhead compared to fixed-sample baselines while preserving or improving accuracy. By unifying statistically optimal stopping with test-time learning, OptPO offers a computationally efficient paradigm for test-time adaptation. The source code will be open upon acceptance at https://open-upon-acceptance.

📄 PDF Abstract BibTeX arXiv:2512.02882

Code (0)

등록된 구현이 없습니다.

Tasks

Test-time Adaptation

Similar Papers 제목 키워드 기반

Where to Spend Rollouts: Hit-Utility Optimal Rollout Allocation for Group-Based RLVR

2026-05-08 · Tao Wang, Shuo Li, Yan Sun, Dongsheng Ding 외 arxiv

Reinforcement learning with verifiable rewards (RLVR) has emerged as a central paradigm for improving the reasoning capabilities of large language models. Group-based policy optimization methods, such as GRPO, typically …

Reinforcement LearningMathematical Reasoning

Adaptive Rollout Allocation for Online Reinforcement Learning with Verifiable Rewards

2026-02-02 · Hieu Trung Nguyen, Bao Nguyen, Wenao Ma, Yuzhi Zhao 외 arxiv

Sampling efficiency is a key bottleneck in reinforcement learning with verifiable rewards. Existing group-based policy optimization methods, such as GRPO, allocate a fixed number of rollouts for all training prompts. Thi…

Reinforcement Learning

IsoCompute Playbook: Optimally Scaling Sampling Compute for LLM RL

2026-03-12 · Zhoujun Cheng, Yutao Xie, Yuxiao Qu, Amrith Setlur 외 arxiv

While scaling laws guide compute allocation for LLM pre-training, analogous prescriptions for reinforcement learning (RL) post-training of large language models (LLMs) remain poorly understood. We study the compute-optim…

Reinforcement Learning

A Rollout-Based Algorithm and Reward Function for Efficient Resource Allocation in Business Processes

2025-04-15 · Jeroen Middelhuis, Zaharah Bukhsh, Ivo Adan, Remco Dijkman

Resource allocation plays a critical role in minimizing cycle time and improving the efficiency of business processes. Recently, Deep Reinforcement Learning (DRL) has emerged as a powerful tool to optimize resource alloc…

Deep Reinforcement Learning

Towards Scaling Law Analysis For Spatiotemporal Weather Data

2026-04-06 · Alexander Kiefer, Prasanna Balaprakash, Xiao Wang arxiv

Compute-optimal scaling laws are relatively well studied for NLP and CV, where objectives are typically single-step and targets are comparatively homogeneous. Weather forecasting is harder to characterize in the same fra…

Weather Forecasting