paper-with-me

홈 › Papers

Small Generalizable Prompt Predictive Models Can Steer Efficient RL Post-Training of Large Reasoning Models

2026-02-02 · Yun Qu, Qi Wang, Yixiu Mao, Heming Zou, Yuhang Jiang, Weijie Liu, Clive Bai, Kai Yang, Yangkun Chen, Saiyong Yang, Xiangyang Ji arxiv

Reinforcement learning enhances the reasoning capabilities of large language models but often involves high computational costs due to rollout-intensive optimization. Online prompt selection presents a plausible solution by prioritizing informative prompts to improve training efficiency. However, current methods either depend on costly, exact evaluations or construct prompt-specific predictive models lacking generalization across prompts. This study introduces Generalizable Predictive Prompt Selection (GPS), which performs Bayesian inference towards prompt difficulty using a lightweight generative model trained on the shared optimization history. Intermediate-difficulty prioritization and history-anchored diversity are incorporated into the batch acquisition principle to select informative prompt batches. The small predictive model also generalizes at test-time for efficient computational allocation. Experiments across varied reasoning benchmarks indicate GPS's substantial improvements in training efficiency, final performance, and test-time efficiency over superior baseline methods.

📄 PDF Abstract BibTeX arXiv:2602.01970

Code (0)

등록된 구현이 없습니다.

Tasks

Reinforcement LearningBayesian Inference

Similar Papers 제목 키워드 기반

Learning to Steer: Input-dependent Steering for Multimodal LLMs

2025-08-18 · Jayneel Parekh, Pegah Khayatan, Mustafa Shukor, Arnaud Dapogny 외 arxiv

Steering has emerged as a practical approach to enable post-hoc guidance of LLMs towards enforcing a specific behavior. However, it remains largely underexplored for multimodal LLMs (MLLMs); furthermore, existing steerin…

Identifying and Transferring Reasoning-Critical Neurons: Improving LLM Inference Reliability via Activation Steering

2026-01-27 · Fangan Dong, Zuming Yan, Xuri Ge, Zhiwei Xu 외 arxiv

Despite the strong reasoning capabilities of recent large language models (LLMs), achieving reliable performance on challenging tasks often requires post-training or computationally expensive sampling strategies, limitin…

Painless Activation Steering: An Automated, Lightweight Approach for Post-Training Large Language Models

2025-09-25 · Sasha Cui, Zhongren Chen arxiv

Language models (LMs) are typically post-trained for desired capabilities and behaviors via weight-based or prompt-based steering, but the former is time-consuming and expensive, and the latter is not precisely controlla…

Reinforcement Learning

Breaking Lock-In: Preserving Steerability under Low-Data VLA Post-Training

2026-04-25 · Suning Huang, Jiaqi Shao, Ke Wang, Qianzhong Chen 외 arxiv

Have you ever post-trained a generalist vision-language-action (VLA) policy on a small demonstration dataset, only to find that it stops responding to new instructions and is limited to behaviors observed during post-tra…

Visual Grounding

When is Your LLM Steerable?

2026-06-10 · Chenrui Fan, Yize Cheng, Ming Li, Soheil Feizi 외 arxiv

Activation steering offers a lightweight approach to control language models' behavior at inference time, but whether it succeeds or fails heavily depends on the prompt, concept, model, and steering configuration. Findin…