Provably Efficient Online RLHF with One-Pass Reward Modeling

• URL: http://arxiv.org/abs/2502.07193v2
• Date: Thu, 19 Jun 2025 06:54:01 GMT
• Title: Provably Efficient Online RLHF with One-Pass Reward Modeling
• Authors: Long-Fei Li, Yu-Yang Qian, Peng Zhao, Zhi-Hua Zhou,
• Abstract summary: We propose a one-pass reward modeling method that does not require storing the historical data and can be computed in constant time.<n>We provide theoretical guarantees showing that our method improves both statistical and computational efficiency.<n>We conduct experiments using Llama-3-8B-Instruct and Qwen2.5-7B-Instruct models on the Ultrafeedback-binarized and Mixture2 datasets.
• Score: 59.30310692855397
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Reinforcement Learning from Human Feedback (RLHF) has shown remarkable success in aligning Large Language Models (LLMs) with human preferences. Traditional RLHF approaches rely on a fixed dataset, which often suffers from limited coverage. To this end, online RLHF has emerged as a promising direction, enabling iterative data collection and model improvement. Despite its potential, this paradigm faces a key bottleneck: the requirement to continuously integrate new data into the historical dataset and re-optimize the model from scratch at each iteration, resulting in computational and storage costs that grow linearly with the number of iterations. In this work, we address this challenge by proposing a one-pass reward modeling method that does not require storing the historical data and can be computed in constant time. Specifically, we first formalize RLHF as a contextual preference bandit problem and design an online mirror descent algorithm with a tailored local norm to replace the standard maximum likelihood estimation for reward modeling. We then apply our method to various online RLHF settings, including passive data collection, active data collection, and deployment-time adaptation. We provide theoretical guarantees showing that our method improves both statistical and computational efficiency. Finally, we provide practical algorithms and conduct experiments using Llama-3-8B-Instruct and Qwen2.5-7B-Instruct models on the Ultrafeedback-binarized and Mixture2 datasets, validating the effectiveness of our proposed method.

Related papers

• Generalized Linear Bandits: Almost Optimal Regret with One-Pass Update [60.414548453838506]
  We study the generalized linear bandit (GLB) problem, a contextual multi-armed bandit framework that extends the classical linear model by incorporating a non-linear link function.<n>GLBs are widely applicable to real-world scenarios, but their non-linear nature introduces significant challenges in achieving both computational and statistical efficiency.<n>We propose a jointly efficient algorithm that attains a nearly optimal regret bound with $mathcalO(1)$ time and space complexities per round.
  arXiv  Detail & Related papers  (2025-07-16T02:24:21Z)
• Ring-lite: Scalable Reasoning via C3PO-Stabilized Reinforcement Learning for LLMs [51.21041884010009]
  Ring-lite is a Mixture-of-Experts (MoE)-based large language model optimized via reinforcement learning (RL)<n>Our approach matches the performance of state-of-the-art (SOTA) small-scale reasoning models on challenging benchmarks.
  arXiv  Detail & Related papers  (2025-06-17T17:12:34Z)
• Exploring Data Scaling Trends and Effects in Reinforcement Learning from Human Feedback [12.7099489697479]
  We introduce a hybrid reward system combining reasoning task verifiers (RTV) and a generative reward model (GenRM) to mitigate reward hacking. We also propose a novel prompt-selection method, Pre-PPO, to maintain response diversity and enhance learning effectiveness.
  arXiv  Detail & Related papers  (2025-03-28T08:26:41Z)
• Distributionally Robust Reinforcement Learning with Human Feedback [13.509499718691016]
  We introduce a distributionally robust RLHF for fine-tuning large language models. Our goal is to ensure that a fine-tuned model retains its performance even when the distribution of prompts significantly differs. We show that our robust training improves the accuracy of the learned reward models on average, and markedly on some tasks, such as reasoning.
  arXiv  Detail & Related papers  (2025-03-01T15:43:39Z)
• Avoiding $\mathbf{exp(R_{max})}$ scaling in RLHF through Preference-based Exploration [20.76451379043945]
  Reinforcement Learning from Human Feedback (RLHF) has emerged as a pivotal technique for large language model (LLM) alignment.<n>This paper studies the setting of online RLHF and focus on improving sample efficiency.
  arXiv  Detail & Related papers  (2025-02-02T04:40:04Z)
• Active RLHF via Best Policy Learning from Trajectory Preference Feedback [15.799929216215672]
  We address the problem of best policy identification in preference-based reinforcement learning (PbRL)<n>We propose Posterior Sampling for Preference Learning ($mathsfPSPL$), a novel algorithm inspired by Top-Two Thompson Sampling.
  arXiv  Detail & Related papers  (2025-01-31T03:55:10Z)
• Optimal Design for Reward Modeling in RLHF [83.3614658277817]
  We formalize the reward training model in Reinforcement Learning from Human Feedback. We frame the selection of an effective dataset as a simple regret minimization task. We derive bounds on the simple regret under appropriate assumptions.
  arXiv  Detail & Related papers  (2024-10-22T14:36:44Z)
• MA-RLHF: Reinforcement Learning from Human Feedback with Macro Actions [46.608747360764035]
  Reinforcement learning from human feedback (RLHF) has demonstrated effectiveness in aligning large language models (LLMs) with human preferences. We propose MA-RLHF, a simple yet effective RLHF framework that incorporates macro actions -- sequences of tokens or higher-level language constructs -- into the learning process.
  arXiv  Detail & Related papers  (2024-10-03T17:55:13Z)
• A Distribution-Aware Flow-Matching for Generating Unstructured Data for Few-Shot Reinforcement Learning [1.0709300917082865]
  We introduce a distribution-aware flow matching approach to generate synthetic unstructured data for few-shot reinforcement learning.<n>Our approach addresses key challenges in traditional model-based RL, such as overfitting and data correlation.<n>Results demonstrate that our method achieves stable convergence in terms of maximum Q-value while enhancing frame rates by 30% in the initial timestamps.
  arXiv  Detail & Related papers  (2024-09-21T15:50:59Z)
• RLHFuse: Efficient RLHF Training for Large Language Models with Inter- and Intra-Stage Fusion [10.165579735221092]
  Existing RLHF systems suffer from low GPU utilization in production deployments. RLHFuse breaks the traditional view of RLHF workflow as a composition of individual tasks. RLHFuse increases the training throughput by up to 3.7x, compared to existing state-of-the-art systems.
  arXiv  Detail & Related papers  (2024-09-20T05:15:38Z)
• Progressively Label Enhancement for Large Language Model Alignment [42.01694160556464]
  Large Language Models (LLM) alignment aims to prevent models from producing content that misaligns with human expectations. We propose PLE, a framework that dynamically adjusts the model's training process based on the evolving quality of the generated data.
  arXiv  Detail & Related papers  (2024-08-05T16:21:17Z)
• BOND: Aligning LLMs with Best-of-N Distillation [63.254031574394965]
  We propose Best-of-N Distillation (BOND), a novel RLHF algorithm that seeks to emulate Best-of-N but without its significant computational overhead at inference time. Specifically, BOND is a distribution matching algorithm that forces the distribution of generations from the policy to get closer to the Best-of-N distribution. We demonstrate the effectiveness of our approach and several design choices through experiments on abstractive summarization and Gemma models.
  arXiv  Detail & Related papers  (2024-07-19T18:38:25Z)
• SAIL: Self-Improving Efficient Online Alignment of Large Language Models [56.59644677997827]
  Reinforcement Learning from Human Feedback is a key method for aligning large language models with human preferences. Recent literature has focused on designing online RLHF methods but still lacks a unified conceptual formulation. Our approach significantly improves alignment performance on open-sourced datasets with minimal computational overhead.
  arXiv  Detail & Related papers  (2024-06-21T18:05:35Z)
• Scaling Laws for Reward Model Overoptimization in Direct Alignment Algorithms [50.808123629394245]
  Direct Alignment Algorithms (DDAs) like Direct Preference Optimization have emerged as alternatives to the classical RLHF pipeline. This work formulates and formalizes the reward over-optimization or hacking problem for DAAs and explores its consequences across objectives, training regimes, and model scales.
  arXiv  Detail & Related papers  (2024-06-05T03:41:37Z)
• Adaptive debiased SGD in high-dimensional GLMs with streaming data [4.704144189806667]
  This paper introduces a novel approach to online inference in high-dimensional generalized linear models.<n>Our method operates in a single-pass mode, making it different from existing methods that require full dataset access or large-dimensional summary statistics storage.<n>The core of our methodological innovation lies in an adaptive descent algorithm tailored for dynamic objective functions, coupled with a novel online debiasing procedure.
  arXiv  Detail & Related papers  (2024-05-28T15:36:48Z)
• RLHF Workflow: From Reward Modeling to Online RLHF [79.83927049253924]
  We present the workflow of Online Iterative Reinforcement Learning from Human Feedback (RLHF) in this technical report. RLHF is widely reported to outperform its offline counterpart by a large margin in the recent large language model (LLM) literature. We show that supervised fine-tuning (SFT) and iterative RLHF can obtain state-of-the-art performance with fully open-source datasets.
  arXiv  Detail & Related papers  (2024-05-13T15:50:39Z)
• Parameter Efficient Reinforcement Learning from Human Feedback [27.687265760622918]
  Reinforcement Learning from Human Feedback (RLHF) effectively aligns pretrained Large Language and Vision-Language Models with human preferences. To alleviate some of the computational burden of fine-tuning, efficient methods, like LoRA were introduced. We benchmark the PE-RLHF setup on six diverse datasets spanning summarization, harmless/helpful response generation, UI automation, and visual question answering.
  arXiv  Detail & Related papers  (2024-03-15T21:43:46Z)
• Improving Reinforcement Learning from Human Feedback with Efficient Reward Model Ensemble [67.4269821365504]
  Reinforcement Learning from Human Feedback (RLHF) is a widely adopted approach for aligning large language models with human values. However, RLHF relies on a reward model that is trained with a limited amount of human preference data. We contribute a reward ensemble method that allows the reward model to make more accurate predictions.
  arXiv  Detail & Related papers  (2024-01-30T00:17:37Z)
• Iterative Preference Learning from Human Feedback: Bridging Theory and Practice for RLHF under KL-Constraint [56.74058752955209]
  This paper studies the alignment process of generative models with Reinforcement Learning from Human Feedback (RLHF) We first identify the primary challenges of existing popular methods like offline PPO and offline DPO as lacking in strategical exploration of the environment. We propose efficient algorithms with finite-sample theoretical guarantees.
  arXiv  Detail & Related papers  (2023-12-18T18:58:42Z)
• ESRL: Efficient Sampling-based Reinforcement Learning for Sequence Generation [43.506732624371786]
  We introduce two-stage sampling and dynamic sampling approaches to improve the sampling efficiency during training sequence generation models via RL. Experimental results show that the efficient sampling-based RL, referred to as ESRL, can outperform all baselines in terms of both training efficiency and memory consumption.
  arXiv  Detail & Related papers  (2023-08-04T09:35:45Z)
• Provable Reward-Agnostic Preference-Based Reinforcement Learning [61.39541986848391]
  Preference-based Reinforcement Learning (PbRL) is a paradigm in which an RL agent learns to optimize a task using pair-wise preference-based feedback over trajectories. We propose a theoretical reward-agnostic PbRL framework where exploratory trajectories that enable accurate learning of hidden reward functions are acquired.
  arXiv  Detail & Related papers  (2023-05-29T15:00:09Z)

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.