Rollout-Training Co-Design for Efficient LLM-Based Multi-Agent Reinforcement Learning

• URL: http://arxiv.org/abs/2602.09578v1
• Date: Tue, 10 Feb 2026 09:27:03 GMT
• Title: Rollout-Training Co-Design for Efficient LLM-Based Multi-Agent Reinforcement Learning
• Authors: Zhida Jiang, Zhaolong Xing, Jiawei Lu, Yipei Niu, Qingyuan Sang, Liangxu Zhang, Wenquan Dai, Junhua Shu, Jiaxing Wang, Qiangyu Pei, Qiong Chen, Xinyu Liu, Fangming Liu, Ai Han, Zhen Chen, Ke Zhang,
• Abstract summary: Existing training frameworks fail to address the unique system-level challenges of multi-agent reinforcement learning.<n>We propose FlexMARL, the first end-to-end training framework that holistically optimize rollout, training, and their orchestration.<n>We show that FlexMARL achieves up to 7.3x speedup and improves hardware utilization by up to 5.6x compared to existing frameworks.
• Score: 21.32826098306222
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Despite algorithm-level innovations for multi-agent reinforcement learning (MARL), the underlying networked infrastructure for large-scale MARL training remains underexplored. Existing training frameworks primarily optimize for single-agent scenarios and fail to address the unique system-level challenges of MARL, including rollout-training synchronization barriers, rollout load imbalance, and training resource underutilization. To bridge this gap, we propose FlexMARL, the first end-to-end training framework that holistically optimizes rollout, training, and their orchestration for large-scale LLM-based MARL. Specifically, FlexMARL introduces the joint orchestrator to manage data flow under the rollout-training disaggregated architecture. Building upon the experience store, a novel micro-batch driven asynchronous pipeline eliminates the synchronization barriers while providing strong consistency guarantees. Rollout engine adopts a parallel sampling scheme combined with hierarchical load balancing, which adapts to skewed inter/intra-agent request patterns. Training engine achieves on-demand hardware binding through agent-centric resource allocation. The training states of different agents are swapped via unified and location-agnostic communication. Empirical results on a large-scale production cluster demonstrate that FlexMARL achieves up to 7.3x speedup and improves hardware utilization by up to 5.6x compared to existing frameworks.

Related papers

• DiRL: An Efficient Post-Training Framework for Diffusion Language Models [54.405206032785706]
  Diffusion Language Models (dLLMs) have emerged as promising alternatives to Auto-Regressive (AR) models.<n>Existing methods suffer from computational inefficiency and objective mismatches between training and inference.<n>We introduce DiRL, an efficient post-training framework that tightly integrates FlexAttention-accelerated blockwise training with LMDeploy-optimized inference.
  arXiv  Detail & Related papers  (2025-12-23T08:33:19Z)
• Part II: ROLL Flash -- Accelerating RLVR and Agentic Training with Asynchrony [78.70328630805041]
  ROLL Flash is a system that extends ROLL with native support for asynchronous RL post-training.<n>We show that ROLL Flash significantly improves resource utilization and scalability over synchronous RL post-training.
  arXiv  Detail & Related papers  (2025-10-13T12:41:27Z)
• CollaPipe: Adaptive Segment-Optimized Pipeline Parallelism for Collaborative LLM Training in Heterogeneous Edge Networks [57.95170323315603]
  We introduce CollaPipe, a distributed learning framework that integrates collaborative pipeline parallelism with federated aggregation to support self-evolving networks.<n>In CollaPipe, the encoder part is adaptively partitioned into variable-sized segments and deployed across mobile devices for pipeline-parallel training, while the decoder is deployed on edge servers to handle generative tasks.<n>To enhance training efficiency, we formulate a joint optimization problem that adaptively allocates model segments, micro-batches, bandwidth, and transmission power.
  arXiv  Detail & Related papers  (2025-09-24T07:54:01Z)
• ElasticMM: Efficient Multimodal LLMs Serving with Elastic Multimodal Parallelism [9.93378263858092]
  Multimodal large language models (MLLMs) handle images, videos, and audio by incorporating feature extractors and projection modules.<n>Current tightly coupled serving architectures struggle to distinguish between mixed request types.<n>We introduce Elastic Multimodal Parallelism (EMP), a new serving paradigm that elastically adapts to resource heterogeneity.
  arXiv  Detail & Related papers  (2025-07-14T08:53:48Z)
• AsyncFlow: An Asynchronous Streaming RL Framework for Efficient LLM Post-Training [24.60677187852425]
  Reinforcement learning (RL) has become a pivotal technology in the post-training phase of large language models (LLMs)<n>Traditional task-colocated RL frameworks suffer from significant scalability bottlenecks.<n>Task-separated RL frameworks face challenges in complex dataflows and the corresponding resource idling and workload imbalance.<n>We propose AsyncFlow, an asynchronous streaming RL framework for efficient post-training.
  arXiv  Detail & Related papers  (2025-07-02T12:45:34Z)
• StreamRL: Scalable, Heterogeneous, and Elastic RL for LLMs with Disaggregated Stream Generation [55.75008325187133]
  Reinforcement learning (RL) has become the core post-training technique for large language models (LLMs)<n>StreamRL is designed with disaggregation from first principles to address two types of performance bottlenecks.<n> Experiments show that StreamRL improves throughput by up to 2.66x compared to existing state-of-the-art systems.
  arXiv  Detail & Related papers  (2025-04-22T14:19:06Z)
• An Offline Multi-Agent Reinforcement Learning Framework for Radio Resource Management [5.771885923067511]
  offline multi-agent reinforcement learning (MARL) addresses key limitations of online MARL.<n>We propose an offline MARL algorithm for radio resource management (RRM)<n>We evaluate three training paradigms: centralized, independent, and centralized training with decentralized execution (CTDE)
  arXiv  Detail & Related papers  (2025-01-22T16:25:46Z)
• Flextron: Many-in-One Flexible Large Language Model [85.93260172698398]
  We introduce Flextron, a network architecture and post-training model optimization framework supporting flexible model deployment. We present a sample-efficient training method and associated routing algorithms for transforming an existing trained LLM into a Flextron model. We demonstrate superior performance over multiple end-to-end trained variants and other state-of-the-art elastic networks, all with a single pretraining run that consumes a mere 7.63% tokens compared to original pretraining.
  arXiv  Detail & Related papers  (2024-06-11T01:16:10Z)
• A Bayesian Framework of Deep Reinforcement Learning for Joint O-RAN/MEC Orchestration [12.914011030970814]
  Multi-access Edge Computing (MEC) can be implemented together with Open Radio Access Network (O-RAN) over commodity platforms to offer low-cost deployment. In this paper, a joint O-RAN/MEC orchestration using a Bayesian deep reinforcement learning (RL)-based framework is proposed.
  arXiv  Detail & Related papers  (2023-12-26T18:04:49Z)
• MALib: A Parallel Framework for Population-based Multi-agent Reinforcement Learning [61.28547338576706]
  Population-based multi-agent reinforcement learning (PB-MARL) refers to the series of methods nested with reinforcement learning (RL) algorithms. We present MALib, a scalable and efficient computing framework for PB-MARL.
  arXiv  Detail & Related papers  (2021-06-05T03:27:08Z)

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.