ECHO-2: A Large-Scale Distributed Rollout Framework for Cost-Efficient Reinforcement Learning

TL;DR

ECHO-2 is a scalable distributed reinforcement learning framework that enhances cost efficiency in large language model post-training by overlapping rollout, dissemination, and training processes.

Contribution

It introduces a novel capacity model and peer-assisted broadcast techniques to optimize distributed RL with latency and cost considerations.

Findings

ECHO-2 achieves significant cost savings in large-scale RL training.

It maintains RL reward performance comparable to strong baselines.

Experiments demonstrate effectiveness on 4B and 8B models under real bandwidth conditions.

Abstract

Reinforcement learning (RL) is a critical stage in post-training large language models (LLMs), involving repeated interaction between rollout generation, reward evaluation, and centralized learning. Distributing rollout execution offers opportunities to leverage more cost-efficient inference resources, but introduces challenges in wide-area coordination and policy dissemination. We present ECHO-2, a distributed RL framework for post-training with remote inference workers and non-negligible dissemination latency. ECHO-2 combines centralized learning with distributed rollouts and treats bounded policy staleness as a user-controlled parameter, enabling rollout generation, dissemination, and training to overlap. We introduce an overlap-based capacity model that relates training time, dissemination latency, and rollout throughput, yielding a practical provisioning rule for sustaining learner utilization. To mitigate dissemination bottlenecks and lower cost, ECHO-2 employs peer-assisted pipelined broadcast and cost-aware activation of heterogeneous workers. Experiments on GRPO post-training of 4B and 8B models under real wide-area bandwidth regimes show that ECHO-2 significantly improves cost efficiency while preserving RL reward comparable to strong baselines.