Tag
Prime Intellect 的 prime-rl 0.10.0 引入了 IPO（Intellect Policy Optimization），旨在解决异步强化学习中训练器与推理引擎对 token 概率估计不一致的真实问题。
推荐 HuggingFace 博客文章系列《Score Function Centering》，从直觉出发系统讲解 LLM 异步强化学习中的 trainer-generator 失配问题以及分数函数中心化（SFC）如何修复策略梯度中零期望性质，只需了解 GRPO 即可阅读。
The authors have created a lightweight, performant asynchronous reinforcement learning stack in pure JAX, sharing a work log with insights on inference, RDMA weight transfer, memory optimizations, and sharding for scaling RL systems.
PRIME-RL is a framework for large-scale asynchronous reinforcement learning, designed to be hackable and scale to 1000+ GPUs with support for various models and environments.
Discusses a new paper from Tsinghua/Zhipu AI on asynchronous reinforcement learning for agents, and notes that their previous GLM-5.2 model uses a critic instead of GRPO.
AsyncWebRL introduces an asynchronous multi-step reinforcement learning system for vision-language web agents, achieving up to 2.9x training speedup and setting a new state-of-the-art on WebGym by replacing per-trajectory normalization with a constant to reduce trajectory length inefficiency.