Qwen 发布 QwenGyre，为超长程智能体训练引入弹性 RL
Introducing QwenGyre, bringing elastic RL to extreme-long-horizon (XLong) agent training. ✨
Qwen 发布 QwenGyre，将弹性 RL 用于超长程（XLong）智能体训练，包含 Elastic Scheduler 和 Trajectory Processor 两项核心设计。Elastic Scheduler 按需在 rollout 与训练之间重新分配 GPU 且不中断执行，相比 Colocate 最高提速 1.85 倍、相比 Async 提速 1.78 倍；Trajectory Processor 把分支执行转为前缀共享的轨迹树，超时后的部分进展仍可被评分。在 Qwen3.8-2.4T-A95B 上以每次 rollout 约 700K tokens 扩展，48 步将 NL2RepoBench 从 52.5% 提升到 58.5%，端到端训练时间相比 Async 减少约 44%。
推荐理由
精选AI 评分 60
QwenGyre 用弹性调度和轨迹树处理超长程智能体 RL，给出了相对 Colocate 与 Async 的加速比和基准提升数据。
主题
正文
当前显示原文，中文全文译文尚未就绪。
Introducing QwenGyre, bringing elastic RL to extreme-long-horizon (XLong) agent training. ✨
It launches with two core designs: 🥳 - Elastic Scheduler: reallocates GPUs between rollout & training as demand shifts—live executions never interrupted. Up to 1.85× speedup over Colocate, 1.78× over Async. - Trajectory Processor: turns branching executions (compaction, sub-agents, retries) into prefix-shared trajectory trees with bounded, deduplicated training samples. Partial progress after timeout still gets scored.
Scaled to Qwen3.8-2.4T-A95B with ~700K tokens per rollout, QwenGyre lifts NL2RepoBench from 52.5% to 58.5% in 48 steps, cutting end-to-end training time by ~44% vs Async.