Qwen 团队提出 Elastic Horizon 智能体 RL 调度法

DAIR.AI · @dair_ai · X·2026-09-11 02:00·44分钟前
AI 导读

Qwen 团队提出 Elastic Horizon,一种闭环控制器,用成功轨迹长度的 90 分位追踪"有效交互前沿",解决智能体 RL 中交互轮数调度这一开环控制问题。在 AppWorld 和 BFCL 上,固定轮数扫描出现明显饱和平台;该方法在 7B 和 14B 骨干上均取得最佳成功率,并节省最多 25% 的每步轨迹 token。

DAIR.AI@dair_ai
42AI 编辑部评分,满分 100

Qwen 团队提出 Elastic Horizon 智能体 RL 调度法

2026-09-11 02:00· 44分钟前
AI 导读

Qwen 团队提出 Elastic Horizon,一种闭环控制器,用成功轨迹长度的 90 分位追踪"有效交互前沿",解决智能体 RL 中交互轮数调度这一开环控制问题。在 AppWorld 和 BFCL 上,固定轮数扫描出现明显饱和平台;该方法在 7B 和 14B 骨干上均取得最佳成功率,并节省最多 25% 的每步轨迹 token。

Recommended read. Interaction horizon scheduling is an underexplored control problem in agentic RL

This paper from the Qwen team takes a closer look at the problem.

Scaling the maximum number of environment interactions per episode improves long-horizon agents, and curriculum methods that expand the horizon beat fixed-horizon training.

But those schedules are open-loop.

They increase monotonically to a manually specified maximum with no way to detect that expansion stopped helping.

The authors propose the effective interaction frontier, a dynamic boundary past which extra interactions give diminishing returns while cost keeps growing linearly. Fixed-horizon sweeps on AppWorld and BFCL show clear saturation plateaus.

Elastic Horizon is a closed-loop controller that tracks the boundary using the 90th percentile of successful trajectory lengths, a statistic already available during training.

It settles inside the saturation band from both under-capacity and over-capacity starts, gets the best success rates across 7B and 14B backbones, and saves up to 25% of per-step trajectory tokens.

Paper: https://academy.dair.ai/papers/elastic-horizon-discovering-the-effective-interaction-frontier-in-agentic-reinfo-2609.07247

来源:DAIR.AI· x.com