PulseAugur
中
实时 12:49:54
English(EN) Trajectory2Gym: Decision-Grounded Compositional Interaction Synthesis for Tool-Use Agents

Trajectory2Gym框架通过决策接地增强工具使用智能体训练 · 跟踪2个来源

研究人员开发了Trajectory2Gym,一个旨在改进工具使用智能体训练的新型框架。该系统专注于构建确保组合覆盖和决策接地的交互,而不是简单地收集成功的轨迹。通过从已实现执行中的接地决策中获得监督,Trajectory2Gym旨在创造更具学习性的交互体验。在Qwen3模型的五个尺度上,这种方法显著提高了在Tau1、Tau2和BFCL基准测试上的性能,在有状态和多轮交互中显示出最强的提升。 AI

影响 增强了AI智能体的训练方法,可能提高了它们处理复杂、多步骤任务的能力。

排序理由 该集群包含一篇详细介绍AI智能体训练新框架的研究论文。

在 Hugging Face Daily Papers 阅读 →

AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →

Trajectory2Gym框架通过决策接地增强工具使用智能体训练 · 跟踪2个来源

本文如何被排名

Signal score
0 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Research
该集群包含一篇详细介绍AI智能体训练新框架的研究论文。
Source corroboration
2 independent sources
Multiple independent publishers reporting the same story raises confidence that it's real and newsworthy.
Topics
paper, model release
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
3 days old
Aged out of breaking-news scoring windows; ranking reflects the durable signal from the full source set.

完整方法见我们的编辑标准。

报道来源 [2]

  1. arXiv cs.CL TIER_1 English(EN) · Yihao Hu, Zhihao Wen, Yitong Lin, Siyuan Lu, Pan Wang, Wei Wu, Jiang-Ming Yang, Tao Lin ·

    Trajectory2Gym: 决策驱动的工具使用智能体的组合交互合成

    arXiv:2610.33568v1 Announce Type: new Abstract: Successful tool-use trajectories are not necessarily useful supervision. Even successful interactions may underrepresent multi-step dependencies or contain actions that are unjustified by the information visible to the agent at deci…

  2. Hugging Face Daily Papers TIER_1 English(EN) ·

    Trajectory2Gym: 决策驱动的工具使用智能体的组合交互合成

    Trajectory2Gym improves tool-use training by constructing interactions with compositional coverage and decision grounding rather than collecting successful trajectories.