AI Resonance · 阅读最新日报 · AI 入门推荐

2026-10-27 · America/Los_Angeles · 论文 · #5

Trajectory2Gym: Decision-Grounded Compositional Interaction Synthesis for Tool-Use Agents

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 decision time. We identify two coupled requirements for effective trajectory supervision: compositional coverage, so that important observation-action dependencies are represented, and decision grounding, so that each supervised action is supported by the learner-visible interaction history. We introduce Trajectory2Gym, which constructs executable interactions around…

Trajectory2Gym: Decision-Grounded Compositional Interaction Synthesis for Tool-Use Agents

热度 59.7 / 100;排名与评分保留该期记录。