Tag
该论文提出Δτ-Bench，一个评估AI智能体在不断演化的工具生态中适应能力的可执行基准，包含42个领域、6个生态状态下的1,588个多轮任务；并提出Steward生态系统管理器，通过增量技能维护和动态供应实现更高的持续任务成功率。
Meta AI 提出 Retrospective Self-Distillation (RSD)，一个利用特权回顾性上下文让模型对自身动作重新打分并回灌策略的多轮自进化框架，在 ALFWorld、ScienceWorld 和 SearchQA 上相对基线提升最高 59.7%。
M-Plan is a closed-loop multi-agent system framework that integrates long-term and short-term memory with global-local and past-current guidance to improve LLM agents' planning on long-horizon multi-turn tasks, outperforming baselines on ALFWorld, Webshop, and PDDL.
The paper proposes Turn-level Multiscale Density Ratio Estimation (tlm-DRE), a post-training method that assigns different weights to turns and uses asymmetric token-level training to enhance LLM agents' performance in multi-turn reasoning tasks, showing competitive results on benchmarks.