AGENTIC R AG-R1: Agentic Reinforcement Learning with Stack Memory for Multi-Step Reasoning, Retrieval and Memorizing
尚未评估计划受阻研究发现claim-component-ablation

The paper reports that removing memory actions, rollout rejection, or both lowers average F1, and that removing both retrieval and memory rewards gives the lowest ablation average.

来源:paper-fixed:page 8, Table 2

报告指标与观测值

average F1

m-ablation-full

论文报告 33.55 percentage_points

实际观测 — percentage_points

average F1

m-ablation-no-rewards

论文报告 19.94 percentage_points

实际观测 — percentage_points

Assessment(0)

这条 Claim 暂无已发布的不可变 Assessment。

关联运行(0)

这条 Claim 暂无关联执行。