When Does Memory Help? A Cost-Aware Evaluation of Long-Term Memory in Tool-Using LLM Agents
A benchmark and harness that measures the marginal utility of memory for task-executing agents under explicit cost accounting, and provides episodic tool-use tasks in three domains whose dependence on earlier-episode facts is verified by an automated leak check.
Shweta Mishra, Shashank Mishra
· 0 citations