The authors evaluated tool graph memory across nine models from different families and sizes on two multi-turn tool-use benchmarks.
Tool graph memory rarely improves multi-turn agents by itself
Across nine language models and two benchmarks, most apparent gains came from additional tool calls rather than accurate memory guidance.
Chinese Tech
Uiji Hwang · Richeng Xuan · Taeuk Kim
Hanyang University · Hunyuan, Tencent
Research Digest··2 min read
Hwang, Xuan and Kim test whether storing previous tool transitions helps language-model agents choose tools over multi-turn interactions.
Why this paper
From Hunyuan, Tencent and Hanyang University
In one line
Tool graph memory rarely improves multi-turn tool use, and apparent gains come from extra computation rather than the stored guidance.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ·No stated limitations found
- ·No benchmark numbers found
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§