arXiv cs.AI· Yepeng Liu, Shengmai Chen, Xuandong Zhao, Chengzhi Liu, Dawn Song, Xin Eric Wang, Yuheng Bu·· 2 小时前AI 评分40
测试时记忆演化（TTME）能否超越测试时训练用于科学发现？
Can Test-Time Memory Evolution Outperform Test-Time Training for Scientific Discovery?
AI 导读
研究者提出即插即用的测试时适应方法 Test-Time Memory Evolution（TTME），作为测试时训练（TTT）的替代方案，通过持续更新外部记忆蒸馏经验来引导后续科学发现，无需更新模型权重。
正文 · 原文
Abstract:Recent \emph{test-time training} (TTT) frameworks have enabled open-weight large language models (LLMs) to outperform inference-only frameworks powered by frontier models on scientific discovery tasks. However, TTT introduces additional computational cost for online weight updates during the discovery process, making computational efficiency an important consideration when scaling to larger models. This motivates us to explore an efficient \emph{test-time adaptation} method for scientific discovery. We introduce \emph{Test-Time Memory Evolution} (TTME), a plug-and-play test-time adaptation method, as an alternative to TTT. TTME continually updates the external memory by distilling the experience to guide subsequent discovery without updating model weights. To have a clean comparison, we evaluate TTME and TTT under the same discovery framework, replacing only the test-time adaptation method. Our evaluation shows that TTME outperforms TTT on all five mathematical tasks, while reducing computational cost by $82.8$-$84.6\%$ in FLOPs and $47.7$-$58.4\%$ in runtime. These results establish TTME as an effective and computationally efficient alternative to TTT in the evaluated discovery settings.
| Subjects: | Artificial Intelligence (cs.AI); Machine Learning (cs.LG) |
| Cite as: | arXiv:2610.32046 [cs.AI] |
| (or arXiv:2610.32046v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.32046 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Yepeng Liu [view email]
[v1]
Fri, 23 Oct 2026 01:37:19 UTC (4,793 KB)
来源：arXiv cs.AI · arxiv.org