J017athan/

ImplicitMemBench

ImplicitMemBench: Measuring Unconscious Behavioral Adaptation in Large Language Models

ACL 2026 Oral Presentation

Chonghan Qin1, Xiachong Feng1*, Weitao Ma2, Xiaocheng Feng2, Lingpeng Kong1

1The University of Hong Kong    2Harbin Institute of Technology

*Corresponding Author

ImplicitMemBench is a benchmark for studying whether large language models can acquire, retain, and express implicit memory through behavior. This… See the full description on the dataset page: https://huggingface.co/datasets/J017athan/ImplicitMemBench.