Research
HABIT-Bench tests whether agent memory systems can infer user habits from weak evidence
Best of eight memory methods reaches 33.61% macro accuracy across food, finance, software and travel domains. Performance stays limited even when annotated evidence is supplied directly.
Every outlet that ran it 1 article
Hugging Face PapersResearchLead
HABIT-Bench: Evaluating Habit Induction from Longitudinal Weak Evidence in Agent Memory