today local_bar

Jun Suzuki

Full Professor, Tohoku University

1 paper at NeurIPS 2025

Homepage· OpenReview· Semantic Scholar· Google Scholar

Poster Session 4

Thursday, December 4, 2025 · 4:30 PM → 7:30 PM

Exhibit Hall C,D,E

Transformer Key-Value Memories Are Nearly as Interpretable as Sparse Autoencoders

#900 · Mengyu Ye, Jun Suzuki, Tatsuro Inaba, Tatsuki Kuribayashi

We find that transformer key-value memories are nearly as interpretable as SAE features