TY  - RPRT
TI  - Sparse Autoencoders Reveal Interpretable and Steerable Features in VLA Models
AU  - Aiden Swann
AU  - Lachlain McGranahan
AU  - Hugo Buurmeijer
AU  - Monroe Kennedy III
AU  - Mac Schwager
PY  - 2026
UR  - https://arxiv.org/abs/2603.19183
ID  - 2603.19183
ER  -