SAE-V: Interpreting Multimodal Models for Enhanced Alignment.

Hantao Lou, Changye Li, Jiaming Ji, Yaodong Yang

Browse the full ICML paper archive.