Controlling Action-Head Co-Adaptation to VLM Representation Anisotropy
Modern learning systems increasingly combine pretrained models with downstream heads. Despite the growing importance of such modular designs, how the downstream model co-adapts to the geometry of pretrained representations remains relatively underexplored.
By Ruijie Zhang, Jiayin Zheng, Yequan Zhao, Liyan Tan, Ziyue Liu, Zhengyang Wang, Kunpeng Song, Nathan Zhao, Jian Yao, Yin Zhang, Zheng ZhangarXiv - cs.AI
Read the full story
Originally published by arXiv - cs.AI on Nov 4, 2026.
Comments
Sign in to join the discussion.