Authors

Junjie Hu

Publications 2

Preprint Open access

Universal Test-Time Training

Zefan Cai, Qinzhe Hu, Ziqiao Ma et al. · 2026

Recent Test-Time Training (TTT) architectures compress context into fast weights that are updated online and queried as memory. Existing TTT designs keep this memory private to each layer: it recurs only over time, and depth merely indexes L separate memories. We argue that memory ownership need not be tied to depth, a …

Preprint Open access

Scaling 3D Visual Grounding in Abdominal CT

Sam Church, Danyal Maqbool, Joshua D. Warner et al. · 2026

Visual grounding models can enhance radiology workflows by linking report findings to image regions. This is particularly valuable for 3D CT, where findings often occupy a tiny fraction of the volume. Training 3D grounding models requires large sets of paired phrases and regions, and building such datasets is expensive …

Co-authors

Zefan Cai Qinzhe Hu Ziqiao Ma Hao Tan Sam Church Danyal Maqbool Joshua D. Warner Andrew Voter Meghan G. Lubner Tyler J. Bradshaw