2 papers
cs.AI2026
Pretraining of Medical Visual Encoders Toward Multi-modal Large Language Models
Tianyou Jiang
Multimodal Large Language Models (MLLMs) commonly reuse visual encoders pretrained with CLIP, although the features of these ViTs are ultimately consumed by autoregressive LLMs. We…
cs.CV2026
TCLA: Training-Free Class-wise Logit Adaptation for Medical Vision-Language Models
Tianyou Jiang, Ziyu Zhou
Medical Vision-Language Models (VLMs) exhibit strong zero-shot performance, yet their effectiveness still declines on out-of-distribution (OOD) data due to domain shifts and class…