LLM2CLIP use LLM to teach CLIP vision encoder.

LLM2CLIP: Powerful Language Model Unlock Richer Visual Representation

Makes SOTA pretrained CLIP model more SOTA ever.

People

Portrait of Yifan Yang

Yifan Yang

Research SDE

Portrait of Xufang Luo

Xufang Luo

Senior Researcher

Portrait of Yuqing Yang

Yuqing Yang

Principal Research SDE Manager

Portrait of Qi Dai

Qi Dai

Senior Researcher

Portrait of Xiyang Dai

Xiyang Dai

Principal Researcher

Portrait of Dongdong Chen

Dongdong Chen

Senior Researcher

Portrait of Chong Luo

Chong Luo

Principal Research Manager

Portrait of Lili Qiu

Lili Qiu

Assistant Managing Director