LLM2CLIP use LLM to teach CLIP vision encoder.

LLM2CLIP

Enabling existing pretrained models to become stronger with minimal fine-tuning

Publications

Publications by Year

Publications by Research Area

Publications by Type