1 result for “contrastive-learning”
CLIP (Contrastive Language-Image Pre-training) is a multimodal neural network model developed by OpenAI that learns visual concepts from natural language descriptions by jointly training an image encoder and a text encoder on 400 million image-text pairs.