contrastive objective

3 papers

• AmorLIP: Efficient Language-Image Pretraining via Amortization
• Removing Concepts from Text-to-Image Models with Only Negative Samples
• UniViT: Unifying Image and Video Understanding in One Vision Encoder