Expand description
Cross-space transfer functions for Joint Embedding Spaces
Zero-shot transfer, knowledge transfer, domain adaptation, CLIP-style contrastive training, data augmentation, curriculum learning.
Structs§
- CLIP
Aligner - CLIP-style contrastive learning implementation
Functions§
- zero_
shot_ retrieval - Zero-shot cross-modal retrieval helper (delegates to JointEmbeddingSpace)