Vision Transformers (ViT)

Transformer Encoder

Fundamental block composed of self-attention layers and feed-forward networks alternating with normalization and residual connections.

← Kembali