Decoder[[cv-decoder]]
Decoder-only vision models are rare because most vision models rely on an encoder to learn an image representation.