What do we call an encoder-decoder with 0 encoder layers and the cross-attention of decoder layers are directed to the outputs of the encoder embedding layer?
- 0-N Enc-decoder
- Decoder-only with cross-attention
- Prefix-LM
- Decoder-only
- Others
What do we call an encoder-decoder with 0 encoder layers and the cross-attention of decoder layers are directed to the outputs of the encoder embedding layer?