Sequence-to-sequence model uses both encoder and decoder part of the transformer. The encoder part can access all the words form the input and the decoder part has access to all the generated words to the left and the encoding from encoder.

It is used for all the tasks that can be done by Decoder Only Transformer, i.e., text generation, summarization etc.

Some examples:

1. BART
2. mBART
3. T5

References