What are the differences between fairseq and fairseq2?
Quotes from the github pages are not very clear
Fairseq(-py) is a sequence modeling toolkit that allows researchers and developers to train custom models for translation, summarization, language modeling and other text generation tasks.
and
fairseq2 is a sequence modeling toolkit that allows researchers and developers to train custom models for translation, summarization, language modeling, and other content generation tasks. It is also the successor of fairseq.
So far I realized there're API changes (for example, model loading from checkpoint), and the addition of new multimodal such as Seamless communication,
but why abandon a well-established framework and start from scratch?