CUNY-CL/yoyodyne

Non-autoregressive sequence generation

Open

#262 aperta il 25 ott 2024

Vedi su GitHub
 (0 commenti) (0 reazioni) (1 assegnatario)Python (21 fork)auto 404
enhancementgood first issuenew architecture

Metriche repository

Star
 (35 star)
Metriche merge PR
 (Metriche PR in attesa)

Descrizione

Right now, our dominant paradigm is autoregressive prediction. While this is SOTA ,it doesn't allow experimentation with other sequence-to-sequence tasks. Technically, we can define an arbitrary sequence length >>>> than any input or output and train a model to project an input to this length and then decode over it for non-autoregressive prediction. Example algorithms would be: Listen Attend Spell or CTC.

Guida contributor