Ahmadzei's picture
added 3 more tables for large emb model
5fa1a76
Our BARTpho uses the "large" architecture and pre-training
scheme of the sequence-to-sequence denoising model BART, thus especially suitable for generative NLP tasks.