Back to events
Model releaseCritical global significanceConfirmed confidence

Meta publishes BART research and pretrained models

Meta published the BART denoising sequence-to-sequence model and made code and pretrained weights available through fairseq.

Event details

Meta published BART research on July 8, 2020. BART corrupts text with a noise function and learns to reconstruct the original text using a sequence-to-sequence Transformer with a bidirectional encoder and an autoregressive decoder. Meta’s official fairseq repository provided pretrained 140-million-parameter base and 400-million-parameter large models.

Why it matters

BART extended denoising autoencoder pretraining to sequence-to-sequence tasks and became a widely used foundation for summarization, generation, and language understanding.

92/100Global significance score. Regional effects are recorded only when the evidence supports a meaningful difference.

Access notes

The paper, code, and pretrained weights were publicly available.