output
20062026
most citedTransformers are RNNs: Fast Autoregressive Transformers with Linear Attention

310 citations

Showing 2020 · math.PRShow all

3 papers · 2 filters