GTR: Gated Token Recurrence for Efficient Dense Prediction
A small neural model for parsing date & time
Implementation of the recurrent looped transformer proposed by Yifan Zhang of Princeton
GTR: Gated Token Recurrence for Efficient Dense Prediction
A small neural model for parsing date & time
Implementation of the recurrent looped transformer proposed by Yifan Zhang of Princeton