CoLT5: Faster Long-Range Transformers with Conditional Computation.
Joshua AinslieTao LeiMichiel de JongSantiago OntañónSiddhartha BrahmaYury ZemlyanskiyDavid C. UthusMandy GuoJames Lee-ThorpYi TayYun-Hsuan SungSumit SanghaiPublished in: CoRR (2023)