ALiBi enables transformer language models to handle longer inputs

Поделиться
HTML-код
  • Опубликовано: 3 окт 2024

Комментарии • 5

  • @MikeMm-n9n
    @MikeMm-n9n Год назад +8

    Very nicely presented. Thank you for sharing this. I became aware of the ALIBI following the release of the MPT-7B. Nice to see the creator of ALIBI presenting it. Congratulations !

  • @QifangZhao
    @QifangZhao 6 месяцев назад

    starting from 21:20, the intuitive explanation is very interesting!

  • @TongLeiTong-Leii
    @TongLeiTong-Leii Год назад +2

    So instructive!

  • @bnglr
    @bnglr 9 месяцев назад

    who is the presenter