NLP for Developers: Transformers | Rasa

Поделиться
HTML-код
  • Опубликовано: 29 янв 2025

Комментарии • 36

  • @glorian111
    @glorian111 4 года назад +36

    This is probaly the most succinct and clear explanation of Transformers that I have found so far. Thanks for sharing!

  • @CodeEmporium
    @CodeEmporium 4 года назад +5

    Nice work!

  • @chamisoufiane1054
    @chamisoufiane1054 3 года назад +4

    Rachel can make things a lot easier to understand as always. Thank you so much!

    • @RasaHQ
      @RasaHQ  3 года назад +2

      aw, ty so much! :) -R

  • @hajhouj
    @hajhouj 3 года назад +2

    Good work, brief but clear

  • @JosephElfassi
    @JosephElfassi 4 года назад +11

    Very clear, concise! Thank you!

  • @rahulmenon3051
    @rahulmenon3051 4 года назад +4

    Very nicely explained .Great work !

  • @akshaybhardwaj10
    @akshaybhardwaj10 4 года назад +6

    Thank you Rachel

  • @jaskaranvirdi6622
    @jaskaranvirdi6622 3 года назад +2

    This was great. Succinct explanation and describing the pros and cons. Thank you!

  • @RamiroG
    @RamiroG 4 года назад +6

    Very well explained keep up the good work!

  • @雨霏吳-s1p
    @雨霏吳-s1p 2 года назад +2

    Very helpful! Thanks a lot

  • @Enerdzizer
    @Enerdzizer 4 года назад

    Thanks for explanation! Still have a question- in 1:45 left side is output and right side is input? It's in contrary to previous narrative, when we are evolving in time from the left to the right

  • @govindbisen388
    @govindbisen388 4 года назад +6

    Wish I had any teacher like you.
    😣

  • @EngRiadAlmadani
    @EngRiadAlmadani 3 года назад +1

    Great work

  • @edbertkhovey
    @edbertkhovey 3 года назад

    question, you say that the drawback is the training part. what if I use a pre-trained model from the hugging face website? is there a flaw in a pre-trained model?

    • @RasaHQ
      @RasaHQ  3 года назад +1

      I would say it's more a drawback to transformers as opposed to other architectures. If you're using a pre-trained model then you're right, it's not something you need to worry about especially. :) -R

    • @edbertkhovey
      @edbertkhovey 3 года назад +1

      @@RasaHQ ah yes okay thank you

  • @auroratijerina6941
    @auroratijerina6941 4 года назад +4

    This was great, thank you!

  • @leixun
    @leixun 4 года назад +1

    Hi, I am wondering if the decoder is more computationally expensive than encoder? or they are at the same level?

    • @RasaHQ
      @RasaHQ  4 года назад +1

      In the original transformer architecture, the decoder is slightly more expensive (in terms of # of parameters) because each of the decoders inside of it has an additional attention layer.

    • @leixun
      @leixun 4 года назад +1

      @@RasaHQ Thanks, how much more would you roughly expect? 2x? 10? 50x?

  • @thenoorer
    @thenoorer 4 года назад +5

    short and informative

  • @alteshaus3149
    @alteshaus3149 2 года назад

    Thank you very much

  • @sandeepkumawat4982
    @sandeepkumawat4982 4 года назад +1

    So my searching ends here..?...or it just starts??

  • @vicrazio
    @vicrazio 4 года назад +3

    Wow ❤️

  • @user-or7ji5hv8y
    @user-or7ji5hv8y 4 года назад

    Can you do a video detailing how self attention and multi-head works.

    • @RasaHQ
      @RasaHQ  4 года назад

      It's not the same format, but there's an algorithm whiteboard video that goes into more detail: ruclips.net/video/yGTUuEx3GkA/видео.html

  • @cristian-bull
    @cristian-bull 4 года назад

    thank you good lady

  • @bhaskartripathi
    @bhaskartripathi 3 года назад +2

    Probably the best summary video of TL. I followed her on twitter : twitter.com/rctatman

  • @2broke2code
    @2broke2code 3 года назад

    Awesome! She should try working for kaggle :/

  • @kyrilcouda
    @kyrilcouda 3 года назад

    does she actually have hair? Nice video, btw.

  • @airepublic9864
    @airepublic9864 2 года назад

    F F

  • @WahranRai
    @WahranRai 4 года назад

    It is better if you reduce your speed !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

    • @RasaHQ
      @RasaHQ  4 года назад

      Thanks for the feedback. If you like, you can reduce the speed of the video yourself on your local device: support.google.com/youtube/answer/7509567?co=GENIE.Platform%3DAndroid&hl=en

    • @WahranRai
      @WahranRai 4 года назад

      @@RasaHQ Do you think i dont know that ! Try to reduce yourself and watch the quality

    • @bellisma1927
      @bellisma1927 4 года назад

      No need to be so rude. In my opinion it sounds fine with speed reduced to 0.75. Also there are closed captions.