LLMs: A Journey Through Time and Architecture

Поделиться
HTML-код
  • Опубликовано: 14 ноя 2024

Комментарии • 42

  • @SebastianRaschka
    @SebastianRaschka  Месяц назад +3

    If someone is interested in a code tutorial converting the GPT model to Llama, I have a step-by-step guide here: github.com/rasbt/LLMs-from-scratch/blob/main/ch05/07_gpt_to_llama/converting-gpt-to-llama2.ipynb (will add it to the description)

    • @SHAMIKII
      @SHAMIKII Месяц назад +1

      Certainly, me, me, me.
      Thank you very much for all your content.

  • @oldmankatan7383
    @oldmankatan7383 7 дней назад +1

    Nice round up! Thank you for this.

  • @hiramcoriarodriguez1252
    @hiramcoriarodriguez1252 Месяц назад +4

    Your book is a master peace, congratulations

  • @SanjaySingh-gj2kq
    @SanjaySingh-gj2kq Месяц назад +3

    Bought your book on manning last year - one of the best book on LLM internals. Looking forward to get the print book

    • @SebastianRaschka
      @SebastianRaschka  Месяц назад +1

      Thanks for the kind words, glad to hear that you've been enjoying it! The print copies started shipping and I hope you get your's soon!

  • @abdulhamidmerii5538
    @abdulhamidmerii5538 Месяц назад +1

    Just received the print version of your book yesterday, I look forward to reading it!

    • @SebastianRaschka
      @SebastianRaschka  Месяц назад

      Good timing! I hope you like it and have a fun weekend ahead!

  • @tee_iam78
    @tee_iam78 Месяц назад +1

    A brilliant content. Thank you.

  • @mahdipourmirzaei1048
    @mahdipourmirzaei1048 Месяц назад +3

    GPT2 training did not train on 40 billion tokens, it was 40 GB of text which is equivalent to roughly 8 billion tokens or less.

  • @vaioslaschos
    @vaioslaschos Месяц назад +2

    I think the grouped-query attention is more than a trick for computational reduction. It says something deep about what is the best way to share information in a multiagent system to have the best performance. And it says something alont the lines that it is better to give little essential info and at the same time request multiple info from many sources.

    • @SebastianRaschka
      @SebastianRaschka  Месяц назад +2

      That's a nice interpretation regarding multi- and grouped-query attention. Thanks for sharing! If you go by the original papers though, the intention was more computation constraints and efficiency (e.g., see arxiv.org/abs/2305.13245), but yeah, perhaps it can actually help with modeling performance as well in certain scenarios (for instance, where there is massive overfitting otherwise).

    • @vaioslaschos
      @vaioslaschos Месяц назад

      @@SebastianRaschka I have no doubt that what you say is true, and in no way I wanted to imply you missed something. Two years ago, I spent couple of months training 100M models with different architectures. I did some weird stuff like putting all the attention layers first and then a big nonlinear layer. You will be surprised with how many monstrosities can actually work without losing too much performance. The two things I got from all this is a) There is some interesting intuition in group querying (that I cant fully articulate), and it will make sense for this to be explored further, b) skip connection, where you pass the value from previous layers to the current, is not a gimmick. If you remove it the performance drops a lot, which for me implies that attention mechanism is actually applied to get only the "new" info. I think that intuitions about the architecture is not passed from the researchers to the community and It is a pity. Also it is a pity that experimenting with architecture is a rich persons hobby. Anyway, I really like your channel. I subscribed :-).

  • @thefatcat-hd6ze
    @thefatcat-hd6ze Месяц назад +3

    Enjoying your book a lot :))

    • @SebastianRaschka
      @SebastianRaschka  Месяц назад

      Thanks! Glad to hear that it was worth all the long hours and weekends!

    • @thefatcat-hd6ze
      @thefatcat-hd6ze Месяц назад

      @@SebastianRaschka 🙏

  • @Ken-de6tp
    @Ken-de6tp Месяц назад +1

    Reading your new book ! 🎉🎉

  • @dc33333
    @dc33333 Месяц назад +1

    my favorite YT channel

  • @maikerodrigo4249
    @maikerodrigo4249 Месяц назад +2

    Llama 3.2 just came out today

    • @SebastianRaschka
      @SebastianRaschka  Месяц назад +1

      Ha yes, I wish I could insert additional slides! What's interesting is that the small model is back from RMSNorm to LayerNorm

  • @cletadjos
    @cletadjos Месяц назад +1

    Thanks for sharing 😊

  • @1msirius
    @1msirius 14 дней назад +1

    Hey, thanks for your videos also can you suggest to me your best book on Gen AI (I want to learn about transformers in detail)

    • @SebastianRaschka
      @SebastianRaschka  14 дней назад

      Glad you found the videos useful! Since you asked for a book recommendation: Build a Large Language Model From Scratch (amzn.to/4fqvn0D), where you build a transformer-based LLM from the ground up, implementing each single component.

  • @Innovatead_Solutions-e4u
    @Innovatead_Solutions-e4u Месяц назад

    Dear Sebastian Raschka, your channel caught our attention and we would like to explore advertising possibilities with you. Looking forward to discussing potential opportunities!

  • @SaiKiran-he5vy
    @SaiKiran-he5vy Месяц назад +1

    What is the pre-requisites knowledge required to explore your new book: `Build a Large Language Model (From Scratch)`

    • @SebastianRaschka
      @SebastianRaschka  Месяц назад +1

      Good question! It would require Python knowledge. PyTorch knowledge is also good to get started quicker, but not strictly necessary. If you are new to PyTorch, you can start with Appendix A, which is a ~50 page intro to PyTorch to get you up to speed

  • @SettimiTommaso
    @SettimiTommaso Месяц назад

    Yes!

  • @subaruhassufferredenough7892
    @subaruhassufferredenough7892 Месяц назад

    What do you mean by high quality annealing?

    • @SebastianRaschka
      @SebastianRaschka  Месяц назад

      They would select a small subset of very high quality data for the final annealing stage.

    • @subaruhassufferredenough7892
      @subaruhassufferredenough7892 Месяц назад +1

      What does annealing mean in the context of LLMs? Is it the same as what we mean by an annealing LR scheduler?

    • @SebastianRaschka
      @SebastianRaschka  Месяц назад +1

      @@subaruhassufferredenough7892 Yes, it's basically the same

    • @subaruhassufferredenough7892
      @subaruhassufferredenough7892 Месяц назад

      Do you know how they determined which data was high quality?

  • @parvesh-rana
    @parvesh-rana Месяц назад

    Explain transformers in detail

    • @SebastianRaschka
      @SebastianRaschka  Месяц назад +2

      That would be a very long video :D. But you might find my book useful in that respect.

  • @SerikPoliasc
    @SerikPoliasc Месяц назад

    Moore Daniel Taylor Brenda Anderson Eric

  • @rafsanjaniLab
    @rafsanjaniLab Месяц назад

    Hi Prof. Raschka, could you please attach the slides?