Lambda Stack - install CUDA, Pytorch, and Tensorflow on Ubuntu with a single line

Поделиться
HTML-код
  • Опубликовано: 4 дек 2024

Комментарии • 28

  • @lfmtube
    @lfmtube 3 года назад +5

    Thank you. Perfect video. Just what I'm needing to stop struggling when I have to upgrade all the DL learning components.

  • @hyunglok
    @hyunglok 3 года назад +3

    Thank you, Steven. This is a very helpful video!

    • @lambdacloud
      @lambdacloud  3 года назад

      Glad you like it! We spent a lot of time developing Lambda Stack and hope it brings value to the DL community.

  • @LearningToLiveNow
    @LearningToLiveNow 2 года назад +1

    Excellent! I just purchased the Tensorbook 2022 and so excited!!

    • @pranavdurai
      @pranavdurai 2 года назад +1

      Dude, that's sweet! How is it?

    • @LearningToLiveNow
      @LearningToLiveNow 2 года назад

      @@pranavdurai So far, I an really enjoying it!

    • @pranavdurai
      @pranavdurai 2 года назад

      @@LearningToLiveNow Lovely! You should make a video talking about your experience using it.

    • @LearningToLiveNow
      @LearningToLiveNow 2 года назад +1

      @@pranavdurai I will definitely consider making a video. Thanks for the suggestion.

    • @pranavdurai
      @pranavdurai 2 года назад

      @@LearningToLiveNow Looking forward to it!

  • @tervancovan
    @tervancovan Год назад

    Wow I am just about to buy Tensorbook

  • @jr_blanco
    @jr_blanco 7 месяцев назад

    And ubuntu 24.04???

  • @JtechAI
    @JtechAI 9 месяцев назад

    I would like to award LamdaStack the Medal of SavingMeRidiculousAmountsOfTime. With V Device ;-)

  • @Diego17511
    @Diego17511 2 года назад

    Google loon may run flairs as necissary

  • @faangsde
    @faangsde 3 года назад +1

    May I install conda on top of it to control the version of Pytorch, TF? Most deeplearning libraries were developed on depreciated versions...

    • @byrospyro4432
      @byrospyro4432 3 года назад

      From what i've heard, Conda doesn't work with it, so you have to install mini-Conda instead which does work with this.

  • @iKeith3000
    @iKeith3000 3 года назад +1

    Pretty cool tech. Does it install Tensorflow from source?

    • @lambdacloud
      @lambdacloud  3 года назад

      It installs a binary that we compile from source. We keep it all up to date so you don't have to worry about it.

    • @iKeith3000
      @iKeith3000 3 года назад +1

      @@lambdacloud Okay. The reason I install from source is because the library (Tensorflow) is aware of the machine it’s being loaded to and will then include optimizations uniquely available for my machine. Will your binary have the ability to recognize the hardware and download additional optimizations available for that hardware?

    • @lambdacloud
      @lambdacloud  3 года назад +2

      @@iKeith3000 We make sure that most of the best optimizations are already enabled. Obviously compiling from source will enable you to use unique CPU-specific features, but, we find that those CPU-specific features don't move the needle that much. All of the GPU related optimizations are done by a combination of CuDNN and the frameworks themselves.

    • @iKeith3000
      @iKeith3000 3 года назад +1

      @@lambdacloud That's good to know. Thanks!

  • @carloshavierv
    @carloshavierv 3 года назад

    Does it resolve combinations of specific versions, say CUDA 9.2 and PyTorch 1.5?

    • @lambdacloud
      @lambdacloud  3 года назад +1

      No, it provides a particular "latest" build that has a compatible version of CUDA, PyTorch, Tensorflow, and drivers. You can't mix-and-match.

  • @Instant_Nerf
    @Instant_Nerf 2 года назад

    Lambda only installs on these computer you sell?

    • @JuanUys
      @JuanUys Год назад

      My guess is that it will be compatible with any NVIDIA-based system with an AMD or Intel CPU.

  • @QuintinMassey
    @QuintinMassey 2 года назад +5

    Speaking form experience here, I recommend that you DO NOT install their packages. You will be sorry if you try to veer from their ecosystem. Also, their customer service is horrible. Our company has more than 4 of their workstations, each costing well over 8k, and both times I have tried to reach out for assistance they never responded. So, don't praise Steve, he's doing a sh** job.

    • @stephenbalaban7706
      @stephenbalaban7706 Год назад

      Hey man, really sorry to hear this. What is your support ticket #? I'm surprised that you got no response, we have an entire team dedicated to support. Always trying to make the company better, so I am personally looking into this. LMK your ticket number.
      Best,
      Stephen