Why is Tensorflow so hard to install?

I'm trying to do some machine learning in Jupyter Notebooks in VSCode, and I always keep getting a ModuleNotFoundError: No module named 'tensorflow.' I did everything, from installing Anaconda to changing my PATH variables to pip install tensorflow, but nothing's working. I'm spending all this time just to install a stupid library. Any tips?

I'm using Windows 10 btw.

Shopping Cart Returner Shirt $21.68

The Kind of Tired That Sleep Won’t Fix Shirt $21.68

Shopping Cart Returner Shirt $21.68

  1. 2 weeks ago
    Anonymous

    Have you tried restarting your computer?

    • 2 weeks ago
      Anonymous

      still doesn't work after rebooting

  2. 2 weeks ago
    Anonymous

    >anaconda
    >windows
    >notebooks
    >tips
    >a single thread for this baby spoonfeeding
    people like you deserve poverty

  3. 2 weeks ago
    Anonymous

    >Windows 10
    here's the problem

  4. 2 weeks ago
    Anonymous

    uninstall anaconda
    choose the correct one: https://code.visualstudio.com/docs/datascience/jupyter-kernel-management#_jupyter-kernels

    • 2 weeks ago
      Anonymous

      >uninstall anaconda
      >choose the correct one: https://code.visualstudio.com/docs/datascience/jupyter-kernel-management#_jupyter-kernels
      this
      and use fedora or ubuntu

    • 2 weeks ago
      Anonymous

      but which kernel should i select?

      • 2 weeks ago
        Anonymous

        just uninstall anaconda and install python normally
        there should only be one kernel

        • 2 weeks ago
          Anonymous

          i tried the default kernel
          still doesn't work

          • 2 weeks ago
            Anonymous

            >default kernel
            no such thing. uninstall anaconda.

          • 2 weeks ago
            Anonymous

            that's what i did

  5. 2 weeks ago
    Anonymous

    >Anaconda
    absolute useless trash. just look at this shit. why does this shit need 7 different chief officers???
    just install python, install tensorflow with pip, and you are good to go.

    • 2 weeks ago
      Anonymous

      imagine the smell

      • 2 weeks ago
        Anonymous

        nitin in arabic means stinky

    • 2 weeks ago
      Anonymous

      Condaslop is made obsolete by python -m venv venv

      • 2 weeks ago
        Anonymous

        um anon, that's too complex for me, please understand.

      • 2 weeks ago
        Anonymous

        until you update your system and all your venvs shit the bed

    • 2 weeks ago
      Anonymous

      spot the israelite

      • 2 weeks ago
        Anonymous

        Hoch~~*man*~~

      • 2 weeks ago
        Anonymous

        He looks like a goblin, it's easy to spot him

    • 2 weeks ago
      Anonymous

      I want Jessica to spit in my face, shove me on the floor, and step on me like a doormat

  6. 2 weeks ago
    Anonymous

    all pyshit is nightmare fuel. you use a container and pray it contains the appropriate userland for your problem.

  7. 2 weeks ago
    Anonymous

    Click on top right and select the kernel that has tensorflow, also run the install command in the exact terminal that opens when you press Ctrl+Shift+~ because it'll use the environment the debugger uses. Also uninstall condaslop and use venv like a sane person

    • 2 weeks ago
      Anonymous

      none of my kernels have tensorflow

      • 2 weeks ago
        Anonymous

        and yes, i did uninstall anaconda and switched to venv
        still not working

        • 2 weeks ago
          Anonymous

          provide me with the chronological sequence of commands you're running, and the necessary import code and I'll do the same on my machine

  8. 2 weeks ago
    Anonymous

    Are you using the command line prompt that comes in the Anaconda start menu? It has all the environment variables set up for you. Use pip install from there.

    In VSCode use the anaconda kernel.
    You shouldn't need to change the PATH, VSCode detected my Anaconda install automatically. Did you use some custom path when installing Anaconda? For me it's installed by default on my user directory.

  9. 2 weeks ago
    Anonymous

    I did my thesis using tf 4 years ago, properly installing it and setting up the right CUDA version to go along with your tf version took me on average a day per install. https://www.tensorflow.org/install/pip this has the basics but from what i see they are cucking the windows users by not supporting windows from 2.10 onwards.

    • 2 weeks ago
      Anonymous

      >CUDA is the only GPU choice
      grim.

  10. 2 weeks ago
    Anonymous

    Consider Tinygrad

  11. 2 weeks ago
    Anonymous

    ditch anaconda, create a venv, install tensorflow in it, select the venv as an explicit python kernel in jupyter notebooks and it should work
    for gpu support, if you did all the logical things and it can't find it, check for all relevant env variables being set, tensorflow seems to rely on some of them, CUDNN_PATH or something

  12. 2 weeks ago
    Anonymous

    op here
    it works finally
    i kept anaconda

  13. 2 weeks ago
    Anonymous

    Ditching anaconda is bad advice. Anaconda is cancerous but it's guaranteed to make your tensorshit go fast. Tensorflow is unironically built for BBA (Big Bloated Anaconda). It's also not that hard to use, just don't let it dickpuppet your bashrc or whatever.

    • 2 weeks ago
      Anonymous

      Imagine having a Turing complete dependency solver

      • 2 weeks ago
        Anonymous

        With the power of Python, you don't need to imagine

      • 2 weeks ago
        Anonymous

        >Turing complete dependency solver
        isn't that basically required always?
        pretty sure all the nonshit depsolvers are basically a SAT solver.

  14. 2 weeks ago
    Anonymous

    To get Tensorflow to work properly you have to install something called PyTorch.

  15. 2 weeks ago
    Anonymous

    Did you activate the conda environment?

  16. 2 weeks ago
    Anonymous

    >I'm using Windows 10 btw
    You answered your own question.
    On Linux it installs jist fine.

    Anything outside of Microsofts domain is generally a bad experience in terms of deployment.

  17. 2 weeks ago
    Anonymous

    dumbass zoomer

  18. 2 weeks ago
    Anonymous

    Windows with WSL just werks. GPU accelerated if you have NVIDIA.

    • 2 weeks ago
      Anonymous

      Don't listen to that guy. Wsl id useless for ai.

      • 2 weeks ago
        Anonymous

        Cope

  19. 2 weeks ago
    Anonymous

    >I'm trying to do some machine learning in Jupyter Notebooks in VSCode, and I always keep getting a ModuleNotFoundError: No module named 'tensorflow.'
    >python cringe webshittery
    >Visual Studio notepad edition
    >can't into fricking managing system python
    You're a dumbass who probably can't install Win 95 from boot floppies, install Gentoo from a zip drive, or write a character device driver for Linux in C.
    It's hard to install because all your ignorant ass knows is how to click next->agree->next->next->finish in InstallShield.

Your email address will not be published. Required fields are marked *