Thanks for this. Some users might want to keep in mind that there are two branches for the git repo probably a bit different for pytorch and tensorflow. Therefore check before cloning. PS: Just ordered the textbook. Looking forward to learning from it :)
Thank you! this got me started but then I kept getting a syntax error when I tried to install Jupyter notebook. I couldn't find the solution on Google so I finally gave ChatGPT a try. I gave it the prompt and copy and pasted the errors I was getting and it walked me through several reasons why it may be happening with the solutions. Essentially, I installed a new environment and installed Jupyter in the new environment and updated Conda. I am very new to all of this so it was like having a personal tutor next to me!
Thank you, this was so helpful! It worked perfectly :) Just one suggestion, you could update the version of python installed from 3.9 to to 3.10 in the .yml file in your git repo.
Thanks a lot Professor. If I may ask this question: I am going for a machine learning laptop and juggling between the choice of a MACbook pro or a Windows GPU laptop. Want a macbook, if it is as good for deep learning projects. Is it? Thanks.
After cloning the repo when I'm trying to run this command I'm getting this, why is that? "SpecNotFound: Invalid name 'torch', try the format: user/package"
please help.. I have done all these operations. also my code is suitable for gpu but model still takes too long to run. And no result is returned. One mistake or one score. Why could it be? I have macbook pro m2 processor.
So I tried to use full fat anaconda to intall pytorch with the metal API and it just didn't work, however when I switched over to a fresh mini conda installation it worked fine.
I am trying to create this and my pip.conf file has some problem. then I have to fix that first. But when I come back when I install this again, I experienced some error. it might be better to tell how to quite the failed install and retry it. currently I am simply try to delete the env and redo it one more time, not sure whether it is a good solution or not.
For torch.cuda.device_count() it says 0 and torch.cuda.is_available() it says false, the model that I want to use need GPU to train and it checks these commands, I tried running it on tensorflow env too ( the one that you taught how to create). what should I do?
Hi Professor, I installed the pytorch as you instruct. But I cannot rename in the jupyter notebook. The error shows: "NameError: name 'os_path' is not defined". Can you help me with this?
python -m ipykernel install --user --name pytorch --display-name "Python 3.9.6 (torch)" does not work. "/Library/Developer/CommandLineTools/usr/bin/python3: No module named ipykernel" is shown. I tried to install ipykernel, no success
Getting a message, although not an error ~ conda activate torch path/miniconda3/envs/torch/etc/conda/activate.d/libxml2_activate.sh:export:12: not valid in this context: XML_CATALOG_FILES+
Learning directly from Jeff Heaton himself, awesome!
Thanks for this. Some users might want to keep in mind that there are two branches for the git repo probably a bit different for pytorch and tensorflow. Therefore check before cloning. PS: Just ordered the textbook. Looking forward to learning from it :)
Thank you! this got me started but then I kept getting a syntax error when I tried to install Jupyter notebook. I couldn't find the solution on Google so I finally gave ChatGPT a try. I gave it the prompt and copy and pasted the errors I was getting and it walked me through several reasons why it may be happening with the solutions. Essentially, I installed a new environment and installed Jupyter in the new environment and updated Conda. I am very new to all of this so it was like having a personal tutor next to me!
as usual, life saver for all of us mac users
Thank you very much! Finally a video straight to the point and clearly explains everything. So helpful. Really Appreciate it.
Very helpful. The most helpful resource I found online on this topic.
Once again, thanks Jeff !
I don't know how we could do without you !
Cheers
Thanks a lot. It helped me to set up a torch env on my M1 Mac. It drove me crazy.
Wow Clear instruction. Thanks a lot Jeff
Hi Thanks a lot. This command worked for me: conda install pytorch::pytorch torchvision torchaudio -c pytorch
thanks - great video - very little out there to set up for M1 macs that is simple and straight forward
Thanks.
With this it was easy peasy!
Thank you, this was so helpful! It worked perfectly :)
Just one suggestion, you could update the version of python installed from 3.9 to to 3.10 in the .yml file in your git repo.
You're doing the lord's work, thanks
Thanks a lot Professor.
If I may ask this question: I am going for a machine learning laptop and juggling between the choice of a MACbook pro or a Windows GPU laptop. Want a macbook, if it is as good for deep learning projects. Is it? Thanks.
Thank you so much, that was really helpful
thx professior, ran the first ai program on my mac!
Thank you Porfessor, this really helps!
Hello professor, How can I install pytorch and cv2 in the same environment which contains tensorflow?
Thank you for the easy to follow instructions.
Have you figured out how to do quantization for M1 like what the bitsandbytes pypi module does? Cheers!
Repository shared in the description isn't available
Which is good according to you sir MacBook Air M1 or any windows laptop for programming
Jeff thank you for all ı got it
Thank you professor, could you give a suggestion about how much ram is needed for ML or DL, for apple slicon 8, 16, 32 or even 64?
Thanks so much for doing this!
After cloning the repo when I'm trying to run this command I'm getting this, why is that? "SpecNotFound: Invalid name 'torch', try the format: user/package"
check that you are loading the correct ( .yml) file in the root of your project.
great video professor!
Hello, how to make mac m1 use more gpu when training, my mac only use 15-20% gpu even i used coreml
please help.. I have done all these operations. also my code is suitable for gpu but model still takes too long to run. And no result is returned. One mistake or one score. Why could it be? I have macbook pro m2 processor.
Can you add egpu RTX 30 or 40 series and it showing GPU available ?
So I tried to use full fat anaconda to intall pytorch with the metal API and it just didn't work, however when I switched over to a fresh mini conda installation it worked fine.
Hi there,
Thnaks for the video!
has_mps is deprecated.
Also, is there any workaround to use CUDA on a mac M1 ?
Thanks !
not install in my macbook pro m1 max there show msg Mac OS not compatibility. what can i do?
By the way, Anaconda has been update for ARM; so I'm not sure that there's still a reason to use Miniconda.
Can I just use the pip to install PyTorch on MacBook Air with M2?
I am trying to create this and my pip.conf file has some problem. then I have to fix that first. But when I come back when I install this again, I experienced some error. it might be better to tell how to quite the failed install and retry it. currently I am simply try to delete the env and redo it one more time, not sure whether it is a good solution or not.
Thank you for this good tutorial, but I can't find the yml file for the PyTorch installation
me too. where is .yml file for the PyTorch installation?
im still unclear, does this mean i can use libraries written for nvidia gpu on my m1 mac?
Hello Professor, I'm trying to install mini conda3 on macOS Ventura 13.1 but I get an error (not supported). but anaconda can run well
For torch.cuda.device_count() it says 0 and torch.cuda.is_available() it says false, the model that I want to use need GPU to train and it checks these commands, I tried running it on tensorflow env too ( the one that you taught how to create). what should I do?
Why can't I find "torch-conda-nightly.yml" file
I don't see it either. WIll try with torch-conda.yml from the repo.
Hi Professor, I installed the pytorch as you instruct. But I cannot rename in the jupyter notebook. The error shows: "NameError: name 'os_path' is not defined". Can you help me with this?
Is there any downside in combining both your tensorflow and PyTorch into a single tank file and environment ? Thank you!
Don't know, but thinking about doing the same.
yup i need the same. @thomasengland1701 @@stack-dhruv
do you have videos on how we can use the M2 on Mac, with PyTorch?
I think it's a similar procedure for the M2
At 1:49 : the clean up removed python but when I hit python3 , its still there
(I'm referring to the video at 5:50)
Thanks!
Hi professor, will this work on miniforge as well? Thank you.
Tell me your questions and I'll answer them
Where can I access the yml file now. It is not available at the given link.
Sorry about that, link updated.
by this will yolo use gpu in m1 mac?
Hi, where is the .yml file?
same issue
download his git repo rather than cloning you will find yml file in the install directory then.
I do not see "torch-conda-nightly.yml " anywhere.
same here
see the comment below your from Phil Hamilton
Professor， I can not find this ipynb in your Github. Do you delete that?
Link in the description, and here: github.com/jeffheaton/t81_558_deep_learning/blob/pytorch/install/pytorch-install-aug-2022.ipynb
where is the yml link?
Is there any chance someone could share the .yml file? I can't seem to find it in the repo
did you end up finding the yml?
wheres the yac file?
You are my angel
how do you run the yml file?
i couldn't do that either, could you figure out?
Yah, Wash U!
You just wasted my time I can't find the yml file and everyone seems to be having the same issue.
python -m ipykernel install --user --name pytorch --display-name "Python 3.9.6 (torch)" does not work. "/Library/Developer/CommandLineTools/usr/bin/python3: No module named ipykernel" is shown. I tried to install ipykernel, no success
Getting a message, although not an error
~ conda activate torch
path/miniconda3/envs/torch/etc/conda/activate.d/libxml2_activate.sh:export:12: not valid in this context: XML_CATALOG_FILES+
Thank you!
Where is the yml file link ?
Found it
@@kelsensantos where?
@@kelsensantos can you share us where it is?
where the fuck is it stop ghosting Belsen