Installation Error,Packages not found

[eric31512]

I don't know how to fix it

C:\Users\User>conda install pytorch==1.4.0 torchvision==0.5.0 cudatoolkit=10.0 -c pytorch -c conda-forge
Collecting package metadata (current_repodata.json): done
Solving environment: failed with initial frozen solve. Retrying with flexible solve.
Collecting package metadata (repodata.json): done
Solving environment: failed with initial frozen solve. Retrying with flexible solve.

PackagesNotFoundError: The following packages are not available from current channels:

• torchvision==0.5.0

Current channels:

• https://conda.anaconda.org/pytorch/win-64
• https://conda.anaconda.org/pytorch/noarch
• https://conda.anaconda.org/conda-forge/win-64
• https://conda.anaconda.org/conda-forge/noarch
• https://repo.anaconda.com/pkgs/main/win-64
• https://repo.anaconda.com/pkgs/main/noarch
• https://repo.anaconda.com/pkgs/r/win-64
• https://repo.anaconda.com/pkgs/r/noarch
• https://repo.anaconda.com/pkgs/msys2/win-64
• https://repo.anaconda.com/pkgs/msys2/noarch

To search for alternate channels that may provide the conda package you're
looking for, navigate to

https://anaconda.org

and use the search bar at the top of the page.

[mr-sudo]

Hey any luck? I'm experiencing the same thing

[cyberlife64]

Same problem with me, you got it working?

[mr-sudo]

No luck on my end 🫠

[mr-sudo]

No luck on my end 🫠, I ended up having trouble with running due to the specs on my machine in general, although one thing that seemed to help progress things a bit was trying to install these packages through pip instead

[Panos-Jr]

Installing with pip is a possible option, have you considered conda update --all? It could possibly work, I’m not sure.

[cyberlife64]

Installing with pip is a possible option, have you considered conda update --all? It could possibly work, I’m not sure.

Updating doesn't work as well, I tried that already

[carlosaln]

Here's what initially worked for me:

pip install torch===1.4.0 torchvision===0.5.0 -f https://download.pytorch.org/whl/torch_stable.html
conda install cudatoolkit=10.0

However, this installs old versions of the torch libraries, which are incompatible with the latest Nvidia NCCL library, so you'll just get an error further down the road. What I ended up doing (which works) is I installed the latest libaries for everything. Taking the instructions in the main README, that would look something like:

# Required: Sampling
conda create --name jukebox python=3.11.8
conda activate jukebox
pip3 install mpi4py
pip3 install torch torchvision torchaudio
conda install cudatoolkit
git clone https://github.com/openai/jukebox.git
cd jukebox
pip install -r requirements.txt
pip install -e .

# Required: Training
conda install av -c conda-forge 
pip install ./tensorboardX

Note I am not a AI/ML expert but do know my way around a terminal, and this worked. Perhaps I could have used conda/pip for everything, but that's just how I ended up doing things. Your results may vary.