Closed DrJonnyT closed 1 year ago
Hey @DrJonnyT! All dependencies except for the linear-algebra backend should be installed automatically.
If you want to use the package with PyTorch, you will need to additionally install PyTorch (pip install torch
). If you want to use the package with TensorFlow, you will need to additionally install TensorFlow and TF Probability (pip install tensorflow tensorflow-probability
).
The reason why PyTorch and TF aren't automatically installed is that you can choose which one to install: users of PyTorch shouldn't be forced to install TF and vice versa.
I'm sorry if this wasn't clear!
Aha cool, I did wonder if it was deliberate. Maybe it would be useful to say that in the readme? Cheers!
@DrJonnyT I've now written this explicitly in the README. Thank you for the suggestion! :)
If I run this: conda create -n neuralproc conda activate neuralproc pip install neuralprocesses
I end up with neuralprocesses installed but dependencies like tensorflow or pytorch are missing. Should they be installed?