setup.py doesnt specify torch as dependency leading to downstream failures
KukumavMozolo opened this issue · comments
Hi there,
is there any chance that you will add torch as a requirement to setup.py or adopt PEP 517?
I am running into problems installing this package while trying to install it using pip-tools as I described here:
jazzband/pip-tools#1662 (comment)
Likely no, since this will result in a lot of issues regarding compilation - if we require torch in setup.py, it will install PyTorch with CUDA 10.2 which might not match your local CUDA version. In addition, you will need to have PyTorch installed in the first place to install this package.
Thank you for your reply,
I am not really knowledgeable on that matter but this https://github.com/huggingface/transformers project also depends on torch and I do not recall running into similar problems installing it via pip-tools.
Maybe there is a solution there?
Well, that‘s because transformers is Python only - it does not need to compile CUDA kernels.
This issue had no activity for 6 months. It will be closed in 2 weeks unless there is some new activity. Is this issue already resolved?