You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Unfortunately, torchlars is only supported for machines with CUDA GPUs. It is specifically designed to speed up the computations in LARS optimizer by combining them into one GPU kernel.
Thank you for your amazing work!
That is understood, but my situation requires installing (not using) the package on a server with no GPU.
If you had a .whl file, or dropped the GPU check on compilation I could just compile it with CUDA. Similar to how one can install the GPU-enabled version of pytorch even without using the GPU functions.
Is it possible to pre-compile the package (make a wheel) for this repository?
I need to install it on a computer without GPUs and it won't let me.
The text was updated successfully, but these errors were encountered: