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
To implement the faster version of RAFT, if I change the default value of --corr_implementation to reg_cuda, I get the error of "name 'corr_sampler' is not defined".
The text was updated successfully, but these errors were encountered:
I met a compatability issue
The detected CUDA version (11.0) mismatches the version that was used to compile
PyTorch (11.7). Please make sure to use the same CUDA versions.
Is it possible to get the a python version of this file please, so that it would be more version compatible?
To implement the faster version of RAFT, if I change the default value of --corr_implementation to reg_cuda, I get the error of "name 'corr_sampler' is not defined".
The text was updated successfully, but these errors were encountered: