Hi, thanks for your great work. There are some explanations in the appendix of the paper regarding the FLOPs calculation. Would it be possible to release the code you used for that? Directly using the fvcore results in some unsupported operations, unfortunately.
Hi, thanks for your great work. There are some explanations in the appendix of the paper regarding the FLOPs calculation. Would it be possible to release the code you used for that? Directly using the fvcore results in some unsupported operations, unfortunately.