Eric Chen

Currently it seems that spconv does not work with torch.compile, for me the first sparse convolution operation raised some error. Is there any chance that spconv can take torch.compile compatibility...

Thanks for the awesome job! Currently nr3d_lib supports float64, float32 and float16, I think bfloat16 support can be introduced without too much effort and for my experience it can stabilize...

Taming gs for faster training and rendering

5

[Taming 3DGS](https://arxiv.org/abs/2406.15643) released an optimized 3dgs implementation which seems to be faster than the current gsplat. The released code mostly related to section 4 of their paper. Maybe similar strategy...