CUDA Torch incompatibility with RTX 5090

[adityadutt09]

I run into an incompatibility issue when running the first cell block in the Jupyter notebook. I am using an RTX 5090 which requires the nightly install of pyTorch.

Before i run down the rabbit hole of modifying the Dockerfile, I wanted to check if the set versions here can be modified?

Thanks!

[GavinYang5]

I have the same problem, have you solved it?

[adityanjothi]

Hey @adityadutt09 apologies for the delay in getting back to you, the new v0.2.0 release supports Blackwell and should support sm_120. Could you try with the latest release?