Variable length lists to a single tensor without padding

Will pytorch support dynamic length in tensors
I would like to convert A=[[1,2,3,4,5],[3,6,5],[7,1,3,5,3,7,6]] to a tensor without changing this lengths
Is it possible to do in pytorch

Thanks in Advance

pytorch/nestedtensor has a prototype of this functionality so you could check it out.

I tried to install a nested tensor and got this error
ERROR: nestedtensor-0.1.1_cu102-cp37-cp37m-linux_x86_64.whl is not a supported wheel on this platform.