Closed bertsky closed 1 year ago
Thanks for reporting. It seems nvidia-nccl-cu12 2.17 has not been published to Pypi yet. I'll look into why that hasn't happened yet.
I just noticed there's a discrepancy between the Py index and the Github release – the latter still has 1.15.5+nv23.02 as tip.
(Perhaps someone forgot to push the 23.03 branch?)
@bertsky the 23.03 branch has now been published. Just waiting on NCCL wheels now...
It worked – thanks!
System information
Describe the problem
When installing via pip:
pip install nvidia-pyindex && pip install nvidia_tensorflow
I end up with:
The problem goes away when downgrading to the previous release,
nvidia_tensorflow==1.15.5+nv23.2
Provide the exact sequence of commands / steps that you executed before running into the problem
see above
Any other info / logs
Environment is a Docker container based on nvidia/cuda:11.3.1-cudnn8-runtime-ubuntu20.04