Even though I do Kubernetes support and applications development, knowing about NGC is key in the HPC/ML world. For example, NVIDIA maintains their own proprietary PyTorch container image that is tuned into all of their GPUs. Everyone from the ML world seems to already know about (and constantly use) NGC. Unfortunately, the PyTorch image doesn’t work as is within JupyterHub in Kubernetes, the focal point of my initial projects here on this K8S infrastructure team.
Related:
Tags:
#ckad #nvidia #ml #hpc