The MPI Operator makes it easy to run allreduce-style distributed training on Kubernetes. Please check out this blog postarrow-up-right for an introduction to MPI Operator and its industry adoption - https://www.kubeflow.org/docs/components/training/mpi/arrow-up-right
To install the MPI operator simply type:
k3ai apply mpi-op
Last updated 5 years ago
Was this helpful?