How can deploy flyte workflows to azure GPU machine . The documentation is not clear to me

@curved-kilobyte-59985 sorry for the lack of clarity. The essential idea is that you should use 1. Taints/labels in your GPU nodes 2. Set tolerations/nodeAffinity rules using a PodTemplate or directly a PodSpec inside your Task I'm not sure who in the community has done so on Azure, but from the K8s point of view the process should be similar regardless of cloud GPU selectors, a recent Flyte feature, also let you request particular devices or partitions: https://docs.flyte.org/en/latest/api/flytekit/extras.accelerators.html#specifying-accelerators Does that help?

Thanks i solved it .

Anything particular to Azure you had to do to make it work? It could be useful info for others

No i changed to GCP .