domains / families / devops / tao-run-on-kubernetes

tao-run-on-kubernetes

separate

Kubernetes execution platform — submits TAO container jobs as k8s Jobs with NVIDIA GPU scheduling; single-pod

When to use

• submitting TAO container jobs as k8s Jobs
• NVIDIA GPU scheduling
• single-pod training on one node
• multi-node distributed training via Indexed Jobs
• running on EKS, GKE, AKS, or on-prem clusters
• integrating TAO into k8s-native ML platforms

When not to use

• Brev or SLURM credential management
• using nvidia-tao-sdk or tao_sdk imports

Skill page

Source