Your TAO jobs, running on a remote GPU cluster.
What it does
Runs TAO training, evaluation, and inference jobs on remote GPU clusters and keeps their results on shared storage.
When to use it
Use when GPU allocation is managed by SLURM and the datasets and outputs are accessible from shared cluster storage.
How to use it
Give it a spec-bundle and cluster access; it runs the workload remotely and stores results on Lustre.
What you provide
- Your files
Uses
SL
EN
LU
Good to know
- cannot be undone: SLURM
- acts without asking
Latest review
What you need · 11
Requires a remote SLURM cluster with a login node reachable through passwordless SSH; scheduler dependencies are checked remotely.
Requires Pyxis container support on the remote cluster, not on the launch host.
Access · 5
This skill
SLURM
Execute
Starts billed GPU allocations (irreversible)
Write
Cancels remote cluster jobs
nvcr.io
Read
Pulls private container images