Training and evaluation
Use a cluster training result
A run on your own cluster keeps its output in your administrator's output volume, under the run ID. Studio does not receive the weights.
Get the weights#
Find the run ID in Kubernetes clusters. Ask your operator to load the weights from the output volume into your company's inference runtime, and to evaluate them before use.
What Studio shows#
Studio receives the run status and optional numeric metrics. It cannot download or deploy private cluster weights. The model reference identifies the run; it is not a public inference endpoint.