Open M-Artemisia opened 7 months ago
Hi, I tried to execute this sample on a local k8s. The KFP version on my cluster is more than 2.0, so I have updated the digit-recognizer code to be compatible with kfp-sdk-v2.
After applying the changes and running the code on kubeflow, all of the pipeline steps are done successfully. But the problem is the created digit-recognizer inference-service on Kubernetes returns an Internal Error complaining about:
No runtime found to support predictor with model type: {tensorflow <nil>}
Any idea about this issue?
Thanks in advance,
Resolved! Nothing was wrong with the pipeline. There was an issue with the ClusterRuntimes on the k8s.
Hi, I tried to execute this sample on a local k8s. The KFP version on my cluster is more than 2.0, so I have updated the digit-recognizer code to be compatible with kfp-sdk-v2.
After applying the changes and running the code on kubeflow, all of the pipeline steps are done successfully. But the problem is the created digit-recognizer inference-service on Kubernetes returns an Internal Error complaining about:
No runtime found to support predictor with model type: {tensorflow <nil>}
Any idea about this issue?
Thanks in advance,