We ran a kubeflow pipeline and created an endpoint(inferenceservice), it is in ready state, but when I curl it to get predictions i get attached issue, i already saved model in proper format and loaded model from storage uri to make predictions which worked, even tensorflow runtime version for training and kserve is same. We have used same tf version(2.16.1) for training and serving.
Here is the link to the Curl command, logs and the yaml file.
Tensorflow error with Kubeflow Pipelines