Jeremy Lewi

Thanks; @hamelsmu Where would we file a feature request for GitHub actions to support this? Could we file such a feature request and then use this issue to track it.

Thanks I filed a request.

We have the kubeflow github organization https://hub.docker.com/orgs/kubeflow We have the kubeflowrobot account for automating pushes to kubeflow I created a personal access token for this account. This is currently stored...

When the action triggered we got a permission error trying to push to DockerHub https://github.com/kubeflow/code-intelligence/commit/ea1e5995d7e9bc8e6326c1adf9c5149af2294ced/checks?check_suite_id=329459221

I'm going to try reproducing it locally; by logging in as the kubeflow robot to docker and building the image. I got the same error.

It looks like I had to do a couple things 1. In Docker hub I had to create the repository kubeflow/kubeflow-triage 1. I had to add the developer team with...

https://hub.docker.com/repository/docker/kubeflow/kubeflow-triage

To try to recover * Delete the gke metadata servers ``` kubectl -n kube-system delete pods -l k8s-app=gke-metadata-server ``` * Restart the label bot pods ``` kubectl delete pods -l...

Ref: keras-team/keras#5640

It looks like doing the following might fix it ``` with self._graph.as_default() as graph: with tf.Session(graph=graph) as sess: init=tf.global_variables_initializer() sess.run(init) probs = self.model.predict(x=[vec_body, vec_title]).tolist()[0] ```