I am setting up a YAML template for Katib, which will follow the same optimizing script and process but with different inputs, so I would like to get the file from a bucket given the path to the file itself.
I used it on a different pipeline with Argo Workflows and it was added like this inside the container parameter:
inputs:
artifacts:
- gcs:
bucket: 'INPUT_BUCKET_NAME'
key: 'FILE_NAME'
name: my-art
path: /my-artifact
parameters:
- name: Bucket
value: 'INPUT_BUCKET_NAME'
- name: File
value: 'FILE_NAME'
But the approach seems different in Katib with Kubeflow