I thought the dataset address was referenced by the dataset name not the clearml.conf? Like the dataset is getting pulled in the code by this command

data_path = Dataset.get(dataset_name="Fashion MNIST", alias="Fashion MNIST").get_local_copy()

So I would assume the code would use the clearml.conf of the running environment?

My local clearml config on my laptop is set to this

This is the issue. Your clearml.conf shouldn't be pointing to localhost but to the address of the server

My local clearml config on my laptop is set to this

api {
  web_server: 

  api_server: 

  files_server: 

  credentials {
    "access_key" = "***"
    "secret_key" = "***"
  }
}

The data is in my laptop in a folder called fashion_mnist .
The contents of my clearml.conf are produced from the values passed to the helm chart as such

agentk8sglue:
  apiServerUrlReference:  "

"
  fileServerUrlReference: "

"
  webServerUrlReference:  "

"
  extraEnvs:
    - name: CLEARML_AGENT_UPDATE_VERSION
      value: "==1.9.2"
    - name: CLEARML_API_HOST
      value: "

"
    - name: CLEARML_FILES_HOST
      value: "

"
    - name: CLEARML_WEB_HOST
      value: "

"

The dataset was created like this

clearml-data create --project "Full Overview" --name "Fashion MNIST"
clearml-data add --files fashion_mnist
clearml-data close

I didn't set the output uri, I think that was created by the cli or something else.

Hi @<1840924578885406720:profile|VictoriousFish46> , how are you uploading the dataset? Did you set output_uri? What is set as your files server in the api section of your clearml.conf ?