There are a lot of datasets available in /databricks-datasets/ that you can look through. You'll have to turn them into a table so that you can access them in automl. There are datasets associated with the spark definitive guide and learning spark 2, which will add some additional guidance.