Does SageMaker AutoGluon support train using multiple CSV in a S3 prefix

Hi,

I have running into the issue, by following using the code example here https://docs.aws.amazon.com/sagemaker/latest/dg/autogluon-tabular.html.

RuntimeError: Found more than 1 file ending in '.csv' under '/opt/ml/input/data/training'. Could not resolve ambiguity on which file to use between: 'part-00067-041d87d1-f3b5-4d3a-926e-b599c0b191a8-c000.csv, part-00123-041d87d1-f3b5-4d3a-926e-b599c0b191a8-c000.csv'. Please update the objects in your S3 buckets to have only one '.csv' file and try again.

But in the documentation, is says: If multiple CSV files are provided for the training or validation channels, the AutoGluon-Tabular algorithm concatenates the files.

Does anyone know how to resolve this?