Thanks for your question!
Under the "Predict" tab for a model, you can compute and download a CSV file of the training predictions for "All data," which will give you out-of-sample predictions for every row in the dataset you uploaded. Alongside the out-of-sample predictions, there's also a column called "Partition" and "row_id." These denote what row was used in each partition. You can use these things to replicate the DataRobot partitions.