Select training data from specific files in one folder


I want to train two models with nearly the same NLU data. All the data is in the same folder in separate files. So one model should use file A, file B and the other model file A, file B and file C. Is there a way to select the separate files while training or do I have to copy / paste the data in a separate folder for the second model?

I was think of something like this for the first model:

rasa train --data data/en/{file A, file B} -c config_en.yml -d domain_en.yml --out models_en_1

and for the second model:

rasa train --data data/en/{file A, file B, file C} -c config_en.yml -d domain_en.yml --out models_en_2

Thank you!

Hi Theresa,

You could create a directory for each combination and setup softlinks to each of the files. You could also write a custom training importer.


@threxx you can also just pass the individual files like this --data data/en/fileA data/en/fileB

1 Like

Thank you! That’s working for me