Release the training set and the code that was used to train the model, or stop calling it open source.
If you can't fork it and take the project in your own direction, it's not open source.
If you can't fork it and take the project in your own direction, it's not open source.
What would you have them do instead? Specifically?