Is a trained DeepSpeech AI model subject to MPL2?

So DeepSpeech has MPL2. Yet I train a TF graph with it (as if I would draw an image in Photoshop). If I use that graph from a standalone application, is my application subject to MPL2 Conditions or I am just a DeepSpeech user and thus my model is only for me to do what ever I want to (not follow the MPL2 conditions)?

BTW: if training data is part of model source and TF graph is a source related to DeepSpeach than it can be really intresting…

I think it also depends on the source of your data. @kdavis have more input to share on that topic, for sure :slight_smile:

Interesting question.

My gut feeling says you can do whatever you want with it. However, I should really ask our legal team to be sure.

I’ve run this request by our legal team and will let you know how they respond.