Hello!
I wanted clarification on this scenario where a model is available with an open license but it was pretrained on a dataset which is not open. The pretraining was not done by anyone in my team.
Would it be okay to use this model’s weights for the competition?
Thanks for your help 
@quantumgeek Yes that is fine. Thanks for checking.
@quantumgeek A team member just pointed out to me that the question has some ambiguity. I assumed you meant the common case where a trained model (i.e. weights) are public but the training data is not.
If the pretrained model weights are available under an open license, then this is fine.
But if it’s just that the model source code is available under an open license but then it was pretrained on closed data, then this would not be allowed.
Thank you for further clarifying! Yes, I meant to ask if the model code + weights have open license.
Cheers