https://community.jmp.com/t5/Discussions/Hyperparameters-and-neural-network-architecture/m-p/769746#M95067 <P>Hello,</P><P>I'm trying to reproduce the results obtained on JMP with the "Neural" model by adding nested cross-validation, which is not possible on the software. However, the architecture is very unclear and I can't understand the calculations performed by the model. I don't have access to certain information such as batch size, optimizer used, loss, learning rate (except the one for the boosting), and any other method or penalization used.<BR />How could I possibly have access to this information? Without it, my results obtained on JMP would be unusable as they would not be reproducible...</P><P>Thank you in advance for your reply!</P> Tue, 02 Jul 2024 12:42:09 GMT Grizzly 2024-07-02T12:42:09Z https://community.jmp.com/t5/Discussions/Hyperparameters-and-neural-network-architecture/m-p/769746#M95067 <P>Hello,</P><P>I'm trying to reproduce the results obtained on JMP with the "Neural" model by adding nested cross-validation, which is not possible on the software. However, the architecture is very unclear and I can't understand the calculations performed by the model. I don't have access to certain information such as batch size, optimizer used, loss, learning rate (except the one for the boosting), and any other method or penalization used.<BR />How could I possibly have access to this information? Without it, my results obtained on JMP would be unusable as they would not be reproducible...</P><P>Thank you in advance for your reply!</P> Tue, 02 Jul 2024 12:42:09 GMT https://community.jmp.com/t5/Discussions/Hyperparameters-and-neural-network-architecture/m-p/769746#M95067 Grizzly 2024-07-02T12:42:09Z