Hi, so when I use the Network.fromJSON function to copy a myNetwork.toJSON result and then RUN the new network with the same input as the original one, the result is different. I also noticed that when I DON'T train the original the outputs DO become the same. Weird. Please help!
EDIT: oh and also I tested making multiple copies not just one and they are all slightly different in their results. Even when I clear them since this is an LSTM
Lines Of Interest In The Code Below
The nn and nn2 activate functions for checking the result:
Issue
Hi, so when I use the Network.fromJSON function to copy a myNetwork.toJSON result and then RUN the new network with the same input as the original one, the result is different. I also noticed that when I DON'T train the original the outputs DO become the same. Weird. Please help!
EDIT: oh and also I tested making multiple copies not just one and they are all slightly different in their results. Even when I clear them since this is an LSTM
Lines Of Interest In The Code Below
The nn and nn2 activate functions for checking the result:
The training function:
Code