Dear all,
I am having a problem training my model. My error metric remains the same throughout. I used the @gcheck to confirm if the gradient calculation is correct but the test give me a false. I am not sure what to do next? It indicates that the problem is at the very first layer.
I am using the following command to train the model:
adam!(loss, [(SlotAttentionModel, batch)], params=params(SlotAttentionModel), lr = 0.0004
macro gcheck1(ex); esc(:(@gcheck $ex (delta=0.000001, nsample=2, rtol=0.05, atol=0.001, verbose=2))); end
@gcheck1 loss(SlotAttentionModel, first(clevrDataset))
Any suggestions?
Best Regards,
Ahmed Imam ShahMS. Computer Science and Engineering
Koç University, Istanbul, Turkey