Stacked LSTM is not improving

#1

Even after using stacked lstm, my accuracy is 32%. What to do? Am I doing something wrong?

14bafb50-cf48-4394-ba8d-31da5e2b68a2.png1366×768

#2

Deep learning is a section of experiments and it require a lot of them to make good models.
so , may be you need to work on model and try agin.

Can you share me your model screenshot too.

#3

Sure!

Screenshot (35)
Screenshot (35)1366×768 91.4 KB

#4

some thoughts:

  1. lower the values of dropouts. 0.5 means you are trying to drop 50% data , and that is very high value and you need to work on it too.

  2. Try changing optimizer ( i hope that would work , not sure on this though )

  3. longer training with reduced learning rate

Try them once.