Anytime I get a shape mismatch, the first question I try to answer is “what is the incorrect shape?” Seeing the difference between the incorrect and correct shapes should be a clue as to where to look next for the solution. Try:
print(hidden_output.shape)
and see what you get. Then the next question is how did it get that way? 
Ok sir.since im new to deep learning imconfused
Debugging is really more of a programming thing, as opposed to a Deep Learning thing. How much programming experience do you have? These courses assume you have reasonable experience in programming in general and in python in particular.
There is also some math familiarity required. Have you worked with arrays that have more than 2 dimensions before?
So what did you see when you printed the shape? Debugging is “one step at a time”. You start from the evidence and work your way backward to find the problem.