C5 W1 A1 lstm_cell_backward, issue with dc_prev calculation

Course Q&A Deep Learning Specialization Sequence Models
1

Hello Friends,
I have written code for all the variables based on the given equations.
Values for all of the variables are correct, except gradients[“dc_prev”][2][3]

My output is:
gradients[“dc_prev”][2][3] = 0.0007137695556389628

Expected output is:
gradients[“dc_prev”][2][3] = 0.797522038797

I have written the code for dc_prev exactly as the given equation 20.
For some reason, only this output is incorrect and rest of them are correct.
Any suggestion on what to look at?

Any suggestions are much appreciated.

3

Hey @Suhas_K_R,
In the mathematical equation for dc_prev, neither I can see the use of dft nor the use of dot, still in your implementation, I can find both. Re-check your implementation for dc_prev against the mathematical equation once. I hope this resolves your error.

Cheers,
Elemento

4

Sure, let me check it.

Thanks,
Suhas