Implemented the backpropagation as well as the gradient descent from scratch in python. Followed Karpathy sensei's lecture throughout to make this.
Ps: The drop used here is the equivalent to tensor in pytorch.
I may change it more in the future.