Weight count on C2_W1_Lab02_CoffeeRoasting_TF lab

[myrium]

Hi there! I’m doing the C2_W1_Lab02_CoffeeRoasting_TF lab and got confused when reading about the parameter counts.

So I now the data has 2 features and we will have 2 layers, one with 3 and another with 1 neurons.

When printing the weights, why is a 2x3 matrix for the first neuron instead of 3x2?
At first I imagined it was to follow the wx + b format but the biases are already being printed separately.

Have I missed something?

[Alireza_Saei]

Hi @myrium

@myrium

So I now the data has 2 features and we will have 2 layers, one with 3 and another with 1 neurons

Number of features in your dataset and the number of layers in a neural network are independant!

@myrium

When printing the weights, why is a 2x3 matrix for the first neuron instead of 3x2?

It depends on your implementation, in these courses you must use 23 matrices, but implementing a code that uses 32 is not wrong!

Hope it helps! Feel free to ask if you need further assistance.