I am not sure if I understood the explanation in the “Using per-item features” video of how the values of the X and b matrices are calculated.
If I am not mistaken, in the dataset of the assigment C3_W2 these values are already calculated but I don’t understand where they come from.
Thanks for your help.
In Practice Lab 1, the “gradient tape” method is used to learn the weights and biases. It’s essentially gradient descent and linear regression, but using TensorFlow.
In Practice Lab 2, it uses scikit-learn’s Standard Scalar method. Again, it’s essentially linear regression.
Hi @TMosh
In practice C3_W2 this data seems to come from the dataset. Or am I misunderstanding something?
As it says there in the note just above the red box, those pre-loaded weights are just so you have some data for some preliminary examples. You’ll train your own weights later in the notebook.
Thank you.
Now it is clearer to me.