In the practice lab C2_W1_Lab02_CoffeeRoasting_TF, for tensor flow model, shape the argument to tf.keras.Input is given as (shape=(2,)). Which means, it is being assumed that the expected shape of input shall be 2 rows and any number of columns. I am not able to get this. The shape of input is (200000,2), which means, it should have 2 columns. Each individual row in this matrix will be having 2 columns. So why are we training the Tensorflow for a shape of (2,0)?

Hello @AJAY_KUMAR_AGRAWAL, welcome to our community!

If you also check out the doc, the shape we put there won’t include the batch (sample) size which is the number of row in your case, instead we are only telling Tensorflow what the number of column is, or better the shape of each sample.

Raymond