Why do we have to squeeze our predictions?

[ashikshafi08]

The reason is pretty straightforward, our y_test is of the shape (10, ) whereas our y_preds the shape is (10, 1).

Using tf.squeeze() will get rid of the extra 1 dimension in our tensor, for us, y_preds have an extra dimension so we get rid of it.

We do this because while calculating evaluation metrics we gotta make sure both y_preds and y_test should be in the same shape.

Hope it helps.