Tensorflow prompts typeerror: unsupported operand type (s) for *: 'float' and 'nonetype‘

The error code is as follows ：

, The variables are declared as follows ：

This is because the wrong type of tensor is used , stay TensorFlow Medium tensor When it comes to finding gradient and updating gradient , Need to use special tensor：Variable, About to be the original tensor Direct force to Variable that will do .

in addition , If the gradient descent algorithm is implemented manually ,tensorflow You cannot use the direct subtraction method of the first screenshot in , It should use TensorFlow Medium assign_sub() function , See the notes in the following figure for the reason ：

Reference link ： Teacher long Liangqu's 《 Deep learning and TensorFlow2 Introduction of actual combat 》 Class hour 39 Forward propagation -4,B Station links ：https://www.bilibili.com/video/BV165411j7ro?from=search&seid=4255840397349822501

This question may have been familiar to me for a long time TensorFlow It's not a problem for the big guys , But for a novice like me, I was confused when I first met him , So write this blog post to record