I can't initialize variable with tensorflow (protobuf error)

Viewed 85

I installed tensorflow 2.2.0 (protobuf 3.12.2) on my ubuntu machine (20.04) through pip. I let pip take care of all the dependencies. My problem is I receive an error everytime I try initializing a Variable. Below the sample code

import tensorflow as tf
 
class LinearModel:
    def __call__(self, x):
        return self.Weight * x + self.Bias

    def __init__(self):
        self.Weight = tf.Variable(11.0)
        self.Bias = tf.Variable(12.0)       
    
if __name__ == "__main__": 
    linear_model = LinearModel()
    

I took it from here

https://morioh.com/p/b12da909d731

The error I receive

AttributeError: 'google.protobuf.pyext._message.RepeatedCompositeCo' object has no attribute 'append'

(if required I can put the whole trace)

Attempt I've made

  1. trying multiple version of protobuf (I've read it could be a version mismatch although I did not find a suitable version)
  2. installing nightly builds of tensorflow
  3. tried on windows, it works

Any help is appreciated

0 Answers
Related