tensorflow - Test accuracy keeps changing for the same data set -


i'm importing pre-trained model , using classify data. normal predicted values/ final accuracy keeps changing every time run model same data set?

i printed softmax values of final layer well. probabilities keep changing.

i encounter same issue recently, , found root cause dropout layer in model, hope helps.


Comments

Popular posts from this blog

php - Vagrant up error - Uncaught Reflection Exception: Class DOMDocument does not exist -

vue.js - Create hooks for automated testing -

Add new key value to json node in java -