I managed to train a DL model with Tensorflow. The format of the model is Softmax. I continuously run this model with the same samples, but the results are not the same, the probabilities of class A or class B change over time. But the final result classification results are always correct, they didn't change. Why exactly?
For example:
At first, the classification result of a testing sample may be 9.9643815e-01 3.5618523e-03 but the next time, the result of this sample may change to [9.93039072e-01 6.96092704e-03].
Comments
Post a Comment