UserWarning: "categorical_crossentropy received from_logits=True, but the output argument was produced by a Softmax activation and thus does not represent logits. Was this intended?

[ankhanh-vni]

Should we modify the from_logits parameters in dvc_workshop/models/classifier.py to False, since we are passing a softmax activation function at the output layer after all?

[TreeOfEast]

Should we modify the from_logits parameters in dvc_workshop/models/classifier.py to False, since we are passing a softmax activation function at the output layer after all?

I encountered this issue while using ipynb, which caused a direct interruption of my kernel. I don't know how you solved this problem?

[ankhanh-vni]

I set the parameter from_logits to False. Hope it helps