Hi,
I tried all possible architectures and my validation accuracy is always 1 and so the loss for val comes as a flat line too and starts somewhere around 2.3 e-7. Can someone help me understand why it is. I followed all the steps and there is no error in any other steps.
And also the train_accuracy become 1 in 1 epoch. Is that what it is supposed to be?
I’m not a mentor for this course, but often the reason for too-high validation accuracy is if you’re actually testing with same set you used for training, instead of using the validation set.
Makes sense. But I use this code to split train and validation, which had worked earlier:
[snippet removed by mentor]
Please don’t post code in public. You’re welcome to share the stacktrace though.
It looks like you’ve satisfied the passing criteria for the assignment: To pass this assignment the slope of your val_loss curve should be 0.0005 at maximum.
What was the grader feedback?
hey thanks, i passed. I was wondering if there if I had manipulated the data was wrong somewhere. Thanks.
Please click my name and message your notebook as an attachment.