Accuracy is correct, but history = 0

Hello,

Week 01 - Assignment 01 (" Programming Assignment: Multiple Output Models using Keras Functional API")

I have a couple issues here

[1] Issue with accuracy training is correct, but history = 0:
please see the picture enclosed

The blue circles show that after the training:

  • my binary accuracy: 0.9937
  • Expected outcome: 0.97-1.00
  • wine_type_accuracy: 0.00 ← This is where I don’t understand

[2] All public tests passed but submit score says there is an issue with history var

Many learners have the same issue, but we still cannot figure out what is wrong

utils.test_history(history)

The test passed but when submit the assignment, they says there is an issue with history var

I suspect is model.compile, the value I enter is:
For loss: 'binary_crossentropy', 'mse'
For metrics: tf.keras.metrics.BinaryAccuracy(), tf.keras.metrics.RootMeanSquaredError()

Please have a look if it is correct

Thank you

Hello,

Please explain further, which assignment is this, which week?

Hello
Week 01- Assignment 01 (" Programming Assignment: Multiple Output Models using Keras Functional API")

My assignment is only 9/10 and they says the issue is about the utils.test_history(history)

However, when running, the assignment says I passed the test, as in picture bellow.

The 1st post is just my suspicion where the cause is. I think it might related to the code of model.compile because I use trial-and-error several times to reach until it is correct.
The inputs for model.compile is:
For loss: 'binary_crossentropy', 'mse'
For metrics: tf.keras.metrics.BinaryAccuracy(), tf.keras.metrics.RootMeanSquaredError()

Please help me have a look
Thank you

Yes, the assignment is telling you the accuracy for wine type and yours is not right.

Try to fix that one first.