C3W3 Assignment: Lossand Val Loss negative values

Course Q&A TensorFlow Developer Professional Natural Language Processing in TensorFlow
1

I’ve tryed different combinations of layers defining the model for the assignment, but all the time I get negative values of loss and validation loss:

Epoch 1/20
4500/4500 [==============================] - 10s 2ms/step - loss: -87652.9844 - accuracy: 1.5278e-04 - val_loss: -236661.2500 - val_accuracy: 1.8750e-04
Epoch 2/20
4500/4500 [==============================] - 10s 2ms/step - loss: -511280.6875 - accuracy: 2.0833e-04 - val_loss: -800583.0625 - val_accuracy: 1.8750e-04
Epoch 3/20
4500/4500 [==============================] - 10s 2ms/step - loss: -1240391.0000 - accuracy: 2.0139e-04 - val_loss: -1641946.0000 - val_accuracy: 1.8750e-04
Epoch 4/20
4500/4500 [==============================] - 10s 2ms/step - loss: -2251042.2500 - accuracy: 2.1528e-04 - val_loss: -2753150.5000 - val_accuracy: 1.8750e-04
Epoch 5/20
4500/4500 [==============================] - 10s 2ms/step - loss: -3537188.5000 - accuracy: 2.2222e-04 - val_loss: -4129936.7500 - val_accuracy: 1.8750e-04
Epoch 6/20
4500/4500 [==============================] - 10s 2ms/step - loss: -5104800.5000 - accuracy: 2.0833e-04 - val_loss: -5781238.5000 - val_accuracy: 1.8750e-04

All previous Notebook cells produced output as expected except for the parse_data_from_file function:

dataset contains 1600000 examples

Text of second example should look like this:
is upset that he can’t update his Facebook by texting it… and might cry as a result School today also. Blah!

Text of fourth example should look like this:
my whole body feels itchy and like its on fire

Labels of last 5 examples should look like this:
[4, 4, 4, 4, 4]

Expected Output:

dataset contains 1600000 examples

Text of second example should look like this:
is upset that he can't update his Facebook by texting it... and might cry as a result  School today also. Blah!

Text of fourth example should look like this:
my whole body feels itchy and like its on fire 

Labels of last 5 examples should look like this:
[1, 1, 1, 1, 1]

I need to understand why I’m getting negative values for loss and validation loss in order to proceed with the assignment.

Can you please help?

  • Week #3
2

your labels do not match, is this course assignment? check if you have correct model compile parameters like optimizer, loss. Check your model.

Most important your labelling is incorrect. Read the instructions given before the grader cell, where you need to recall the labels.

Regards
DP