UNQ_C8: Accuracy too low - failed test

Course Q&A Natural Language Processing NLP with Sequence Models
1

My accuracy was ~ .977 < .98 on the unit test.
I printed the “batch accuracies”, and most of them are 1.0.
Some are .9375 (missed one out of 16).
But a few are much lower, e.g. .5625.
Is there anything I can do to increase the accuracy of the model so I can pass this test?

Here’s some output I created:
batch accuracy = 0.9375
batch accuracy = 1.0
batch accuracy = 0.9375
batch accuracy = 0.5625
batch accuracy = 1.0
batch accuracy = 1.0
batch accuracy = 0.9375
batch accuracy = 1.0
batch accuracy = 1.0
batch accuracy = 1.0
batch accuracy = 1.0

2

Hi John8,

Did you submit the assignment? If it passes I would not worry about the public test. If not, the batch accuracy of 0.5625 is peculiar and may point to an underlying problem - though it is hard to say offhand what this could be. In that case feel free to send me your notebook as an attachment to a direct message so I can have a look.