UNQ_C10 partially failed

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

This is the grader output :

Failed simple_test_check.
Expected:
(‘Ich iss Suppe.’, 0, {0: 0.8571428571428572, 1: 0.8571428571428572, 2: 0.7619047619047619, 3: 0.8571428571428571}),
but got:
(‘Ich esse Suppe.’, 3, {0: 0.8571387701816032, 1: 0.8571387701816032, 2: 0.7619111199457392, 3: 0.857142857142857}).

But how can the grader ensure that the model trained in the notebook has exactly produced the same weights that the one which has produced the expected outputs ??

Thanks

Jerome

2

Hi Jerome,

The weights used in testing are loaded from file through
model.init_from_file(“model.pkl.gz”, weights_only=True).