Hi, I just submitted my Face recognition assignment and I got 67/100, however during the assignment my solutions seemed to match the expected output. except for the case when there is a triplet loss of 527.2598. I have no idea how the other triplet tests pass and how to get this output
What output do you get for that section? Probably the most common mistake is omitting the “axis” parameter on the reduce_sum calls for pos_dist and neg_dist. Those are supposed to be vectors with one entry per “sample”.
I think there is a bug. This line of code will help to solve triplet loss problem.
loss = tf.maximum(basic_loss,0.0)
loss = tf.reduce_sum(loss)