Tencent / tencent-ml-images

Largest multi-label image database; ResNet-101 model; 80.73% top-1 acc on ImageNet
Other
3.06k stars 515 forks source link

non_neg_mask not work #60

Open jasstionzyf opened 5 years ago

jasstionzyf commented 5 years ago

as i read loss function, notice:

b. get non-negative mask

non_neg_mask = tf.fill(tf.shape(labels), -1.0, name='non_neg') non_neg_mask = tf.cast(tf.not_equal(labels, non_neg_mask), tf.float32) tf.summary.histogram('non_neg', non_neg_mask)

but based on above code , non_neg_mask is alwayes like : [1,1,1,1,1,1,1] wonder non_neg_mask is useless?