zkjqw139 comments

Repositories
Issues
Comments

Results 1 comments of


                                            zkjqw139

After some steps while training, the weights and loss value will be NaN .

try use labels = tf.one_hot(labels_placeholder, c3d_model.NUM_CLASSES) loss = -tf.reduce_sum(labels*tf.log(tf.clip_by_value(tf.nn.softmax(logit),1e-10,1.0)),1.0) instead