jupyter notebook - Duplicate TensorFlow placeholder/variables? -
when write tensorflow code jupyter notebook:
a = tf.placeholder(tf.int32, name="a") # <tf.tensor 'a:0' shape=<unknown> dtype=int32> # <tf.tensor 'a_1:0' shape=<unknown> dtype=int32> ... i found if run cell multiple times, name of placeholder(variable) different, , affect future use of these placeholders, i.e., if other operations depend on these inputs, may feed values wrong placeholders. if update future operations, behaviors weird.
my question is, best practice when coding tensorflow on jupyter in such scenarios?
Comments
Post a Comment