While debugging, how to print all variables (which is in list format) who are trainable in Tensorflow?

Question

For instance,

    tvars = tf.trainable_variables()

I want to check all the variables in tvars (which is list type).

I've already tried the below code which returns error,

    myvars = session.run([tvars])
    print(myvars)

Can you be more specific as to what error you are getting?! – shekkizh Jul 21 '16 at 18:23 — shekkizh, Jul 21 '16 at 18:23
print([v.name for v in tf.trainable_variables()]) – Yaroslav Bulatov Jul 21 '16 at 18:34 — Yaroslav Bulatov, Jul 21 '16 at 18:34

score 15 · Answer 1 · answered Jul 26 '16 at 03:28

Since tf.trainable_variables() returns a list of tf.Variable objects, you should be able to pass its result straight to Session.run():

tvars = tf.trainable_variables()
tvars_vals = sess.run(tvars)

for var, val in zip(tvars, tvars_vals):
    print(var.name, val)  # Prints the name of the variable alongside its value.

score 5 · Answer 2 · answered Apr 12 '18 at 18:58

5

To print the complete list of all all variables or nodes of a tensor-flow graph, you may try this:

[n.name for n in tf.get_default_graph().as_graph_def().node]

I copied this from here.

answered Apr 12 '18 at 18:58

rocksyne

1,264
15
17

While debugging, how to print all variables (which is in list format) who are trainable in Tensorflow?

2 Answers2