import_meta_graph: ValueError: At least two variables have the same name: Variable_1

[snbuchholz]

What related GitHub issues or StackOverflow threads have you found by searching the web for your problem?

None

Environment info

Operating System: Ubuntu 16.04.1 LTS
Installed version of CUDA and cuDNN: none
pip package: https://storage.googleapis.com/tensorflow/linux/cpu/tensorflow-0.10.0-cp27-none-linux_x86_64.whl
$ python -c "import tensorflow; print(tensorflow.version)"
0.10.0

If possible, provide a minimal reproducible example (We usually don't have time to read hundreds of lines of your code)

Using commands from https://www.tensorflow.org/versions/r0.9/get_started/index.html to create a model:

import tensorflow as tf
import numpy as np
x_data = np.random.rand(100).astype(np.float32)
y_data = x_data * 0.1 + 0.3
W = tf.Variable(tf.random_uniform([1], -1.0, 1.0))
b = tf.Variable(tf.zeros([1]))
y = W * x_data + b
loss = tf.reduce_mean(tf.square(y - y_data))
optimizer = tf.train.GradientDescentOptimizer(0.5)
train = optimizer.minimize(loss)
init = tf.initialize_all_variables()
sess = tf.Session()
sess.run(init)
sess.run(train)

Then this works:

saver = tf.train.Saver()
saver.save(sess, 'my-model')
sess = tf.Session()
new_saver = tf.train.import_meta_graph('my-model.meta')

But this doesn't:

tf.train.export_meta_graph(filename='my-model.meta')
new_saver = tf.train.import_meta_graph('my-model.meta')

Traceback (most recent call last):
File "", line 1, in
File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/training/saver.py", line 1458, in import_meta_graph
return _import_meta_graph_def(read_meta_graph_file(meta_graph_or_file))
File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/training/saver.py", line 1386, in _import_meta_graph_def
return Saver()
File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/training/saver.py", line 861, in init
restore_sequentially=restore_sequentially)
File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/training/saver.py", line 502, in build
vars_to_save = self._ValidateAndSliceInputs(names_to_variables)
File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/training/saver.py", line 399, in _ValidateAndSliceInputs
names_to_variables = self._VarListToDict(names_to_variables)
File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/training/saver.py", line 377, in _VarListToDict
name)
ValueError: At least two variables have the same name: Variable_1

What I actually want to use is this, which also does not work (same error):

meta_graph_def = tf.train.export_meta_graph()
saver = tf.train.import_meta_graph(meta_graph_def)

I need a solution that does not rely on the file system.
Am I doing something wrong?

What other attempted solutions have you tried?

I tried to look at the files to see where these duplicated names occur but did not find any.

[concretevitamin]

You should call tf.train.import_meta_graph() with a cleared Graph. Example:

    # Create a clean graph and import the MetaGraphDef nodes.
    new_graph = tf.Graph()
    with tf.Session(graph=new_graph) as sess:
      # Import the previously export meta graph.
      saver = tf.train.import_meta_graph(meta_graph_def)

Please feel free to reopen if this doesn't solve the issue.

[snbuchholz]

Thanks! This information was missing from https://www.tensorflow.org/versions/r0.9/how_tos/meta_graph/index.html