python - tensorflow: initialization of variables inside function -

newbee tensorflow. i'm trying write simple net following code:

import tensorflow tf  import tensorflow.contrib tfc import tensorflow.contrib.layers tfcl  def generator_deconv(z, kernel):     tf.variable_scope("generator", reuse=true):         weights = tf.get_variable("weights")         biases = tf.get_variable("biases")         result = tf.matmul(z, weights)         result = tf.add(result, biases)         result = tf.reshape(result, tf.stack([tf.shape(result)[0],13,4,1]))         result = tf.nn.conv2d_transpose(result, kernel,                  output_shape=[tf.shape(result)[0],25,8,1],                  strides=[1,2,2,1],                  padding="same")         result = tf.nn.conv2d_transpose(result, kernel,                  output_shape=[tf.shape(result)[0],50,15,1],                  strides=[1,2,2,1],                  padding="same")         result = tf.nn.conv2d_transpose(result, kernel,                  output_shape=[tf.shape(result)[0],100,30,1],                  strides=[1,2,2,1],                  padding="same")             return result  kernel = tf.constant(1.0, shape=[4,4,1,1]) protype = tf.constant(1.0, shape=[3,4]) init = tf.global_variables_initializer()  config = tf.configproto() config.gpu_options.allocator_type = 'bfc' config.gpu_options.allow_growth=true  tf.variable_scope("generator"):     t1 = tf.get_variable("weights",shape=[4,52])     t2 = tf.get_variable("biases", shape=[52])  test = generator_deconv(protype,kernel)  tf.session(config=config) sess:     sess.run(init)     sess.run(tf.shape(t1))     sess.run(tf.shape(t2))     sess.run(tf.shape(test)) 

but got error:

tensorflow.python.framework.errors_impl.failedpreconditionerror: attempting use uninitialized value generator/weights

for last line

sess.run(tf.shape(test)) 

checked official api of tensorflow still don't know what's wrong code.

================================update==========================

found 2 ways fix it

1.if replace

sess.run(init) 

by

sess.run(tf.global_variables_initializer()) 

then whole code works.

or

2.run

init = tf.global_variables_initializer() tf.session(config=config) sess:     sess.run(init)     sess.run(tf.shape(t1))     sess.run(tf.shape(t2))     sess.run(tf.shape(test)) 

again works.

but don't understand why

i removed parts of code you:

init = tf.global_variables_initializer()  tf.variable_scope("generator"):     t1 = tf.get_variable("weights",shape=[4,52])     t2 = tf.get_variable("biases", shape=[52])  tf.session(config=config) sess:     sess.run(init)     sess.run(tf.shape(t1)) 

you add variables graph after saved result of calling global_variables_initializer(). in fix call function after added variables want initialize graph, , initialized.

hope helps!