爲tf.cond不確定性的行爲上is_variable_initialized

Question

考慮這個演示代碼：

import logging 
logging.getLogger('tensorflow').disabled = True 
import tensorflow as tf 
import sys 

sess = tf.InteractiveSession() 


def test_var_2(): 
    v = tf.Variable(initial_value=True, trainable=False, name="test_var_3") 
    print("is initialized:", tf.is_variable_initialized(v).eval(), file=sys.__stdout__) 

    def make_init(): 
    with tf.control_dependencies([tf.Print(0, ["var is not initialized", tf.is_variable_initialized(v)])]): 
     return tf.variables_initializer([v]) 

    def on_already_init(): 
    return tf.group(tf.no_op(), tf.Print(0, ["var is initialized", tf.is_variable_initialized(v)])) 

    maybe_init = tf.cond(
    tf.is_variable_initialized(v), 
    on_already_init, 
    make_init, 
    name="init") 
    with tf.control_dependencies([maybe_init, tf.Print(0, ["cd init"])]): 
    x = tf.where(v.read_value(), True, False) 

    print("x:", x.eval()) 
    print("is initialized:", tf.is_variable_initialized(v).eval(), file=sys.__stdout__) 
    print("x:", x.eval()) 
    print("is initialized:", tf.is_variable_initialized(v).eval(), file=sys.__stdout__) 


if __name__ == "__main__": 
    test_var_2() 

輸出具有不確定性。我要麼得到這個（1）：

is initialized: False 
.../logging_ops.cc:79] [cd init] 
.../logging_ops.cc:79] [var is not initialized][0] 
x: True 
is initialized: True 
.../logging_ops.cc:79] [cd init] 
.../logging_ops.cc:79] [var is initialized][1] 
x: True 
is initialized: True 

還是這個（2）：

is initialized: False 
.../logging_ops.cc:79] [cd init] 
.../logging_ops.cc:79] [var is not initialized][1] 
x: True 
is initialized: True 
.../logging_ops.cc:79] [cd init] 
.../logging_ops.cc:79] [var is initialized][1] 
x: True 
is initialized: True 

還是這個（3）：

is initialized: False 
.../logging_ops.cc:79] [cd init] 
.../logging_ops.cc:79] [var is initialized][1] 
x: True 
is initialized: True 
.../logging_ops.cc:79] [cd init] 
.../logging_ops.cc:79] [var is initialized][1] 
x: True 
is initialized: True 

的變體3（其中VAR初始化在第一個x.eval()）對我來說毫無意義。它怎麼會在那個cond-branch中結束？爲什麼突然初始化？

變體1和2之間的區別可能是因爲tf.Print沒有完全評估爲tf.control_dependencies的一部分？

Answer 1

這是一個更加棘手的案例this和this。關鍵是tf.variables_initializer([v])並沒有真正爲init創建一個新的操作（即分配給初始值），但是這個assign-op已經在之前創建（在tf.Variable的構造函數中）並在此處使用。因此，我的解決辦法是不使用這個初始化而是要創造我自己的，這必須在make_init函數內部創建，即：

def make_init(): 
    with tf.control_dependencies([tf.Print(0, ["var is not initialized", tf.is_variable_initialized(v)])]): 
    return tf.assign(v, True) 

這解決了問題。