从文件打开张量流图

Question

我正在尝试使用 tensorflow 进行研究，但我不明白如何打开和使用早期保存在文件中的类型为 tf.Graph 的图形。像这样的东西：

import tensorflow as tf

my_graph = tf.Graph()

with g.as_default():
    x = tf.Variable(0)
    b = tf.constant(-5)
    k = tf.constant(2)

    y = k*x + b

tf.train.write_graph(my_graph, '.', 'graph.pbtxt')

f = open('graph.pbtxt', "r")

# Do something with "f" to get my saved graph and use it below in
# tf.Session(graph=...) instead of dots

with tf.Session(graph=...) as sess:
    tf.initialize_all_variables().run()

    y1 = sess.run(y, feed_dict={x: 5})
    y2 = sess.run(y, feed_dict={x: 10})
    print(y1, y2)

Answer 1

您必须加载文件内容，将其解析为 GraphDef，然后导入。它将被导入到当前图形中。您可能想用graph.as_default():上下文管理器包装它。

import tensorflow as tf
from tensorflow.core.framework import graph_pb2 as gpb
from google.protobuf import text_format as pbtf

gdef = gpb.GraphDef()

with open('my-graph.pbtxt', 'r') as fh:
    graph_str = fh.read()

pbtf.Parse(graph_str, gdef)

tf.import_graph_def(gdef)

Answer 2

我这样解决了这个问题：首先，我在图形“输出”中命名所需的计算，然后将该模型保存在下面的代码中......

import tensorflow as tf

x = tf.placeholder(dtype=tf.float64, shape=[], name="input")
a = tf.Variable(111, name="var1", dtype=tf.float64)
b = tf.Variable(-666, name="var2", dtype=tf.float64)

y = tf.add(x, a, name="output")

saver = tf.train.Saver()

with tf.Session() as sess:
    tf.initialize_all_variables().run()

    print(sess.run(y, feed_dict={x: 555}))

    save_path = saver.save(sess, "model.ckpt", meta_graph_suffix='meta', write_meta_graph=True)
    print("Model saved in file: %s" % save_path)

其次，我需要在图中运行某些操作，我知道该操作的名称为“输出”。因此，我只需在另一个代码中恢复模型，并通过采用名称为“输入”和“输出”的必要图形部分来运行恢复的计算：

import tensorflow as tf

# Restore graph to another graph (and make it default graph) and variables
graph = tf.Graph()
with graph.as_default():
    saver = tf.train.import_meta_graph("model.ckpt.meta")

    y = graph.get_tensor_by_name("output:0")
    x = graph.get_tensor_by_name("input:0")

    with tf.Session() as sess:

        saver.restore(sess, "model.ckpt")

        print(sess.run(y, feed_dict={x: 888}))

        # Variable out:
        for var in tf.all_variables():
            print("%s %.2f" % (var.name, var.eval()))