tensorflow - 将常量永久注入(inject)到 Tensorflow 图中进行推理

Question

我用 is_training 的占位符训练模型:

is_training_ph = tf.placeholder(tf.bool)

但是，一旦完成训练和验证，我想永久注入(inject) false 的常量输入该值，然后“重新优化”图形(即使用 optimize_for_inference )。是否有类似 freeze_graph 的内容？那会这样做吗？

Answer 1

一种可能性是使用 tf.import_graph_def() 函数及其input_map参数来重写图中该张量的值。例如，您可以按如下方式构建程序:

with tf.Graph().as_default() as training_graph:
  # Build model.
  is_training_ph = tf.placeholder(tf.bool, name="is_training")
  # ...
training_graph_def = training_graph.as_graph_def()

with tf.Graph().as_default() as temp_graph:
  tf.import_graph_def(training_graph_def,
                      input_map={is_training_ph.name: tf.constant(False)})
temp_graph_def = temp_graph.as_graph_def()

建好后 temp_graph_def ，您可以将其用作 freeze_graph 的输入.

一种替代方案，它可能与 freeze_graph 更兼容和 optimize_for_inference脚本(对变量名和检查点键进行假设)将修改 TensorFlow 的 graph_util.convert_variables_to_constants()函数，以便它转换占位符:

def convert_placeholders_to_constants(input_graph_def,
                                      placeholder_to_value_map):
  """Replaces placeholders in the given tf.GraphDef with constant values.

  Args:
    input_graph_def: GraphDef object holding the network.
    placeholder_to_value_map: A map from the names of placeholder tensors in
      `input_graph_def` to constant values.

  Returns:
    GraphDef containing a simplified version of the original.
  """

  output_graph_def = tf.GraphDef()

  for node in input_graph_def.node:
    output_node = tf.NodeDef()
    if node.op == "Placeholder" and node.name in placeholder_to_value_map:
      output_node.op = "Const"
      output_node.name = node.name
      dtype = node.attr["dtype"].type
      data = np.asarray(placeholder_to_value_map[node.name],
                        dtype=tf.as_dtype(dtype).as_numpy_dtype)
      output_node.attr["dtype"].type = dtype
      output_node.attr["value"].CopyFrom(tf.AttrValue(
          tensor=tf.contrib.util.make_tensor_proto(data,
                                                   dtype=dtype,
                                                   shape=data.shape)))
    else:
      output_node.CopyFrom(node)

    output_graph_def.node.extend([output_node])

  return output_graph_def

...然后你可以建立 training_graph_def如上，并写:

temp_graph_def = convert_placeholders_to_constants(training_graph_def,
                                                   {is_training_ph.op.name: False})