freeze_graph in tensorflow: AssertionError: y_ is not in graph

Question

In Tensorflow, after training the model, I saved it use:

with tf.Session() as session:
/** 
    ------- Model training code goes here ------
**/
tf.train.write_graph(session.graph_def, '.', '../har.pbtxt')  
saver.save(session,save_path = "../har.ckpt")

And to freeze and save the optimized model:

from tensorflow.python.tools import freeze_graph
from tensorflow.python.tools import optimize_for_inference_lib

freeze_graph.freeze_graph(input_graph = "../har.pbtxt",  input_saver = "",
             input_binary = False, input_checkpoint = "../har.ckpt", output_node_names = "y_",
             restore_op_name = "save/restore_all", filename_tensor_name = "save/Const:0",
             output_graph = "frozen_har.pb", clear_devices = True, initializer_nodes = "")

input_graph_def = tf.GraphDef()
with tf.gfile.Open(output_frozen_graph_name, "r") as f:
    data = f.read()
    input_graph_def.ParseFromString(data)

output_graph_def = optimize_for_inference_lib.optimize_for_inference(
        input_graph_def,
        ["input"], 
        ["y_"],
        tf.float32.as_datatype_enum)

f = tf.gfile.FastGFile("optimized_frozen_har.pb", "w")
f.write(output_graph_def.SerializeToString())

However, I get the error:

Traceback (most recent call last):
File "C:\Users\asus\Desktop\cnn.py", line 176, in output_graph = "frozen_har.pb", clear_devices = True, initializer_nodes = "")
File "C:\Users\asus\AppData\Local\Programs\Python\Python35\lib\site-packages\tensorflow\python\tools\freeze_graph.py", line 122, in freeze_graph variable_names_blacklist=variable_names_blacklist) File "C:\Users\asus\AppData\Local\Programs\Python\Python35\lib\site-packages\tensorflow\python\framework\graph_util_impl.py", line 202, in convert_variables_to_constants inference_graph = extract_sub_graph(input_graph_def, output_node_names)
File "C:\Users\asus\AppData\Local\Programs\Python\Python35\lib\site-packages\tensorflow\python\framework\graph_util_impl.py", line 141, in extract_sub_graph assert d in name_to_node_map, "%s is not in graph" % d AssertionError: y_ is not in graph

I defined y_in my code as the output:

y_ = tf.nn.softmax(tf.matmul(f, out_weights) + out_biases)

What seens to be the problem?

Answer 1

When you use,

y_ = tf.nn.softmax(tf.matmul(f, out_weights) + out_biases)

y_ is not the name of the tensor. Please use the following, to explicitly name the tensor as y_.

y_ = tf.nn.softmax(tf.matmul(f, out_weights) + out_biases, name="y_")