Tensorflow eval() without session or move variable to an other session

Question

I'm using tensorflow-models like it is described in the iris predict examples. Because of that I do not have a session object. Now I want to convert the labels to a numpy-array with .eval(). Without a session there comes an error.

Traceback (most recent call last):
 File "myfile.py", line 273, in <module>
   tf.app.run()
 File "/usr/local/lib/python3.4/site-packages/tensorflow/python/platform/app.py", line 30, in run
   sys.exit(main(sys.argv))
 File "myfile.py", line 270, in main
   train_and_eval()
 File "myfile.py", line 258, in train_and_eval
   label.eval()
 File "/usr/local/lib/python3.4/site-packages/tensorflow/python/framework/ops.py", line 559, in eval
   return _eval_using_default_session(self, feed_dict, self.graph, session)
 File "/usr/local/lib/python3.4/site-packages/tensorflow/python/framework/ops.py", line 3642, in _eval_using_default_session
   raise ValueError("Cannot evaluate tensor using `eval()`: No default "
ValueError: Cannot evaluate tensor using `eval()`: No default session is registered. Use `with sess.as_default()` or pass an explicit session to `eval(session=sess)`

Is there a possibility to access / get the session the model used in background? Or is there an other possibility to convert the tensor to a numpy-array?

If I create a new session, then it seems that tensorflow moves to this session but has no access to the variable. A python print() is displayed, but then it runs inifite. How can I parse a variable to this new session?

The other part of the net works well - it's only this special thing convert the tensor to a numpy-array

    COLUMNS = ["col1", "col2", "col3", "target"]
    LABEL_COLUMN = "target"
    CATEGORICAL_COLUMNS = ["col1", "col2", "col3"]

    def build_estimator(model_dir):
        col1 = tf.contrib.layers.sparse_column_with_hash_bucket(
            "col1", hash_bucket_size=10000)
        col2........

        wide_columns = [col1, col2, col3]
        deep_columns = [
            tf.contrib.layers.embedding_column(col1, dimension=7),
            tf.contrib.layers.embedding_column(col2, dimension=7),
            tf.contrib.layers.embedding_column(col3, dimension=7)
        ]

        m = tf.contrib.learn.DNNLinearCombinedClassifier(...)
        return m

    def input_fn(file_names, batch_size):
        ...
        label = tf.string_to_number(examples_dict[LABEL_COLUMN], out_type=tf.int32)
        return feature_cols, label

    def train_and_eval():
        model_dir = "./model/"
        print(model_dir)

        m = build_estimator(model_dir)
        m.fit(input_fn=lambda: input_fn(train_file_name, batch_size), steps=steps)
        results = m.evaluate(input_fn=lambda: input_fn(test_file_name, batch_size),
            steps=1)
        pred_m = m.predict(input_fn=lambda: input_fn(test_file_name, batch_size))


        sess = tf.InteractiveSession()
        with sess.as_default():
            print("Is a session there?")
            _, label = input_fn(test_file_name, batch_size)
            label.eval()
            print(label)

    def main(_):
        train_and_eval()

    if __name__ == "__main__":
        tf.app.run()

The new session starts at the end of the code-snippet:

        sess = tf.InteractiveSession()
        with sess.as_default():
            print("Is a session there?")
            _, label = input_fn(test_file_name, batch_size)
            label.eval()
            print(label)

Solution 1

You need a Session and you need to initialize your variables before being able to access them:

with Session() as sess:
    sess.run(tf.global_variables_initializer())
    ...  
    label_numpy = label.eval()

Solution 2

I changed the order so as to solved my problem.

     import tensorflow as tf
     v= tf.constant([[1.0, 2.0, 3.0], [4.0, 5.0, 6.0]])
     result = tf.clip_by_value(v, 2.5, 4.5).eval()
     with tf.Session() as sess:
          print(sess.run(result))

Then my IDE warned"ValueError: Cannot evaluate tensor using eval(): No default session is registered. Use with sess.as_default() or pass an explicit session to eval(session=sess)"

Afterwards, I changed it into:

    import tensorflow as tf
    v= tf.constant([[1.0, 2.0, 3.0], [4.0, 5.0, 6.0]])
    with tf.Session() as sess:
        result = tf.clip_by_value(v, 2.5, 4.5).eval()
        print(sess.run(result))

Then the problem is solved.

Comments

[Gersee]

Thanks for your support. But this does not work like it was asked. I can use the new session (this already worked with my code), but I do not have access to the variables of the other session. How can I pass them to the new session? Or alternatively how can I change my code / the iris predict example to use an explicit session or how can I get access to the default session of this code?

[ribbit]

Did you ever find how to do this?