Tensorflow: empty computational graph & garbage collection

Question

1 EMPTY GRAPH ERROR

Hey I am trying to run multiple tensorflow graphs completely separate and am running into the following inheritance problem.

Also does

import tensorflow as tf


class A:

    g = tf.Graph()
    g.as_default()
    s = tf.Session(graph=g)

    x = tf.placeholder(tf.float32)

    __call__ = lambda self,X: self.s.run(self.y, {self.x:X})


class B(A):

    y = 2 * A.x


test = B()
print test([1,1,2])

error

RuntimeError: The Session graph is empty.  Add operations to the graph before calling run()

---

2 - GARBAGE COLLECTION

I am also curious to find out about deleting these distinct graphs, if I close the session with Session().close() and it is the only session aware of the graph will this graph now disappear and be garbage collected?

Answer 1

Question 1

If you want your operations to be added to a specific graph, you need to use with g.as_default() as context:

class A:

    g = tf.Graph()
    with g.as_default():
        x = tf.placeholder(tf.float32)

    s = tf.Session(graph=g)

    __call__ = lambda self,X: self.s.run(self.y, {self.x:X})


class B(A):

    with g.as_default():
        y = 2 * A.x


test = B()
print test([1,1,2])

(PS: your code is really badly written, hope it's just for testing)

---

Question 2

A graph is not affected by a corresponding session.

You can create a graph, open a session and close it, the graph will remain intact:

g = tf.Graph()
with g.as_default():
    # build graph...
    x = tf.constant(1)

sess = tf.Session(graph=g)
sess.run(x)
sess.close()

# Now we can create a new session with the same graph
sess2 = tf.Session(graph=g)
sess2.run(x)
sess2.close()