Python meta_graph.create_meta_graph_def方法代碼示例

# 需要導入模塊: from tensorflow.python.framework import meta_graph [as 別名]
# 或者: from tensorflow.python.framework.meta_graph import create_meta_graph_def [as 別名]
def before_run(self, run_context):  # pylint: disable=unused-argument
    if self._timer.last_triggered_step() is None:
      # We do write graph and saver_def at the first call of before_run.
      # We cannot do this in begin, since we let other hooks to change graph and
      # add variables in begin. Graph is finalized after all begin calls.
      training_util.write_graph(
          ops.get_default_graph().as_graph_def(add_shapes=True),
          self._checkpoint_dir,
          "graph.pbtxt")
      saver_def = self._get_saver().saver_def if self._get_saver() else None
      graph = ops.get_default_graph()
      meta_graph_def = meta_graph.create_meta_graph_def(
          graph_def=graph.as_graph_def(add_shapes=True),
          saver_def=saver_def)
      self._summary_writer.add_graph(graph)
      self._summary_writer.add_meta_graph(meta_graph_def)

    return SessionRunArgs(self._global_step_tensor) 

# 需要導入模塊: from tensorflow.python.framework import meta_graph [as 別名]
# 或者: from tensorflow.python.framework.meta_graph import create_meta_graph_def [as 別名]
def after_create_session(self, session, coord):
    global_step = session.run(self._global_step_tensor)

    # We do write graph and saver_def at the first call of before_run.
    # We cannot do this in begin, since we let other hooks to change graph and
    # add variables in begin. Graph is finalized after all begin calls.
    def _write_graph_fn(self):
      training_util.write_graph(
          ops.get_default_graph().as_graph_def(add_shapes=True),
          self._checkpoint_dir, "graph.pbtxt")
    self._write_graph_thread = threading.Thread(target=_write_graph_fn,
                                                args=[self])
    self._write_graph_thread.start()

    saver_def = self._get_saver().saver_def if self._get_saver() else None
    graph = ops.get_default_graph()
    meta_graph_def = meta_graph.create_meta_graph_def(
        graph_def=graph.as_graph_def(add_shapes=True), saver_def=saver_def)
    self._summary_writer.add_graph(graph)
    self._summary_writer.add_meta_graph(meta_graph_def)
    # The checkpoint saved here is the state at step "global_step".
    self._save(session, global_step)
    self._timer.update_last_triggered_step(global_step) 

# 需要導入模塊: from tensorflow.python.framework import meta_graph [as 別名]
# 或者: from tensorflow.python.framework.meta_graph import create_meta_graph_def [as 別名]
def before_run(self, run_context):  # pylint: disable=unused-argument
    if self._timer.last_triggered_step() is None:
      # Write graph in the first call.
      training_util.write_graph(
          ops.get_default_graph().as_graph_def(add_shapes=True),
          self._checkpoint_dir,
          "graph.pbtxt")
      saver_def = self._saver.saver_def if self._saver else None
      graph = ops.get_default_graph()
      meta_graph_def = meta_graph.create_meta_graph_def(
          graph_def=graph.as_graph_def(add_shapes=True),
          saver_def=saver_def)
      self._summary_writer.add_graph(graph)
      self._summary_writer.add_meta_graph(meta_graph_def)

    return SessionRunArgs(self._global_step_tensor) 

# 需要導入模塊: from tensorflow.python.framework import meta_graph [as 別名]
# 或者: from tensorflow.python.framework.meta_graph import create_meta_graph_def [as 別名]
def test_train(self):
    with tf.Graph().as_default() as g, self.test_session(g):
      with tf.control_dependencies(self._build_inference_graph()):
        train_op = tf.assign_add(tf.contrib.framework.get_global_step(), 1)
      self._assert_summaries(self._output_dir)
      self._assert_ckpt(self._output_dir, False)
      loss = learn.graph_actions._monitored_train(  # pylint: disable=protected-access
          g,
          output_dir=self._output_dir,
          train_op=train_op,
          loss_op=tf.constant(2.0),
          steps=1)
      meta_graph_def = meta_graph.create_meta_graph_def()
      self.assertEqual(2.0, loss)
      self._assert_summaries(self._output_dir, expected_graphs=[g],
                             expected_meta_graphs=[meta_graph_def])
      self._assert_ckpt(self._output_dir, True) 

# 需要導入模塊: from tensorflow.python.framework import meta_graph [as 別名]
# 或者: from tensorflow.python.framework.meta_graph import create_meta_graph_def [as 別名]
def test_train_loss(self):
    with tf.Graph().as_default() as g, self.test_session(g):
      tf.contrib.framework.create_global_step()
      loss_var = tf.contrib.framework.local_variable(10.0)
      train_op = tf.group(
          tf.assign_add(tf.contrib.framework.get_global_step(), 1),
          tf.assign_add(loss_var, -1.0))
      self._assert_summaries(self._output_dir)
      self._assert_ckpt(self._output_dir, False)
      loss = learn.graph_actions._monitored_train(  # pylint: disable=protected-access
          g,
          output_dir=self._output_dir,
          train_op=train_op,
          loss_op=loss_var.value(),
          steps=6)
      meta_graph_def = meta_graph.create_meta_graph_def()
      self.assertEqual(4.0, loss)
      self._assert_summaries(self._output_dir, expected_graphs=[g],
                             expected_meta_graphs=[meta_graph_def])
      self._assert_ckpt(self._output_dir, True) 

# 需要導入模塊: from tensorflow.python.framework import meta_graph [as 別名]
# 或者: from tensorflow.python.framework.meta_graph import create_meta_graph_def [as 別名]
def test_train_summaries(self):
    with tf.Graph().as_default() as g, self.test_session(g):
      with tf.control_dependencies(self._build_inference_graph()):
        train_op = tf.assign_add(tf.contrib.framework.get_global_step(), 1)
      loss_op = tf.constant(2.0)
      tf.summary.scalar('loss', loss_op)
      self._assert_summaries(self._output_dir)
      self._assert_ckpt(self._output_dir, False)
      loss = learn.graph_actions._monitored_train(  # pylint: disable=protected-access
          g,
          output_dir=self._output_dir,
          train_op=train_op,
          loss_op=loss_op,
          steps=1)
      meta_graph_def = meta_graph.create_meta_graph_def()
      self.assertEqual(2.0, loss)
      self._assert_summaries(self._output_dir,
                             expected_graphs=[g],
                             expected_meta_graphs=[meta_graph_def],
                             expected_summaries={1: {'loss': 2.0}})
      self._assert_ckpt(self._output_dir, True) 

# 需要導入模塊: from tensorflow.python.framework import meta_graph [as 別名]
# 或者: from tensorflow.python.framework.meta_graph import create_meta_graph_def [as 別名]
def test_summary_writer_defs(self):
    fake_summary_writer.FakeSummaryWriter.install()
    tf.compat.v1.summary.FileWriterCache.clear()
    summary_writer = tf.compat.v1.summary.FileWriterCache.get(self.model_dir)

    with self.graph.as_default():
      hook = basic_session_run_hooks.CheckpointSaverHook(
          self.model_dir, save_steps=2, scaffold=self.scaffold)
      hook.begin()
      self.scaffold.finalize()
      with tf.compat.v1.Session() as sess:
        sess.run(self.scaffold.init_op)
        mon_sess = monitored_session._HookedSession(sess, [hook])
        hook.after_create_session(sess, None)
        mon_sess.run(self.train_op)
      summary_writer.assert_summaries(
          test_case=self,
          expected_logdir=self.model_dir,
          expected_added_meta_graphs=[
              meta_graph.create_meta_graph_def(
                  graph_def=self.graph.as_graph_def(add_shapes=True),
                  saver_def=self.scaffold.saver.saver_def)
          ])

    fake_summary_writer.FakeSummaryWriter.uninstall() 

# 需要導入模塊: from tensorflow.python.framework import meta_graph [as 別名]
# 或者: from tensorflow.python.framework.meta_graph import create_meta_graph_def [as 別名]
def before_run(self, run_context):
        """ Dumps graphs and loads checkpoint if there exits.

        Called before each call to run().

        Args:
            run_context: A `SessionRunContext` object.

        Returns: A `SessionRunArgs` object containing global_step.
        """
        # We do write graph and saver_def at the first call of before_run.
        # We cannot do this in begin, since we let other hooks to change graph and
        # add variables in begin. Graph is finalized after all begin calls.
        if self._is_chief and self._first_call:
            training_util.write_graph(
                ops.get_default_graph().as_graph_def(add_shapes=True),
                self._checkpoint_dir,
                "graph.pbtxt")
            # dump model details "model_analysis.txt"
            dump_model_analysis(self._checkpoint_dir)  # dump model configs
            graph = ops.get_default_graph()
            meta_graph_def = meta_graph.create_meta_graph_def(
                graph_def=graph.as_graph_def(add_shapes=True),
                saver_def=self._saver.saver_def)
            if self._summary_writer is not None:
                self._summary_writer.add_graph(graph)
                self._summary_writer.add_meta_graph(meta_graph_def)
            tf.logging.info("CheckpointSaverHook (before_run): dump graph...")
        self._first_call = False
        return tf.train.SessionRunArgs(self._global_step) 

# 需要導入模塊: from tensorflow.python.framework import meta_graph [as 別名]
# 或者: from tensorflow.python.framework.meta_graph import create_meta_graph_def [as 別名]
def __init__(self, event_writer, graph=None, graph_def=None):
    """Creates a `SummaryWriter` and an event file.

    On construction the summary writer creates a new event file in `logdir`.
    This event file will contain `Event` protocol buffers constructed when you
    call one of the following functions: `add_summary()`, `add_session_log()`,
    `add_event()`, or `add_graph()`.

    If you pass a `Graph` to the constructor it is added to
    the event file. (This is equivalent to calling `add_graph()` later).

    TensorBoard will pick the graph from the file and display it graphically so
    you can interactively explore the graph you built. You will usually pass
    the graph from the session in which you launched it:

    ```python
    ...create a graph...
    # Launch the graph in a session.
    sess = tf.Session()
    # Create a summary writer, add the 'graph' to the event file.
    writer = tf.summary.FileWriter(<some-directory>, sess.graph)
    ```


    Args:
      event_writer: An EventWriter. Implements add_event and get_logdir.
      graph: A `Graph` object, such as `sess.graph`.
      graph_def: DEPRECATED: Use the `graph` argument instead.
    """
    self.event_writer = event_writer
    # For storing used tags for session.run() outputs.
    self._session_run_tags = {}
    if graph is not None or graph_def is not None:
      # Calling it with both graph and graph_def for backward compatibility.
      self.add_graph(graph=graph, graph_def=graph_def)
      # Also export the meta_graph_def in this case.
      # graph may itself be a graph_def due to positional arguments
      maybe_graph_as_def = (graph.as_graph_def(add_shapes=True)
                            if isinstance(graph, ops.Graph) else graph)
      self.add_meta_graph(
          meta_graph.create_meta_graph_def(graph_def=graph_def or
                                           maybe_graph_as_def)) 

# 需要導入模塊: from tensorflow.python.framework import meta_graph [as 別名]
# 或者: from tensorflow.python.framework.meta_graph import create_meta_graph_def [as 別名]
def __init__(self, event_writer, graph=None, graph_def=None):
    """Creates a `SummaryWriter` and an event file.

    On construction the summary writer creates a new event file in `logdir`.
    This event file will contain `Event` protocol buffers constructed when you
    call one of the following functions: `add_summary()`, `add_session_log()`,
    `add_event()`, or `add_graph()`.

    If you pass a `Graph` to the constructor it is added to
    the event file. (This is equivalent to calling `add_graph()` later).

    TensorBoard will pick the graph from the file and display it graphically so
    you can interactively explore the graph you built. You will usually pass
    the graph from the session in which you launched it:

    ```python
    ...create a graph...
    # Launch the graph in a session.
    sess = tf.Session()
    # Create a summary writer, add the 'graph' to the event file.
    writer = tf.summary.FileWriter(<some-directory>, sess.graph)
    ```


    Args:
      event_writer: An EventWriter. Implements add_event method.
      graph: A `Graph` object, such as `sess.graph`.
      graph_def: DEPRECATED: Use the `graph` argument instead.
    """
    self.event_writer = event_writer
    # For storing used tags for session.run() outputs.
    self._session_run_tags = {}
    if graph is not None or graph_def is not None:
      # Calling it with both graph and graph_def for backward compatibility.
      self.add_graph(graph=graph, graph_def=graph_def)
      # Also export the meta_graph_def in this case.
      # graph may itself be a graph_def due to positional arguments
      maybe_graph_as_def = (
          graph.as_graph_def(add_shapes=True) if isinstance(graph, ops.Graph)
          else graph)
      self.add_meta_graph(
          meta_graph.create_meta_graph_def(
              graph_def=graph_def or maybe_graph_as_def)) 

# 需要導入模塊: from tensorflow.python.framework import meta_graph [as 別名]
# 或者: from tensorflow.python.framework.meta_graph import create_meta_graph_def [as 別名]
def testChiefCanWriteEvents(self):
    logdir = _test_dir("can_write")
    with tf.Graph().as_default():
      tf.summary.scalar("c1", tf.constant(1))
      tf.summary.scalar("c2", tf.constant(2))
      tf.summary.scalar("c3", tf.constant(3))
      summ = tf.summary.merge_all()
      sv = tf.train.Supervisor(is_chief=True, logdir=logdir, summary_op=None)
      meta_graph_def = meta_graph.create_meta_graph_def()
      sess = sv.prepare_or_wait_for_session("")
      sv.summary_computed(sess, sess.run(summ))
      sess.close()
      # Wait to make sure everything is written to file before stopping.
      time.sleep(1)
      sv.stop()

    rr = _summary_iterator(logdir)

    # The first event should list the file_version.
    ev = next(rr)
    self.assertEquals("brain.Event:2", ev.file_version)

    # The next one has the graph.
    ev = next(rr)
    ev_graph = tf.GraphDef()
    ev_graph.ParseFromString(ev.graph_def)
    self.assertProtoEquals(sess.graph.as_graph_def(add_shapes=True), ev_graph)

    # Stored MetaGraphDef
    ev = next(rr)
    ev_meta_graph = meta_graph_pb2.MetaGraphDef()
    ev_meta_graph.ParseFromString(ev.meta_graph_def)
    self.assertProtoEquals(meta_graph_def, ev_meta_graph)
    self.assertProtoEquals(
        sess.graph.as_graph_def(add_shapes=True), ev_meta_graph.graph_def)
    # The next one should have the values from the summary.
    ev = next(rr)
    self.assertProtoEquals("""
      value { tag: 'c1' simple_value: 1.0 }
      value { tag: 'c2' simple_value: 2.0 }
      value { tag: 'c3' simple_value: 3.0 }
      """, ev.summary)

    # The next one should be a stop message if we closed cleanly.
    ev = next(rr)
    self.assertEquals(tf.SessionLog.STOP, ev.session_log.status)

    # We should be done.
    self.assertRaises(StopIteration, lambda: next(rr)) 

# 需要導入模塊: from tensorflow.python.framework import meta_graph [as 別名]
# 或者: from tensorflow.python.framework.meta_graph import create_meta_graph_def [as 別名]
def testNoLogdirButExplicitSummaryWriter(self):
    logdir = _test_dir("explicit_summary_writer")
    with tf.Graph().as_default():
      tf.summary.scalar("c1", tf.constant(1))
      tf.summary.scalar("c2", tf.constant(2))
      tf.summary.scalar("c3", tf.constant(3))
      summ = tf.summary.merge_all()
      sw = tf.train.SummaryWriter(logdir)
      sv = tf.train.Supervisor(logdir="", summary_op=None, summary_writer=sw)
      meta_graph_def = meta_graph.create_meta_graph_def()
      sess = sv.prepare_or_wait_for_session("")
      sv.summary_computed(sess, sess.run(summ))
      sess.close()
      # Wait to make sure everything is written to file before stopping.
      time.sleep(1)
      sv.stop()

    # Check the summary was written to 'logdir'
    rr = _summary_iterator(logdir)

    # The first event should list the file_version.
    ev = next(rr)
    self.assertEquals("brain.Event:2", ev.file_version)

    # The next one has the graph.
    ev = next(rr)
    ev_graph = tf.GraphDef()
    ev_graph.ParseFromString(ev.graph_def)
    self.assertProtoEquals(sess.graph.as_graph_def(add_shapes=True), ev_graph)

    # Stored MetaGraphDef
    ev = next(rr)
    ev_meta_graph = meta_graph_pb2.MetaGraphDef()
    ev_meta_graph.ParseFromString(ev.meta_graph_def)
    self.assertProtoEquals(meta_graph_def, ev_meta_graph)
    self.assertProtoEquals(
        sess.graph.as_graph_def(add_shapes=True), ev_meta_graph.graph_def)

    # The next one should have the values from the summary.
    ev = next(rr)
    self.assertProtoEquals("""
      value { tag: 'c1' simple_value: 1.0 }
      value { tag: 'c2' simple_value: 2.0 }
      value { tag: 'c3' simple_value: 3.0 }
      """, ev.summary)

    # The next one should be a stop message if we closed cleanly.
    ev = next(rr)
    self.assertEquals(tf.SessionLog.STOP, ev.session_log.status)

    # We should be done.
    self.assertRaises(StopIteration, lambda: next(rr)) 

# 需要導入模塊: from tensorflow.python.framework import meta_graph [as 別名]
# 或者: from tensorflow.python.framework.meta_graph import create_meta_graph_def [as 別名]
def testStandardServicesWithoutGlobalStep(self):
    logdir = _test_dir("standard_services_without_global_step")
    # Create a checkpoint.
    with tf.Graph().as_default():
      v = tf.Variable([1.0], name="foo")
      tf.summary.scalar("v", v[0])
      sv = tf.train.Supervisor(logdir=logdir)
      meta_graph_def = meta_graph.create_meta_graph_def(
          saver_def=sv.saver.saver_def)
      sess = sv.prepare_or_wait_for_session("")
      save_path = sv.save_path
      self._wait_for_glob(save_path, 3.0)
      self._wait_for_glob(
          os.path.join(logdir, "*events*"), 3.0, for_checkpoint=False)
      # Wait to make sure everything is written to file before stopping.
      time.sleep(1)
      sv.stop()
    # There should be an event file with a version number.
    rr = _summary_iterator(logdir)
    ev = next(rr)
    self.assertEquals("brain.Event:2", ev.file_version)
    ev = next(rr)
    ev_graph = tf.GraphDef()
    ev_graph.ParseFromString(ev.graph_def)
    self.assertProtoEquals(sess.graph.as_graph_def(add_shapes=True), ev_graph)

    # Stored MetaGraphDef
    ev = next(rr)
    ev_meta_graph = meta_graph_pb2.MetaGraphDef()
    ev_meta_graph.ParseFromString(ev.meta_graph_def)
    self.assertProtoEquals(meta_graph_def, ev_meta_graph)
    self.assertProtoEquals(
        sess.graph.as_graph_def(add_shapes=True), ev_meta_graph.graph_def)

    ev = next(rr)
    self.assertProtoEquals("value { tag: 'v' simple_value: 1.0 }", ev.summary)

    ev = next(rr)
    self.assertEquals(tf.SessionLog.STOP, ev.session_log.status)

    self.assertRaises(StopIteration, lambda: next(rr))
    # There should be a checkpoint file with the variable "foo"
    with tf.Graph().as_default(), self.test_session() as sess:
      v = tf.Variable([10.10], name="foo")
      sav = tf.train.Saver([v])
      sav.restore(sess, save_path)
      self.assertEqual(1.0, v.eval()[0])

  # Same as testStandardServicesNoGlobalStep but with a global step.
  # We should get a summary about the step time. 

# 需要導入模塊: from tensorflow.python.framework import meta_graph [as 別名]
# 或者: from tensorflow.python.framework.meta_graph import create_meta_graph_def [as 別名]
def testStandardServicesWithGlobalStep(self):
    logdir = _test_dir("standard_services_with_global_step")
    # Create a checkpoint.
    with tf.Graph().as_default():
      v = tf.Variable([123], name="global_step")
      sv = tf.train.Supervisor(logdir=logdir)
      meta_graph_def = meta_graph.create_meta_graph_def(
          saver_def=sv.saver.saver_def)
      sess = sv.prepare_or_wait_for_session("")
      # This is where the checkpoint will appear, with step number 123.
      save_path = "%s-123" % sv.save_path
      self._wait_for_glob(save_path, 3.0)
      self._wait_for_glob(
          os.path.join(logdir, "*events*"), 3.0, for_checkpoint=False)
      # Wait to make sure everything is written to file before stopping.
      time.sleep(1)
      sv.stop()
    # There should be an event file with a version number.
    rr = _summary_iterator(logdir)
    ev = next(rr)
    self.assertEquals("brain.Event:2", ev.file_version)
    ev = next(rr)
    ev_graph = tf.GraphDef()
    ev_graph.ParseFromString(ev.graph_def)
    self.assertProtoEquals(sess.graph.as_graph_def(add_shapes=True), ev_graph)
    ev = next(rr)
    ev_meta_graph = meta_graph_pb2.MetaGraphDef()
    ev_meta_graph.ParseFromString(ev.meta_graph_def)
    self.assertProtoEquals(meta_graph_def, ev_meta_graph)
    self.assertProtoEquals(
        sess.graph.as_graph_def(add_shapes=True), ev_meta_graph.graph_def)
    ev = next(rr)
    # It is actually undeterministic whether SessionLog.START gets written
    # before the summary or the checkpoint, but this works when run 10000 times.
    self.assertEquals(123, ev.step)
    self.assertEquals(tf.SessionLog.START, ev.session_log.status)
    first = next(rr)
    second = next(rr)
    # It is undeterministic whether the value gets written before the checkpoint
    # since they are on separate threads, so we check for both conditions.
    if first.HasField("summary"):
      self.assertProtoEquals("""value { tag: 'global_step/sec'
                                        simple_value: 0.0 }""",
                             first.summary)
      self.assertEquals(123, second.step)
      self.assertEquals(tf.SessionLog.CHECKPOINT, second.session_log.status)
    else:
      self.assertEquals(123, first.step)
      self.assertEquals(tf.SessionLog.CHECKPOINT, first.session_log.status)
      self.assertProtoEquals("""value { tag: 'global_step/sec'
                                        simple_value: 0.0 }""",
                             second.summary)
    ev = next(rr)
    self.assertEquals(tf.SessionLog.STOP, ev.session_log.status)
    self.assertRaises(StopIteration, lambda: next(rr))
    # There should be a checkpoint file with the variable "foo"
    with tf.Graph().as_default(), self.test_session() as sess:
      v = tf.Variable([-12], name="global_step")
      sav = tf.train.Saver([v])
      sav.restore(sess, save_path)
      self.assertEqual(123, v.eval()[0]) 

# 需要導入模塊: from tensorflow.python.framework import meta_graph [as 別名]
# 或者: from tensorflow.python.framework.meta_graph import create_meta_graph_def [as 別名]
def __init__(self, event_writer, graph=None, graph_def=None):
    """Creates a `SummaryWriter` and an event file.

    On construction the summary writer creates a new event file in `logdir`.
    This event file will contain `Event` protocol buffers constructed when you
    call one of the following functions: `add_summary()`, `add_session_log()`,
    `add_event()`, or `add_graph()`.

    If you pass a `Graph` to the constructor it is added to
    the event file. (This is equivalent to calling `add_graph()` later).

    TensorBoard will pick the graph from the file and display it graphically so
    you can interactively explore the graph you built. You will usually pass
    the graph from the session in which you launched it:

    ```python
    ...create a graph...
    # Launch the graph in a session.
    sess = tf.Session()
    # Create a summary writer, add the 'graph' to the event file.
    writer = tf.train.SummaryWriter(<some-directory>, sess.graph)
    ```


    Args:
      event_writer: An EventWriter. Implements add_event method.
      graph: A `Graph` object, such as `sess.graph`.
      graph_def: DEPRECATED: Use the `graph` argument instead.
    """
    self.event_writer = event_writer
    # For storing used tags for session.run() outputs.
    self._session_run_tags = {}
    if graph is not None or graph_def is not None:
      # Calling it with both graph and graph_def for backward compatibility.
      self.add_graph(graph=graph, graph_def=graph_def)
      # Also export the meta_graph_def in this case.
      # graph may itself be a graph_def due to positional arguments
      maybe_graph_as_def = (
          graph.as_graph_def(add_shapes=True) if isinstance(graph, ops.Graph)
          else graph)
      self.add_meta_graph(
          meta_graph.create_meta_graph_def(
              graph_def=graph_def or maybe_graph_as_def))