Python v1.Graph方法代码示例

# 需要导入模块: from tensorflow.compat import v1 [as 别名]
# 或者: from tensorflow.compat.v1 import Graph [as 别名]
def _load_frozen_graph(self, frozen_graph_path):
    frozen_graph = tf.GraphDef()
    with open(frozen_graph_path, 'rb') as f:
      frozen_graph.ParseFromString(f.read())

    self.graph = tf.Graph()
    with self.graph.as_default():
      self.output_node = tf.import_graph_def(
          frozen_graph, return_elements=[
              'probabilities:0',
          ])
    self.session = tf.InteractiveSession(graph=self.graph)

    tf_probabilities = self.graph.get_tensor_by_name('import/probabilities:0')
    self._output_nodes = [tf_probabilities]
    self.sliding_window = None
    self.frames_since_last_inference = self.config.inference_rate
    self.last_annotations = [] 

# 需要导入模块: from tensorflow.compat import v1 [as 别名]
# 或者: from tensorflow.compat.v1 import Graph [as 别名]
def _benchmark_train(self):
    """Run cnn in benchmark mode. Skip the backward pass if forward_only is on.

    Returns:
      Dictionary containing training statistics (num_workers, num_steps,
      average_wall_time, images_per_sec).
    """
    graph = tf.Graph()
    with graph.as_default():
      build_result = self._build_graph()
      if self.mode == constants.BenchmarkMode.TRAIN_AND_EVAL:
        with self.variable_mgr.reuse_variables():
          with tf.name_scope('Evaluation') as ns:
            eval_build_results = self._build_eval_graph(ns)
      else:
        eval_build_results = None
    (graph, result_to_benchmark) = self._preprocess_graph(graph, build_result)
    with graph.as_default():
      return self._benchmark_graph(result_to_benchmark, eval_build_results) 

# 需要导入模块: from tensorflow.compat import v1 [as 别名]
# 或者: from tensorflow.compat.v1 import Graph [as 别名]
def evaluate(self, env_fn, hparams, sampling_temp):
    with tf.Graph().as_default():
      with tf.name_scope("rl_eval"):
        eval_env = env_fn(in_graph=True)
        (collect_memory, _, collect_init) = _define_collect(
            eval_env,
            hparams,
            "ppo_eval",
            eval_phase=True,
            frame_stack_size=self.frame_stack_size,
            force_beginning_resets=False,
            sampling_temp=sampling_temp,
            distributional_size=self._distributional_size,
        )
        model_saver = tf.train.Saver(
            tf.global_variables(hparams.policy_network + "/.*")
            # tf.global_variables("clean_scope.*")  # Needed for sharing params.
        )

        with tf.Session() as sess:
          sess.run(tf.global_variables_initializer())
          collect_init(sess)
          trainer_lib.restore_checkpoint(self.agent_model_dir, model_saver,
                                         sess)
          sess.run(collect_memory) 

# 需要导入模块: from tensorflow.compat import v1 [as 别名]
# 或者: from tensorflow.compat.v1 import Graph [as 别名]
def __init__(self, hparams, action_space, observation_space, policy_dir):
    assert hparams.base_algo == "ppo"
    ppo_hparams = trainer_lib.create_hparams(hparams.base_algo_params)

    frame_stack_shape = (1, hparams.frame_stack_size) + observation_space.shape
    self._frame_stack = np.zeros(frame_stack_shape, dtype=np.uint8)

    with tf.Graph().as_default():
      self.obs_t = tf.placeholder(shape=self.frame_stack_shape, dtype=np.uint8)
      self.logits_t, self.value_function_t = get_policy(
          self.obs_t, ppo_hparams, action_space
      )
      model_saver = tf.train.Saver(
          tf.global_variables(scope=ppo_hparams.policy_network + "/.*")  # pylint: disable=unexpected-keyword-arg
      )
      self.sess = tf.Session()
      self.sess.run(tf.global_variables_initializer())
      trainer_lib.restore_checkpoint(policy_dir, model_saver,
                                     self.sess) 

# 需要导入模块: from tensorflow.compat import v1 [as 别名]
# 或者: from tensorflow.compat.v1 import Graph [as 别名]
def __init__(
      self, batch_size, observation_space, action_space, policy_hparams,
      policy_dir, sampling_temp
  ):
    super(PolicyAgent, self).__init__(
        batch_size, observation_space, action_space
    )
    self._sampling_temp = sampling_temp
    with tf.Graph().as_default():
      self._observations_t = tf.placeholder(
          shape=((batch_size,) + self.observation_space.shape),
          dtype=self.observation_space.dtype
      )
      (logits, self._values_t) = rl.get_policy(
          self._observations_t, policy_hparams, self.action_space
      )
      actions = common_layers.sample_with_temperature(logits, sampling_temp)
      self._probs_t = tf.nn.softmax(logits / sampling_temp)
      self._actions_t = tf.cast(actions, tf.int32)
      model_saver = tf.train.Saver(
          tf.global_variables(policy_hparams.policy_network + "/.*")  # pylint: disable=unexpected-keyword-arg
      )
      self._sess = tf.Session()
      self._sess.run(tf.global_variables_initializer())
      trainer_lib.restore_checkpoint(policy_dir, model_saver, self._sess) 

# 需要导入模块: from tensorflow.compat import v1 [as 别名]
# 或者: from tensorflow.compat.v1 import Graph [as 别名]
def run_in_graph_mode_only(func=None, config=None, use_gpu=True):
  """Runs a test in graph mode only, when eager is enabled by default."""
  def decorator(f):
    """Decorator for a method."""
    def decorated(self, *args, **kwargs):
      """Run the decorated test method."""
      self.tearDown()
      # Run in graph mode block
      with tf.Graph().as_default():
        self.setUp()
        with self.test_session(use_gpu=use_gpu, config=config):
          f(self, *args, **kwargs)

    return decorated

  if func is not None:
    return decorator(func)

  return decorator 

# 需要导入模块: from tensorflow.compat import v1 [as 别名]
# 或者: from tensorflow.compat.v1 import Graph [as 别名]
def test_invertibility(self, op, name, dropout=0.0):
    with tf.Graph().as_default():
      tf.set_random_seed(42)
      x = tf.random_uniform(shape=(16, 32, 32, 4))

      if op in [glow_ops.affine_coupling, glow_ops.additive_coupling]:
        with arg_scope([glow_ops.get_dropout], init=False):
          x_inv, _ = op(name, x, reverse=False, dropout=dropout)
          x_inv_inv, _ = op(name, x_inv, reverse=True, dropout=dropout)
      else:
        x_inv, _ = op(name, x, reverse=False)
        x_inv_inv, _ = op(name, x_inv, reverse=True)
      with tf.Session() as session:
        session.run(tf.global_variables_initializer())
        diff = session.run(x - x_inv_inv)
        self.assertTrue(np.allclose(diff, 0.0, atol=1e-5)) 

# 需要导入模块: from tensorflow.compat import v1 [as 别名]
# 或者: from tensorflow.compat.v1 import Graph [as 别名]
def test_conv2d(self):
    with tf.Graph().as_default():
      x = 10.0 * tf.random_uniform(shape=(16, 5, 5, 32))

      with arg_scope([glow_ops.actnorm], init=True):
        actnorm_conv2d = glow_ops.conv(
            "actnorm_conv2d", x, output_channels=64, apply_actnorm=True)
        actnorm_zeros2d = glow_ops.conv(
            "actnorm_zeros2d", x, output_channels=64, apply_actnorm=False)

      with tf.Session() as session:
        session.run(tf.global_variables_initializer())

        # test if apply_actnorm is set to True, the first minibatch has
        # zero mean and unit variance.
        actnorm_np, zeros_np = session.run([actnorm_conv2d, actnorm_zeros2d])
        self.assertEqual(actnorm_np.shape, (16, 5, 5, 64))
        mean = np.mean(actnorm_np, axis=(0, 1, 2))
        var = np.var(actnorm_np, axis=(0, 1, 2))
        self.assertTrue(np.allclose(mean, 0.0, atol=1e-5))
        self.assertTrue(np.allclose(var, 1.0, atol=1e-5))

        # test shape in case apply_actnorm is set to False,
        self.assertEqual(zeros_np.shape, (16, 5, 5, 64)) 

# 需要导入模块: from tensorflow.compat import v1 [as 别名]
# 或者: from tensorflow.compat.v1 import Graph [as 别名]
def test_temporal_latent_to_dist(self, apply_dilation, activation,
                                   dropout=0.0, noise=0.1, num_steps=5):
    with tf.Graph().as_default():
      hparams = self.get_glow_hparams()
      hparams.latent_apply_dilations = apply_dilation
      hparams.latent_activation = activation
      hparams.latent_dropout = dropout
      hparams.latent_noise = noise
      latent_shape = (16, num_steps, 32, 32, 48)
      latents = tf.random_normal(latent_shape)
      dist = glow_ops.temporal_latent_to_dist(
          "tensor_to_dist", latents, hparams)
      with tf.Session() as sess:
        sess.run(tf.global_variables_initializer())
        # dilated conv_3d is not available on CPU.
        is_gpu = tf.test.is_gpu_available()
        if not apply_dilation or is_gpu:
          mean, scale = dist.loc, dist.scale
          mean_np, scale_np = sess.run([mean, scale])
          self.assertTrue(np.allclose(mean_np, 0.0))
          self.assertTrue(np.allclose(scale_np, 1.0)) 

# 需要导入模块: from tensorflow.compat import v1 [as 别名]
# 或者: from tensorflow.compat.v1 import Graph [as 别名]
def linear_interpolate_rank(self):
    with tf.Graph().as_default():
      # Since rank is 1, the first channel should remain 1.0.
      # and the second channel should be interpolated between 1.0 and 6.0
      z1 = np.ones(shape=(4, 4, 2))
      z2 = np.copy(z1)
      z2[:, :, 0] += 0.01
      z2[:, :, 1] += 5.0
      coeffs = np.linspace(0.0, 1.0, 11)
      z1 = np.expand_dims(z1, axis=0)
      z2 = np.expand_dims(z2, axis=0)
      tensor1 = tf.convert_to_tensor(z1, dtype=tf.float32)
      tensor2 = tf.convert_to_tensor(z2, dtype=tf.float32)
      lin_interp_max = glow_ops.linear_interpolate_rank(
          tensor1, tensor2, coeffs)
      with tf.Session() as sess:
        lin_interp_np_max = sess.run(lin_interp_max)
        for lin_interp_np, coeff in zip(lin_interp_np_max, coeffs):
          exp_val = 1.0 + coeff * (6.0 - 1.0)
          self.assertTrue(np.allclose(lin_interp_np[:, :, 0], 1.0))
          self.assertTrue(np.allclose(lin_interp_np[:, :, 1], exp_val)) 

# 需要导入模块: from tensorflow.compat import v1 [as 别名]
# 或者: from tensorflow.compat.v1 import Graph [as 别名]
def testVarNames(self):
    with tf.Graph().as_default():
      model, features = get_model(
          mode=tf.estimator.ModeKeys.PREDICT,
          model_cls=transformer.TransformerScorer)
      _ = model.infer(features)
      scorer_vars = [v.name for v in tf.global_variables()]

    with tf.Graph().as_default():
      model, features = get_model(
          mode=tf.estimator.ModeKeys.EVAL,
          model_cls=transformer.TransformerScorer)
      _ = model(features)
      scorer_eval_vars = [v.name for v in tf.global_variables()]

    with tf.Graph().as_default():
      model, features = get_model(
          mode=tf.estimator.ModeKeys.EVAL,
          model_cls=transformer.Transformer)
      _ = model(features)
      transformer_vars = [v.name for v in tf.global_variables()]

    self.assertEqual(sorted(scorer_vars), sorted(transformer_vars))
    self.assertEqual(sorted(scorer_eval_vars), sorted(transformer_vars)) 

# 需要导入模块: from tensorflow.compat import v1 [as 别名]
# 或者: from tensorflow.compat.v1 import Graph [as 别名]
def testSpectralNorm(self):
    # Test that after 20 calls to apply_spectral_norm, the spectral
    # norm of the normalized matrix is close to 1.0
    with tf.Graph().as_default():
      weights = tf.get_variable("w", dtype=tf.float32, shape=[2, 3, 50, 100])
      weights = tf.multiply(weights, 10.0)
      normed_weight, assign_op = common_layers.apply_spectral_norm(weights)

      with tf.Session() as sess:
        sess.run(tf.global_variables_initializer())

        for _ in range(20):
          sess.run(assign_op)
          normed_weight, assign_op = common_layers.apply_spectral_norm(
              weights)
        normed_weight = sess.run(normed_weight).reshape(-1, 100)
        _, s, _ = np.linalg.svd(normed_weight)
        self.assertTrue(np.allclose(s[0], 1.0, rtol=0.1)) 

# 需要导入模块: from tensorflow.compat import v1 [as 别名]
# 或者: from tensorflow.compat.v1 import Graph [as 别名]
def __init__(self, batch_size, *args, **kwargs):
    self._store_rollouts = kwargs.pop("store_rollouts", True)

    super(T2TEnv, self).__init__(*args, **kwargs)

    self.batch_size = batch_size
    self._rollouts_by_epoch_and_split = collections.OrderedDict()
    self.current_epoch = None
    self._should_preprocess_on_reset = True
    with tf.Graph().as_default() as tf_graph:
      self._tf_graph = _Noncopyable(tf_graph)
      self._decoded_image_p = _Noncopyable(
          tf.placeholder(dtype=tf.uint8, shape=(None, None, None))
      )
      self._encoded_image_t = _Noncopyable(
          tf.image.encode_png(self._decoded_image_p.obj)
      )
      self._encoded_image_p = _Noncopyable(tf.placeholder(tf.string))
      self._decoded_image_t = _Noncopyable(
          tf.image.decode_png(self._encoded_image_p.obj)
      )
      self._session = _Noncopyable(tf.Session()) 

# 需要导入模块: from tensorflow.compat import v1 [as 别名]
# 或者: from tensorflow.compat.v1 import Graph [as 别名]
def generate_samples(self, data_dir, tmp_dir, dataset_split):
    with tf.Graph().as_default():
      # train and eval set are generated on-the-fly.
      # test set is the official test-set.
      if dataset_split == problem.DatasetSplit.TEST:
        moving_ds = self.get_test_iterator(tmp_dir)
      else:
        moving_ds = self.get_train_iterator()

      next_video = moving_ds.get_next()
      with tf.Session() as sess:
        sess.run(moving_ds.initializer)

        n_samples = SPLIT_TO_SIZE[dataset_split]
        for _ in range(n_samples):
          next_video_np = sess.run(next_video)
          for frame_number, frame in enumerate(next_video_np):
            yield {
                "frame_number": [frame_number],
                "frame": frame,
            } 

# 需要导入模块: from tensorflow.compat import v1 [as 别名]
# 或者: from tensorflow.compat.v1 import Graph [as 别名]
def export_module_spec_with_checkpoint(module_spec,
                                       checkpoint_path,
                                       export_path,
                                       scope_prefix=""):
  """Exports given checkpoint as tfhub module with given spec."""

  # The main requirement is that it is possible to know how to map from
  # module variable name to checkpoint variable name.
  # This is trivial if the original code used variable scopes,
  # but can be messy if the variables to export are interwined
  # with variables not export.
  with tf.Graph().as_default():
    m = hub.Module(module_spec)
    assign_map = {
        scope_prefix + name: value for name, value in m.variable_map.items()
    }
    tf.train.init_from_checkpoint(checkpoint_path, assign_map)
    init_op = tf.initializers.global_variables()
    with tf.Session() as session:
      session.run(init_op)
      m.export(export_path, session)