Python absl.flags.DEFINE_boolean() Examples

def define_flags(specs=None):
  """Define a command line flag for each ParamSpec in flags.param_specs."""
  specs = specs or param_specs
  define_flag = {
      'boolean': absl_flags.DEFINE_boolean,
      'float': absl_flags.DEFINE_float,
      'integer': absl_flags.DEFINE_integer,
      'string': absl_flags.DEFINE_string,
      'enum': absl_flags.DEFINE_enum,
      'list': absl_flags.DEFINE_list
  }
  for name, param_spec in six.iteritems(specs):
    if param_spec.flag_type not in define_flag:
      raise ValueError('Unknown flag_type %s' % param_spec.flag_type)
    else:
      define_flag[param_spec.flag_type](name, param_spec.default_value,
                                        help=param_spec.description,
                                        **param_spec.kwargs) 

def _define_flags(options):
  for option in _filter_options(options):
    name, type_, default_ = option[:3]
    if type_ == 'boolean':
      flags.DEFINE_boolean(name, default_, '')
    elif type_ == 'integer':
      flags.DEFINE_integer(name, default_, '')
    elif type_ == 'float':
      flags.DEFINE_float(name, default_, '')
    elif type_ == 'string':
      flags.DEFINE_string(name, default_, '')
    elif type_ == 'list_of_ints':
      flags.DEFINE_string(name, default_, '')
    else:
      assert 'Unexpected option type %s' % type_


# Define command-line flags for all options (unless `internal`). 

def define_wide_deep_flags():
  """Add supervised learning flags, as well as wide-deep model type."""
  flags_core.define_base()
  flags_core.define_benchmark()
  flags_core.define_performance(
      num_parallel_calls=False, inter_op=True, intra_op=True,
      synthetic_data=False, max_train_steps=False, dtype=False,
      all_reduce_alg=False)

  flags.adopt_module_key_flags(flags_core)

  flags.DEFINE_enum(
      name="model_type", short_name="mt", default="wide_deep",
      enum_values=['wide', 'deep', 'wide_deep'],
      help="Select model topology.")
  flags.DEFINE_boolean(
      name="download_if_missing", default=True, help=flags_core.help_wrap(
          "Download data to data_dir if it is not already present.")) 

def define_wide_deep_flags():
  """Add supervised learning flags, as well as wide-deep model type."""
  flags_core.define_base()
  flags_core.define_benchmark()
  flags_core.define_performance(
      num_parallel_calls=False, inter_op=True, intra_op=True,
      synthetic_data=False, max_train_steps=False, dtype=False,
      all_reduce_alg=False)

  flags.adopt_module_key_flags(flags_core)

  flags.DEFINE_enum(
      name="model_type", short_name="mt", default="wide_deep",
      enum_values=['wide', 'deep', 'wide_deep'],
      help="Select model topology.")
  flags.DEFINE_boolean(
      name="download_if_missing", default=True, help=flags_core.help_wrap(
          "Download data to data_dir if it is not already present.")) 

def define_flags():
  """Define a command line flag for each ParamSpec in flags.param_specs."""
  define_flag = {
      'boolean': absl_flags.DEFINE_boolean,
      'float': absl_flags.DEFINE_float,
      'integer': absl_flags.DEFINE_integer,
      'string': absl_flags.DEFINE_string,
      'enum': absl_flags.DEFINE_enum,
      'list': absl_flags.DEFINE_list
  }
  for name, param_spec in six.iteritems(param_specs):
    if param_spec.flag_type not in define_flag:
      raise ValueError('Unknown flag_type %s' % param_spec.flag_type)
    else:
      define_flag[param_spec.flag_type](name, param_spec.default_value,
                                        help=param_spec.description,
                                        **param_spec.kwargs) 

def define_wide_deep_flags():
  """Add supervised learning flags, as well as wide-deep model type."""
  flags_core.define_base(clean=True, train_epochs=True,
                         epochs_between_evals=True)
  flags_core.define_benchmark()
  flags_core.define_performance(
      num_parallel_calls=False, inter_op=True, intra_op=True,
      synthetic_data=False, max_train_steps=False, dtype=False,
      all_reduce_alg=False)

  flags.adopt_module_key_flags(flags_core)

  flags.DEFINE_enum(
      name="model_type", short_name="mt", default="wide_deep",
      enum_values=['wide', 'deep', 'wide_deep'],
      help="Select model topology.")
  flags.DEFINE_boolean(
      name="download_if_missing", default=True, help=flags_core.help_wrap(
          "Download data to data_dir if it is not already present.")) 

def duplicate_flags(flagnames=None):
  """Returns a new FlagValues object with the requested flagnames.

  Used to test DuplicateFlagError detection.

  Args:
    flagnames: str, A list of flag names to create.

  Returns:
    A FlagValues object with one boolean flag for each name in flagnames.
  """
  flag_values = flags.FlagValues()
  for name in flagnames:
    flags.DEFINE_boolean(name, False, 'Flag named %s' % (name,),
                         flag_values=flag_values)
  return flag_values 

def define_flags(flag_values=FLAGS):
  """Defines some flags.

  Args:
    flag_values: The FlagValues object we want to register the flags
      with.
  """
  # The 'tmod_bar_' prefix (short for 'test_module_bar') ensures there
  # is no name clash with the existing flags.
  flags.DEFINE_boolean('tmod_bar_x', True, 'Boolean flag.',
                       flag_values=flag_values)
  flags.DEFINE_string('tmod_bar_y', 'default', 'String flag.',
                      flag_values=flag_values)
  flags.DEFINE_boolean('tmod_bar_z', False,
                       'Another boolean flag from module bar.',
                       flag_values=flag_values)
  flags.DEFINE_integer('tmod_bar_t', 4, 'Sample int flag.',
                       flag_values=flag_values)
  flags.DEFINE_integer('tmod_bar_u', 5, 'Sample int flag.',
                       flag_values=flag_values)
  flags.DEFINE_integer('tmod_bar_v', 6, 'Sample int flag.',
                       flag_values=flag_values) 

def define_flags():
  """Define the flags for the Shakespeare character LSTM."""
  flags_core.define_base(data_dir=False,
                         clean=False,
                         train_epochs=True,
                         epochs_between_evals=False,
                         stop_threshold=False,
                         hooks=False,
                         export_dir=False,
                         run_eagerly=True)

  flags_core.define_performance(num_parallel_calls=False,
                                inter_op=False,
                                intra_op=False,
                                synthetic_data=False,
                                max_train_steps=False,
                                dtype=False,
                                enable_xla=True,
                                force_v2_in_keras_compile=True)

  flags_core.set_defaults(train_epochs=43,
                          batch_size=64)

  flags.DEFINE_boolean(name='enable_eager', default=True, help='Enable eager?')
  flags.DEFINE_boolean(
      name='train', default=True,
      help='If true trains the model.')
  flags.DEFINE_string(
      name='predict_context', default=None,
      help='If set, makes a prediction with the given context.')
  flags.DEFINE_integer(
      name='predict_length', default=1000,
      help='Length of the predicted text including the context.')
  flags.DEFINE_integer(
      name='log_steps', default=100,
      help='For every log_steps, we log the timing information such as '
      'examples per second.')
  flags.DEFINE_string(
      name='training_data', default=None,
      help='Path to file containing the training data.')
  flags.DEFINE_boolean(name='cudnn', default=True, help='Use CuDNN LSTM.') 

def define_keras_flags():
  flags.DEFINE_boolean(name='enable_eager', default=False, help='Enable eager?')
  flags.DEFINE_boolean(name='skip_eval', default=False, help='Skip evaluation?')
  flags.DEFINE_integer(
      name='train_steps', default=None,
      help='The number of steps to run for training. If it is larger than '
      '# batches per epoch, then use # batches per epoch. When this flag is '
      'set, only one epoch is going to run for training.')
  flags.DEFINE_integer(
      name='log_steps', default=100,
      help='For every log_steps, we log the timing information such as '
      'examples per second. Besides, for every log_steps, we store the '
      'timestamp of a batch end.') 

def DEFINE_boolean(
  name: str,
  default: Optional[bool],
  help: str,
  required: bool = False,
  validator: Callable[[bool], bool] = None,
):
  """Registers a flag whose value must be a boolean."""
  absl_flags.DEFINE_boolean(
    name, default, help, module_name=get_calling_module_name(),
  )
  if required:
    absl_flags.mark_flag_as_required(name)
  if validator:
    RegisterFlagValidator(name, validator) 

def DEFINE_boolean(name, default, help):  # pylint: disable=invalid-name,redefined-builtin
  param_specs[name] = ParamSpec('boolean', default, help, {}) 

def define_ctl_flags():
  """Define flags for CTL."""

  flags.DEFINE_boolean(name='use_tf_function', default=True,
                       help='Wrap the train and test step inside a '
                       'tf.function.')
  flags.DEFINE_boolean(name='single_l2_loss_op', default=False,
                       help='Calculate L2_loss on concatenated weights, '
                       'instead of using Keras per-layer L2 loss.') 

def DEFINE_boolean(name, default, help):  # pylint: disable=invalid-name,redefined-builtin
  param_specs[name] = ParamSpec('boolean', default, help, {}) 

def test_find_module_defining_flag_passing_module_name(self):
    my_flags = flags.FlagValues()
    module_name = sys.__name__  # Must use an existing module.
    flags.DEFINE_boolean('flag_name', True,
                         'Flag with a different module name.',
                         flag_values=my_flags,
                         module_name=module_name)
    self.assertEqual(module_name,
                     my_flags.find_module_defining_flag('flag_name')) 

def setUp(self):
    self.flag_values = flags.FlagValues()
    flags.DEFINE_string('unittest_message1', 'Foo!', 'You Add Here.',
                        flag_values=self.flag_values)
    flags.DEFINE_string('unittest_message2', 'Bar!', 'Hello, Sailor!',
                        flag_values=self.flag_values)
    flags.DEFINE_boolean('unittest_boolflag', 0, 'Some Boolean thing',
                         flag_values=self.flag_values)
    flags.DEFINE_integer('unittest_number', 12345, 'Some integer',
                         lower_bound=0, flag_values=self.flag_values)
    flags.DEFINE_list('UnitTestList', '1,2,3', 'Some list',
                      flag_values=self.flag_values)
    self.tmp_path = None
    self.flag_values.mark_as_parsed() 

def define_flags(flag_values=FLAGS):
  """Defines a few flags."""
  module_bar.define_flags(flag_values=flag_values)
  # The 'tmod_foo_' prefix (short for 'test_module_foo') ensures that we
  # have no name clash with existing flags.
  flags.DEFINE_boolean('tmod_foo_bool', True, 'Boolean flag from module foo.',
                       flag_values=flag_values)
  flags.DEFINE_string('tmod_foo_str', 'default', 'String flag.',
                      flag_values=flag_values)
  flags.DEFINE_integer('tmod_foo_int', 3, 'Sample int flag.',
                       flag_values=flag_values) 

def test_find_module_id_defining_flag_passing_module_name(self):
    my_flags = flags.FlagValues()
    module_name = sys.__name__  # Must use an existing module.
    flags.DEFINE_boolean('flag_name', True,
                         'Flag with a different module name.',
                         flag_values=my_flags,
                         module_name=module_name)
    self.assertEqual(id(sys),
                     my_flags.find_module_id_defining_flag('flag_name')) 

def define_keras_flags():
  flags.DEFINE_boolean(name='enable_eager', default=False, help='Enable eager?')
  flags.DEFINE_boolean(name='skip_eval', default=False, help='Skip evaluation?')
  flags.DEFINE_integer(
      name='train_steps', default=None,
      help='The number of steps to run for training. If it is larger than '
      '# batches per epoch, then use # batches per epoch. When this flag is '
      'set, only one epoch is going to run for training.')
  flags.DEFINE_integer(
      name='log_steps', default=100,
      help='For every log_steps, we log the timing information such as '
      'examples per second. Besides, for every log_steps, we store the '
      'timestamp of a batch end.') 

def define_flags():
    """Define a command line flag for each ParamSpec in _DEFAULT_PARAMS."""
    define_flag = {
        'boolean': flags.DEFINE_boolean,
        'float': flags.DEFINE_float,
        'integer': flags.DEFINE_integer,
        'string': flags.DEFINE_string,
    }
    for name, param_spec in six.iteritems(_DEFAULT_PARAMS):
        if param_spec.flag_type not in define_flag:
            raise ValueError('Unknown flag_type %s' % param_spec.flag_type)
        else:
            define_flag[param_spec.flag_type](name, param_spec.default_value,
                                              param_spec.description)
            flags.declare_key_flag(name) 

def define_keras_flags():
  flags.DEFINE_boolean(name='enable_eager', default=False, help='Enable eager?')
  flags.DEFINE_boolean(name='skip_eval', default=False, help='Skip evaluation?')
  flags.DEFINE_integer(
      name='train_steps', default=None,
      help='The number of steps to run for training. If it is larger than '
      '# batches per epoch, then use # batches per epoch. When this flag is '
      'set, only one epoch is going to run for training.')
  flags.DEFINE_integer(
      name='log_steps', default=100,
      help='For every log_steps, we log the timing information such as '
      'examples per second. Besides, for every log_steps, we store the '
      'timestamp of a batch end.') 

def define_resnet_flags(resnet_size_choices=None):
  """Add flags and validators for ResNet."""
  flags_core.define_base()
  flags_core.define_performance(num_parallel_calls=False)
  flags_core.define_image()
  flags_core.define_benchmark()
  flags.adopt_module_key_flags(flags_core)

  flags.DEFINE_enum(
      name='resnet_version', short_name='rv', default='1',
      enum_values=['1', '2'],
      help=flags_core.help_wrap(
          'Version of ResNet. (1 or 2) See README.md for details.'))
  flags.DEFINE_bool(
      name='fine_tune', short_name='ft', default=False,
      help=flags_core.help_wrap(
          'If True do not train any parameters except for the final layer.'))
  flags.DEFINE_string(
      name='pretrained_model_checkpoint_path', short_name='pmcp', default=None,
      help=flags_core.help_wrap(
          'If not None initialize all the network except the final layer with '
          'these values'))
  flags.DEFINE_boolean(
      name='eval_only', default=False,
      help=flags_core.help_wrap('Skip training and only perform evaluation on '
                                'the latest checkpoint.'))

  choice_kwargs = dict(
      name='resnet_size', short_name='rs', default='50',
      help=flags_core.help_wrap('The size of the ResNet model to use.'))

  if resnet_size_choices is None:
    flags.DEFINE_string(**choice_kwargs)
  else:
    flags.DEFINE_enum(enum_values=resnet_size_choices, **choice_kwargs) 

def define_keras_benchmark_flags():
  """Add flags for keras built-in application models."""
  flags_core.define_base(hooks=False)
  flags_core.define_performance()
  flags_core.define_image()
  flags_core.define_benchmark()
  flags.adopt_module_key_flags(flags_core)

  flags_core.set_defaults(
      data_format="channels_last",
      use_synthetic_data=True,
      batch_size=32,
      train_epochs=2)

  flags.DEFINE_enum(
      name="model", default=None,
      enum_values=MODELS.keys(), case_sensitive=False,
      help=flags_core.help_wrap(
          "Model to be benchmarked."))

  flags.DEFINE_integer(
      name="num_train_images", default=1000,
      help=flags_core.help_wrap(
          "The number of synthetic images for training. The default value is "
          "1000."))

  flags.DEFINE_integer(
      name="num_eval_images", default=50,
      help=flags_core.help_wrap(
          "The number of synthetic images for evaluation. The default value is "
          "50."))

  flags.DEFINE_boolean(
      name="eager", default=False, help=flags_core.help_wrap(
          "To enable eager execution. Note that if eager execution is enabled, "
          "only one GPU is utilized even if multiple GPUs are provided and "
          "multi_gpu_model is used."))

  flags.DEFINE_boolean(
      name="dist_strat", default=False, help=flags_core.help_wrap(
          "To enable distribution strategy for model training and evaluation. "
          "Number of GPUs used for distribution strategy can be set by the "
          "argument --num_gpus."))

  flags.DEFINE_list(
      name="callbacks",
      default=["ExamplesPerSecondCallback", "LoggingMetricCallback"],
      help=flags_core.help_wrap(
          "A list of (case insensitive) strings to specify the names of "
          "callbacks. For example: `--callbacks ExamplesPerSecondCallback,"
          "LoggingMetricCallback`"))

  @flags.multi_flags_validator(
      ["eager", "dist_strat"],
      message="Both --eager and --dist_strat were set. Only one can be "
              "defined, as DistributionStrategy is not supported in Eager "
              "execution currently.")
  # pylint: disable=unused-variable
  def _check_eager_dist_strat(flag_dict):
    return not(flag_dict["eager"] and flag_dict["dist_strat"]) 

def define_resnet_flags(resnet_size_choices=None):
  """Add flags and validators for ResNet."""
  flags_core.define_base()
  flags_core.define_performance(num_parallel_calls=False,
                                tf_gpu_thread_mode=True,
                                datasets_num_private_threads=True,
                                datasets_num_parallel_batches=True)
  flags_core.define_image()
  flags_core.define_benchmark()
  flags.adopt_module_key_flags(flags_core)

  flags.DEFINE_enum(
      name='resnet_version', short_name='rv', default='1',
      enum_values=['1', '2'],
      help=flags_core.help_wrap(
          'Version of ResNet. (1 or 2) See README.md for details.'))
  flags.DEFINE_bool(
      name='fine_tune', short_name='ft', default=False,
      help=flags_core.help_wrap(
          'If True do not train any parameters except for the final layer.'))
  flags.DEFINE_string(
      name='pretrained_model_checkpoint_path', short_name='pmcp', default=None,
      help=flags_core.help_wrap(
          'If not None initialize all the network except the final layer with '
          'these values'))
  flags.DEFINE_boolean(
      name='eval_only', default=False,
      help=flags_core.help_wrap('Skip training and only perform evaluation on '
                                'the latest checkpoint.'))
  flags.DEFINE_boolean(
      name='image_bytes_as_serving_input', default=False,
      help=flags_core.help_wrap(
          'If True exports savedmodel with serving signature that accepts '
          'JPEG image bytes instead of a fixed size [HxWxC] tensor that '
          'represents the image. The former is easier to use for serving at '
          'the expense of image resize/cropping being done as part of model '
          'inference. Note, this flag only applies to ImageNet and cannot '
          'be used for CIFAR.'))

  choice_kwargs = dict(
      name='resnet_size', short_name='rs', default='50',
      help=flags_core.help_wrap('The size of the ResNet model to use.'))

  if resnet_size_choices is None:
    flags.DEFINE_string(**choice_kwargs)
  else:
    flags.DEFINE_enum(enum_values=resnet_size_choices, **choice_kwargs) 

def define_flags():
  """Construct flags for the server."""
  flags.DEFINE_integer(name="num_workers", default=multiprocessing.cpu_count(),
                       help="Size of the negative generation worker pool.")
  flags.DEFINE_string(name="data_dir", default=None,
                      help="The data root. (used to construct cache paths.)")
  flags.DEFINE_string(name="cache_id", default=None,
                      help="The cache_id generated in the main process.")
  flags.DEFINE_integer(name="num_readers", default=4,
                       help="Number of reader datasets in training. This sets"
                            "how the epoch files are sharded.")
  flags.DEFINE_integer(name="num_neg", default=None,
                       help="The Number of negative instances to pair with a "
                            "positive instance.")
  flags.DEFINE_integer(name="num_train_positives", default=None,
                       help="The number of positive training examples.")
  flags.DEFINE_integer(name="num_items", default=None,
                       help="Number of items from which to select negatives.")
  flags.DEFINE_integer(name="num_users", default=None,
                       help="The number of unique users. Used for evaluation.")
  flags.DEFINE_integer(name="epochs_per_cycle", default=1,
                       help="The number of epochs of training data to produce"
                            "at a time.")
  flags.DEFINE_integer(name="num_cycles", default=None,
                       help="The number of cycles to produce training data "
                            "for.")
  flags.DEFINE_integer(name="train_batch_size", default=None,
                       help="The batch size with which training TFRecords will "
                            "be chunked.")
  flags.DEFINE_integer(name="eval_batch_size", default=None,
                       help="The batch size with which evaluation TFRecords "
                            "will be chunked.")
  flags.DEFINE_boolean(name="redirect_logs", default=False,
                       help="Catch logs and write them to a file. "
                            "(Useful if this is run as a subprocess)")
  flags.DEFINE_boolean(name="use_tf_logging", default=False,
                       help="Use tf.logging instead of log file.")
  flags.DEFINE_integer(name="seed", default=None,
                       help="NumPy random seed to set at startup. If not "
                            "specified, a seed will not be set.")
  flags.DEFINE_boolean(name="ml_perf", default=None,
                       help="Match MLPerf. See ncf_main.py for details.")
  flags.DEFINE_bool(name="output_ml_perf_compliance_logging", default=None,
                    help="Output the MLPerf compliance logging. See "
                         "ncf_main.py for details.")

  flags.mark_flags_as_required(["data_dir", "cache_id"]) 

def SetupFlags():
  flags.DEFINE_string(
      'base_directory',
      './policies',
      'The base directory to look for acls; '
      'typically where you\'d find ./corp and ./prod')
  flags.DEFINE_string(
      'definitions_directory',
      './def',
      'Directory where the definitions can be found.')
  flags.DEFINE_string(
      'policy_file',
      None,
      'Individual policy file to generate.')
  flags.DEFINE_string(
      'output_directory',
      './',
      'Directory to output the rendered acls.')
  flags.DEFINE_boolean(
      'optimize',
      False,
      'Turn on optimization.',
      short_name='o')
  flags.DEFINE_boolean(
      'recursive',
      True,
      'Descend recursively from the base directory rendering acls')
  flags.DEFINE_boolean(
      'debug',
      False,
      'Debug messages')
  flags.DEFINE_boolean(
      'verbose',
      False,
      'Verbose messages')
  flags.DEFINE_list(
      'ignore_directories',
      'DEPRECATED, def',
      'Don\'t descend into directories that look like this string')
  flags.DEFINE_integer(
      'max_renderers',
      10,
      'Max number of rendering processes to use.')
  flags.DEFINE_boolean(
      'shade_check',
      False,
      'Raise an error when a term is completely shaded by a prior term.')
  flags.DEFINE_integer(
      'exp_info',
      2,
      'Print a info message when a term is set to expire in that many weeks.') 

def define_flags():
  """Define the flags for the Shakespeare character LSTM."""
  flags_core.define_base(data_dir=False,
                         clean=False,
                         train_epochs=True,
                         epochs_between_evals=False,
                         stop_threshold=False,
                         num_gpu=True,
                         export_dir=False,
                         run_eagerly=True,
                         distribution_strategy=True)

  flags_core.define_performance(num_parallel_calls=False,
                                inter_op=False,
                                intra_op=False,
                                synthetic_data=False,
                                max_train_steps=False,
                                dtype=True,
                                loss_scale=True,
                                enable_xla=True)

  flags_core.set_defaults(train_epochs=43,
                          batch_size=64)

  flags.DEFINE_boolean(name='enable_eager', default=True, help='Enable eager?')
  flags.DEFINE_boolean(
      name='train', default=True,
      help='If true trains the model.')
  flags.DEFINE_string(
      name='predict_context', default=None,
      help='If set, makes a prediction with the given context.')
  flags.DEFINE_integer(
      name='predict_length', default=1000,
      help='Length of the predicted text including the context.')
  flags.DEFINE_integer(name='train_steps', default=None,
                       help='Overrides train_steps per epoch if not None.')
  flags.DEFINE_integer(
      name='log_steps', default=100,
      help='For every log_steps, we log the timing information such as '
      'examples per second.')
  flags.DEFINE_string(
      name='training_data', default=None,
      help='Path to file containing the training data.')
  flags.DEFINE_boolean(name='cudnn', default=True, help='Use CuDNN LSTM.') 

def define_keras_benchmark_flags():
  """Add flags for keras built-in application models."""
  flags_core.define_base(hooks=False)
  flags_core.define_performance()
  flags_core.define_image()
  flags_core.define_benchmark()
  flags.adopt_module_key_flags(flags_core)

  flags_core.set_defaults(
      data_format="channels_last",
      use_synthetic_data=True,
      batch_size=32,
      train_epochs=2)

  flags.DEFINE_enum(
      name="model", default=None,
      enum_values=MODELS.keys(), case_sensitive=False,
      help=flags_core.help_wrap(
          "Model to be benchmarked."))

  flags.DEFINE_integer(
      name="num_train_images", default=1000,
      help=flags_core.help_wrap(
          "The number of synthetic images for training. The default value is "
          "1000."))

  flags.DEFINE_integer(
      name="num_eval_images", default=50,
      help=flags_core.help_wrap(
          "The number of synthetic images for evaluation. The default value is "
          "50."))

  flags.DEFINE_boolean(
      name="eager", default=False, help=flags_core.help_wrap(
          "To enable eager execution. Note that if eager execution is enabled, "
          "only one GPU is utilized even if multiple GPUs are provided and "
          "multi_gpu_model is used."))

  flags.DEFINE_boolean(
      name="dist_strat", default=False, help=flags_core.help_wrap(
          "To enable distribution strategy for model training and evaluation. "
          "Number of GPUs used for distribution strategy can be set by the "
          "argument --num_gpus."))

  flags.DEFINE_list(
      name="callbacks",
      default=["ExamplesPerSecondCallback", "LoggingMetricCallback"],
      help=flags_core.help_wrap(
          "A list of (case insensitive) strings to specify the names of "
          "callbacks. For example: `--callbacks ExamplesPerSecondCallback,"
          "LoggingMetricCallback`"))

  @flags.multi_flags_validator(
      ["eager", "dist_strat"],
      message="Both --eager and --dist_strat were set. Only one can be "
              "defined, as DistributionStrategy is not supported in Eager "
              "execution currently.")
  # pylint: disable=unused-variable
  def _check_eager_dist_strat(flag_dict):
    return not(flag_dict["eager"] and flag_dict["dist_strat"]) 

def define_resnet_flags(resnet_size_choices=None):
  """Add flags and validators for ResNet."""
  flags_core.define_base()
  flags_core.define_performance(num_parallel_calls=False,
                                tf_gpu_thread_mode=True,
                                datasets_num_private_threads=True,
                                datasets_num_parallel_batches=True)
  flags_core.define_image()
  flags_core.define_benchmark()
  flags.adopt_module_key_flags(flags_core)

  flags.DEFINE_enum(
      name='resnet_version', short_name='rv', default='1',
      enum_values=['1', '2'],
      help=flags_core.help_wrap(
          'Version of ResNet. (1 or 2) See README.md for details.'))
  flags.DEFINE_bool(
      name='fine_tune', short_name='ft', default=False,
      help=flags_core.help_wrap(
          'If True do not train any parameters except for the final layer.'))
  flags.DEFINE_string(
      name='pretrained_model_checkpoint_path', short_name='pmcp', default=None,
      help=flags_core.help_wrap(
          'If not None initialize all the network except the final layer with '
          'these values'))
  flags.DEFINE_boolean(
      name='eval_only', default=False,
      help=flags_core.help_wrap('Skip training and only perform evaluation on '
                                'the latest checkpoint.'))
  flags.DEFINE_boolean(
      name='image_bytes_as_serving_input', default=False,
      help=flags_core.help_wrap(
          'If True exports savedmodel with serving signature that accepts '
          'JPEG image bytes instead of a fixed size [HxWxC] tensor that '
          'represents the image. The former is easier to use for serving at '
          'the expense of image resize/cropping being done as part of model '
          'inference. Note, this flag only applies to ImageNet and cannot '
          'be used for CIFAR.'))
  flags.DEFINE_boolean(
      name='turn_off_distribution_strategy', default=False,
      help=flags_core.help_wrap('Set to True to not use distribution '
                                'strategies.'))
  choice_kwargs = dict(
      name='resnet_size', short_name='rs', default='50',
      help=flags_core.help_wrap('The size of the ResNet model to use.'))

  if resnet_size_choices is None:
    flags.DEFINE_string(**choice_kwargs)
  else:
    flags.DEFINE_enum(enum_values=resnet_size_choices, **choice_kwargs) 

def define_resnet_flags(resnet_size_choices=None):
  """Add flags and validators for ResNet."""
  flags_core.define_base()
  flags_core.define_performance(num_parallel_calls=False,
                                tf_gpu_thread_mode=True,
                                datasets_num_private_threads=True,
                                datasets_num_parallel_batches=True)
  flags_core.define_image()
  flags_core.define_benchmark()
  flags.adopt_module_key_flags(flags_core)

  flags.DEFINE_enum(
      name='resnet_version', short_name='rv', default='1',
      enum_values=['1', '2'],
      help=flags_core.help_wrap(
          'Version of ResNet. (1 or 2) See README.md for details.'))
  flags.DEFINE_bool(
      name='fine_tune', short_name='ft', default=False,
      help=flags_core.help_wrap(
          'If True do not train any parameters except for the final layer.'))
  flags.DEFINE_string(
      name='pretrained_model_checkpoint_path', short_name='pmcp', default=None,
      help=flags_core.help_wrap(
          'If not None initialize all the network except the final layer with '
          'these values'))
  flags.DEFINE_boolean(
      name='eval_only', default=False,
      help=flags_core.help_wrap('Skip training and only perform evaluation on '
                                'the latest checkpoint.'))
  flags.DEFINE_boolean(
      name='image_bytes_as_serving_input', default=False,
      help=flags_core.help_wrap(
          'If True exports savedmodel with serving signature that accepts '
          'JPEG image bytes instead of a fixed size [HxWxC] tensor that '
          'represents the image. The former is easier to use for serving at '
          'the expense of image resize/cropping being done as part of model '
          'inference. Note, this flag only applies to ImageNet and cannot '
          'be used for CIFAR.'))

  choice_kwargs = dict(
      name='resnet_size', short_name='rs', default='50',
      help=flags_core.help_wrap('The size of the ResNet model to use.'))

  if resnet_size_choices is None:
    flags.DEFINE_string(**choice_kwargs)
  else:
    flags.DEFINE_enum(enum_values=resnet_size_choices, **choice_kwargs)