当前位置: 首页>>代码示例>>Python>>正文


Python logging.log_first_n方法代码示例

本文整理汇总了Python中absl.logging.log_first_n方法的典型用法代码示例。如果您正苦于以下问题:Python logging.log_first_n方法的具体用法?Python logging.log_first_n怎么用?Python logging.log_first_n使用的例子?那么, 这里精选的方法代码示例或许可以为您提供帮助。您也可以进一步了解该方法所在absl.logging的用法示例。


在下文中一共展示了logging.log_first_n方法的3个代码示例,这些例子默认根据受欢迎程度排序。您可以为喜欢或者感觉有用的代码点赞,您的评价将有助于系统推荐出更棒的Python代码示例。

示例1: gym_env_wrapper

# 需要导入模块: from absl import logging [as 别名]
# 或者: from absl.logging import log_first_n [as 别名]
def gym_env_wrapper(env, rl_env_max_episode_steps, maxskip_env, rendered_env,
                    rendered_env_resize_to, sticky_actions, output_dtype,
                    num_actions):
  """Wraps a gym environment. see make_gym_env for details."""
  # rl_env_max_episode_steps is None or int.
  assert ((not rl_env_max_episode_steps) or
          isinstance(rl_env_max_episode_steps, int))

  wrap_with_time_limit = ((not rl_env_max_episode_steps) or
                          rl_env_max_episode_steps >= 0)

  if wrap_with_time_limit:
    env = remove_time_limit_wrapper(env)

  if num_actions is not None:
    logging.log_first_n(
        logging.INFO, "Number of discretized actions: %d", 1, num_actions)
    env = ActionDiscretizeWrapper(env, num_actions=num_actions)

  if sticky_actions:
    env = StickyActionEnv(env)

  if maxskip_env:
    env = MaxAndSkipEnv(env)  # pylint: disable=redefined-variable-type

  if rendered_env:
    env = RenderedEnv(
        env, resize_to=rendered_env_resize_to, output_dtype=output_dtype)

  if wrap_with_time_limit and rl_env_max_episode_steps is not None:
    env = gym.wrappers.TimeLimit(
        env, max_episode_steps=rl_env_max_episode_steps)
  return env 
开发者ID:tensorflow,项目名称:tensor2tensor,代码行数:35,代码来源:gym_utils.py

示例2: _compute_gradient

# 需要导入模块: from absl import logging [as 别名]
# 或者: from absl.logging import log_first_n [as 别名]
def _compute_gradient(self, loss, dense_features, gradient_tape=None):
    """Computes the gradient given a loss and dense features."""
    feature_values = list(dense_features.values())
    if gradient_tape is None:
      grads = tf.gradients(loss, feature_values)
    else:
      grads = gradient_tape.gradient(loss, feature_values)

    # The order of elements returned by .values() and .keys() are guaranteed
    # corresponding to each other.
    keyed_grads = dict(zip(dense_features.keys(), grads))

    invalid_grads, valid_grads = self._split_dict(keyed_grads,
                                                  lambda grad: grad is None)
    # Two cases that grad can be invalid (None):
    # (1) The feature is not differentiable, like strings or integers.
    # (2) The feature is not involved in loss computation.
    if invalid_grads:
      if self._raise_invalid_gradient:
        raise ValueError('Cannot perturb features ' + str(invalid_grads.keys()))
      logging.log_first_n(logging.WARNING, 'Cannot perturb features %s', 1,
                          invalid_grads.keys())

    # Guards against numerical errors. If the gradient is malformed (inf, -inf,
    # or NaN) on a dimension, replace it with 0, which has the effect of not
    # perturbing the original sample along that perticular dimension.
    return tf.nest.map_structure(
        lambda g: tf.where(tf.math.is_finite(g), g, tf.zeros_like(g)),
        valid_grads)

  # The _compose_as_dict and _decompose_as functions are similar to
  # tf.nest.{flatten, pack_sequence_as} except that the composed representation
  # is a dictionary of (name, value) pairs instead of a list of values. The
  # names are needed for joining values from different inputs (e.g. input
  # features and feature masks) with possibly missing values (e.g. no mask for
  # some features). 
开发者ID:tensorflow,项目名称:neural-structured-learning,代码行数:38,代码来源:adversarial_neighbor.py

示例3: tfhub_cache_dir

# 需要导入模块: from absl import logging [as 别名]
# 或者: from absl.logging import log_first_n [as 别名]
def tfhub_cache_dir(default_cache_dir=None, use_temp=False):
  """Returns cache directory.

  Returns cache directory from either TFHUB_CACHE_DIR environment variable
  or --tfhub_cache_dir or default, if set.

  Args:
    default_cache_dir: Default cache location to use if neither TFHUB_CACHE_DIR
                       environment variable nor --tfhub_cache_dir are
                       not specified.
    use_temp: bool, Optional to enable using system's temp directory as a
              module cache directory if neither default_cache_dir nor
              --tfhub_cache_dir nor TFHUB_CACHE_DIR environment variable are
              specified .
  """

  # Note: We are using FLAGS["tfhub_cache_dir"] (and not FLAGS.tfhub_cache_dir)
  # to access the flag value in order to avoid parsing argv list. The flags
  # should have been parsed by now in main() by tf.app.run(). If that was not
  # the case (say in Colab env) we skip flag parsing because argv may contain
  # unknown flags.
  cache_dir = (
      os.getenv(_TFHUB_CACHE_DIR, "") or FLAGS["tfhub_cache_dir"].value or
      default_cache_dir)
  if not cache_dir and use_temp:
    # Place all TF-Hub modules under <system's temp>/tfhub_modules.
    cache_dir = os.path.join(tempfile.gettempdir(), "tfhub_modules")
  if cache_dir:
    logging.log_first_n(logging.INFO, "Using %s to cache modules.", 1,
                        cache_dir)
  return cache_dir 
开发者ID:tensorflow,项目名称:hub,代码行数:33,代码来源:resolver.py


注:本文中的absl.logging.log_first_n方法示例由纯净天空整理自Github/MSDocs等开源代码及文档管理平台,相关代码片段筛选自各路编程大神贡献的开源项目,源码版权归原作者所有,传播和使用请参考对应项目的License;未经允许,请勿转载。