experiments/STT-tensorflow
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Sample tf.learn API doc formatting change. Moves the "Parameters:" section in the class documentation to an "Args:" section under __init__.

Browse Source 
Change: 124865749
This commit is contained in:
A. Unique TensorFlower 2016-06-14 11:01:37 -08:00 committed by TensorFlower Gardener
parent b97968aa0d
commit 0060abbda6
2 changed files with 52 additions and 40 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

60
tensorflow/contrib/learn/python/learn/estimators/estimator.py
View File

@ -97,9 +97,6 @@ class BaseEstimator(sklearn.BaseEstimator):
* _get_predict_ops
`Estimator` implemented below is a good example of how to use this class.
Parameters:
model_dir: Directory to save model parameters, graph and etc.
"""
__metaclass__ = abc.ABCMeta
@ -107,6 +104,12 @@ class BaseEstimator(sklearn.BaseEstimator):
_Config = run_config.RunConfig # pylint: disable=invalid-name
def __init__(self, model_dir=None, config=None):
"""Initializes a BaseEstimator instance.
Args:
model_dir: Directory to save model parameters, graph and etc.
config: A RunConfig instance.
"""
# Model directory.
self._model_dir = model_dir
if self._model_dir is None:
@ -621,29 +624,6 @@ class BaseEstimator(sklearn.BaseEstimator):
class Estimator(BaseEstimator):
"""Estimator class is the basic TensorFlow model trainer/evaluator.
Parameters:
model_fn: Model function, takes features and targets tensors or dicts of
tensors and returns predictions and loss tensors.
Supports next three signatures for the function:
* `(features, targets) -> (predictions, loss, train_op)`
* `(features, targets, mode) -> (predictions, loss, train_op)`
* `(features, targets, mode, params) ->
(predictions, loss, train_op)`
Where:
* `features` are single `Tensor` or `dict` of `Tensor`s
(depending on data passed to `fit`),
* `targets` are `Tensor` or
`dict` of `Tensor`s (for multi-head model).
* `mode` represents if this training, evaluation or prediction.
See `ModeKeys` for example keys.
* `params` is a `dict` of hyperparameters. Will receive what is
passed to Estimator in `params` parameter. This allows to
configure Estimators from hyper parameter tunning.
model_dir: Directory to save model parameters, graph and etc.
config: Configuration object.
params: `dict` of hyper parameters that will be passed into `model_fn`.
Keys are names of parameters, values are basic python types.
"""
def __init__(self,
@ -651,6 +631,34 @@ class Estimator(BaseEstimator):
model_dir=None,
config=None,
params=None):
"""Constructs an Estimator instance.
Args:
model_fn: Model function, takes features and targets tensors or dicts of
tensors and returns predictions and loss tensors.
Supports next three signatures for the function:
* `(features, targets) -> (predictions, loss, train_op)`
* `(features, targets, mode) -> (predictions, loss, train_op)`
* `(features, targets, mode, params) ->
(predictions, loss, train_op)`
Where:
* `features` are single `Tensor` or `dict` of `Tensor`s
(depending on data passed to `fit`),
* `targets` are `Tensor` or
`dict` of `Tensor`s (for multi-head model).
* `mode` represents if this training, evaluation or
prediction. See `ModeKeys` for example keys.
* `params` is a `dict` of hyperparameters. Will receive what
is passed to Estimator in `params` parameter. This allows
to configure Estimators from hyper parameter tunning.
model_dir: Directory to save model parameters, graph and etc.
config: Configuration object.
params: `dict` of hyper parameters that will be passed into `model_fn`.
Keys are names of parameters, values are basic python types.
Raises:
ValueError: parameters of `model_fn` don't match `params`.
"""
super(Estimator, self).__init__(model_dir=model_dir, config=config)
if model_fn is not None:
# Check number of arguments of the given function matches requirements.

32
tensorflow/g3doc/api_docs/python/contrib.learn.md
View File

@ -21,14 +21,16 @@ Concrete implementation of this class should provide following functions:
* _get_predict_ops
`Estimator` implemented below is a good example of how to use this class.
Parameters:
model_dir: Directory to save model parameters, graph and etc.
- - -
#### `tf.contrib.learn.BaseEstimator.__init__(model_dir=None, config=None)` {#BaseEstimator.__init__}
Initializes a BaseEstimator instance.
##### Parameters:
* <b>`model_dir`</b>: Directory to save model parameters, graph and etc.
- - -
@ -270,16 +272,23 @@ component of a nested object.
### `class tf.contrib.learn.Estimator` {#Estimator}
Estimator class is the basic TensorFlow model trainer/evaluator.
- - -
Parameters:
model_fn: Model function, takes features and targets tensors or dicts of
#### `tf.contrib.learn.Estimator.__init__(model_fn=None, model_dir=None, config=None, params=None)` {#Estimator.__init__}
Constructs an Estimator instance.
##### Args:
* <b>`model_fn`</b>: Model function, takes features and targets tensors or dicts of
tensors and returns predictions and loss tensors.
Supports next three signatures for the function:
* `(features, targets) -> (predictions, loss, train_op)`
* `(features, targets, mode) -> (predictions, loss, train_op)`
* `(features, targets, mode, params) ->
(predictions, loss, train_op)`
Where:
* <b>`Where`</b>:
* `features` are single `Tensor` or `dict` of `Tensor`s
(depending on data passed to `fit`),
* `targets` are `Tensor` or
@ -289,15 +298,10 @@ Parameters:
* `params` is a `dict` of hyperparameters. Will receive what is
passed to Estimator in `params` parameter. This allows to
configure Estimators from hyper parameter tunning.
model_dir: Directory to save model parameters, graph and etc.
config: Configuration object.
params: `dict` of hyper parameters that will be passed into `model_fn`.
* <b>`model_dir`</b>: Directory to save model parameters, graph and etc.
* <b>`config`</b>: Configuration object.
* <b>`params`</b>: `dict` of hyper parameters that will be passed into `model_fn`.
Keys are names of parameters, values are basic python types.
- - -
#### `tf.contrib.learn.Estimator.__init__(model_fn=None, model_dir=None, config=None, params=None)` {#Estimator.__init__}
- - -