![]() |
![]() |
Class BaselineEstimator
An estimator that can establish a simple baseline.
Inherits From: Estimator
Aliases:
The estimator uses a user-specified head.
This estimator ignores feature values and will learn to predict the average value of each label. E.g. for single-label classification problems, this will predict the probability distribution of the classes as seen in the labels. For multi-label classification problems, it will predict the ratio of examples that contain each class.
Example:
# Build baseline multi-label classifier.
estimator = BaselineEstimator(
head=tf.contrib.estimator.multi_label_head(n_classes=3))
# Input builders
def input_fn_train:
# Returns tf.data.Dataset of (x, y) tuple where y represents label's class
# index.
pass
def input_fn_eval:
# Returns tf.data.Dataset of (x, y) tuple where y represents label's class
# index.
pass
# Fit model.
estimator.train(input_fn=input_fn_train)
# Evaluates cross entropy between the test and train labels.
loss = estimator.evaluate(input_fn=input_fn_eval)["loss"]
# For each class, predicts the ratio of training examples that contain the
# class.
predictions = estimator.predict(new_samples)
Input of train
and evaluate
should have following features,
otherwise there will be a KeyError
:
- if
weight_column
is specified in thehead
constructor (and not None) for the head passed to BaselineEstimator's constructor, a feature withkey=weight_column
whose value is aTensor
.
__init__
__init__(
head,
model_dir=None,
optimizer='Ftrl',
config=None
)
Constructs an Estimator
instance.
See estimators for more information.
To warm-start an Estimator
:
estimator = tf.estimator.DNNClassifier(
feature_columns=[categorical_feature_a_emb, categorical_feature_b_emb],
hidden_units=[1024, 512, 256],
warm_start_from="/path/to/checkpoint/dir")
For more details on warm-start configuration, see
tf.estimator.WarmStartSettings
.
Args:
model_fn
: Model function. Follows the signature:Args:
features
: This is the first item returned from theinput_fn
passed totrain
,evaluate
, andpredict
. This should be a singletf.Tensor
ordict
of same.labels
: This is the second item returned from theinput_fn
passed totrain
,evaluate
, andpredict
. This should be a singletf.Tensor
ordict
of same (for multi-head models). If mode istf.estimator.ModeKeys.PREDICT
,labels=None
will be passed. If themodel_fn
's signature does not acceptmode
, themodel_fn
must still be able to handlelabels=None
.mode
: Optional. Specifies if this is training, evaluation or prediction. Seetf.estimator.ModeKeys
.params
: Optionaldict
of hyperparameters. Will receive what is passed to Estimator inparams
parameter. This allows to configure Estimators from hyper parameter tuning.config
: Optionalestimator.RunConfig
object. Will receive what is passed to Estimator as itsconfig
parameter, or a default value. Allows setting up things in yourmodel_fn
based on configuration such asnum_ps_replicas
, ormodel_dir
.Returns:
tf.estimator.EstimatorSpec
model_dir
: Directory to save model parameters, graph and etc. This can also be used to load checkpoints from the directory into an estimator to continue training a previously saved model. IfPathLike
object, the path will be resolved. IfNone
, the model_dir inconfig
will be used if set. If both are set, they must be same. If both areNone
, a temporary directory will be used.config
:estimator.RunConfig
configuration object.params
:dict
of hyper parameters that will be passed intomodel_fn
. Keys are names of parameters, values are basic python types.warm_start_from
: Optional string filepath to a checkpoint or SavedModel to warm-start from, or atf.estimator.WarmStartSettings
object to fully configure warm-starting.If None, only TRAINABLE variables are warm-started. If the string filepath is provided instead of a <a href="../../tf/estimator/WarmStartSettings"><code>tf.estimator.WarmStartSettings</code></a>, then all variables are warm-started, and it is assumed that vocabularies and <a href="../../tf/Tensor"><code>tf.Tensor</code></a> names are unchanged.
Raises:
ValueError
: parameters ofmodel_fn
don't matchparams
.ValueError
: if this is called via a subclass and if that class overrides a member ofEstimator
.
Properties
config
model_dir
model_fn
Returns the model_fn
which is bound to self.params
.
Returns:
The model_fn
with following signature:
def model_fn(features, labels, mode, config)
params
Methods
eval_dir
eval_dir(name=None)
Shows the directory name where evaluation metrics are dumped.
Args:
name
: Name of the evaluation if user needs to run multiple evaluations on different data sets, such as on training data vs test data. Metrics for different evaluations are saved in separate folders, and appear separately in tensorboard.
Returns:
A string which is the path of directory contains evaluation metrics.
evaluate
evaluate(
input_fn,
steps=None,
hooks=None,
checkpoint_path=None,
name=None
)
Evaluates the model given evaluation data input_fn
.
For each step, calls input_fn
, which returns one batch of data.
Evaluates until:
- steps
batches are processed, or
- input_fn
raises an end-of-input exception (tf.errors.OutOfRangeError
or
StopIteration
).
Args:
input_fn
: A function that constructs the input data for evaluation. See Premade Estimators for more information. The function should construct and return one of the following: * Atf.data.Dataset
object: Outputs ofDataset
object must be a tuple(features, labels)
with same constraints as below. * A tuple(features, labels)
: Wherefeatures
is atf.Tensor
or a dictionary of string feature name toTensor
andlabels
is aTensor
or a dictionary of string label name toTensor
. Bothfeatures
andlabels
are consumed bymodel_fn
. They should satisfy the expectation ofmodel_fn
from inputs.steps
: Number of steps for which to evaluate model. IfNone
, evaluates untilinput_fn
raises an end-of-input exception.hooks
: List oftf.train.SessionRunHook
subclass instances. Used for callbacks inside the evaluation call.checkpoint_path
: Path of a specific checkpoint to evaluate. IfNone
, the latest checkpoint inmodel_dir
is used. If there are no checkpoints inmodel_dir
, evaluation is run with newly initializedVariables
instead of ones restored from checkpoint.name
: Name of the evaluation if user needs to run multiple evaluations on different data sets, such as on training data vs test data. Metrics for different evaluations are saved in separate folders, and appear separately in tensorboard.
Returns:
A dict containing the evaluation metrics specified in model_fn
keyed by
name, as well as an entry global_step
which contains the value of the
global step for which this evaluation was performed. For canned
estimators, the dict contains the loss
(mean loss per mini-batch) and
the average_loss
(mean loss per sample). Canned classifiers also return
the accuracy
. Canned regressors also return the label/mean
and the
prediction/mean
.
Raises:
ValueError
: Ifsteps <= 0
.
experimental_export_all_saved_models
experimental_export_all_saved_models(
export_dir_base,
input_receiver_fn_map,
assets_extra=None,
as_text=False,
checkpoint_path=None
)
Exports a SavedModel
with tf.MetaGraphDefs
for each requested mode.
For each mode passed in via the input_receiver_fn_map
,
this method builds a new graph by calling the input_receiver_fn
to obtain
feature and label Tensor
s. Next, this method calls the Estimator
's
model_fn
in the passed mode to generate the model graph based on
those features and labels, and restores the given checkpoint
(or, lacking that, the most recent checkpoint) into the graph.
Only one of the modes is used for saving variables to the SavedModel
(order of preference: tf.estimator.ModeKeys.TRAIN
,
tf.estimator.ModeKeys.EVAL
, then
tf.estimator.ModeKeys.PREDICT
), such that up to three
tf.MetaGraphDefs
are saved with a single set of variables in a single
SavedModel
directory.
For the variables and tf.MetaGraphDefs
, a timestamped export directory
below
export_dir_base
, and writes a SavedModel
into it containing
the tf.MetaGraphDef
for the given mode and its associated signatures.
For prediction, the exported MetaGraphDef
will provide one SignatureDef
for each element of the export_outputs
dict returned from the model_fn
,
named using the same keys. One of these keys is always
tf.saved_model.signature_constants.DEFAULT_SERVING_SIGNATURE_DEF_KEY
,
indicating which
signature will be served when a serving request does not specify one.
For each signature, the outputs are provided by the corresponding
tf.estimator.export.ExportOutput
s, and the inputs are always the input
receivers provided by
the serving_input_receiver_fn
.
For training and evaluation, the train_op
is stored in an extra
collection,
and loss, metrics, and predictions are included in a SignatureDef
for the
mode in question.
Extra assets may be written into the SavedModel
via the assets_extra
argument. This should be a dict, where each key gives a destination path
(including the filename) relative to the assets.extra directory. The
corresponding value gives the full path of the source file to be copied.
For example, the simple case of copying a single file without renaming it
is specified as {'my_asset_file.txt': '/path/to/my_asset_file.txt'}
.
Args:
export_dir_base
: A string containing a directory in which to create timestamped subdirectories containing exportedSavedModel
s.input_receiver_fn_map
: dict oftf.estimator.ModeKeys
toinput_receiver_fn
mappings, where theinput_receiver_fn
is a function that takes no arguments and returns the appropriate subclass ofInputReceiver
.assets_extra
: A dict specifying how to populate the assets.extra directory within the exportedSavedModel
, orNone
if no extra assets are needed.as_text
: whether to write theSavedModel
proto in text format.checkpoint_path
: The checkpoint path to export. IfNone
(the default), the most recent checkpoint found within the model directory is chosen.
Returns:
The string path to the exported directory.
Raises:
ValueError
: if anyinput_receiver_fn
isNone
, noexport_outputs
are provided, or no checkpoint can be found.
export_saved_model
export_saved_model(
export_dir_base,
serving_input_receiver_fn,
assets_extra=None,
as_text=False,
checkpoint_path=None,
experimental_mode=ModeKeys.PREDICT
)
Exports inference graph as a SavedModel
into the given dir.
For a detailed guide, see Using SavedModel with Estimators.
This method builds a new graph by first calling the
serving_input_receiver_fn
to obtain feature Tensor
s, and then calling
this Estimator
's model_fn
to generate the model graph based on those
features. It restores the given checkpoint (or, lacking that, the most
recent checkpoint) into this graph in a fresh session. Finally it creates
a timestamped export directory below the given export_dir_base
, and writes
a SavedModel
into it containing a single tf.MetaGraphDef
saved from this
session.
The exported MetaGraphDef
will provide one SignatureDef
for each
element of the export_outputs
dict returned from the model_fn
, named
using
the same keys. One of these keys is always
tf.saved_model.signature_constants.DEFAULT_SERVING_SIGNATURE_DEF_KEY
,
indicating which
signature will be served when a serving request does not specify one.
For each signature, the outputs are provided by the corresponding
tf.estimator.export.ExportOutput
s, and the inputs are always the input
receivers provided by
the serving_input_receiver_fn
.
Extra assets may be written into the SavedModel
via the assets_extra
argument. This should be a dict, where each key gives a destination path
(including the filename) relative to the assets.extra directory. The
corresponding value gives the full path of the source file to be copied.
For example, the simple case of copying a single file without renaming it
is specified as {'my_asset_file.txt': '/path/to/my_asset_file.txt'}
.
The experimental_mode parameter can be used to export a single
train/eval/predict graph as a SavedModel
.
See experimental_export_all_saved_models
for full docs.
Args:
export_dir_base
: A string containing a directory in which to create timestamped subdirectories containing exportedSavedModel
s.serving_input_receiver_fn
: A function that takes no argument and returns atf.estimator.export.ServingInputReceiver
ortf.estimator.export.TensorServingInputReceiver
.assets_extra
: A dict specifying how to populate the assets.extra directory within the exportedSavedModel
, orNone
if no extra assets are needed.as_text
: whether to write theSavedModel
proto in text format.checkpoint_path
: The checkpoint path to export. IfNone
(the default), the most recent checkpoint found within the model directory is chosen.experimental_mode
:tf.estimator.ModeKeys
value indicating with mode will be exported. Note that this feature is experimental.
Returns:
The string path to the exported directory.
Raises:
ValueError
: if noserving_input_receiver_fn
is provided, noexport_outputs
are provided, or no checkpoint can be found.
export_savedmodel
export_savedmodel(
export_dir_base,
serving_input_receiver_fn,
assets_extra=None,
as_text=False,
checkpoint_path=None,
strip_default_attrs=False
)
Exports inference graph as a SavedModel
into the given dir. (deprecated)
For a detailed guide, see Using SavedModel with Estimators.
This method builds a new graph by first calling the
serving_input_receiver_fn
to obtain feature Tensor
s, and then calling
this Estimator
's model_fn
to generate the model graph based on those
features. It restores the given checkpoint (or, lacking that, the most
recent checkpoint) into this graph in a fresh session. Finally it creates
a timestamped export directory below the given export_dir_base
, and writes
a SavedModel
into it containing a single tf.MetaGraphDef
saved from this
session.
The exported MetaGraphDef
will provide one SignatureDef
for each
element of the export_outputs
dict returned from the model_fn
, named
using
the same keys. One of these keys is always
tf.saved_model.signature_constants.DEFAULT_SERVING_SIGNATURE_DEF_KEY
,
indicating which
signature will be served when a serving request does not specify one.
For each signature, the outputs are provided by the corresponding
tf.estimator.export.ExportOutput
s, and the inputs are always the input
receivers provided by
the serving_input_receiver_fn
.
Extra assets may be written into the SavedModel
via the assets_extra
argument. This should be a dict, where each key gives a destination path
(including the filename) relative to the assets.extra directory. The
corresponding value gives the full path of the source file to be copied.
For example, the simple case of copying a single file without renaming it
is specified as {'my_asset_file.txt': '/path/to/my_asset_file.txt'}
.
Args:
export_dir_base
: A string containing a directory in which to create timestamped subdirectories contain