Module: tf.contrib.estimator

Defined in tensorflow/contrib/estimator/__init__.py.

Experimental utilities re:tf.estimator.*.

Classes

class DNNEstimator: An estimator for TensorFlow DNN models with user-specified head.

class DNNLinearCombinedEstimator: An estimator for TensorFlow Linear and DNN joined models with custom head.

class LinearEstimator: An estimator for TensorFlow linear models with user-specified head.

class TowerOptimizer: Gathers gradients from all towers and reduces them in the last one.

Functions

add_metrics(...): Creates a new tf.estimator.Estimator which has given metrics.

binary_classification_head(...): Creates a _Head for single label binary classification.

boosted_trees_classifier_train_in_memory(...): Trains a boosted tree classifier with in memory dataset.

boosted_trees_regressor_train_in_memory(...): Trains a boosted tree regressor with in memory dataset.

call_logit_fn(...): Calls logit_fn.

clip_gradients_by_norm(...): Returns an optimizer which clips gradients before applying them.

dnn_logit_fn_builder(...): Function builder for a dnn logit_fn.

forward_features(...): Forward features to predictions dictionary.

linear_logit_fn_builder(...): Function builder for a linear logit_fn.

multi_class_head(...): Creates a _Head for multi class classification.

multi_head(...): Creates a _Head for multi-objective learning.

multi_label_head(...): Creates a _Head for multi-label classification.

poisson_regression_head(...): Creates a _Head for poisson regression using tf.nn.log_poisson_loss.

regression_head(...): Creates a _Head for regression using the mean_squared_error loss.

replicate_model_fn(...): Replicate Estimator.model_fn over GPUs.