# Module: tf.contrib.training

<br />

|---|
| [![](https://www.tensorflow.org/images/GitHub-Mark-32px.png) View source on GitHub](https://github.com/tensorflow/tensorflow/blob/v1.15.0/tensorflow/contrib/training/__init__.py) |

Training and input utilities.

See
[Contrib Training](https://tensorflow.org/api_guides/python/contrib.training)
guide.

## Classes

[`class FeedingQueueRunner`](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/contrib/training/FeedingQueueRunner): A queue runner that allows the feeding of values such as numpy arrays.

[`class GreedyLoadBalancingStrategy`](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/contrib/training/GreedyLoadBalancingStrategy): Returns the least-loaded ps task for op placement.

[`class HParams`](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/contrib/training/HParams): Class to hold a set of hyperparameters as name-value pairs.

[`class NextQueuedSequenceBatch`](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/contrib/training/NextQueuedSequenceBatch): NextQueuedSequenceBatch stores deferred SequenceQueueingStateSaver data.

[`class RandomStrategy`](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/contrib/training/RandomStrategy): Returns a random PS task for op placement.

[`class SequenceQueueingStateSaver`](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/contrib/training/SequenceQueueingStateSaver): SequenceQueueingStateSaver provides access to stateful values from input.

[`class StopAfterNEvalsHook`](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/contrib/training/StopAfterNEvalsHook): Run hook used by the evaluation routines to run the `eval_ops` N times.

[`class SummaryAtEndHook`](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/contrib/training/SummaryAtEndHook): A run hook that saves a summary with the results of evaluation.

## Functions

[`add_gradients_summaries(...)`](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/contrib/training/add_gradients_summaries): Add summaries to gradients.

[`batch_sequences_with_states(...)`](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/contrib/training/batch_sequences_with_states): Creates batches of segments of sequential input.

[`bucket(...)`](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/contrib/training/bucket): Lazy bucketing of input tensors according to `which_bucket`.

[`bucket_by_sequence_length(...)`](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/contrib/training/bucket_by_sequence_length): Lazy bucketing of inputs according to their length.

[`byte_size_load_fn(...)`](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/contrib/training/byte_size_load_fn): Load function that computes the byte size of a single-output `Operation`.

[`checkpoints_iterator(...)`](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/contrib/training/checkpoints_iterator): Continuously yield new checkpoint files as they appear.

[`clip_gradient_norms(...)`](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/contrib/training/clip_gradient_norms): Clips the gradients by the given value.

[`clip_gradient_norms_fn(...)`](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/contrib/training/clip_gradient_norms_fn): Returns a `transform_grads_fn` function for gradient clipping.

[`create_train_op(...)`](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/contrib/training/create_train_op): Creates an `Operation` that evaluates the gradients and returns the loss.

[`evaluate_once(...)`](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/contrib/training/evaluate_once): Evaluates the model at the given checkpoint path.

[`evaluate_repeatedly(...)`](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/contrib/training/evaluate_repeatedly): Repeatedly searches for a checkpoint in `checkpoint_dir` and evaluates it.

[`get_or_create_eval_step(...)`](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/contrib/training/get_or_create_eval_step): Gets or creates the eval step `Tensor`.

[`multiply_gradients(...)`](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/contrib/training/multiply_gradients): Multiply specified gradients.

[`parse_values(...)`](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/contrib/training/parse_values): Parses hyperparameter values from a string into a python map.

[`rejection_sample(...)`](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/contrib/training/rejection_sample): Stochastically creates batches by rejection sampling.

[`resample_at_rate(...)`](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/contrib/training/resample_at_rate): Given `inputs` tensors, stochastically resamples each at a given rate.

[`stratified_sample(...)`](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/contrib/training/stratified_sample): Stochastically creates batches based on per-class probabilities.

[`train(...)`](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/contrib/training/train): Runs the training loop.

[`wait_for_new_checkpoint(...)`](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/contrib/training/wait_for_new_checkpoint): Waits until a new checkpoint file is found.

[`weighted_resample(...)`](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/contrib/training/weighted_resample): Performs an approximate weighted resampling of `inputs`.