o i e@sdZddlmZddlmZddlmZddlmmZddl m Z ddl m Z ddlmZed Gd d d e jZdd d ZdS)zThis API defines FeatureColumn for sequential input. NOTE: This API is a work in progress and will likely be changing frequently. )absolute_import)division)print_functionN)backend)base_feature_layer) keras_exportz#keras.experimental.SequenceFeaturescs@eZdZdZd fdd ZeddZdd Zd d d ZZ S)SequenceFeaturesaA layer for sequence input. All `feature_columns` must be sequence dense columns with the same `sequence_length`. The output of this method can be fed into sequence networks, such as RNN. The output of this method is a 3D `Tensor` of shape `[batch_size, T, D]`. `T` is the maximum sequence length for this batch, which could differ from batch to batch. If multiple `feature_columns` are given with `Di` `num_elements` each, their outputs are concatenated. So, the final `Tensor` has shape `[batch_size, T, D0 + D1 + ... + Dn]`. Example: ```python import tensorflow as tf # Behavior of some cells or feature columns may depend on whether we are in # training or inference mode, e.g. applying dropout. training = True rating = tf.feature_column.sequence_numeric_column('rating') watches = tf.feature_column.sequence_categorical_column_with_identity( 'watches', num_buckets=1000) watches_embedding = tf.feature_column.embedding_column(watches, dimension=10) columns = [rating, watches_embedding] features = { 'rating': tf.sparse.from_dense([[1.0,1.1, 0, 0, 0], [2.0,2.1,2.2, 2.3, 2.5]]), 'watches': tf.sparse.from_dense([[2, 85, 0, 0, 0],[33,78, 2, 73, 1]]) } sequence_input_layer = tf.keras.experimental.SequenceFeatures(columns) sequence_input, sequence_length = sequence_input_layer( features, training=training) sequence_length_mask = tf.sequence_mask(sequence_length) hidden_size = 32 rnn_cell = tf.keras.layers.SimpleRNNCell(hidden_size) rnn_layer = tf.keras.layers.RNN(rnn_cell) outputs, state = rnn_layer(sequence_input, mask=sequence_length_mask) ``` TNc s&tjd|||tjjjd|dS)a "Constructs a SequenceFeatures layer. Args: feature_columns: An iterable of dense sequence columns. Valid columns are - `embedding_column` that wraps a `sequence_categorical_column_with_*` - `sequence_numeric_column`. trainable: Boolean, whether the layer's variables will be updated via gradient descent during training. name: Name to give to the SequenceFeatures. **kwargs: Keyword arguments to construct a layer. Raises: ValueError: If any of the `feature_columns` is not a `SequenceDenseColumn`. )feature_columns trainablenameexpected_column_typeN)super__init__tf __internal__feature_columnSequenceDenseColumn)selfr r r kwargs __class__r _/var/www/myenv/lib/python3.10/site-packages/keras/src/feature_column/sequence_feature_column.pyrRs zSequenceFeatures.__init__cCsdS)NTr )rr r r_is_feature_layerlsz"SequenceFeatures._is_feature_layercCs|d|d|fS)Nrr )r input_shapetotal_elementsr r r _target_shapepszSequenceFeatures._target_shapec Cst|ts td||durt}tjj|}g}g}|j D]F}t |j 6z |j ||j |d\}}WntyH| ||j \}}Ynw||||||Wdn1sawYq t||j t|}|||fS)aReturns sequence input corresponding to the `feature_columns`. Args: features: A dict mapping keys to tensors. training: Python boolean or None, indicating whether to the layer is being run in training mode. This argument is passed to the call method of any `FeatureColumn` that takes a `training` argument. For example, if a `FeatureColumn` performed dropout, the column could expose a `training` argument to control whether the dropout should be applied. If `None`, becomes `tf.keras.backend.learning_phase()`. Defaults to `None`. Returns: An `(input_layer, sequence_length)` tuple where: - input_layer: A float `Tensor` of shape `[batch_size, T, D]`. `T` is the maximum sequence length for this batch, which could differ from batch to batch. `D` is the sum of `num_elements` for all `feature_columns`. - sequence_length: An int `Tensor` of shape `[batch_size]`. The sequence length for each example. Raises: ValueError: If features are not a dictionary. z/We expected a dictionary here. Instead we got: N)training) isinstancedict ValueErrorrlearning_phaserrrFeatureTransformationCache_feature_columns name_scoper get_sequence_dense_tensor_state_manager TypeErrorappend_process_dense_tensorkfc"_verify_static_batch_size_equality_assert_all_equal_and_return_verify_and_concat_tensors) rfeaturesrtransformation_cacheoutput_tensorssequence_lengthscolumn dense_tensorsequence_lengthr r rcallssP      zSequenceFeatures.call)TNN) __name__ __module__ __qualname____doc__rpropertyrrr6 __classcell__r r rrr!s/ rc Cst|pdUt|dkr|dWdSg}|ddD]}|tjj|d|q!t|t |dWdWdS1sOwYWddS1s_wYdS)z=Asserts that all tensors are equal and returns the first one.assert_all_equalrrN) rr%lenr)rcompatv1 assert_equalcontrol_dependenciesidentity)tensorsr assert_equal_opstr r rr-s   "r-r7)r; __future__rrrtensorflow.compat.v2r@v2r keras.srcrkeras.src.feature_columnrr+ tensorflow.python.util.tf_exportr_BaseFeaturesLayerrr-r r r rs