B «»ˆdaMã@sêdZddlmmZddlmZddlmZddlmZddlm Z ddlm Z ddl m Z dd l mZdd lmZdd lmZdd lmZdd lmZddlmZddlmZedƒGdd„dee jƒƒZedƒGdd„deƒƒZdS)zFully connected RNN layer.éN)Ú activations)Úbackend)Ú constraints)Ú initializers)Ú regularizers)Ú base_layer)Ú InputSpec)Ú rnn_utils)ÚRNN)ÚDropoutRNNCellMixin)Útf_utils)Ú tf_logging)Ú keras_exportzkeras.layers.SimpleRNNCellc sTeZdZdZd‡fd d „ Zej‡fd d „ƒZdd d„Zddd„Z ‡fdd„Z ‡Z S)Ú SimpleRNNCella« Cell class for SimpleRNN. See [the Keras RNN API guide](https://www.tensorflow.org/guide/keras/rnn) for details about the usage of RNN API. This class processes one step within the whole time sequence input, whereas `tf.keras.layer.SimpleRNN` processes the whole sequence. Args: units: Positive integer, dimensionality of the output space. activation: Activation function to use. Default: hyperbolic tangent (`tanh`). If you pass `None`, no activation is applied (ie. "linear" activation: `a(x) = x`). use_bias: Boolean, (default `True`), whether the layer uses a bias vector. kernel_initializer: Initializer for the `kernel` weights matrix, used for the linear transformation of the inputs. Default: `glorot_uniform`. recurrent_initializer: Initializer for the `recurrent_kernel` weights matrix, used for the linear transformation of the recurrent state. Default: `orthogonal`. bias_initializer: Initializer for the bias vector. Default: `zeros`. kernel_regularizer: Regularizer function applied to the `kernel` weights matrix. Default: `None`. recurrent_regularizer: Regularizer function applied to the `recurrent_kernel` weights matrix. Default: `None`. bias_regularizer: Regularizer function applied to the bias vector. Default: `None`. kernel_constraint: Constraint function applied to the `kernel` weights matrix. Default: `None`. recurrent_constraint: Constraint function applied to the `recurrent_kernel` weights matrix. Default: `None`. bias_constraint: Constraint function applied to the bias vector. Default: `None`. dropout: Float between 0 and 1. Fraction of the units to drop for the linear transformation of the inputs. Default: 0. recurrent_dropout: Float between 0 and 1. Fraction of the units to drop for the linear transformation of the recurrent state. Default: 0. Call arguments: inputs: A 2D tensor, with shape of `[batch, feature]`. states: A 2D tensor with shape of `[batch, units]`, which is the state from the previous time step. For timestep 0, the initial state provided by user will be feed to cell. training: Python boolean indicating whether the layer should behave in training mode or in inference mode. Only relevant when `dropout` or `recurrent_dropout` is used. Examples: ```python inputs = np.random.random([32, 10, 8]).astype(np.float32) rnn = tf.keras.layers.RNN(tf.keras.layers.SimpleRNNCell(4)) output = rnn(inputs) # The output has shape `[32, 4]`. rnn = tf.keras.layers.RNN( tf.keras.layers.SimpleRNNCell(4), return_sequences=True, return_state=True) # whole_sequence_output has shape `[32, 10, 4]`. # final_state has shape `[32, 4]`. whole_sequence_output, final_state = rnn(inputs) ``` ÚtanhTÚglorot_uniformÚ orthogonalÚzerosNçc s |dkrtd|›dƒ‚tjj ¡r4| dd¡|_n| dd¡|_tƒjf|Ž||_ t   |¡|_ ||_ t  |¡|_t  |¡|_t  |¡|_t  |¡|_t  |¡|_t  | ¡|_t  | ¡|_t  | ¡|_t  | ¡|_tdtd| ƒƒ|_tdtd|ƒƒ|_|j |_|j |_dS) NrzQReceived an invalid value for argument `units`, expected a positive integer, got Ú.Úenable_caching_deviceTFgð?g) Ú ValueErrorÚtfÚcompatZv1Z#executing_eagerly_outside_functionsÚpopZ_enable_caching_deviceÚsuperÚ__init__ÚunitsrÚgetÚ activationÚuse_biasrÚkernel_initializerÚrecurrent_initializerÚbias_initializerrÚkernel_regularizerÚrecurrent_regularizerÚbias_regularizerrÚkernel_constraintÚrecurrent_constraintÚbias_constraintÚminÚmaxÚdropoutÚrecurrent_dropoutZ state_sizeZ output_size)Úselfrrr r!r"r#r$r%r&r'r(r)r,r-Úkwargs)Ú __class__©úM/var/www/html/venv/lib/python3.7/site-packages/keras/layers/rnn/simple_rnn.pyrjs2             zSimpleRNNCell.__init__csžtƒ |¡t |¡}|j|d|jfd|j|j|j|d|_ |j|j|jfd|j |j |j |d|_ |jrŽ|j|jfd|j|j|j|d|_nd|_d|_dS)NéÿÿÿÿÚkernel)ÚshapeÚnameZ initializerZ regularizerÚ constraintÚcaching_deviceÚrecurrent_kernelÚbiasT)rÚbuildr r8Z add_weightrr!r$r'r4r"r%r(r9r r#r&r)r:Zbuilt)r.Z input_shapeZdefault_caching_device)r0r1r2r; s4       zSimpleRNNCell.buildc CsÆtj |¡r|dn|}| ||¡}| ||¡}|dk rLt |||j¡}nt ||j¡}|jdk rrt  ||j¡}|dk r‚||}|t ||j ¡}|j dk r¨|  |¡}tj |¡rº|gn|} || fS)Nr) rÚnestZ is_nestedZget_dropout_mask_for_cellZ#get_recurrent_dropout_mask_for_cellrÚdotr4r:Zbias_addr9r) r.ÚinputsZstatesÚtrainingZ prev_outputZdp_maskZ rec_dp_maskÚhÚoutputÚ new_stater1r1r2ÚcallÁs     zSimpleRNNCell.callcCst ||||¡S)N)r Z#generate_zero_filled_state_for_cell)r.r>Z batch_sizeÚdtyper1r1r2Úget_initial_stateØszSimpleRNNCell.get_initial_statecs°|jt |j¡|jt |j¡t |j¡t |j¡t  |j ¡t  |j ¡t  |j ¡t  |j¡t  |j¡t  |j¡|j|jdœ}| t |¡¡tƒ ¡}tt| ¡ƒt| ¡ƒƒS)N)rrr r!r"r#r$r%r&r'r(r)r,r-)rrÚ serializerr rr!r"r#rr$r%r&rr'r(r)r,r-Úupdater Ú config_for_enable_caching_devicerÚ get_configÚdictÚlistÚitems)r.ÚconfigÚ base_config)r0r1r2rIÝs,       zSimpleRNNCell.get_config) rTrrrNNNNNNrr)N)NNN) Ú__name__Ú __module__Ú __qualname__Ú__doc__rr Zshape_type_conversionr;rCrErIÚ __classcell__r1r1)r0r2r%s$C'!  rzkeras.layers.SimpleRNNcsðeZdZdZd.‡fd d „ Zd/‡fd d „ Zedd„ƒZedd„ƒZedd„ƒZ edd„ƒZ edd„ƒZ edd„ƒZ edd„ƒZ edd„ƒZedd„ƒZed d!„ƒZed"d#„ƒZed$d%„ƒZed&d'„ƒZed(d)„ƒZ‡fd*d+„Zed,d-„ƒZ‡ZS)0Ú SimpleRNNa6Fully-connected RNN where the output is to be fed back to input. See [the Keras RNN API guide](https://www.tensorflow.org/guide/keras/rnn) for details about the usage of RNN API. Args: units: Positive integer, dimensionality of the output space. activation: Activation function to use. Default: hyperbolic tangent (`tanh`). If you pass None, no activation is applied (ie. "linear" activation: `a(x) = x`). use_bias: Boolean, (default `True`), whether the layer uses a bias vector. kernel_initializer: Initializer for the `kernel` weights matrix, used for the linear transformation of the inputs. Default: `glorot_uniform`. recurrent_initializer: Initializer for the `recurrent_kernel` weights matrix, used for the linear transformation of the recurrent state. Default: `orthogonal`. bias_initializer: Initializer for the bias vector. Default: `zeros`. kernel_regularizer: Regularizer function applied to the `kernel` weights matrix. Default: `None`. recurrent_regularizer: Regularizer function applied to the `recurrent_kernel` weights matrix. Default: `None`. bias_regularizer: Regularizer function applied to the bias vector. Default: `None`. activity_regularizer: Regularizer function applied to the output of the layer (its "activation"). Default: `None`. kernel_constraint: Constraint function applied to the `kernel` weights matrix. Default: `None`. recurrent_constraint: Constraint function applied to the `recurrent_kernel` weights matrix. Default: `None`. bias_constraint: Constraint function applied to the bias vector. Default: `None`. dropout: Float between 0 and 1. Fraction of the units to drop for the linear transformation of the inputs. Default: 0. recurrent_dropout: Float between 0 and 1. Fraction of the units to drop for the linear transformation of the recurrent state. Default: 0. return_sequences: Boolean. Whether to return the last output in the output sequence, or the full sequence. Default: `False`. return_state: Boolean. Whether to return the last state in addition to the output. Default: `False` go_backwards: Boolean (default False). If True, process the input sequence backwards and return the reversed sequence. stateful: Boolean (default False). If True, the last state for each sample at index i in a batch will be used as initial state for the sample of index i in the following batch. unroll: Boolean (default False). If True, the network will be unrolled, else a symbolic loop will be used. Unrolling can speed-up a RNN, although it tends to be more memory-intensive. Unrolling is only suitable for short sequences. Call arguments: inputs: A 3D tensor, with shape `[batch, timesteps, feature]`. mask: Binary tensor of shape `[batch, timesteps]` indicating whether a given timestep should be masked. An individual `True` entry indicates that the corresponding timestep should be utilized, while a `False` entry indicates that the corresponding timestep should be ignored. training: Python boolean indicating whether the layer should behave in training mode or in inference mode. This argument is passed to the cell when calling it. This is only relevant if `dropout` or `recurrent_dropout` is used. initial_state: List of initial state tensors to be passed to the first call of the cell. Examples: ```python inputs = np.random.random([32, 10, 8]).astype(np.float32) simple_rnn = tf.keras.layers.SimpleRNN(4) output = simple_rnn(inputs) # The output has shape `[32, 4]`. simple_rnn = tf.keras.layers.SimpleRNN( 4, return_sequences=True, return_state=True) # whole_sequence_output has shape `[32, 10, 4]`. # final_state has shape `[32, 4]`. whole_sequence_output, final_state = simple_rnn(inputs) ``` rTrrrNçFc s´d|kr| d¡t d¡d|kr4d| d¡i}ni}t|f|||||||| | | | ||| d¡| dd¡dœ|—Ž}tƒj|f|||||dœ|—Žt | ¡|_t d d g|_ dS) NÚimplementationzhThe `implementation` argument in `SimpleRNN` has been deprecated. Please remove it from your layer call.rrDÚ trainableT)rr r!r"r#r$r%r&r'r(r)r,r-rDrW)Úreturn_sequencesÚ return_stateÚ go_backwardsÚstatefulÚunrollé)Úndim) rÚloggingÚwarningrrrrrÚactivity_regularizerrZ input_spec)r.rrr r!r"r#r$r%r&rar'r(r)r,r-rXrYrZr[r\r/Z cell_kwargsÚcell)r0r1r2rUsF  zSimpleRNN.__init__cstƒj||||dS)N)Úmaskr?Ú initial_state)rrC)r.r>rcr?rd)r0r1r2rC™szSimpleRNN.callcCs|jjS)N)rbr)r.r1r1r2ržszSimpleRNN.unitscCs|jjS)N)rbr)r.r1r1r2r¢szSimpleRNN.activationcCs|jjS)N)rbr )r.r1r1r2r ¦szSimpleRNN.use_biascCs|jjS)N)rbr!)r.r1r1r2r!ªszSimpleRNN.kernel_initializercCs|jjS)N)rbr")r.r1r1r2r"®szSimpleRNN.recurrent_initializercCs|jjS)N)rbr#)r.r1r1r2r#²szSimpleRNN.bias_initializercCs|jjS)N)rbr$)r.r1r1r2r$¶szSimpleRNN.kernel_regularizercCs|jjS)N)rbr%)r.r1r1r2r%ºszSimpleRNN.recurrent_regularizercCs|jjS)N)rbr&)r.r1r1r2r&¾szSimpleRNN.bias_regularizercCs|jjS)N)rbr')r.r1r1r2r'ÂszSimpleRNN.kernel_constraintcCs|jjS)N)rbr()r.r1r1r2r(ÆszSimpleRNN.recurrent_constraintcCs|jjS)N)rbr))r.r1r1r2r)ÊszSimpleRNN.bias_constraintcCs|jjS)N)rbr,)r.r1r1r2r,ÎszSimpleRNN.dropoutcCs|jjS)N)rbr-)r.r1r1r2r-ÒszSimpleRNN.recurrent_dropoutcsÂ|jt |j¡|jt |j¡t |j¡t |j¡t  |j ¡t  |j ¡t  |j ¡t  |j ¡t |j¡t |j¡t |j¡|j|jdœ}tƒ ¡}| t |j¡¡|d=tt| ¡ƒt| ¡ƒƒS)N)rrr r!r"r#r$r%r&rar'r(r)r,r-rb)rrrFrr rr!r"r#rr$r%r&rarr'r(r)r,r-rrIrGr rHrbrJrKrL)r.rMrN)r0r1r2rIÖs2       zSimpleRNN.get_configcCsd|kr| d¡|f|ŽS)NrV)r)ÚclsrMr1r1r2Ú from_configùs zSimpleRNN.from_config)rTrrrNNNNNNNrUrUFFFFF)NNN)rOrPrQrRrrCÚpropertyrrr r!r"r#r$r%r&r'r(r)r,r-rIÚ classmethodrfrSr1r1)r0r2rTýsJV/               #rT)rRZtensorflow.compat.v2rZv2rZkerasrrrrrZ keras.enginerZkeras.engine.input_specrZkeras.layers.rnnr Zkeras.layers.rnn.base_rnnr Z'keras.layers.rnn.dropout_rnn_cell_mixinr Z keras.utilsr Ztensorflow.python.platformr r_Z tensorflow.python.util.tf_exportrZBaseRandomLayerrrTr1r1r1r2Ús&              X