o 2h3@s`ddlZddlmZddlmZddlmZddlmZddlm Z edGdd d eZ dS) N)ops)utils) keras_export)Layer)serialization_libzkeras.layers.BidirectionalcseZdZdZ   dfdd ZddZddd Z   dd d Zd d ZddZ e ddZ dddZ ddZ fddZedddZZS) Bidirectionala~ Bidirectional wrapper for RNNs. Args: layer: `keras.layers.RNN` instance, such as `keras.layers.LSTM` or `keras.layers.GRU`. It could also be a `keras.layers.Layer` instance that meets the following criteria: 1. Be a sequence-processing layer (accepts 3D+ inputs). 2. Have a `go_backwards`, `return_sequences` and `return_state` attribute (with the same semantics as for the `RNN` class). 3. Have an `input_spec` attribute. 4. Implement serialization via `get_config()` and `from_config()`. Note that the recommended way to create new RNN layers is to write a custom RNN cell and use it with `keras.layers.RNN`, instead of subclassing `keras.layers.Layer` directly. When `return_sequences` is `True`, the output of the masked timestep will be zero regardless of the layer's original `zero_output_for_mask` value. merge_mode: Mode by which outputs of the forward and backward RNNs will be combined. One of `{"sum", "mul", "concat", "ave", None}`. If `None`, the outputs will not be combined, they will be returned as a list. Defaults to `"concat"`. backward_layer: Optional `keras.layers.RNN`, or `keras.layers.Layer` instance to be used to handle backwards input processing. If `backward_layer` is not provided, the layer instance passed as the `layer` argument will be used to generate the backward layer automatically. Note that the provided `backward_layer` layer should have properties matching those of the `layer` argument, in particular it should have the same values for `stateful`, `return_states`, `return_sequences`, etc. In addition, `backward_layer` and `layer` should have different `go_backwards` argument values. A `ValueError` will be raised if these requirements are not met. Call arguments: The call arguments for this layer are the same as those of the wrapped RNN layer. Beware that when passing the `initial_state` argument during the call of this layer, the first half in the list of elements in the `initial_state` list will be passed to the forward RNN call and the last half in the list of elements will be passed to the backward RNN call. Note: instantiating a `Bidirectional` layer from an existing RNN layer instance will not reuse the weights state of the RNN layer instance -- the `Bidirectional` layer will have freshly initialized weights. Examples: ```python model = Sequential([ Input(shape=(5, 10)), Bidirectional(LSTM(10, return_sequences=True), Bidirectional(LSTM(10)), Dense(5, activation="softmax"), ]) model.compile(loss='categorical_crossentropy', optimizer='rmsprop') # With custom backward layer forward_layer = LSTM(10, return_sequences=True) backward_layer = LSTM(10, activation='relu', return_sequences=True, go_backwards=True) model = Sequential([ Input(shape=(5, 10)), Bidirectional(forward_layer, backward_layer=backward_layer), Dense(5, activation="softmax"), ]) model.compile(loss='categorical_crossentropy', optimizer='rmsprop') ``` concatNc  sft|ts td||durt|tstd||dvr(td|dtjdi|t|}dt|j d|dd<t ||_ |durlt|}d |dd <d t|j d |dd<t ||_ n||_ | d d }||j ||j ||_|rt|}|d|d|j _||dd|j _|j|_|j|_|j|_d |_|j|_dS)NzXPlease initialize `Bidirectional` layer with a `keras.layers.Layer` instance. Received: zG`backward_layer` need to be a `keras.layers.Layer` instance. Received: )summulaverNzInvalid merge mode. Received: zC. Merge mode should be one of {"sum", "mul", "ave", "concat", None}forward_confignameT go_backwards backward_cSs t|dddur|j|_dSdS)Nzero_output_for_mask)getattrreturn_sequencesr)layerr]/var/www/html/chatgem/venv/lib/python3.10/site-packages/keras/src/layers/rnn/bidirectional.pyforce_zero_output_for_masks z:Bidirectional.__init__..force_zero_output_for_maskr) isinstancer ValueErrorsuper__init__rserialize_keras_objectr removeprefixrdeserialize_keras_object forward_layerbackward_layer_verify_layer_config merge_modeleninitial_weightsstatefulr return_statesupports_masking input_spec) selfrr#weightsr!kwargsr rnw __class__rrrSs\         zBidirectional.__init__c Csz|jj|jjkrtd|jjd|jjd}|D] }t|j|}t|j|}||kr:td|d|d|dqdS) zBEnsure the forward and backward layers have valid common property.zrForward layer and backward layer should have different `go_backwards` value. Received: forward_layer.go_backwards z, backward_layer.go_backwards=)r&rr'zTForward layer and backward layer are expected to have the same value for attribute "z", got "z" for forward layer and "z" for backward layerN)r rr!rr)r*common_attributesa forward_valuebackward_valuerrrr"s0   z"Bidirectional._verify_layer_configcCs|j|}|jr|d|dd}}|jdkr*t|}|dd9<t|}n |jdur3||g}|jrL|jdurCt|||St|g||St|S)Nrrr)r compute_output_shaper'r#listtuple)r*sequences_shapeinitial_state_shape output_shape state_shaperrrr6s     z"Bidirectional.compute_output_shapecCsi}|jjr ||d<|jjr||d<|dur.||}}t|d}|d|} ||d} n ||}}d\} } |j|fd| i|} |j|fd| i|} |jrft| dd| dd} | d} | d} t| |j } t| |j } |j r~tj | dd} |j d krtj | | gd d}n3|j d kr| | }n)|j d kr| | d}n|j d kr| | }n|j dur| | f}n td|j d|jr|j dur|| S|f| S|S)NtrainingmaskrNN initial_stater4r)axisrr5r r r z/Unrecognized value for `merge_mode`. Received: z0Expected one of {"concat", "sum", "ave", "mul"}.)r _call_has_training_arg_call_has_mask_argr$r!r'r8rcast compute_dtyperflipr# concatenater)r* sequencesr@r>r=r,forward_inputsbackward_inputshalf forward_statebackward_stateyy_revstatesoutputrrrcallsl              zBidirectional.callcCs |dSN) reset_stater*rrr reset_statess zBidirectional.reset_statescCs&|jstd|j|jdS)NzLayer must be stateful.)r&AttributeErrorr rTr!rUrrrrTs zBidirectional.reset_statecCs(|jjr|jjrt|jj|jjSdSrS)r rPr!r8rUrrrrP szBidirectional.statescCs0|jjs |j||jjs|j|dSdSrS)r builtbuildr!)r*r9r:rrrrYs  zBidirectional.buildcCst|tr |d}|jr|js||f}n |}n|jsdnd}|jr@|jdur@dd|jD}t|tr9||dS|f|dS|S)Nrr?css|]}dVqdSrSr).0_rrr #sz-Bidirectional.compute_mask..r)rr7rr#r'rP)r*r[r> output_mask state_maskrrr compute_masks    zBidirectional.compute_maskcs@d|ji}t|j|d<t|j|d<t}i||S)Nr#rr!)r#rrr r!r get_config)r*r base_configr.rrr`)s   zBidirectional.get_configcCs\t|}tj|d|d|d<|dd}|dur%tj||d}||d<|di|}|S)Nr)custom_objectsr!r)copydeepcopyrrpop)clsr rbbackward_layer_configr!rrrr from_config4s  zBidirectional.from_config)rNNrS)NNN)__name__ __module__ __qualname____doc__rr"r6rRrVrTpropertyrPrYr_r` classmethodrh __classcell__rrr.rr s*JB  A    r) rc keras.srcrrkeras.src.api_exportrkeras.src.layers.layerrkeras.src.savingrrrrrrs