6bi(+lSrSSKJs Jr SSKJr SSKJr SSKJ r SSKJ r Sr SrS rS rS rS rg) z%Utilities for keras functional model.N)backend) input_layer) keras_tensor)nodezFound unexpected instance while processing input tensors for keras functional model. Expecting KerasTensor which is from tf.keras.Input() or output from keras layer call(). Got: {}c[R"U5(d[[R U55eUR R $)aCheck if tensor is directly generated from `tf.keras.Input`. This check is useful when constructing the functional model, since we will need to clone Nodes and KerasTensors if the model is building from non input tensor. Args: tensor: A `KerasTensor` as inputs to the functional model. Returns: bool. Whether the tensor is directly generated from `tf.keras.Input`. Raises: ValueError: if the tensor is not a KerasTensor instance. ) node_moduleis_keras_tensor ValueError"_KERAS_TENSOR_TYPE_CHECK_ERROR_MSGformatris_inputtensors ^/home/james-whalen/.local/lib/python3.13/site-packages/tf_keras/src/engine/functional_utils.pyis_input_keras_tensorrs<  & &v . .;BB6JKK ;;  c*[RRU5n[RRU5nX#-H<n[R"U5(aM [ [ RU55e [UVs/sHn[U5PM sn5n[5n/n/n [5n UHnURUR5 M U(aURS5n [U 5U ;aM)U R[U 55 U RU 5 U RHvn[U5U;aUR[U55 M.URn U R(a[ SRU55eURU 5 Mx U(aMXg:wa=UVs/sHn[U5U;dMUPM n n[ SRU 55eU $s snfs snf)a Fetch all Nodes in the graph defined by "inputs" and "outputs". This method is used to find and then clone Nodes when creating a new sub-model from an existing functional model. Args: inputs: A nested structure of KerasTensor to use as model inputs. outputs: A nested structure of KerasTensor to use as model outputs. Returns: A list of Nodes that are connected to the inputs and outputs. Raises: ValueError: when inputs and outputs are disconnected or in case of unexpected objects in the inputs/outputs. rzFound input tensor cannot be reached given provided output tensors. Please make sure the tensor {} is included in the model inputs when building functional model.zHFound unvisited input tensors that are disconnected from the outputs: {})tfnestflattenrr r r r setidappendrpopadd keras_inputsr )inputsoutputsstart_keras_tensorsend_keras_tensorstktend_ids end_ids_foundnodes_to_visitnodes_in_graphnode_id_visitedr inbound_nodeunvisited_inputss r find_nodes_by_inputs_and_outputsr*4s4''//'2/ 4**1--?FFqIJ J5$56$5b2b6$567GEMNNeO aff%! !!!$ d8 & BtH%d###B"v !!"R&)77L $$ ()/r   ! !, /%$ .:* *2bfM.IB*   $f%56   c7T s H H'Hcz[X5n/n/n0n[RRU5H{nURR (a UR U5 Xe[U5'M>[U5n[R"US9nUR U5 X[U5'M} [RRX5n[RRU5H=n [U 5nU RUl UR U5 Xu[U 5'M? [RRX5nUHfn [U RU5n [U RU5n [U R U5n ["R$"U R&U U U S9 Mh X44$)axClone the `Node` between the inputs and output tensors. This function is used to create a new functional model from any intermediate keras tensors. The clone of the nodes mimic the behavior of reconstructing the functional graph network by re-executing all the __call__ methods. The cloned nodes will be appended to the layers. Note that a new tf.keras.Inputs will be created for any items in the `inputs` Args: inputs: A nested structure of keras_tensors. outputs: A nested structure of keras_tensors. Returns: A pair of inputs and outputs, with cloned keras_tensors. They can be used to create a new functional model. r) call_args call_kwargsr)r*rrrrr rr_clone_keras_tensorinput_layer_moduleInputpack_sequence_as_keras_historyclone_keras_tensorsoutput_tensorsr,r-rNodelayer)rrnodes_to_clone cloned_inputscloned_outputs kt_id_mappingkt_inputcpy cloned_input kt_outputr output_copycall_args_copycall_kwargs_copys rclone_graph_nodesrBsx&6fFNMN MGGOOF+ == ! !   **2"X, '&h/C-333?L   .*6"X, ',GG,,VCMWW__W- !), '55c"'*bm$.WW--gFN*$*=*=}M ,T^^]K.t/?/?O  JJ$(  "  ((rc/n[RRU5Hn[R"U5(aZ[ U5U;aU[ U5nO)[ U5nURUlXA[ U5'URU5 MxURU5 M [RRX5$)aClone the keras tensors from the inputs. For any KerasTensor instance in the `args`, a new copy of KerasTensor will be created if it has not been cloned yet (by checking the `keras_tensor_mapping`). For any other types, the instance will be unchanged. This function is useful for cloning the Nodes since KerasTensor can't be reused across the models. Args: args: A nested structure of objects, which could contain KerasTensor. keras_tensor_mapping: A dict contains the ID of original KerasTensor, and the cloned KerasTensor instance. The dict will be updated with newly copied KerasTensor instances within this method. Returns: Same structure as inputs, with KerasTensor cloned. ) rrrrr rr.r2rr1)argskeras_tensor_mappingresultobjr<s rr3r3s"Fwwt$  & &s + +#w..*2c73*#.%(%7%7"03RW- MM#  MM# % 77 # #D 11rc$[R"5nUR5 [R"U5n[R "U5sSSS5 sSSS5 $!,(df  O=fSSS5 g!,(df  g=f)a Create an identical keras_tensor based on the input. We use keras_tensor_to_placeholder and keras_tensor_from_tensor to make sure inferred shape are not lost during the copy. Args: kt: the input KerasTensor. Returns: An identical copy of the input KerasTensor. N)r_scratch_graph as_defaultrkeras_tensor_to_placeholderkeras_tensor_from_tensor)r" scratch_graph placeholders rr.r.sb    !]  % % '&BB2FK88E( ' " ! ' ' ' " ! !s"B,A& B& A4 0B B)__doc__tensorflow.compat.v2compatv2r tf_keras.srcrtf_keras.src.enginerr/rrrr rr*rBr3r.rrrVsE,!! A,31# *QhH)V2BFr