2Vh)dZddlmZddlmZddlmZddlmZddlmZddlmZddl m Z dd l m Z dd l mZdd lmZdd lmZGd deZy)z2Keras base class for transpose convolution layers.) activations) constraints) initializers)ops) regularizers)standardize_data_format)#compute_conv_transpose_output_shape) InputSpec)Layer)standardize_padding)standardize_tuplec`eZdZdZ dfd ZdZdZdZfdZxZ S)BaseConvTransposea) Abstract N-D transposed convolution layer. The need for transposed convolutions generally arises from the desire to use a transformation going in the opposite direction of a normal convolution, i.e., from something that has the shape of the output of some convolution to something that has the shape of its input while maintaining a connectivity pattern that is compatible with said convolution. Args: rank: int, the rank of the transposed convolution, e.g. 2 for 2D transposed convolution. filters: int, the dimension of the output space (the number of filters in the transposed convolution). kernel_size: int or tuple/list of `rank` integers, specifying the size of the transposed convolution window. strides: int or tuple/list of `rank` integers, specifying the stride length of the transposed convolution. If only one int is specified, the same stride size will be used for all dimensions. `strides > 1` is incompatible with `dilation_rate > 1`. padding: string, either `"valid"` or `"same"` (case-insensitive). `"valid"` means no padding. `"same"` results in padding evenly to the left/right or up/down of the input such that output has the same height/width dimension as the input. data_format: string, either `"channels_last"` or `"channels_first"`. The ordering of the dimensions in the inputs. `"channels_last"` corresponds to inputs with shape `(batch, steps, features)` while `"channels_first"` corresponds to inputs with shape `(batch, features, steps)`. It defaults to the `image_data_format` value found in your Keras config file at `~/.keras/keras.json`. If you never set it, then it will be `"channels_last"`. dilation_rate: int or tuple/list of `rank` integers, specifying the dilation rate to use for dilated convolution. If only one int is specified, the same dilation rate will be used for all dimensions. activation: Activation function. If `None`, no activation is applied. use_bias: bool, if `True`, bias will be added to the output. kernel_initializer: Initializer for the convolution kernel. If `None`, the default initializer (`"glorot_uniform"`) will be used. bias_initializer: Initializer for the bias vector. If `None`, the default initializer (`"zeros"`) will be used. kernel_regularizer: Optional regularizer for the convolution kernel. bias_regularizer: Optional regularizer for the bias vector. activity_regularizer: Optional regularizer function for the output. kernel_constraint: Optional projection function to be applied to the kernel after being updated by an `Optimizer` (e.g. used to implement norm constraints or value constraints for layer weights). The function must take as input the unprojected variable and must return the projected variable (which must have the same shape). Constraints are not safe to use when doing asynchronous distributed training. bias_constraint: Optional projection function to be applied to the bias after being updated by an `Optimizer`. c t|d|||d|||_||_t ||d|_t ||d|_t ||d|_t||_ |d|_ nt ||d|_ t||_ tj| |_| |_t#j| |_t#j| |_t)j| |_t)j||_t/j||_t/j||_t5|jdz|_|j|_ |j(|jdkrt9d |jd t;|j st9d |j d t;|j st9d |j d t=|j d kDr>t=|jd kDr%t9d|j d|jyy)N) trainablenameactivity_regularizer kernel_sizestrides dilation_rateoutput_padding)min_ndimrz[Invalid value for argument `filters`. Expected a strictly positive value. Received filters=.zBThe argument `kernel_size` cannot contain 0. Received kernel_size=z;The argument `strides` cannot contains 0. Received strides=zW`strides > 1` not supported in conjunction with `dilation_rate > 1`. Received: strides=z and dilation_rate=)super__init__rankfiltersr rrrr paddingrr data_formatrget activationuse_biasrkernel_initializerbias_initializerrkernel_regularizerbias_regularizerrkernel_constraintbias_constraintr input_spec ValueErrorallmax)selfrr rrr!rr"rr$r%r&r'r(r)rr*r+rrkwargs __class__s b/home/dcms/DCMS/lib/python3.12/site-packages/keras/src/layers/convolutional/base_conv_transpose.pyrzBaseConvTranspose.__init__GsY.  !5     ,[$ N($ B . 4 +73  !"&D "3 #D  3;?%//*5  "."2"23E"F , 0 01A B"."2"23E"F , 0 01A B!,1B!C*?#TYY];++ << # (9448LL>D  4##$#//03  4<< <<.+  t|| q S););%,,H!!%!3!3 46 &A c  |jdk(rd}|d}nd}|d}t|jdz||i|_|j|j |fz}|j d||j|j|jd|j|_ |jrQ|j d |j f|j|j|jd|j|_yd|_y) N channels_lastrr)raxeskernelT)rshape initializer regularizer constraintrdtypebias)r"r rr,rr add_weightr&r(r*r>r9r%r'r)r+r?)r0 input_shape channel_axis input_channel kernel_shapes r3buildzBaseConvTranspose.builds    .L'OML'NM#YY], )F '' LL +  oo////--**&   ==||o 11 11//jj(DIDIr4c "tj||jt|j|j |j |j|j}|jr|jdk(r!d|jdzz|jfz}nd|jfd|jzz}tj|j|}tj||}|j|j|S|S)N)rr!rrr"r6)rr)rconv_transposer9listrr!rrr"r%rr reshaper?addr$)r0inputsoutputs bias_shaper?s r3callzBaseConvTranspose.calls$$  KK&LL..,,((  ==?2!TYY]3t||oE . 1AA ;;tyy*5Dgggt,G ?? &??7+ +r4c t||j|j|j|j|j |j |jS)N)rr!rr"r)r rr rr!rr"r)r0rAs r3compute_output_shapez&BaseConvTranspose.compute_output_shapesK2     LLLLLL..((,,  r4ct|}|j|j|j|j |j |j|jtj|j|jtj|jtj|jt!j|j"t!j|j$t!j|j&t)j|j*t)j|j,d|S)N)r rrr!r"rr$r%r&r'r(r)rr*r+)r get_configupdater rrr!r"rr serializer$r%rr&r'rr(r)rrr*r+)r0configr2s r3rRzBaseConvTranspose.get_configs#% <<#//<<<<#//!%!3!3)33DOOD MM&2&<&<++'%1$:$:))%'3&<&<++'%1$:$:))%)5(>(>--)&1%:%:**&$/#8#89M9M#N7  > r4)rvalidNNrNTglorot_uniformzerosNNNNNTN) __name__ __module__ __qualname____doc__rrErNrPrR __classcell__)r2s@r3rrs\2r+ ! )Pd#J.  !!r4rN)r\ keras.srcrrrrrkeras.src.backendr&keras.src.backend.common.backend_utilsr keras.src.layers.input_specr keras.src.layers.layerr #keras.src.utils.argument_validationr r rrr4r3rds;8!!""52(CAppr4