jMddlZddlmZddlmZddlmZGddeZGddeZ Gdd eZ dS) N)Union)Modulec\eZdZdZ ddedededed ed ed ed effd ZdZdZxZ S)Conv1daApplies a 1-dimensional convolution over the multi-channel input sequence. The channels are expected to be last i.e. the input shape should be ``NLC`` where: * ``N`` is the batch dimension * ``L`` is the sequence length * ``C`` is the number of input channels Args: in_channels (int): The number of input channels out_channels (int): The number of output channels kernel_size (int): The size of the convolution filters stride (int, optional): The stride when applying the filter. Default: ``1``. padding (int, optional): How many positions to 0-pad the input with. Default: ``0``. dilation (int, optional): The dilation of the convolution. groups (int, optional): The number of groups for the convolution. Default: ``1``. bias (bool, optional): If ``True`` add a learnable bias to the output. Default: ``True`` rT in_channels out_channels kernel_sizestridepaddingdilationgroupsbiasc t||zdkrtd|d|dtjd||zz } t j| | ||||zf|_|rt j |f|_ ||_ ||_ ||_ ||_dS)NrThe number of input channels (-) must be divisible by the number of groups ()rlowhighshape)super__init__ ValueErrormathsqrtmxrandomuniformweightzerosrr r r r selfrr r r r r rrscale __class__s c/lsinfo/ai/hellotax_ai/base_platform/venv/lib/python3.11/site-packages/mlx/nn/layers/convolution.pyrzConv1d.__init__"s   1 $ $@@@6<@@@   !{[89::i''kV.CD(    2,11DI     c|jjdd|jjdd|jjdd|jd|jd|jd |jd d |vS) N, r, kernel_size=r , stride= , padding= , dilation= , groups=, bias=rr rr r r rr#s r& _extra_reprzConv1d._extra_reprCs{ $ % % (9!(< % %;,Q/ % %:>+ % %| % %04  % %k % %dN  % % r'ctj||j|j|j|j|j}d|vr ||jz}|SNr)rconv1dr r r r rrr#xys r&__call__zConv1d.__call__LE I t{DKt}dk   T>>DI Ar'rrrrT) __name__ __module__ __qualname____doc__intboolrr3r: __classcell__r%s@r&rr s8     B   r'rceZdZdZ ddededeeefdeeefd eeefd eeefd ed effd ZdZ dZ xZ S)Conv2daApplies a 2-dimensional convolution over the multi-channel input image. The channels are expected to be last i.e. the input shape should be ``NHWC`` where: * ``N`` is the batch dimension * ``H`` is the input image height * ``W`` is the input image width * ``C`` is the number of input channels Args: in_channels (int): The number of input channels. out_channels (int): The number of output channels. kernel_size (int or tuple): The size of the convolution filters. stride (int or tuple, optional): The size of the stride when applying the filter. Default: ``1``. padding (int or tuple, optional): How many positions to 0-pad the input with. Default: ``0``. dilation (int or tuple, optional): The dilation of the convolution. groups (int, optional): The number of groups for the convolution. Default: ``1``. bias (bool, optional): If ``True`` add a learnable bias to the output. Default: ``True`` rrTrr r r r r rrc t||zdkrtd|d|dtd|||f\}}}t jd||dz|dzz } t j| | |g|||zR|_ |rt j |f|_ ||_ ||_ ||_||_dS)Nrrrrc8t|tr||fn|SN isinstancerAr8s r&z!Conv2d.__init__..s 1c 2 29q!ffr'rr)rrrmaprrrrrr r!rr r r rr"s r&rzConv2d.__init__ns1   1 $ $@@@6<@@@  (+ 9 9 &' *( ( $ VW !{[^;k!nLMNNi''E+E{f/DEE(    2,11DI     r'c|jjdd|jjdd|jjddd|jd|jd |jd |jd d |vS) Nr)r*rr+rr,r-r.r/r0rr1r2s r&r3zConv2d._extra_reprs{ $ % % (9!(< % %;,QqS1 % %<@K % %| % %04  % %k % %dN  % % r'ctj||j|j|j|j|j}d|vr ||jz}|Sr5)rconv2dr r r r rrr7s r&r:zConv2d.__call__r;r'r< r=r>r?r@rArtuplerBrr3r:rCrDs@r&rFrFUs:%&%&&'####3:& # c5j! # sEz" #U #########J   r'rFceZdZdZ ddededeeefdeeefd eeefd eeefd effd Zd Z dZ xZ S)Conv3daApplies a 3-dimensional convolution over the multi-channel input image. The channels are expected to be last i.e. the input shape should be ``NDHWC`` where: * ``N`` is the batch dimension * ``D`` is the input image depth * ``H`` is the input image height * ``W`` is the input image width * ``C`` is the number of input channels Args: in_channels (int): The number of input channels. out_channels (int): The number of output channels. kernel_size (int or tuple): The size of the convolution filters. stride (int or tuple, optional): The size of the stride when applying the filter. Default: ``1``. dilation (int or tuple, optional): The dilation of the convolution. padding (int or tuple, optional): How many positions to 0-pad the input with. Default: ``0``. bias (bool, optional): If ``True`` add a learnable bias to the output. Default: ``True`` rrTrr r r r r rcttd|||f\}}}tjd||dz|dz|dzz }t j| ||g||R|_|rt j |f|_ ||_ ||_ ||_ dS)Nc:t|tr|||fn|SrIrJrLs r&rMz!Conv3d.__init__..s:a#5#5QO P  i'';+;{;;(    2,11DI    r'c|jjdd|jjdd|jjddd|jd|jd |jd d |v S) Nr)r*rr+rr,r-r.r0r)r rr r r r2s r&r3zConv3d._extra_reprs{ $ % % (9!(< % %;,QqS1 % %<@K % %| % %04  % %dN % % r'cztj||j|j|j|j}d|vr ||jz}|Sr5)rconv3dr r r r rr7s r&r:zConv3d.__call__s: Iadk4< O O T>>DI Ar')rrrTrSrDs@r&rVrVs8%&%&&'!!!!3:& ! c5j! ! sEz" !U #!!!!!!!>   r'rV) rtypingrmlx.corecorermlx.nn.layers.baserrrFrVr'r&rcs %%%%%%HHHHHVHHHVMMMMMVMMM`CCCCCVCCCCCr'