o i e:@sdZddlmZddlmZddlmZddlmZddlmZddlmZddl m Z dd l m Z dd l mZdd l mZdd lmZed GdddeZdS)z%Locally-connected layer for 1D input.) activations)backend) constraints) initializers) regularizers)Layer) InputSpec)locally_connected_utils) conv_utils)tf_utils) keras_exportzkeras.layers.LocallyConnected1DcsxeZdZdZ             dfdd Zed d Zejd d Z ejddZ ddZ fddZ Z S)LocallyConnected1DaLocally-connected layer for 1D inputs. The `LocallyConnected1D` layer works similarly to the `Conv1D` layer, except that weights are unshared, that is, a different set of filters is applied at each different patch of the input. Note: layer attributes cannot be modified after the layer has been called once (except the `trainable` attribute). Example: ```python # apply a unshared weight convolution 1d of length 3 to a sequence with # 10 timesteps, with 64 output filters model = Sequential() model.add(LocallyConnected1D(64, 3, input_shape=(10, 32))) # now model.output_shape == (None, 8, 64) # add a new conv1d on top model.add(LocallyConnected1D(32, 3)) # now model.output_shape == (None, 6, 32) ``` Args: filters: Integer, the dimensionality of the output space (i.e. the number of output filters in the convolution). kernel_size: An integer or tuple/list of a single integer, specifying the length of the 1D convolution window. strides: An integer or tuple/list of a single integer, specifying the stride length of the convolution. padding: Currently only supports `"valid"` (case-insensitive). `"same"` may be supported in the future. `"valid"` means no padding. data_format: A string, one of `channels_last` (default) or `channels_first`. The ordering of the dimensions in the inputs. `channels_last` corresponds to inputs with shape `(batch, length, channels)` while `channels_first` corresponds to inputs with shape `(batch, channels, length)`. When unspecified, uses `image_data_format` value found in your Keras config file at `~/.keras/keras.json` (if exists) else 'channels_last'. Defaults to 'channels_last'. activation: Activation function to use. If you don't specify anything, no activation is applied (ie. "linear" activation: `a(x) = x`). use_bias: Boolean, whether the layer uses a bias vector. kernel_initializer: Initializer for the `kernel` weights matrix. bias_initializer: Initializer for the bias vector. kernel_regularizer: Regularizer function applied to the `kernel` weights matrix. bias_regularizer: Regularizer function applied to the bias vector. activity_regularizer: Regularizer function applied to the output of the layer (its "activation").. kernel_constraint: Constraint function applied to the kernel matrix. bias_constraint: Constraint function applied to the bias vector. implementation: implementation mode, either `1`, `2`, or `3`. `1` loops over input spatial locations to perform the forward pass. It is memory-efficient but performs a lot of (small) ops. `2` stores layer weights in a dense but sparsely-populated 2D matrix and implements the forward pass as a single matrix-multiply. It uses a lot of RAM but performs few (large) ops. `3` stores layer weights in a sparse tensor and implements the forward pass as a single sparse matrix-multiply. How to choose: `1`: large, dense models, `2`: small models, `3`: large, sparse models, where "large" stands for large input/output activations (i.e. many `filters`, `input_filters`, large `input_size`, `output_size`), and "sparse" stands for few connections between inputs and outputs, i.e. small ratio `filters * input_filters * kernel_size / (input_size * strides)`, where inputs to and outputs of the layer are assumed to have shapes `(input_size, input_filters)`, `(output_size, filters)` respectively. It is recommended to benchmark each in the setting of interest to pick the most efficient one (in terms of speed and memory usage). Correct choice of implementation can lead to dramatic speed improvements (e.g. 50X), potentially at the expense of RAM. Also, only `padding="valid"` is supported by `implementation=1`. Input shape: 3D tensor with shape: `(batch_size, steps, input_dim)` Output shape: 3D tensor with shape: `(batch_size, new_steps, filters)` `steps` value might have changed due to padding or strides. validNTglorot_uniformzerosc stjd i|||_t|dd|_tj|dddd|_t||_|jdkr3|dkr3t d|t ||_ t ||_||_t ||_t | |_t | |_t | |_t | |_t | |_t ||_||_tdd |_dS) Nr kernel_sizestridesT) allow_zerorz_Invalid border mode for LocallyConnected1D (only "valid" is supported if implementation is 1): )ndim)super__init__filtersr normalize_tuplerrnormalize_paddingpadding ValueErrornormalize_data_format data_formatrget activationuse_biasrkernel_initializerbias_initializerrkernel_regularizerbias_regularizeractivity_regularizerrkernel_constraintbias_constraintimplementationr input_spec)selfrrrrr r"r#r$r%r&r'r(r)r*r+kwargs __class__re/var/www/myenv/lib/python3.10/site-packages/keras/src/layers/locally_connected/locally_connected1d.pyrts6          zLocallyConnected1D.__init__cCsdS)NFr)r-rrr1!_use_input_spec_as_call_signaturesz4LocallyConnected1D._use_input_spec_as_call_signaturec Cs.|jdkr|d|d}}n |d|d}}|dur!td|t||jd|j|jd|_|jdkrBtd|jd|d|j dkrd|j|jd||j f|_ |j |j |j d |j|jd |_n||j dkr|jdkrx|||j |jf|_ n |||j|j f|_ |j |j |j d |j|jd |_tj|f|j|j|j|jd |_n?|j d kr|j|j ||f|_ ttj|f|j|j|j||j |jd |_|j t|jf|j d |j|jd |_ntd|j |jr|j |j|j f|jd|j|jd |_nd|_|jdkr td d|id|_n td d|id|_d|_dS)Nchannels_firstrz5Axis 2 of input should be fully-defined. Found shape:rzCOne of the dimensions in the output is <= 0 due to downsampling in z;. Consider increasing the input size. Received input shape zO which would produce output shape with a zero or negative value in a dimension.kernel)shape initializername regularizer constraint) input_shape kernel_shaperrr r)r;r<rr filters_in filters_outr %Unrecognized implementation mode: %d.bias)raxesT) r rr conv_output_lengthrrr output_lengthr8r+rr< add_weightr$r&r)r5r get_locallyconnected_mask kernel_masksortedconv_kernel_idxs kernel_idxslenr#r%r'r*r@rr,built)r-r; input_dim input_lengthrrr1builds               zLocallyConnected1D.buildcCsr|jdkr |d}n|d}t||jd|j|jd}|jdkr*|d|j|fS|jdkr7|d||jfSdS)Nr3r4rr channels_last)r r rCrrrr)r-r;rNlengthrrr1compute_output_shape!s    z'LocallyConnected1D.compute_output_shapec Cs|jdkrt||j|j|j|jf|j}n1|jdkr*t ||j|j | |j }n|jdkr@t ||j|j|j| |j }ntd|j|jrTtj||j|jd}||}|S)Nrr4rr?)r )r+r local_convr5rrrDr r local_conv_matmulrGrRr6local_conv_sparse_matmulrJr<rr#bias_addr@r")r-inputsoutputrrr1call1sB        zLocallyConnected1D.callcs|j|j|j|j|jt|j|jt |j t |j t |j t |jt |jt|jt|j|jd}t}tt|t|S)N)rrrrr r"r#r$r%r&r'r(r)r*r+)rrrrr r serializer"r#rr$r%rr&r'r(rr)r*r+r get_configdictlistitems)r-config base_configr/rr1r[Zs0      zLocallyConnected1D.get_config) rrNNTrrNNNNNr)__name__ __module__ __qualname____doc__rpropertyr2r shape_type_conversionrOrRrYr[ __classcell__rrr/r1r !s0U.  z )r N)rd keras.srcrrrrrkeras.src.engine.base_layerrkeras.src.engine.input_specr"keras.src.layers.locally_connectedr keras.src.utilsr r tensorflow.python.util.tf_exportr r rrrr1s