24 #ifndef ARM_COMPUTE_NEFULLYCONNECTEDLAYER_H 25 #define ARM_COMPUTE_NEFULLYCONNECTEDLAYER_H 74 namespace weights_transformations
82 _output.allocator()->allocate();
89 _output.allocator()->free();
97 uint32_t
uid()
override 104 _func.configure(input, &_output);
108 static constexpr uint32_t _uid = 0x0;
191 Tensor _converted_weights_output;
192 Tensor _reshape_weights_output;
193 const ITensor *_original_weights;
194 bool _are_weights_converted;
195 bool _are_weights_reshaped;
196 bool _is_fc_after_conv;
197 bool _is_quantized_asymmetric;
~NEFullyConnectedLayerReshapeWeights()=default
Default destructor.
Base class for all functions.
void run() override final
Run the kernels contained in the function.
Basic function to reshape the weights of Fully Connected layer with Neon.
Basic function to execute GEMM on Neon.
NEFullyConnectedLayerReshapeWeights & operator=(const NEFullyConnectedLayerReshapeWeights &)=delete
Prevent instances of this class from being copied (As this class contains pointers) ...
Fully connected layer info.
Store the tensor's metadata.
Activation Layer Information class.
Interface for Neon tensor.
Copyright (c) 2017-2021 Arm Limited.
static Status validate(const ITensorInfo *input, const ITensorInfo *output)
Static function to check if given info will lead to a valid configuration of NEFullyConnectedLayerRes...
Basic function to run NEConvertFullyConnectedWeightsKernel.
Basic interface for functions which have a single Neon kernel and no border.
virtual void prepare()
Prepare the function for executing.
NEFullyConnectedLayerReshapeWeights()=default
Constructor.
Basic implementation of the tensor interface.
Weights manager interface to handle weights transformations.
Basic function to compute a Fully Connected layer on Neon.
Basic function to execute flatten layer kernel.
void configure(const ITensor *input, ITensor *output)
Set the input and output tensors.
Basic function to execute GEMMLowpMatrixMultiplyCore on Neon.