ONE - On-device Neural Engine
Loading...
Searching...
No Matches
FullyConnected.cpp
Go to the documentation of this file.
1/*
2 * Copyright (c) 2021 Samsung Electronics Co., Ltd. All Rights Reserved
3 *
4 * Licensed under the Apache License, Version 2.0 (the "License");
5 * you may not use this file except in compliance with the License.
6 * You may obtain a copy of the License at
7 *
8 * http://www.apache.org/licenses/LICENSE-2.0
9 *
10 * Unless required by applicable law or agreed to in writing, software
11 * distributed under the License is distributed on an "AS IS" BASIS,
12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13 * See the License for the specific language governing permissions and
14 * limitations under the License.
15 */
16
17#include "Builders.h"
18
20
21namespace luci_interpreter
22{
23
24std::unique_ptr<Kernel> build_kernel_CircleFullyConnected(const luci::CircleNode *circle_node,
25 KernelBuilderHelper &helper)
26{
27 const auto *node = loco::must_cast<const luci::CircleFullyConnected *>(circle_node);
28 assert(node->arity() == 3);
29
30 const Tensor *input = helper.getInputTensor(node->input());
31 const Tensor *weights = helper.getInputTensor(node->weights());
32 const Tensor *bias = helper.getOptionalInputTensor(node->bias());
33 Tensor *output = helper.getOutputTensor(node);
34
35 FullyConnectedParams params{};
36 params.activation = node->fusedActivationFunction();
37 params.keep_num_dims = node->keep_num_dims();
38 if (weights->element_type() == loco::DataType::S4 ||
39 weights->element_type() == loco::DataType::U4)
40 {
41 auto scratchpad =
42 std::make_unique<Tensor>(input->element_type(), weights->shape(), AffineQuantization{}, "");
43 scratchpad->set_observable(false);
44 scratchpad->set_data_buffer(nullptr);
45 Tensor *scratchpad_tmp =
46 helper.getRuntimeGraph(node->graph())->addTensor(std::move(scratchpad));
47 helper.getRuntimeGraph(node->graph())->configureAllocations(scratchpad_tmp);
48 return std::make_unique<kernels::FullyConnected>(input, weights, bias, output, scratchpad_tmp,
49 params);
50 }
51 return std::make_unique<kernels::FullyConnected>(input, weights, bias, output, params);
52}
53
54} // namespace luci_interpreter
Tensor * getOutputTensor(const loco::Node *node) const
const Tensor * getOptionalInputTensor(const loco::Node *node) const
RuntimeGraph * getRuntimeGraph(const loco::Graph *graph) const
const Tensor * getInputTensor(const loco::Node *node) const
Tensor * addTensor(std::unique_ptr< Tensor > &&tensor)
void configureAllocations(Tensor *tensor)
const Shape & shape() const
Definition Tensor.h:107
DataType element_type() const
Definition Tensor.h:105
std::unique_ptr< Kernel > build_kernel_CircleFullyConnected(const luci::CircleNode *circle_node, KernelBuilderHelper &helper)