Added gather_elements & unit test (#363)

Signed-off-by: Chen Xin <jack.chen@verisilicon.com>

Co-authored-by: Chen Xin <jack.chen@verisilicon.com>
This commit is contained in:
chxin66 2022-05-10 09:55:50 +08:00 committed by GitHub
parent 60cfea53a0
commit 0d8ac3dc2b
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
5 changed files with 297 additions and 0 deletions

View File

@ -0,0 +1,57 @@
/****************************************************************************
*
* Copyright (c) 2022 Vivante Corporation
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the "Software"),
* to deal in the Software without restriction, including without limitation
* the rights to use, copy, modify, merge, publish, distribute, sublicense,
* and/or sell copies of the Software, and to permit persons to whom the
* Software is furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in
* all copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
* FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
* DEALINGS IN THE SOFTWARE.
*
*****************************************************************************/
#ifndef TIM_VX_OPS_GATHER_H_
#define TIM_VX_OPS_GATHER_H_
#include "tim/vx/direct_map_op.h"
namespace tim {
namespace vx {
namespace ops {
/**
* ## Gather_elements
*
* Gather_elements slices from input, **axis** according to **indices**.
* out[i][j][k] = input[index[i][j][k]][j][k] if axis = 0,
* out[i][j][k] = input[i][index[i][j][k]][k] if axis = 1,
* out[i][j][k] = input[i][j][index[i][j][k]] if axis = 2,
* https://github.com/onnx/onnx/blob/main/docs/Operators.md#GatherElements
*/
class Gather_elements : public DirectMapOp {
public:
Gather_elements(Graph* Graph, int axis);
std::shared_ptr<Operation> Clone(
std::shared_ptr<Graph>& graph) const override;
protected:
int axis_;
};
} // namespace ops
} // namespace vx
} // namespace tim
#endif /* TIM_VX_OPS_GATHER_H_ */

View File

@ -20,6 +20,7 @@ Relu|RELU|Mapped|[tf.nn.relu](https://tensorflow.google.cn/api_docs/python/tf/nn
Reorg|REORG|Mapped|[darknet.reorg](https://github.com/pjreddie/darknet/blob/master/src/reorg_layer.c)
L2Normalization|L2_NORMALIZE|Mapped|[tf.math.l2_normalize](https://tensorflow.google.cn/api_docs/python/tf/math/l2_normalize)
FullyConnected|FCL2|Mapped|[tf.keras.layers.Dense](https://tensorflow.google.cn/api_docs/python/tf/keras/layers/Dense)
Dense|FCL|Mapped|[tf.keras.layers.Dense](https://tensorflow.google.cn/api_docs/python/tf/keras/layers/Dense)
MaxpoolWithArgmax|POOLWITHARGMAX|Mapped|[tf.nn.max_pool_with_argmax](https://tensorflow.google.cn/api_docs/python/tf/nn/max_pool_with_argmax)
ArgMax|ARGMAX|Mapped|[tf.math.argmax](https://tensorflow.google.cn/api_docs/python/tf/math/argmax)
Maximum|MAXIMUM|Mapped|[tf.math.maximum](https://tensorflow.google.cn/api_docs/python/tf/math/maximum)
@ -76,6 +77,7 @@ Exp|EXP|Mapped|[tf.math.exp](https://tensorflow.google.cn/api_docs/python/tf/mat
Clip|CLIP|Mapped|[tf.clip_by_value](https://tensorflow.google.cn/api_docs/python/tf/clip_by_value)
AddN|ADDN|Mapped|[tf.math.add_n](https://tensorflow.google.cn/api_docs/python/tf/math/add_n)
Gather|GATHER|Mapped|[tf.gather](https://tensorflow.google.cn/api_docs/python/tf/gather)
Gather_elements|GATHER_ELEMENTS|Mapped|[onnx/GatherElements](https://github.com/onnx/onnx/blob/main/docs/Operators.md#gatherelements)
LogicalNot|LOGICAL_NOT|Mapped|[tf.math.logical_not](https://tensorflow.google.cn/api_docs/python/tf/math/logical_not)
Sin|SIN|Mapped|[tf.math.sin](https://tensorflow.google.cn/api_docs/python/tf/math/sin)
Log|LOG|Mapped|[tf.math.log](https://tensorflow.google.cn/api_docs/python/tf/math/log)

View File

@ -0,0 +1,45 @@
/****************************************************************************
*
* Copyright (c) 2022 Vivante Corporation
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the "Software"),
* to deal in the Software without restriction, including without limitation
* the rights to use, copy, modify, merge, publish, distribute, sublicense,
* and/or sell copies of the Software, and to permit persons to whom the
* Software is furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in
* all copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
* FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
* DEALINGS IN THE SOFTWARE.
*
*****************************************************************************/
#include "tim/vx/ops/gather_elements.h"
#include "direct_map_op_impl.h"
#include "vsi_nn_pub.h"
namespace tim {
namespace vx {
namespace ops {
Gather_elements::Gather_elements(Graph* graph, int axis)
: DirectMapOp(graph, VSI_NN_OP_GATHER_ELEMENTS), axis_(axis) {
this->impl()->node()->nn_param.gather_elements.axis = axis_;
}
std::shared_ptr<Operation> Gather_elements::Clone(
std::shared_ptr<Graph>& graph) const {
return graph->CreateOperation<Gather_elements>(this->axis_);
}
} // namespace ops
} // namespace vx
} // namespace tim

View File

@ -0,0 +1,170 @@
/****************************************************************************
*
* Copyright (c) 2022 Vivante Corporation
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the "Software"),
* to deal in the Software without restriction, including without limitation
* the rights to use, copy, modify, merge, publish, distribute, sublicense,
* and/or sell copies of the Software, and to permit persons to whom the
* Software is furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in
* all copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
* FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
* DEALINGS IN THE SOFTWARE.
*
*****************************************************************************/
#include "tim/vx/context.h"
#include "tim/vx/graph.h"
#include "tim/vx/ops/gather_elements.h"
#include <iostream>
#include "gtest/gtest.h"
#include "test_utils.h"
TEST(Gather_elements, shape_3_2_1_int32_axis_0) {
auto ctx = tim::vx::Context::Create();
auto graph = ctx->CreateGraph();
tim::vx::ShapeType in_shape({3, 2, 1});
tim::vx::ShapeType indices_shape({2, 2, 1});
tim::vx::ShapeType out_shape({2, 2, 1});
tim::vx::TensorSpec input_spec(tim::vx::DataType::INT32, in_shape,
tim::vx::TensorAttribute::INPUT);
tim::vx::TensorSpec indices_spec(tim::vx::DataType::INT32, indices_shape,
tim::vx::TensorAttribute::INPUT);
tim::vx::TensorSpec output_spec(tim::vx::DataType::INT32, out_shape,
tim::vx::TensorAttribute::OUTPUT);
auto input_tensor = graph->CreateTensor(input_spec);
auto indices_tensor = graph->CreateTensor(indices_spec);
auto output_tensor = graph->CreateTensor(output_spec);
std::vector<int32_t> in_data = {
1, 2, 3, 4, 5, 6,
};
//The index value greater than rank-1 is regarded as rank-1
std::vector<int32_t> indices = {
1,
2,
0,
2,
};
std::vector<int32_t> golden = {
2, 3, 4, 6,
};
EXPECT_TRUE(
input_tensor->CopyDataToTensor(in_data.data(), in_data.size() * 4));
EXPECT_TRUE(
indices_tensor->CopyDataToTensor(indices.data(), indices.size() * 4));
auto op = graph->CreateOperation<tim::vx::ops::Gather_elements>(0);
(*op).BindInputs({input_tensor, indices_tensor}).BindOutputs({output_tensor});
EXPECT_TRUE(graph->Compile());
EXPECT_TRUE(graph->Run());
std::vector<int32_t> output(golden.size());
EXPECT_TRUE(output_tensor->CopyDataFromTensor(output.data()));
EXPECT_EQ(golden, output);
}
TEST(Gather_elements, shape_3_2_1_int32_axis_1) {
auto ctx = tim::vx::Context::Create();
auto graph = ctx->CreateGraph();
tim::vx::ShapeType in_shape({3, 2, 1});
tim::vx::ShapeType indices_shape({2, 2, 1});
tim::vx::ShapeType out_shape({2, 2, 1});
tim::vx::TensorSpec input_spec(tim::vx::DataType::INT32, in_shape,
tim::vx::TensorAttribute::INPUT);
tim::vx::TensorSpec indices_spec(tim::vx::DataType::INT32, indices_shape,
tim::vx::TensorAttribute::INPUT);
tim::vx::TensorSpec output_spec(tim::vx::DataType::INT32, out_shape,
tim::vx::TensorAttribute::OUTPUT);
auto input_tensor = graph->CreateTensor(input_spec);
auto indices_tensor = graph->CreateTensor(indices_spec);
auto output_tensor = graph->CreateTensor(output_spec);
std::vector<int32_t> in_data = {
1, 2, 3, 4, 5, 6,
};
//The index value greater than rank-1 is regarded as rank-1
std::vector<int32_t> indices = {
1,
2,
0,
2,
};
std::vector<int32_t> golden = {
4, 5, 1, 5,
};
EXPECT_TRUE(
input_tensor->CopyDataToTensor(in_data.data(), in_data.size() * 4));
EXPECT_TRUE(
indices_tensor->CopyDataToTensor(indices.data(), indices.size() * 4));
auto op = graph->CreateOperation<tim::vx::ops::Gather_elements>(1);
(*op).BindInputs({input_tensor, indices_tensor}).BindOutputs({output_tensor});
EXPECT_TRUE(graph->Compile());
EXPECT_TRUE(graph->Run());
std::vector<int32_t> output(golden.size());
EXPECT_TRUE(output_tensor->CopyDataFromTensor(output.data()));
EXPECT_EQ(golden, output);
}
TEST(Gather_elements, shape_3_2_1_float32_axis_2) {
auto ctx = tim::vx::Context::Create();
auto graph = ctx->CreateGraph();
tim::vx::ShapeType in_shape({3, 2, 1});
tim::vx::ShapeType indices_shape({2, 2, 1});
tim::vx::ShapeType out_shape({2, 2, 1});
tim::vx::TensorSpec input_spec(tim::vx::DataType::INT32, in_shape,
tim::vx::TensorAttribute::INPUT);
tim::vx::TensorSpec indices_spec(tim::vx::DataType::INT32, indices_shape,
tim::vx::TensorAttribute::INPUT);
tim::vx::TensorSpec output_spec(tim::vx::DataType::INT32, out_shape,
tim::vx::TensorAttribute::OUTPUT);
auto input_tensor = graph->CreateTensor(input_spec);
auto indices_tensor = graph->CreateTensor(indices_spec);
auto output_tensor = graph->CreateTensor(output_spec);
std::vector<float> in_data = {
1, 2, 3, 4, 5, 6,
};
//The index value greater than rank-1 is regarded as rank-1
std::vector<int32_t> indices = {
1,
2,
0,
2,
};
std::vector<float> golden = {
1, 2, 3, 4,
};
EXPECT_TRUE(
input_tensor->CopyDataToTensor(in_data.data(), in_data.size() * 4));
EXPECT_TRUE(
indices_tensor->CopyDataToTensor(indices.data(), indices.size() * 4));
auto op = graph->CreateOperation<tim::vx::ops::Gather_elements>(2);
(*op).BindInputs({input_tensor, indices_tensor}).BindOutputs({output_tensor});
EXPECT_TRUE(graph->Compile());
EXPECT_TRUE(graph->Run());
std::vector<float> output(golden.size());
EXPECT_TRUE(output_tensor->CopyDataFromTensor(output.data()));
EXPECT_EQ(golden, output);
}

View File

@ -1,3 +1,26 @@
/****************************************************************************
*
* Copyright (c) 2022 Vivante Corporation
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the "Software"),
* to deal in the Software without restriction, including without limitation
* the rights to use, copy, modify, merge, publish, distribute, sublicense,
* and/or sell copies of the Software, and to permit persons to whom the
* Software is furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in
* all copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
* FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
* DEALINGS IN THE SOFTWARE.
*
*****************************************************************************/
#include "tim/vx/context.h"
#include "tim/vx/graph.h"
#include "tim/vx/ops/pool2d.h"