--- /dev/null
+/*
+ * Copyright (c) 2019 Samsung Electronics Co., Ltd. All Rights Reserved
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "Convert.h"
+#include "GraphBuilder.h"
+#include "GraphBuilderContext.h"
+
+#include <moco/tf/Names.h>
+
+#include <loco.h>
+#include <loco/IR/PermutingCodec.h>
+#include <stdex/Memory.h>
+
+#include <tensorflow/core/framework/graph.pb.h>
+
+#include <cassert>
+#include <stdexcept>
+
+namespace
+{
+using namespace moco::tf;
+
+class ValueInputUpdate final : public GraphUpdate
+{
+public:
+ ValueInputUpdate(loco::BiasAdd<loco::Domain::Tensor> *bias_add, const TensorName &&input_name)
+ : _bias_add(bias_add), _input_name(input_name)
+ {
+ }
+
+ void input(const SymbolTable *) const override;
+
+private:
+ loco::BiasAdd<loco::Domain::Tensor> *_bias_add;
+ const TensorName _input_name;
+};
+
+void ValueInputUpdate::input(const SymbolTable *node_table) const
+{
+ loco::Node *input_node = node_table->node(_input_name);
+ _bias_add->value(input_node);
+}
+
+class BiasInputUpdate final : public GraphUpdate
+{
+public:
+ BiasInputUpdate(loco::BiasEncode *bias_enc, const TensorName &&input_name)
+ : _bias_enc(bias_enc), _input_name(input_name)
+ {
+ }
+
+ void input(const SymbolTable *) const override;
+
+private:
+ loco::BiasEncode *_bias_enc;
+ const TensorName _input_name;
+};
+
+void BiasInputUpdate::input(const SymbolTable *node_table) const
+{
+ loco::Node *input_node = node_table->node(_input_name);
+ _bias_enc->input(input_node);
+}
+
+} // namespace
+
+namespace moco
+{
+namespace tf
+{
+
+/**
+ * @brief GraphBuilder for BiasAdd node
+ */
+class BiasAddGraphBuilder final : public GraphBuilder
+{
+public:
+ bool validate(const tensorflow::NodeDef &) const override;
+ void build(const tensorflow::NodeDef &, GraphBuilderContext *) const override;
+};
+
+bool BiasAddGraphBuilder::validate(const tensorflow::NodeDef &node) const
+{
+ assert(node.input_size() == 2);
+
+ // note: even though "data_format" is not entered when a model is written,
+ // TF seems to generate "data_format" field into a pb file
+ if (!has_attrs(node, {"T", "data_format"}))
+ return false;
+
+ // TODO add type check
+ // type of input and bias should be same (except using quantization)
+
+ // Note In case of TF.nn.bias_add,
+ // "value may have any number of dimensions." ...
+ // but "data_format: A string. 'NHWC' and 'NCHW' are supported."
+ // Not sure if value should be 4-D tensor. Let's skip this check for now.
+
+ return true;
+}
+
+void BiasAddGraphBuilder::build(const tensorflow::NodeDef &node, GraphBuilderContext *context) const
+{
+ assert(context != nullptr);
+
+ loco::Graph *graph = context->graph();
+ SymbolTable *tensor_names = context->tensor_names();
+ UpdateQueue *updates = context->updates();
+
+ // tensorflow data_format: one of NHWC or NCHW.
+ auto data_layout = get_data_layout(node, "data_format");
+
+ // creating loco nodes
+ auto bias_enc = graph->nodes()->create<loco::BiasEncode>();
+
+ auto bias_add = graph->nodes()->create<loco::BiasAdd<loco::Domain::Tensor>>();
+ {
+ if (data_layout == DataLayout::NHWC)
+ {
+ bias_add->axis(3);
+ }
+ else if (data_layout == DataLayout::NCHW)
+ {
+ bias_add->axis(1); // Channel
+ // Note: the following descrition of TF 1.13 at
+ // https://www.tensorflow.org/api_docs/python/tf/nn/bias_add seems wrong:
+ // "bias: A 1-D Tensor with size matching the last dimension of value."
+ // because providing the size of W (last dimension) to bias throws an error with Tensorflow
+ }
+ }
+
+ // link nodes
+ bias_add->bias(bias_enc);
+
+ // To set the input node of encode_node with biasAdd_name
+ TensorName output_name(node.name(), 0);
+ tensor_names->enroll(output_name, bias_add);
+
+ // Record ifm inputs to featureEncode_node
+ auto value_update = stdex::make_unique<ValueInputUpdate>(bias_add, TensorName(node.input(0)));
+ auto bias_update = stdex::make_unique<BiasInputUpdate>(bias_enc, TensorName(node.input(1)));
+
+ updates->enroll(std::move(value_update));
+ updates->enroll(std::move(bias_update));
+}
+
+} // namespace tf
+} // namespace moco
+
+#include "GraphBuilderRegistry.h"
+
+REGISTER_OP_BUILDER(BiasAdd, BiasAddGraphBuilder)
--- /dev/null
+/*
+ * Copyright (c) 2019 Samsung Electronics Co., Ltd. All Rights Reserved
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "TestHelper.h"
+
+#include <moco/tf/Frontend.h>
+
+#include <loco.h>
+
+#include <gtest/gtest.h>
+
+#include <cstring>
+
+using namespace moco::tf::test;
+
+namespace
+{
+
+// clang-format off
+const char *bias_add_01_pbtxtdata = STRING_CONTENT(
+
+node {
+ name: "val"
+ op: "Const"
+ attr {
+ key: "dtype"
+ value { type: DT_FLOAT }
+ }
+ attr {
+ key: "value"
+ value {
+ tensor {
+ dtype: DT_FLOAT
+ tensor_shape {
+ dim { size: 1 }
+ dim { size: 5 }
+ dim { size: 5 }
+ dim { size: 3 }
+ }
+ float_val: 2.1
+ }
+ }
+ }
+}
+node {
+ name: "bias"
+ op: "Const"
+ attr {
+ key: "dtype"
+ value { type: DT_FLOAT }
+ }
+ attr {
+ key: "value"
+ value {
+ tensor {
+ dtype: DT_FLOAT
+ tensor_shape {
+ dim { size: 3 }
+ }
+ float_val: 1.1
+ }
+ }
+ }
+}
+node {
+ name: "out"
+ op: "BiasAdd"
+ input: "val"
+ input: "bias"
+ attr {
+ key: "T"
+ value { type: DT_FLOAT }
+ }
+ attr {
+ key: "data_format"
+ value { s: "NHWC" }
+ }
+}
+
+);
+// clang-format on
+
+} // namespace
+
+TEST(TensorFlowFrontend, bias_add_01)
+{
+ moco::tf::Frontend frontend;
+ moco::tf::ModelSignature signature;
+ signature.add_output(moco::tf::TensorName("out", 0));
+
+ imemstream mempb(bias_add_01_pbtxtdata, std::strlen(bias_add_01_pbtxtdata));
+ std::unique_ptr<loco::Graph> graph =
+ frontend.load(signature, &mempb, moco::tf::Frontend::FileType::Text);
+
+ // test 1.
+ // loco node : 1. ConstGen ------------------+-- 4. BiasAdd<Domain::Tensor> -- 5. Push
+ // 2. ConstGen - 3. BiasEncode -/
+ loco::Graph::NodeContext *loco_nodes = graph->nodes();
+
+ loco::Graph::InputContext *loco_inputs = graph->inputs();
+ ASSERT_EQ(loco_inputs->size(), 0);
+ ASSERT_EQ(loco_nodes->size(), 5);
+
+ int idx = 0;
+
+ loco::ConstGen *value = dynamic_cast<loco::ConstGen *>(loco_nodes->at(idx++));
+ loco::ConstGen *bias = dynamic_cast<loco::ConstGen *>(loco_nodes->at(idx++));
+ loco::BiasEncode *enc = dynamic_cast<loco::BiasEncode *>(loco_nodes->at(idx++));
+ loco::BiasAdd<loco::Domain::Tensor> *bias_add =
+ dynamic_cast<loco::BiasAdd<loco::Domain::Tensor> *>(loco_nodes->at(idx++));
+ loco::Push *push = dynamic_cast<loco::Push *>(loco_nodes->at(idx++));
+
+ ASSERT_NE(value, nullptr);
+ ASSERT_NE(bias, nullptr);
+ ASSERT_NE(enc, nullptr);
+ ASSERT_NE(bias_add, nullptr);
+ ASSERT_NE(push, nullptr);
+
+ // check their connection is all OK
+ ASSERT_TRUE(enc->input() == bias);
+ ASSERT_TRUE(bias_add->bias() == enc);
+ ASSERT_TRUE(bias_add->value() == value);
+ ASSERT_TRUE(push->from() == bias_add);
+
+ // test 2.
+ // attrs inside BiasAdd
+
+ // axis
+ ASSERT_EQ(bias_add->axis(), 3); // NHWC
+}
+
+namespace
+{
+
+// clang-format off
+const char *bias_add_NCHW_pbtxtdata = STRING_CONTENT(
+
+node {
+ name: "val"
+ op: "Const"
+ attr {
+ key: "dtype"
+ value { type: DT_FLOAT }
+ }
+ attr {
+ key: "value"
+ value {
+ tensor {
+ dtype: DT_FLOAT
+ tensor_shape {
+ dim { size: 1 }
+ dim { size: 3 }
+ dim { size: 299 }
+ dim { size: 299 }
+ }
+ float_val: 2.1
+ }
+ }
+ }
+}
+node {
+ name: "bias"
+ op: "Const"
+ attr {
+ key: "dtype"
+ value { type: DT_FLOAT }
+ }
+ attr {
+ key: "value"
+ value {
+ tensor {
+ dtype: DT_FLOAT
+ tensor_shape {
+ dim { size: 3 }
+ }
+ float_val: 1.1
+ }
+ }
+ }
+}
+node {
+ name: "out"
+ op: "BiasAdd"
+ input: "val"
+ input: "bias"
+ attr {
+ key: "T"
+ value { type: DT_FLOAT }
+ }
+ attr {
+ key: "data_format"
+ value { s: "NCHW" }
+ }
+}
+
+);
+// clang-format on
+
+} // namespace
+
+TEST(TensorFlowFrontend, bias_add_NCHW_axis)
+{
+ moco::tf::Frontend frontend;
+ moco::tf::ModelSignature signature;
+ signature.add_output(moco::tf::TensorName("out", 0));
+
+ imemstream mempb(bias_add_NCHW_pbtxtdata, std::strlen(bias_add_NCHW_pbtxtdata));
+ std::unique_ptr<loco::Graph> graph =
+ frontend.load(signature, &mempb, moco::tf::Frontend::FileType::Text);
+
+ // testing axis value of biasAdd
+ loco::Graph::NodeContext *loco_nodes = graph->nodes();
+ loco::BiasAdd<loco::Domain::Tensor> *bias_add =
+ dynamic_cast<loco::BiasAdd<loco::Domain::Tensor> *>(loco_nodes->at(3));
+ ASSERT_NE(bias_add, nullptr);
+
+ ASSERT_EQ(bias_add->axis(), 1); // NCHW
+}
return true;
}
+bool fix_padding(loco::BiasAdd<loco::Domain::Tensor> *node)
+{
+ // Nothing to do with padding
+ return false;
+}
+
+bool fix_padding(loco::BiasEncode *node)
+{
+ // Nothing to do with padding
+ return false;
+}
+
bool fix_padding(loco::ConstGen *node)
{
// Nothing to do with padding
return true;
}
+bool fix_shape(loco::BiasAdd<loco::Domain::Tensor> *node)
+{
+ // Output shape is same as the input
+ auto input = node->value(); // BiasAdd accepts n dimension tensor
+ return copy_shapedata(input, node); // after adding Bias, output shape is same with input
+}
+
+bool fix_shape(loco::BiasEncode *node)
+{
+ auto input = node->input();
+ return copy_shapedata(input, node);
+}
+
bool fix_shape(loco::ConstGen *node)
{
auto shapedata = node->annot<ShapeInferenceData>();
// MOCONODE(Name) : alphabetic order please
MOCONODE(AvgPool2D)
+MOCONODE(BiasAdd<loco::Domain::Tensor>)
+MOCONODE(BiasEncode)
MOCONODE(ConstGen)
MOCONODE(Conv2D)
MOCONODE(FeatureDecode)
projects / platform / core / ml / nnfw.git / commitdiff