使用限制

放入variant中的对象满足如下条件：

有拷贝构造函数
有默认构造函数
是protocol buffer, 或是tensor或者定义了

string TypeName() const;
void Encode(VariantTensorData* data) const;
bool Decode(VariantTensorData data);

如下是一个示例：tensorflow/learn/test/variant_test.cpp

//#include <any>     //https://en.cppreference.com/w/cpp/utility/any
//#include <variant> //https://en.cppreference.com/w/cpp/utility/variant
//c++17才有 any variant
#include "tensorflow/core/platform/logging.h"
#include "tensorflow/core/platform/test.h"
#include "tensorflow/core/framework/variant.h"
using namespace tensorflow;
struct MyVariantValue {int x;
};bool DecodeVariant(VariantTensorData* data, MyVariantValue* value)
{return true;
}bool DecodeVariant(std::string* buf, MyVariantValue* value)
{return true;
}
void EncodeVariant(const MyVariantValue& value, VariantTensorData* data)
{}void EncodeVariant(const MyVariantValue& value, std::string* buf)
{}
string DebugStringVariant(const MyVariantValue &value)
{return "debug my variant value";
}
string TypeNameVariant(const MyVariantValue &value)
{return "MyVariant";
}
TEST(VariantTest, Variant)
{LOG(INFO) << "test tensorflow core framework variant";MyVariantValue t;t.x = 3;Variant x = t;LOG(INFO) << x.get<MyVariantValue>()->x;EXPECT_EQ(x.get<MyVariantValue>()->x, 3);
}
//bazel test --test_output all -c opt  //tensorflow/learn:test_variant_test

BUILD文件在tensorflow/learn下

load("//tensorflow:tensorflow.bzl", "tf_custom_op_library")
load("//tensorflow:tensorflow.bzl","tf_cc_test","tf_cc_tests","tf_copts","tf_cuda_library",
)
load("//tensorflow/core/platform:build_config.bzl","tf_kernel_tests_linkstatic","tf_proto_library","tf_pyclif_proto_library",
)tf_cc_tests(name = "learn_test",size = "small",srcs = ["test/variant_test.cpp"],linkopts = select({"//tensorflow:macos": ["-headerpad_max_install_names"],"//conditions:default": [],}),linkstatic = tf_kernel_tests_linkstatic(),visibility = ["//tensorflow:internal","//tensorflow/core:__pkg__",],deps = ["//tensorflow/cc:cc_ops","//tensorflow/cc:cc_ops_internal","//tensorflow/cc:function_ops","//tensorflow/cc:ops","//tensorflow/cc:scope","//tensorflow/cc:sendrecv_ops","//tensorflow/cc:while_loop","//tensorflow/compiler/tf2xla/ops:xla_ops","//tensorflow/core","//tensorflow/core:core_cpu","//tensorflow/core:core_cpu_internal","//tensorflow/core:framework","//tensorflow/core:framework_internal","//tensorflow/core:lib","//tensorflow/core:lib_internal","//tensorflow/core:ops","//tensorflow/core:protos_all_cc","//tensorflow/core:test","//tensorflow/core:test_main","//tensorflow/core:testlib","//tensorflow/core/common_runtime:direct_session_internal","//tensorflow/core/kernels:ops_util","//tensorflow/core/lib/monitoring:cell_reader","//tensorflow/core/platform:regexp","//tensorflow/core/platform:status_matchers","//tensorflow/core/profiler/lib:profiler_session","//tensorflow/core/profiler/protobuf:memory_profile_proto_cc","//tensorflow/core/profiler/utils:xplane_schema","//tensorflow/core/profiler/utils:xplane_visitor","//tensorflow/core/util:protos_test_cc","//third_party/eigen3","@com_google_absl//absl/base","@com_google_absl//absl/memory","@com_google_absl//absl/strings",],
)

运行测试

bazel test --test_output all -c opt  //tensorflow/learn:test_variant_test

tensorflow 之variant相关推荐

cnn 预测过程代码_FPN的Tensorflow代码详解——特征提取
@TOC 特征金字塔网络最早于2017年发表于CVPR,与Faster RCNN相比其在多池度特征预测的方式使得其在小目标预测上取得了较好的效果.FPN也作为mmdeteciton的Neck模块, ...
Tensorflow遇到的问题InvalidArgumentError: Graph execution error:2 root error(s) found.解决方法
文章目录遇到问题解决参考遇到问题平台google colab 使用GPU 训练dl模型时,遇到报错 --------------------------------------------- ...
TensorFlow学习笔记（二十六）CNN的9大模型之LeNet5的原理讲解
在机器视觉,图像处理领域,卷积神经网络取得了巨大的成功.本文将参考UFLDL和DEEPLEARNING.NET的教程,结合自己的理解,梳理一下卷积神经网络的构成以及其BP算法的求解.虽然利用thean ...
CNN: TensorFlow 1.14.0 更新
版本contribute发布:https://fossies.org/linux/tensorflow/RELEASE.md https://github.com/tensorflow/tensorf ...
使用TensorFlow 2.0+和Keras实现AlexNet CNN架构
技术 (Technical) 介绍 (Introduction) The main content of this article will present how the AlexNet Convo ...
【TensorFlow实战】TensorFlow实现经典卷积神经网络之ResNet
ResNet ResNet(Residual Neural Network)通过使用Residual Unit成功训练152层深的神经网络,在ILSVRC 2015比赛中获得冠军,取得3.57%的to ...
GraphSAGE NIPS 2017 代码分析（Tensorflow版）
文章目录数据集 ppi数据集信息 toy-ppi-G.json 图的信息 toy-ppi-class_map.json toy-ppi-id_map.json toy-ppi-walks.txt t ...
CNN网络架构学习：Chapter-5-ResNet(附代码tensorflow)
ResNet(Residual Neural Network)由微软研究院的Kaiming He等人提出,通过使用ResNet Unit成功训练出了152层的神经网络,并在ILSVRC2015比赛中取 ...
ubantu + anaconda + TensorFlow 1.13.1
tensorflow2.0尝鲜版本发布了,但是tensorflow1.13.1毕竟是稳定版本,由于网络问题安装.更新一直失败. 系统:ubantu 18 anaconda 3 首先查询版本: (bas ...

tensorflow 之variant

使用限制

如下是一个示例：tensorflow/learn/test/variant_test.cpp

BUILD文件在tensorflow/learn下

运行测试

tensorflow 之variant相关推荐

最新文章

热门文章

tensorflow 之variant

使用限制

如下是一个示例 ：tensorflow/learn/test/variant_test.cpp

BUILD文件在tensorflow/learn下

运行测试

tensorflow 之variant相关推荐

最新文章

热门文章

如下是一个示例：tensorflow/learn/test/variant_test.cpp