本文整理汇总了C++中tensor_type类的典型用法代码示例。如果您正苦于以下问题：C++ tensor_type类的具体用法？C++ tensor_type怎么用？C++ tensor_type使用的例子？那么恭喜您, 这里精选的类代码示例或许可以为您提供帮助。
在下文中一共展示了tensor_type类的15个代码示例，这些例子默认根据受欢迎程度排序。您可以为喜欢或者感觉有用的代码点赞，您的评价将有助于系统推荐出更棒的C++代码示例。
示例1: backward
void Min::backward(const tensor_type& data_input, const tensor_type& gradient_output)
{
gradient_input.resizeLike(data_input);
gradient_input.setZero();
if (dimension) {
for (int row = 0; row != data_input.rows(); ++ row)
gradient_input.row(row)[indices[row]] = gradient_output.col(0)[row];
} else {
for (int col = 0; col != data_input.cols(); ++ col)
gradient_input.col(col)[indices[col]] = gradient_output.col(0)[col];
}
}示例2: apply
static void apply( const tensor_type & tensor ,
const MatrixValue * const a ,
const VectorValue * const x ,
VectorValue * const y )
{
const size_type nDim = tensor.dimension();
// Loop over i
for ( size_type i = 0; i < nDim; ++i) {
VectorValue ytmp = 0;
// Loop over k for this i
const size_type nk = tensor.num_k(i);
const size_type kBeg = tensor.k_begin(i);
const size_type kEnd = kBeg + nk;
for (size_type kEntry = kBeg; kEntry < kEnd; ++kEntry) {
const size_type k = tensor.k_coord(kEntry);
const MatrixValue ak = a[k];
const VectorValue xk = x[k];
// Loop over j for this i,k
const size_type nj = tensor.num_j(kEntry);
const size_type jBeg = tensor.j_begin(kEntry);
const size_type jEnd = jBeg + nj;
for (size_type jEntry = jBeg; jEntry < jEnd; ++jEntry) {
const size_type j = tensor.j_coord(jEntry);
ytmp += tensor.value(jEntry) * ( a[j] * xk + ak * x[j] );
}
}
y[i] += ytmp ;
}
}示例3: apply
KOKKOS_INLINE_FUNCTION
static void apply( const tensor_type & tensor ,
const MatrixValue * const a ,
const VectorValue * const x ,
VectorValue * const y )
{
const size_type nk = tensor.num_k();
// Loop over k
for ( size_type k = 0; k < nk; ++k) {
const MatrixValue ak = a[k];
const VectorValue xk = x[k];
// Loop over j for this k
const size_type nj = tensor.num_j(k);
const size_type jBeg = tensor.j_begin(k);
const size_type jEnd = jBeg + nj;
for (size_type jEntry = jBeg; jEntry < jEnd; ++jEntry) {
const size_type j = tensor.j_coord(jEntry);
VectorValue tmp = a[j] * xk + ak * x[j];
// Loop over i for this k,j
const size_type ni = tensor.num_i(jEntry);
const size_type iBeg = tensor.i_begin(jEntry);
const size_type iEnd = iBeg + ni;
for (size_type iEntry = iBeg; iEntry < iEnd; ++iEntry) {
const size_type i = tensor.i_coord(iEntry);
y[i] += tensor.value(iEntry) * tmp;
}
}
}
}示例4: fill_from_file
void fill_from_file(tensor_type& tens, const std::string& str)
{
std::ifstream ifs(str);
for(int k = 0 ; k < tens.extent<2>(); ++k)
for(int j = 0 ; j < tens.extent<1>(); ++j)
for(int i = 0 ; i < tens.extent<0>(); ++i)
{
ifs >> tens.at(i,j,k);
}
}示例5: operator
void operator ()(const tensor_type &delta, const tensor_type &x, tensor_type &x_gradient) {
CHECK_EQ(x.order(), delta.order());
// TODO(robertsdionne): Support arbitrary tensor order with an n-dimensional iterator.
for (auto i = 0; i < x.shape().at(0); ++i) {
for (auto j = 0; j < x.shape().at(1); ++j) {
x_gradient.set({i, j}, delta.at({i, j}) * (x.at({i, j}) > F(0)));
}
}
}示例6: forward
void SoftMax::forward(const tensor_type& data_input)
{
// use of redux for computing logsum...
const double infty = - std::numeric_limits<double>::infinity();
double logsum = infty;
for (difference_type i = 0; i != data_input.rows(); ++ i) {
const double value = data_input.col(0)[i];
if (logsum == infty)
logsum = value;
else if (value > infty) {
if (logsum >= value)
logsum = logsum + utils::mathop::log1p(std::exp(value - logsum));
else
logsum = value + utils::mathop::log1p(std::exp(logsum - value));
}
}
data_output = (data_input.array() - logsum).exp();
}示例7: print
void print(const tensor_type& tens, const std::string& str)
{
std::ofstream of(str);
of << std::showpos;
using limit_type = std::numeric_limits<typename tensor_type::value_type>;
of << std::scientific;
of << std::setprecision(limit_type::max_digits10);
for(int k = 0 ; k < tens.extent<2>(); ++k)
for(int j = 0 ; j < tens.extent<1>(); ++j)
for(int i = 0 ; i < tens.extent<0>(); ++i)
{
of << tens.at(i,j,k) << std::endl;
}
}示例8: operator
void operator ()(const tensor_type &x, tensor_type &y) {
using std::max;
CHECK_EQ(x.order(), y.order());
// TODO(robertsdionne): Support arbitrary tensor order with an n-dimensional iterator.
for (auto i = 0; i < x.shape().at(0); ++i) {
for (auto j = 0; j < x.shape().at(1); ++j) {
y.set({i, j}, max(F(0), x.at({i, j})));
}
}
}示例9: forward
void Min::forward(const tensor_type& data_input)
{
if (dimension) {
// select row and compute column-min
data_output.resize(data_input.rows(), 1);
indices.resize(data_input.rows());
for (size_type row = 0; row != data_input.rows(); ++ row) {
int col_min = 0;
data_output.col(0)[row] = data_input.row(row).minCoeff(&col_min);
indices[row] = col_min;
}
} else {
// select column and compute row-min!
data_output.resize(data_input.cols(), 1);
indices.resize(data_input.cols());
for (size_type col = 0; col != data_input.cols(); ++ col) {
int row_min = 0;
data_output.col(0)[col] = data_input.col(col).minCoeff(&row_min);
indices[col] = row_min;
}
}
}示例10: matrix_size
static size_type matrix_size( const tensor_type & tensor )
{
return tensor.dimension();
}示例11: vector_size
KOKKOS_INLINE_FUNCTION
static size_type vector_size( const tensor_type & tensor )
{ return tensor.dimension(); }示例12: forward
void Exp::forward(const tensor_type& data_input)
{
data_output = data_input.array().exp();
}示例13: backward
void Exp::backward(const tensor_type& data_input, const tensor_type& gradient_output)
{
gradient_input.array() = gradient_output.array() * data_output.array();
}示例14: backward
void Log::backward(const tensor_type& data_input, const tensor_type& gradient_output)
{
gradient_input = data_input.array() / gradient_output.array();
}示例15: vector_size
static size_type vector_size( const tensor_type & tensor )
{
return tensor.dimension();
}