gradientdescent_internal.h

Go to the documentation of this file.

#pragma once

#include "tensor/tensor.h"

namespace magmadnn {
namespace internal {

template <typename T>
magmadnn_error_t gradientdescent_update_internal(Tensor<T> *var, Tensor<T> *grad, T learning_rate);

#if defined(_HAS_CUDA_)
template <typename T>
magmadnn_error_t gradientdescent_update_internal_device(Tensor<T> *var, Tensor<T> *grad, T learning_rate);
#endif

}   // namespace internal
}   // namespace magmadnn