提交 47aaac00 编写于 作者: X xutianbing

add some comments...

上级 a948eea3
...@@ -17,6 +17,15 @@ limitations under the License. */ ...@@ -17,6 +17,15 @@ limitations under the License. */
#include "paddle/math/Vector.h" #include "paddle/math/Vector.h"
namespace paddle { namespace paddle {
/**
* Cosine Similarity for CpuMatrix
*
* \param out_mat, output value, size: nSamples * 1.
* \param in1_mat, input value 1, size: nSamples * dim.
* \param in2_mat, input value 2, size: n2 * dim (n2 == 1 or n2 == nSamples).
* \param scale, default 1.0
*
*/
template <> template <>
void CosSimForward<DEVICE_TYPE_CPU>(CpuMatrix* out_mat, void CosSimForward<DEVICE_TYPE_CPU>(CpuMatrix* out_mat,
const CpuMatrix* in1_mat, const CpuMatrix* in1_mat,
...@@ -48,6 +57,13 @@ void CosSimForward<DEVICE_TYPE_CPU>(CpuMatrix* out_mat, ...@@ -48,6 +57,13 @@ void CosSimForward<DEVICE_TYPE_CPU>(CpuMatrix* out_mat,
} }
/** /**
* Cosine Similarity
* for each row i,
* out[i] = scale * cos(input1[i], input2[i])
* = scale * <input1[i], input2[i]>/sqrt(|input1[i]|^2 * |input2[i]|^2)
* when input2 only has one row, then for each row i,
* out[i] = cos(input1[i], input2[0])
*
* \param inputs[0] input matrix 1, size: nSamples * dim. * \param inputs[0] input matrix 1, size: nSamples * dim.
* \param inputs[1] input matrix 2, size: n2 * dim (n2 == 1 or n2 == nSamples). * \param inputs[1] input matrix 2, size: n2 * dim (n2 == 1 or n2 == nSamples).
* \param outputs[0] output matrix, size : nSamples * 1. * \param outputs[0] output matrix, size : nSamples * 1.
...@@ -85,6 +101,20 @@ private: ...@@ -85,6 +101,20 @@ private:
real scale_; real scale_;
}; };
/**
* Cosine Similarity Derivative for CpuMatrix
*
* \param in1_grad forward input grad 1, size: nSamples * dim.
* \param in2_grad forward input grad 2,
* size: n2 * dim (n2 == 1 or n2 == nSamples).
*
* \param out_grad backward loss output grad, size : nSamples * 1.
* \param out_val forward output value, size: nSamples * 1.
* \param in1_val forward input value 1, size: nSamples * dim.
* \param in2_val forward input value 2,
* size: n2 * dim (n2 == 1 or n2 == nSamples).
* \param scale, default 1.0
*/
template <> template <>
void CosSimBackward<DEVICE_TYPE_CPU>(const CpuMatrix* out_grad, void CosSimBackward<DEVICE_TYPE_CPU>(const CpuMatrix* out_grad,
const CpuMatrix* out_val, const CpuMatrix* out_val,
...@@ -146,6 +176,8 @@ void CosSimBackward<DEVICE_TYPE_CPU>(const CpuMatrix* out_grad, ...@@ -146,6 +176,8 @@ void CosSimBackward<DEVICE_TYPE_CPU>(const CpuMatrix* out_grad,
} }
/** /**
* Cosine Similarity backward Derivative
*
* \param inouts[0] forward input grad 1, size: nSamples * dim. * \param inouts[0] forward input grad 1, size: nSamples * dim.
* \param inouts[1] forward input grad 2, * \param inouts[1] forward input grad 2,
* size: n2 * dim (n2 == 1 or n2 == nSamples). * size: n2 * dim (n2 == 1 or n2 == nSamples).
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册