Quant.h 文件参考

#include "tensor/Tensor.h"
#include <algorithm>
#include <cmath>
#include <cstdint>
#include <cstring>
#include <vector>

浏览源代码.

类
struct	eve::tensor::q::QuantPayload
	QuantPayload public API. 更多...

命名空间
namespace	eve
	Build metadata (engine git commit, build time, third-party version).
namespace	eve::tensor
namespace	eve::tensor::q

函数
bool	eve::tensor::q::isQuantDType (DType dt)
	True when quant d type.
int	eve::tensor::q::quantByteSize (DType dt, int count)
	Quant byte size.
uint16_t	eve::tensor::q::f32ToF16 (float x)
	F 32 to f 16.
float	eve::tensor::q::f16ToF32 (uint16_t h)
	F 16 to f 32.
float	eve::tensor::q::fp8E4M3ToF32 (uint8_t v)
	Fp 8 e 4 m 3 to f 32.
float	eve::tensor::q::fp4E2M1ToF32 (uint8_t nib)
	Fp 4 e 2 m 1 to f 32.
uint32_t	eve::tensor::q::floatToEfm (float x, int expBits, int manBits, int bias)
	Float to efm.
uint8_t	eve::tensor::q::f32ToFp8E4M3 (float x)
	F 32 to fp 8 e 4 m 3.
uint8_t	eve::tensor::q::f32ToFp4E2M1 (float x)
	F 32 to fp 4 e 2 m 1.
float	eve::tensor::q::efmMaxMagnitude (int expBits, int manBits, int bias)
	Efm max magnitude.
float	eve::tensor::q::dequantValue (DType dt, const uint8_t *bytes, const float *scales, int group, int idx)
	Dequant value.
void	eve::tensor::q::dequantizeAll (DType dt, const uint8_t *bytes, const float *scales, int group, int count, float *out)
	Dequantize all.
QuantPayload	eve::tensor::q::quantize (const float *src, int count, DType dt, int group)
	Quantize.