Struct syn_nn_quant_t

ClassList > syn_nn_quant_t

Affine Quantization Scaling Parameters (Scale & Zero-Point). More...

• #include <syn_nn.h>

Public Attributes

Type	Name
uint16_t	multiplier
uint8_t	shift
int8_t	zero_point

Detailed Description

Implements TFLite-compatible affine output scaling: output = clamp_int8( ((acc * multiplier) >> (16 + shift)) + zero_point )

Public Attributes Documentation

variable multiplier

uint16_t syn_nn_quant_t::multiplier;

Fixed-point Q0.16 scale multiplier (1 to 65535)

---

variable shift

uint8_t syn_nn_quant_t::shift;

Right bit-shift factor (0 to 16)

---

variable zero_point

int8_t syn_nn_quant_t::zero_point;

Quantization zero-point offset (-128 to 127)

---

---

The documentation for this class was generated from the following file src/syntropic/util/syn_nn.h