Conv Struct Reference

File-backed convolution parameter bundle. More...

#include <noodle.h>

Public Attributes
uint16_t	K = 3
	Kernel width, or tap count for 1D convolution.
uint16_t	P = 0
	Padding per side; 65535 requests SAME-style 2D padding.
uint16_t	S = 1
	Convolution stride.
uint16_t	OP = 0
	User-computed output padding for transpose convolution.
const char *	weight_fn = nullptr
	Weight filename.
const char *	bias_fn = nullptr
	Bias filename.
Activation	act = ACT_RELU
	Activation applied after adding bias.

Detailed Description

File-backed convolution parameter bundle.

Weight and bias files are read sequentially. For 2D convolution, weights use [O][I][K][K]; for 1D convolution, [O][I][K]; for depthwise convolution, [C][K][K]. Bias files contain one scalar per output channel, or one scalar per channel for depthwise convolution.

For transpose convolution with explicit padding, callers choose OP to match the desired output width: V = (W - 1) * S - 2 * P + K + OP.

Member Data Documentation

act

Activation Conv::act = ACT_RELU

Activation applied after adding bias.

bias_fn

const char* Conv::bias_fn = nullptr

Bias filename.

K

uint16_t Conv::K = 3

Kernel width, or tap count for 1D convolution.

OP

uint16_t Conv::OP = 0

User-computed output padding for transpose convolution.

P

uint16_t Conv::P = 0

Padding per side; 65535 requests SAME-style 2D padding.

S

uint16_t Conv::S = 1

Convolution stride.

weight_fn

const char* Conv::weight_fn = nullptr

Weight filename.

---

The documentation for this struct was generated from the following file:

• noodle.h