2026-01-20 10:14:31 +08:00
|
|
|
#include "../../../vendors/functions.h"
|
2026-01-19 10:38:50 +08:00
|
|
|
#include "quantization/w8a8/per_token_group_quant_8bit.h"
|
|
|
|
|
|
|
|
|
|
void per_token_group_quant_int8(const torch::Tensor& input,
|
|
|
|
|
torch::Tensor& output_q,
|
|
|
|
|
torch::Tensor& output_s, int64_t group_size,
|
|
|
|
|
double eps, double int8_min, double int8_max) {
|
|
|
|
|
per_token_group_quant_8bit(input, output_q, output_s, group_size, eps,
|
|
|
|
|
int8_min, int8_max);
|
|
|
|
|
}
|