vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

No second least squares pass

Browse source
This commit is contained in:
Henrik Forstén 2023-12-30 18:31:46 +02:00
parent 8386034e08
commit 5a02328d1f
1 changed files with 14 additions and 25 deletions
Download patch file Download diff file Expand all files Collapse all files

39
ggml-quants.c
View file

@ -5,6 +5,7 @@
#include <string.h> #include <string.h>
#include <assert.h> #include <assert.h>
#include <float.h> #include <float.h>
#include <stdio.h>
#ifdef __ARM_NEON #ifdef __ARM_NEON
@ -457,20 +458,6 @@ static void lstsq_q_1(const uint8_t * restrict q, const float * restrict x, int
} }
} }
static float lstsq_q_0(const float * restrict q, const float * restrict x, int qk) {
// Least squares fits `d * q = x` for d.
float qs2 = 0.0f;
float xq = 0.0f;
for (int i = 0; i < qk; i++) {
qs2 += q[i]*q[i];
xq += x[i]*q[i];
}
if (qs2 == 0.0f) {
return 0.0f;
}
return xq / qs2;
}
static float lstsq_q_0_u8(const uint8_t * restrict q, const float * restrict x, int qk) { static float lstsq_q_0_u8(const uint8_t * restrict q, const float * restrict x, int qk) {
// Least squares fits `d * q = x` for d. // Least squares fits `d * q = x` for d.
float qs2 = 0.0f; float qs2 = 0.0f;
@ -1445,18 +1432,20 @@ static void quantize_q_k_1(const float * x, int bits, int scale_bits, int block_
} }
} }
// Least squares fit min and scale. if (pass == 0) {
float min, scale; // Least squares fit min and scale.
lstsq_q_k(q_fit, x, q_m, block_size, &min, &scale); float min, scale;
// Check for nans. lstsq_q_k(q_fit, x, q_m, block_size, &min, &scale);
assert(min == min); // Check for nans.
assert(scale == scale); assert(min == min);
// Quantize to fp16 for the next pass. assert(scale == scale);
max_scale = GGML_FP16_TO_FP32(GGML_FP32_TO_FP16(scale)) * max_group; // Quantize to fp16 for the next pass.
max_min = GGML_FP16_TO_FP32(GGML_FP32_TO_FP16(min)) * max_group; max_scale = GGML_FP16_TO_FP32(GGML_FP32_TO_FP16(scale)) * max_group;
max_min = GGML_FP16_TO_FP32(GGML_FP32_TO_FP16(min)) * max_group;
*block_scale = GGML_FP32_TO_FP16(scale); *block_scale = GGML_FP32_TO_FP16(scale);
*block_min = GGML_FP32_TO_FP16(min); *block_min = GGML_FP32_TO_FP16(min);
}
} }
} }