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

sync : ggml (conv 1d + 2d updates)

Browse source 
ggml-ci
This commit is contained in:
Georgi Gerganov 2023-10-04 12:21:20 +03:00
parent f72f8f22c9
commit 23277774df
No known key found for this signature in database
GPG key ID: 449E073F9DC10735
5 changed files with 707 additions and 360 deletions
Download patch file Download diff file Expand all files Collapse all files

953
ggml.c
View file

File diff suppressed because it is too large Load diff

13
ggml.h
View file

@ -401,10 +401,14 @@ extern "C" {
GGML_OP_CLAMP,
GGML_OP_CONV_1D,
GGML_OP_CONV_2D,
GGML_OP_CONV_TRANSPOSE_1D,
GGML_OP_CONV_TRANSPOSE_2D,
GGML_OP_POOL_1D,
GGML_OP_POOL_2D,
GGML_OP_CONV_1D_STAGE_0, // internal
GGML_OP_CONV_1D_STAGE_1, // internal
GGML_OP_UPSCALE, // nearest interpolate
GGML_OP_FLASH_ATTN,
@ -1386,6 +1390,14 @@ extern "C" {
int s,
int d);
GGML_API struct ggml_tensor * ggml_conv_transpose_1d(
struct ggml_context * ctx,
struct ggml_tensor * a,
struct ggml_tensor * b,
int s0,
int p0,
int d0);
GGML_API struct ggml_tensor * ggml_conv_2d(
struct ggml_context * ctx,
struct ggml_tensor * a,
@ -1759,6 +1771,7 @@ extern "C" {
GGML_OPT_NO_CONTEXT,
GGML_OPT_INVALID_WOLFE,
GGML_OPT_FAIL,
GGML_OPT_CANCEL,
GGML_LINESEARCH_FAIL = -128,
GGML_LINESEARCH_MINIMUM_STEP,

2
k_quants.c
View file

@ -69,7 +69,6 @@ inline static int32_t vaddvq_s32(int32x4_t v) {
// 2-6 bit quantization in super-blocks
//
//
// ===================== Helper functions
//
@ -348,7 +347,6 @@ void quantize_row_q2_K_reference(const float * restrict x, block_q2_K * restrict
const float q4scale = 15.f;
for (int i = 0; i < nb; i++) {
float max_scale = 0; // as we are deducting the min, scales are always positive
float max_min = 0;
for (int j = 0; j < QK_K/16; ++j) {

20
tests/test-grad0.cpp
View file

@ -208,26 +208,6 @@ static struct ggml_tensor * get_random_tensor_i32(
return result;
}
static void print_elements(const char* label, const struct ggml_tensor * t) {
if (!t) {
printf("%s: %s = null\n", __func__, label);
return;
}
const int nelements = ggml_nelements(t);
printf("%s: %s = [", __func__, label);
for (int k = 0; k < nelements; ++k) {
if (k > 0) { printf(", "); }
printf("%.5f", ggml_get_f32_1d(t, k));
}
printf("] shape: [");
for (int k = 0; k < t->n_dims; ++k) {
if (k > 0) { printf(", "); }
printf("%d", (int)t->ne[k]);
}
printf("]\n");
}
static bool check_gradient(
const char * op_name,
struct ggml_context * ctx0,

29
tests/test-opt.cpp
View file

@ -40,27 +40,6 @@ static float frand(void) {
return (float)rand()/(float)RAND_MAX;
}
static int irand(int n) {
return rand()%n;
}
static void get_random_dims(int64_t * dims, int ndims) {
dims[0] = dims[1] = dims[2] = dims[3] = 1;
for (int i = 0; i < ndims; i++) {
dims[i] = 1 + irand(4);
}
}
static void get_random_dims_minmax(int64_t * dims, int ndims, int min, int max) {
dims[0] = dims[1] = dims[2] = dims[3] = 1;
for (int i = 0; i < ndims; i++) {
dims[i] = min + irand(max-min);
}
}
static struct ggml_tensor * get_random_tensor(
struct ggml_context * ctx0, int ndims, int64_t ne[], float fmin, float fmax
) {
@ -106,14 +85,6 @@ static struct ggml_tensor * get_random_tensor(
return result;
}
static float get_element(const struct ggml_tensor * t, int idx) {
return ((float *)t->data)[idx];
}
static void set_element(struct ggml_tensor * t, int idx, float value) {
((float *)t->data)[idx] = value;
}
int main(void) {
struct ggml_init_params params = {
/* .mem_size = */ 1024*1024*1024,