mirrors/ollama-for-amd
1
0
Fork 0
mirror of https://github.com/likelovewant/ollama-for-amd.git synced 2025-12-23 23:18:26 +00:00
Code Issues Packages Projects Releases Wiki Activity

update llama.cpp to e782c9e735f93ab4767ffc37462c523b73a17ddc

Browse Source
This commit is contained in:
Michael Yang
2023-07-19 16:47:05 -07:00
parent 5156e48c2a
commit a83eaa7a9f
12 changed files with 1724 additions and 666 deletions
Download Patch File Download Diff File Expand all files Collapse all files

51
llama/ggml.h
View File

@@ -1,5 +1,5 @@
/**
* llama.cpp - git 5bf2a2771886ee86137e01dbc7492f78fb392066
* llama.cpp - git e782c9e735f93ab4767ffc37462c523b73a17ddc
*
* MIT License
*
@@ -227,8 +227,13 @@
#define GGML_MAX_NAME 48
#define GGML_DEFAULT_N_THREADS 4
#define GGML_EXIT_SUCCESS 0
#define GGML_EXIT_ABORTED 1
#define GGML_UNUSED(x) (void)(x)
#define GGML_ASSERT(x) \
do { \
if (!(x)) { \
@@ -389,6 +394,8 @@ extern "C" {
GGML_OP_CLAMP,
GGML_OP_CONV_1D,
GGML_OP_CONV_2D,
GGML_OP_POOL_1D,
GGML_OP_POOL_2D,
GGML_OP_FLASH_ATTN,
GGML_OP_FLASH_FF,
@@ -468,6 +475,10 @@ extern "C" {
// the `n_tasks` of nodes, 1:1 mapping to cgraph nodes
int n_tasks[GGML_MAX_NODES];
// abort ggml_graph_compute when true
bool (*abort_callback)(void * data);
void * abort_callback_data;
};
// computation graph
@@ -1136,6 +1147,17 @@ extern "C" {
int mode,
int n_ctx);
// custom RoPE, in-place, returns view(a)
GGML_API struct ggml_tensor * ggml_rope_custom_inplace(
struct ggml_context * ctx,
struct ggml_tensor * a,
int n_past,
int n_dims,
int mode,
float freq_base,
float freq_scale,
int n_ctx);
// rotary position embedding backward, i.e compute dx from dy
// a - dy
GGML_API struct ggml_tensor * ggml_rope_back(
@@ -1190,6 +1212,31 @@ extern "C" {
int s,
int d);
enum ggml_op_pool {
GGML_OP_POOL_MAX,
GGML_OP_POOL_AVG,
GGML_OP_POOL_COUNT,
};
GGML_API struct ggml_tensor* ggml_pool_1d(
struct ggml_context * ctx,
struct ggml_tensor * a,
enum ggml_op_pool op,
int k0, // kernel size
int s0, // stride
int p0); // padding
GGML_API struct ggml_tensor* ggml_pool_2d(
struct ggml_context * ctx,
struct ggml_tensor * a,
enum ggml_op_pool op,
int k0,
int k1,
int s0,
int s1,
int p0,
int p1);
GGML_API struct ggml_tensor * ggml_flash_attn(
struct ggml_context * ctx,
struct ggml_tensor * q,
@@ -1329,7 +1376,7 @@ extern "C" {
// ggml_graph_plan() has to be called before ggml_graph_compute()
// when plan.work_size > 0, caller must allocate memory for plan.work_data
GGML_API struct ggml_cplan ggml_graph_plan (struct ggml_cgraph * cgraph, int n_threads /*= GGML_DEFAULT_N_THREADS*/);
GGML_API void ggml_graph_compute(struct ggml_cgraph * cgraph, struct ggml_cplan * cplan);
GGML_API int ggml_graph_compute(struct ggml_cgraph * cgraph, struct ggml_cplan * cplan);
GGML_API void ggml_graph_reset (struct ggml_cgraph * cgraph);
// same as ggml_graph_compute() but the work data is allocated as a part of the context