ikawrakow/ik_llama.cpp
1
0
Fork 0
mirror of https://github.com/ikawrakow/ik_llama.cpp.git synced 2026-04-30 11:21:56 +00:00
Code Issues Packages Projects Releases Wiki Activity

SER - Smart Expert Reduction (#239)

Browse Source 
* A better way to measure the cost of ggml_barrier

* Smart expert selection

* Add ser option to llama-bench

---------

Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
This commit is contained in:
Kawrakow
2025-03-02 13:47:38 +02:00
committed by GitHub
parent 101c888724
commit 9424c80ab1
11 changed files with 330 additions and 27 deletions
Download Patch File Download Diff File Expand all files Collapse all files

13
ggml/include/ggml.h
View File

@@ -597,6 +597,7 @@ extern "C" {
GGML_OP_ARANGE,
GGML_OP_TIMESTEP_EMBEDDING,
GGML_OP_ARGSORT,
GGML_OP_ARGSORT_THRESH,
GGML_OP_LEAKY_RELU,
GGML_OP_SOFTCAP,
GGML_OP_SOFT_CAP_MAX,
@@ -1913,6 +1914,12 @@ extern "C" {
struct ggml_tensor * a,
enum ggml_sort_order order);
GGML_API struct ggml_tensor * ggml_argsort_thresh(
struct ggml_context * ctx,
struct ggml_tensor * a,
int min_entries,
float threshold);
GGML_API struct ggml_tensor * ggml_arange(
struct ggml_context * ctx,
float start,
@@ -1924,6 +1931,12 @@ extern "C" {
struct ggml_context * ctx,
struct ggml_tensor * a,
int k);
GGML_API struct ggml_tensor * ggml_top_k_thresh(
struct ggml_context * ctx,
struct ggml_tensor * a,
int k,
int min_entries,
float thresh);
#define GGML_KQ_MASK_PAD 32