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

GGUF: C++ refactor, backend support, misc fixes (#11030)

Browse source 
* GGUF: C++ refactor, backend support, misc fixes

remove ggml_tensor.backend

update CODEOWNERS [no ci]

remove gguf_get_data from API

revise GGUF API data types
This commit is contained in:
Johannes Gäßler 2025-01-07 18:01:58 +01:00 • committed by GitHub
parent 017cc5f446
commit 53ff6b9b9f
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
21 changed files with 1795 additions and 1627 deletions
Download patch file Download diff file Expand all files Collapse all files

6
examples/llava/clip.cpp
View file

@ -7,6 +7,7 @@
#include "ggml-cpu.h"
#include "ggml-alloc.h"
#include "ggml-backend.h"
#include "gguf.h"
//#ifdef GGML_USE_CUDA
//#include "ggml-cuda.h"
@ -262,7 +263,7 @@ static std::string gguf_kv_to_str(const struct gguf_context * ctx_gguf, int i) {
{
const enum gguf_type arr_type = gguf_get_arr_type(ctx_gguf, i);
int arr_n = gguf_get_arr_n(ctx_gguf, i);
const void * data = gguf_get_arr_data(ctx_gguf, i);
const void * data = arr_type == GGUF_TYPE_STRING ? nullptr : gguf_get_arr_data(ctx_gguf, i);
std::stringstream ss;
ss << "[";
for (int j = 0; j < arr_n; j++) {
@ -2734,7 +2735,8 @@ bool clip_model_quantize(const char * fname_inp, const char * fname_out, const i
total_size_org += orig_size;
total_size_new += new_size;
gguf_set_tensor_type(ctx_out, name.c_str(), new_type);
gguf_set_tensor_data(ctx_out, name.c_str(), new_data, new_size);
GGML_ASSERT(gguf_get_tensor_size(ctx_out, gguf_find_tensor(ctx_out, name.c_str())) == new_size);
gguf_set_tensor_data(ctx_out, name.c_str(), new_data);
fout.write((const char *)new_data, new_size);
size_t pad = GGML_PAD(new_size, gguf_get_alignment(ctx_out)) - new_size;
for (size_t j = 0; j < pad; ++j) {