vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2025-09-15 19:39:42 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Merge commit 'f1e3eb4249' into concedo_experimental

Browse source 
# Conflicts:
#	.github/workflows/build.yml
#	README.md
#	docs/backend/SYCL.md
#	examples/llava/clip.cpp
#	ggml/src/ggml-sycl/CMakeLists.txt
#	ggml/src/ggml-vulkan/cmake/host-toolchain.cmake.in
This commit is contained in:
Concedo 2025-04-08 20:48:53 +08:00
commit 822cf2430e
26 changed files with 1822 additions and 2088 deletions
Download patch file Download diff file Expand all files Collapse all files

31
examples/gguf-split/gguf-split.cpp
View file

@ -409,8 +409,6 @@ static void gguf_merge(const split_params & split_params) {
exit(EXIT_FAILURE);
}
std::ofstream fout(split_params.output.c_str(), std::ios::binary);
fout.exceptions(std::ofstream::failbit); // fail fast on write errors
auto * ctx_out = gguf_init_empty();
@ -454,7 +452,6 @@ static void gguf_merge(const split_params & split_params) {
gguf_free(ctx_gguf);
ggml_free(ctx_meta);
gguf_free(ctx_out);
fout.close();
exit(EXIT_FAILURE);
}
@ -467,7 +464,6 @@ static void gguf_merge(const split_params & split_params) {
gguf_free(ctx_gguf);
ggml_free(ctx_meta);
gguf_free(ctx_out);
fout.close();
exit(EXIT_FAILURE);
}
@ -480,7 +476,6 @@ static void gguf_merge(const split_params & split_params) {
gguf_free(ctx_gguf);
ggml_free(ctx_meta);
gguf_free(ctx_out);
fout.close();
exit(EXIT_FAILURE);
}
@ -501,9 +496,11 @@ static void gguf_merge(const split_params & split_params) {
fprintf(stderr, "\033[3Ddone\n");
}
// placeholder for the meta data
{
std::ofstream fout;
if (!split_params.dry_run) {
fout.open(split_params.output.c_str(), std::ios::binary);
fout.exceptions(std::ofstream::failbit); // fail fast on write errors
// placeholder for the meta data
auto meta_size = gguf_get_meta_size(ctx_out);
::zeros(fout, meta_size);
}
@ -519,7 +516,9 @@ static void gguf_merge(const split_params & split_params) {
ggml_free(ctx_metas[i]);
}
gguf_free(ctx_out);
fout.close();
if (!split_params.dry_run) {
fout.close();
}
exit(EXIT_FAILURE);
}
fprintf(stderr, "%s: writing tensors %s ...", __func__, split_path);
@ -541,10 +540,11 @@ static void gguf_merge(const split_params & split_params) {
auto offset = gguf_get_data_offset(ctx_gguf) + gguf_get_tensor_offset(ctx_gguf, i_tensor);
f_input.seekg(offset);
f_input.read((char *)read_data.data(), n_bytes);
// write tensor data + padding
fout.write((const char *)read_data.data(), n_bytes);
zeros(fout, GGML_PAD(n_bytes, GGUF_DEFAULT_ALIGNMENT) - n_bytes);
if (!split_params.dry_run) {
// write tensor data + padding
fout.write((const char *)read_data.data(), n_bytes);
zeros(fout, GGML_PAD(n_bytes, GGUF_DEFAULT_ALIGNMENT) - n_bytes);
}
}
gguf_free(ctx_gguf);
@ -553,16 +553,15 @@ static void gguf_merge(const split_params & split_params) {
fprintf(stderr, "\033[3Ddone\n");
}
{
if (!split_params.dry_run) {
// go back to beginning of file and write the updated metadata
fout.seekp(0);
std::vector<uint8_t> data(gguf_get_meta_size(ctx_out));
gguf_get_meta_data(ctx_out, data.data());
fout.write((const char *)data.data(), data.size());
fout.close();
gguf_free(ctx_out);
}
gguf_free(ctx_out);
fprintf(stderr, "%s: %s merged from %d split with %d tensors.\n",
__func__, split_params.output.c_str(), n_split, total_tensors);