CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-11-01 09:01:57 +00:00
Code Issues Packages Projects Releases Wiki Activity

gguf : start implementing quantization (WIP)

Browse Source
This commit is contained in:
M. Yusuf Sarıgöz
2023-08-12 10:40:56 +03:00
parent 186c496fdf
commit 4fa017a1f9
2 changed files with 20 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

23
gguf-util.h
View File

@@ -106,18 +106,21 @@ struct gguf_file {
}
void write_str(const std::string & val) {
size_t write_str(const std::string & val) {
size_t total_written = 0;
const int32_t n = val.size();
fwrite((const char *) &n, sizeof(n), 1, fp);
fwrite(val.c_str(), n, 1, fp);
total_written += fwrite((const char *) &n, sizeof(n), 1, fp);
total_written += fwrite(val.c_str(), n, 1, fp);
return total_written;
}
void write_i32(int32_t val) {
fwrite((const char *) &val, sizeof(val), 1, fp);
size_t write_i32(int32_t val) {
return fwrite((const char *) &val, sizeof(val), 1, fp);
}
void write_u64(size_t val) {
fwrite((const char *) &val, sizeof(val), 1, fp);
size_t write_u64(size_t val) {
return fwrite((const char *) &val, sizeof(val), 1, fp);
}
template<typename T>
@@ -167,6 +170,12 @@ struct gguf_file {
fwrite(val[i].c_str(), nstr, 1, fp);
}
}
void write_zeros(size_t count) {
for (size_t i = 0; i < count; ++i) {
fputc(0, fp);
}
}
};
#if defined(_WIN32)