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

quantize: add imatrix filename in KV

Browse source
This commit is contained in:
Pierrick HYMBERT 2024-04-13 19:06:50 +02:00
parent 851de160dd
commit 0d82da6f79
1 changed files with 8 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

8
examples/quantize/quantize.cpp
View file

@ -52,6 +52,7 @@ static const std::vector<struct quant_option> QUANT_OPTIONS = {
{ "COPY", LLAMA_FTYPE_ALL_F32, "only copy tensors, no quantizing", }, { "COPY", LLAMA_FTYPE_ALL_F32, "only copy tensors, no quantizing", },
}; };
static const char * const LLM_KV_QUANTIZE_IMATRIX_FILE = "quantize.imatrix.file";
static const char * const LLM_KV_QUANTIZE_IMATRIX_DATASET = "quantize.imatrix.dataset"; static const char * const LLM_KV_QUANTIZE_IMATRIX_DATASET = "quantize.imatrix.dataset";
static const char * const LLM_KV_QUANTIZE_IMATRIX_N_ENTRIES = "quantize.imatrix.entries_count"; static const char * const LLM_KV_QUANTIZE_IMATRIX_N_ENTRIES = "quantize.imatrix.entries_count";
static const char * const LLM_KV_QUANTIZE_IMATRIX_N_CHUNKS = "quantize.imatrix.chunks_count"; static const char * const LLM_KV_QUANTIZE_IMATRIX_N_CHUNKS = "quantize.imatrix.chunks_count";
@ -300,6 +301,13 @@ int main(int argc, char ** argv) {
int m_last_call = prepare_imatrix(imatrix_file, imatrix_dataset, included_weights, excluded_weights, imatrix_data); int m_last_call = prepare_imatrix(imatrix_file, imatrix_dataset, included_weights, excluded_weights, imatrix_data);
if (!imatrix_data.empty()) { if (!imatrix_data.empty()) {
params.imatrix = &imatrix_data; params.imatrix = &imatrix_data;
{
llama_model_kv_override kvo;
std::strcpy(kvo.key, LLM_KV_QUANTIZE_IMATRIX_FILE);
kvo.tag = LLAMA_KV_OVERRIDE_TYPE_STR;
kvo.str_value = strdup(imatrix_file.c_str());
kv_overrides.emplace_back(std::move(kvo));
}
if (!imatrix_dataset.empty()) { if (!imatrix_dataset.empty()) {
llama_model_kv_override kvo; llama_model_kv_override kvo;
std::strcpy(kvo.key, LLM_KV_QUANTIZE_IMATRIX_DATASET); std::strcpy(kvo.key, LLM_KV_QUANTIZE_IMATRIX_DATASET);