happyz
/
llama.cpp
mirror of https://github.com/ggerganov/llama.cpp.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Populate activations_data with imatrix activations if present

This commit is contained in:
Ed Addario 2025-08-19 10:50:37 +01:00
parent e6d55dc47b
commit 77b818c040
No known key found for this signature in database
GPG Key ID: E7875815A3230993
1 changed files with 5 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
tools/quantize/quantize.cpp
View File

@ -561,10 +561,11 @@ int main(int argc, char ** argv) {
}
std::vector<std::string> imatrix_datasets;
std::unordered_map<std::string, std::vector<float>> imatrix_data;
int m_last_call = prepare_imatrix(imatrix_file, imatrix_datasets, included_weights, excluded_weights, imatrix_data);
if (!imatrix_data.empty()) {
params.imatrix = &imatrix_data;
std::unordered_map<std::string, std::vector<float>> values_data;
std::unordered_map<std::string, std::vector<float>> activations_data;
int m_last_call = prepare_imatrix(imatrix_file, imatrix_datasets, included_weights, excluded_weights, values_data, activations_data);
if (!values_data.empty()) {
params.imatrix = &values_data;
{
llama_model_kv_override kvo;
std::strcpy(kvo.key, LLM_KV_QUANTIZE_IMATRIX_FILE);