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

Fix merge conflicts

This commit is contained in:
Colin Kealty 2026-03-10 11:45:43 -04:00
parent 544745c034
commit 182cbe5d64
1 changed files with 5 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
src/llama-quant.cpp
View File

@ -913,7 +913,8 @@ static void llama_model_quantize_impl(const std::string & fname_inp, const std::
imatrix_data = static_cast<const std::unordered_map<std::string, std::vector<float>>*>(params->imatrix);
if (imatrix_data) {
LLAMA_LOG_INFO("\n%s: have importance matrix data with %d entries\n",
__func__, (int)imatrix_data->size()); qs.has_imatrix = true;
__func__, (int)imatrix_data->size());
qs.has_imatrix = true;
// check imatrix for nans or infs
for (const auto & kv : *imatrix_data) {
for (float f : kv.second) {
@ -995,7 +996,7 @@ static void llama_model_quantize_impl(const std::string & fname_inp, const std::
});
}
{ // Based on old loop
{
std::vector<std::string> tensor_names;
tensor_names.reserve(tensors.size());
for (const auto * it : tensors) {
@ -1005,7 +1006,6 @@ static void llama_model_quantize_impl(const std::string & fname_inp, const std::
}
int idx = 0;
uint16_t n_split = 1;
// Assume split index is continuous
@ -1048,6 +1048,8 @@ static void llama_model_quantize_impl(const std::string & fname_inp, const std::
const struct ggml_tensor * tensor = it->tensor;
const std::string name = ggml_get_name(tensor);
metadata[i].category = tensor_get_category(name);
uint16_t i_split = params->keep_split ? it->idx : 0;
if (!ctx_outs[i_split]) {
ctx_outs[i_split].reset(gguf_init_empty());