CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-11-04 09:32:00 +00:00
Code Issues Packages Projects Releases Wiki Activity

cleaning debug quant

Browse Source
This commit is contained in:
ibrahimkhadraoui
2025-07-07 17:21:54 +04:00
parent 624699c53f
commit 042e5ff90b
1 changed files with 0 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
src/llama-quant.cpp
View File

@@ -315,11 +315,6 @@ static ggml_type llama_tensor_get_type(quantize_state_impl & qs, ggml_type new_t
else if (ftype == LLAMA_FTYPE_MOSTLY_IQ3_XXS) {
new_type = GGML_TYPE_IQ2_S;
}
} else if (name.find("ssm_in.weight") != std::string::npos) {
// For mamba-based models it's better to not quantize the ssm-proj layers
if (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_S) {
new_type = GGML_TYPE_BF16;
}
} else if (name.find("attn_q.weight") != std::string::npos) {
if (ftype == LLAMA_FTYPE_MOSTLY_IQ3_XS) {
new_type = GGML_TYPE_IQ3_XXS;