tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-06-30 12:55:17 +00:00
Code Issues Packages Projects Releases Wiki Activity

quantize : fix precedence of cli args (#6541)

Browse Source
This commit is contained in:
Georgi Gerganov
2024-04-08 16:23:01 +03:00
committed by GitHub
parent e3c337d87c
commit b73e564b16
1 changed files with 2 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
llama.cpp
View File

@ -13562,10 +13562,10 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
if (!params->pure && ggml_is_quantized(default_type)) { if (!params->pure && ggml_is_quantized(default_type)) {
new_type = llama_tensor_get_type(qs, new_type, tensor, ftype); new_type = llama_tensor_get_type(qs, new_type, tensor, ftype);
} }
else if (params->token_embedding_type < GGML_TYPE_COUNT && strcmp(tensor->name, "token_embd.weight") == 0) { if (params->token_embedding_type < GGML_TYPE_COUNT && strcmp(tensor->name, "token_embd.weight") == 0) {
new_type = params->token_embedding_type; new_type = params->token_embedding_type;
} }
else if (params->output_tensor_type < GGML_TYPE_COUNT && strcmp(tensor->name, "output.weight") == 0) { if (params->output_tensor_type < GGML_TYPE_COUNT && strcmp(tensor->name, "output.weight") == 0) {
new_type = params->output_tensor_type; new_type = params->output_tensor_type;
} }