LLM inference in C/C++
C++ ggml
llama : use LLM_KV for quantization_version & file_type (#24802)
Adrien Gallouët84de01a1f1c8
diff --git a/src/llama-quant.cpp b/src/llama-quant.cppindex cf92ce4bb8..89b7fe8d43 100644--- a/src/llama-quant.cpp+++ b/src/llama-quant.cpp@@ -932,8 +932,8 @@ static void llama_model_quantize_impl(const std::string & fname_inp, const std::// copy the KV pairs from the input filegguf_set_kv (ctx_out.get(), ml.metadata);- gguf_set_val_u32(ctx_out.get(), "general.quantization_version", GGML_QNT_VERSION); // TODO: use LLM_KV- gguf_set_val_u32(ctx_out.get(), "general.file_type", ftype); // TODO: use LLM_KV+ gguf_set_val_u32(ctx_out.get(), ml.llm_kv(LLM_KV_GENERAL_QUANTIZATION_VERSION).c_str(), GGML_QNT_VERSION);+ gguf_set_val_u32(ctx_out.get(), ml.llm_kv(LLM_KV_GENERAL_FILE_TYPE).c_str(), ftype);// Remove split metadatagguf_remove_key(ctx_out.get(), ml.llm_kv(LLM_KV_SPLIT_NO).c_str());