quantize : fix precedence of cli args (#6541)

diff --git a/llama.cpp b/llama.cpp
index 9dde3efd03f34d185064f87d98d9e5476b594485..89ea3fe1ae7fab7b79786f709635967a6bdf0750 100644 (file)
--- a/llama.cpp
+++ b/llama.cpp
@@ -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)) {
                 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;
             }
-            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;
             }