fix: reorder is_embedded_model and params.embedding

Author: vansangpfiev

controllers/llamaCPP.cc

@@ -654,9 +654,9 @@ bool llamaCPP::LoadModelImpl(std::shared_ptr<Json::Value> jsonBody) {
 
     params.n_gpu_layers = jsonBody->get("ngl", 100).asInt();
     params.n_ctx = jsonBody->get("ctx_len", 2048).asInt();
+    params.embedding = jsonBody->get("embedding", true).asBool();
     is_embedded_model =
         !(*jsonBody)["embedding"].isNull() && (*jsonBody)["embedding"].asBool();
-    params.embedding = jsonBody->get("embedding", true).asBool();
     // Check if n_parallel exists in jsonBody, if not, set to drogon_thread
     params.n_batch = jsonBody->get("n_batch", 512).asInt();
     params.n_parallel = jsonBody->get("n_parallel", 1).asInt();