enable cache prompt

Author: tikikun

controllers/llamaCPP.cc

@@ -164,7 +164,7 @@ void llamaCPP::chatCompletion(
     data["frequency_penalty"] =
         (*jsonBody).get("frequency_penalty", 0).asFloat();
     data["presence_penalty"] = (*jsonBody).get("presence_penalty", 0).asFloat();
-
+    data["cache_prompt"] = true;
     const Json::Value &messages = (*jsonBody)["messages"];
     for (const auto &message : messages) {
       std::string input_role = message["role"].asString();