feat: cortex-cpp with cortex.tensorrt-llm

cortex-cpp/controllers/server.cc

@@ -17,6 +17,7 @@ namespace {
 constexpr static auto kLlamaEngine = "cortex.llamacpp";
 constexpr static auto kPythonRuntimeEngine = "cortex.python";
 constexpr static auto kOnnxEngine = "cortex.onnx";
+constexpr static auto kTensorrtLlmEngine = "cortex.tensorrt-llm";
 }  // namespace
 
 server::server(){
@@ -264,6 +265,8 @@ void server::LoadModel(const HttpRequestPtr& req,
         return cortex_utils::kLlamaLibPath;
       } else if(e == kOnnxEngine) {
         return cortex_utils::kOnnxLibPath;
+      } else if(e == kTensorrtLlmEngine) {
+        return cortex_utils::kTensorrtLlmPath;
       }
       return cortex_utils::kLlamaLibPath;
     };
@@ -371,4 +374,4 @@ void server::UnloadEngines() {
   }
 }
 
-}  // namespace inferences
+}  // namespace inferences

cortex-cpp/controllers/server.h

@@ -149,4 +149,4 @@ class server : public drogon::HttpController<server>,
   std::unordered_map<std::string, EngineInfo> engines_;
   std::string cur_engine_type_;
 };
-};  // namespace inferences
+};  // namespace inferences

cortex-cpp/utils/cortex_utils.h

@@ -28,6 +28,7 @@ namespace cortex_utils {
 constexpr static auto kLlamaLibPath = "/engines/cortex.llamacpp";
 constexpr static auto kPythonRuntimeLibPath = "/engines/cortex.python";
 constexpr static auto kOnnxLibPath = "/engines/cortex.onnx";
+constexpr static auto kTensorrtLlmPath = "/engines/cortex.tensorrt-llm";
 
 inline std::string models_folder = "./models";