Change debug prints to function syntax (#946)

Author: vshampor

modules/llama_cpp_plugin/src/compiled_model.cpp

@@ -27,11 +27,11 @@ LlamaCppModel::LlamaCppModel(const std::string& gguf_fname,
     : ICompiledModel(nullptr, plugin),
       m_gguf_fname(gguf_fname),
       m_num_threads(num_threads) {
-    OPENVINO_DEBUG << "llama_cpp_plugin: loading llama model directly from GGUF... " << std::endl;
+    OPENVINO_DEBUG("llama_cpp_plugin: loading llama model directly from GGUF... \n");
     llama_model_params mparams = llama_model_default_params();
     mparams.n_gpu_layers = 99;
     m_llama_model_ptr = llama_load_model_from_file(gguf_fname.c_str(), mparams);
-    OPENVINO_DEBUG << "llama_cpp_plugin: llama model loaded successfully from GGUF..." << std::endl;
+    OPENVINO_DEBUG("llama_cpp_plugin: llama model loaded successfully from GGUF... \n");
 
     auto input_ids = std::make_shared<ov::opset13::Parameter>(ov::element::Type_t::i64, ov::PartialShape({-1, -1}));
     auto fake_convert = std::make_shared<ov::opset13::Convert>(input_ids->output(0), ov::element::Type_t::f32);
@@ -71,7 +71,7 @@ std::shared_ptr<const ov::Model> LlamaCppModel::get_runtime_model() const {
 }
 
 void LlamaCppModel::set_property(const ov::AnyMap& properties) {
-    OPENVINO_DEBUG << "llama_cpp_plugin: attempted to set_property (did nothing)";
+    OPENVINO_DEBUG("llama_cpp_plugin: attempted to set_property (did nothing)");
 }
 
 ov::Any LlamaCppModel::get_property(const std::string& name) const {

modules/llama_cpp_plugin/src/infer_request.cpp

@@ -28,7 +28,7 @@ void allocate_tensor_impl(ov::SoPtr<ov::ITensor>& tensor,
 LlamaCppSyncInferRequest::LlamaCppSyncInferRequest(const std::shared_ptr<const LlamaCppModel>& compiled_model,
                                                    size_t num_threads)
     : ov::ISyncInferRequest(compiled_model) {
-    OPENVINO_DEBUG << "llama_cpp_plugin: infer request ctor called\n";
+    OPENVINO_DEBUG("llama_cpp_plugin: infer request ctor called\n");
     llama_context_params cparams = llama_context_default_params();
     cparams.n_threads = num_threads ? num_threads : std::thread::hardware_concurrency();
     cparams.n_ctx = 0;  // this means that the actual n_ctx will be taken equal to the model's train-time value
@@ -51,7 +51,7 @@ LlamaCppSyncInferRequest::LlamaCppSyncInferRequest(const std::shared_ptr<const L
 }
 void LlamaCppSyncInferRequest::set_tensors_impl(const ov::Output<const ov::Node> port,
                                                 const std::vector<ov::SoPtr<ov::ITensor>>& tensors) {
-    OPENVINO_DEBUG << "llama_cpp_plugin: set_tensors_impl called\n";
+    OPENVINO_DEBUG("llama_cpp_plugin: set_tensors_impl called\n");
 }
 
 void llama_batch_add_reimpl(struct llama_batch& batch,
@@ -131,12 +131,12 @@ void LlamaCppSyncInferRequest::infer() {
     llama_batch_free(batch);
 };
 std::vector<ov::ProfilingInfo> LlamaCppSyncInferRequest::get_profiling_info() const {
-    OPENVINO_DEBUG << "llama_cpp_plugin: get_profiling_info() called\n";
+    OPENVINO_DEBUG("llama_cpp_plugin: get_profiling_info() called\n");
     return std::vector<ov::ProfilingInfo>{};
 };
 
 std::vector<ov::SoPtr<ov::IVariableState>> LlamaCppSyncInferRequest::query_state() const {
-    OPENVINO_DEBUG << "llama_cpp_plugin: query_state() called\n";
+    OPENVINO_DEBUG("llama_cpp_plugin: query_state() called\n");
     return {std::static_pointer_cast<ov::IVariableState>(std::make_shared<LlamaCppState>(m_llama_ctx))};
 }