Merge pull request #2526 from gcuendet/fix-memory-leaks

Fix memory leaks

Author: narendasan

core/conversion/conversionctx/ConversionCtx.cpp

@@ -164,7 +164,7 @@ void ConversionCtx::RecordNewITensor(const torch::jit::Value* value, nvinfer1::I
 
 std::string ConversionCtx::SerializeEngine() {
 #if NV_TENSORRT_MAJOR > 7
-  auto serialized_network = builder->buildSerializedNetwork(*net, *cfg);
+  auto serialized_network = make_trt(builder->buildSerializedNetwork(*net, *cfg));
   if (!serialized_network) {
     TORCHTRT_THROW_ERROR("Building serialized network failed in TensorRT");
   }

core/runtime/register_jit_hooks.cpp

@@ -87,7 +87,7 @@ static auto TORCHTRT_UNUSED TRTEngineTSRegistrtion =
         .def_pickle(
             [](const c10::intrusive_ptr<TRTEngine>& self) -> std::vector<std::string> {
               // Serialize TensorRT engine
-              auto serialized_trt_engine = self->cuda_engine->serialize();
+              auto serialized_trt_engine = make_trt(self->cuda_engine->serialize());
 
               // Adding device info related meta data to the serialized file
               auto trt_engine = std::string((const char*)serialized_trt_engine->data(), serialized_trt_engine->size());