Update notebook to use postprocessing_use_tensorrt_nms:bool instead o…

…f engine

src/super_gradients/examples/model_export/models_export.ipynb

@@ -2209,31 +2209,30 @@
    "source": [
     "### Supported backends\n",
     "\n",
-    "Currently, we support two backends for exporting models:\n",
+    "Currently, model can be exported to ONNX format and be run on these inference engines:\n",
     "\n",
     "* ONNX Runtime\n",
     "* TensorRT\n",
+    "* OpenVINO\n",
     "\n",
-    "The only difference between these two backends is what NMS implementation will be used.\n",
-    "ONNX Runtime uses NMS implementation from ONNX opset, while TensorRT uses its own NMS implementation which is expected to be faster.\n",
+    "When model is exported with postprocessing enabled, the NMS module is attached to the ONNX graph.\n",
+    "There are two implementations of NMS that can be used: ONNX NMS and TensorRT NMS.\n",
     "\n",
-    "A disadvantage of TensorRT backend is that you cannot run model exported for TensorRT backend by ONNX Runtime.\n",
-    "You can, however, run models exported for ONNX Runtime backend inside TensorRT.\n",
+    "A TensorRT NMS as named suggests is NMS implementation available exclusively for TensorRT runtime.\n",
+    "It is generally slightly faster than ONNX NMS. \n",
+    "A disadvantage of TensorRT NMS is that you cannot run model exported with this NMS on other inference engines like ONNX Runtime.\n",
+    "You can, however, run models exported with ONNX NMS inside TensorRT.\n",
     "\n",
     "Therefore, ONNX Runtime backend is recommended for most use-cases and is used by default.\n",
     "\n",
-    "You can specify the desired execution backend by setting the `execution_backend` argument of `export()` method:\n",
+    "You can specify the desired NMS by setting the `postprocessing_use_tensorrt_nms_` argument of `export()` method:\n",
     "\n",
     "```python\n",
-    "from super_gradients.conversion import ExportTargetBackend\n",
-    "\n",
-    "model.export(..., engine=ExportTargetBackend.ONNXRUNTIME)\n",
+    "model.export(..., postprocessing_use_tensorrt_nms=False)\n",
     "```\n",
     "\n",
     "```python\n",
-    "from super_gradients.conversion import ExportTargetBackend\n",
-    "\n",
-    "model.export(..., engine=ExportTargetBackend.TENSORRT)\n",
+    "model.export(..., postprocessing_use_tensorrt_nms=True)\n",
     "```"
    ],
    "metadata": {