Random forest deployment in jetson

Robotics & Edge Computing Jetson Systems Jetson Nano
1

Hi,

I have jetson orin nano edge device. A random forest model was trained using tfdf library but some of the layers are not supported in order to convert it into .trt. Then we trained the model using sklearn and converted the model successfully into .onnx but again conversion to .trt was not possible.
Using cuml, random forest can be trained but will it work in jetson orin as it is specific to tensorRT.
If yes is there any documentation else is there any other edge device which supports deployment of random forest or any decision tree architecture.

Thanks

5

Dear @atul.pant,
You can write TensorRT plugin to support non supported layers. Could you check Developer Guide :: NVIDIA Deep Learning TensorRT Documentation