TensorRT version

[trungpham2606]

hi @wenbowen123
I had successfully converted the pytorch model to onnx model (based on your script). I also did some test with that onnx model and so far it showed similar results as pytorch one.
But after converting onnx model to trt model, the trt model output all nan values.
Can you provide some info about the tensorrt version you were using when converted to trt model ?

[VidhiRambhia]

I am experiencing this too

[zhiguo-rbs]

After I convert the onnx to tensorrt engine file,
I load the file as following:

TRT_LOGGER = trt.Logger(trt.Logger.INFO)
with trt.Runtime(TRT_LOGGER) as runtime:
    with open(engine_path, "rb") as f:
        engine_data = f.read()
        engine = runtime.deserialize_cuda_engine(engine_data)
        if engine is None:
            print("Failed to deserialize engine.")

I get Failed to deserialize engine