traced_module_float = torch.jit.trace(model, images_float)
    compile_settings = {
        "inputs": [
            torch_tensorrt.Input(
                min_shape=[max_batch, in_chans, img_size, img_size],
                opt_shape=[max_batch, in_chans, img_size, img_size],
                max_shape=[max_batch, in_chans, img_size, img_size],
            )],
        "enabled_precisions": torch.float,
    }
    trt_traced_module_float = torch_tensorrt.compile(traced_module_float, **compile_settings)

ERROR: [Torch-TensorRT] - Unsupported operator: aten::roll(Tensor self, int[1] shifts, int[1] dims=[]) -> (Tensor)
…
ERROR: [Torch-TensorRT] - Unsupported operator: aten::gelu(Tensor self, bool approximate) -> (Tensor)
…
ERROR: [Torch-TensorRT] - Unsupported operator: aten::Int.Tensor(Tensor a) -> (int)
…
ERROR: [Torch-TensorRT] - Unsupported operator: aten::div.Tensor_mode(Tensor self, Tensor other, *, str? rounding_mode) -> (Tensor)
…
ERROR: [Torch-TensorRT] - Method requested cannot be compiled by Torch-TensorRT.TorchScript.
Unsupported operators listed below:
  - aten::roll(Tensor self, int[1] shifts, int[1] dims=[]) -> (Tensor)
  - aten::gelu(Tensor self, bool approximate) -> (Tensor)
  - aten::Int.Tensor(Tensor a) -> (int)
  - aten::div.Tensor_mode(Tensor self, Tensor other, *, str? rounding_mode) -> (Tensor)
You can either implement converters for these ops in your application or request implementation
https://www.github.com/nvidia/Torch-TensorRT/issues
…