run_node¶

dragon.vm.tensorrt.onnx.run_node(
  node,
  inputs,
  device='CUDA:0',
  **kwargs
)[source]¶

Execute an onnx node once.

Parameters:

• node (onnx.NodeProto) – The onnx node.
• inputs (Union[Sequence, Dict]) – The input arrays.
• device (str, optional, default='CUDA:0') – The executing device.

Returns:

namedtuple – The model outputs.