Using TensorRT for inference acceleration

[TAKAGNU]

Could the author please provide an explanation of using tensorrt for inference acceleration? At present, the real-time inference performance of the model directly on the RTX 4090 graphics card still fails to meet the requirements.

[ynma-hanvo]

hi, what is your input size and process time