Can I use both variable(dynamic) input shape and batching? And what’s the strategy to padding input when batching?
TensorRT version: lastest