Is there any example like that over the web?
I’d like to benchmark tvm with my model. It’s a bit non-standard because it takes two images of the fixed size (one RGB and another gray mask).
I’ve managed to run autotuning and export it like described here.
Now I’d like to have the small command line app which I could run to measure time (sanity check), memory usage and get results (sanity check) to confirm that tvm can be used as the replacement of the current inference engine.
I know android_deploy example but it’d be nicer to have the minimum command line app before going further because it’s easier to benchmark and update.