After successfully completed the train part for supernet, and I want to test supernet. However, it seems that the testsupernet.py needs an unexcepted .json file which named model-space. I try to generate it by run the countflops.py, is that the model-space file? Why test needs the flops count? Is this project provided a single program to just predict images and save their results?
I also want to know, if the parameter in args '--show' is set and '--show-dir', how did the test going to do? Just show results or show and save results?
After successfully completed the train part for supernet, and I want to test supernet. However, it seems that the testsupernet.py needs an unexcepted .json file which named model-space. I try to generate it by run the countflops.py, is that the model-space file? Why test needs the flops count? Is this project provided a single program to just predict images and save their results?