defining DETECTIONS_PER_IMAGE in the lazy config

I wanna export my model that has been trained by the lazyconfig_train_net.py. The default value for DETECTIONS_PER_IMAGE= 100 i assume. How can i set this via lazy? When the training was done in the normal way, i could just override the value cfg.TEST.DETECTIONS_PER_IMAGE = 1000.

Start with lazyconfig_train_net.py and search for DETECTIONS_PER_IMAGE and the lazy configuration entry points. Confirm how the value is supplied for model export, then document that workflow so a user can set the detection limit without the normal cfg override.

Written by the indexing model from the issue text.