tlt train faster rcnn checkpoint interval


just wondering is there a way to set the checkpoint interval in the faster rcnn training spec so it doesnt save after every epoch? I tried to add the checkpoint_interval parameter like the detectnet inside the training_config block but i got a ParseError


Hi kaede_ai,
Unfortunately faster-rcnn doesn’t support checkpoint interval. Its training epochs are usually 10-20, so it is not designed to set an interval.