Hyperparameter tuning for detectron2 maskrcnn

Question

For the hyperparameter tuning, below code shows the configuration of the model, i can change its learning rate, iterations, batch size but i am stuck in changing its filter size, activation fuction,padding etc. How to do modification of the code to change above parameters? Model used: MASK_RCNN_R101_FPN_3x

# HYPERPARAMETERS
ARCHITECTURE = "mask_rcnn_R_101_FPN_3x"
CONFIG_FILE_PATH = f"COCO-InstanceSegmentation/{ARCHITECTURE}.yaml"
MAX_ITER = 2000
EVAL_PERIOD = 1500
BASE_LR = 0.001
NUM_CLASSES = 2

cfg = get_cfg()
cfg.merge_from_file(model_zoo.get_config_file(CONFIG_FILE_PATH))
cfg.MODEL.WEIGHTS = model_zoo.get_checkpoint_url(CONFIG_FILE_PATH)
cfg.DATASETS.TRAIN = (TRAIN_DATA_SET_NAME,)
cfg.DATASETS.TEST = (TEST_DATA_SET_NAME,)
cfg.MODEL.ROI_HEADS.BATCH_SIZE_PER_IMAGE = 128
cfg.TEST.EVAL_PERIOD = EVAL_PERIOD
cfg.DATALOADER.NUM_WORKERS = 2
cfg.SOLVER.IMS_PER_BATCH = 2
cfg.INPUT.MASK_FORMAT='bitmask'
cfg.SOLVER.BASE_LR = BASE_LR
cfg.SOLVER.MAX_ITER = MAX_ITER
cfg.MODEL.ROI_HEADS.NUM_CLASSES = NUM_CLASSES
cfg.OUTPUT_DIR = OUTPUT_DIR_PATH

I have tried to find the model architecture files but cant find it when downloading the detectron2 from github.