Xiao

it seems that the 'test' and 'trainval' of trian.py is not fit the data loader. ### train.py ``` PCN_TRAIN_DATASET, PCN_TEST_DATASET, AE_TRAIN_DATASET, AE_TEST_DATASET, NUM_PARTS = data_utils.load_data(DATA_PATH, NUM_POINTS, CATEGORY, 'test', 'trainval') ```...

Hello, I have noticed that when you are preparing the data set, the threshold written in the code when calculating contact points is 1e-3, and when evaluating, the threshold is...