split = list(tfds.Split.TRAIN.subsplit(weighted = (8, 2))). ) # loads the training set only and then splits it n_samples = meta.splits['train'].num_examples.
確定! 回上一頁