[ 仅 API 调用方式不一致 ]torch.utils.data.random_split

torch.utils.data.random_split(dataset, lengths, generator=<torch._C.Generator object>)

paddle.io.random_split(dataset, lengths, generator=None)

# PyTorch 写法
class Data(torch.utils.data.Dataset):
    def __init__(self):
        self.x = [0, 1, 2, 3, 4, 5, 6, 7, 8, 9]

    def __getitem__(self, idx):
        return self.x[idx]

    def __len__(self):
        return len(self.x)

data = Data()
datasets = torch.utils.data.random_split(data, [3, 7])

# Paddle 写法
class Data(paddle.io.Dataset):
    def __init__(self):
        self.x = [0, 1, 2, 3, 4, 5, 6, 7, 8, 9]

    def __getitem__(self, idx):
        return self.x[idx]

    def __len__(self):
        return len(self.x)

data = Data()
datasets = paddle.io.random_split(data, [3, 7])