Data division in incremental fashion

1

I want to divide the data set in following fashion
30000 input size
1st iteration : 15000
2nd iteration:30000
in incremental fashion for MNIST dataset.

num_clients = 2

traindata = datasets.MNIST('./data', train=True, download=True,
                       transform=transforms.Compose([transforms.ToTensor(),transforms.Normalize((0.1307,), (0.3081,))])
                       )
traindata_split = torch.utils.data.random_split(traindata, [int(traindata.data.shape[0] / num_clients) for _ in range(num_clients)])

i wish to divide the traindata_split into two or more chunk in incremental fashion eg,1st chunk 15000,second 30000