Pytorch's Dataloader reimplemented using SeqTools

dataloader.py

"""A reimplementeation of PyTorch's DataLoader to showcase seqtools.

:author: Nicolas Granger
:license: 0BSD (~public domain)
"""

import numbers
import random
from functools import singledispatch
from multiprocessing import sharedctypes

import numpy as np
import torch

import seqtools


@singledispatch
def into_tensors(value):
    return torch.tensor(value)


@into_tensors.register(torch.Tensor)
def _(value):
    return value


@into_tensors.register(np.ndarray)
def _(value):
    return torch.from_numpy(value)


@into_tensors.register(tuple)
def _(value):
    return tuple(into_tensors(v) for v in value)


@into_tensors.register(list)
def _(value):
    return [into_tensors(v) for v in value]


@into_tensors.register(dict)
def _(value):
    return {k: into_tensors(v) for k, v in value.items()}


@singledispatch
def pin_memory(value):
    return value.pin_memory()


@pin_memory.register(tuple)
def _(value):
    return tuple(pin_memory(v) for v in value)


@pin_memory.register(list)
def _(value):
    return [pin_memory(v) for v in value]


@pin_memory.register(dict)
def _(value):
    return {k: pin_memory(v) for k, v in value.items()}


@singledispatch
def default_collate_fn(values):
    if not isinstance(values, (list, tuple)):
        values = list(values)

    sample = values[0]

    if isinstance(sample, torch.Tensor):
        return torch.stack(values)

    elif isinstance(sample, np.ndarray):
        return np.stack(values)

    elif isinstance(sample, numbers.Integral):
        return np.array(values)

    elif isinstance(sample, tuple):
        return tuple(default_collate_fn(row) for row in zip(*values))

    elif isinstance(sample, list):
        return [default_collate_fn(*row) for row in zip(*values)]

    elif isinstance(sample, dict):
        return {k: default_collate_fn(*(v[k] for v in values))
                for k in values[0].keys()}


class DataLoader:
    def __init__(self, dataset, batch_size=1, shuffle=False, sampler=None,
           batch_sampler=None, num_workers=0, collate_fn=None,
           pin_memory=False, drop_last=False, timeout=0,
           worker_init_fn=None):
        """

        Natbale differences:

        - all dataset elements must have the same shape
        - custom samplers are not implemented
        - timeout is irrelevant
        """
        if sampler is not None or batch_sampler is not None:
            raise NotImplementedError("custom samplers are not supported yet")

        # shuffle
        if shuffle:
            self.shuffling_indexes = memoryview(sharedctypes.RawArray('L', range(len(dataset))))
            dataset = seqtools.gather(dataset, self.shuffling_indexes)
        else:
            self.shuffling_indexes = None

        if batch_size is not None:
            dataset = seqtools.batch(
                dataset,
                k=batch_size, drop_last=drop_last,
                collate_fn=collate_fn or default_collate_fn)

        if num_workers > 0:
            dataset = seqtools.prefetch(
                dataset,
                max_buffered=num_workers * 10, nworkers=num_workers, method='sharedmem',
                start_hook=worker_init_fn)

        # convert values into tensors
        dataset = seqtools.smap(into_tensors, dataset)

        if pin_memory:
            dataset = seqtools.smap(pin_memory, dataset)

        self.dataset = dataset

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

    def __iter__(self):
        if self.shuffling_indexes is not None:
            random.shuffle(self.shuffling_indexes)

        return iter(self.dataset)

main.py

# Adapted from https://github.com/pytorch/examples/blob/master/mnist/main.py

import argparse

import numpy as np
import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim
from torchvision import datasets, transforms
from torch.optim.lr_scheduler import StepLR

import seqtools
from docs.examples.dataloader import DataLoader


class Net(nn.Module):
    def __init__(self):
        super(Net, self).__init__()
        self.conv1 = nn.Conv2d(1, 32, 3, 1)
        self.conv2 = nn.Conv2d(32, 64, 3, 1)
        self.dropout1 = nn.Dropout2d(0.25)
        self.dropout2 = nn.Dropout2d(0.5)
        self.fc1 = nn.Linear(9216, 128)
        self.fc2 = nn.Linear(128, 10)

    def forward(self, x):
        x = self.conv1(x)
        x = F.relu(x)
        x = self.conv2(x)
        x = F.max_pool2d(x, 2)
        x = self.dropout1(x)
        x = torch.flatten(x, 1)
        x = self.fc1(x)
        x = F.relu(x)
        x = self.dropout2(x)
        x = self.fc2(x)
        output = F.log_softmax(x, dim=1)
        return output


def train(args, model, device, train_loader, optimizer, epoch):
    model.train()
    for batch_idx, (data, target) in enumerate(train_loader):
        data, target = data.to(device), target.to(device)
        optimizer.zero_grad()
        output = model(data)
        loss = F.nll_loss(output, target)
        loss.backward()
        optimizer.step()
        if batch_idx % args.log_interval == 0:
            print('Train Epoch: {} [{}/{} ({:.0f}%)]\tLoss: {:.6f}'.format(
                epoch, batch_idx * len(data), len(train_loader.dataset),
                100. * batch_idx / len(train_loader), loss.item()))


def test(args, model, device, test_loader):
    model.eval()
    test_loss = 0
    correct = 0
    with torch.no_grad():
        for data, target in test_loader:
            data, target = data.to(device), target.to(device)
            output = model(data)
            test_loss += F.nll_loss(output, target, reduction='sum').item()  # sum up batch loss
            pred = output.argmax(dim=1, keepdim=True)  # get the index of the max log-probability
            correct += pred.eq(target.view_as(pred)).sum().item()

    test_loss /= len(test_loader.dataset)

    print('\nTest set: Average loss: {:.4f}, Accuracy: {}/{} ({:.0f}%)\n'.format(
        test_loss, correct, len(test_loader.dataset),
        100. * correct / len(test_loader.dataset)))


def main():
    # Training settings
    parser = argparse.ArgumentParser(description='PyTorch MNIST Example')
    parser.add_argument('--batch-size', type=int, default=64, metavar='N',
                        help='input batch size for training (default: 64)')
    parser.add_argument('--test-batch-size', type=int, default=1000, metavar='N',
                        help='input batch size for testing (default: 1000)')
    parser.add_argument('--epochs', type=int, default=14, metavar='N',
                        help='number of epochs to train (default: 14)')
    parser.add_argument('--lr', type=float, default=1.0, metavar='LR',
                        help='learning rate (default: 1.0)')
    parser.add_argument('--gamma', type=float, default=0.7, metavar='M',
                        help='Learning rate step gamma (default: 0.7)')
    parser.add_argument('--no-cuda', action='store_true', default=False,
                        help='disables CUDA training')
    parser.add_argument('--seed', type=int, default=1, metavar='S',
                        help='random seed (default: 1)')
    parser.add_argument('--log-interval', type=int, default=10, metavar='N',
                        help='how many batches to wait before logging training status')

    parser.add_argument('--save-model', action='store_true', default=False,
                        help='For Saving the current Model')
    args = parser.parse_args()
    use_cuda = not args.no_cuda and torch.cuda.is_available()

    torch.manual_seed(args.seed)

    device = torch.device("cuda" if use_cuda else "cpu")

    kwargs = {'num_workers': 1, 'pin_memory': True} if use_cuda else {'num_workers': 1}
    train_loader = DataLoader(
        seqtools.starmap(
            lambda img, label: ((np.asarray(img, dtype=np.float32)[None] - 0.1307) / 0.3081, label),
            datasets.MNIST('/tmp/MNIST', train=True, download=True)),
        batch_size=args.batch_size, shuffle=True, **kwargs)
    test_loader = DataLoader(
        seqtools.starmap(
            lambda img, label: ((np.asarray(img, dtype=np.float32)[None] - 0.1307) / 0.3081, label),
            datasets.MNIST('/tmp/MNIST', train=False, download=True)),
        batch_size=args.test_batch_size, shuffle=True, **kwargs)

    model = Net().to(device)
    optimizer = optim.Adadelta(model.parameters(), lr=args.lr)

    scheduler = StepLR(optimizer, step_size=1, gamma=args.gamma)
    for epoch in range(1, args.epochs + 1):
        train(args, model, device, train_loader, optimizer, epoch)
        test(args, model, device, test_loader)
        scheduler.step()

    if args.save_model:
        torch.save(model.state_dict(), "mnist_cnn.pt")


if __name__ == '__main__':
    main()