models/lib/utils/data/distributed.py

import math

import torch
from torch.distributed import get_rank, get_world_size

from .sampler import Sampler


class DistributedSampler(Sampler):
    """Sampler that restricts data loading to a subset of the dataset.

    It is especially useful in conjunction with
    :class:`torch.nn.parallel.DistributedDataParallel`. In such case, each
    process can pass a DistributedSampler instance as a DataLoader sampler,
    and load a subset of the original dataset that is exclusive to it.

    .. note::
        Dataset is assumed to be of constant size.

    Arguments:
        dataset: Dataset used for sampling.
        num_replicas (optional): Number of processes participating in
            distributed training.
        rank (optional): Rank of the current process within num_replicas.
    """

    def __init__(self, dataset, num_replicas=None, rank=None):
        if num_replicas is None:
            num_replicas = get_world_size()
        if rank is None:
            rank = get_rank()
        self.dataset = dataset
        self.num_replicas = num_replicas
        self.rank = rank
        self.epoch = 0
        self.num_samples = int(math.ceil(len(self.dataset) * 1.0 / self.num_replicas))
        self.total_size = self.num_samples * self.num_replicas

    def __iter__(self):
        # deterministically shuffle based on epoch
        g = torch.Generator()
        g.manual_seed(self.epoch)
        indices = list(torch.randperm(len(self.dataset), generator=g))

        # add extra samples to make it evenly divisible
        indices += indices[:(self.total_size - len(indices))]
        assert len(indices) == self.total_size

        # subsample
        offset = self.num_samples * self.rank
        indices = indices[offset:offset + self.num_samples]
        assert len(indices) == self.num_samples

        return iter(indices)

    def __len__(self):
        return self.num_samples

    def set_epoch(self, epoch):
        self.epoch = epoch