mmpretrain/mmcls/datasets/samplers/repeat_aug.py

107 lines
4.1 KiB
Python

import math
import torch
from mmengine.runner import get_dist_info
from torch.utils.data import Sampler
from mmcls.core.utils import sync_random_seed
from mmcls.datasets import SAMPLERS
@SAMPLERS.register_module()
class RepeatAugSampler(Sampler):
"""Sampler that restricts data loading to a subset of the dataset for
distributed, with repeated augmentation. It ensures that different each
augmented version of a sample will be visible to a different process (GPU).
Heavily based on torch.utils.data.DistributedSampler.
This sampler was taken from
https://github.com/facebookresearch/deit/blob/0c4b8f60/samplers.py
Used in
Copyright (c) 2015-present, Facebook, Inc.
"""
def __init__(self,
dataset,
num_replicas=None,
rank=None,
shuffle=True,
num_repeats=3,
selected_round=256,
selected_ratio=0,
seed=0):
default_rank, default_world_size = get_dist_info()
rank = default_rank if rank is None else rank
num_replicas = (
default_world_size if num_replicas is None else num_replicas)
self.dataset = dataset
self.num_replicas = num_replicas
self.rank = rank
self.shuffle = shuffle
self.num_repeats = num_repeats
self.epoch = 0
self.num_samples = int(
math.ceil(len(self.dataset) * num_repeats / self.num_replicas))
self.total_size = self.num_samples * self.num_replicas
# Determine the number of samples to select per epoch for each rank.
# num_selected logic defaults to be the same as original RASampler
# impl, but this one can be tweaked
# via selected_ratio and selected_round args.
selected_ratio = selected_ratio or num_replicas # ratio to reduce
# selected samples by, num_replicas if 0
if selected_round:
self.num_selected_samples = int(
math.floor(
len(self.dataset) // selected_round * selected_round /
selected_ratio))
else:
self.num_selected_samples = int(
math.ceil(len(self.dataset) / selected_ratio))
# In distributed sampling, different ranks should sample
# non-overlapped data in the dataset. Therefore, this function
# is used to make sure that each rank shuffles the data indices
# in the same order based on the same seed. Then different ranks
# could use different indices to select non-overlapped data from the
# same data list.
self.seed = sync_random_seed(seed)
def __iter__(self):
# deterministically shuffle based on epoch
if self.shuffle:
if self.num_replicas > 1: # In distributed environment
# deterministically shuffle based on epoch
g = torch.Generator()
# When :attr:`shuffle=True`, this ensures all replicas
# use a different random ordering for each epoch.
# Otherwise, the next iteration of this sampler will
# yield the same ordering.
g.manual_seed(self.epoch + self.seed)
indices = torch.randperm(
len(self.dataset), generator=g).tolist()
else:
indices = torch.randperm(len(self.dataset)).tolist()
else:
indices = list(range(len(self.dataset)))
# produce repeats e.g. [0, 0, 0, 1, 1, 1, 2, 2, 2....]
indices = [x for x in indices for _ in range(self.num_repeats)]
# add extra samples to make it evenly divisible
padding_size = self.total_size - len(indices)
indices += indices[:padding_size]
assert len(indices) == self.total_size
# subsample per rank
indices = indices[self.rank:self.total_size:self.num_replicas]
assert len(indices) == self.num_samples
# return up to num selected samples
return iter(indices[:self.num_selected_samples])
def __len__(self):
return self.num_selected_samples
def set_epoch(self, epoch):
self.epoch = epoch