Pytorch DataLoader 迭代顺序稳定吗?

Question

Pytorch Dataloader 的迭代顺序是否保证相同(在温和条件下(？

例如：

dataloader = DataLoader(my_dataset, batch_size=4,
shuffle=True, num_workers=4)
print("run 1")
for batch in dataloader:
print(batch["index"])
print("run 2")
for batch in dataloader:
print(batch["index"])

到目前为止，我已经尝试对其进行测试，它似乎不是固定的，两次运行的顺序相同。有没有办法使订单相同？谢谢

编辑：我也试过做

unlabeled_sampler = data.sampler.SubsetRandomSampler(unlabeled_indices)
unlabeled_dataloader = data.DataLoader(train_dataset, 
sampler=unlabeled_sampler, batch_size=args.batch_size, drop_last=False)

然后遍历数据加载器两次，但会产生相同的非确定性。

Answer 1

当shuffle=TrueDataLoader的迭代顺序在迭代之间不稳定时，简短的回答是否定的。每次迭代加载器时，内部RandomSampler都会创建一个新的随机顺序。

获得稳定的随机DataLoader的一种方法是使用一组随机排列的索引创建Subset数据集。

shuffled_dataset = torch.utils.data.Subset(my_dataset, torch.randperm(len(my_dataset)).tolist())
dataloader = DataLoader(shuffled_dataset, batch_size=4, num_workers=4, shuffled=False)

Answer 2

我实际上接受了jodag的评论答案：

torch.manual_seed("0")
for i,elt in enumerate(unlabeled_dataloader):
order.append(elt[2].item())
print(elt)
if i > 10:
break
torch.manual_seed("0")
print("new dataloader")
for i,elt in enumerate( unlabeled_dataloader):
print(elt)
if i > 10:
break
exit(1)                       

和输出：

[tensor([[-0.3583, -0.6944]]), tensor([3]), tensor([1610])]
[tensor([[-0.6623, -0.3790]]), tensor([3]), tensor([1958])]
[tensor([[-0.5046, -0.6399]]), tensor([3]), tensor([1814])]
[tensor([[-0.5349,  0.2365]]), tensor([2]), tensor([1086])]
[tensor([[-0.1310,  0.1158]]), tensor([0]), tensor([321])]
[tensor([[-0.2085,  0.0727]]), tensor([0]), tensor([422])]
[tensor([[ 0.1263, -0.1597]]), tensor([0]), tensor([142])]
[tensor([[-0.1387,  0.3769]]), tensor([1]), tensor([894])]
[tensor([[-0.0500,  0.8009]]), tensor([3]), tensor([1924])]
[tensor([[-0.6907,  0.6448]]), tensor([4]), tensor([2016])]
[tensor([[-0.2817,  0.5136]]), tensor([2]), tensor([1267])]
[tensor([[-0.4257,  0.8338]]), tensor([4]), tensor([2411])]
new dataloader
[tensor([[-0.3583, -0.6944]]), tensor([3]), tensor([1610])]
[tensor([[-0.6623, -0.3790]]), tensor([3]), tensor([1958])]
[tensor([[-0.5046, -0.6399]]), tensor([3]), tensor([1814])]
[tensor([[-0.5349,  0.2365]]), tensor([2]), tensor([1086])]
[tensor([[-0.1310,  0.1158]]), tensor([0]), tensor([321])]
[tensor([[-0.2085,  0.0727]]), tensor([0]), tensor([422])]
[tensor([[ 0.1263, -0.1597]]), tensor([0]), tensor([142])]
[tensor([[-0.1387,  0.3769]]), tensor([1]), tensor([894])]
[tensor([[-0.0500,  0.8009]]), tensor([3]), tensor([1924])]
[tensor([[-0.6907,  0.6448]]), tensor([4]), tensor([2016])]
[tensor([[-0.2817,  0.5136]]), tensor([2]), tensor([1267])]
[tensor([[-0.4257,  0.8338]]), tensor([4]), tensor([2411])]

这是如愿以偿的。但是，我认为约达格的主要答案仍然更好;这只是一个快速的黑客，现在;)