比较与torch.utils.data.WeightedRandomSampler的差异

torch.utils.data.WeightedRandomSampler

class torch.utils.data.WeightedRandomSampler(weights, num_samples, replacement=True, generator=None)

更多内容详见torch.utils.data.WeightedRandomSampler。

mindspore.dataset.WeightedRandomSampler

class mindspore.dataset.WeightedRandomSampler(weights, num_samples=None, replacement=True)

更多内容详见mindspore.dataset.WeightedRandomSampler。

差异对比

PyTorch：给定样本的权重列表，根据权重的大小对样本进行采样，支持指定采样逻辑。

MindSpore：给定样本的权重列表，根据权重的大小对样本进行采样，不支持指定采样逻辑。

分类	子类	PyTorch	MindSpore	差异
参数	参数1	weights	weights	-
	参数2	num_samples	num_samples	-
	参数3	replacement	replacement	-
	参数4	generator	-	指定额外的采样逻辑，MindSpore为全局随机采样

代码示例

import torch
from torch.utils.data import WeightedRandomSampler

torch.manual_seed(0)

class MyMapDataset(torch.utils.data.Dataset):
    def __init__(self):
        super(MyMapDataset).__init__()
        self.data = [i for i in range(1, 5)]
    def __getitem__(self, index):
        return self.data[index]
    def __len__(self):
        return len(self.data)

ds = MyMapDataset()
sampler = WeightedRandomSampler(weights=[0.1, 0.1, 0.9, 0.9], num_samples=4)
dataloader = torch.utils.data.DataLoader(ds, sampler=sampler)

for data in dataloader:
    print(data)
# Out:
# tensor([4])
# tensor([3])
# tensor([4])
# tensor([4])

import mindspore as ms
from mindspore.dataset import WeightedRandomSampler

ms.dataset.config.set_seed(4)

class MyMapDataset():
    def __init__(self):
        super(MyMapDataset).__init__()
        self.data = [i for i in range(1, 5)]
    def __getitem__(self, index):
        return self.data[index]
    def __len__(self):
        return len(self.data)

ds = MyMapDataset()
sampler = WeightedRandomSampler(weights=[0.1, 0.1, 0.9, 0.9], num_samples=4)
dataloader = ms.dataset.GeneratorDataset(ds, column_names=["data"], sampler=sampler)

for data in dataloader:
    print(data)
# Out:
# [Tensor(shape=[], dtype=Int64, value= 4)]
# [Tensor(shape=[], dtype=Int64, value= 3)]
# [Tensor(shape=[], dtype=Int64, value= 4)]
# [Tensor(shape=[], dtype=Int64, value= 4)]