这篇文章主要介绍了Pytorch中TensorDataset与DataLoader怎么联合使用的相关知识,内容详细易懂,操作简单快捷,具有一定借鉴价值,相信大家阅读完这篇Pytorch中TensorDataset与DataLoader怎么联合使用文章都会有所收获,下面我们一起来看看吧。
Pytorch中TensorDataset,DataLoader的联合使用
首先从字面意义上来理解TensorDataset和DataLoader,TensorDataset是个只用来存放tensor(张量)的数据集,而DataLoader是一个数据加载器,一般用到DataLoader的时候就说明需要遍历和操作数据了。
TensorDataset(tensor1,tensor2)的功能就是形成数据tensor1和标签tensor2的对应,也就是说tensor1中是数据,而tensor2是tensor1所对应的标签。
来个小例子
from torch.utils.data import TensorDataset,DataLoader
import torch
a = torch.tensor([[1, 2, 3],
[4, 5, 6],
[7, 8, 9],
[1, 2, 3],
[4, 5, 6],
[7, 8, 9],
[1, 2, 3],
[4, 5, 6],
[7, 8, 9],
[1, 2, 3],
[4, 5, 6],
[7, 8, 9]])
b = torch.tensor([44, 55, 66, 44, 55, 66, 44, 55, 66, 44, 55, 66])
train_ids = TensorDataset(a,b)
# 切片输出
print(train_ids[0:4]) # 第0,1,2,3行
# 循环取数据
for x_train,y_label in train_ids:
print(x_train,y_label)
下面是对应的输出:
(tensor([[1, 2, 3],
[4, 5, 6],
[7, 8, 9],
[1, 2, 3]]), tensor([44, 55, 66, 44]))
===============================================
tensor([1, 2, 3]) tensor(44)
tensor([4, 5, 6]) tensor(55)
tensor([7, 8, 9]) tensor(66)
tensor([1, 2, 3]) tensor(44)
tensor([4, 5, 6]) tensor(55)
tensor([7, 8, 9]) tensor(66)
tensor([1, 2, 3]) tensor(44)
tensor([4, 5, 6]) tensor(55)
tensor([7, 8, 9]) tensor(66)
tensor([1, 2, 3]) tensor(44)
tensor([4, 5, 6]) tensor(55)
tensor([7, 8, 9]) tensor(66)
从输出结果我们就可以很好的理解,tensor型数据和tensor型标签的对应了,这就是TensorDataset的基本应用。
接下来我们把构造好的TensorDataset封装到DataLoader来操作里面的数据:
# 参数说明,dataset=train_ids表示需要封装的数据集,batch_size表示一次取几个
# shuffle表示乱序取数据,设为False表示顺序取数据,True表示乱序取数据
train_loader = DataLoader(dataset=train_ids,batch_size=4,shuffle=False)
# 注意enumerate返回值有两个,一个是序号,一个是数据(包含训练数据和标签)
for i,data in enumerate(train_loader,1):
train_data, label = data
print(' batch:{0} train_data:{1} label: {2}'.format(i+1, train_data, label))
下面是对应的输出:
batch:1 x_data:tensor([[1, 2, 3],
[4, 5, 6],
[7, 8, 9],
[1, 2, 3]]) label: tensor([44, 55, 66, 44])
batch:2 x_data:tensor([[4, 5, 6],
[7, 8, 9],
[1, 2, 3],
[4, 5, 6]]) label: tensor([55, 66, 44, 55])
batch:3 x_data:tensor([[7, 8, 9],
[1, 2, 3],
[4, 5, 6],
[7, 8, 9]]) label: tensor([66, 44, 55, 66])
至此,TensorDataset和DataLoader的联合使用就介绍完了。
我们再看一下这两种方法的源码:
class TensorDataset(Dataset[Tuple[Tensor, ...]]):
r"""Dataset wrapping tensors.
Each sample will be retrieved by indexing tensors along the first dimension.
Arguments:
*tensors (Tensor): tensors that have the same size of the first dimension.
"""
tensors: Tuple[Tensor, ...]
def __init__(self, *tensors: Tensor) -> None:
assert all(tensors[0].size(0) == tensor.size(0) for tensor in tensors)
self.tensors = tensors
def __getitem__(self, index):
return tuple(tensor[index] for tensor in self.tensors)
def __len__(self):
return self.tensors[0].size(0)
# 由于此类内容过多,故仅列举了与本文相关的参数,其余参数可以自行去查看源码
class DataLoader(Generic[T_co]):
r"""
Data loader. Combines a dataset and a sampler, and provides an iterable over
the given dataset.
The :class:`~torch.utils.data.DataLoader` supports both map-style and
iterable-style datasets with single- or multi-process loading, customizing
loading order and optional automatic batching (collation) and memory pinning.
See :py:mod:`torch.utils.data` documentation page for more details.
Arguments:
dataset (Dataset): dataset from which to load the data.
batch_size (int, optional): how many samples per batch to load
(default: ``1``).
shuffle (bool, optional): set to ``True`` to have the data reshuffled
at every epoch (default: ``False``).
"""
dataset: Dataset[T_co]
batch_size: Optional[int]
def __init__(self, dataset: Dataset[T_co], batch_size: Optional[int] = 1,
shuffle: bool = False):
self.dataset = dataset
self.batch_size = batch_size
Pytorch的DataLoader和Dataset以及TensorDataset的源码分析
1.为什么要用DataLoader和Dataset
要对大量数据进行加载和处理时因为可能会出现内存不够用的情况,这时候就需要用到数据集类Dataset或TensorDataset和数据集加载类DataLoader了。
使用这些类后可以将原本的数据分成小块,在需要使用的时候再一部分一本分读进内存中,而不是一开始就将所有数据读进内存中。
2.Dateset的使用
pytorch中的torch.utils.data.Dataset是表示数据集的抽象类,但它一般不直接使用,而是通过自定义一个数据集来使用。
来自定义数据集应该继承Dataset并应该有实现返回数据集尺寸的__len__方法和用来获取索引数据的__getitem__方法。
Dataset类的源码如下:
class Dataset(object):
r"""An abstract class representing a :class:`Dataset`.
All datasets that represent a map from keys to data samples should subclass
it. All subclasses should overwrite :meth:`__getitem__`, supporting fetching a
data sample for a given key. Subclasses could also optionally overwrite
:meth:`__len__`, which is expected to return the size of the dataset by many
:class:`~torch.utils.data.Sampler` implementations and the default options
of :class:`~torch.utils.data.DataLoader`.
.. note::
:class:`~torch.utils.data.DataLoader` by default constructs a index
sampler that yields integral indices. To make it work with a map-style
dataset with non-integral indices/keys, a custom sampler must be provided.
"""
def __getitem__(self, index):
raise NotImplementedError
def __add__(self, other):
return ConcatDataset([self, other])
# No `def __len__(self)` default?
# See NOTE [ Lack of Default `__len__` in Python Abstract Base Classes ]
# in pytorch/torch/utils/data/sampler.py
可以看到Dataset类中没有__len__方法,虽然有__getitem__方法,但是并没有实现啥有用的功能。
所以要写一个Dataset类的子类来实现其应有的功能。
自定义类的实现举例:
import torch
from torch.utils.data import Dataset, DataLoader, TensorDataset
from torch.autograd import Variable
import numpy as np
import pandas as pd
value_df = pd.read_csv('data1.csv')
value_array = np.array(value_df)
print("value_array.shape =", value_array.shape) # (73700, 300)
value_size = value_array.shape[0] # 73700
train_size = int(0.7*value_size)
train_array = val_array[:train_size]
train_label_array = val_array[60:train_size+60]
class DealDataset(Dataset):
"""
下载数据、初始化数据,都可以在这里完成
"""
def __init__(self, *arrays):
assert all(arrays[0].shape[0] == array.shape[0] for array in arrays)
self.arrays = arrays
def __getitem__(self, index):
return tuple(array[index] for array in self.arrays)
def __len__(self):
return self.arrays[0].shape[0]
# 实例化这个类,然后我们就得到了Dataset类型的数据,记下来就将这个类传给DataLoader,就可以了。
train_dataset = DealDataset(train_array, train_label_array)
train_loader2 = DataLoader(dataset=train_dataset,
batch_size=32,
shuffle=True)
for epoch in range(2):
for i, data in enumerate(train_loader2):
# 将数据从 train_loader 中读出来,一次读取的样本数是32个
inputs, labels = data
# 将这些数据转换成Variable类型
inputs, labels = Variable(inputs), Variable(labels)
# 接下来就是跑模型的环节了,我们这里使用print来代替
print("epoch:", epoch, "的第", i, "个inputs", inputs.data.size(), "labels", labels.data.size())
结果:
epoch: 0 的第 0 个inputs torch.Size([32, 300]) labels torch.Size([32, 300])
epoch: 0 的第 1 个inputs torch.Size([32, 300]) labels torch.Size([32, 300])
epoch: 0 的第 2 个inputs torch.Size([32, 300]) labels torch.Size([32, 300])
epoch: 0 的第 3 个inputs torch.Size([32, 300]) labels torch.Size([32, 300])
epoch: 0 的第 4 个inputs torch.Size([32, 300]) labels torch.Size([32, 300])
epoch: 0 的第 5 个inputs torch.Size([32, 300]) labels torch.Size([32, 300])
...
3.TensorDataset的使用
TensorDataset是可以直接使用的数据集类,它的源码如下:
class TensorDataset(Dataset):
r"""Dataset wrapping tensors.
Each sample will be retrieved by indexing tensors along the first dimension.
Arguments:
*tensors (Tensor): tensors that have the same size of the first dimension.
"""
def __init__(self, *tensors):
assert all(tensors[0].size(0) == tensor.size(0) for tensor in tensors)
self.tensors = tensors
def __getitem__(self, index):
return tuple(tensor[index] for tensor in self.tensors)
def __len__(self):
return self.tensors[0].size(0)
可以看到TensorDataset类是Dataset类的子类,且拥有返回数据集尺寸的__len__方法和用来获取索引数据的__getitem__方法,所以可以直接使用。
它的结构跟上面自定义的子类的结构是一样的,惟一的不同是TensorDataset已经规定了传入的数据必须是torch.Tensor类型的,而自定义子类可以自由设定。
使用举例:
import torch
from torch.utils.data import Dataset, DataLoader, TensorDataset
from torch.autograd import Variable
import numpy as np
import pandas as pd
value_df = pd.read_csv('data1.csv')
value_array = np.array(value_df)
print("value_array.shape =", value_array.shape) # (73700, 300)
value_size = value_array.shape[0] # 73700
train_size = int(0.7*value_size)
train_array = val_array[:train_size]
train_tensor = torch.tensor(train_array, dtype=torch.float32).to(device)
train_label_array = val_array[60:train_size+60]
train_labels_tensor = torch.tensor(train_label_array,dtype=torch.float32).to(device)
train_dataset = TensorDataset(train_tensor, train_labels_tensor)
train_loader = DataLoader(dataset=train_dataset,
batch_size=100,
shuffle=False,
num_workers=0)
for epoch in range(2):
for i, data in enumerate(train_loader):
inputs, labels = data
inputs, labels = Variable(inputs), Variable(labels)
print(epoch, i, "inputs", inputs.data.size(), "labels", labels.data.size())
结果:
0 0 inputs torch.Size([100, 300]) labels torch.Size([100, 300])
0 1 inputs torch.Size([100, 300]) labels torch.Size([100, 300])
0 2 inputs torch.Size([100, 300]) labels torch.Size([100, 300])
0 3 inputs torch.Size([100, 300]) labels torch.Size([100, 300])
0 4 inputs torch.Size([100, 300]) labels torch.Size([100, 300])
0 5 inputs torch.Size([100, 300]) labels torch.Size([100, 300])
0 6 inputs torch.Size([100, 300]) labels torch.Size([100, 300])
0 7 inputs torch.Size([100, 300]) labels torch.Size([100, 300])
0 8 inputs torch.Size([100, 300]) labels torch.Size([100, 300])
0 9 inputs torch.Size([100, 300]) labels torch.Size([100, 300])
0 10 inputs torch.Size([100, 300]) labels torch.Size([100, 300])
...
以上就是Pytorch中TensorDataset与DataLoader怎么联合使用的详细内容,更多关于Pytorch中TensorDataset与DataLoader怎么联合使用的资料请关注九品源码其它相关文章!