数据加载器快速查找

import torch
import torchvision
# 将数据集合下载到指定目录下
train_dataset = torchvision.datasets.MNIST(root='./data',
                                           train=True,
                                           transform=torchvision.transforms.ToTensor(),
                                           download=True)
train_dataset
"""加载MNIST数据
Dataset MNIST
    Number of datapoints: 60000
    Root location: ./data
    Split: Train
    StandardTransform
Transform: ToTensor()
"""



import numpy as np
import matplotlib.pyplot as plt
%matplotlib inline
a_data, a_label = train_dataset[0]
print("\n打印第一个batch的第一个图像，对应标签数字为{}".format(a_label))
# 显示第一个 batch 中的第一个图像，六千张上的第一张
# 原始数据是归一化后的数据，因此这里需要反归一化
img = np.array(a_data+1)*127.5
img = np.reshape(img, [28, 28]).astype(np.uint8)
plt.imshow(img, 'gray')
plt.show()




from torch.utils.data import DataLoader
train_loader = DataLoader(dataset=train_dataset, batch_size=100, shuffle=True)
num_epochs = 1  # 迭代次数
for epoch in range(num_epochs):
    for i, (inputs, labels) in enumerate(train_loader):
        # 每 10 个批次展示一次
        if (i+1) % 10 == 0:
            print(
                f'Epoch: {epoch+1}/{num_epochs},Step {i+1}/{len(train_dataset)/100}| Inputs {inputs.shape} | Labels {labels.shape}')
            
      
    
    

import pandas as pd
df = pd.read_csv(
    "https://labfile.oss.aliyuncs.com/courses/2316/wine.csv", header=None)





from torch.utils.data import Dataset
class WineDataset(Dataset):
    # 建立一个数据集合继承  Dataset 即可
    def __init__(self):
        # I初始化数据
        # 以pandas的形式读入数据
        xy = pd.read_csv(
            "https://labfile.oss.aliyuncs.com/courses/2316/wine.csv", header=None)
        self.n_samples = xy.shape[0]

        # 将 pandas 类型的数据转换成 numpy 类型
        # size [n_samples, n_features]
        self.x_data = torch.from_numpy(xy.values[:, 1:])
        self.y_data = torch.from_numpy(
            xy.values[:, [0]])  # size [n_samples, 1]

    # 返回 dataset[index]
    def __getitem__(self, index):
        return self.x_data[index], self.y_data[index]

    # 返回数据长度
    def __len__(self):
        return self.n_samples
# 测试
# 创造 dataset
dataset = WineDataset()
dataset[0]




import math
# 传入加载器
train_loader = DataLoader(dataset=dataset, batch_size=4, shuffle=True)
# 分批训练
# 迭代次数
num_epochs = 2
total_samples = len(dataset)
# 批次
n_iterations = math.ceil(total_samples/4)
print("该数据集合共有{}条数据，被分成了{}个批次".format(total_samples, n_iterations))
for epoch in range(num_epochs):
    for i, (inputs, labels) in enumerate(train_loader):

        # 178 个样本, batch_size = 4, n_iters=178/4=44.5 -> 45 个批次
        if (i+1) % 5 == 0:
            print(
                f'Epoch: {epoch+1}/{num_epochs}, Step {i+1}/{n_iterations}| Inputs {inputs.shape} | Labels {labels.shape}')