Pytorch专题实战——批训练数据(DataLoader)

1.计算流程

# Implement a custom Dataset:
# inherit Dataset
# implement __init__ , __getitem__ , and __len__

数据链接
密码:yrit

2.Pytorch构造批处理数据

2.1.导入必要模块

import torch
from torch.utils.data import Dataset, DataLoader 
import numpy as np
import math

2.2.定义数据类

class wineDataset(Dataset):
    def __init__(self):
        xy = np.loadtxt('./wine.csv', delimiter=',',dtype=np.float32,skiprows=1)  #读取数据并跳过第一行
        self.n_samples = xy.shape[0]    #计算样本数
        self.x_data = torch.from_numpy(xy[:,1:])     #数据
        self.y_data = torch.from_numpy(xy[:,[0]])    #标签
    
    def __getitem__(self, index):    #取数据函数
        return self.x_data[index], self.y_data[index]
    
    def __len__(self):       #计算样本大小函数
        return self.n_samples  

2.3.定义DataLoader

dataset = wineDataset()   #实例化

train_loader = DataLoader(dataset=dataset,    #一个迭代4个数据并且打乱
                         batch_size=4,
                         shuffle=True,
                         num_workers=2)

2.4.打印效果

num_epochs = 2
total_samples = len(dataset)
n_iterations = math.ceil(total_samples/4)
print(total_samples, n_iterations)

for epoch in range(num_epochs):
    for i, (inputs, labels) in enumerate(train_loader):
        if (i+1)%5 == 0:
            print(f'Epoch:{epoch+1}/{num_epochs}, Step {i+1}/{n_iterations} | Inputs {inputs.shape} | Labels{labels.shape}')