我對PyTorch dataloader裡的shuffle=True的理解
對shuffle=True的理解:
之前不瞭解shuffle的實際效果,假設有數據a,b,c,d,不知道batch_size=2後打亂,具體是如下哪一種情況:
1.先按順序取batch,對batch內打亂,即先取a,b,a,b進行打亂;
2.先打亂,再取batch。
證明是第二種
shuffle (bool, optional): set to ``True`` to have the data reshuffled at every epoch (default: ``False``). if shuffle: sampler = RandomSampler(dataset) #此時得到的是索引
補充:簡單測試一下pytorch dataloader裡的shuffle=True是如何工作的
看代碼吧~
import sys import torch import random import argparse import numpy as np import pandas as pd import torch.nn as nn from torch.nn import functional as F from torch.optim import lr_scheduler from torchvision import datasets, transforms from torch.utils.data import TensorDataset, DataLoader, Dataset class DealDataset(Dataset): def __init__(self): xy = np.loadtxt(open('./iris.csv','rb'), delimiter=',', dtype=np.float32) #data = pd.read_csv("iris.csv",header=None) #xy = data.values self.x_data = torch.from_numpy(xy[:, 0:-1]) self.y_data = torch.from_numpy(xy[:, [-1]]) self.len = xy.shape[0] def __getitem__(self, index): return self.x_data[index], self.y_data[index] def __len__(self): return self.len dealDataset = DealDataset() train_loader2 = DataLoader(dataset=dealDataset, batch_size=2, shuffle=True) #print(dealDataset.x_data) for i, data in enumerate(train_loader2): inputs, labels = data #inputs, labels = Variable(inputs), Variable(labels) print(inputs) #print("epoch:", epoch, "的第" , i, "個inputs", inputs.data.size(), "labels", labels.data.size())
簡易數據集
shuffle之後的結果,每次都是隨機打亂,然後分成大小為n的若幹個mini-batch.
以上為個人經驗,希望能給大傢一個參考,也希望大傢多多支持WalkonNet。
推薦閱讀:
- pytorch DataLoader的num_workers參數與設置大小詳解
- Pytorch DataLoader shuffle驗證方式
- Pytorch實現圖像識別之數字識別(附詳細註釋)
- 解決pytorch讀取自制數據集出現過的問題
- pytorch鎖死在dataloader(訓練時卡死)