Memory consumption increases while training

The memory consumption increases while training,it ran a few times to the third epoch reported MemoryError.And my train dataset is stored in the TrainData folder by .csv type.The inputs of the net is a pair of samples and the corresponding label.
the error :

my as follows:
import torch
import numpy as np
from import Dataset
import pandas

class Mydataset(Dataset):
def init(self, csvlist):
self.csvlist = csvlist

def __len__(self):
    return len(self.csvlist)

def __getitem__(self,idx):
    csvfile = self.csvlist[idx]
    inputs = np.loadtxt(open(csvfile,'rb'),delimiter=",",skiprows=0) 
    inputs = torch.Tensor(inputs)
    input1 = inputs[:35937].view(1,33,33,33)
    input2 = inputs[35937:-1].view(1,33,33,33)
    label = torch.zeros(1)
    label[0] = inputs[-1]
    return input1, input2, label

this is a windows pytorch bug. Follow this thread for updates: