我在做一个超大数据集的多分类,设备Ubuntu 22.04+i9 13900K+Nvidia 4090+64GB RAM,第一次的训练的训练集有700万张,训练成功。后面收集到更多数据集,数据增强后达到了1000万张。但第二次训练4个小时后,就被系统杀掉进程了,原因是Out of Memory。找了很久的原因,发现内存随着训练step的增加而线性增加,猜测是内存泄露,最后定位到了DataLoader的num_workers参数(只要num_workers=0就没有问题)。
Python(Pytorch)中的list转换成tensor时,会发生内存泄漏,要避免list的使用,可以通过使用np.array来代替list。
自定义DataLoader中的Dataset类,然后Dataset类中的list全部用np.array来代替。这样的话,DataLoader将np.array转换成Tensor的过程就不会发生内存泄露。
# 加载数据
train_data = datasets.ImageFolder(root=TRAIN_DIR_ARG, transform=transform)
valid_data = datasets.ImageFolder(root=VALIDATION_DIR, transform=transform)
test_data = datasets.ImageFolder(root=TEST_DIR, transform=transform)
train_loader = DataLoader(train_data, batch_size=BATCH_SIZE, shuffle=True, num_workers=8)
valid_loader = DataLoader(valid_data, batch_size=BATCH_SIZE, shuffle=False, num_workers=8)
test_loader = DataLoader(test_data, batch_size=BATCH_SIZE, shuffle=False, num_workers=8)
class CustomDataset(Dataset):
def __init__(self, data_dir, transform=None):
self.data_dir = data_dir
self.transform = transform
self.image_paths = []
self.labels = []
# 遍历数据目录并收集图像文件路径和对应的标签
classes = os.listdir(data_dir)
for i, class_name in enumerate(classes):
class_dir = os.path.join(data_dir, class_name)
if os.path.isdir(class_dir):
for image_name in os.listdir(class_dir):
image_path = os.path.join(class_dir, image_name)
self.image_paths.append(image_path)
self.labels.append(i)
def __len__(self):
return len(self.image_paths)
def __getitem__(self, idx):
image_path = self.image_paths[idx]
label = self.labels[idx]
# # 在需要时加载图像
image = Image.open(image_path)
if self.transform:
image = self.transform(image)
return image, label
train_data = CustomDataset(data_dir=TRAIN_DIR_ARG, transform=transform)
valid_data = CustomDataset(data_dir=VALIDATION_DIR, transform=transform)
test_data = CustomDataset(data_dir=TEST_DIR, transform=transform)
train_loader = DataLoader(train_data, batch_size=BATCH_SIZE, shuffle=True, num_workers=18)
valid_loader = DataLoader(valid_data, batch_size=BATCH_SIZE, shuffle=False, num_workers=8)
test_loader = DataLoader(test_data, batch_size=BATCH_SIZE, shuffle=False, num_workers=8, pin_memory=False)
class CustomDataset(Dataset):
def __init__(self, data_dir, transform=None):
self.data_dir = data_dir
self.transform = transform
self.image_paths = [] # 使用Python列表
self.labels = [] # 使用Python列表
# 遍历数据目录并收集图像文件路径和对应的标签
classes = os.listdir(data_dir)
for i, class_name in enumerate(classes):
class_dir = os.path.join(data_dir, class_name)
if os.path.isdir(class_dir):
for image_name in os.listdir(class_dir):
image_path = os.path.join(class_dir, image_name)
self.image_paths.append(image_path) # 添加到Python列表
self.labels.append(i) # 添加到Python列表
# 转换为NumPy数组,这里就是解决内存泄露的关键代码
self.image_paths = np.array(self.image_paths)
self.labels = np.array(self.labels)
def __len__(self):
return len(self.image_paths)
def __getitem__(self, idx):
image_path = self.image_paths[idx]
label = self.labels[idx]
# 在需要时加载图像
image = Image.open(image_path)
if self.transform:
image = self.transform(image)
# 将图像数据转换为NumPy数组
image = np.array(image)
return image, label
train_data = CustomDataset(data_dir=TRAIN_DIR_ARG, transform=transform)
valid_data = CustomDataset(data_dir=VALIDATION_DIR, transform=transform)
test_data = CustomDataset(data_dir=TEST_DIR, transform=transform)
train_loader = DataLoader(train_data, batch_size=BATCH_SIZE, shuffle=True, num_workers=18)
valid_loader = DataLoader(valid_data, batch_size=BATCH_SIZE, shuffle=False, num_workers=8)
test_loader = DataLoader(test_data, batch_size=BATCH_SIZE, shuffle=False, num_workers=8, pin_memory=False)