机器之心转载

来源:知乎

作者:张皓


本文代码基于 PyTorch 1.0 版本,需要用到以下包

import collectionsimport osimport shutilimport tqdmimport numpy as npimport PIL.Imageimport torchimport torchvision

基础配置

检查 PyTorch 版本

torch.__version__               # PyTorch versiontorch.version.cuda              # Corresponding CUDA versiontorch.backends.cudnn.version()  # Corresponding cuDNN versiontorch.cuda.get_device_name(0)   # GPU type

更新 PyTorch

PyTorch 将被安装在 anaconda3/lib/python3.7/site-packages/torch/目录下。

conda update pytorch torchvision -c pytorch

固定随机种子

torch.manual_seed(0)torch.cuda.manual_seed_all(0)

指定程序运行在特定 GPU 卡上

在命令行指定环境变量

CUDA_VISIBLE_DEVICES=0,1 python train.py

或在代码中指定

os.environ[ CUDA_VISIBLE_DEVICES ] =  0,1

判断是否有 CUDA 支持

torch.cuda.is_available()

设置为 cuDNN benchmark 模式

Benchmark 模式会提升计算速度,但是由于计算中有随机性,每次网络前馈结果略有差异。

torch.backends.cudnn.benchmark = True

如果想要避免这种结果波动,设置

torch.backends.cudnn.deterministic = True

清除 GPU 存储

有时 Control-C 中止运行后 GPU 存储没有及时释放,需要手动清空。在 PyTorch 内部可以

torch.cuda.empty_cache()

或在命令行可以先使用 ps 找到程序的 PID,再使用 kill 结束该进程

ps aux | grep pythonkill -9 [pid]

或者直接重置没有被清空的 GPU

nvidia-smi --gpu-reset -i [gpu_id]

张量处理

张量基本信息

tensor.type()   # Data typetensor.size()   # Shape of the tensor. It is a subclass of Python tupletensor.dim()    # Number of dimensions.

数据类型转换

# Set default tensor type. Float in PyTorch is much faster than double.torch.set_default_tensor_type(torch.FloatTensor)# Type convertions.tensor = tensor.cuda()tensor = tensor.cpu()tensor = tensor.float()tensor = tensor.long()

torch.Tensor 与 np.ndarray 转换

# torch.Tensor -> np.ndarray.ndarray = tensor.cpu().numpy()# np.ndarray -> torch.Tensor.tensor = torch.from_numpy(ndarray).float()tensor = torch.from_numpy(ndarray.copy()).float()  # If ndarray has negative stride

torch.Tensor 与 PIL.Image 转换

PyTorch 中的张量默认采用 N×D×H×W 的顺序,并且数据范围在 [0, 1],需要进行转置和规范化。

# torch.Tensor -> PIL.Image.image = PIL.Image.fromarray(torch.clamp(tensor * 255, min=0, max=255    ).byte().permute(1, 2, 0).cpu().numpy())image = torchvision.transforms.functional.to_pil_image(tensor)  # Equivalently way# PIL.Image -> torch.Tensor.tensor = torch.from_numpy(np.asarray(PIL.Image.open(path))    ).permute(2, 0, 1).float() / 255tensor = torchvision.transforms.functional.to_tensor(PIL.Image.open(path))  # Equivalently way###np.ndarray 与 PIL.Image 转换# np.ndarray -> PIL.Image.image = PIL.Image.fromarray(ndarray.astypde(np.uint8))# PIL.Image -> np.ndarray.ndarray = np.asarray(PIL.Image.open(path))

从只包含一个元素的张量中提取值

这在训练时统计 loss 的变化过程中特别有用。否则这将累积计算图,使 GPU 存储占用量越来越大。

value = tensor.item()

张量形变

张量形变常常需要用于将卷积层特征输入全连接层的情形。相比 torch.view,torch.reshape 可以自动处理输入张量不连续的情况。

tensor = torch.reshape(tensor, shape)

打乱顺序

tensor = tensor[torch.randperm(tensor.size(0))]  # Shuffle the first dimension

水平翻转

PyTorch 不支持 tensor[::-1] 这样的负步长操作,水平翻转可以用张量索引实现。

Assume tensor has shape NDH*W.tensor = tensor[:, :, :, torch.arange(tensor.size(3) - 1, -1, -1).long()]

复制张量

有三种复制的方式,对应不同的需求。

# Operation                 |  New/Shared memory | Still in computation graph |tensor.clone()            # |        New         |          Yes               |tensor.detach()           # |      Shared        |          No                |tensor.detach.clone()()   # |        New         |          No                |

拼接张量

注意 torch.cat 和 torch.stack 的区别在于 torch.cat 沿着给定的维度拼接,而 torch.stack 会新增一维。例如当参数是 3 个 10×5 的张量,torch.cat 的结果是 30×5 的张量,而 torch.stack 的结果是 3×10×5 的张量。

tensor = torch.cat(list_of_tensors, dim=0)tensor = torch.stack(list_of_tensors, dim=0)

将整数标记转换成独热(one-hot)编码

PyTorch 中的标记默认从 0 开始。

N = tensor.size(0)one_hot = torch.zeros(N, num_classes).long()one_hot.scatter_(dim=1, index=torch.unsqueeze(tensor, dim=1), src=torch.ones(N, num_classes).long())

得到非零/零元素

torch.nonzero(tensor)               # Index of non-zero elementstorch.nonzero(tensor == 0)          # Index of zero elementstorch.nonzero(tensor).size(0)       # Number of non-zero elementstorch.nonzero(tensor == 0).size(0)  # Number of zero elements

张量扩展

# Expand tensor of shape 64*512 to shape 64*512*7*7.torch.reshape(tensor, (64, 512, 1, 1)).expand(64, 512, 7, 7)

矩阵乘法

Matrix multiplication: (mn) (np) -> (mp).

result = torch.mm(tensor1, tensor2)

Batch matrix multiplication: (bmn) (bnp) -> (bm*p).

result = torch.bmm(tensor1, tensor2)

Element-wise multiplication.

result = tensor1 * tensor2

计算两组数据之间的两两欧式距离

# X1 is of shape m*d.X1 = torch.unsqueeze(X1, dim=1).expand(m, n, d)# X2 is of shape n*d.X2 = torch.unsqueeze(X2, dim=0).expand(m, n, d)# dist is of shape m*n, where dist[i][j] = sqrt(|X1[i, :] - X[j, :]|^2)dist = torch.sqrt(torch.sum((X1 - X2) ** 2, dim=2))

模型定义

卷积层

最常用的卷积层配置是

conv = torch.nn.Conv2d(in_channels, out_channels, kernel_size=3, stride=1, padding=1, bias=True)conv = torch.nn.Conv2d(in_channels, out_channels, kernel_size=1, stride=1, padding=0, bias=True)

如果卷积层配置比较复杂,不方便计算输出大小时,可以利用如下可视化工具辅助

链接:https://ezyang.github.io/convolution-visualizer/index.html

###0GAP(Global average pooling)层gap = torch.nn.AdaptiveAvgPool2d(output_size=1)###双线性汇合(bilinear pooling)X = torch.reshape(N, D, H * W)                        # Assume X has shape N*D*H*WX = torch.bmm(X, torch.transpose(X, 1, 2)) / (H * W)  # Bilinear poolingassert X.size() == (N, D, D)X = torch.reshape(X, (N, D * D))X = torch.sign(X) * torch.sqrt(torch.abs(X) + 1e-5)   # Signed-sqrt normalizationX = torch.nn.functional.normalize(X)                  # L2 normalization

多###卡同步 BN(Batch normalization)

当使用 torch.nn.DataParallel 将代码运行在多张 GPU 卡上时,PyTorch 的 BN 层默认操作是各卡上数据独立地计算均值和标准差,同步 BN 使用所有卡上的数据一起计算 BN 层的均值和标准差,缓解了当批量大小(batch size)比较小时对均值和标准差估计不准的情况,是在目标检测等任务中一个有效的提升性能的技巧。

链接:https://github.com/vacancy/Synchronized-BatchNorm-PyTorch

类似 BN 滑动平均

如果要实现类似 BN 滑动平均的操作,在 forward 函数中要使用原地(inplace)操作给滑动平均赋值。

class BN(torch.nn.Module)    def __init__(self):        ...        self.register_buffer( running_mean , torch.zeros(num_features))    def forward(self, X):        ...        self.running_mean += momentum * (current - self.running_mean)

计算模型整体参数量

num_parameters = sum(torch.numel(parameter) for parameter in model.parameters())

类似 Keras 的 model.summary() 输出模型信息

链接:https://github.com/sksq96/pytorch-summary

模型权值初始化

注意 model.modules() 和 model.children() 的区别:model.modules() 会迭代地遍历模型的所有子层,而 model.children() 只会遍历模型下的一层。

# Common practise for initialization.for   in model.modules():    if isinstance( , torch.nn.Conv2d):        torch.nn.init.kaiming_normal_( .weight, mode= fan_out ,                                      nonlinearity= relu )        if  .bias is not None:            torch.nn.init.constant_( .bias, val=0.0)    elif isinstance( , torch.nn.BatchNorm2d):        torch.nn.init.constant_( .weight, val=1.0)        torch.nn.init.constant_( .bias, val=0.0)    elif isinstance( , torch.nn.Linear):        torch.nn.init.xavier_normal_( .weight)        if  .bias is not None:            torch.nn.init.constant_( .bias, val=0.0)# Initialization with given tensor. .weight = torch.nn.Parameter(tensor)

部分层使用预训练模型

注意如果保存的模型是 torch.nn.DataParallel,则当前的模型也需要是

model.load_state_dict(torch.load( model,pth ), strict=False)

将在 GPU 保存的模型加载到 CPU

model.load_state_dict(torch.load( model,pth , map_location= cpu ))

数据准备、特征提取与微调

得到视频数据基本信息

import cv2video = cv2.VideoCapture(mp4_path)height = int(video.get(cv2.CAP_PROP_ _HEIGHT))width = int(video.get(cv2.CAP_PROP_ _WIDTH))num_ s = int(video.get(cv2.CAP_PROP_ _COUNT))fps = int(video.get(cv2.CAP_PROP_FPS))video.release()

TSN 每段(segment)采样一帧视频

K = self._num_segmentsif is_train:    if num_ s > K:        # Random index for each segment.         _indices = torch.randint(            high=num_ s // K, size=(K,), dtype=torch.long)         _indices += num_ s // K * torch.arange(K)    else:         _indices = torch.randint(            high=num_ s, size=(K - num_ s,), dtype=torch.long)         _indices = torch.sort(torch.cat((            torch.arange(num_ s),  _indices)))[0]else:    if num_ s > K:        # Middle index for each segment.         _indices = num_ s / K // 2         _indices += num_ s // K * torch.arange(K)    else:         _indices = torch.sort(torch.cat((                                          torch.arange(num_ s), torch.arange(K - num_ s))))[0]assert  _indices.size() == (K,)return [ _indices[i] for i in range(K)]

提取 ImageNet 预训练模型某层的卷积特征

# VGG-16 relu5-3 feature.model = torchvision.models.vgg16(pretrained=True).features[:-1]# VGG-16 pool5 feature.model = torchvision.models.vgg16(pretrained=True).features# VGG-16 fc7 feature.model = torchvision.models.vgg16(pretrained=True)model.classifier = torch.nn.Sequential(*list(model.classifier.children())[:-3])# ResNet GAP feature.model = torchvision.models.resnet18(pretrained=True)model = torch.nn.Sequential(collections.OrderedDict(    list(model.named_children())[:-1]))with torch.no_grad():    model.eval()    conv_representation = model(image)

提取 ImageNet 预训练模型多层的卷积特征

class FeatureExtractor(torch.nn.Module):    """Helper class to extract several convolution features from the given    pre-trained model.    Attributes:        _model, torch.nn.Module.        _ s_to_extract, list<str> or set<str>    Example:        >>> model = torchvision.models.resnet152(pretrained=True)        >>> model = torch.nn.Sequential(collections.OrderedDict(                list(model.named_children())[:-1]))        >>> conv_representation = FeatureExtractor(                pretrained_model=model,                 s_to_extract={  1 ,   2 ,   3 ,   4 })(image)    """    def __init__(self, pretrained_model,  s_to_extract):        torch.nn.Module.__init__(self)        self._model = pretrained_model        self._model.eval()        self._ s_to_extract = set( s_to_extract)    def forward(self, x):        with torch.no_grad():            conv_representation = []            for name,   in self._model.named_children():                x =  (x)                if name in self._ s_to_extract:                    conv_representation.append(x)            return conv_representation

其他预训练模型

链接:https://github.com/Cadene/pretrained-models.pytorch

微调全连接层

model = torchvision.models.resnet18(pretrained=True)for param in model.parameters():    param.requires_grad = Falsemodel.fc = nn.Linear(512, 100)  # Replace the last fc  optimizer = torch.optim.SGD(model.fc.parameters(), lr=1e-2, momentum=0.9, weight_decay=1e-4)

以较大学习率微调全连接层,较小学习率微调卷积层

model = torchvision.models.resnet18(pretrained=True)finetuned_parameters = list(map(id, model.fc.parameters()))conv_parameters = (p for p in model.parameters() if id(p) not in finetuned_parameters)parameters = [{ params : conv_parameters,  lr : 1e-3},               { params : model.fc.parameters()}]optimizer = torch.optim.SGD(parameters, lr=1e-2, momentum=0.9, weight_decay=1e-4)

模型训练

常用训练和验证数据预处理

其中 ToTensor 操作会将 PIL.Image 或形状为 H×W×D,数值范围为 [0, 255] 的 np.ndarray 转换为形状为 D×H×W,数值范围为 [0.0, 1.0] 的 torch.Tensor。

train_transform = torchvision.transforms.Compose([    torchvision.transforms.RandomResizedCrop(size=224,                                             scale=(0.08, 1.0)),    torchvision.transforms.RandomHorizontalFlip(),    torchvision.transforms.ToTensor(),    torchvision.transforms.Normalize(mean=(0.485, 0.456, 0.406),                                     std=(0.229, 0.224, 0.225)), ]) val_transform = torchvision.transforms.Compose([    torchvision.transforms.Resize(224),    torchvision.transforms.CenterCrop(224),    torchvision.transforms.ToTensor(),    torchvision.transforms.Normalize(mean=(0.485, 0.456, 0.406),                                     std=(0.229, 0.224, 0.225)),])

训练基本代码框架

for t in epoch(80):    for images, labels in tqdm.tqdm(train_loader, desc= Epoch %3d  % (t + 1)):        images, labels = images.cuda(), labels.cuda()        scores = model(images)        loss = loss_function(scores, labels)        optimizer.zero_grad()        loss.backward()        optimizer.step()

标记平滑(label smoothing)

for images, labels in train_loader:    images, labels = images.cuda(), labels.cuda()    N = labels.size(0)    # C is the number of classes.    smoothed_labels = torch.full(size=(N, C), fill_value=0.1 / (C - 1)).cuda()    smoothed_labels.scatter_(dim=1, index=torch.unsqueeze(labels, dim=1), value=0.9)    score = model(images)    log_prob = torch.nn.functional.log_softmax(score, dim=1)    loss = -torch.sum(log_prob * smoothed_labels) / N    optimizer.zero_grad()    loss.backward()    optimizer.step()

Mixup

beta_distribution = torch.distributions.beta.Beta(alpha, alpha)for images, labels in train_loader:    images, labels = images.cuda(), labels.cuda()    # Mixup images.    lambda_ = beta_distribution.sample([]).item()    index = torch.randperm(images.size(0)).cuda()    mixed_images = lambda_ * images + (1 - lambda_) * images[index, :]    # Mixup loss.        scores = model(mixed_images)    loss = (lambda_ * loss_function(scores, labels)             + (1 - lambda_) * loss_function(scores, labels[index]))    optimizer.zero_grad()    loss.backward()    optimizer.step()

L1 正则化

l1_regularization = torch.nn.L1Loss(reduction= sum )loss = ...  # Standard cross-entropy lossfor param in model.parameters():    loss += torch.sum(torch.abs(param))loss.backward()

不对偏置项进行 L2 正则化/权值衰减(weight decay)

bias_list = (param for name, param in model.named_parameters() if name[-4:] ==  bias )others_list = (param for name, param in model.named_parameters() if name[-4:] !=  bias )parameters = [{ parameters : bias_list,  weight_decay : 0},                              { parameters : others_list}]optimizer = torch.optim.SGD(parameters, lr=1e-2, momentum=0.9, weight_decay=1e-4)

梯度裁剪(gradient clipping)

torch.nn.utils.clip_grad_norm_(model.parameters(), max_norm=20)

计算 Softmax 输出的准确率

score = model(images)prediction = torch.argmax(score, dim=1)num_correct = torch.sum(prediction == labels).item()accuruacy = num_correct / labels.size(0)

可视化模型前馈的计算图

链接:https://github.com/szagoruyko/pytorchviz

可视化学习曲线

有 Facebook 自己开发的 Visdom 和 Tensorboard 两个选择。

https://github.com/facebookresearch/visdom

https://github.com/lanpa/tensorboardX

# Example using Visdom.vis = visdom.Visdom(env= Learning curve , use_incoming_socket=False)assert self._visdom.check_connection()self._visdom.close()options = collections.namedtuple( Options , [ loss ,  acc ,  lr ])(    loss={ xlabel :  Epoch ,  ylabel :  Loss ,  showlegend : True},    acc={ xlabel :  Epoch ,  ylabel :  Accuracy ,  showlegend : True},    lr={ xlabel :  Epoch ,  ylabel :  Learning rate ,  showlegend : True})for t in epoch(80):    tran(...)    val(...)    vis.line(X=torch.Tensor([t + 1]), Y=torch.Tensor([train_loss]),             name= train , win= Loss , update= append , opts=options.loss)    vis.line(X=torch.Tensor([t + 1]), Y=torch.Tensor([val_loss]),             name= val , win= Loss , update= append , opts=options.loss)    vis.line(X=torch.Tensor([t + 1]), Y=torch.Tensor([train_acc]),             name= train , win= Accuracy , update= append , opts=options.acc)    vis.line(X=torch.Tensor([t + 1]), Y=torch.Tensor([val_acc]),             name= val , win= Accuracy , update= append , opts=options.acc)    vis.line(X=torch.Tensor([t + 1]), Y=torch.Tensor([lr]),             win= Learning rate , update= append , opts=options.lr)

得到当前学习率

# If there is one global learning rate (which is the common case).lr = next(iter(optimizer.param_groups))[ lr ]# If there are multiple learning rates for different  s.all_lr = []for param_group in optimizer.param_groups:    all_lr.append(param_group[ lr ])

学习率衰减

# Reduce learning rate when validation accuarcy plateau.scheduler = torch.optim.lr_scheduler.ReduceLROnPlateau(optimizer, mode= max , patience=5, verbose=True)for t in range(0, 80):    train(...); val(...)    scheduler.step(val_acc)# Cosine annealing learning rate.scheduler = torch.optim.lr_scheduler.CosineAnnealingLR(optimizer, T_max=80)# Reduce learning rate by 10 at given epochs.scheduler = torch.optim.lr_scheduler.MultiStepLR(optimizer, milestones=[50, 70], gamma=0.1)for t in range(0, 80):    scheduler.step()        train(...); val(...)# Learning rate warmup by 10 epochs.scheduler = torch.optim.lr_scheduler.LambdaLR(optimizer, lr_lambda=lambda t: t / 10)for t in range(0, 10):    scheduler.step()    train(...); val(...)

保存与加载断点

注意为了能够恢复训练,我们需要同时保存模型和优化器的状态,以及当前的训练轮数。

# Save checkpoint.is_best = current_acc > best_accbest_acc = max(best_acc, current_acc)checkpoint = {     best_acc : best_acc,         epoch : t + 1,     model : model.state_dict(),     optimizer : optimizer.state_dict(),}model_path = os.path.join( model ,  checkpoint.pth.tar )torch.save(checkpoint, model_path)if is_best:    shutil.copy( checkpoint.pth.tar , model_path)# Load checkpoint.if resume:    model_path = os.path.join( model ,  checkpoint.pth.tar )    assert os.path.isfile(model_path)    checkpoint = torch.load(model_path)    best_acc = checkpoint[ best_acc ]    start_epoch = checkpoint[ epoch ]    model.load_state_dict(checkpoint[ model ])    optimizer.load_state_dict(checkpoint[ optimizer ])    print( Load checkpoint at epoch %d.  % start_epoch)

计算准确率、查准率(precision)、查全率(recall)

# data[ label ] and data[ prediction ] are groundtruth label and prediction # for each image, respectively.accuracy = np.mean(data[ label ] == data[ prediction ]) * 100# Compute recision and recall for each class.for c in range(len(num_classes)):    tp = np.dot((data[ label ] == c).astype(int),                (data[ prediction ] == c).astype(int))    tp_fp = np.sum(data[ prediction ] == c)    tp_fn = np.sum(data[ label ] == c)    precision = tp / tp_fp * 100    recall = tp / tp_fn * 100

PyTorch 其他注意事项

模型定义

建议有参数的层和汇合(pooling)层使用 torch.nn 模块定义,激活函数直接使用 torch.nn.functional。torch.nn 模块和 torch.nn.functional 的区别在于,torch.nn 模块在计算时底层调用了 torch.nn.functional,但 torch.nn 模块包括该层参数,还可以应对训练和测试两种网络状态。使用 torch.nn.functional 时要注意网络状态,如

def forward(self, x):    ...    x = torch.nn.functional.dropout(x, p=0.5, training=self.training)

model(x) 前用 model.train() 和 model.eval() 切换网络状态。

不需要计算梯度的代码块用 with torch.no_grad() 包含起来。model.eval() 和 torch.no_grad() 的区别在于,model.eval() 是将网络切换为测试状态,例如 BN 和随机失活(dropout)在训练和测试阶段使用不同的计算方法。torch.no_grad() 是关闭 PyTorch 张量的自动求导机制,以减少存储使用和加速计算,得到的结果无法进行 loss.backward()。

torch.nn.CrossEntropyLoss 的输入不需要经过 Softmax。torch.nn.CrossEntropyLoss 等价于 torch.nn.functional.log_softmax + torch.nn.NLLLoss。loss.backward() 前用 optimizer.zero_grad() 清除累积梯度。optimizer.zero_grad() 和 model.zero_grad() 效果一样。

PyTorch 性能与调试

torch.utils.data.DataLoader 中尽量设置 pin_memory=True,对特别小的数据集如 MNIST 设置 pin_memory=False 反而更快一些。num_workers 的设置需要在实验中找到最快的取值。

用 del 及时删除不用的中间变量,节约 GPU 存储。

使用 inplace 操作可节约 GPU 存储,如

x = torch.nn.functional.relu(x, inplace=True)

减少 CPU 和 GPU 之间的数据传输。例如如果你想知道一个 epoch 中每个 mini-batch 的 loss 和准确率,先将它们累积在 GPU 中等一个 epoch 结束之后一起传输回 CPU 会比每个 mini-batch 都进行一次 GPU 到 CPU 的传输更快。

使用半精度浮点数 half() 会有一定的速度提升,具体效率依赖于 GPU 型号。需要小心数值精度过低带来的稳定性问题。

时常使用 assert tensor.size() == (N, D, H, W) 作为调试手段,确保张量维度和你设想中一致。

除了标记 y 外,尽量少使用一维张量,使用 n*1 的二维张量代替,可以避免一些意想不到的一维张量计算结果。

统计代码各部分耗时

with torch.autograd.profiler.profile(enabled=True, use_cuda=False) as profile:    ...print(profile)

或者在命令行运行

python -m torch.utils.bottleneck main.py

张皓:南京大学计算机系机器学习与数据挖掘所(LAMDA)硕士生,研究方向为计算机视觉和机器学习,特别是视觉识别和深度学习。个人主页:http://lamda.nju.edu.cn/zhangh/

文章来源:微信公众号 机器学习算法与Python学习

收藏 打印