【深度学习PyTorch入门】7.Save and Load the Model 保存和加载模型

Save and Load the Model 保存和加载模型

在本节中，我们将了解如何通过保存、加载和运行模型预测来持久保存模型状态。

import torch
import torchvision.models as models

Saving and Loading Model Weights 保存和加载模型权重

PyTorch 模型将学习到的参数存储在内部状态字典中，称为state_dict。这些可以通过torch.save方法保存 ：

model = models.vgg16(weights='IMAGENET1K_V1')
torch.save(model.state_dict(), 'model_weights.pth')

Out:

Downloading: "https://download.pytorch.org/models/vgg16-397923af.pth" to /var/lib/jenkins/.cache/torch/hub/checkpoints/vgg16-397923af.pth

  0%|          | 0.00/528M [00:00<?, ?B/s]
  3%|2         | 13.6M/528M [00:00<00:03, 143MB/s]
  5%|5         | 28.0M/528M [00:00<00:03, 147MB/s]
  8%|8         | 42.5M/528M [00:00<00:03, 149MB/s]
 11%|#         | 56.9M/528M [00:00<00:03, 150MB/s]
 13%|#3        | 71.2M/528M [00:00<00:03, 150MB/s]
 16%|#6        | 85.7M/528M [00:00<00:03, 151MB/s]
 19%|#8        | 100M/528M [00:00<00:02, 151MB/s]
 22%|##1       | 115M/528M [00:00<00:02, 151MB/s]
 24%|##4       | 129M/528M [00:00<00:02, 151MB/s]
 27%|##7       | 143M/528M [00:01<00:02, 151MB/s]
 30%|##9       | 158M/528M [00:01<00:02, 152MB/s]
 33%|###2      | 172M/528M [00:01<00:02, 151MB/s]
 35%|###5      | 187M/528M [00:01<00:02, 151MB/s]
 38%|###8      | 201M/528M [00:01<00:02, 151MB/s]
 41%|####      | 216M/528M [00:01<00:02, 151MB/s]
 44%|####3     | 230M/528M [00:01<00:02, 152MB/s]
 46%|####6     | 245M/528M [00:01<00:01, 152MB/s]
 49%|####9     | 259M/528M [00:01<00:01, 152MB/s]
 52%|#####1    | 274M/528M [00:01<00:01, 152MB/s]
 55%|#####4    | 288M/528M [00:02<00:01, 152MB/s]
 57%|#####7    | 303M/528M [00:02<00:01, 152MB/s]
 60%|######    | 317M/528M [00:02<00:01, 152MB/s]
 63%|######2   | 332M/528M [00:02<00:01, 152MB/s]
 66%|######5   | 346M/528M [00:02<00:01, 151MB/s]
 68%|######8   | 361M/528M [00:02<00:01, 151MB/s]
 71%|#######1  | 375M/528M [00:02<00:01, 151MB/s]
 74%|#######3  | 390M/528M [00:02<00:00, 151MB/s]
 77%|#######6  | 404M/528M [00:02<00:00, 151MB/s]
 79%|#######9  | 418M/528M [00:02<00:00, 151MB/s]
 82%|########2 | 433M/528M [00:03<00:00, 151MB/s]
 85%|########4 | 447M/528M [00:03<00:00, 151MB/s]
 88%|########7 | 462M/528M [00:03<00:00, 151MB/s]
 90%|######### | 476M/528M [00:03<00:00, 151MB/s]
 93%|#########2| 491M/528M [00:03<00:00, 151MB/s]
 96%|#########5| 505M/528M [00:03<00:00, 151MB/s]
 98%|#########8| 520M/528M [00:03<00:00, 151MB/s]
100%|##########| 528M/528M [00:03<00:00, 151MB/s]

要加载模型权重，您需要先创建同一模型的实例，然后使用load_state_dict()方法加载参数。

model = models.vgg16() # we do not specify ``weights``, i.e. create untrained model
model.load_state_dict(torch.load('model_weights.pth'))
model.eval()

Out:

VGG(
  (features): Sequential(
    (0): Conv2d(3, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (1): ReLU(inplace=True)
    (2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (3): ReLU(inplace=True)
    (4): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
    (5): Conv2d(64, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (6): ReLU(inplace=True)
    (7): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (8): ReLU(inplace=True)
    (9): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
    (10): Conv2d(128, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (11): ReLU(inplace=True)
    (12): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (13): ReLU(inplace=True)
    (14): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (15): ReLU(inplace=True)
    (16): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
    (17): Conv2d(256, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (18): ReLU(inplace=True)
    (19): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (20): ReLU(inplace=True)
    (21): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (22): ReLU(inplace=True)
    (23): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
    (24): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (25): ReLU(inplace=True)
    (26): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (27): ReLU(inplace=True)
    (28): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (29): ReLU(inplace=True)
    (30): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
  )
  (avgpool): AdaptiveAvgPool2d(output_size=(7, 7))
  (classifier): Sequential(
    (0): Linear(in_features=25088, out_features=4096, bias=True)
    (1): ReLU(inplace=True)
    (2): Dropout(p=0.5, inplace=False)
    (3): Linear(in_features=4096, out_features=4096, bias=True)
    (4): ReLU(inplace=True)
    (5): Dropout(p=0.5, inplace=False)
    (6): Linear(in_features=4096, out_features=1000, bias=True)
  )
)

笔记

请务必在推理之前，调用model.eval()方法，将 dropout 和批量归一化层设置为评估模式。如果不这样做将会产生不一致的推理结果。

Saving and Loading Models with Shapes 保存和加载带有形状的模型

加载模型权重时，我们需要首先实例化模型类，因为该类定义了网络的结构。我们可能希望将此类的结构与模型一起保存，在这种情况下，我们可以将model（而不是model.state_dict()）传递给保存函数：

torch.save(model, 'model.pth')

然后我们可以像这样加载模型：

model = torch.load('model.pth')

笔记

此方法在序列化模型时使用 Python pickle模块，因此它依赖于加载模型时可用的实际类定义。

Related Tutorials 相关教程

• Saving and Loading a General Checkpoint in PyTorch
• [Tips for loading an nn.Module from a checkpoint](https://pytorch.org/tutorials/recipes/recipes/module_load_state_dict_tips.html?highlight=loading nn module from checkpoint)

References 参考资料

Save and Load the Model — PyTorch Tutorials 2.2.0+cu121 documentation

Save and Load the Model — PyTorch Tutorials 2.2.0+cu121 documentation

Github

storm-ice/Get_started_with_PyTorch

storm-ice/Get_started_with_PyTorch