6-2,训练模型的3种方法

Pytorch通常需要用户编写自定义训练循环，训练循环的代码风格因人而异。

有3类典型的训练循环代码风格：脚本形式训练循环，函数形式训练循环，类形式训练循环。

下面以minist数据集的分类模型的训练为例，演示这3种训练模型的风格。

〇，准备数据

import torch 
from torch import nn 
from torchkeras import summary,Model 

import torchvision 
from torchvision import transforms

transform = transforms.Compose([transforms.ToTensor()])

ds_train = torchvision.datasets.MNIST(root="../data/minist/",train=True,download=True,transform=transform)
ds_valid = torchvision.datasets.MNIST(root="../data/minist/",train=False,download=True,transform=transform)

dl_train =  torch.utils.data.DataLoader(ds_train, batch_size=128, shuffle=True, num_workers=4)
dl_valid =  torch.utils.data.DataLoader(ds_valid, batch_size=128, shuffle=False, num_workers=4)

print(len(ds_train))
print(len(ds_valid))

60000
10000

%matplotlib inline
%config InlineBackend.figure_format = 'svg'

#查看部分样本
from matplotlib import pyplot as plt 

plt.figure(figsize=(8,8)) 
for i in range(9):
    img,label = ds_train[i]
    img = torch.squeeze(img)
    ax=plt.subplot(3,3,i+1)
    ax.imshow(img.numpy())
    ax.set_title("label = %d"%label)
    ax.set_xticks([])
    ax.set_yticks([]) 
plt.show()

一，脚本风格

脚本风格的训练循环最为常见。

net = nn.Sequential()
net.add_module("conv1",nn.Conv2d(in_channels=1,out_channels=32,kernel_size = 3))
net.add_module("pool1",nn.MaxPool2d(kernel_size = 2,stride = 2))
net.add_module("conv2",nn.Conv2d(in_channels=32,out_channels=64,kernel_size = 5))
net.add_module("pool2",nn.MaxPool2d(kernel_size = 2,stride = 2))
net.add_module("dropout",nn.Dropout2d(p = 0.1))
net.add_module("adaptive_pool",nn.AdaptiveMaxPool2d((1,1)))
net.add_module("flatten",nn.Flatten())
net.add_module("linear1",nn.Linear(64,32))
net.add_module("relu",nn.ReLU())
net.add_module("linear2",nn.Linear(32,10))

print(net)

Sequential(
  (conv1): Conv2d(1, 32, kernel_size=(3, 3), stride=(1, 1))
  (pool1): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
  (conv2): Conv2d(32, 64, kernel_size=(5, 5), stride=(1, 1))
  (pool2): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
  (dropout): Dropout2d(p=0.1, inplace=False)
  (adaptive_pool): AdaptiveMaxPool2d(output_size=(1, 1))
  (flatten): Flatten()
  (linear1): Linear(in_features=64, out_features=32, bias=True)
  (relu): ReLU()
  (linear2): Linear(in_features=32, out_features=10, bias=True)
)

summary(net,input_shape=(1,32,32))

----------------------------------------------------------------
        Layer (type)               Output Shape         Param #
================================================================
            Conv2d-1           [-1, 32, 30, 30]             320
         MaxPool2d-2           [-1, 32, 15, 15]               0
            Conv2d-3           [-1, 64, 11, 11]          51,264
         MaxPool2d-4             [-1, 64, 5, 5]               0
         Dropout2d-5             [-1, 64, 5, 5]               0
 AdaptiveMaxPool2d-6             [-1, 64, 1, 1]               0
           Flatten-7                   [-1, 64]               0
            Linear-8                   [-1, 32]           2,080
              ReLU-9                   [-1, 32]               0
           Linear-10                   [-1, 10]             330
================================================================
Total params: 53,994
Trainable params: 53,994
Non-trainable params: 0
----------------------------------------------------------------
Input size (MB): 0.003906
Forward/backward pass size (MB): 0.359695
Params size (MB): 0.205971
Estimated Total Size (MB): 0.569572
----------------------------------------------------------------

import datetime
import numpy as np 
import pandas as pd 
from sklearn.metrics import accuracy_score

def accuracy(y_pred,y_true):
    y_pred_cls = torch.argmax(nn.Softmax(dim=1)(y_pred),dim=1).data
    return accuracy_score(y_true,y_pred_cls)

loss_func = nn.CrossEntropyLoss()
optimizer = torch.optim.Adam(params=net.parameters(),lr = 0.01)
metric_func = accuracy
metric_name = "accuracy"

epochs = 3
log_step_freq = 100

dfhistory = pd.DataFrame(columns = ["epoch","loss",metric_name,"val_loss","val_"+metric_name]) 
print("Start Training...")
nowtime = datetime.datetime.now().strftime('%Y-%m-%d %H:%M:%S')
print("=========="*8 + "%s"%nowtime)

for epoch in range(1,epochs+1):  

    # 1，训练循环-------------------------------------------------
    net.train()
    loss_sum = 0.0
    metric_sum = 0.0
    step = 1

    for step, (features,labels) in enumerate(dl_train, 1):

        # 梯度清零
        optimizer.zero_grad()

        # 正向传播求损失
        predictions = net(features)
        loss = loss_func(predictions,labels)
        metric = metric_func(predictions,labels)

        # 反向传播求梯度
        loss.backward()
        optimizer.step()

        # 打印batch级别日志
        loss_sum += loss.item()
        metric_sum += metric.item()
        if step%log_step_freq == 0:   
            print(("[step = %d] loss: %.3f, "+metric_name+": %.3f") %
                  (step, loss_sum/step, metric_sum/step))

    # 2，验证循环-------------------------------------------------
    net.eval()
    val_loss_sum = 0.0
    val_metric_sum = 0.0
    val_step = 1

    for val_step, (features,labels) in enumerate(dl_valid, 1):
        with torch.no_grad():
            predictions = net(features)
            val_loss = loss_func(predictions,labels)
            val_metric = metric_func(predictions,labels)

        val_loss_sum += val_loss.item()
        val_metric_sum += val_metric.item()

    # 3，记录日志-------------------------------------------------
    info = (epoch, loss_sum/step, metric_sum/step, 
            val_loss_sum/val_step, val_metric_sum/val_step)
    dfhistory.loc[epoch-1] = info

    # 打印epoch级别日志
    print(("\nEPOCH = %d, loss = %.3f,"+ metric_name + \
          "  = %.3f, val_loss = %.3f, "+"val_"+ metric_name+" = %.3f") 
          %info)
    nowtime = datetime.datetime.now().strftime('%Y-%m-%d %H:%M:%S')
    print("\n"+"=========="*8 + "%s"%nowtime)

print('Finished Training...')

Start Training...
================================================================================2020-06-26 12:49:16
[step = 100] loss: 0.742, accuracy: 0.745
[step = 200] loss: 0.466, accuracy: 0.843
[step = 300] loss: 0.363, accuracy: 0.880
[step = 400] loss: 0.310, accuracy: 0.898

EPOCH = 1, loss = 0.281,accuracy  = 0.908, val_loss = 0.087, val_accuracy = 0.972

================================================================================2020-06-26 12:50:32
[step = 100] loss: 0.103, accuracy: 0.970
[step = 200] loss: 0.114, accuracy: 0.966
[step = 300] loss: 0.112, accuracy: 0.967
[step = 400] loss: 0.108, accuracy: 0.968

EPOCH = 2, loss = 0.111,accuracy  = 0.967, val_loss = 0.082, val_accuracy = 0.976

================================================================================2020-06-26 12:51:47
[step = 100] loss: 0.093, accuracy: 0.972
[step = 200] loss: 0.095, accuracy: 0.971
[step = 300] loss: 0.092, accuracy: 0.972
[step = 400] loss: 0.093, accuracy: 0.972

EPOCH = 3, loss = 0.098,accuracy  = 0.971, val_loss = 0.113, val_accuracy = 0.970

================================================================================2020-06-26 12:53:09
Finished Training...

二，函数风格

该风格在脚本形式上作了简单的函数封装。

class Net(nn.Module):
    def __init__(self):
        super(Net, self).__init__()
        self.layers = nn.ModuleList([
            nn.Conv2d(in_channels=1,out_channels=32,kernel_size = 3),
            nn.MaxPool2d(kernel_size = 2,stride = 2),
            nn.Conv2d(in_channels=32,out_channels=64,kernel_size = 5),
            nn.MaxPool2d(kernel_size = 2,stride = 2),
            nn.Dropout2d(p = 0.1),
            nn.AdaptiveMaxPool2d((1,1)),
            nn.Flatten(),
            nn.Linear(64,32),
            nn.ReLU(),
            nn.Linear(32,10)]
        )
    def forward(self,x):
        for layer in self.layers:
            x = layer(x)
        return x
net = Net()
print(net)

Net(
  (layers): ModuleList(
    (0): Conv2d(1, 32, kernel_size=(3, 3), stride=(1, 1))
    (1): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
    (2): Conv2d(32, 64, kernel_size=(5, 5), stride=(1, 1))
    (3): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
    (4): Dropout2d(p=0.1, inplace=False)
    (5): AdaptiveMaxPool2d(output_size=(1, 1))
    (6): Flatten()
    (7): Linear(in_features=64, out_features=32, bias=True)
    (8): ReLU()
    (9): Linear(in_features=32, out_features=10, bias=True)
  )
)

summary(net,input_shape=(1,32,32))

----------------------------------------------------------------
        Layer (type)               Output Shape         Param #
================================================================
            Conv2d-1           [-1, 32, 30, 30]             320
         MaxPool2d-2           [-1, 32, 15, 15]               0
            Conv2d-3           [-1, 64, 11, 11]          51,264
         MaxPool2d-4             [-1, 64, 5, 5]               0
         Dropout2d-5             [-1, 64, 5, 5]               0
 AdaptiveMaxPool2d-6             [-1, 64, 1, 1]               0
           Flatten-7                   [-1, 64]               0
            Linear-8                   [-1, 32]           2,080
              ReLU-9                   [-1, 32]               0
           Linear-10                   [-1, 10]             330
================================================================
Total params: 53,994
Trainable params: 53,994
Non-trainable params: 0
----------------------------------------------------------------
Input size (MB): 0.003906
Forward/backward pass size (MB): 0.359695
Params size (MB): 0.205971
Estimated Total Size (MB): 0.569572
----------------------------------------------------------------

import datetime
import numpy as np 
import pandas as pd 
from sklearn.metrics import accuracy_score

def accuracy(y_pred,y_true):
    y_pred_cls = torch.argmax(nn.Softmax(dim=1)(y_pred),dim=1).data
    return accuracy_score(y_true,y_pred_cls)

model = net
model.optimizer = torch.optim.SGD(model.parameters(),lr = 0.01)
model.loss_func = nn.CrossEntropyLoss()
model.metric_func = accuracy
model.metric_name = "accuracy"

def train_step(model,features,labels):

    # 训练模式，dropout层发生作用
    model.train()

    # 梯度清零
    model.optimizer.zero_grad()

    # 正向传播求损失
    predictions = model(features)
    loss = model.loss_func(predictions,labels)
    metric = model.metric_func(predictions,labels)

    # 反向传播求梯度
    loss.backward()
    model.optimizer.step()

    return loss.item(),metric.item()

@torch.no_grad()
def valid_step(model,features,labels):

    # 预测模式，dropout层不发生作用
    model.eval()

    predictions = model(features)
    loss = model.loss_func(predictions,labels)
    metric = model.metric_func(predictions,labels)

    return loss.item(), metric.item()


# 测试train_step效果
features,labels = next(iter(dl_train))
train_step(model,features,labels)

(2.32741117477417, 0.1015625)

def train_model(model,epochs,dl_train,dl_valid,log_step_freq):

    metric_name = model.metric_name
    dfhistory = pd.DataFrame(columns = ["epoch","loss",metric_name,"val_loss","val_"+metric_name]) 
    print("Start Training...")
    nowtime = datetime.datetime.now().strftime('%Y-%m-%d %H:%M:%S')
    print("=========="*8 + "%s"%nowtime)

    for epoch in range(1,epochs+1):  

        # 1，训练循环-------------------------------------------------
        loss_sum = 0.0
        metric_sum = 0.0
        step = 1

        for step, (features,labels) in enumerate(dl_train, 1):

            loss,metric = train_step(model,features,labels)

            # 打印batch级别日志
            loss_sum += loss
            metric_sum += metric
            if step%log_step_freq == 0:   
                print(("[step = %d] loss: %.3f, "+metric_name+": %.3f") %
                      (step, loss_sum/step, metric_sum/step))

        # 2，验证循环-------------------------------------------------
        val_loss_sum = 0.0
        val_metric_sum = 0.0
        val_step = 1

        for val_step, (features,labels) in enumerate(dl_valid, 1):

            val_loss,val_metric = valid_step(model,features,labels)

            val_loss_sum += val_loss
            val_metric_sum += val_metric

        # 3，记录日志-------------------------------------------------
        info = (epoch, loss_sum/step, metric_sum/step, 
                val_loss_sum/val_step, val_metric_sum/val_step)
        dfhistory.loc[epoch-1] = info

        # 打印epoch级别日志
        print(("\nEPOCH = %d, loss = %.3f,"+ metric_name + \
              "  = %.3f, val_loss = %.3f, "+"val_"+ metric_name+" = %.3f") 
              %info)
        nowtime = datetime.datetime.now().strftime('%Y-%m-%d %H:%M:%S')
        print("\n"+"=========="*8 + "%s"%nowtime)

    print('Finished Training...')
    return dfhistory

epochs = 3
dfhistory = train_model(model,epochs,dl_train,dl_valid,log_step_freq = 100)

Start Training...
================================================================================2020-06-26 13:10:00
[step = 100] loss: 2.298, accuracy: 0.137
[step = 200] loss: 2.288, accuracy: 0.145
[step = 300] loss: 2.278, accuracy: 0.165
[step = 400] loss: 2.265, accuracy: 0.183

EPOCH = 1, loss = 2.254,accuracy  = 0.195, val_loss = 2.158, val_accuracy = 0.301

================================================================================2020-06-26 13:11:23
[step = 100] loss: 2.127, accuracy: 0.302
[step = 200] loss: 2.080, accuracy: 0.338
[step = 300] loss: 2.025, accuracy: 0.374
[step = 400] loss: 1.957, accuracy: 0.411

EPOCH = 2, loss = 1.905,accuracy  = 0.435, val_loss = 1.469, val_accuracy = 0.710

================================================================================2020-06-26 13:12:43
[step = 100] loss: 1.435, accuracy: 0.615
[step = 200] loss: 1.324, accuracy: 0.647
[step = 300] loss: 1.221, accuracy: 0.672
[step = 400] loss: 1.132, accuracy: 0.696

EPOCH = 3, loss = 1.074,accuracy  = 0.711, val_loss = 0.582, val_accuracy = 0.878

================================================================================2020-06-26 13:13:59
Finished Training...

三，类风格

此处使用torchkeras中定义的模型接口构建模型，并调用compile方法和fit方法训练模型。

使用该形式训练模型非常简洁明了。推荐使用该形式。

class CnnModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.layers = nn.ModuleList([
            nn.Conv2d(in_channels=1,out_channels=32,kernel_size = 3),
            nn.MaxPool2d(kernel_size = 2,stride = 2),
            nn.Conv2d(in_channels=32,out_channels=64,kernel_size = 5),
            nn.MaxPool2d(kernel_size = 2,stride = 2),
            nn.Dropout2d(p = 0.1),
            nn.AdaptiveMaxPool2d((1,1)),
            nn.Flatten(),
            nn.Linear(64,32),
            nn.ReLU(),
            nn.Linear(32,10)]
        )
    def forward(self,x):
        for layer in self.layers:
            x = layer(x)
        return x
model = torchkeras.Model(CnnModel())
print(model)

CnnModel(
  (layers): ModuleList(
    (0): Conv2d(1, 32, kernel_size=(3, 3), stride=(1, 1))
    (1): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
    (2): Conv2d(32, 64, kernel_size=(5, 5), stride=(1, 1))
    (3): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
    (4): Dropout2d(p=0.1, inplace=False)
    (5): AdaptiveMaxPool2d(output_size=(1, 1))
    (6): Flatten()
    (7): Linear(in_features=64, out_features=32, bias=True)
    (8): ReLU()
    (9): Linear(in_features=32, out_features=10, bias=True)
  )
)

model.summary(input_shape=(1,32,32))

----------------------------------------------------------------
        Layer (type)               Output Shape         Param #
================================================================
            Conv2d-1           [-1, 32, 30, 30]             320
         MaxPool2d-2           [-1, 32, 15, 15]               0
            Conv2d-3           [-1, 64, 11, 11]          51,264
         MaxPool2d-4             [-1, 64, 5, 5]               0
         Dropout2d-5             [-1, 64, 5, 5]               0
 AdaptiveMaxPool2d-6             [-1, 64, 1, 1]               0
           Flatten-7                   [-1, 64]               0
            Linear-8                   [-1, 32]           2,080
              ReLU-9                   [-1, 32]               0
           Linear-10                   [-1, 10]             330
================================================================
Total params: 53,994
Trainable params: 53,994
Non-trainable params: 0
----------------------------------------------------------------
Input size (MB): 0.003906
Forward/backward pass size (MB): 0.359695
Params size (MB): 0.205971
Estimated Total Size (MB): 0.569572
----------------------------------------------------------------

from sklearn.metrics import accuracy_score

def accuracy(y_pred,y_true):
    y_pred_cls = torch.argmax(nn.Softmax(dim=1)(y_pred),dim=1).data
    return accuracy_score(y_true.numpy(),y_pred_cls.numpy())

model.compile(loss_func = nn.CrossEntropyLoss(),
             optimizer= torch.optim.Adam(model.parameters(),lr = 0.02),
             metrics_dict={"accuracy":accuracy})

dfhistory = model.fit(3,dl_train = dl_train, dl_val=dl_valid, log_step_freq=100) 

Start Training ...

================================================================================2020-06-26 13:22:39
{'step': 100, 'loss': 0.976, 'accuracy': 0.664}
{'step': 200, 'loss': 0.611, 'accuracy': 0.795}
{'step': 300, 'loss': 0.478, 'accuracy': 0.841}
{'step': 400, 'loss': 0.403, 'accuracy': 0.868}

 +-------+-------+----------+----------+--------------+
| epoch |  loss | accuracy | val_loss | val_accuracy |
+-------+-------+----------+----------+--------------+
|   1   | 0.371 |  0.879   |  0.087   |    0.972     |
+-------+-------+----------+----------+--------------+

================================================================================2020-06-26 13:23:59
{'step': 100, 'loss': 0.182, 'accuracy': 0.948}
{'step': 200, 'loss': 0.176, 'accuracy': 0.949}
{'step': 300, 'loss': 0.173, 'accuracy': 0.95}
{'step': 400, 'loss': 0.174, 'accuracy': 0.951}

 +-------+-------+----------+----------+--------------+
| epoch |  loss | accuracy | val_loss | val_accuracy |
+-------+-------+----------+----------+--------------+
|   2   | 0.175 |  0.951   |  0.152   |    0.958     |
+-------+-------+----------+----------+--------------+

================================================================================2020-06-26 13:25:22
{'step': 100, 'loss': 0.143, 'accuracy': 0.961}
{'step': 200, 'loss': 0.151, 'accuracy': 0.959}
{'step': 300, 'loss': 0.149, 'accuracy': 0.96}
{'step': 400, 'loss': 0.152, 'accuracy': 0.959}

 +-------+-------+----------+----------+--------------+
| epoch |  loss | accuracy | val_loss | val_accuracy |
+-------+-------+----------+----------+--------------+
|   3   | 0.153 |  0.959   |  0.086   |    0.975     |
+-------+-------+----------+----------+--------------+

================================================================================2020-06-26 13:26:48
Finished Training...

如果对本书内容理解上有需要进一步和作者交流的地方，欢迎在公众号"Python与算法之美"下留言。作者时间和精力有限，会酌情予以回复。

也可以在公众号后台回复关键字：加群，加入读者交流群和大家讨论。