炒股资金放大 PyTorch实现简易文本模型训练跟预测

宏利金利3个月定开债券发起式为债券型-长债基金，根据最新一期基金季报显示，该基金资产配置：无股票类资产，债券占净值比103.99%，现金占净值比0.03%。

PyTorch可以将文本数据集打标，然后进行训练，本文分享一段简单文本模型的训练跟预测方法。

一、 创建样本

创建样本要用到Pandas工具，确保你电脑上安装了Python环境，没有安装的请查看我前面的博客。

安装pandas方法：

pip install pandas

打开PyCharm终端输入命令:pip install pandas

等待安装完成。

然后编写创建csv文件样本集代码，新建generate_data.py文件，内容为:

import pandas as pd

# 创建示例数据

data = {

'text': [

'This is a good movie.',

'I did not like the book.',

'The weather is nice.',

'Terrible service.',

'MyName is AI Teacher.'

],

'label': [1.0, 0.0, 1.0, 0.0, 1.0]

}

# 创建 DataFrame

df = pd.DataFrame(data)

# 保存为 CSV 文件

df.to_csv('text_data.csv', index=False)

根据以上代码，在PyCharm工具右键点击运行，就会在当前目录生成text_data.csv文件啦：

在PyCharm打开text_data.csv文件可查看数据集:

这样就完成了样本创建。

二、 安装PyTorch

打开PyCharm终端安装PyTorch:

pip install torch torchvision

安装过程:

完成PyTorch安装

三、 安装sklearn

打开PyCharm终端安装sklearn:

pip install scikit-learn

使用镜像加速的话:

pip install scikit-learn -i https://mirrors.aliyun.com/pypi/simple

安装过程:

四、 导入PyTorch相关库

新建train_new1.py文件，导入库:

import torch

import torch.nn as nn

import torch.optim as optim

from torch.utils.data import DataLoader, Dataset

from sklearn.model_selection import train_test_split

from sklearn.feature_extraction.text import CountVectorizer

from sklearn.preprocessing import LabelEncoder

import pandas as pd

五、 创建数据集加载器

class TextDataset(Dataset):

def __init__(self, texts, labels):

self.texts = texts

self.labels = labels

def __len__(self):

return len(self.labels)

def __getitem__(self, idx):

text = torch.tensor(self.texts[idx], dtype=torch.long)

label = torch.tensor(self.labels[idx], dtype=torch.long)

return text, label

六、 创建自定义模型结构

class TextClassifier(nn.Module):

def __init__(self, input_dim, hidden_dim, output_dim):

super(TextClassifier, self).__init__()

self.fc1 = nn.Linear(input_dim, hidden_dim)

self.relu = nn.ReLU()

self.fc2 = nn.Linear(hidden_dim, output_dim)

def forward(self, x):

out = self.fc1(x)

out = self.relu(out)

out = self.fc2(out)

return out

七、 加载数据并训练

7.1加载数据

加载第一部分提到的文本样本集:

# 加载数据

data = pd.read_csv('text_data.csv')

texts = data['text'].values

labels = data['label'].values

进行分词和向量化:

# 分词和向量化

vectorizer = CountVectorizer(max_features=5000) # 可以调整特征数量

X = vectorizer.fit_transform(texts).toarray()

进行数据预处理:

# 标签编码

label_encoder = LabelEncoder()

y = label_encoder.fit_transform(labels)

# 划分训练集和验证集

X_train, X_val, y_train, y_val = train_test_split(X, y, test_size=0.2, random_state=42)

代码如图:

7.2创建数据加载器

# 创建数据集实例

train_dataset = TextDataset(X_train, y_train)

val_dataset = TextDataset(X_val, y_val)

# 创建数据加载器

train_loader = DataLoader(train_dataset, batch_size=32, shuffle=True)

val_loader = DataLoader(val_dataset, batch_size=32, shuffle=False)

7.3设置训练参数

input_dim = X_train.shape[1] # 输入维度

hidden_dim = 128

output_dim = len(label_encoder.classes_) # 输出维度

model = TextClassifier(input_dim, hidden_dim, output_dim)

optimizer = optim.Adam(model.parameters(), lr=0.001)

criterion = nn.CrossEntropyLoss()

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

model.to(device)

7.4编写训练函数并进行训练

def train(model, dataloader, optimizer, criterion, device):

model.train()

total_loss = 0

for inputs, labels in dataloader:

inputs, labels = inputs.to(device), labels.to(device)

inputs = inputs.float()

optimizer.zero_grad()

outputs = model(inputs)

loss = criterion(outputs, labels)

loss.backward()

optimizer.step()

total_loss += loss.item()

return total_loss / len(dataloader)

# 训练模型

num_epochs = 10

for epoch in range(num_epochs):

train_loss = train(model, train_loader, optimizer, criterion, device)

print(f'Epoch [{epoch+1}/{num_epochs}], Loss: {train_loss:.4f}')

八、 数据预测

8.1预测函数及预测

编写预测函数：

def predict(model, dataloader, device):

model.eval()

predictions = []

with torch.no_grad():

for inputs, _ in dataloader:

inputs = inputs.to(device)

inputs = inputs.float()

outputs = model(inputs)

_, preds = torch.max(outputs, 1)

predictions.extend(preds.tolist())

return predictions

使用测试集预测结果:

# 使用验证集进行预测

predictions = predict(model, val_loader, device)

print(predictions[:5])

8.2完整数据训练预测代码

import torch

import torch.nn as nn

import torch.optim as optim

from torch.utils.data import DataLoader, Dataset

from sklearn.model_selection import train_test_split

from sklearn.feature_extraction.text import CountVectorizer

from sklearn.preprocessing import LabelEncoder

import pandas as pd

class TextDataset(Dataset):

def __init__(self, texts, labels):

self.texts = texts

self.labels = labels

def __len__(self):

return len(self.labels)

def __getitem__(self, idx):

text = torch.tensor(self.texts[idx], dtype=torch.long)

label = torch.tensor(self.labels[idx], dtype=torch.long)

return text, label

class TextClassifier(nn.Module):

def __init__(self, input_dim, hidden_dim, output_dim):

super(TextClassifier, self).__init__()

self.fc1 = nn.Linear(input_dim, hidden_dim)

self.relu = nn.ReLU()

self.fc2 = nn.Linear(hidden_dim, output_dim)

def forward(self, x):

out = self.fc1(x)

out = self.relu(out)

out = self.fc2(out)

return out

# 加载数据

data = pd.read_csv('text_data.csv')

texts = data['text'].values

labels = data['label'].values

# 分词和向量化

vectorizer = CountVectorizer(max_features=5000) # 可以调整特征数量

X = vectorizer.fit_transform(texts).toarray()

# 标签编码

label_encoder = LabelEncoder()

y = label_encoder.fit_transform(labels)

# 划分训练集和验证集

X_train, X_val, y_train, y_val = train_test_split(X, y, test_size=0.2, random_state=42)

# 创建数据集实例

train_dataset = TextDataset(X_train, y_train)

val_dataset = TextDataset(X_val, y_val)

# 创建数据加载器

train_loader = DataLoader(train_dataset, batch_size=32, shuffle=True)

val_loader = DataLoader(val_dataset, batch_size=32, shuffle=False)

input_dim = X_train.shape[1] # 输入维度

hidden_dim = 128

output_dim = len(label_encoder.classes_) # 输出维度

model = TextClassifier(input_dim, hidden_dim, output_dim)

optimizer = optim.Adam(model.parameters(), lr=0.001)

criterion = nn.CrossEntropyLoss()

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

model.to(device)

def train(model, dataloader, optimizer, criterion, device):

model.train()

total_loss = 0

for inputs, labels in dataloader:

inputs, labels = inputs.to(device), labels.to(device)

inputs = inputs.float()

optimizer.zero_grad()

outputs = model(inputs)

loss = criterion(outputs, labels)

loss.backward()

optimizer.step()

total_loss += loss.item()

return total_loss / len(dataloader)

# 训练模型

num_epochs = 10

for epoch in range(num_epochs):

train_loss = train(model, train_loader, optimizer, criterion, device)

print(f'Epoch [{epoch+1}/{num_epochs}], Loss: {train_loss:.4f}')

def predict(model, dataloader, device):

model.eval()

predictions = []

with torch.no_grad():

for inputs, _ in dataloader:

inputs = inputs.to(device)

inputs = inputs.float()

outputs = model(inputs)

_, preds = torch.max(outputs, 1)

predictions.extend(preds.tolist())

return predictions

# 使用验证集进行预测

predictions = predict(model, val_loader, device)

print(predictions[:5])

8.3运行结果

右键运行train1_new.py文件，展开简易训练和得出结果:

共完成十轮训练，由于数据样本较少，训练速度较快。实际生产环境数据集一次完整训练可能需要花费一周，最好采用NVIDA显卡4080 Super GPU深度学习服务器训练。

欢迎关注博主炒股资金放大，获取更多AI技术学习文章。

valinputsmodeldevicelabels发布于：湖北省声明：该文观点仅代表作者本人，搜狐号系信息发布平台，搜狐仅提供信息存储空间服务。