pyTorch · 測試

# pytorch -使用 Dependencies 分析報錯 ddl 文件的依賴 - [pytorch｜找不到 fbgemm.dll 問題處理](https://blog.csdn.net/Changxing_J/article/details/140489278) - [pytorch 缺少 libomp140.x86\_64.dll 下載](https://blog.csdn.net/Enexj/article/details/140870389) > - pip install torch torchvision nltk -i https://pypi.tuna.tsinghua.edu.cn/simple - pip install scikit-learn -i https://pypi.tuna.tsinghua.edu.cn/simple ~~~ import torch import torch.nn as nn import torch.optim as optim from sklearn.feature_extraction.text import CountVectorizer from sklearn.preprocessing import LabelEncoder import numpy as np # import nltk # nltk.download('punkt') # 數據準備 data = [ ("訂單狀態", "訂單查詢"), ("我的訂單什么時候到？", "訂單查詢"), ("退貨政策是什么？", "退貨"), ("我想取消訂單", "取消訂單"), ("如何申請退款？", "退款") ] questions, labels = zip(*data) vectorizer = CountVectorizer() X = vectorizer.fit_transform(questions).toarray() label_encoder = LabelEncoder() y = label_encoder.fit_transform(labels) # 定義模型 class SimpleClassifier(nn.Module): def __init__(self, input_size, num_classes): super(SimpleClassifier, self).__init__() self.fc1 = nn.Linear(input_size, 50) self.fc2 = nn.Linear(50, num_classes) def forward(self, x): x = torch.relu(self.fc1(x)) x = self.fc2(x) return x input_size = X.shape[1] num_classes = len(label_encoder.classes_) model = SimpleClassifier(input_size, num_classes) # 定義損失函數和優化器 criterion = nn.CrossEntropyLoss() optimizer = optim.SGD(model.parameters(), lr=0.01) # 將數據轉換為張量 X_tensor = torch.tensor(X, dtype=torch.float32) y_tensor = torch.tensor(y, dtype=torch.long) # 訓練模型 epochs = 100 for epoch in range(epochs): outputs = model(X_tensor) loss = criterion(outputs, y_tensor) optimizer.zero_grad() loss.backward() optimizer.step() # 保存模型的狀態字典 torch.save(model.state_dict(), 'model_state_dict.pth') # 加載模型的狀態字典 loaded_model = SimpleClassifier(input_size, num_classes) loaded_model.load_state_dict(torch.load('model_state_dict.pth', weights_only=True)) loaded_model.eval() # 測試加載的模型 def predict(question): question_vec = vectorizer.transform([question]).toarray() with torch.no_grad(): question_tensor = torch.tensor(question_vec, dtype=torch.float32) output = loaded_model(question_tensor) _, predicted = torch.max(output, 1) return label_encoder.inverse_transform(predicted.numpy())[0] test_question = "我的訂單狀態是怎樣" response = predict(test_question) print(f'預測的客服回復類型: {response}') ~~~