Evaluate lacks an activation function in TransferLearning.py of chapter11

[Bettergo567]

outputs = model(inputs)
outputs = torch.sigmoid(outputs) # 添加这一行
preds = (outputs > 0.5).float()