from torch import nn
from torch.nn import functional
import torch
import numpy as np


class Entropy_Loss(nn.Module):
    def __init__(self):
        super(Entropy_Loss, self).__init__()

    def forward(self, outputs, labels):
        # 转换为张量并确保在同一設備上
        outputs_New = torch.as_tensor(outputs, dtype=torch.float32)
        labels_New = torch.as_tensor(labels, dtype=torch.float32)  # 標籤應該是 long 類型用於索引
        
        loss = functional.cross_entropy(outputs_New, labels_New)
            
        return torch.as_tensor(loss, dtype=torch.float32)