这几天关于accuracy和loss的计算有一些疑惑,原来是自己还没有弄清楚. 给出实例 def train(train_loader, model, criteon, optimizer, epoch): train_loss = 0 train_acc = 0 num_correct= 0 for step, (x,y) in enumerate(train_loader): # x: [b, 3, 224, 224], y: [b] x, y = x.to(device), y.to(de