本文主要是介绍【Python】一个最基础，但是超难看出来的类声明为tuple的BUG，希望对大家解决编程问题提供一定的参考价值，需要的开发者们随着小编来一起学习吧！
直接上代码，我在layer.py中定义了这样一个类：
class QConv2d(nn.Conv2d):''' A standard Conv2d layer that quantizes input and weights before performing mult(weights, input). It quantizes outputs also. '''def __init__(self, inCh: int, outCh: int, kDim: int, stride: int = 1, quantization: dict = None):super(QConv2d,self).__init__(inCh, outCh, kDim, stride=stride, bias = False, padding = int((kDim-1)/2))self.isWinograd = Falseself.quantize = True if quantization['q'] else Falseif self.quantize:self.bits = quantization['bits']self.Quantize_weights = Quant(self.bits)self.Quantize_input = Quant(self.bits)self.Quantize_output = Quant(self.bits)else:self.bits = 'FP'def forward(self, input):if self.quantize:qinput = self.Quantize_input(input)qweight = self.Quantize_weights(self.weight)return self.Quantize_output(nn.functional.conv2d(qinput, qweight, self.bias, self.stride, self.padding, self.dilation, self.groups))else:return nn.functional.conv2d(input, self.weight, self.bias, self.stride, self.padding, self.dilation, self.groups)
然后我在model.py中对其进行声明：
self.conv1 = QConv2d(in_planes, planes, kDim=3, stride=stride, quantization=Q_args),
结果，出现了这样的报错：
TypeError: 'tuple' object is not callable
打印self.conv1可以看到：
<class 'tuple'>
(QConv2d(16, 32, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False(Quantize_weights): Quant()(Quantize_input): Quant()(Quantize_output): Quant()
),)
为啥这个class是个tuple类型呀！？
原因非常简单，我在声明类的时候， 末尾多了一个逗号！！！！！！ 修改后如下：
self.conv1 = QConv2d(in_planes, planes, kDim=3, stride=stride, quantization=Q_args)
再打印self.conv1看看：
<class 'src.layers.Conv2d'>
QConv2d( 32, 64, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False(Quantize_weights): Quant()(Quantize_input): Quant()(Quantize_output): Quant()
)
小结一句，Python末尾是没有分号或者逗号的，要防止误敲无关字符！
这篇关于【Python】一个最基础，但是超难看出来的类声明为tuple的BUG的文章就介绍到这儿，希望我们推荐的文章对编程师们有所帮助！
