dea6518310856dcc35d3f877f75a2502200cad40,batchflow/models/torch/layers/conv.py,BaseConv,__init__,#BaseConv#Any#Any#Any#Any#Any#Any#Any#Any#,14

Before Change

        padding = calc_padding(inputs, padding=padding, transposed=self.TRANSPOSED, **args)

        if isinstance(padding, int) or \
            (isinstance(padding, tuple) and all(isinstance(item, int) for item in padding)) or \
            all(all(inner == item[0] and inner > 0 for inner in item) for item in padding):
            // If every element is the same and non-negative, we can use the `padding` of Conv layers
            if isinstance(padding, tuple) and all(isinstance(item, tuple) for item in padding):
                padding = tuple(item[0] for item in padding)
            args["padding"] = padding
            self.padding = False
        else:
            // Otherwise, use custom padding
            self.padding = sum(padding, ())

        self.layer = self.LAYERS[get_num_dims(inputs)](**args)

    def forward(self, x):
        if self.padding:

After Change

        padding = calc_padding(inputs, padding=padding, transposed=self.TRANSPOSED, **args)

        if isinstance(padding, tuple) and isinstance(padding[0], tuple):
            args["padding"] = 0
            self.padding = sum(padding, ())
        else:
            args["padding"] = padding
            self.padding = 0

        self.layer = self.LAYERS[get_num_dims(inputs)](**args)

    def forward(self, x):
        if self.padding:

Instances