3x3 convolution with padding

Save checkpoint when `KeyboardInterrupt` or other errors occur.

Args: data['img'] (Tensor): Float tensor of size (B, C, H, W) data['label'] (Tensor): Integer tensor of size (B, )

Args: batch (Tensor): Float tensor of size (B, C, H, W) target (Tensor): Integer tensor of size (B, ) Returns

:param x: [N D W H] :return: [N C], [N P C]

:param x: Shape: [b, 2048, h, w] :return: out: Feature maps. Shape: [b, 2048, h, w]

bottom1: 1st input, 4D Tensor of shape [batch_size, input_dim1, height, width]. bottom2: 2nd input, 4D Tensor of shape [batch_size, i

Perform convolution over the input using the squared L2 distance for all prototypes in the layer :param xs: A batch of input images o

Args: inputs (torch.Tensor): feature matrix with shape (batch_size, part_num, feat_dim). targets (torch.LongTensor):