4b3ca2794fe54a4b71ab2bdaea57cfd01c2f2849,models/model_utils.py,psi,forward,#psi#Any#,58

Before Change

        output = input.permute(0, 2, 3, 1)
        (batch_size, s_height, s_width, s_depth) = output.size()
        d_depth = s_depth * self.block_size_sq
        d_height = int(s_height / self.block_size)
        t_1 = output.split(self.block_size, 2)
        stack = [t_t.contiguous().view(batch_size, d_height, d_depth) for t_t in t_1]
        output = torch.stack(stack, 1)
        output = output.permute(0, 2, 1, 3)
        output = output.permute(0, 3, 1, 2)
        return output.contiguous()

After Change

        return input.reshape(bs, bl, bl, new_d, h, w).permute(0, 3, 4, 1, 5, 2).reshape(bs, new_d, h * bl, w * bl)

    def forward(self, input):
        bl, bl_sq = self.block_size, self.block_size_sq
        bs, d, new_h, new_w = input.shape[0], input.shape[1], input.shape[2] // bl, input.shape[3] // bl
        return input.reshape(bs, d, new_h, bl, new_w, bl).permute(0, 3, 5, 1, 2, 4).reshape(bs, d * bl_sq, new_h, new_w)

Instances