what did it means using self.aet=True

[cqray1990]

        else:
            beziers2 = [p.top_feat for p in proposals]
            for k in range(len(targets)):
                rec_assign = [int(torch.argmin(torch.abs(beziers[k] - beziers2[k][i]).sum(dim=1))) for i in range(len(beziers2[k]))]
                targets[k] = torch.cat([targets[k].text, targets[k].text[rec_assign]], dim = 0)
            targets = torch.cat([x for x in targets], dim = 0)
            cat_beziers = []
            for ix in range(len(beziers)):
                cat_beziers.append(cat((beziers[ix], beziers2[ix]), dim=0))
            beziers = cat_beziers

[km1562]

see the papre, the paper had said that