import torchimport torch_mlu.core.mlu_model as ctct.set_core_number(16)ct.set_core_version("MLU270")torch.set_grad_enabled(False)a = torch.tensor(((1,2,3),(4,5,6)),dtype=torch.float32).to(ct.mlu_device())print(a.cpu())v,indx = a.topk(3,1,largest=False)#v,indx = a.topk(3,1,largest=True)print("v:",v.cpu())print("indx:",indx.cpu())[图片]对于二维以上的tensor设置largest=False,结果是从大到小排的设置largest= True,结果也是从大到小排的,和上面是一样的怎么让结果从小到大排??是我对largest的理解有错吗展开