NTS-Net复现出bug

最近看这篇论文《Learning to Navigate for Fine-grained Classification》，下载了源码想要复现，下载的cuda是9.2版本，python3.7，把鸟类数据集放进去之后，运行train.py,开始报错

top_n_prob = torch.gather(rpn_score, dim=1, index=top_n_index)
RuntimeError: Expected object of scalar type Long but got scalar type Int for argument #3 'index' in call to _th_gather_out

把top_n_index改为long之后，直接退出：Process finished with exit code -1073741819 (0xC0000005)

有无大佬指点一下