不同的cluster之间点不相连,共用1个邻接矩阵,不同cluster单独做max_pool# try:assertout_data.x.shape[0] %int(sub_data.time_step_len[0]) ==0# except:# from pdb import set_trace; set_trace()out_data.x = out_data.x / out_data.x.norm(dim=0)returnout_data...
cuPyNumeric enables a distributed implementation of TorchSWE that avoids the complexities of an MPI implementation. After porting TorchSWE to cuPyNumeric by removing all domain decomposition logic, it scaled effortlessly across multiple GPUs and nodes without further code modifications. This scalability ena...