Porblem with dataset

[Yongze-zzz]

Dear author, thank you for sharing the excellent open source code, I ran into a problem when reproducing the experiments in your paper, when I run "./run.sh", it looks like I'm training the model, which is used by default It is syn dataset, where is the syn dataset saved? When I run "python3 -m subgraph_matching.train --dataset aids", it reports RuntimeError: unable to open shared memory object </torch_494089_134325577_997> in read-write mode: Too many open files (24)
[W CudaIPCTypes.cpp:15] Producer process has been terminated before all shared CUDA tensors released. See Note [Sharing CUDA tensors]