• I have marked all applicable categories:
  • exception-raising bug
  • RL algorithm bug
  • documentation request (i.e. "X is missing from the documentation.")
  • new feature request
• I have visited the source website
• I have searched through the issue tracker for duplicates
• I have mentioned version numbers, operating system and environment, where applicable:

  import tianshou, torch, numpy, sys
  print(tianshou.__version__, torch.__version__, numpy.__version__, sys.version, sys.platform)
  0.4.1 1.8.1+cpu 1.20.2 3.8.5 (default, Sep  3 2020, 21:29:08) [MSC v.1916 64 bit (AMD64)] win32

Hi, @Trinkle23897
From the usage here, it seems the nx.Graph() cannot be used as before. May I ask that, how canI still use this feature, since directly saving the graph object is more efficient, instead of the edge_list array.

Maybe in the future, tianshou can support the dgl.graph object. It stores the tensor internally.

Thanks!

这个例子可能现在不太能work，因为numpy升级了，以及nx.Graph重写了__getitem__，导致np.array([nx.Graph()])会出来空的> 数组……
不过正常的自定义class应该没啥问题
import networkx as nx
b = ReplayBuffer(size=3)
b.add(obs=nx.Graph(), act=0, rew=0, done=0)
print(b)