Emanuele Pesce

Results 6 comments of Emanuele Pesce

> > > https://github.com/shariqiqbal2810/multiagent-particle-envs > > Shariq has added walls in his version of MPE. That's definitely great. Is there any way to make the walls visible to the agents,...

@Kimonili to limit the agents from running to infinity you can use the wall class defined by Shariq ([multiagent](https://github.com/shariqiqbal2810/multiagent-particle-envs/blob/master/multiagent/core.py)) as suggested by @HassamSheikh .

Did you try to use the relu with the batch normalization? It looks that now NaN have disappeared.

If I change relu to tanh, the training has been running for almost 3 days and NaN have not appeared (neither in the actions or rewards). Looks like it also...