Open freebooterish opened 11 months ago
Well, I'm not sure whether torch2 can work for our project as we adopt pytorch 1.13.1. In fact, we have never met the problems shown in the log file. In fact, we provide a shell file called start_all.sh in ./learner_torch directory and we recommend to use it to start the training. Also, our code are executed using dockers, a learner docker and multiple actor dockers. I'm not sure whether other settings can run the code. As for the second point, as the gamecore we use only supports command input, I think it's not feasible to run in graphical interface. If you want to participate in the game, you can change the code of one actor as each actor sends a json command to the gamecore client.
I'm sorry to bother you again. 1、After testing, I may not be seeing the hoped for results, if it's convenient please give me some hints, I followed the readme.md execution:
2、How can I see the results of the run in a graphical interface and can I participate in the game, thank you very much!