search

PWhiddy / PokemonRedExperiments

Playing Pokemon Red with Reinforcement Learning
MIT License
7k stars 645 forks source link

Want to save Actions and Frames of the RL #188

Open M-Nusrat-Ullah opened 1 month ago

M-Nusrat-Ullah commented 1 month ago

Hello, I have recently found out about this RL. This is really a great project. I am doing some experiment on this. I am trying to save the frames, actions and episodes of the RL taking in every parallel processing. What is the easiest way to achieve this? My current approach is: I am modifying the code in StreamWrapper class, saving frames, actions etc. At the end, I am just saving them in a file. Is there any better way to achieve this? Any type of advice is appreciable. Thanks in advance.

PWhiddy commented 3 weeks ago

Hello 👋 That seems like a fine approach. Technically it could be cleaner to create a separate wrapper to save this data. It would also be fine to just modify the base environment to save it as well.