Help getting started with "Transfer Learning Between Different Games via Generative Models"


#1

Related to: [Transfer Learning Between Different Games via Generative Models]

So I plan to use the openAI baseline, probably A2C, to learn on 10 Atari games and save 10000 trajectories from the policy for each game. I am just getting started with gym and was wondering if there was already example code of loading a model and generating and saving these trajectories for a given game.

Also if anyone knows of any code resources where the trajectories are loaded in to fit a generative model please let me know as well.