Make current learner and KD-R-learner compatible/swappable

schuderer / bprl

Business Process Control using Reinforcement Learning. Work in progress.

MIT License

3 stars 0 forks source link

Make current learner and KD-R-learner compatible/swappable #5

Open schuderer opened 5 years ago

schuderer commented 5 years ago

also look into getting Stefano's RLACOSarsaLambda-Learner to run this environment

schuderer commented 5 years ago

Not really an agent thing, but did a few runs of the environment using stable-baselines' A2C (commit e994ff80c51d47283cd23c44c59c274019b96cd0), but the preliminary tests show no effective learning (10.000.000 timesteps).