Load or restore offline algorithm

RLlib Offline RL
1

Low: It annoys or frustrates me for a moment.

When i am training an offline algorithm and saving its checkpoint afterwards it seems not possible to load it and continue to training with offline data. The resulting error is

AttributeError: ‘NoneType’ object has no attribute ‘sync_weights’ in restore_from_path() in algorithm.py

due to the fact that no env_runner_group is created when using offline data for training. If i want to do online training it works. Is this the desired behavior? if so, is there any work around to coninue offline training from a checkpoint?

thank you and kind regards
Philipp

2

@phil , thanks for raising this issue. I will check this week if there is a bug in the code. Usually it should revover the module from the learner.

3

Thank you very much! I would appreciate if you would let me know the result :slight_smile: