Continual reinforcement learning baselines: experiment specifications, implementation of existing methods, and common metrics. Easily extensible to new methods.
I just wanted to check that the replay buffer size for Clear (denoted "replay_buffer_frames" in the config files) is based on the number of transitions and not the number of unrolled trajectories?
Sorry for missing this! Yes, it is the number of transitions. It gets divided by the unroll length to determine how many unrolled trajectories to store.
I just wanted to check that the replay buffer size for Clear (denoted "replay_buffer_frames" in the config files) is based on the number of transitions and not the number of unrolled trajectories?
Thanks in advance!