Would it be possible to benchmark the newly implemented algorithms, especially the off-policy ones like SAC and TD3? I had some bad experiences tuning the batch size and updating numbers, as IsaacGym can generate too many samples per iteration. It would be nice if you could provide some insight.
Dear RSL,
Would it be possible to benchmark the newly implemented algorithms, especially the off-policy ones like SAC and TD3? I had some bad experiences tuning the batch size and updating numbers, as IsaacGym can generate too many samples per iteration. It would be nice if you could provide some insight.
Thanks, Jin