better model

[nutsintheshell]

I 'm using your quick demo. I wonder is there a better model can be open. I really need a more rubust and smooth model.Thanks.

[cypypccpy]

Hi @nutsintheshell, could you provide more details about your requirement? The checkpoint we released is the one we used to obtain the results in the paper. Some motions, such as digging, might seem jerky, but this is what the RL algorithm found to work best. You can fine-tune our checkpoint and include a smoothing reward function term to achieve the behavior you want.

[nutsintheshell]

Hi @nutsintheshell, could you provide more details about your requirement? The checkpoint we released is the one we used to obtain the results in the paper. Some motions, such as digging, might seem jerky, but this is what the RL algorithm found to work best. You can fine-tune our checkpoint and include a smoothing reward function term to achieve the behavior you want.

well, thanks for your reply.