训练时使用夹爪数据是用action/effector/position 还是state/effector/position

[zbzyjya]

感谢你们的开源贡献，
想请问下你们预训练时使用夹爪数据是用action/effector/position 还是state/effector/position
我注意到他们的数值在数据集中的表示不太一样
一个的值域为[0,1]另一个值域为[33,200]左右

[ModiShi]

We used /action/effector/position, though the choice is inconsequential—just ensure proper normalization.

[zbzyjya]

We used /action/effector/position, though the choice is inconsequential—just ensure proper normalization.

感谢