About the pre-trained model GO-1

[cjxtu]

Is the pre-trained GO-1 model based on joint training or end-effector (eef) training? My own collected data is end-effector data and stored in the lerobot format. Will fine-tuning the GO-1 model be effective?

[ModiShi]

The pre-trained GO-1 model is based on joint-space training. However, fine-tuning with end-effector data is also effective which we have verified.