xbpeng
xbpeng
thanks for pointing that out, that shouldn't be there. Removed.
those are not ready yet, this is just an initial commit of the code. We are still working on releasing the data and models.
We are still working on releasing the motion data. Those are not quite ready yet.
the motion data has been added to ase/data/motions
which character config did you use? you need to use amp_humanoid.xml for those clips since they don't have a sword and shield.
as mentioned in the readme, the model is trained using a dataset of multiple motion clips specified in `ase/data/motions/reallusion_sword_shield/dataset_reallusion_sword_shield.yaml`
yes that's intentional. Adding the mlp to for the latent seems to work a bit better for the actor, but didn't make much of a difference for the critic. So...
Yea, seems like the tasks are not working as well as I expected. There might be an issue with the retrained ASE model. Looking into it now. How many samples...
The task training issues should be fixed now. I've also uploaded new pre-trained models for the low-level controllers and high-level controllers.
It wasn't so much an issue with the pre-trained model. I just had to tweak the reward functions for the downstream tasks. Had to shape the reward a bit to...