Is icra_benchmark/rgl.py trained with ground truth position?

[zerosansan]

Hi,

I've ran training for all the benchmarks and it appears that rgl.py has the best results. Is this trained with ground truth positions (no state predictions)?

Regards,
Hafiq

[zafaralikhan9260]

Hello @zerosansan . Could you please lmk How did you trained the models other than RGL. ? Every time I am training the models using python train.py --policy rgl/sarl. It gives me the same result. i mean the same output cumulative reward curve.