how to set a low batch size value for running eval_sim.eval_rdt_maniskill

[amiltonwong]

Hi, all,

When I run python -m eval_sim.eval_rdt_maniskill --pretrained_path data/maniskill-model/mp_rank_00_model_states.pt, I got CUDA out of memory issue. How could I set a low batch size value for running eval_sim.eval_rdt_maniskill ?

Thanks~