datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
qsim-shelf-restocking-mujoco-300
Qsim MuJoCo Shelf Restocking
300 seeded synthetic shelf-restocking episodes with the stock bimanual OpenArm
robot, physical gripper contacts, smooth scripted motion, and synchronized robot
head and wrist camera observations.
Release status: complete — 300/300 episodes passed the task and dataset checks.
600 placements; 542,518 observation/action samples; 108,640 RGB frames per camera; 3.01 hours of simulated trajectories; 9.68 GB of HDF5 episodes.
Arm selections: {'left': 297… See the full description on the dataset page: https://huggingface.co/datasets/qualiadev/qsim-shelf-restocking-mujoco-300.mujoco_act_53epsmujoco-gymnasium-v5-benchmark
MuJoCo Gymnasium v5 Benchmark
PPO evaluation benchmark on MuJoCo Gymnasium v5 locomotion and manipulation environments.
Tasks
Task ID
Name
hopper-v5
Hopper-v5
halfcheetah-v5
HalfCheetah-v5
walker2d-v5
Walker2d-v5
ant-v5
Ant-v5
humanoid-v5
Humanoid-v5
reacher-v5
Reacher-v5
Usage
Submit evaluation results to this benchmark by adding .eval_results/*.yaml files
to your model repo, referencing this dataset's task IDs.
© 2026 ParamTatva.org
lerobot_mujoco_so100mujoco_so100-2