← Leaderboard

Vidar

World-action
Tsinghua University & ShengShu · Jul 17, 2025

Video-diffusion world model paired with a masked inverse-dynamics adapter that turns generated video into actions, aimed at low-data adaptation to new robots. It self-reports RoboTwin 2.0 results.

Type
World-action
Parameters
—
Backbone
Video diffusion (Vidu 2.0 on real robots, Wan2.2 in sim) + masked inverse dynamics
Weights
Unknown
License
–
Commercial use
Unclear
Released
Jul 17, 2025
Robots
Aloha-AgileX (sim)
SO-101
–
LeRobot
–
Inputs
–
VRAM
–

Benchmark results

1 of 13 benchmarks
RoboTwin 2.0Near saturationSimulation
  • 41.6%, self-reported by the authors, not a leaderboard entryPaper / model card ↗Jul 2025· self-reported· n = 10000· non-standard
41.6%, self-reported by the authors, not a leaderboard entrynot ranked
Not reported on 12 benchmarks
Wrong or missing number? Report an error