← Leaderboard

ChatVLA

VLA
Midea Group & ECNU · Feb 20, 2025

VLA trained to keep multimodal chat and understanding ability alongside control (MMStar 47.2), using phased alignment and a mixture-of-experts. Real-robot results are self-reported.

Type
VLA
Parameters
—
Backbone
Qwen2-VL-2B with mixture-of-experts
Weights
Open
License
MIT
Commercial use
Unclear
Released
Feb 20, 2025
Robots
–
SO-101
–
LeRobot
–
Inputs
–
VRAM
–

Benchmark results

0 of 13 benchmarks

No benchmark results recorded yet.

Not reported on 13 benchmarks
Wrong or missing number? Report an error