List your company

Continual Android Device Control on AndroidControl-CL Task Order 1

81.73Shopping AccuracySFT-Joint-Training
46.671655.773364.87573.9767Mar 3, 2026

Evaluation results

MethodLinks
SFT-Joint-TrainingBackbone=QwenVL2.5-3b, Task Order=12026.0381.7386.7485.7582.2488.177782.7683.4841.66—
OursBackbone=QwenVL2.5-3b, Task Order=12026.0380.3285.4681.9982.0388.1775.8182.5382.3338.03-0.02
GRPOBackbone=QwenVL2.5-3b, Task Order=12026.0379.6384.7483.4381.3485.7474.0881.7381.5336.78-0.62
RIF-RFTBackbone=QwenVL2.5-3b, Task Order=12026.0379.5383.3580.7780.2585.947380.2580.4432.91-0.58
SFT+KLBackbone=QwenVL2.5-3b, Task Order=12026.0379.383.8582.3279.8485.8374.0880.7180.8434.69-1.01
SFT+ReplayBackbone=QwenVL2.5-3b, Task Order=12026.0374.6183.3581.6679.0285.6172.2682.0879.830.19-3.11
SFTBackbone=QwenVL2.5-3b, Task Order=12026.0370.5479.5278.976.4183.2469.2480.4276.923.53-5.73
Zero-ShotBackbone=QwenVL2.5-3b, Task Order=12026.0348.0257.7149.552.8155.8742.3348.8650.731.86—