List your company

Procedure Planning on CrossTask T=3 (test)

59.35SRCLAD
-2.37413.650529.67545.6995May 4, 2022Dec 24, 2022Aug 15, 2023Apr 6, 2024Nov 26, 2024Jul 18, 2025Mar 10, 2026

Evaluation results

MethodLinks
CLADUse ground truth a0 and at=false2025.0359.3584.56—80.61
PDPPUse ground truth a0 and at=true2025.0350.9483.65—79.44
KEPPUse ground truth a0 and at=true2025.0347.9882.66—77.94
SCHEMAUse ground truth a0 and at=true2025.0347.0682.35——
ActionDiffusionUse ground truth a0 and at=true2025.0346.3582.12—77.48
LAPData curation setting=KEPP setting2026.0341.1470.13—65.98
ViterbiPlanNetParams (M)=5.572026.0338.4563.0783.89—
SCHEMAParams (M)=6.132026.0337.2462.6983.94—
PlanLLMParams (M)=384.942026.0336.8461.5683.23—
PDPPParams (M)=41.872026.0336.7361.9683.2—
LAP-TextData curation setting=KEPP setting, Input modality=text description2026.0335.8465.65—62.48
KEPPParams (M)=42.182026.0334.9360.3482.67—
PDPPData curation setting=KEPP setting2026.0333.961.39——
KEPPData curation setting=KEPP setting2026.0333.2760.16——
ActionDiffusionData curation setting=KEPP setting2026.0333.2561.47—65.09
SCHEMAData curation setting=KEPP setting2026.0332.3457.3——
Gemini 2.5 ProParams (M)=>100,0002026.0329.1857.981.48—
SkipPlanData curation setting=KEPP setting2026.0327.659.9——
Qwen2.5-32BParams (M)=32,0002026.0325.1456.180.92—
Qwen3-30BParams (M)=30,0002026.0323.3755.9681.16—
P3IVSupervision=L2022.0523.3449.9673.89—
Qwen3-30B + PKGParams (M)=30,000, PKG=true2026.0323.3156.1581.06—
PKG beam searchParams (M)=41.872026.0322.3855.7480.92—
Ext-GAILSupervision=V2022.0521.2749.4661.7—
Ext-GAIL w/o Aug.Supervision=V2022.0518.0143.8657.16—
DDNSupervision=V2022.0512.1831.2947.48—
Qwen2.5-VL-32BParams (M)=32,0002026.0311.4836.3569.52—
Retrieval-BasedSupervision=-2022.058.0523.332.06—
UPNSupervision=V2022.052.8924.3931.56—
UAAASupervision=-2022.052.1520.2130.87—
WLTDOSupervision=-2022.051.8721.6431.7—
RandomSupervision=-2022.050.010.941.66—
RandomUse ground truth a0 and at=true2025.0301.02—1.7