List your company

Multi-turn agent interaction on ScienceWorld

29.5ScoreGuided-OPD
4.217610.781317.34523.9087Jun 14, 2026

Evaluation results

MethodLinks
Guided-OPDParams=4B2026.0629.511.522.86
Zero-shotParams=30B-A3B2026.0627.737.2521.62
Vanilla OPDParams=4B2026.0626.457.523.88
TCODParams=4B2026.0625.869.2523.36
AdaSwitchParams=4B2026.0625.246.2524.48
Guided-OPDParams=1.7B2026.0623.245.2523.84
Guided-OPDParams=0.6B2026.0619.633.7524.63
TCODParams=1.7B2026.0619.083.2525.26
AdaSwitchParams=1.7B2026.0617.022.7526.45
Vanilla OPDParams=1.7B2026.0616.94325.27
Zero-shotParams=4B2026.0616.213.7515.97
TCODParams=0.6B2026.0615.71327.48
Vanilla OPDParams=0.6B2026.0610.421.526.87
AdaSwitchParams=0.6B2026.069.561.2529.43
Zero-shotParams=1.7B2026.066.480.7524.1
Zero-shotParams=0.6B2026.065.192.2520.26