List your company

Language Agent Task on WebShop (OOD)

49.6ScoreTask-Perturbed NLL Optimization
45.02446.21247.448.588Jun 25, 2026

Evaluation results

MethodLinks
Task-Perturbed NLL OptimizationBackbone=Qwen3-8B2026.0649.6
Task AugBackbone=Qwen3-8B2026.0647.5
SFTBackbone=Qwen3-8B2026.0647.1
COINBackbone=Qwen3-8B2026.0645.2