List your company

Embodied Agent Action Prediction on ALFWorld

65AccuracyDPO+
54.657.36062.7Jun 25, 2026

Evaluation results

MethodLinks
DPO+Backbone=Qwen3-8B2026.0665
SFT+Backbone=Qwen3-8B2026.0663
FRODOBackbone=Qwen3-8B2026.0661
DPOBackbone=Qwen3-8B2026.0659
SFTBackbone=Qwen3-8B2026.0655