List your company

Agentic Action Prediction on ALFWorld (out domain)

89Success Rateours
42.254.3566.578.65Jun 25, 2026

Evaluation results

MethodLinks
oursModel=Qwen3-8B2026.0689
GRPOModel=Qwen3-8B2026.0686
oursModel=Qwen3-4B2026.0652
GRPOModel=Qwen3-4B2026.0644