List your company

Agentic Decision-Making on ALFWorld (Out-of-Distribution)

86.3Out-Success RateC-RF w/ NTF
67.99672.74877.582.252Jun 15, 2026

Evaluation results

MethodLinks
C-RF w/ NTFModel=Qwen2.5-7B-Instruct, Type=RL Training2026.0686.3
C-RF w/ NTFModel=Qwen2.5-1.5B-Instruct, Type=RL Training2026.0677.9
GRPOModel=Qwen2.5-7B-Instruct, Type=RL Training2026.0676.8
PPO (with critic)Model=Qwen2.5-7B-Instruct, Type=RL Training2026.0676.2
RLOOModel=Qwen2.5-7B-Instruct, Type=RL Training2026.0674
GRPOModel=Qwen2.5-1.5B-Instruct, Type=RL Training2026.0670.1
RLOOModel=Qwen2.5-1.5B-Instruct, Type=RL Training2026.0668.7