List your company

Embodied Household Tasks on ALFWorld Seen

90.7Average Accumulated RewardQ-Evolve
-3.62820.86145.3569.839Jun 5, 2026

Evaluation results

MethodLinks
Q-EvolveBackbone=Llama-2-7B-Chat2026.0690.7
QLASSBackbone=Llama-2-7B-Chat2026.0677.9
ETOBackbone=Llama-2-7B-Chat2026.0668.6
RFTBackbone=Llama-2-7B-Chat2026.0662.9
Best-of-NBackbone=Llama-2-7B-Chat, N=62026.0662.1
SFTBackbone=Llama-2-7B-Chat2026.0660
ReflexionPrompting=ReAct2026.0645.7
GPT-4Prompting=ReAct2026.0642.9
PPOBackbone=Llama-2-7B-Chat2026.0622.1
GPT-3.5-TurboPrompting=ReAct2026.067.9
Base AgentBackbone=Llama-2-7B-Chat2026.060