Share your thoughts, 1 month free Claude Pro on usSee more

Interactive Environment Task Completion on ALFWorld (Seen)

90.2Average Reward

EAGLET + GPT-5

Updated 15d ago

Evaluation Results

Method	Links
EAGLET + GPT-5 2025.10		90.2
EAGLET + GiGPO 2025.10		88.6
MPO + GPT-5 2025.10		88.2
GPT-5 2025.10		87.9
BPO 2025.08		87.9
EAGLET + ETO 2025.10		87.3
MPO + GiGPO 2025.10		86.6
GiGPO 2025.10		85.2
MPO + ETO 2025.10		85
EAGLET + GPT-4.1 2025.10		84.3
MPO 2025.08		82.9
EAGLET + AgentTuning 2025.10		82.3
DeepSeek-V3.1-Think 2025.10		81.4
MPO + GPT-4.1 2025.10		81.4
MPO + AgentTuning 2025.10		80.7
KnowAgent 2025.10		80
SFT 2025.08		80
AgentTuning 2025.10		79.3
GPT-4.1 2025.10		78.6
ETO 2025.08		78.6
WKM 2025.10		77.5
ETO 2025.10		77.1
Qwen-2.5-7B-Instruct 2025.08		72.1
Deepseek-R1 2025.08		61.4
o3-mini 2025.08		58.9
EAGLET + Llama-3.1-8B-Instruct 2025.10		54.3
DeepSeek-V3.1-Non-Think 2025.10		50
MPO + Llama-3.1-8B-Instruct 2025.10		50
Qwen-3-Thinking 2025.08		49.3
Llama-3.1-8B-Instruct 2025.08		32.9
Llama-3.1-8B-Instruct 2025.10		22.9