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

Agent Interaction on Aggregate FTWP, ScienceWorld, WebShop

89.74Format Faithfulness Rate

Gemma2-9B + MISE

Updated 4d ago

Evaluation Results

Method	Links
Gemma2-9B + MISE 2026.04		89.74	28.07	23.25
Gemma2-9B + PRM 2026.04		88.9	26.63	21.45
Gemma2-9B + PPO 2026.04		86.42	25.87	20.67
GPT-4o 2026.04		82.13	40.4	38.29
LLaMA3-8B + PRM 2026.04		80.7	32.75	27.36
LLaMA3-8B + PPO 2026.04		78.9	28.2	22.33
LLaMA3-8B + MISE 2026.04		77.57	36.48	30.47
LLaMA3-8B + RFT 2026.04		77.47	31.29	26.1
Qwen2-7B + MISE 2026.04		73.14	32.12	26.22
GPT-4o-mini 2026.04		72.54	32.55	29.64
Qwen2-7B + PPO 2026.04		71.34	27.86	22.15
Qwen2-7B + PRM 2026.04		70.92	28.28	22.95
LLaMA3-8B + ReAct 2026.04		68.68	28.3	23.63
LLaMA3-8B 2026.04		67.83	28.92	23.06
LLaMA3-8B + online DPO 2026.04		67.47	29.54	24.4
Gemma2-9B 2026.04		61.23	9.93	9.74
Qwen2-7B 2026.04		44.85	19.6	15.72