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

Hallucination Evaluation on A-HaluEval

91.5A-Accuracy

Multi-DPOP

Updated 22d ago

Evaluation Results

Method	Links
Multi-DPOP 2025.10		91.5
Qwen3-30B 2025.10		91.5
Multi-DPOP 2025.10		86.5
Qwen3-4B 2025.10		86.5
Multi-DPOP 2025.10		85
Tulu-3.1-8B 2025.10		85
Multi-DPOP 2025.10		84
GLM-4-9B 2025.10		84
Multi-DPOP 2025.10		82.5
Gemma-3-4B 2025.10		82.5
GRPO 2025.10		61.4
Qwen3-30B 2025.10		61.4
DPO 2025.10		60.2
GRPO 2025.10		59.3
Qwen3-4B 2025.10		59.3
SFT 2025.10		58.9
GRPO 2025.10		58.3
Tulu-3.1-8B 2025.10		58.3
DPO 2025.10		58
Base 2025.10		57.6
Qwen3-30B 2025.10		57.6
GRPO 2025.10		57.4
GLM-4-9B 2025.10		57.4
DPO 2025.10		57.1
SFT 2025.10		56.4
DPO 2025.10		56.3
GRPO 2025.10		56.1
Gemma-3-4B 2025.10		56.1
SFT 2025.10		55.8
DPO 2025.10		55.2
Base 2025.10		55.2
Qwen3-4B 2025.10		55.2
SFT 2025.10		54.9
Base 2025.10		54.7
Tulu-3.1-8B 2025.10		54.7
Base 2025.10		53.8
GLM-4-9B 2025.10		53.8
SFT 2025.10		53.5
Base 2025.10		52.3
Gemma-3-4B 2025.10		52.3