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

Medical on HealthBench-500

43.6Score

HeRL

Updated 4mo ago

Evaluation Results

Method	Links
HeRL 2026.03		43.6
Qwen3-4B-Instruct-2507 2026.03		42
RLVR 2026.03		41.7
DPO 2026.03		39.1
SFT 2026.03		36
HeRL 2026.03		34.3
RLVR 2026.03		30.5
DPO 2026.03		28
SFT 2026.03		27.2
HeRL 2026.03		26.6
Qwen2.5-7B-Instruct 2026.03		24.4
SFT 2026.03		21
RLVR 2026.03		17.8
Llama-3.2-3B-Instruct 2026.03		14.5
DPO 2026.03		13.5