List your company

Medical Reasoning on MedBullets, MedMCQA, and MedQA

54.22MedBullets Score+ WIST (Reward)
22.822430.973739.12547.2763Mar 22, 2026

Evaluation results

MethodLinks
+ WIST (Reward)Backbone=Qwen3-14B-Base2026.0354.2260.2962.6659.06
+ WIST (Random)Backbone=Qwen3-8B-Base2026.0350.1655.6357.2254.34
+ SPICEBackbone=Qwen3-14B-Base2026.0350.1656.8761.9656.33
+ WIST (Reward)Backbone=Qwen3-8B-Base2026.035056.3558.7355.03
+ WIST (Random)Backbone=Qwen3-14B-Base2026.035053.9662.355.42
+ R-ZeroBackbone=Qwen3-14B-Base2026.0346.4351.8361.4853.25
+ SPICEBackbone=Qwen3-8B-Base2026.0345.9454.3658.9453.08
Base ModelBackbone=Qwen3-14B-Base2026.0338.6448.5160.1549.1
+ R-ZeroBackbone=Qwen3-8B-Base2026.0334.5846.0754.5445.06
Base ModelBackbone=Qwen3-8B-Base2026.0330.0339.850.8740.24
+ SPICEBackbone=Qwen3-4B-Base2026.0324.3532.3952.6936.48
+ WIST (Reward)Backbone=Qwen3-4B-Base2026.0324.3532.3753.4236.71
+ WIST (Random)Backbone=Qwen3-4B-Base2026.0324.1932.352.7636.42
Base ModelBackbone=Qwen3-4B-Base2026.0324.0332.2748.5234.94
+ R-ZeroBackbone=Qwen3-4B-Base2026.0324.0332.2751.535.93