List your company

Medical Knowledge Question Answering on Medical Domain (MedQA, MMLU, MedMCQA) (test)

92.5MedQA ScoreGPT-4.1 (25-04-14)
28.74845.29961.8578.401Jan 27, 2026Jan 28, 2026Jan 29, 2026Jan 30, 2026

Evaluation results

MethodLinks
GPT-4.1 (25-04-14)Params=N/A, Label=false, Knowledge=false2026.0192.597.381.390.4
DeepSeek-R1Params=671B, Label=false, Knowledge=false2026.01919978.789.6
Gemini-2.5-Pro (25-06-05)Params=N/A, Label=false, Knowledge=false2026.0189.798.182.190
Med-PaLM2Params=340B, Label=true, Knowledge=false2026.0185.49272.383.2
Llama3.1-InstructParams=70B, Label=false, Knowledge=false2026.0182.392.472.282.3
Qwen2.5-InstructParams=72B, Label=false, Knowledge=false2026.0179.694.168.680.8
Qwen2.5-InstructParams=32B, Label=false, Knowledge=false2026.0173.193.263.176.5
DGRC (Ours)Params=7B, Label=false, Knowledge=false2026.0172.890.967.577.1
RLAIF (GRPO)Params=7B, Label=false, Knowledge=false2026.0171.489.168.176.2
Baseline-w/-labelParams=7B, Label=true, Knowledge=false2026.0170.190.164.574.9
Baseline-w/o-labelParams=7B, Label=false, Knowledge=false2026.0169.487.264.373.7
Llama3.1-InstructParams=8B, Label=false, Knowledge=false2026.0165.880.556.567.6
MegaScienceParams=7B, Label=false, Knowledge=true2026.016187.157.468.5
DGRC (Ours)Params=3B, Label=false, Knowledge=false2026.0160.68159.767.1
RLAIF (GRPO)Params=3B, Label=false, Knowledge=false2026.0160.176.558.264.9
Qwen2.5-InstructParams=7B, Label=false, Knowledge=false2026.0159.886.655.667.3
MediTronParams=70B, Label=true, Knowledge=true2026.0159.871.553.361.5
Baseline-w/-labelParams=3B, Label=true, Knowledge=false2026.0159.177.357.664.7
Baseline-w/o-labelParams=3B, Label=false, Knowledge=false2026.0158.473.556.562.8
InstructDiffBackbone=LLaMA3-8B, Selection Ratio=10%2026.0154.6764.4850.1156.42
EntropyMaxBackbone=LLaMA3-8B, Selection Ratio=10%2026.0152.466.5252.1657.03
ZIPBackbone=LLaMA3-8B, Selection Ratio=10%2026.0151.5362.8748.8454.41
Inst/RespMaxBackbone=LLaMA3-8B, Selection Ratio=10%2026.0150.9858.7143.9251.2
DGRC (Ours)Params=1.5B, Label=false, Knowledge=false2026.0150.872.251.758.3
ALLBackbone=LLaMA3-8B, Selection Ratio=10%2026.0149.9661.0448.4153.14
Baseline-w/-labelParams=1.5B, Label=true, Knowledge=false2026.0149.872.249.257.1
RLAIF (GRPO)Params=1.5B, Label=false, Knowledge=false2026.0149.567.550.255.7
Inst LenMaxBackbone=LLaMA3-8B, Selection Ratio=10%2026.0149.2557.145.5250.62
Baseline-w/o-labelParams=1.5B, Label=false, Knowledge=false2026.0148.36548.954.1
PPLMinBackbone=LLaMA3-8B, Selection Ratio=10%2026.0148.2357.6943.5149.81
PPLMaxBackbone=LLaMA3-8B, Selection Ratio=10%2026.0147.7658.3845.3750.5
EntropyMidBackbone=LLaMA3-8B, Selection Ratio=10%2026.0146.4353.342.1947.31
SelectITBackbone=LLaMA3-8B, Selection Ratio=10%2026.0146.1146.8435.9342.96
RandomBackbone=LLaMA3-8B, Selection Ratio=10%2026.0145.9554.2240.1146.76
PPLMidBackbone=LLaMA3-8B, Selection Ratio=10%2026.0143.7548.4535.0942.43
Base ModelBackbone=LLaMA3-8B, Selection Ratio=0%2026.0143.651.0448.9847.87
IFDBackbone=LLaMA3-8B, Selection Ratio=10%2026.0143.1348.2742.3644.59
Resp LenMaxBackbone=LLaMA3-8B, Selection Ratio=10%2026.0142.1141.9932.4438.85
SuperfilteringBackbone=LLaMA3-8B, Selection Ratio=10%2026.0141.6341.1532.0838.29
Qwen2.5-InstructParams=3B, Label=false, Knowledge=false2026.0141.663.738.447.9
Qwen2.5-InstructParams=1.5B, Label=false, Knowledge=false2026.0140.154.837.144
MegaScienceParams=3B, Label=false, Knowledge=true2026.014073.344.352.5
EntropyMinBackbone=LLaMA3-8B, Selection Ratio=10%2026.0136.7642.2434.2337.74
Inst/RespMinBackbone=LLaMA3-8B, Selection Ratio=10%2026.0135.5144.6540.1640.11
MegaScienceParams=1.5B, Label=false, Knowledge=true2026.0131.261.236.543