List your company

Knowledge gap detection on MATH

89.1Accuracy (Knowledge Gap)GRADEpos
38.03651.29364.5577.807Apr 3, 2026

Evaluation results

MethodLinks
GRADEposBackbone=Llama3.1-8b2026.0489.196.2
GRADEposBackbone=Qwen2.5-7b2026.0487.992.1
GRADEposBackbone=Gemma2-9b2026.0487.193.4
GRADEpreBackbone=Qwen2.5-7b2026.0480.487.9
GRADEpreBackbone=Gemma2-9b2026.0479.481.3
GRADEpreBackbone=Llama3.1-8b2026.047784.1
Align-PBackbone=Llama3.1-8b2026.0475.360.7
Align-PBackbone=Qwen2.5-7b2026.0474.779.7
ICBackbone=Gemma2-9b2026.0471.575.7
P-EntropyBackbone=Qwen2.5-7b2026.0470.254.2
P-EntropyBackbone=Gemma2-9b2026.0469.175.1
ICBackbone=Qwen2.5-7b2026.0467.672.2
JudgeBackbone=Qwen2.5-7b2026.0466.2—
P-EntropyBackbone=Llama3.1-8b2026.0466.170.4
Align-PBackbone=Gemma2-9b2026.0461.363.6
ICBackbone=Llama3.1-8b2026.0457.562.7
JudgeBackbone=Llama3.1-8b2026.0450.1—
JudgeBackbone=Gemma2-9b2026.0440—