List your company

Math Problem Solving on Aggregated Benchmarks (AIME24, Math-OAI, Minerva, Olympiad, ACM23)

7.71AIME24 ScoreInstructDiff
1.40763.04384.686.3162Jan 30, 2026

Evaluation results

MethodLinks
InstructDiffBackbone=Qwen2.5-7B, Selection Ratio=10%2026.017.7161.7921.4226.9440.3131.63
PPLMidBackbone=Qwen2.5-7B, Selection Ratio=10%2026.016.0554.7613.6121.0132.8125.65
EntropyMinBackbone=Qwen2.5-7B, Selection Ratio=10%2026.015.8360.7418.5927.1336.7229.8
PPLMinBackbone=Qwen2.5-7B, Selection Ratio=10%2026.015.6357.816.3523.6831.7227.04
RandomBackbone=Qwen2.5-7B, Selection Ratio=10%2026.015.4355.4114.8620.1833.1325.8
ALLBackbone=Qwen2.5-7B, Selection Ratio=100%2026.01557.9317.8322.332.1927.05
SuperfilteringBackbone=Qwen2.5-7B, Selection Ratio=10%2026.014.855.5514.0621.3830.7825.31
Inst LenMaxBackbone=Qwen2.5-7B, Selection Ratio=10%2026.014.7954.9812.2720.9929.5324.51
SelectITBackbone=Qwen2.5-7B, Selection Ratio=10%2026.014.1552.8615.0620.2929.3824.35
ZIPBackbone=Qwen2.5-7B, Selection Ratio=10%2026.014.1550.312.2316.693022.61
PPLMaxBackbone=Qwen2.5-7B, Selection Ratio=10%2026.013.9550.5612.7718.5728.2822.83
Inst/RespMinBackbone=Qwen2.5-7B, Selection Ratio=10%2026.013.7452.512.2720.1427.9723.32
EntropyMidBackbone=Qwen2.5-7B, Selection Ratio=10%2026.013.5454.512.2720.1229.3823.96
Inst/RespMaxBackbone=Qwen2.5-7B, Selection Ratio=10%2026.013.3350.3312.0816.9628.2822.2
EntropyMaxBackbone=Qwen2.5-7B, Selection Ratio=10%2026.013.1250.512.4118.5423.5921.63
IFDBackbone=Qwen2.5-7B, Selection Ratio=10%2026.013.1250.3613.9818.1424.3821.99
Resp LenMaxBackbone=Qwen2.5-7B, Selection Ratio=10%2026.012.2844.6510.0215.3821.5618.78
Base ModelBackbone=Qwen2.5-7B2026.011.6528.799.267.6915.6512.61