MATH
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
MATH
44.8pass@1
6
MATH-500
84.78Pass@10
6
MATH 500
88.4Pass@20
6
MATH 500
88.8Maj Score
6
MATH500
34.8Exact-match Accuracy (Ngen=128)
6
MATH500
38.1Avg Diffusion Steps
6
MATH500
100Pass@16
6
math 1.0 (test)
90.1Overall Accuracy
6
MATH 500
63.87Selection Accuracy
6
MATH-500
2.57Error Rate
6
Math (50% 10% 40%)
75.13Accuracy
6
Math (40%/10%/50%)
74.79Accuracy
6
Math
57.5Win Rate
6
MATH500
40.6Accuracy
5
MATH500 @4 (test)
87.33Accuracy
5
MATH500
39.6Accuracy
5
MATH 500
95Sel-AUC
5
Math
87.6MTP Acceptance Rate
5
MATH
117.3Threshold Score
5
MATH
123.3Throughput (tokens/s)
5
MATH 500
61.6Accuracy
5
MATH
79.2Pass@1
5
MATH 500
62.7Pass@1
5
MATH-HARD
77.3Accuracy (MATH-HARD)
5
Math
80Accuracy
5