MATH
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
MATH
55.9Multi-step pass@5 Accuracy
16
MATH (test)
51.8Hit@1
16
MATH 519 problems (test)
76.75Accuracy
16
MATH 500
91.6Accuracy
16
MATH 500
78.2Accuracy (M→M-500)
16
MATH (test)
76.1Exact Match (EM)
16
MATH 500
82.1Avg@2
16
MATH500
82.2Pass@1 Accuracy
16
MATH-500
99.4Pass@5 Rate
16
MATH
8,849.7Avg Context Length (tokens)
16
MATH 500
67Pass@1
15
Math500
82.25Pass@1
15
MATH-500 simple-numeric-answer (Level 5)
96Accuracy
15
MATH
90Pass@1 Accuracy
15
MATH500
78.8Pass@1
15
MATH500
96.4Accuracy
15
MATH 500
89.05Accuracy (Avg@8)
15
Math500
83.1Average Accuracy @16
15
MATH 500
74.95Avg@k
15
MATH 500
93.5Accuracy
15
MATH 500
90.4MDR
15
MATH 500
92.51Pass@1 Rate
15
MATH 500
95.4Accuracy
15
MATH-500 (val)
90.07Avg@5 Accuracy
15
MATH500
0.778VBAL Score
15