MATH
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
MATH
46.74Accuracy
5
MATH GSM8K → Comp.
3.6ECE
5
MATH 500
53Accuracy
5
MATH 500
98.4Pass@12
5
MATH Minerva
55.6Score
5
MATH 500
92.16Pass@1
5
MATH-500
89.9Accuracy
5
Math500
97.8Pass@1
5
MATH levels 3–5 50/50 calibration (test)
81.2CR (%)
5
MATH
13.8Accuracy
5
Math500
80.4Pass@1
5
MATH en
49.46Accuracy
5
MATH-500
89.8Accuracy
5
MATH-500
83.2VAcc (Verification Accuracy)
5
MATH
87.48avg.@8 Score
5
MATH
10.2Accuracy Improvement Delta
5
MATH
56.75Acc (mean@8)
5
MATH 500
5.94TPF
5
MATH
92.8Exact Match (EM)
5
Math
30Average Iterations
5
MATH
82.98Accuracy
5
MATH
17.6Latency (MATH)
5
MATH
54.71Mean@16
5
MATH 500
62.6Accuracy Pass@1
5
MATH 500
1.4Avg Rank
5