Minerva
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
MINERVA 600+s (test)
31.8Accuracy
8
MINERVA 0-600s (test)
37.8Accuracy
8
MINERVA overall (test)
33.8Accuracy
8
Minerva
32Accuracy
7
Minerva
33.32Pass@1
7
Minerva
25.5Average@4
7
Minerva
66.5Accuracy
6
Minerva
37.4Accuracy
6
Minerva
39Pass@1
6
Minerva
60.29Pass@32
6
Minerva
37.58Avg@32
6
Minerva-Algebra
7.03TPF
5
Minerva
1.41Average Rank
5
MINERVA
65.6Accuracy (%)
5
Minerva Math
42.6Score
4
Minerva
19.1Pass@1
4
Minerva
67.6Score
4
Minerva
26.9Pass@1
3
Minerva
40.3Mean@16
3
Minerva 272 undergraduate-level STEM problems
56.57Avg@32 Score
3
Minerva OpenR1-Math Harder
48.9Accuracy
3
Minerva
49.6Avg@16 Accuracy
2
Minerva
272Total Score
2
Minerva
3.65Tau (Accepted Tokens)
2
Minerva
24.6Average Score (@32)
2