MATH
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
MATH 500
48.5Average@10 Score
12
MATH500
160Training Steps to Target Accuracy
12
MATH 500
91Peak Accuracy
12
MATH500
87.93Accuracy
12
Math Family Average
67Accuracy
12
MATH-500 (test)
42.6Reference Score S
12
MATH
58.2Solve Rate (MATH)
12
MATH 500
93.6Accuracy
12
MATH500
69.7Accuracy @ 1
12
MATH
467Max Response Length
12
MATH
90.7Response Similarity
12
MATH500
31Accuracy
12
Math500
38.2Accuracy
12
MATH
89.6pass@1
12
MATH-500
79Mean Score (k=32)
12
MATH
48.24Score
12
MATH500
38.6Accuracy (MATH500)
12
MATH-500
7.52Speedup
12
Math
34.34Score
12
MATH500
88.8Accuracy
12
MATH 500
87.4Accuracy@4
12
MATH
3.3ECE
12
MATH500 4-shot max_gen_len=512
44.6Accuracy
12
MATH500
79.8Accuracy
12
MATH 500
93.8Accuracy (%)
12