Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Math Reasoning

Benchmarks

Task NameDataset NameSOTA ResultTrend
Mathematical ReasoningMath Reasoning AIME24, AIME25, HMMT25
AIME24 Score78.4
30
Mathematical ReasoningMath Reasoning Overall
Mean@1663.8
24
Preference ModelingMath Reasoning
Accuracy87.6
20
Math ReasoningMath Reasoning Long Q, Long A (test)
Pass@10.65
15
Mathematical ReasoningMath Reasoning Out-domain (SVAMP, Mathematics, SimulEq) (test)
SVAMP Accuracy79.6
8
Mathematical ReasoningMath Reasoning In-domain (GSM8K, MATH, NumGLUE) (test)
GSM8K Accuracy69.1
8
Math ReasoningOverall Average Math Reasoning
Pass@154.54
6
Math ReasoningMath Reasoning Aggregate
Avg@3240.08
6
Math-ReasoningMath-Reasoning (test)
Gap20.83
4
Math-ReasoningMath-Reasoning (dev)
Gap24
4
Preference ClassificationMath Reasoning (test)
Classification Accuracy85.4
4
Math ReasoningMath Reasoning 1.5B model (val)
Validation Accuracy69.4
3
Mathematical Visual ReasoningMath-Reasoning I
Accuracy96.62
1
Showing 13 of 13 rows