README.md

mlfoundations-dev/a1_math_metamath_eval_1331

Precomputed model outputs for evaluation.

Evaluation Results

Summary

Metric	AIME24	AMC23	MATH500	GPQADiamond	MMLUPro	LiveCodeBench	CodeElo	JEEBench
Accuracy	13.7	58.0	74.2	39.6	28.8	9.8	2.5	34.4

AIME24

• Average Accuracy: 13.67% ± 1.29%
• Number of Runs: 10

Run	Accuracy	Questions Solved	Total Questions
1	13.33%	4	30
2	16.67%	5	30
3	13.33%	4	30
4	13.33%	4	30
5	13.33%	4	30
6	3.33%	1	30
7	13.33%	4	30
8	13.33%	4	30
9	16.67%	5	30
10	20.00%	6	30

AMC23

• Average Accuracy: 58.00% ± 2.17%
• Number of Runs: 10

Run	Accuracy	Questions Solved	Total Questions
1	57.50%	23	40
2	67.50%	27	40
3	57.50%	23	40
4	62.50%	25	40
5	50.00%	20	40
6	55.00%	22	40
7	70.00%	28	40
8	52.50%	21	40
9	60.00%	24	40
10	47.50%	19	40

MATH500

• Accuracy: 74.20%

  Accuracy	Questions Solved	Total Questions
  74.20%	371	500

GPQADiamond

• Average Accuracy: 39.56% ± 1.67%
• Number of Runs: 3

Run	Accuracy	Questions Solved	Total Questions
1	42.93%	85	198
2	39.90%	79	198
3	35.86%	71	198

MMLUPro

• Average Accuracy: 28.80% ± 0.00%
• Number of Runs: 1

Run	Accuracy	Questions Solved	Total Questions
1	28.80%	144	500

LiveCodeBench

• Average Accuracy: 9.78% ± 1.18%
• Number of Runs: 3

Run	Accuracy	Questions Solved	Total Questions
1	8.81%	45	511
2	8.41%	43	511
3	12.13%	62	511

CodeElo

• Average Accuracy: 2.47% ± 0.37%
• Number of Runs: 3

Run	Accuracy	Questions Solved	Total Questions
1	1.79%	7	391
2	3.07%	12	391
3	2.56%	10	391

JEEBench

• Average Accuracy: 34.42% ± 0.46%
• Number of Runs: 3

Run	Accuracy	Questions Solved	Total Questions
1	34.51%	177.75	515
2	35.34%	182.0	515
3	33.40%	172.0	515