README.md

mlfoundations-dev/a1_math_gair_math_eval_636d

Precomputed model outputs for evaluation.

Evaluation Results

Summary

Metric	AIME24	AMC23	MATH500	MMLUPro	JEEBench	GPQADiamond	LiveCodeBench	CodeElo	CodeForces
Accuracy	21.7	64.2	79.4	32.4	43.8	45.5	18.5	7.5	8.7

AIME24

• Average Accuracy: 21.67% ± 1.35%
• Number of Runs: 10

Run	Accuracy	Questions Solved	Total Questions
1	16.67%	5	30
2	20.00%	6	30
3	23.33%	7	30
4	23.33%	7	30
5	16.67%	5	30
6	26.67%	8	30
7	23.33%	7	30
8	20.00%	6	30
9	16.67%	5	30
10	30.00%	9	30

AMC23

• Average Accuracy: 64.25% ± 1.06%
• Number of Runs: 10

Run	Accuracy	Questions Solved	Total Questions
1	57.50%	23	40
2	67.50%	27	40
3	65.00%	26	40
4	60.00%	24	40
5	67.50%	27	40
6	62.50%	25	40
7	67.50%	27	40
8	65.00%	26	40
9	67.50%	27	40
10	62.50%	25	40

MATH500

• Accuracy: 79.40%

  Accuracy	Questions Solved	Total Questions
  79.40%	397	500

MMLUPro

• Average Accuracy: 32.40% ± 0.00%
• Number of Runs: 1

Run	Accuracy	Questions Solved	Total Questions
1	32.40%	162	500

JEEBench

• Average Accuracy: 43.83% ± 1.28%
• Number of Runs: 3

Run	Accuracy	Questions Solved	Total Questions
1	46.31%	238.5	515
2	44.27%	228.0	515
3	40.92%	210.75	515

GPQADiamond

• Average Accuracy: 45.45% ± 1.72%
• Number of Runs: 3

Run	Accuracy	Questions Solved	Total Questions
1	49.49%	98	198
2	42.42%	84	198
3	44.44%	88	198

LiveCodeBench

• Average Accuracy: 18.46% ± 1.65%
• Number of Runs: 3

Run	Accuracy	Questions Solved	Total Questions
1	16.44%	84	511
2	17.22%	88	511
3	21.72%	111	511

CodeElo

• Average Accuracy: 7.50% ± 0.96%
• Number of Runs: 3

Run	Accuracy	Questions Solved	Total Questions
1	7.42%	29	391
2	5.88%	23	391
3	9.21%	36	391

CodeForces

• Average Accuracy: 8.68% ± 0.78%
• Number of Runs: 3

Run	Accuracy	Questions Solved	Total Questions
1	8.39%	38	453
2	7.51%	34	453
3	10.15%	46	453