README.md

mlfoundations-dev/fig1_math_openr1_eval_636d

Precomputed model outputs for evaluation.

Evaluation Results

Summary

Metric	AIME24	AMC23	MATH500	MMLUPro	JEEBench	GPQADiamond	LiveCodeBench	CodeElo	CodeForces
Accuracy	37.3	76.8	85.0	29.8	52.0	46.0	20.2	5.0	5.4

AIME24

• Average Accuracy: 37.33% ± 1.23%
• Number of Runs: 10

Run	Accuracy	Questions Solved	Total Questions
1	33.33%	10	30
2	40.00%	12	30
3	30.00%	9	30
4	33.33%	10	30
5	40.00%	12	30
6	36.67%	11	30
7	43.33%	13	30
8	36.67%	11	30
9	40.00%	12	30
10	40.00%	12	30

AMC23

• Average Accuracy: 76.75% ± 1.37%
• Number of Runs: 10

Run	Accuracy	Questions Solved	Total Questions
1	80.00%	32	40
2	77.50%	31	40
3	75.00%	30	40
4	75.00%	30	40
5	80.00%	32	40
6	85.00%	34	40
7	77.50%	31	40
8	70.00%	28	40
9	77.50%	31	40
10	70.00%	28	40

MATH500

• Accuracy: 85.00%

  Accuracy	Questions Solved	Total Questions
  85.00%	425	500

MMLUPro

• Average Accuracy: 29.80% ± 0.00%
• Number of Runs: 1

Run	Accuracy	Questions Solved	Total Questions
1	29.80%	149	500

JEEBench

• Average Accuracy: 52.04% ± 0.30%
• Number of Runs: 3

Run	Accuracy	Questions Solved	Total Questions
1	52.77%	271.75	515
2	51.60%	265.75	515
3	51.75%	266.5	515

GPQADiamond

• Average Accuracy: 45.96% ± 0.82%
• Number of Runs: 3

Run	Accuracy	Questions Solved	Total Questions
1	44.95%	89	198
2	47.98%	95	198
3	44.95%	89	198

LiveCodeBench

• Average Accuracy: 20.16% ± 0.74%
• Number of Runs: 3

Run	Accuracy	Questions Solved	Total Questions
1	20.35%	104	511
2	21.33%	109	511
3	18.79%	96	511

CodeElo

• Average Accuracy: 5.03% ± 0.23%
• Number of Runs: 3

Run	Accuracy	Questions Solved	Total Questions
1	5.37%	21	391
2	4.60%	18	391
3	5.12%	20	391

CodeForces

• Average Accuracy: 5.37% ± 0.64%
• Number of Runs: 3

Run	Accuracy	Questions Solved	Total Questions
1	5.52%	25	453
2	4.19%	19	453
3	6.40%	29	453