README.md

mlfoundations-dev/reasoning_6_chat_eval_636d

Precomputed model outputs for evaluation.

Evaluation Results

Summary

Metric	AIME24	AMC23	MATH500	MMLUPro	JEEBench	GPQADiamond	LiveCodeBench	CodeElo	CodeForces
Accuracy	28.3	69.0	83.8	28.4	49.8	38.9	37.3	10.4	15.7

AIME24

• Average Accuracy: 28.33% ± 1.72%
• Number of Runs: 10

Run	Accuracy	Questions Solved	Total Questions
1	30.00%	9	30
2	23.33%	7	30
3	30.00%	9	30
4	26.67%	8	30
5	33.33%	10	30
6	20.00%	6	30
7	36.67%	11	30
8	33.33%	10	30
9	20.00%	6	30
10	30.00%	9	30

AMC23

• Average Accuracy: 69.00% ± 1.47%
• Number of Runs: 10

Run	Accuracy	Questions Solved	Total Questions
1	67.50%	27	40
2	75.00%	30	40
3	62.50%	25	40
4	70.00%	28	40
5	60.00%	24	40
6	70.00%	28	40
7	75.00%	30	40
8	70.00%	28	40
9	72.50%	29	40
10	67.50%	27	40

MATH500

• Accuracy: 83.80%

  Accuracy	Questions Solved	Total Questions
  83.80%	419	500

MMLUPro

• Average Accuracy: 28.40% ± 0.00%
• Number of Runs: 1

Run	Accuracy	Questions Solved	Total Questions
1	28.40%	142	500

JEEBench

• Average Accuracy: 49.84% ± 0.29%
• Number of Runs: 3

Run	Accuracy	Questions Solved	Total Questions
1	50.24%	258.75	515
2	49.13%	253.0	515
3	50.15%	258.25	515

GPQADiamond

• Average Accuracy: 38.89% ± 1.33%
• Number of Runs: 3

Run	Accuracy	Questions Solved	Total Questions
1	35.86%	71	198
2	41.41%	82	198
3	39.39%	78	198

LiveCodeBench

• Average Accuracy: 37.31% ± 0.91%
• Number of Runs: 3

Run	Accuracy	Questions Solved	Total Questions
1	35.81%	183	511
2	38.94%	199	511
3	37.18%	190	511

CodeElo

• Average Accuracy: 10.40% ± 0.67%
• Number of Runs: 3

Run	Accuracy	Questions Solved	Total Questions
1	11.51%	45	391
2	10.49%	41	391
3	9.21%	36	391

CodeForces

• Average Accuracy: 15.67% ± 0.25%
• Number of Runs: 3

Run	Accuracy	Questions Solved	Total Questions
1	16.11%	73	453
2	15.67%	71	453
3	15.23%	69	453