List your company

Mathematical Reasoning on HMMT 2024

23.5Pass@1 AccuracyOPD
1.0366.86812.718.532May 8, 2026

Evaluation results

MethodLinks
OPDStudent-Teacher Pair=DeepSeek-R1-Distill-Qwen-1.5B / JustRL-DeepSeek-1.5B2026.0523.56.90—
PRUNE-OPD (top-p)Student-Teacher Pair=DeepSeek-R1-Distill-Qwen-1.5B / JustRL-DeepSeek-1.5B2026.0523.34.337.7—
OPD (Truncate 4k)Student-Teacher Pair=DeepSeek-R1-Distill-Qwen-1.5B / JustRL-DeepSeek-1.5B2026.0522.74.830.4—
PRUNE-OPD (overlap)Student-Teacher Pair=DeepSeek-R1-Distill-Qwen-1.5B / JustRL-DeepSeek-1.5B2026.0522.74.140.6—
PRUNE-OPD (top-p)Student-Teacher Pair=DeepSeek-R1-Distill-Qwen-1.5B / DeepSeek-R1-Distill-Qwen-7B2026.0514.64.861.6—
PRUNE-OPD (overlap)Student-Teacher Pair=DeepSeek-R1-Distill-Qwen-1.5B / DeepSeek-R1-Distill-Qwen-7B2026.0514.6468—
OPDStudent-Teacher Pair=DeepSeek-R1-Distill-Qwen-1.5B / DeepSeek-R1-Distill-Qwen-7B2026.0514.412.50—
OPD (Truncate 4k)Student-Teacher Pair=DeepSeek-R1-Distill-Qwen-1.5B / DeepSeek-R1-Distill-Qwen-7B2026.0514.26.647.2—
PRUNE-OPD (overlap)Student-Teacher Pair=Qwen3-4B-Base / Qwen3-4B (Non-thinking)2026.056.713.552.6—
OPDStudent-Teacher Pair=Qwen3-4B-Base / Qwen3-4B (Non-thinking)2026.055.428.50—
OPD (Truncate 4k)Student-Teacher Pair=Qwen3-4B-Base / Qwen3-4B (Non-thinking)2026.054.822.321.8—
PRUNE-OPD (top-p)Student-Teacher Pair=Qwen3-4B-Base / Qwen3-4B (Non-thinking)2026.054.82029.8—
PRUNE-OPD (overlap)Student-Teacher Pair=Qwen3-1.7B-Base / Qwen3-4B (Non-thinking)2026.052.910.337.6—
OPD (Truncate 4k)Student-Teacher Pair=Qwen3-1.7B-Base / Qwen3-4B (Non-thinking)2026.052.312.921.8—
OPDStudent-Teacher Pair=Qwen3-1.7B-Base / Qwen3-4B (Non-thinking)2026.052.116.50—
PRUNE-OPD (top-p)Student-Teacher Pair=Qwen3-1.7B-Base / Qwen3-4B (Non-thinking)2026.051.911.530.3—
OPDTeacher-Student Pair=DeepSeek-R1-Distill-Qwen-7B / Skywork-OR1-7B, Time=17.5h, Red.=0%2026.05———33.5
OPD (Truncate 4k)Teacher-Student Pair=DeepSeek-R1-Distill-Qwen-7B / Skywork-OR1-7B, Time=11.3h, Red.=35.4%2026.05———31
PRUNE-OPD (overlap)Teacher-Student Pair=DeepSeek-R1-Distill-Qwen-7B / Skywork-OR1-7B, Time=17.0h, Red.=2.9%2026.05———33.1