List your company

Mathematical Reasoning on AIME 2024 1

30AccuracyTA-OPD
0.53688.185915.83523.4841May 26, 2026

Evaluation results

MethodLinks
TA-OPDSetting=Qwen3-8B (GRPO) -> Qwen3-4B, Supervised-token budget=10%2026.0530
TA-OPD+Ent.Setting=Qwen3-8B (GRPO) -> Qwen3-4B, Supervised-token budget=10%2026.0530
EntropySetting=Qwen3-8B (GRPO) -> Qwen3-4B, Supervised-token budget=10%2026.0526.33
TIPSetting=Qwen3-8B (GRPO) -> Qwen3-4B, Supervised-token budget=10%2026.0526.33
Pure OPDSetting=Qwen3-14B -> Qwen3-4B, Supervised-token budget=10%2026.0523.34
TA-OPDSetting=Qwen3-14B -> Qwen3-4B, Supervised-token budget=10%2026.0523.34
EntropySetting=Qwen3-14B -> Qwen3-4B, Supervised-token budget=10%2026.0521.67
TA-OPD+Ent.Setting=Qwen3-14B -> Qwen3-4B, Supervised-token budget=10%2026.0521.67
BaseSetting=Qwen3-8B (GRPO) -> Qwen3-4B, Supervised-token budget=10%2026.0521.66
BaseSetting=Qwen3-14B -> Qwen3-4B, Supervised-token budget=10%2026.0521.66
TA-OPDSetting=Qwen3-4B -> Qwen3-1.7B, Supervised-token budget=10%2026.0520
Pure OPDSetting=Qwen3-8B (GRPO) -> Qwen3-4B, Supervised-token budget=10%2026.0520
TIPSetting=Qwen3-14B -> Qwen3-4B, Supervised-token budget=10%2026.0520
TIPSetting=Qwen3-4B -> Qwen3-1.7B, Supervised-token budget=10%2026.0518.34
TA-OPD+Ent.Setting=Qwen3-4B -> Qwen3-1.7B, Supervised-token budget=10%2026.0515
EntropySetting=Qwen3-4B -> Qwen3-1.7B, Supervised-token budget=10%2026.0513.34
Pure OPDSetting=Qwen3-4B -> Qwen3-1.7B, Supervised-token budget=10%2026.0511.67
BaseSetting=Qwen3-4B -> Qwen3-1.7B, Supervised-token budget=10%2026.0510
TA-OPD+Ent.Setting=DeepSeek-R1-Distill-Qwen-14B -> Qwen2.5-3B, Supervised-token budget=10%2026.058.34
Pure OPDSetting=DeepSeek-R1-Distill-Qwen-14B -> Qwen2.5-3B, Supervised-token budget=10%2026.055
TA-OPDSetting=DeepSeek-R1-Distill-Qwen-14B -> Qwen2.5-3B, Supervised-token budget=10%2026.055
BaseSetting=DeepSeek-R1-Distill-Qwen-14B -> Qwen2.5-3B, Supervised-token budget=10%2026.053.34
TIPSetting=DeepSeek-R1-Distill-Qwen-14B -> Qwen2.5-3B, Supervised-token budget=10%2026.053.33
EntropySetting=DeepSeek-R1-Distill-Qwen-14B -> Qwen2.5-3B, Supervised-token budget=10%2026.051.67