Mathematical Reasoning on AIME (Pass@1, Pass@2)
71.42Pass@1Phi-4-mini + Mistral3-3B
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Phi-4-mini + Mistral3-3BTraining Framework=CORE, Model Configuration=Team2026.01 | 71.42 | 79.65 | |
| Phi-4-mini-reasoningTraining Framework=CORE, Model Configuration=Single2026.01 | 68.16 | 76.64 | |
| Mistral3-3B-ReasoningTraining Framework=CORE, Model Configuration=Single2026.01 | 67.15 | 75.9 | |
| Phi-4-mini + Mistral3-3B + OracleTraining Framework=SD-E2, Model Configuration=Oracle2026.01 | 54.42 | 60.19 | |
| Phi-4-mini-reasoningTraining Framework=SD-E2, Model Configuration=Single2026.01 | 53.92 | 57.58 | |
| Mistral3-3B-ReasoningTraining Framework=SD-E2, Model Configuration=Single2026.01 | 49 | 55.42 | |
| Qwen2.5-3B + Qwen3-4B (Team)Paradigm=CORE2026.01 | 19.5 | 20.45 | |
| Qwen2.5-3B + Qwen3-4B + OracleParadigm=SD-E22026.01 | 17.14 | 17.9 | |
| Qwen3-4B-InstructParadigm=CORE2026.01 | 16.1 | 17.04 | |
| Qwen2.5-3B-InstructParadigm=CORE2026.01 | 15.7 | 16.82 | |
| Qwen3-4B-InstructParadigm=SD-E22026.01 | 13.3 | 14.9 | |
| Qwen2.5-3B-InstructParadigm=SD-E22026.01 | 13.28 | 14.85 | |
| Phi-4-mini + Mistral3-3B + OracleTraining Framework=Base, Model Configuration=Oracle2026.01 | 13.02 | 17.92 | |
| Phi-4-mini-reasoningTraining Framework=Base, Model Configuration=Single2026.01 | 11.91 | 16.58 | |
| Mistral3-3B-ReasoningTraining Framework=Base, Model Configuration=Single2026.01 | 10.18 | 14.86 | |
| Qwen2.5-3B + Qwen3-4B + OracleParadigm=Base2026.01 | 7.9 | 9.2 | |
| Qwen3-4B-InstructParadigm=Base2026.01 | 7.04 | 7.8 | |
| Qwen2.5-3B-InstructParadigm=Base2026.01 | 6.74 | 7.5 |