Multimodal Mathematical Reasoning on MathVision (Mean@5)
79.5Mean@5Naive-KD
Evaluation Results
| Method | Links | |
|---|---|---|
| Naive-KDTrain Set=MathVerse, Student=Qwen3-VL-235B2026.03 | 79.5 | |
| TEDTrain Set=MathVerse, Student=Qwen3-VL-235B2026.03 | 76.2 | |
| MementoTrain Set=MathVerse, Student=Qwen3-VL-235B2026.03 | 75.8 | |
| RAGTrain Set=–, Student=Qwen3-VL-235B2026.03 | 75.1 | |
| ReflexionTrain Set=MathVerse, Student=Qwen3-VL-235B2026.03 | 75.1 | |
| DirectTrain Set=–, Student=Qwen3-VL-235B2026.03 | 74.6 | |
| Few-shotTrain Set=–, Student=Qwen3-VL-235B2026.03 | 74.4 | |
| MemComTrain Set=MathVerse, Student=Qwen3-VL-235B2026.03 | 74.1 | |
| Naive-KDTrain Set=MathVerse, Student=Qwen3-VL-8B2026.03 | 72.9 | |
| TEDTrain Set=MathVerse, Student=Qwen3-VL-8B2026.03 | 70.2 | |
| MementoTrain Set=MathVerse, Student=Qwen3-VL-8B2026.03 | 67.4 | |
| ReflexionTrain Set=MathVerse, Student=Qwen3-VL-8B2026.03 | 66.2 | |
| MemComTrain Set=MathVerse, Student=Qwen3-VL-8B2026.03 | 64.6 | |
| RAGTrain Set=–, Student=Qwen3-VL-8B2026.03 | 63.9 | |
| Few-shotTrain Set=–, Student=Qwen3-VL-8B2026.03 | 63.1 | |
| DirectTrain Set=–, Student=Qwen3-VL-8B2026.03 | 62.7 |