Question Answering on M3SciQA
93.1AccuracyInternVL-2.5
Evaluation results
| Method | Links | ||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| InternVL-2.5Model Size=8B2026.03 | 93.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Percep. LMModel Size=3B2026.03 | 89.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| InterVL2Model Size=4B2026.03 | 89.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Percep. LMModel Size=8B2026.03 | 87.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| InterVL2Model Size=8B2026.03 | 87.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| InternVL-2.5Model Size=4B2026.03 | 86.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| v-LCMvariant=v-LCM2026.03 | 76.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Percep. LMModel Size=1B2026.03 | 73.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen2-VLModel Size=7B2026.03 | 70.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| InternVL-2.5Model Size=1B2026.03 | 69 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen2.5-VLModel Size=7B2026.03 | 61.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen2.5-VLModel Size=3B2026.03 | 55 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen2-VLModel Size=2B2026.03 | 54.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| InterVL2Model Size=1B2026.03 | 53.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| LCMvariant=LCM2026.03 | 44.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DeepSeek-R1RAG Framework=Classic-RAG2025.05 | — | 11.9 | 9.5 | 11.2 | — | — | — | — | — | — | — | — | — | — | |
| DeepSeek-R1RAG Framework=NeuSym-RAG2025.05 | — | 19 | 13.7 | 17.4 | — | — | — | — | — | — | — | — | — | — | |
| GPT-4o-miniRAG Framework=Classic-RAG2025.05 | — | 17.9 | 10.6 | 15.6 | — | — | — | — | — | — | — | — | — | — | |
| GPT-4o-miniRAG Framework=NeuSym-RAG2025.05 | — | 18.7 | 16.6 | 18 | — | — | — | — | — | — | — | — | — | — | |
| GPT-4VRAG Framework=Classic-RAG2025.05 | — | 12.1 | 8.8 | 11.1 | — | — | — | — | — | — | — | — | — | — | |
| GPT-4VRAG Framework=NeuSym-RAG2025.05 | — | 13.7 | 13.4 | 13.6 | — | — | — | — | — | — | — | — | — | — | |
| GraphRAG2026.02 | — | — | — | — | 16.27 | 10.31 | 6.92 | 19.23 | 10.11 | 15.22 | 5.13 | 8.11 | 6.46 | 6.65 | |
| LightRAG2026.02 | — | — | — | — | 27.98 | 21.04 | 14.47 | 31.97 | 17.02 | 26.02 | 30.26 | 4.53 | 11.57 | 11.91 | |
| Llama-3.3-70B-InstructRAG Framework=Classic-RAG2025.05 | — | 12.7 | 8.1 | 11.3 | — | — | — | — | — | — | — | — | — | — | |
| Llama-3.3-70B-InstructRAG Framework=NeuSym-RAG2025.05 | — | 26.3 | 17.6 | 23.6 | — | — | — | — | — | — | — | — | — | — | |
| Naive RAG2026.02 | — | — | — | — | 29.03 | 21.82 | 14.38 | 33.23 | 18.6 | 27.35 | 26.94 | 28.8 | 4.28 | 4.29 | |
| PT-RAG2026.02 | — | — | — | — | 37.68 | 30.04 | 18.71 | 41.29 | 20.12 | 36.26 | 42.57 | 35.81 | 0.66 | 0.72 | |
| Qwen2.5-VL-72B-InstructRAG Framework=Classic-RAG2025.05 | — | 11.6 | 11.6 | 11.6 | — | — | — | — | — | — | — | — | — | — | |
| Qwen2.5-VL-72B-InstructRAG Framework=NeuSym-RAG2025.05 | — | 20.2 | 22.7 | 21.1 | — | — | — | — | — | — | — | — | — | — | |
| RAPTOR2026.02 | — | — | — | — | 32.12 | 24.95 | 17.99 | 33.34 | 19.5 | 28.64 | 35.37 | 28.33 | 1.85 | 1.9 |