Knowledge-Based Visual Question Answering on OVEN (val)
25.1Accuracy (All)Qwen3-VL-4B
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen3-VL-4BBackbone=Qwen3-VL-4B, Model Scale=Small, Method=Zero-shot retrieval-augmented2026.04 | 25.1 | |
| Qwen3-VL-4B + LoTBackbone=Qwen3-VL-4B, Model Scale=Small, Method=LoT2026.04 | 24.5 | |
| Qwen2.5-VL-7B + LoTBackbone=Qwen2.5-VL-7B, Model Scale=Medium, Method=LoT2026.04 | 21.2 | |
| Qwen2.5-VL-7BBackbone=Qwen2.5-VL-7B, Model Scale=Medium, Method=Zero-shot retrieval-augmented2026.04 | 21.1 | |
| InternVL3.5-38BBackbone=InternVL3.5-38B, Model Scale=Large, Method=Zero-shot retrieval-augmented2026.04 | 20.2 | |
| Qwen3-VL-8B + LoTBackbone=Qwen3-VL-8B, Model Scale=Medium, Method=LoT2026.04 | 19.6 | |
| InternVL3.5-38B + LoTBackbone=InternVL3.5-38B, Model Scale=Large, Method=LoT2026.04 | 18.5 | |
| Qwen2.5-VL-3B + LoTBackbone=Qwen2.5-VL-3B, Model Scale=Small, Method=LoT2026.04 | 18.3 | |
| Qwen3-VL-8BBackbone=Qwen3-VL-8B, Model Scale=Medium, Method=Zero-shot retrieval-augmented2026.04 | 17.7 | |
| InternVL3.5-8BBackbone=InternVL3.5-8B, Model Scale=Medium, Method=Zero-shot retrieval-augmented2026.04 | 17.7 | |
| InternVL3.5-8B + LoTBackbone=InternVL3.5-8B, Model Scale=Medium, Method=LoT2026.04 | 17.6 | |
| Qwen2-VL-7B + LoTBackbone=Qwen2-VL-7B, Model Scale=Medium, Method=LoT2026.04 | 16.6 | |
| Qwen2.5-VL-32B + LoTBackbone=Qwen2.5-VL-32B, Model Scale=Large, Method=LoT2026.04 | 14.2 | |
| Qwen2.5-VL-32BBackbone=Qwen2.5-VL-32B, Model Scale=Large, Method=Zero-shot retrieval-augmented2026.04 | 13.3 | |
| Qwen2.5-VL-3BBackbone=Qwen2.5-VL-3B, Model Scale=Small, Method=Zero-shot retrieval-augmented2026.04 | 11.5 | |
| InternVL3.5-4B + LoTBackbone=InternVL3.5-4B, Model Scale=Small, Method=LoT2026.04 | 11.5 | |
| Qwen2-VL-7BBackbone=Qwen2-VL-7B, Model Scale=Medium, Method=Zero-shot retrieval-augmented2026.04 | 11.1 | |
| InternVL3.5-4BBackbone=InternVL3.5-4B, Model Scale=Small, Method=Zero-shot retrieval-augmented2026.04 | 10.8 | |
| Qwen2-VL-2B + LoTBackbone=Qwen2-VL-2B, Model Scale=Small, Method=LoT2026.04 | 2.8 | |
| Qwen2-VL-2BBackbone=Qwen2-VL-2B, Model Scale=Small, Method=Zero-shot retrieval-augmented2026.04 | 1.2 |