Knowledge-based VQA on E-VQA
70.27Single-Hop AccuracyMG2-RAG
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| MG2-RAGModel Category=Graph Retrieval-Augmented Models, LLM/MLLM=Qwen3.5-27B, Retrieval Pool Size=5k2026.04 | 70.27 | 60.24 | |
| MG2-RAGModel Category=Graph Retrieval-Augmented Models, LLM/MLLM=GPT-5.22026.04 | 68.96 | 60.3 | |
| MG2-RAGModel Category=Graph Retrieval-Augmented Models, LLM/MLLM=Qwen2.5-VL-7B, Retrieval Pool Size=5k2026.04 | 62.88 | 53.36 | |
| MG2-RAGModel Category=Graph Retrieval-Augmented Models, LLM/MLLM=Qwen3.5-27B2026.04 | 61.65 | 52.19 | |
| MG2-RAGModel Category=Graph Retrieval-Augmented Models, LLM/MLLM=Qwen2.5-VL-7B2026.04 | 57.33 | 48.59 | |
| MG2-RAGModel Category=Graph Retrieval-Augmented Models, LLM/MLLM=LLaMA-3.1-8B2026.04 | 55.57 | 47.77 | |
| EchoSightModel Category=Retrieval-Augmented Models, LLM/MLLM=LLaMA-3.1-8B, Modality=V→T2026.04 | 52.96 | 47.23 | |
| EchoSightModel Category=Retrieval-Augmented Models, LLM/MLLM=LLaMA-3.1-8B, Modality=V→V2026.04 | 46.32 | 41.29 | |
| GPT-5.2Model Category=Zero-Shot MLLMs, LLM/MLLM=GPT-5.22026.04 | 44.19 | 40.3 | |
| Qwen3.5-27BModel Category=Zero-Shot MLLMs, LLM/MLLM=Qwen3.5-27B2026.04 | 29.79 | 25.72 | |
| Gemini-3.1-ProModel Category=Zero-Shot MLLMs, LLM/MLLM=Gemini-3.1-Pro2026.04 | 29.36 | 27.9 | |
| Qwen2.5-VL-7BModel Category=Zero-Shot MLLMs, LLM/MLLM=Qwen2.5-VL-7B2026.04 | 23.6 | 23.2 | |
| MMGraphRAG (5k)Model Category=Graph Retrieval-Augmented Models, LLM/MLLM=Qwen2.5-VL-7B, Retrieval Pool Size=5k2026.04 | 19.12 | 16.52 | |
| VaLiK (5k)Model Category=Graph Retrieval-Augmented Models, LLM/MLLM=Qwen2.5-VL-7B, Retrieval Pool Size=5k2026.04 | 16.16 | 15.22 | |
| LLaVA-v1.5Model Category=Zero-Shot MLLMs, LLM/MLLM=LLaMA-3.1-8B2026.04 | 16 | 16.9 | |
| InstructBLIPModel Category=Zero-Shot MLLMs, LLM/MLLM=Flan-T5XL2026.04 | 11.9 | 12 |