Visual Question Answering on INFOSEEK Unseen-E
42.1AccuracyVLM-PRF w/ RL
Evaluation Results
| Method | Links | |
|---|---|---|
| VLM-PRF w/ RLLLM/MLLM=InternVL3-8B, Retriever=EVA-CLIP-8B, Feature=V+T2026.05 | 42.1 | |
| CgRAGLLM/MLLM=InternVL3-8B, Retriever=EVA-CLIP-8B, Feature=V+T2026.05 | 42 | |
| CgRAGLLM/MLLM=LLaVA-NeXT-8B, Retriever=EVA-CLIP-8B, Feature=V+T2026.05 | 40.9 | |
| CgRAGLLM/MLLM=LLaVA-NeXT-7B, Retriever=EVA-CLIP-8B, Feature=V+T2026.05 | 40.5 | |
| CgRAGLLM/MLLM=Qwen2.5-VL-7B, Retriever=EVA-CLIP-8B, Feature=V+T2026.05 | 39.5 | |
| VLM-PRFLLM/MLLM=Qwen2.5-VL-7B, Retriever=EVA-CLIP-8B, Feature=V+T2026.05 | 39.4 | |
| CgRAGLLM/MLLM=Qwen2-VL-7B, Retriever=EVA-CLIP-8B, Feature=V+T2026.05 | 39.2 | |
| MMKB-RAGLLM/MLLM=Qwen2-VL-7B, Retriever=EVA-CLIP-8B, Feature=V2026.05 | 36.3 | |
| ReflectiVALLM/MLLM=LLaMA-3.1-8B, Retriever=EVA-CLIP-8B, Feature=V2026.05 | 28.1 | |
| Wiki-LLaVALLM/MLLM=Vicuna-7B, Retriever=CLIP ViT-L/14 + Contriever, Feature=T2026.05 | 27.8 | |
| RORA-VLMLLM/MLLM=Vicuna-7B, Retriever=CLIP + Google Search, Feature=V2026.05 | 27.3 | |
| Wiki-LLaVALLM/MLLM=LLaMA-3.1-8B, Retriever=CLIP ViT-L/14 + Contriever, Feature=T2026.05 | 25.7 | |
| LLaVA-NeXT-8BMode=Zero-shot MLLMs2026.05 | 24 | |
| Qwen2.5-VL-7BMode=Zero-shot MLLMs2026.05 | 24 | |
| LLaVA-NeXT-7BMode=Zero-shot MLLMs2026.05 | 23.4 | |
| Qwen2.5-VL-3BMode=Zero-shot MLLMs2026.05 | 21.7 | |
| EchoSightLLM/MLLM=LLaMA-3.1-8B, Retriever=EVA-CLIP-8B, Feature=V2026.05 | 19.8 | |
| Qwen2-VL-7BMode=Zero-shot MLLMs2026.05 | 16.8 | |
| GPT-4VMode=Zero-shot MLLMs2026.05 | 14.3 | |
| BLIP-2LLM/MLLM=Flan-T5XL, Mode=Zero-shot MLLMs2026.05 | 12.3 | |
| LLaVA-1.5LLM/MLLM=Vicuna-7B, Mode=Zero-shot MLLMs2026.05 | 9.4 | |
| LLaVA-1.5LLM/MLLM=LLaMA-3.1-8B, Mode=Zero-shot MLLMs2026.05 | 8.9 | |
| InstructBLIPLLM/MLLM=Flan-T5XL, Mode=Zero-shot MLLMs2026.05 | 7.4 |