List your company

Visual Retrieval-Augmented Generation on Visual-RAG

66.04ScoreCLIP-B
33.966442.293250.6258.9468Jan 31, 2026Feb 17, 2026Mar 6, 2026Mar 23, 2026Apr 9, 2026Apr 26, 2026May 13, 2026

Evaluation results

MethodLinks
CLIP-BMain Model=Ovis2.5-9B, Top K=1, Retriever Params=151M2026.0566.04
Ours (Ovis2.5-2B Surrogate)Main Model=MiniCPM-V4.5, Top K=3, Retriever Params=2.6B2026.0565.11
VLM2Vec-V2.0Main Model=Qwen3-VL-8B, Top K=4, Retriever Params=2.2B2026.0561.63
OpenCLIPMain Model=InternVL3.5-8B, Top K=1, Retriever Params=151M2026.0560.29
GT (Human-Annotated Image as Input)Main Model=InternVL3.5-8B, Top K=52026.0558.82
BGE-MLLM-S2Main Model=Gemma3-12B, Top K=2, Retriever Params=7.6B2026.0555.75
ViLoMemBackbone=QWEN2.5-VL-32B, Augmentation=ViLoMem2026.0154.6
PolarMemBackbone=QWEN2.5-VL-32B, Augmentation=PolarMem2026.0153.4
Zero-ShotMain Model=Ovis2.5-9B2026.0552.67
ViLoMemBackbone=QWEN2.5-VL-7B, Augmentation=ViLoMem2026.0150.8
Vanilla RAGBackbone=QWEN2.5-VL-32B, Augmentation=Vanilla RAG2026.0150
PolarMemBackbone=QWEN2.5-VL-7B, Augmentation=PolarMem2026.0149.9
PolarMemBackbone=LLaVA-Next-Llama-3-8B, Augmentation=PolarMem2026.0148.6
QWEN2.5-VL-32BBackbone=QWEN2.5-VL-32B, Augmentation=None2026.0147.9
ViLoMemBackbone=LLaVA-Next-Llama-3-8B, Augmentation=ViLoMem2026.0147.9
Vanilla RAGBackbone=QWEN2.5-VL-7B, Augmentation=Vanilla RAG2026.0144.1
PolarMemBackbone=DeepSeek-VL2, Augmentation=PolarMem2026.0144.1
PolarMemBackbone=InternVL-3.5-4B, Augmentation=PolarMem2026.0143.6
ViLoMemBackbone=DeepSeek-VL2, Augmentation=ViLoMem2026.0142.7
PolarMemBackbone=LLaVA-Next-Mistral-7B, Augmentation=PolarMem2026.0142.6
ViLoMemBackbone=InternVL-3.5-8B, Augmentation=ViLoMem2026.0142
ViLoMemBackbone=LLaVA-Next-Mistral-7B, Augmentation=ViLoMem2026.0141.8
PolarMemBackbone=InternVL-3.5-8B, Augmentation=PolarMem2026.0141.3
QWEN2.5-VL-7BBackbone=QWEN2.5-VL-7B, Augmentation=None2026.0140.2
ViLoMemBackbone=DeepSeek-VL2-Small, Augmentation=ViLoMem2026.0140.2
InternVL-3.5-8BBackbone=InternVL-3.5-8B, Augmentation=None2026.0140.1
ViLoMemBackbone=InternVL-3.5-4B, Augmentation=ViLoMem2026.0140
Vanilla RAGBackbone=InternVL-3.5-8B, Augmentation=Vanilla RAG2026.0139.6
PolarMemBackbone=DeepSeek-VL2-Small, Augmentation=PolarMem2026.0139.5
Vanilla RAGBackbone=DeepSeek-VL2, Augmentation=Vanilla RAG2026.0139.4
DeepSeek-VL2Backbone=DeepSeek-VL2, Augmentation=None2026.0138.2
Vanilla RAGBackbone=InternVL-3.5-4B, Augmentation=Vanilla RAG2026.0138.1
Vanilla RAGBackbone=LLaVA-Next-Llama-3-8B, Augmentation=Vanilla RAG2026.0137.9
InternVL-3.5-4BBackbone=InternVL-3.5-4B, Augmentation=None2026.0137.8
LLaVA-Next-Llama-3-8BBackbone=LLaVA-Next-Llama-3-8B, Augmentation=None2026.0136.8
Vanilla RAGBackbone=LLaVA-Next-Mistral-7B, Augmentation=Vanilla RAG2026.0135.9
Vanilla RAGBackbone=DeepSeek-VL2-Small, Augmentation=Vanilla RAG2026.0135.6
DeepSeek-VL2-SmallBackbone=DeepSeek-VL2-Small, Augmentation=None2026.0135.2
LLaVA-Next-Mistral-7BBackbone=LLaVA-Next-Mistral-7B, Augmentation=None2026.0135.2