Multi-page Document Question Answering on MP-DocVQA
88.2ANLSURaG
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| URaGBackbone=Qwen2.5-VL, Param=7B, Paradigm=RAG2026.04 | 88.2 | — | |
| Doc-V* (GRPO)Backbone=Qwen2.5-VL, Param=7B, Paradigm=Agent2026.04 | 86.2 | — | |
| DocVLMBackbone=Qwen2-VL, Param=7B, Paradigm=E2E2026.04 | 84.5 | — | |
| M3DocRAGBackbone=Qwen2-VL, Param=7B, Paradigm=RAG2026.04 | 84.4 | — | |
| GRAM2024.10 | 83 | — | |
| DocopilotBackbone=InternVL2, Param=8B, Paradigm=E2E2026.04 | 81.3 | — | |
| Doc-V* (SFT)Backbone=Qwen2.5-VL, Param=7B, Paradigm=Agent2026.04 | 81.3 | — | |
| InternVL3Backbone=InternViT / Qwen2.5, Param=8B, Paradigm=E2E2026.04 | 80.8 | — | |
| Qwen2.5-VL (RAG Top-5)Backbone=Qwen2.5-VL, Param=7B, Paradigm=RAG2026.04 | 77.4 | — | |
| PDF-WuKong2024.10 | 76.9 | — | |
| DocFormerv22024.10 | 76.4 | — | |
| Qwen2.5-VL (Baseline)Backbone=Qwen2.5-VL, Param=7B, Paradigm=E2E2026.04 | 75.2 | — | |
| CogDocBackbone=Qwen2.5-VL, Param=7B, Paradigm=Agent2026.04 | 75 | — | |
| LayTokenLLM-8BBackbone=8B2025.03 | 74.31 | — | |
| CREAMBackbone=Pix2Struct / LLaMa2, Param=7B, Paradigm=RAG2026.04 | 74.3 | — | |
| LayTokenLLM-7BBackbone=7B2025.03 | 72.81 | — | |
| SV-RAGBackbone=InternVL2, Param=4B, Paradigm=RAG2026.04 | 71 | — | |
| SV-RAG2024.10 | 71 | — | |
| DocOwl2-8BToken^V=3242024.09 | 69.42 | 0.95 | |
| mPLUG-DocOwl2-8BBackbone=8B2025.03 | 69.42 | — | |
| mPLUG-DocOwl2Backbone=ViT / LLaMa, Param=8B, Paradigm=E2E2026.04 | 69.4 | — | |
| DocOwl22024.10 | 69.4 | — | |
| InternVL2-8BBackbone=8B2025.03 | 68 | — | |
| GPT-4oBackbone=-, Param=-, Paradigm=E2E2026.04 | 67.4 | — | |
| Idefics3-8B2024.10 | 67.2 | — | |
| Idefics3-8BToken^V=~8382024.09 | 67.15 | 2.26 | |
| Idefics3-8BBackbone=8B2025.03 | 67.15 | — | |
| CREAM2024.10 | 65.3 | — | |
| VDocRAGBackbone=Phi3-Vision, Param=4B, Paradigm=RAG2026.04 | 62.6 | — | |
| HiVT5 (PR)Backbone=DiT / T5, Param=0.3B, Paradigm=E2E2026.04 | 62 | — | |
| Hi-VT52024.10 | 61.8 | — | |
| LongVA-7BToken^V=~2,0292024.09 | 60.8 | 2.13 | |
| LongVA-7BBackbone=7B2025.03 | 60.8 | — | |
| BigBird2024.10 | 58.5 | — | |
| LayoutLMv32024.10 | 55.1 | — | |
| Longformer2024.10 | 55.1 | — | |
| LLaVA-next-interleave-7BToken^V=7292024.09 | 44.87 | 1.56 | |
| LLaVA-next-interleave-7BBackbone=7B2025.03 | 44.87 | — |