List your company

Multi-page Document Question Answering on MP-DocVQA

88.2ANLSURaG
43.136854.835966.53578.2341Sep 5, 2024Dec 11, 2024Mar 19, 2025Jun 25, 2025Oct 1, 2025Jan 7, 2026Apr 15, 2026

Evaluation results

MethodLinks
URaGBackbone=Qwen2.5-VL, Param=7B, Paradigm=RAG2026.0488.2—
Doc-V* (GRPO)Backbone=Qwen2.5-VL, Param=7B, Paradigm=Agent2026.0486.2—
DocVLMBackbone=Qwen2-VL, Param=7B, Paradigm=E2E2026.0484.5—
M3DocRAGBackbone=Qwen2-VL, Param=7B, Paradigm=RAG2026.0484.4—
GRAM2024.1083—
DocopilotBackbone=InternVL2, Param=8B, Paradigm=E2E2026.0481.3—
Doc-V* (SFT)Backbone=Qwen2.5-VL, Param=7B, Paradigm=Agent2026.0481.3—
InternVL3Backbone=InternViT / Qwen2.5, Param=8B, Paradigm=E2E2026.0480.8—
Qwen2.5-VL (RAG Top-5)Backbone=Qwen2.5-VL, Param=7B, Paradigm=RAG2026.0477.4—
PDF-WuKong2024.1076.9—
DocFormerv22024.1076.4—
Qwen2.5-VL (Baseline)Backbone=Qwen2.5-VL, Param=7B, Paradigm=E2E2026.0475.2—
CogDocBackbone=Qwen2.5-VL, Param=7B, Paradigm=Agent2026.0475—
LayTokenLLM-8BBackbone=8B2025.0374.31—
CREAMBackbone=Pix2Struct / LLaMa2, Param=7B, Paradigm=RAG2026.0474.3—
LayTokenLLM-7BBackbone=7B2025.0372.81—
SV-RAGBackbone=InternVL2, Param=4B, Paradigm=RAG2026.0471—
SV-RAG2024.1071—
DocOwl2-8BToken^V=3242024.0969.420.95
mPLUG-DocOwl2-8BBackbone=8B2025.0369.42—
mPLUG-DocOwl2Backbone=ViT / LLaMa, Param=8B, Paradigm=E2E2026.0469.4—
DocOwl22024.1069.4—
InternVL2-8BBackbone=8B2025.0368—
GPT-4oBackbone=-, Param=-, Paradigm=E2E2026.0467.4—
Idefics3-8B2024.1067.2—
Idefics3-8BToken^V=~8382024.0967.152.26
Idefics3-8BBackbone=8B2025.0367.15—
CREAM2024.1065.3—
VDocRAGBackbone=Phi3-Vision, Param=4B, Paradigm=RAG2026.0462.6—
HiVT5 (PR)Backbone=DiT / T5, Param=0.3B, Paradigm=E2E2026.0462—
Hi-VT52024.1061.8—
LongVA-7BToken^V=~2,0292024.0960.82.13
LongVA-7BBackbone=7B2025.0360.8—
BigBird2024.1058.5—
LayoutLMv32024.1055.1—
Longformer2024.1055.1—
LLaVA-next-interleave-7BToken^V=7292024.0944.871.56
LLaVA-next-interleave-7BBackbone=7B2025.0344.87—