List your company
ResearchFormerly WizWand

Visual Question Answering on DocVQA (val)

89.2ANLSQwen2-VL-2B
21.70439.22756.7574.273May 29, 2023Nov 17, 2023May 7, 2024Oct 26, 2024Apr 16, 2025Oct 5, 2025Mar 27, 2026
Updated 7mo ago

Evaluation results

MethodLinks
Qwen2-VL-2BParams (B)=2.12024.1089.2
NSCEvaluation Protocol=Merged, Backbone=Vicuna 7B, LoRA rank=162026.0387.1
PaLI-XOCR pipeline input=true2023.0586.8
InternVL2-2BParams (B)=2.12024.1086.2
GPT-4o-202408062024.1086.1
AdaMergingEvaluation Protocol=Merged, Token Strategy=Full Token, Backbone=Vicuna 7B, LoRA rank=162026.0385.8
SVDEvaluation Protocol=Merged, Backbone=Vicuna 7B, LoRA rank=162026.0385.5
KnOTS-TIESEvaluation Protocol=Merged, Backbone=Vicuna 7B, LoRA rank=162026.0385.5
KnOTS-DARE-TIESEvaluation Protocol=Merged, Backbone=Vicuna 7B, LoRA rank=162026.0385.3
TAEvaluation Protocol=Merged, Backbone=Vicuna 7B, LoRA rank=162026.0385.2
Method [54]OCR pipeline input=true2023.0584.7
DARE-TIESEvaluation Protocol=Merged, Backbone=Vicuna 7B, LoRA rank=162026.0384.4
Phi-3-Vision - MicrosoftParams (B)=4.22024.1084.3
EMR-MergingEvaluation Protocol=Merged, Backbone=Vicuna 7B, LoRA rank=162026.0384.3
H2OVL-Mississippi-2BParams (B)=2.12024.1083.8
LoRA-LEGOEvaluation Protocol=Merged, Backbone=Vicuna 7B, LoRA rank=162026.0382.5
TIESEvaluation Protocol=Merged, Backbone=Vicuna 7B, LoRA rank=162026.0382.3
RobustMergeEvaluation Protocol=Merged, Backbone=Vicuna 7B, LoRA rank=162026.0382.1
AdaMergingEvaluation Protocol=Merged, Token Strategy=Single Token, Backbone=Vicuna 7B, LoRA rank=162026.0382.1
DeepStack-HD+LLM=Vicuna-13B, Visual Tokens=14400, Context Length=2880, Pre-training=748K, Instruction-tuning=748K2024.0681
PaLI-XOCR pipeline input=false2023.0580
LinearEvaluation Protocol=Merged, Backbone=Vicuna 7B, LoRA rank=162026.0379.2
DeepStack-HD+LLM=Vicuna-7B, Visual Tokens=14400, Context Length=2880, Pre-training=558K, Instruction-tuning=748K2024.0678.8
Vary-baseSFT data=LLaVA-80k2023.1278.2
Vary-baseSFT data=LLaVA-CC665k2023.1278.1
LLaVA-NextLLM=Vicuna-13B, Visual Tokens=2880, Context Length=2880, Pre-training=558K, Instruction-tuning=765K2024.0677.5
Method [27]OCR pipeline input=false2023.0576.6
FR-MergingEvaluation Protocol=Merged, Backbone=Vicuna 7B, LoRA rank=162026.0376.2
LLaVA-NextLLM=Vicuna-7B, Visual Tokens=2880, Context Length=2880, Pre-training=558K, Instruction-tuning=765K2024.0674.4
PaliGemma-3B-mix-448Params (B)=2.92024.1073.9
Cambrian-13BParams (B)=132024.1073.7
MiniCPM-V-2Params (B)=2.82024.1069.6
Vary-toySize=1.8B2024.0165.6
Qwen-VL-chatSize=7B2024.0165.1
mPLUG-DocOwlSize=7B2024.0162.2
VisInContextText Source=Rendered Image2024.0648.5
Dessurt2023.1246.5
Dessurt2024.0146.5
Open-Flamingo-9B BaselineText Source=Raw Text2024.0645.3
DeepStack-LLLM=Vicuna-13B, Visual Tokens=2880, Context Length=576, Pre-training=558K, Instruction-tuning=665K2024.0643.1
DeepStack-VLLM=Vicuna-13B, Visual Tokens=2880, Context Length=576, Pre-training=558K, Instruction-tuning=665K2024.0641.7
DeepStack-VLLM=Vicuna-7B, Visual Tokens=2880, Context Length=576, Pre-training=558K, Instruction-tuning=665K2024.0641
FinetunedEvaluation Protocol=Finetuned, Backbone=Vicuna 7B, LoRA rank=162026.0340.7
DeepStack-LLLM=Vicuna-7B, Visual Tokens=2880, Context Length=576, Pre-training=558K, Instruction-tuning=665K2024.0639.3
LLaVA-1.5LLM=Vicuna-13B, Visual Tokens=576, Context Length=576, Pre-training=558K, Instruction-tuning=665K2024.0630.3
LLaVA-1.5LLM=Vicuna-7B, Visual Tokens=576, Context Length=576, Pre-training=558K, Instruction-tuning=665K2024.0628.1
Zero-ShotEvaluation Protocol=Zero-Shot2026.0324.3