List your company
ResearchFormerly WizWand

Table Question Answering on WTQ (test)

64.7Denotation AccuracyT5 + TC + P
13.32426.6624053.338Mar 13, 2024Jul 11, 2024Nov 9, 2024Mar 10, 2025Jul 9, 2025Nov 7, 2025Mar 8, 2026
Updated 7mo ago

Evaluation results

MethodLinks
T5 + TC + PModel Category=Text-to-SQL, Table Content=true, Post-processing=true2024.0964.7
OmniTabModel Category=E2E TQA2024.0962.6
Readi-GPT3.5category=Inference-based Method, LLM=gpt-3.5-turbo2024.0361.7
Readi-GPT4category=Inference-based Method, LLM=GPT42024.0361.3
DocLayLLMSize(B)=82026.0358.6
DocCogitoSize(B)=82026.0358.3
TAPEXcategory=Training-based Method2024.0357.5
Qwen3-VL-InstructSize(B)=82026.0357.1
GPT4category=Inference-based Method2024.0357
TAPEXModel Category=E2E TQA2024.0957
GPT3.5category=Inference-based Method2024.0355.8
DocCogitoSize(B)=42026.0354.8
FRESBase Model=Pixtral 12B2025.0554.4
Qwen3-VL-InstructSize(B)=42026.0354.4
MM1.5-30BModel Scale=30B2024.0954.1
PixtralModel Size=12B, Representation=text and image2025.0552.5
MartenSize(B)=8.12026.0352.4
StructGPTcategory=Inference-based Method2024.0352.2
PixtralModel Size=12B, Representation=text2025.0551.3
UnifiedSKGcategory=Training-based Method, backbone=T5-3B2024.0349.3
TAPAScategory=Training-based Method2024.0348.8
DocFormerv2Size(B)=0.752026.0348.3
TabPediaInput Size=25602024.0647.8
Phi-3-Vision-4BModel Scale=4B2024.0947.4
TextHawk2Size(B)=7.42026.0346.2
MM1.5-7BModel Scale=7B2024.0946
GPT4VInput Size=6452024.0645.5
PixtralModel Size=12B, Representation=image2025.0542.2
MM1.5-3BModel Scale=3B2024.0941.8
DocOwl-1.5-ChatModel Scale=7B2024.0940.6
mPLUG-DocOwl 1.5backbone=mPLUG-Owl 2, fine-tuning dataset=4M2024.1140.6
DocOwl-1.5-ChatSize(B)=8.12026.0340.6
DocOwl 1.5Input Size=13442024.0639.8
DocOwl-1.5Size(B)=8.12026.0339.8
MM1.5-3B-MOEModel Scale=3B, Architecture=MoE2024.0939.1
MM1.5-1B-MOEModel Scale=1B, Architecture=MoE2024.0938.9
FRESBase Model=TableLlaVA 7B2025.0538.8
Zhang et al.Size(B)=8.12026.0338.6
TextMonkeyInput Size=8962024.0637.9
DocOwl2Size(B)=82026.0336.5
InternVL2-2BModel Scale=1B2024.0935.8
Davinci-003category=Inference-based Method2024.0334.8
HVFAbackbone=mPLUG-Owl, fine-tuning dataset=650K2024.1134.5
Park et al.Size(B)=7.22026.0334.5
TableLlaVAModel Size=7B, Representation=text2025.0534.4
MM1.5-1BModel Scale=1B2024.0934.1
MM1-30BModel Scale=30B2024.0933.3
DocKylinSize(B)=7.12026.0332.4
Gemini ProInput Size=6592024.0632.3
HRVDASize(B)=7.12026.0331.2
CogagentInput Size=11202024.0630.2
MM1-7BModel Scale=7B2024.0928.8
Xcomposer2Input Size=5112024.0628.7
MonkeyInput Size=8962024.0625.3
MonkeySize(B)=9.82026.0325.3
MiniCPM-V 2.0-3BModel Scale=3B2024.0924.2
MM1-3BModel Scale=3B2024.0923.6
MM1-1BModel Scale=1B2024.0919.9
TableLlaVAModel Size=7B, Representation=image2025.0519.2
DonutSize(B)=0.262026.0318.8
TableLlaVAModel Size=7B, Representation=text and image2025.0517.2
SPHINX-TinyModel Scale=1B2024.0915.3