List your company
ResearchFormerly WizWand

Table Question Answering on WikiTableQuestions (test)

76.6AccuracyTabLaP
-0.04819.85139.7559.649Sep 9, 2019Aug 9, 2020Jul 11, 2021Jun 11, 2022May 13, 2023Apr 12, 2024Mar 14, 2025
Updated 8mo ago

Evaluation results

MethodLinks
TabLaP2025.0376.6——
RePandamode=Finetuned-Pandas for QA, training examples=1,2002025.0375.1——
SynTQAengine=GPT2025.0374.4——
Mix SC2025.0373.6——
SynTQAengine=RF2025.0371.6——
CABINET2025.0369.1——
GPT-42025.0168.4——
Chain-of-Table2025.0367.31——
Tab-PoT2025.0366.78——
Codex w/ DATEREvaluation Protocol=LLM based2023.0165.9——
OmniTabf-shot=full, pretraining_data=all2022.0762.8——
OmniTab w/ DATEREvaluation Protocol=Fine-tuning2023.0162.5——
BinderEvaluation Protocol=LLM based2023.0161.9——
OmniTabf-shot=full, pretraining_data=natural2022.0761.3——
OmniTabf-shot=full, pretraining_data=synthetic2022.0761.3——
OmniTabEvaluation Protocol=Fine-tuning2023.0161.2——
TaCubeEvaluation Protocol=Fine-tuning2023.0160.8——
TAPEX*f-shot=full2022.0760.1——
TAPEXf-shot=full2022.0759.5——
ReasTAPEvaluation Protocol=Fine-tuning2023.0158.6——
TAPEXEvaluation Protocol=Fine-tuning2023.0157.2——
TableRAG2024.1057.03——
Binder2024.1056.74——
GPT-3.52025.0153.13——
Text-to-SQL2024.1052.9——
TAMAbackbone=LLaMA 3.1 8B Instruct2025.0152.88——
Dater2024.1052.81——
Wang et al. (2019) + GRAPPA (MLM+SSP)Encoder=GRAPPA, Pre-training objective=MLM+SSP2020.0952.7——
TableFormerEvaluation Protocol=Fine-tuning2023.0152.6——
TaBERT+MAPOf-shot=full2022.0752.3——
TaBERT2024.1052.3——
OmniTabf-shot=1024, pretraining_data=all2022.0751.9——
Yin et al. (2020b)2020.0951.8——
Wang et al. (2019) + GRAPPA (MLM)Encoder=GRAPPA, Pre-training objective=MLM2020.0951.7——
Wang et al. (2019) + GRAPPA (SSP)Encoder=GRAPPA, Pre-training objective=SSP2020.0951.1——
Wang et al. (2019) + RoBERTa-largeEncoder=RoBERTa-large2020.0950.9——
TaPasEvaluation Protocol=Fine-tuning2023.0150.4——
OmniTabf-shot=1024, pretraining_data=natural2022.0749.8——
T5-3BEvaluation Protocol=Fine-tuning2023.0149.3——
Herzig et al. (2020b)2020.0948.8——
TAPASf-shot=full2022.0748.8——
OmniTabf-shot=1024, pretraining_data=synthetic2022.0748.8——
CodexEvaluation Protocol=LLM based2023.0147.6——
Structured-Attention ParserEnsemble size=102019.0947.3——
Agarwal et al. (2019)Ensemble size=102019.0946.9——
Liang et al. (2018)Ensemble size=102019.0946.3——
TAPEXf-shot=10242022.0745.5——
IterativeSearchEvaluation Protocol=Fine-tuning2023.0144.7——
TAPEX*f-shot=10242022.0744.6——
Wang et al. (2019)2020.0944.5——
LatentAlignmentEvaluation Protocol=Fine-tuning2023.0144.5——
Agarwal et al. (2019)2020.0944.1——
MeRLEvaluation Protocol=Fine-tuning2023.0144.1——
Dasigi et al. (2019)2020.0943.9——
MAPOEvaluation Protocol=Fine-tuning2023.0143.8——
LLaMA 3.1 8B Instructstatus=base model2025.0143.46——
Liang et al. (2018)2020.0943.1——
Wang et al. (2019) + GRAPPA (MLM+SSP) (10% data)Encoder=GRAPPA, Pre-training objective=MLM+SSP, Training data percentage=10%2020.0942——
OmniTabf-shot=128, pretraining_data=all2022.0741.4——
OmniTabf-shot=128, pretraining_data=natural2022.0738.4——
Wang et al. (2019) + RoBERTa-large (10% data)Encoder=RoBERTa-large, Training data percentage=10%2020.0938.1——
BARTf-shot=full2022.0738——
OmniTabf-shot=128, pretraining_data=synthetic2022.0737.5——
mPLUG-Owl-7B + OursModel Params=7.2B, Trainable Params=96M, Pre-training Data=1.1B, Fine-tuning Data=650K2024.1134.5——
TAPASf-shot=10242022.0733.6——
TaBERT+MAPOf-shot=10242022.0733.3——
MonkeyModel Params=9.8B, Trainable Params=207M, Pre-training Data=1.4B+76.8M, Fine-tuning Data=1.44M2024.1132.8——
mPLUG-Owl-7B + UReaderModel Params=7.2B, Trainable Params=86M, Pre-training Data=1.1B, Fine-tuning Data=650K2024.1129.4——
OmniTabf-shot=16, pretraining_data=all2022.0726.8——
TAPEX*f-shot=1282022.0725.2——
TAPEXf-shot=1282022.0723.1——
OmniTabf-shot=16, pretraining_data=natural2022.0722.8——
OmniTabf-shot=16, pretraining_data=synthetic2022.0721.5——
BLIP-2-OPT-2.7B + OursModel Params=3.8B, Trainable Params=14M, Pre-training Data=129M, Fine-tuning Data=650K2024.1121.2——
TAPASf-shot=1282022.0718.9——
DonutModel Params=143M, Trainable Params=143M, Pre-training Data=13M, Fine-tuning Data=Task-specific2024.1118.8——
BLIP-2-OPT-2.7B + UReaderModel Params=3.8B, Trainable Params=8M, Pre-training Data=129M, Fine-tuning Data=650K2024.1117.4——
BARTf-shot=10242022.0717.3——
Qwen-VLModel Params=9.6B, Trainable Params=207M, Pre-training Data=1.4B+76.8M, Fine-tuning Data=Task-specific2024.1115.9——
TAPEX*f-shot=162022.0715.7——
TaBERT+MAPOf-shot=1282022.0715.1——
TAPEXf-shot=162022.0710.4——
TAPASf-shot=162022.079.8——
BARTf-shot=1282022.078.4——
TaBERT+MAPOf-shot=162022.077.7——
BARTf-shot=162022.072.9——
Agarwal et al. (2019)Category=Previous Systems2021.07—44.1—
Agarwal et al. (2019)2020.04—44.1—
BARTCategory=Pre-trained Language Models2021.07—38—
BinderLLM=Codex2024.06——61.9
CoTLLM=Mixtral-8x7B2024.06——53.48
CoTLLM=Mistral-7B2024.06——30.46
CoTLLM=DeepSeek-67B2024.06——55.57
CoTLLM=DeepSeek-7B2024.06——34.65
Dasigi et al. (2019)Category=Previous Systems2021.07—44.3—
Dasigi et al. (2019)2020.04—43.9—
DaterLLM=Codex2024.06——65.9
DirectLLM=Mixtral-8x7B2024.06——53.08
DirectLLM=Mistral-7B2024.06——27.19
DirectLLM=DeepSeek-67B2024.06——54.72