Table Question Answering on WikiTableQuestions (WTQ) (test)
81.4AccuracyMOCA-AGENT
Evaluation results
| Method | Links | ||
|---|---|---|---|
| MOCA-AGENTBackbone=Qwen3.6-27B2026.06 | 81.4 | — | |
| ARTEMIS-DABackbone=GPT-4, Evaluation Protocol=Prompting / in-context learning with LLMs2026.06 | 80.8 | — | |
| TableMasterBackbone=GPT-4o-mini, Evaluation Protocol=Prompting / in-context learning with LLMs2026.06 | 78.13 | — | |
| SS-CoTBackbone=Llama-3.1-70B, Evaluation Protocol=Prompting / in-context learning with LLMs2026.06 | 76.8 | — | |
| TIDE-AgentBackbone=GPT-3.5, Evaluation Protocol=Prompting / in-context learning with LLMs2026.06 | 75 | — | |
| TabAFBackbone=Qwen2.5-Coder-7B, Evaluation Protocol=Fine-tuning approaches2026.06 | 74.72 | — | |
| SynTQABackbone=GPT-4, Evaluation Protocol=Prompting / in-context learning with LLMs2026.06 | 74.4 | — | |
| Norm-DP&AgentBackbone=GPT-3.5, Evaluation Protocol=Prompting / in-context learning with LLMs2026.06 | 73.65 | — | |
| Mix-SCBackbone=GPT-3.5, Evaluation Protocol=Prompting / in-context learning with LLMs2026.06 | 73.6 | — | |
| CABINETBackbone=FLAN-T5/LLM, Evaluation Protocol=Prompting / in-context learning with LLMs2026.06 | 69.1 | — | |
| ReAcTableBackbone=Codex, Evaluation Protocol=Prompting / in-context learning with LLMs2026.06 | 68 | — | |
| Chain-of-TableBackbone=PaLM-2, Evaluation Protocol=Prompting / in-context learning with LLMs2026.06 | 67.31 | — | |
| DaterBackbone=Codex, Evaluation Protocol=Prompting / in-context learning with LLMs2026.06 | 65.9 | — | |
| LEVERBackbone=Codex, Evaluation Protocol=Prompting / in-context learning with LLMs2026.06 | 65.8 | — | |
| BinderBackbone=Codex, Evaluation Protocol=Prompting / in-context learning with LLMs2026.06 | 64.6 | — | |
| OmniTabBackbone=BART-Large, Evaluation Protocol=Fine-tuning approaches2026.06 | 62.8 | — | |
| TableGPT2Backbone=Qwen2.5-7B, Evaluation Protocol=Fine-tuning approaches2026.06 | 61.42 | — | |
| TAPEX-LargeBackbone=BART-Large, Evaluation Protocol=Fine-tuning approaches2026.06 | 59.1 | — | |
| TableLLMBackbone=Qwen2-7B, Evaluation Protocol=Fine-tuning approaches2026.06 | 53.59 | — | |
| TAPAS-largeBackbone=BERT-large, Evaluation Protocol=Fine-tuning approaches2026.06 | 48.7 | — | |
| Full TripletSetting=Full Triplet2026.05 | 46.59 | 1,279.6 | |
| PaddleOCR-VL-1.5Setting=PaddleOCR-VL-1.52026.05 | 44.42 | 3,270.4 | |
| MinerUSetting=MinerU2026.05 | 42.46 | 1,435.4 | |
| TripletQL Agent†Setting=TripletQL Agent†2026.05 | 40.03 | 1,201.5 | |
| HTML BaselineSetting=HTML Baseline2026.05 | 37.43 | 1,643.4 | |
| TableLlamaBackbone=Llama-2-7B, Evaluation Protocol=Fine-tuning approaches2026.06 | 32.14 | — |