Table Question Answering on WikiTableQuestions (WTQ) complete (test)
73.6AccuracyOurs (NORM + Mix Self-Consistency)
Evaluation results
| Method | Links | |
|---|---|---|
| Ours (NORM + Mix Self-Consistency)Type=LLMs Based Methods, Backbone=GPT-3.5, Mode=zero-shot2023.12 | 73.6 | |
| DATERType=LLMs Based Methods, Backbone=Codex2023.12 | 65.9 | |
| LEVERType=LLMs Based Methods, Backbone=Codex2023.12 | 65.8 | |
| BINDERType=LLMs Based Methods, Backbone=Codex2023.12 | 64.6 | |
| OMNITABType=Fine-tuning Based Models2023.12 | 63.3 | |
| REASTAPType=Fine-tuning Based Models2023.12 | 58.7 | |
| TAPAXType=Fine-tuning Based Models2023.12 | 57.5 | |
| BINDERType=LLMs Based Methods, Backbone=GPT-3.52023.12 | 55.5 | |
| T5-3BType=Fine-tuning Based Models2023.12 | 49.3 | |
| TAPASType=Fine-tuning Based Models2023.12 | 48.8 | |
| STRUCTGPTType=LLMs Based Methods, Backbone=GPT-3.52023.12 | 48.4 |