Table Question Answering on ArcadeQA
49.2Exact Match AccuracyTableRAG
Evaluation results
| Method | Links | |
|---|---|---|
| TableRAGLanguage Model=GPT 3.5 Turbo2024.10 | 49.2 | |
| TableRAGLanguage Model=Mistral Nemo2024.10 | 46.2 | |
| ReadSchemaLanguage Model=GPT 3.5 Turbo2024.10 | 43.1 | |
| RandRowSamplingLanguage Model=GPT 3.5 Turbo2024.10 | 42.3 | |
| TableRAGLanguage Model=Gemini 1.0 Pro2024.10 | 42.3 | |
| RowColRetrievalLanguage Model=GPT 3.5 Turbo2024.10 | 37.7 | |
| ReadSchemaLanguage Model=Mistral Nemo2024.10 | 32.3 | |
| ReadSchemaLanguage Model=Gemini 1.0 Pro2024.10 | 30.8 | |
| RowColRetrievalLanguage Model=Mistral Nemo2024.10 | 30 | |
| RandRowSamplingLanguage Model=Mistral Nemo2024.10 | 28.5 | |
| RandRowSamplingLanguage Model=Gemini 1.0 Pro2024.10 | 20.8 | |
| RowColRetrievalLanguage Model=Gemini 1.0 Pro2024.10 | 16.9 | |
| ReadTableLanguage Model=Mistral Nemo2024.10 | 5.4 | |
| ReadTableLanguage Model=GPT 3.5 Turbo2024.10 | 4.6 | |
| ReadTableLanguage Model=Gemini 1.0 Pro2024.10 | 1.5 |