Text-to-SQL on Geoquery
83Exact Match AccuracyLIR+RIR
Evaluation results
| Method | Links | ||
|---|---|---|---|
| LIR+RIRBackbone=T5-base2023.05 | 83 | — | |
| SpanBasedSP2023.05 | 82.2 | — | |
| BaselineBackbone=T5-large2023.05 | 81.9 | — | |
| LSPBackbone=T5-base2023.05 | 81.3 | — | |
| BaselineBackbone=T5-base2023.05 | 79.7 | — | |
| BaselineBackbone=T5-3B2023.05 | 79.7 | — | |
| LIR+RIRBackbone=T5-large2023.05 | 79.7 | — | |
| LIR+RIRBackbone=T5-3B2023.05 | 75.8 | — | |
| Seq2Seq+ST2023.05 | 63.6 | — | |
| Transformer+ST2023.05 | 61.9 | — | |
| Transformer2023.05 | 53.9 | — | |
| GECA2023.05 | 52.1 | — | |
| Seq2Seq (Zheng and Lapata)2023.05 | 48.5 | — | |
| TopK + LPRBackbone=Llama-2-7B-Chat, Noise Type=Relevant Noise, Noise Ratio=60%2024.05 | 37.74 | — | |
| DPP + LPRBackbone=Llama-2-7B-Chat, Noise Type=Relevant Noise, Noise Ratio=60%2024.05 | 36.81 | — | |
| Random + LPRBackbone=Llama-2-7B-Chat, Noise Type=Relevant Noise, Noise Ratio=60%2024.05 | 26.73 | — | |
| Seq2Seq (Finegan-Dollak et al.)2023.05 | 20 | — |