Language-to-Code Generation on WikiTQ official (dev)
70.9Execution AccuracyOracle
Evaluation results
| Method | Links | |
|---|---|---|
| Oraclegenerator_model=code-davinci-0022023.02 | 70.9 | |
| Codex Binderfinetuning=false2023.02 | 62.6 | |
| LEVERgenerator_model=code-davinci-0022023.02 | 61.1 | |
| TaCubefinetuning=true2023.02 | 59.7 | |
| Codex SQLfinetuning=false2023.02 | 57.6 | |
| TaPEX*finetuning=true2023.02 | 57 | |
| EP + Votinggenerator_model=code-davinci-0022023.02 | 50.6 | |
| EP + MLgenerator_model=code-davinci-0022023.02 | 50.1 | |
| Codex QAfinetuning=false2023.02 | 49.3 | |
| MLgenerator_model=code-davinci-0022023.02 | 48.3 | |
| Greedygenerator_model=code-davinci-0022023.02 | 47.2 |