Codebase generation on Online Social Media App (test)
82.4Feature Completion (%)Code-L2MAC
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Code-L2MAC2023.10 | 82.4 | 0 | 395 | 1,830 | |
| AutoGPTBackbone=GPT-42023.10 | 33.3 | 0.6 | 148 | 300 | |
| CodeT2023.10 | 19.5 | 16.4 | 110 | 181 | |
| Reflexion2023.10 | 19.5 | 10.2 | 10.5 | 0 | |
| Self-Refine2023.10 | 15.2 | 2.53 | 122 | 133 | |
| GPT42023.10 | 13.5 | 4.09 | 116 | 81.8 |