Unit test generation on MBPP+ (test)
0.53Error RateCVeDRL
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| CVeDRLScale=0.6B2026.01 | 0.53 | 15.79 | 83.68 | 97.37 | 3.13 | |
| CodeRMScale=8.0B2026.01 | 2.44 | 52.86 | 44.7 | 97.11 | 7.88 | |
| GPT-4oScale=-2026.01 | 3.98 | 29.89 | 66.13 | 96.91 | 6.12 | |
| GPT-3.5Scale=-2026.01 | 5.14 | 40.15 | 54.71 | 96.65 | 5.97 | |
| Qwen3Scale=32B2026.01 | 11.42 | 31.39 | 57.19 | 92.44 | 7.47 | |
| LLaMA3.1Scale=8.0B2026.01 | 15.79 | 47.53 | 36.68 | 95.93 | 4.13 | |
| Qwen3Scale=0.6B2026.01 | 28.18 | 31.53 | 40.29 | 90.12 | 3.48 |