Unit test generation on LeetCode (test)
2.77Error Rate (%)GPT-4o
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| GPT-4oScale=-2026.01 | 2.77 | 25.14 | 72.09 | 83.64 | 5.77 | |
| CVeDRLScale=0.6B2026.01 | 3.49 | 20.98 | 75.53 | 91.61 | 2.84 | |
| GPT-3.5Scale=-2026.01 | 3.53 | 37.28 | 59.19 | 76.53 | 5.63 | |
| CodeRMScale=8.0B2026.01 | 3.7 | 58.16 | 38.14 | 75.37 | 6.43 | |
| LLaMA3.1Scale=8.0B2026.01 | 12.47 | 54.77 | 32.76 | 70.49 | 3.88 | |
| Qwen3Scale=32B2026.01 | 23.18 | 27.63 | 49.19 | 68.62 | 7.34 | |
| Qwen3Scale=0.6B2026.01 | 37.31 | 42.26 | 20.43 | 61.47 | 3.76 |