Uncertainty Estimation on HumanEval-X C++ (test)
0.649AUROCMax Entropy
Evaluation Results
| Method | Links | |
|---|---|---|
| Max EntropyModel=Qwen3-Coder-30B2026.06 | 0.649 | |
| Top-5 EntropyModel=Qwen3-Coder-30B2026.06 | 0.648 | |
| Consistency (BLEU)Model=Qwen3-Coder-30B2026.06 | 0.581 | |
| Consistency (VR)Model=Qwen3-Coder-30B2026.06 | 0.572 | |
| Mean EntropyModel=Qwen3-Coder-30B2026.06 | 0.543 | |
| Mean ProbabilityModel=Qwen3-Coder-30B2026.06 | 0.523 |