Code Generation on HumanEval (Score, APF, TPF, APT Profile)
4.7Speedup (x)DOUBLE
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| DOUBLEBase Model=LLaMA3.3-70B, Temperature=1, Decoding Setting=non-greedy2026.01 | 4.7 | — | — | — | — | 12.66 | |
| EAGLE3Base Model=LLaMA3.3-70B, Temperature=1, Decoding Setting=non-greedy2026.01 | 4.24 | — | — | — | — | 6.05 | |
| PEARLBase Model=LLaMA3.3-70B, Temperature=1, Decoding Setting=non-greedy2026.01 | 4.13 | — | — | — | — | 10.06 | |
| DOUBLEBase Model=Qwen3-32B, Temperature=1, Decoding Setting=non-greedy2026.01 | 2.5 | — | — | — | — | 8.9 | |
| PEARLBase Model=Qwen3-32B, Temperature=1, Decoding Setting=non-greedy2026.01 | 2.21 | — | — | — | — | 7.31 | |
| TEAMEvaluation Protocol=0-shot2026.02 | 2.2 | 79.88 | 34.48 | 5.07 | 6.8 | — | |
| EAGLE3Base Model=Qwen3-32B, Temperature=1, Decoding Setting=non-greedy2026.01 | 2.11 | — | — | — | — | 2.63 | |
| VanillaEvaluation Protocol=0-shot2026.02 | 1 | 79.27 | 53.34 | 2.91 | 18.33 | — |