Code Translation on CodeNet
93.6Computational Accuracy (C)SWIFTTRANS
Evaluation results
| Method | Links | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| SWIFTTRANSLLM=Qwen2.5-7B2026.06 | 93.6 | 95 | 96.1 | 94.9 | 94.6 | 190 | 216 | 145 | 106 | 122 | 90.2 | 292 | |
| SWIFTTRANSLLM=StarCoder-7B2026.06 | 92.3 | 93.8 | 95.4 | 94.3 | 94.8 | 203 | 242 | 168 | 123 | 126 | 89.4 | 311 | |
| SWIFTTRANSLLM=Qwen2.5-3B2026.06 | 91.8 | 92.7 | 89.7 | 93.4 | 94 | 218 | 269 | 223 | 138 | 146 | 86.9 | 339 | |
| F2STransLLM=Qwen2.5-7B2026.06 | 91 | 91.4 | 86.8 | 88.5 | 91.1 | 470 | 711 | 303 | 175 | 320 | 84.6 | 731 | |
| Cor.-OnlyLLM=GPT-52026.06 | 87.8 | 91.4 | 91.9 | 81.8 | 90.3 | 391 | 435 | 355 | 161 | 187 | 86.4 | 528 | |
| F2STransLLM=Qwen2.5-3B2026.06 | 86.4 | 89.8 | 85.6 | 86.5 | 83.6 | 494 | 613 | 340 | 164 | 336 | 82 | 744 | |
| Cor.+Eff.LLM=GPT-52026.06 | 82.9 | 88.5 | 89.1 | 81.2 | 80.5 | 376 | 357 | 338 | 137 | 167 | 79.8 | 453 | |
| Cor.+Eff.LLM=Qwen3-Next-80B2026.06 | 79.7 | 79.2 | 66.8 | 74.6 | 77.9 | 455 | 529 | 295 | 173 | 256 | 68.6 | 667 | |
| Cor.-OnlyLLM=Qwen3-Next-80B2026.06 | 79.3 | 81.5 | 71.2 | 77.3 | 80.9 | 514 | 685 | 363 | 174 | 315 | 73.3 | 776 | |
| Cor.→Eff.LLM=Qwen3-Next-80B2026.06 | 68.9 | 72.9 | 69.5 | 69.3 | 70.1 | 364 | 504 | 285 | 121 | 228 | 59.3 | 565 | |
| Cor.→Eff.LLM=GPT-52026.06 | 68.4 | 62.9 | 66.9 | 70.2 | 61.3 | 322 | 329 | 328 | 126 | 143 | 62.7 | 399 |