Code Debug on InfiniteBench Code Debug
74.37AccuracyFull Prompt
Evaluation Results
| Method | Links | |
|---|---|---|
| Full PromptTarget Model=DeepSeek-R1-0528, Keep rate=–2026.03 | 74.37 | |
| SP w/ Llama-3.1-8B-InsTarget Model=DeepSeek-R1-0528, Draft Model=Llama-3.1-8B-Ins, Keep rate=30%2026.03 | 70.3 | |
| SP w/ Llama-3.1-8B-InsTarget Model=DeepSeek-R1-0528, Draft Model=Llama-3.1-8B-Ins, Keep rate=25%2026.03 | 68.02 | |
| Full PromptTarget Model=DeepSeek-V3.1, Keep rate=–2026.03 | 67.51 | |
| SP w/ Llama-3.1-8B-InsTarget Model=DeepSeek-V3.1, Draft Model=Llama-3.1-8B-Ins, Keep rate=20%2026.03 | 64.72 | |
| SP w/ Llama-3.1-8B-InsTarget Model=DeepSeek-R1-0528, Draft Model=Llama-3.1-8B-Ins, Keep rate=15%2026.03 | 62.44 | |
| SP w/ Llama-3.1-8B-InsTarget Model=DeepSeek-V3.1, Draft Model=Llama-3.1-8B-Ins, Keep rate=15%2026.03 | 59.13 |