Code Generation on MBPP+ (accuracy %)
75.1Accuracy (%)Qwen3-30B-A3B-Instruct-2507
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen3-30B-A3B-Instruct-2507Role=Teacher Model2026.06 | 75.1 | |
| Teacher ModelModel=Qwen3-4B-Instruct-2507, Teacher-Student Overlap=Large2026.06 | 71.3 | |
| IW-OPDStudent=Qwen3-4B2026.06 | 70.9 | |
| OPDStudent=Qwen3-4B2026.06 | 69.1 | |
| IW-OPDStudent Model=Qwen3-4B, Teacher Model=Qwen3-4B-Instruct-25072026.06 | 68.2 | |
| IW-OPDStudent=Qwen3-4B, Training Step=102026.06 | 68.1 | |
| OPDStudent Model=Qwen3-4B, Teacher Model=Qwen3-4B-Instruct-25072026.06 | 67.8 | |
| IW-OPD10Student Model=Qwen3-4B, Training Step=10, Teacher Model=Qwen3-4B-Instruct-25072026.06 | 67.5 | |
| OPDStudent=Qwen3-4B, Training Step=102026.06 | 66.7 | |
| OPD10Student Model=Qwen3-4B, Training Step=10, Teacher Model=Qwen3-4B-Instruct-25072026.06 | 66.1 | |
| DynaSteerBackbone=Qwen3-8B2026.06 | 65.61 | |
| BaseStudent Model=Qwen3-4B, Teacher Model=Qwen3-4B-Instruct-25072026.06 | 64.5 | |
| BaseStudent=Qwen3-4B2026.06 | 64.5 | |
| ROMEBackbone=Qwen3-8B2026.06 | 55.82 | |
| IW-OPDStudent=Qwen3-1.7B2026.06 | 55 | |
| IW-OPDStudent Model=Qwen3-1.7B, Teacher Model=Qwen3-4B-Instruct-25072026.06 | 54.9 | |
| IW-OPDStudent=Qwen3-1.7B, Training Step=102026.06 | 54.8 | |
| DynaSteerBackbone=Qwen3-1.7B2026.06 | 54.5 | |
| IW-OPD10Student Model=Qwen3-1.7B, Training Step=10, Teacher Model=Qwen3-4B-Instruct-25072026.06 | 54 | |
| OPDStudent Model=Qwen3-1.7B, Teacher Model=Qwen3-4B-Instruct-25072026.06 | 53.7 | |
| OPDStudent=Qwen3-1.7B2026.06 | 53.7 | |
| MEMITBackbone=Qwen3-8B2026.06 | 53.7 | |
| OPD10Student Model=Qwen3-1.7B, Training Step=10, Teacher Model=Qwen3-4B-Instruct-25072026.06 | 53.4 | |
| OPDStudent=Qwen3-1.7B, Training Step=102026.06 | 52.6 | |
| BaseStudent Model=Qwen3-1.7B, Teacher Model=Qwen3-4B-Instruct-25072026.06 | 52.5 | |
| BaseStudent=Qwen3-1.7B2026.06 | 52.5 | |
| TruthFlowBackbone=Qwen3-1.7B2026.06 | 52.12 | |
| TruthXBackbone=Qwen3-1.7B2026.06 | 51.06 | |
| IW-OPDStudent Model=Qwen3-0.6B, Teacher Model=Qwen3-4B-Instruct-25072026.06 | 35.7 | |
| IW-OPD10Student Model=Qwen3-0.6B, Training Step=10, Teacher Model=Qwen3-4B-Instruct-25072026.06 | 33.9 | |
| OPDStudent Model=Qwen3-0.6B, Teacher Model=Qwen3-4B-Instruct-25072026.06 | 33.3 | |
| IW-OPDStudent=Qwen3-0.6B2026.06 | 31.9 | |
| OPD10Student Model=Qwen3-0.6B, Training Step=10, Teacher Model=Qwen3-4B-Instruct-25072026.06 | 31 | |
| OPDStudent=Qwen3-0.6B2026.06 | 28.7 | |
| BaseStudent Model=Qwen3-0.6B, Teacher Model=Qwen3-4B-Instruct-25072026.06 | 28.4 | |
| BaseStudent=Qwen3-0.6B2026.06 | 28.4 | |
| IW-OPDStudent=Qwen3-0.6B, Training Step=102026.06 | 27.5 | |
| OPDStudent=Qwen3-0.6B, Training Step=102026.06 | 23.1 |