Code Generation on CodeContests official (val)
43.6Pass@4CPPO
Evaluation Results
| Method | Links | |
|---|---|---|
| CPPOBase model=Qwen3.5-9B2026.05 | 43.6 | |
| Tuple Planner SFTBase model=Qwen3.5-9B2026.05 | 28.5 | |
| CPPOBase model=Qwen3.5-4B2026.05 | 24.3 | |
| CPPOBase model=Qwen3.5-2B2026.05 | 20.1 | |
| Pass@K Training / RLVRBase model=Qwen3.5-9B2026.05 | 18.5 | |
| PKPOBase model=Qwen3.5-9B2026.05 | 18.3 | |
| Direct SolveBase model=Qwen3.5-9B2026.05 | 17.5 | |
| Plan-and-SolveBase model=Qwen3.5-9B2026.05 | 17.1 | |
| PlanSearchBase model=Qwen3.5-9B2026.05 | 16.8 | |
| PKPOBase model=Qwen3.5-4B2026.05 | 15.6 | |
| Tuple Planner SFTBase model=Qwen3.5-4B2026.05 | 14.2 | |
| UpSkillBase model=Qwen3.5-9B2026.05 | 13.6 | |
| PlanSearchBase model=Qwen3.5-4B2026.05 | 12.8 | |
| UpSkillBase model=Qwen3.5-4B2026.05 | 12.1 | |
| Pass@K Training / RLVRBase model=Qwen3.5-4B2026.05 | 11.5 | |
| Plan-and-SolveBase model=Qwen3.5-4B2026.05 | 9.8 | |
| Direct SolveBase model=Qwen3.5-4B2026.05 | 9.4 | |
| Tuple Planner SFTBase model=Qwen3.5-2B2026.05 | 8.7 | |
| Pass@K Training / RLVRBase model=Qwen3.5-2B2026.05 | 8.5 | |
| Plan-and-SolveBase model=Qwen3.5-2B2026.05 | 6.5 | |
| PKPOBase model=Qwen3.5-2B2026.05 | 6.2 | |
| PlanSearchBase model=Qwen3.5-2B2026.05 | 4.3 | |
| UpSkillBase model=Qwen3.5-2B2026.05 | 4.1 | |
| Direct SolveBase model=Qwen3.5-2B2026.05 | 4 |