Mathematical Problem Solving on Olympiad (Pass@k)
47.1Pass@kGCPO
Evaluation Results
| Method | Links | |
|---|---|---|
| GCPOk=64, Base Model=Qwen3-1.7B2026.05 | 47.1 | |
| GCPOk=32, Base Model=Qwen3-1.7B2026.05 | 46.5 | |
| DQOk=64, Base Model=Qwen3-1.7B2026.05 | 45.4 | |
| GCPOk=16, Base Model=Qwen3-1.7B2026.05 | 45.2 | |
| GRPOk=64, Base Model=Qwen3-1.7B2026.05 | 44.3 | |
| DQOk=32, Base Model=Qwen3-1.7B2026.05 | 44.1 | |
| GCPOk=8, Base Model=Qwen3-1.7B2026.05 | 43.9 | |
| Basek=64, Base Model=Qwen3-1.7B2026.05 | 43.8 | |
| DQOk=16, Base Model=Qwen3-1.7B2026.05 | 43.7 | |
| GRPOk=32, Base Model=Qwen3-1.7B2026.05 | 43.6 | |
| GRPOk=16, Base Model=Qwen3-1.7B2026.05 | 42.8 | |
| DQOk=8, Base Model=Qwen3-1.7B2026.05 | 42.4 | |
| Basek=32, Base Model=Qwen3-1.7B2026.05 | 42.3 | |
| GCPOk=4, Base Model=Qwen3-1.7B2026.05 | 41.6 | |
| GRPOk=8, Base Model=Qwen3-1.7B2026.05 | 41.1 | |
| Basek=16, Base Model=Qwen3-1.7B2026.05 | 41 | |
| DQOk=4, Base Model=Qwen3-1.7B2026.05 | 40.8 | |
| GRPOk=4, Base Model=Qwen3-1.7B2026.05 | 39.4 | |
| GCPOk=2, Base Model=Qwen3-1.7B2026.05 | 39.2 | |
| Basek=8, Base Model=Qwen3-1.7B2026.05 | 38.3 | |
| DQOk=2, Base Model=Qwen3-1.7B2026.05 | 37.5 | |
| GRPOk=2, Base Model=Qwen3-1.7B2026.05 | 36 | |
| Basek=4, Base Model=Qwen3-1.7B2026.05 | 35.1 | |
| Basek=2, Base Model=Qwen3-1.7B2026.05 | 30.8 |