Mathematical Problem Solving on MINERVA (Pass@k)
37.5Pass@kGCPO
Evaluation Results
| Method | Links | |
|---|---|---|
| GCPOk=64, Base Model=Qwen3-1.7B2026.05 | 37.5 | |
| DQOk=64, Base Model=Qwen3-1.7B2026.05 | 37 | |
| GRPOk=64, Base Model=Qwen3-1.7B2026.05 | 36.6 | |
| GCPOk=32, Base Model=Qwen3-1.7B2026.05 | 36.4 | |
| DQOk=32, Base Model=Qwen3-1.7B2026.05 | 35.8 | |
| GRPOk=32, Base Model=Qwen3-1.7B2026.05 | 35.1 | |
| GCPOk=16, Base Model=Qwen3-1.7B2026.05 | 34.6 | |
| DQOk=16, Base Model=Qwen3-1.7B2026.05 | 34.1 | |
| GRPOk=16, Base Model=Qwen3-1.7B2026.05 | 33.2 | |
| GCPOk=8, Base Model=Qwen3-1.7B2026.05 | 32.8 | |
| Basek=64, Base Model=Qwen3-1.7B2026.05 | 32 | |
| DQOk=8, Base Model=Qwen3-1.7B2026.05 | 31.9 | |
| GRPOk=8, Base Model=Qwen3-1.7B2026.05 | 31.5 | |
| GCPOk=4, Base Model=Qwen3-1.7B2026.05 | 30 | |
| Basek=32, Base Model=Qwen3-1.7B2026.05 | 29.7 | |
| DQOk=4, Base Model=Qwen3-1.7B2026.05 | 29.3 | |
| GRPOk=4, Base Model=Qwen3-1.7B2026.05 | 28.6 | |
| Basek=16, Base Model=Qwen3-1.7B2026.05 | 27.6 | |
| GCPOk=2, Base Model=Qwen3-1.7B2026.05 | 26.9 | |
| DQOk=2, Base Model=Qwen3-1.7B2026.05 | 26.6 | |
| GRPOk=2, Base Model=Qwen3-1.7B2026.05 | 25.7 | |
| Basek=8, Base Model=Qwen3-1.7B2026.05 | 25.4 | |
| Basek=4, Base Model=Qwen3-1.7B2026.05 | 22.5 | |
| Basek=2, Base Model=Qwen3-1.7B2026.05 | 20.1 |