Mathematical Reasoning on AIME24 (ACC/maj@8)
43.3ACC/maj@8DiPO
Evaluation Results
| Method | Links | |
|---|---|---|
| DiPOBackbone=Qwen3-8B-Base2026.04 | 43.3 | |
| DAPO w/ ELBackbone=Qwen3-8B-Base2026.04 | 40 | |
| CDEBackbone=Qwen3-8B-Base2026.04 | 40 | |
| GRPOBackbone=Qwen3-4B-Base2026.04 | 36.67 | |
| CDEBackbone=Qwen3-4B-Base2026.04 | 36.67 | |
| DiPOBackbone=Qwen3-4B-Base2026.04 | 36.67 | |
| GRPOBackbone=Qwen3-8B-Base2026.04 | 36.67 | |
| DAPOBackbone=Qwen3-8B-Base2026.04 | 36.67 | |
| DAPOBackbone=Qwen3-4B-Base2026.04 | 33.33 | |
| DAPO w/ ELBackbone=Qwen3-4B-Base2026.04 | 33.33 | |
| DiPOBackbone=Qwen2.5-7B2026.04 | 33.33 | |
| DAPOBackbone=Qwen2.5-7B2026.04 | 30 | |
| DAPO w/ ELBackbone=Qwen2.5-7B2026.04 | 30 | |
| GRPOBackbone=Qwen2.5-7B2026.04 | 26.67 | |
| CDEBackbone=Qwen2.5-7B2026.04 | 23.3 | |
| Base modelBackbone=Qwen3-8B-Base2026.04 | 16.67 | |
| Base modelBackbone=Qwen3-4B-Base2026.04 | 13.3 | |
| Base modelBackbone=Qwen2.5-7B2026.04 | 6.67 |