Mathematical Reasoning on Olympiad (pass@1, pass@256)
42.4Pass@1F-CISPO
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| F-CISPOBackbone=Qwen2.5-7B, Group Size (N)=82026.02 | 42.4 | 77.5 | |
| CISPOBackbone=Qwen2.5-7B, Group Size (N)=82026.02 | 41.5 | 76.9 | |
| F-DAPOBackbone=Qwen2.5-7B, Group Size (N)=82026.02 | 40.9 | 75.6 | |
| DAPOBackbone=Qwen2.5-7B, Group Size (N)=82026.02 | 40.5 | 71.8 | |
| GRPOBackbone=Qwen2.5-7B, Group Size (N)=82026.02 | 37.8 | 65.8 | |
| CISPOBackbone=Qwen2.5-1.5B-Math, Group Size (N)=82026.02 | 37.6 | 71.9 | |
| F-GRPOBackbone=Qwen2.5-7B, Group Size (N)=82026.02 | 37.5 | 71.6 | |
| F-DAPOBackbone=Qwen2.5-1.5B-Math, Group Size (N)=82026.02 | 37.2 | 76.2 | |
| F-GRPOBackbone=Qwen2.5-1.5B-Math, Group Size (N)=82026.02 | 37 | 75.5 | |
| DAPOBackbone=Qwen2.5-1.5B-Math, Group Size (N)=82026.02 | 36.4 | 73.2 | |
| F-CISPOBackbone=Qwen2.5-1.5B-Math, Group Size (N)=82026.02 | 36.4 | 74 | |
| GRPOBackbone=Qwen2.5-1.5B-Math, Group Size (N)=82026.02 | 36.3 | 74.3 | |
| F-DAPOBackbone=Llama-3.2-3B-Instruct, Group Size (N)=82026.02 | 23.2 | 61.3 | |
| CISPOBackbone=Llama-3.2-3B-Instruct, Group Size (N)=82026.02 | 22.5 | 55.4 | |
| F-CISPOBackbone=Llama-3.2-3B-Instruct, Group Size (N)=82026.02 | 21.5 | 58.7 | |
| DAPOBackbone=Llama-3.2-3B-Instruct, Group Size (N)=82026.02 | 21 | 48.4 | |
| F-GRPOBackbone=Llama-3.2-3B-Instruct, Group Size (N)=82026.02 | 20.1 | 61.3 | |
| GRPOBackbone=Llama-3.2-3B-Instruct, Group Size (N)=82026.02 | 19.4 | 59.3 |