Logical Reasoning on SynLogic
8.7pass@1F-GRPO
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| F-GRPOBackbone=Qwen2.5-7B, Group Size (N)=82026.02 | 8.7 | 57 | |
| F-CISPOBackbone=Qwen2.5-7B, Group Size (N)=82026.02 | 8.2 | 60 | |
| CISPOBackbone=Qwen2.5-7B, Group Size (N)=82026.02 | 8 | 53.6 | |
| GRPOBackbone=Qwen2.5-7B, Group Size (N)=82026.02 | 7.9 | 51.3 | |
| F-DAPOBackbone=Qwen2.5-7B, Group Size (N)=82026.02 | 7.9 | 62.4 | |
| DAPOBackbone=Qwen2.5-7B, Group Size (N)=82026.02 | 7.5 | 53.3 | |
| F-CISPOBackbone=Llama-3.2-3B-Instruct, Group Size (N)=82026.02 | 5.4 | 33.9 | |
| CISPOBackbone=Qwen2.5-1.5B-Math, Group Size (N)=82026.02 | 5.2 | 26.3 | |
| DAPOBackbone=Qwen2.5-1.5B-Math, Group Size (N)=82026.02 | 5 | 26.9 | |
| GRPOBackbone=Qwen2.5-1.5B-Math, Group Size (N)=82026.02 | 4.9 | 27.4 | |
| F-DAPOBackbone=Qwen2.5-1.5B-Math, Group Size (N)=82026.02 | 4.9 | 26.4 | |
| F-CISPOBackbone=Qwen2.5-1.5B-Math, Group Size (N)=82026.02 | 4.9 | 26.1 | |
| F-GRPOBackbone=Qwen2.5-1.5B-Math, Group Size (N)=82026.02 | 4.8 | 27.5 | |
| DAPOBackbone=Llama-3.2-3B-Instruct, Group Size (N)=82026.02 | 4.8 | 28.9 | |
| GRPOBackbone=Llama-3.2-3B-Instruct, Group Size (N)=82026.02 | 4.7 | 36.4 | |
| F-GRPOBackbone=Llama-3.2-3B-Instruct, Group Size (N)=82026.02 | 4.6 | 35.5 | |
| F-DAPOBackbone=Llama-3.2-3B-Instruct, Group Size (N)=82026.02 | 4.3 | 33 | |
| CISPOBackbone=Llama-3.2-3B-Instruct, Group Size (N)=82026.02 | 4.3 | 29.4 |