Mathematical Reasoning on AIME25 (P@1, P@4)
5P@1Base
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| BaseModel Size=Qwen2.5-7B2026.06 | 5 | 16.67 | |
| ICTModel Size=Qwen2.5-7B2026.06 | 5 | 17.17 | |
| STAPOModel Size=Qwen2.5-7B2026.06 | 4.17 | 13.33 | |
| GRPOModel Size=Qwen2.5-7B2026.06 | 2.5 | 6.67 | |
| BaseModel Size=Qwen2.5-1.5B2026.06 | 1.67 | 6.67 | |
| ICTModel Size=Qwen2.5-1.5B2026.06 | 1.67 | 6.67 | |
| 20-EntropyModel Size=Qwen2.5-7B2026.06 | 1.67 | 6.67 | |
| STAPOModel Size=Qwen2.5-1.5B2026.06 | 0.83 | 3.33 | |
| 20-EntropyModel Size=Qwen2.5-1.5B2026.06 | 0.43 | 2.11 | |
| GRPOModel Size=Qwen2.5-1.5B2026.06 | 0.21 | 2.47 | |
| BaseModel Size=Qwen2.5-0.5B2026.06 | 0 | 0 | |
| GRPOModel Size=Qwen2.5-0.5B2026.06 | 0 | 0 | |
| 20-EntropyModel Size=Qwen2.5-0.5B2026.06 | 0 | 0 | |
| STAPOModel Size=Qwen2.5-0.5B2026.06 | 0 | 0 | |
| ICTModel Size=Qwen2.5-0.5B2026.06 | 0 | 0 |