Mathematical Reasoning on AIME 2024 (M->AM and G8->AM Accuracy)
3.3Accuracy (G8->AM)Base
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| BaseModel Size=7B2025.12 | 3.3 | 3.3 | |
| GRPOModel Size=7B2025.12 | 3.3 | 20 | |
| FT-EModel Size=7B2025.12 | 3.3 | 10 | |
| FT-SModel Size=7B2025.12 | 3.3 | 16.7 | |
| iCLPModel Size=7B2025.12 | 3.3 | 20 | |
| BaseModel Size=0.5B2025.12 | 0 | 0 | |
| GRPOModel Size=0.5B2025.12 | 0 | 0 | |
| FT-EModel Size=0.5B2025.12 | 0 | 0 | |
| FT-SModel Size=0.5B2025.12 | 0 | 0 | |
| FT-LEModel Size=0.5B2025.12 | 0 | 0 | |
| iCLPModel Size=0.5B2025.12 | 0 | 0 | |
| BaseModel Size=3B2025.12 | 0 | 0 | |
| GRPOModel Size=3B2025.12 | 0 | 10 | |
| FT-EModel Size=3B2025.12 | 0 | 0 | |
| FT-SModel Size=3B2025.12 | 0 | 7.6 | |
| FT-LEModel Size=3B2025.12 | 0 | — | |
| iCLPModel Size=3B2025.12 | 0 | 10 |