Mathematical Reasoning on MATH500 (Avg@8, Pass@8)
45.01Avg@8 Score (MATH500)GRPO
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| GRPOStudent Model=Llama-3.2-3B-Instruct, Teacher Model=Llama-3.1-8B-Instruct2026.03 | 45.01 | 66 | |
| EOPDStudent Model=Llama-3.2-3B-Instruct, Teacher Model=Llama-3.1-8B-Instruct2026.03 | 44.9 | 67.2 | |
| OPDStudent Model=Llama-3.2-3B-Instruct, Teacher Model=Llama-3.1-8B-Instruct2026.03 | 43.6 | 64.8 | |
| KDStudent Model=Llama-3.2-3B-Instruct, Teacher Model=Llama-3.1-8B-Instruct2026.03 | 42.6 | 65.8 |