Math Reasoning on AIME 25 (accuracy (%))
30Accuracy (%)EGLR-OracleL
Evaluation Results
| Method | Links | |
|---|---|---|
| EGLR-OracleLModel=Qwen2.5-Math-1.5B-Instruct2026.06 | 30 | |
| EGLR-OracleLModel=Qwen2.5-7B-Instruct2026.06 | 26.7 | |
| EGLR-OracleLModel=Qwen2.5-14B-Instruct2026.06 | 26.7 | |
| EGLR-OracleLModel=Qwen2.5-Math-7B-Instruct2026.06 | 26.7 | |
| EGLR (best L)Model=Qwen2.5-14B-Instruct2026.06 | 20 | |
| EGLR-SCL (B = 10)Model=Qwen2.5-14B-Instruct2026.06 | 20 | |
| EGLR (best L)Model=Qwen2.5-Math-1.5B-Instruct2026.06 | 20 | |
| EGLR-SCL (B = 10)Model=Qwen2.5-7B-Instruct2026.06 | 16.7 | |
| GreedyModel=Qwen2.5-14B-Instruct2026.06 | 16.7 | |
| EGLR-OracleLModel=Qwen2.5-3B-Instruct2026.06 | 13.3 | |
| EGLR (best L)Model=Qwen2.5-7B-Instruct2026.06 | 13.3 | |
| GreedyModel=Qwen2.5-Math-1.5B-Instruct2026.06 | 13.3 | |
| EGLR-SCL (B = 10)Model=Qwen2.5-Math-1.5B-Instruct2026.06 | 13.3 | |
| GreedyModel=Qwen2.5-Math-7B-Instruct2026.06 | 13.3 | |
| EGLR (best L)Model=Qwen2.5-Math-7B-Instruct2026.06 | 13.3 | |
| EGLR-SCL (B = 10)Model=Qwen2.5-Math-7B-Instruct2026.06 | 13.3 | |
| GreedyModel=Qwen2.5-7B-Instruct2026.06 | 10 | |
| EGLR (best L)Model=Qwen2.5-3B-Instruct2026.06 | 6.7 | |
| EGLR-SCL (B = 10)Model=Qwen2.5-3B-Instruct2026.06 | 6.7 | |
| EGLR-SCL (B = 10)Model=Llama-3.1-8B-Instruct2026.06 | 6.7 | |
| EGLR-OracleLModel=Llama-3.1-8B-Instruct2026.06 | 6.7 | |
| GreedyModel=Qwen2.5-3B-Instruct2026.06 | 3.3 | |
| EGLR (best L)Model=Llama-3.1-8B-Instruct2026.06 | 3.3 | |
| GreedyModel=Qwen2.5-0.5B-Instruct2026.06 | 0 | |
| EGLR (best L)Model=Qwen2.5-0.5B-Instruct2026.06 | 0 | |
| EGLR-SCL (B = 10)Model=Qwen2.5-0.5B-Instruct2026.06 | 0 | |
| EGLR-OracleLModel=Qwen2.5-0.5B-Instruct2026.06 | 0 | |
| GreedyModel=Llama-3.1-8B-Instruct2026.06 | 0 | |
| GreedyModel=Mistral-7B-Instruct-v0.32026.06 | 0 | |
| EGLR (best L)Model=Mistral-7B-Instruct-v0.32026.06 | 0 | |
| EGLR-SCL (B = 10)Model=Mistral-7B-Instruct-v0.32026.06 | 0 | |
| EGLR-OracleLModel=Mistral-7B-Instruct-v0.32026.06 | 0 |