Math Reasoning on AMC 23 (accuracy (%))
82.5Accuracy (%)EGLR-OracleL
Evaluation results
| Method | Links | |
|---|---|---|
| EGLR-OracleLModel=Qwen2.5-7B-Instruct2026.06 | 82.5 | |
| EGLR-OracleLModel=Qwen2.5-14B-Instruct2026.06 | 82.5 | |
| EGLR-OracleLModel=Qwen2.5-Math-7B-Instruct2026.06 | 82.5 | |
| EGLR-OracleLModel=Qwen2.5-Math-1.5B-Instruct2026.06 | 77.5 | |
| GreedyModel=Qwen2.5-Math-7B-Instruct2026.06 | 70 | |
| EGLR-OracleLModel=Qwen2.5-3B-Instruct2026.06 | 67.5 | |
| GreedyModel=Qwen2.5-14B-Instruct2026.06 | 67.5 | |
| EGLR (best L)Model=Qwen2.5-14B-Instruct2026.06 | 67.5 | |
| EGLR-SCL (B = 10)Model=Qwen2.5-14B-Instruct2026.06 | 67.5 | |
| EGLR (best L)Model=Qwen2.5-Math-7B-Instruct2026.06 | 67.5 | |
| EGLR-SCL (B = 10)Model=Qwen2.5-Math-1.5B-Instruct2026.06 | 65 | |
| EGLR (best L)Model=Qwen2.5-Math-1.5B-Instruct2026.06 | 62.5 | |
| EGLR-SCL (B = 10)Model=Qwen2.5-Math-7B-Instruct2026.06 | 62.5 | |
| EGLR (best L)Model=Qwen2.5-7B-Instruct2026.06 | 60 | |
| EGLR-SCL (B = 10)Model=Qwen2.5-7B-Instruct2026.06 | 60 | |
| EGLR-SCL (B = 10)Model=Qwen2.5-3B-Instruct2026.06 | 55 | |
| GreedyModel=Qwen2.5-Math-1.5B-Instruct2026.06 | 55 | |
| GreedyModel=Qwen2.5-7B-Instruct2026.06 | 52.5 | |
| EGLR-OracleLModel=Llama-3.1-8B-Instruct2026.06 | 50 | |
| EGLR (best L)Model=Qwen2.5-3B-Instruct2026.06 | 47.5 | |
| EGLR-SCL (B = 10)Model=Llama-3.1-8B-Instruct2026.06 | 47.5 | |
| EGLR-OracleLModel=Qwen2.5-0.5B-Instruct2026.06 | 40 | |
| GreedyModel=Qwen2.5-3B-Instruct2026.06 | 37.5 | |
| EGLR (best L)Model=Llama-3.1-8B-Instruct2026.06 | 30 | |
| EGLR-OracleLModel=Mistral-7B-Instruct-v0.32026.06 | 25 | |
| GreedyModel=Llama-3.1-8B-Instruct2026.06 | 20 | |
| EGLR-SCL (B = 10)Model=Qwen2.5-0.5B-Instruct2026.06 | 15 | |
| EGLR (best L)Model=Qwen2.5-0.5B-Instruct2026.06 | 12.5 | |
| EGLR (best L)Model=Mistral-7B-Instruct-v0.32026.06 | 12.5 | |
| GreedyModel=Qwen2.5-0.5B-Instruct2026.06 | 10 | |
| EGLR-SCL (B = 10)Model=Mistral-7B-Instruct-v0.32026.06 | 2.5 | |
| GreedyModel=Mistral-7B-Instruct-v0.32026.06 | 0 |