Mathematical Reasoning on AMC2023 Kaggle (test)
75Base Accuracyentropy-guided decoding
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| entropy-guided decodingModel=GPT4o2026.03 | 75 | 98.25 | 1.9 | 12 | 75.9 | 1.33 | 51 | |
| entropy-guided decodingModel=Llama3.1-8B2026.03 | 45 | 97.5 | 1.6 | 21 | 78.6 | 1.21 | 16.9 |