Mathematical Reasoning on AIME 2024 (Accuracy and Solution Length)
54.33AccuracyVanilla
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| VanillaBackbone=DeepSeek-R1-Distill-Qwen-7B2026.03 | 54.33 | 10,874 | |
| ConciseHintBackbone=DeepSeek-R1-Distill-Qwen-7B2026.03 | 53.33 | 9,053 | |
| Step EntropyBackbone=DeepSeek-R1-Distill-Qwen-7B2026.03 | 53.33 | 8,567 | |
| SPOT-stage2Backbone=DeepSeek-R1-Distill-Qwen-7B2026.03 | 53.33 | 7,885 | |
| DEERBackbone=DeepSeek-R1-Distill-Qwen-7B2026.03 | 52.67 | 9,306 | |
| SPOT-stage1Backbone=DeepSeek-R1-Distill-Qwen-7B2026.03 | 52.67 | 8,281 | |
| CCoTBackbone=DeepSeek-R1-Distill-Qwen-7B2026.03 | 52.33 | 9,884 | |
| LightThinkerBackbone=DeepSeek-R1-Distill-Qwen-7B2026.03 | 38.67 | 8,354 | |
| L1-MaxBackbone=DeepSeek-R1-Distill-Qwen-7B2026.03 | 30.33 | 3,866 | |
| Latent-SFTBackbone=DeepSeek-R1-Distill-Qwen-7B2026.03 | 20.33 | 1,537 | |
| CODIBackbone=DeepSeek-R1-Distill-Qwen-7B2026.03 | 10.67 | 1,237 | |
| COCONUTBackbone=DeepSeek-R1-Distill-Qwen-7B2026.03 | 8.67 | 963 |