Mathematical Reasoning on AIME 2024 (Acceptance Length and Decoding Speedup)
5.41Average Acceptance Length (τ)EDA (Ours)
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| EDA (Ours)Temperature=0, Target Model=Qwen2.5-Math-7B2026.03 | 5.41 | 3.2 | |
| Full-FTTemperature=0, Target Model=Qwen2.5-Math-7B2026.03 | 5.05 | 3.14 | |
| EDA (Base)Temperature=0, Target Model=Qwen2.5-Math-7B2026.03 | 5.02 | 3.12 | |
| LoRATemperature=0, Target Model=Qwen2.5-Math-7B2026.03 | 4.9 | 2.93 | |
| EDA (Ours)Temperature=1, Target Model=Qwen2.5-Math-7B2026.03 | 3.38 | 2.05 | |
| EDA (Base)Temperature=1, Target Model=Qwen2.5-Math-7B2026.03 | 2.92 | 1.86 | |
| Full-FTTemperature=1, Target Model=Qwen2.5-Math-7B2026.03 | 2.89 | 1.84 | |
| LoRATemperature=1, Target Model=Qwen2.5-Math-7B2026.03 | 2.8 | 1.78 | |
| Training-FreeTemperature=0, Target Model=Qwen2.5-Math-7B2026.03 | 1.18 | 0.76 | |
| Training-FreeTemperature=1, Target Model=Qwen2.5-Math-7B2026.03 | 1.15 | 0.72 |