Mathematical Reasoning on MATH500 (Speedup and Tau metrics)
6.76SpeedupDFLARE
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| DFLAREModel=Q3-4B, Temperature=02026.06 | 6.76 | 8.97 | |
| DFLAREModel=Q3-8B, Temperature=02026.06 | 6.72 | 8.95 | |
| DFlashModel=Q3-8B, Temperature=02026.06 | 6.29 | 7.88 | |
| DFlashModel=Q3-4B, Temperature=02026.06 | 6.2 | 7.8 | |
| DFLAREModel=Q3-4B, Temperature=12026.06 | 5.33 | 7.34 | |
| DFLAREModel=Q3-8B, Temperature=12026.06 | 5.25 | 7.18 | |
| DFlashModel=Q3-4B, Temperature=12026.06 | 5.22 | 6.75 | |
| DFlashModel=Q3-8B, Temperature=12026.06 | 4.99 | 6.39 | |
| DFLAREModel=GPT-20B, Temperature=02026.06 | 4.2 | 5.19 | |
| DFlashModel=GPT-20B, Temperature=02026.06 | 3.96 | 4.86 | |
| DFLAREModel=GPT-20B, Temperature=12026.06 | 3.08 | 3.91 | |
| DFlashModel=GPT-20B, Temperature=12026.06 | 2.99 | 3.76 | |
| EAGLE3Model=Q3-4B, Temperature=02026.06 | 2.57 | 3.7 | |
| EAGLE3Model=Q3-4B, Temperature=12026.06 | 2.43 | 3.57 | |
| EAGLE3Model=Q3-8B, Temperature=02026.06 | 2.42 | 3.55 | |
| EAGLE3Model=Q3-8B, Temperature=12026.06 | 2.2 | 3.4 | |
| PRRModel=GLM-4, DSA mechanism=InfLLM-v22026.06 | 1.59 | — | |
| PRRModel=GLM-Z1, DSA mechanism=InfLLM-v22026.06 | 1.55 | — | |
| PRRModel=Qwen3-14B, DSA mechanism=InfLLM-v22026.06 | 1.53 | — | |
| PRRModel=GLM-Z1, DSA mechanism=Quest2026.06 | 1.45 | — | |
| PRRModel=Qwen3-14B, DSA mechanism=Quest2026.06 | 1.45 | — | |
| PRRModel=GLM-4, DSA mechanism=Quest2026.06 | 1.42 | — | |
| PRRModel=DeepSeek-R1, DSA mechanism=Quest2026.06 | 1.39 | — | |
| PRRModel=Qwen3-32B, DSA mechanism=InfLLM-v22026.06 | 1.35 | — | |
| PRRModel=Llama3, DSA mechanism=Quest2026.06 | 1.34 | — | |
| PRRModel=Llama3, DSA mechanism=InfLLM-v22026.06 | 1.3 | — | |
| PRRModel=DeepSeek-R1, DSA mechanism=InfLLM-v22026.06 | 1.28 | — | |
| PRRModel=Qwen3-32B, DSA mechanism=Quest2026.06 | 1.26 | — |