Dialogue Evaluation on Arena-Hard
3.29Accepted Length (τ)DSpark
Evaluation Results
| Method | Links | |
|---|---|---|
| DSparktarget_model=Qwen3-4B2026.07 | 3.29 | |
| DSparktarget_model=Qwen3-8B2026.07 | 3.21 | |
| DSparktarget_model=Qwen3-14B2026.07 | 3.13 | |
| DSparktarget_model=Gemma4-12B2026.07 | 2.92 | |
| DFlashtarget_model=Qwen3-4B2026.07 | 2.83 | |
| DFlashtarget_model=Qwen3-8B2026.07 | 2.81 | |
| DFlashtarget_model=Qwen3-14B2026.07 | 2.72 | |
| Eagle3target_model=Gemma4-12B2026.07 | 2.72 | |
| DFlashtarget_model=Gemma4-12B2026.07 | 2.59 | |
| Eagle3target_model=Qwen3-4B2026.07 | 2.55 | |
| Eagle3target_model=Qwen3-8B2026.07 | 2.54 | |
| Eagle3target_model=Qwen3-14B2026.07 | 2.48 |