Explanation Refinement on EntailmentBank
25.33Initial ScoreFaithful-Refiner
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Faithful-RefinerBackbone=Deepseek-V3.12026.01 | 25.33 | 48 | |
| Faithful-RefinerBackbone=GPT-4o2026.01 | 23.33 | 44 | |
| LLM-TP TreeBackbone=Qwen3-max2026.01 | 22.67 | 78.67 | |
| LLM-TP TreeBackbone=Deepseek-V3.12026.01 | 22 | 71.33 | |
| Faithful-RefinerBackbone=Qwen3-max2026.01 | 22 | 52 | |
| LLM-TP TreeBackbone=GPT-4o2026.01 | 21.33 | 70 | |
| LLM-TP TreeBackbone=GPT-5 nano2026.01 | 18 | 74 | |
| Faithful-RefinerBackbone=GPT-5 nano2026.01 | 17.33 | 47.33 | |
| LLM-TP TreeBackbone=Grok-4 fast2026.01 | 16.67 | 68.67 | |
| Explanation-RefinerBackbone=Qwen3-max2026.01 | 16.67 | 28 | |
| Explanation-RefinerBackbone=GPT-4o2026.01 | 15.33 | 26.67 | |
| Explanation-RefinerBackbone=Deepseek-V3.12026.01 | 15.33 | 26.67 | |
| Faithful-RefinerBackbone=Grok-4 fast2026.01 | 14 | 46 | |
| Explanation-RefinerBackbone=GPT-5 nano2026.01 | 12 | 23.33 | |
| Explanation-RefinerBackbone=Grok-4 fast2026.01 | 8.67 | 13.33 |