Knowledge Acquisition on LawQA
44.71ROUGE-1External-Only
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| External-OnlyUntrusted External Model=GPT-4o mini2026.04 | 44.71 | 20.72 | 30.9 | 46.28 | 85.98 | |
| GTKATrusted Local Model=Llama-3.1-8B, Untrusted External Model=GPT-4o mini2026.04 | 42.67 | 19.58 | 29.72 | 45.36 | 85.39 | |
| Soft-POTrusted Local Model=Llama-3.1-8B, Untrusted External Model=GPT-4o mini2026.04 | 40.86 | 18.12 | 28.24 | 43.68 | 84.76 | |
| Local-OnlyTrusted Local Model=Llama-3.1-8B2026.04 | 40.49 | 18.64 | 28.78 | 43.97 | 81.81 | |
| Hard-POTrusted Local Model=Llama-3.1-8B, Untrusted External Model=GPT-4o mini2026.04 | 40.35 | 17.86 | 27.92 | 43.45 | 83.72 | |
| IOITrusted Local Model=Llama-3.1-8B, Untrusted External Model=GPT-4o mini2026.04 | 39.72 | 17.56 | 27.48 | 43.12 | 83.08 | |
| PP-TSTrusted Local Model=Llama-3.1-8B, Untrusted External Model=GPT-4o mini2026.04 | 38.56 | 16.82 | 26.54 | 42.28 | 82.13 |