Question Answering on BioQA
21.41ROUGE-1External-Only (GPT-4o mini)
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| External-Only (GPT-4o mini)Untrusted External Model=GPT-4o mini2026.04 | 21.41 | 5.38 | 16.34 | 23.83 | 84.1 | |
| GTKATrusted Local Model=Llama-3.1-8B, Untrusted External Model=GPT-4o mini2026.04 | 18.79 | 4.52 | 14.38 | 21.94 | 83.72 | |
| Soft-POTrusted Local Model=Llama-3.1-8B, Untrusted External Model=GPT-4o mini2026.04 | 17.25 | 4.05 | 13.62 | 20.78 | 83.34 | |
| Hard-POTrusted Local Model=Llama-3.1-8B, Untrusted External Model=GPT-4o mini2026.04 | 17.04 | 3.98 | 13.47 | 20.61 | 82.94 | |
| Local-Only (Llama-3.1-8B)Trusted Local Model=Llama-3.1-8B2026.04 | 16.35 | 3.26 | 11.62 | 18.17 | 81.02 | |
| IOITrusted Local Model=Llama-3.1-8B, Untrusted External Model=GPT-4o mini2026.04 | 16.28 | 3.54 | 12.24 | 19.72 | 82.86 | |
| PP-TSTrusted Local Model=Llama-3.1-8B, Untrusted External Model=GPT-4o mini2026.04 | 15.62 | 3.25 | 11.76 | 19.24 | 81.67 |