Long-form QA on BioASQ (test)
34.3ROUGE-1Fine-Tuned GPT-4o + MedBioRAG
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Fine-Tuned GPT-4o + MedBioRAGFine-tuned=true, MedBioRAG=true2025.12 | 34.3 | 18.81 | 27.74 | 6.12 | 35.43 | -15.44 | |
| Fine-Tuned GPT-4oFine-tuned=true, MedBioRAG=false2025.12 | 32.69 | 16.84 | 25.11 | 6.52 | 32.97 | -2.41 | |
| GPT-4o + MedBioRAGFine-tuned=false, MedBioRAG=true2025.12 | 22.29 | 8.21 | 15.64 | 2.27 | 11.6 | -12.5 | |
| GPT-4oFine-tuned=false, MedBioRAG=false2025.12 | 13.97 | 5.51 | 10.08 | 1.27 | 0.22 | -24.84 |