Long-form Question Answering on GroundBench (test)
87.5Faithfulness (Full)RHIO-13B
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| RHIO-13BParameters=13B2025.01 | 87.5 | 7.6 | 4.9 | 3.8 | |
| GPT-4oModel Type=Proprietary2025.01 | 86.5 | 8.1 | 5.4 | 4.2 | |
| Llama-3.1-70B-InstructBackbone=Llama-3.1, Parameters=70B2025.01 | 82.6 | 10.2 | 7.2 | 3.7 | |
| SFT-13BParameters=13B, Type=Supervised Fine-Tuned Baseline2025.01 | 76.8 | 13.8 | 9.4 | 3.2 |