Question Answering on ConFiQA MR
89.6F1 ScoreProbeRAG
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ProbeRAGArchitecture=Qwen2.5-7B-Instruct2025.10 | 89.6 | 86.2 | |
| ProbeRAGArchitecture=LLaMA-2-7B-Chat-HF2025.10 | 80.2 | 77 | |
| CANOEArchitecture=LLaMA-2-7B-Chat-HF2025.10 | 75.2 | 72.6 | |
| CANOEArchitecture=Qwen2.5-7B-Instruct2025.10 | 71.7 | 67.8 | |
| Context-DPOArchitecture=Qwen2.5-7B-Instruct2025.10 | 71.1 | 58.8 | |
| Context-DPOArchitecture=LLaMA-2-7B-Chat-HF2025.10 | 58.5 | 32.7 |