Open-ended QA on Tele-Eval JCC domain
77.63ScoreBase RAG
Evaluation Results
| Method | Links | |
|---|---|---|
| Base RAGretrieval=top-16, judge=GPT-5.22026.06 | 77.63 | |
| ARMORretrieval=top-16, judge=GPT-5.22026.06 | 77.19 | |
| InfoNCE QE FTretrieval=top-16, judge=GPT-5.22026.06 | 74.25 | |
| Mix QE FTretrieval=top-16, judge=GPT-5.22026.06 | 73.6 | |
| RAG QE FTretrieval=top-16, judge=GPT-5.22026.06 | 72.3 | |
| Base Genjudge=GPT-5.22026.06 | 29.8 |