RAG Robustness Evaluation on SIG Trivial
90.5Style RobustnessMistral-7B-Instruct
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Mistral-7B-InstructEvaluation paradigm=LLM-as-Judge2025.03 | 90.5 | 91.5 | 92 | 93.8 | 96 | |
| Mistral-7B-InstructEvaluation paradigm=string-based2025.03 | 88 | 94 | 94.5 | 94 | 99 | |
| Llama-3.1-8B-Inst.Evaluation paradigm=string-based2025.03 | 87.5 | 93.5 | 93 | 90.8 | 97 | |
| Llama-3.1-8B-Inst.Evaluation paradigm=LLM-as-Judge2025.03 | 85 | 92 | 91 | 90.8 | 93.3 |