Retrieval-Augmented Generation on PubMedQA
77.9AccuracyMetaGen BlendedRAG
Evaluation Results
| Method | Links | |
|---|---|---|
| MetaGen BlendedRAGDescription=Leverages metadata augmentation for retrieval, Fine-tuning status=Non-Fine-Tuned2025.05 | 77.9 | |
| GPT-3.5 + RAGDescription=RAG pipeline with GPT-3.5 model, Fine-tuning status=Non-Fine-Tuned2025.05 | 71.6 | |
| LLaMA2-7B + RAGDescription=RAG pipeline with LLaMA2-7B model, Fine-tuning status=Non-Fine-Tuned2025.05 | 58.8 | |
| TTARAGBackbone=Llama-2-7b-chat2026.01 | 54 | |
| Naive-RAGBackbone=Llama-2-7b-chat2026.01 | 47.6 | |
| RAATBackbone=Llama-2-7b-chat2026.01 | 46.6 | |
| Self-ragBackbone=Llama-2-7b-chat2026.01 | 43.4 | |
| Ret-RobustBackbone=Llama-2-7b-chat2026.01 | 28.4 |