Question Answering on Bamboogle 500 samples (val)
34.6EMASSISTRAG
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| ASSISTRAGMain LLM=ChatGPT3.5, Retrieval Setting=Baselines with retrieval2024.11 | 34.6 | 41.4 | 41.1 | |
| ASSISTRAGMain LLM=LLaMA2-chat 7B, Retrieval Setting=Baselines with retrieval2024.11 | 33 | 39.6 | 38.7 | |
| ASSISTRAGMain LLM=ChatGLM3 6B, Retrieval Setting=Baselines with retrieval2024.11 | 32.8 | 39.8 | 39 | |
| IRCoTMain LLM=ChatGPT3.5, Retrieval Setting=Baselines with retrieval2024.11 | 30.2 | 38.8 | 37.9 | |
| SELF-RAGMain LLM=SELF-RAG 7B, Retrieval Setting=Baselines with retrieval2024.11 | 29.8 | 35.5 | 37.8 | |
| ReActMain LLM=ChatGPT3.5, Retrieval Setting=Baselines with retrieval2024.11 | 28.8 | 37.7 | 38.2 | |
| LLMLinguaMain LLM=ChatGPT3.5, Retrieval Setting=Baselines with retrieval2024.11 | 25.2 | 31.3 | 30.8 | |
| Self-AskMain LLM=ChatGPT3.5, Retrieval Setting=Baselines with retrieval2024.11 | 23.2 | 32.8 | 30.8 | |
| Naive RAGMain LLM=ChatGPT3.5, Retrieval Setting=Baselines with retrieval2024.11 | 18.4 | 24.4 | 24.7 | |
| Naive RAGMain LLM=ChatGLM3 6B, Retrieval Setting=Baselines with retrieval2024.11 | 15.8 | 21.1 | 20.8 | |
| Naive RAGMain LLM=LLaMA2-chat 7B, Retrieval Setting=Baselines with retrieval2024.11 | 15.2 | 20.4 | 20.3 | |
| CloseBookMain LLM=ChatGPT3.5, Retrieval Setting=Baselines without retrieval2024.11 | 14.4 | 22 | 22.3 | |
| CloseBookMain LLM=ChatGLM3 6B, Retrieval Setting=Baselines without retrieval2024.11 | 12.6 | 17.6 | 16.9 | |
| CloseBookMain LLM=LLaMA2-chat 7B, Retrieval Setting=Baselines without retrieval2024.11 | 10.4 | 16.3 | 16.7 |