Question Answering on HotpotQA (F1, Performance Gain)
53.2F1 ScorePatchRAG
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| PatchRAGStep=t, Generator=Llama-3 8B, Retriever=bge-m32026.04 | 53.2 | 8.4 | |
| RAFTStep=t, Generator=Llama-3 8B, Retriever=bge-m32026.04 | 49.4 | 4.8 | |
| Golden-onlyStep=t, Generator=Llama-3 8B, Retriever=bge-m32026.04 | 48.1 | 4.4 | |
| Auto-RAGStep=-, Generator=Llama-3 8B, Retriever=bge-m32026.04 | 44.9 | — | |
| PatchRAGStep=< t, Generator=Llama-3 8B, Retriever=bge-m32026.04 | 44.8 | — | |
| ChatQA-1.5Step=-, Generator=Llama-3 8B, Retriever=bge-m32026.04 | 44.6 | — | |
| RAFTStep=< t, Generator=Llama-3 8B, Retriever=bge-m32026.04 | 44.6 | — | |
| Golden-onlyStep=< t, Generator=Llama-3 8B, Retriever=bge-m32026.04 | 43.7 | — | |
| Self-RAGStep=-, Generator=Llama-3 8B, Retriever=bge-m32026.04 | 29.6 | — | |
| Standard RAGStep=-, Generator=Llama-3 8B, Retriever=bge-m32026.04 | 28.5 | — |