Multi-step Retrieval on HotpotQA (val)
90.2R@2KeyExtract + CoopRAG (GPT-4o-mini)
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| KeyExtract + CoopRAG (GPT-4o-mini)Framework=KeyExtract, Retriever=CoopRAG, Model=GPT-4o-mini2025.12 | 90.2 | 97.2 | |
| KeyExtract + CoopRAG (Gemma2-9B)Framework=KeyExtract, Retriever=CoopRAG, Model=Gemma2-9B2025.12 | 88.6 | 93.9 | |
| IRCoT + CoopRAG (GPT-4o-mini)Framework=IRCoT, Retriever=CoopRAG, Model=GPT-4o-mini2025.12 | 87.3 | 90.9 | |
| IRCoT + ColBERTv2Framework=IRCoT, Retriever=ColBERTv22025.12 | 67.9 | 82 | |
| IRCoT + HippoRAG (ColBERTv2)Framework=IRCoT, Retriever=HippoRAG, Backbone=ColBERTv22025.12 | 67 | 83 | |
| IRCoT + ContrieverFramework=IRCoT, Retriever=Contriever2025.12 | 65.9 | 81.6 | |
| IRCoT + HippoRAG (Contriever)Framework=IRCoT, Retriever=HippoRAG, Backbone=Contriever2025.12 | 65.8 | 82.3 | |
| IRCoT + BM25 (Default)Framework=IRCoT, Retriever=BM252025.12 | 65.6 | 79 |