Retrieval on HotpotQA (test)
96.8Recall@5CoopRAG
Evaluation results
| Method | Links | ||
|---|---|---|---|
| CoopRAGBackbone LLM=GPT-4o-mini2025.12 | 96.8 | 88.8 | |
| CoopRAGBackbone LLM=Gemma2-27B2025.12 | 96.6 | 88.3 | |
| CoopRAGBackbone LLM=Llama3.3-70B2025.12 | 96.6 | 86.9 | |
| HippoRAG2Backbone LLM=Llama3.3-70B2025.12 | 96.3 | 83.5 | |
| HippoRAG 2Knowledge Representation Type=Structured Knowledge Representation Augmented2026.06 | 96.3 | — | |
| HippoRAG 2 (reproduced)Knowledge Representation Type=Structured Knowledge Representation Augmented, Reproduction Status=reproduced2026.06 | 96.3 | — | |
| SemFlowRAGKnowledge Representation Type=Structured Knowledge Representation Augmented2026.06 | 96.1 | — | |
| HopRAGBackbone LLM=GPT-4o-mini2025.12 | 96 | 81.1 | |
| HippoRAG2Backbone LLM=GPT-4o-mini2025.12 | 95.7 | 80.5 | |
| CoopRAGBackbone LLM=Gemma2-9B2025.12 | 95.6 | 87.9 | |
| SiReRAGBackbone LLM=GPT-4o-mini2025.12 | 94.8 | 80 | |
| NV-Embed-v2Parameters=7B2025.12 | 94.5 | 84.1 | |
| NV-Embed-v2 (7B)Knowledge Representation Type=Embedding-based Knowledge Representation Augmented, Number of Parameters=7B2026.06 | 94.5 | — | |
| GritLM-7B2025.12 | 92.4 | 79.2 | |
| GritLM-7BKnowledge Representation Type=Embedding-based Knowledge Representation Augmented, Backbone=GritLM, Number of Parameters=7B2026.06 | 92.4 | — | |
| RAPTORBackbone LLM=GPT-4o-mini2025.12 | 90.2 | 78.6 | |
| GTE-Qwen2-7B-Instruct2025.12 | 89.1 | 75.8 | |
| GTE-Qwen2-7B-InstructKnowledge Representation Type=Embedding-based Knowledge Representation Augmented, Backbone=Qwen2-7B-Instruct, Number of Parameters=7B2026.06 | 89.1 | — | |
| RAPTORBackbone LLM=Llama3.3-70B2025.12 | 86.9 | 76.8 | |
| RAPTORKnowledge Representation Type=Structured Knowledge Representation Augmented2026.06 | 86.9 | — | |
| ColBERTv22025.12 | 79.3 | 64.7 | |
| HippoRAGBackbone LLM=GPT-4o-mini2025.12 | 78.5 | 60.1 | |
| PropositionBase Retriever=ColBERTv22025.12 | 78.1 | 63.9 | |
| HippoRAGBackbone LLM=Llama3.3-70B2025.12 | 77.3 | 60.4 | |
| HippoRAG (reproduced)Knowledge Representation Type=Structured Knowledge Representation Augmented, Reproduction Status=reproduced2026.06 | 77.3 | — | |
| Contriever2025.12 | 75.3 | 58.4 | |
| ContrieverKnowledge Representation Type=Unstructured Knowledge Representation Augmented2026.06 | 75.3 | — | |
| BM252025.12 | 74.8 | 57.3 | |
| BM25Knowledge Representation Type=Unstructured Knowledge Representation Augmented2026.06 | 74.8 | — | |
| GTRBackbone=T5-base2025.12 | 73.9 | 59.3 | |
| GTRKnowledge Representation Type=Unstructured Knowledge Representation Augmented2026.06 | 73.9 | — |