Multi-hop Question Answering on LV-Eval (test)
12.9F1 ScoreHippoRAG2
Evaluation results
| Method | Links | |
|---|---|---|
| HippoRAG2Backbone=Llama-3.3-70B-Instruct2026.02 | 12.9 | |
| HELPBackbone=Llama-3.3-70B-Instruct2026.02 | 12.5 | |
| GraphRAGBackbone=Llama-3.3-70B-Instruct2026.02 | 11.2 | |
| LinearRAGBackbone=Llama-3.3-70B-Instruct, Local Reproduction=true2026.02 | 10.3 | |
| GritLM-7BBackbone=Llama-3.3-70B-Instruct2026.02 | 9.8 | |
| NV-Embed-v2Backbone=Llama-3.3-70B-Instruct2026.02 | 9.8 | |
| HippoRAGBackbone=Llama-3.3-70B-Instruct2026.02 | 8.4 | |
| ContrieverBackbone=Llama-3.3-70B-Instruct2026.02 | 8.1 | |
| GTRBackbone=Llama-3.3-70B-Instruct2026.02 | 7.1 | |
| GTE-Qwen2-7BBackbone=Llama-3.3-70B-Instruct2026.02 | 7.1 | |
| NoneBackbone=Llama-3.3-70B-Instruct2026.02 | 6 | |
| BM25Backbone=Llama-3.3-70B-Instruct2026.02 | 5.9 | |
| RAPTORBackbone=Llama-3.3-70B-Instruct2026.02 | 5 | |
| LightRAGBackbone=Llama-3.3-70B-Instruct2026.02 | 1 |