Question Answering on Book Corpus Multi-hop Private (176 questions)
89.8AccuracyNWM Graph Retrieval
Evaluation Results
| Method | Links | |
|---|---|---|
| NWM Graph RetrievalAnswerer model=Opus 4.8, Extractor model=Sonnet 4.52026.07 | 89.8 | |
| GraphitiAnswerer model=Opus 4.8, Extractor model=Sonnet 4.52026.07 | 57.4 | |
| NWM State MemoryAnswerer model=Opus 4.8, Extractor model=Sonnet 4.52026.07 | 35.8 | |
| GraphRAGAnswerer model=Opus 4.82026.07 | 18.8 | |
| RAG (BGE+BM25+RRF)Answerer model=Opus 4.82026.07 | 17.6 | |
| Simple SearchAnswerer model=Opus 4.82026.07 | 8.5 | |
| No MemoryAnswerer model=Opus 4.82026.07 | 0 |