Question Answering on NQ (sampled)
69.5AccuracyVanilla RAG
Evaluation Results
| Method | Links | |
|---|---|---|
| Vanilla RAGBackbone=GLM-4.5, Sample Size=2002026.01 | 69.5 | |
| PAGERBackbone=GLM-4.5, Sample Size=2002026.01 | 62 | |
| Vanilla LLMBackbone=GLM-4.5, Sample Size=2002026.01 | 58 | |
| DeepNoteBackbone=GLM-4.5, Sample Size=2002026.01 | 52 |