Multi-hop Question Answering on HotpotQA 1600 documents
55.76F1E-mem
Evaluation Results
| Method | Links | |
|---|---|---|
| E-memBackbone=GPT4o-mini2026.01 | 55.76 | |
| E-memBackbone=Qwen2.5-14B2026.01 | 54.87 | |
| RAGBackbone=GPT4o-mini2026.01 | 54.01 | |
| Long-ContextBackbone=GPT4o-mini2026.01 | 53.92 | |
| GAMBackbone=GPT4o-mini2026.01 | 53.71 | |
| RAGBackbone=Qwen2.5-14B2026.01 | 48.36 | |
| GAMBackbone=Qwen2.5-14B2026.01 | 44.32 | |
| Long-ContextBackbone=Qwen2.5-14B2026.01 | 43.17 | |
| A-memBackbone=GPT4o-mini2026.01 | 31.37 | |
| LIGHTMEMBackbone=GPT4o-mini2026.01 | 30.02 | |
| LIGHTMEMBackbone=Qwen2.5-14B2026.01 | 28.25 | |
| Mem0Backbone=GPT4o-mini2026.01 | 27.41 | |
| Mem0Backbone=Qwen2.5-14B2026.01 | 26.55 | |
| MEMORYOSBackbone=GPT4o-mini2026.01 | 24.16 | |
| MEMORYOSBackbone=Qwen2.5-14B2026.01 | 23.13 | |
| A-memBackbone=Qwen2.5-14B2026.01 | 22.92 |