Multi-hop Question Answering on HotpotQA 800 documents
55.46F1 ScoreE-mem
Evaluation Results
| Method | Links | |
|---|---|---|
| E-memBackbone=GPT4o-mini2026.01 | 55.46 | |
| GAMBackbone=GPT4o-mini2026.01 | 52.86 | |
| RAGBackbone=GPT4o-mini2026.01 | 51.84 | |
| Long-ContextBackbone=GPT4o-mini2026.01 | 49.71 | |
| E-memBackbone=Qwen2.5-14B2026.01 | 47.91 | |
| Long-ContextBackbone=Qwen2.5-14B2026.01 | 46.82 | |
| RAGBackbone=Qwen2.5-14B2026.01 | 46.72 | |
| GAMBackbone=Qwen2.5-14B2026.01 | 41.1 | |
| LIGHTMEMBackbone=GPT4o-mini2026.01 | 35.28 | |
| Mem0Backbone=Qwen2.5-14B2026.01 | 32.44 | |
| Mem0Backbone=GPT4o-mini2026.01 | 31.74 | |
| MEMORYOSBackbone=Qwen2.5-14B2026.01 | 30.25 | |
| A-memBackbone=GPT4o-mini2026.01 | 30.2 | |
| LIGHTMEMBackbone=Qwen2.5-14B2026.01 | 27.72 | |
| A-memBackbone=Qwen2.5-14B2026.01 | 25.65 | |
| MEMORYOSBackbone=GPT4o-mini2026.01 | 23.1 |