Question Answering on LongMemEval MAB
72LLM-judge AccuracySEGTREEMEM
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| SEGTREEMEMVariant=Best, Backbone=qwen3.5-flash2026.06 | 72 | 14.4 | |
| SEGTREEMEMVariant=No-Propagation, Backbone=qwen3.5-flash2026.06 | 66.9 | 12.8 | |
| HippoRAGVariant=Default, Backbone=qwen3.5-flash2026.06 | 61.2 | 14.2 | |
| A-MEMVariant=Default, Backbone=qwen3.5-flash2026.06 | 59.9 | 13.9 | |
| RAPTORVariant=Default, Backbone=qwen3.5-flash2026.06 | 59.3 | 14 | |
| DenseVariant=Default, Backbone=qwen3.5-flash2026.06 | 59 | 14 | |
| MemTreeVariant=Default, Backbone=qwen3.5-flash2026.06 | 54.3 | 13.7 | |
| BM25Variant=Default, Backbone=qwen3.5-flash2026.06 | 54.2 | 13.7 | |
| Mem0Variant=Default, Backbone=qwen3.5-flash2026.06 | 51.7 | 14.7 |