Long-context Memory Retrieval and Reasoning on HotpotQA 128K
24.95F1 ScoreMemSifter
Evaluation Results
| Method | Links | |
|---|---|---|
| MemSifterBackbone LLM=DeepSeek V3.22026.03 | 24.95 | |
| HippoRAGBackbone LLM=DeepSeek V3.22026.03 | 23.2 | |
| A-MEMBackbone LLM=DeepSeek V3.22026.03 | 23.12 | |
| NemoriBackbone LLM=DeepSeek V3.22026.03 | 22.76 | |
| MemSifterBackbone LLM=Qwen3-30B-A3B-Ins2026.03 | 22.71 | |
| ReasonRankBackbone LLM=Qwen3-30B-A3B-Ins2026.03 | 22.37 | |
| ReasonRankBackbone LLM=DeepSeek V3.22026.03 | 22.25 | |
| Qwen3-30BBackbone LLM=Qwen3-30B-A3B-Ins2026.03 | 21.59 | |
| HippoRAGBackbone LLM=Qwen3-30B-A3B-Ins2026.03 | 21.41 | |
| GemmaEmbBackbone LLM=DeepSeek V3.22026.03 | 21.01 | |
| NemoriBackbone LLM=Qwen3-30B-A3B-Ins2026.03 | 20.91 | |
| BGE-M3Backbone LLM=DeepSeek V3.22026.03 | 20.79 | |
| Mem0Backbone LLM=Qwen3-30B-A3B-Ins2026.03 | 20.57 | |
| BGE-M3Backbone LLM=Qwen3-30B-A3B-Ins2026.03 | 20.49 | |
| Mem0Backbone LLM=DeepSeek V3.22026.03 | 20.35 | |
| RearankBackbone LLM=Qwen3-30B-A3B-Ins2026.03 | 19.95 | |
| DS-V3.2Backbone LLM=DeepSeek V3.22026.03 | 19.88 | |
| A-MEMBackbone LLM=Qwen3-30B-A3B-Ins2026.03 | 19.85 | |
| GemmaEmbBackbone LLM=Qwen3-30B-A3B-Ins2026.03 | 19.84 | |
| RearankBackbone LLM=DeepSeek V3.22026.03 | 19.76 |