Long-context Memory Retrieval and Reasoning on ZH4O 128K
51.68F1 ScoreQwen3-30B
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen3-30BBackbone LLM=Qwen3-30B-A3B-Ins2026.03 | 51.68 | |
| MemSifterBackbone LLM=Qwen3-30B-A3B-Ins2026.03 | 50.91 | |
| GemmaEmbBackbone LLM=Qwen3-30B-A3B-Ins2026.03 | 50.25 | |
| RearankBackbone LLM=Qwen3-30B-A3B-Ins2026.03 | 49.02 | |
| Mem0Backbone LLM=Qwen3-30B-A3B-Ins2026.03 | 48.4 | |
| HippoRAGBackbone LLM=Qwen3-30B-A3B-Ins2026.03 | 48.36 | |
| BGE-M3Backbone LLM=Qwen3-30B-A3B-Ins2026.03 | 48.31 | |
| MemSifterBackbone LLM=DeepSeek V3.22026.03 | 48.13 | |
| Mem0Backbone LLM=DeepSeek V3.22026.03 | 47.19 | |
| ReasonRankBackbone LLM=Qwen3-30B-A3B-Ins2026.03 | 47.16 | |
| RearankBackbone LLM=DeepSeek V3.22026.03 | 47.14 | |
| DS-V3.2Backbone LLM=DeepSeek V3.22026.03 | 46.73 | |
| HippoRAGBackbone LLM=DeepSeek V3.22026.03 | 46.56 | |
| A-MEMBackbone LLM=DeepSeek V3.22026.03 | 45.69 | |
| ReasonRankBackbone LLM=DeepSeek V3.22026.03 | 45.55 | |
| NemoriBackbone LLM=Qwen3-30B-A3B-Ins2026.03 | 45.53 | |
| BGE-M3Backbone LLM=DeepSeek V3.22026.03 | 45.38 | |
| NemoriBackbone LLM=DeepSeek V3.22026.03 | 45.07 | |
| A-MEMBackbone LLM=Qwen3-30B-A3B-Ins2026.03 | 44.52 | |
| GemmaEmbBackbone LLM=DeepSeek V3.22026.03 | 41.84 |