Long-horizon Question Answering on LoCoMo Single-Hop 1.0
16.77EMMemWeaver
Evaluation results
| Method | Links | ||
|---|---|---|---|
| MemWeaverBackbone=GPT-4o-mini2026.01 | 16.77 | 33.46 | |
| MemWeaverBackbone=Qwen2.5-1.5B2026.01 | 12.96 | 25.66 | |
| A-MemBackbone=GPT-4o-mini2026.01 | 9.27 | 32.79 | |
| LoCoMoBackbone=GPT-4o-mini2026.01 | 6.54 | 39.75 | |
| MemWeaverBackbone=Llama3.2-3B2026.01 | 5.35 | 16.26 | |
| A-MemBackbone=Llama3.2-3B2026.01 | 3.92 | 16.68 | |
| A-MemBackbone=Qwen2.5-1.5B2026.01 | 3.8 | 20.55 | |
| MemWeaverBackbone=Llama3.2-1B2026.01 | 1.78 | 13.14 | |
| LoCoMoBackbone=Llama3.2-1B2026.01 | 1.07 | 10.77 | |
| ReadAgentBackbone=Llama3.2-1B2026.01 | 1.07 | 5.36 | |
| LoCoMoBackbone=Qwen2.5-1.5B2026.01 | 1.07 | 13.48 | |
| LoCoMoBackbone=Llama3.2-3B2026.01 | 0.71 | 9.47 | |
| ReadAgentBackbone=Qwen2.5-1.5B2026.01 | 0.71 | 5.52 | |
| MemoryBankBackbone=GPT-4o-mini2026.01 | 0 | 13.04 | |
| ReadAgentBackbone=GPT-4o-mini2026.01 | 0 | 8.01 | |
| MemoryBankBackbone=Llama3.2-3B2026.01 | 0 | 11.15 | |
| ReadAgentBackbone=Llama3.2-3B2026.01 | 0 | 2.46 | |
| MemoryBankBackbone=Llama3.2-1B2026.01 | 0 | 9.27 | |
| A-MemBackbone=Llama3.2-1B2026.01 | 0 | 12.33 | |
| MemoryBankBackbone=Qwen2.5-1.5B2026.01 | 0 | 14.08 |