Long-horizon Question Answering on LoCoMo Open-Domain 1.0
7.29EMMemWeaver
Evaluation results
| Method | Links | ||
|---|---|---|---|
| MemWeaverBackbone=Qwen2.5-1.5B2026.01 | 7.29 | 10.99 | |
| MemWeaverBackbone=GPT-4o-mini2026.01 | 6.25 | 9.7 | |
| A-MemBackbone=GPT-4o-mini2026.01 | 4.17 | 8.65 | |
| MemWeaverBackbone=Llama3.2-1B2026.01 | 4.17 | 7.21 | |
| LoCoMoBackbone=Llama3.2-1B2026.01 | 3.12 | 7.25 | |
| LoCoMoBackbone=GPT-4o-mini2026.01 | 2.08 | 7.67 | |
| MemoryBankBackbone=GPT-4o-mini2026.01 | 2.08 | 7.26 | |
| MemWeaverBackbone=Llama3.2-3B2026.01 | 2.08 | 3.98 | |
| LoCoMoBackbone=Llama3.2-3B2026.01 | 1.04 | 4.57 | |
| ReadAgentBackbone=Llama3.2-3B2026.01 | 1.04 | 3.39 | |
| A-MemBackbone=Llama3.2-3B2026.01 | 1.04 | 3.51 | |
| MemoryBankBackbone=Llama3.2-1B2026.01 | 1.04 | 6.66 | |
| ReadAgentBackbone=Llama3.2-1B2026.01 | 1.04 | 7.13 | |
| LoCoMoBackbone=Qwen2.5-1.5B2026.01 | 1.04 | 9.44 | |
| ReadAgentBackbone=Qwen2.5-1.5B2026.01 | 1.04 | 8.97 | |
| A-MemBackbone=Qwen2.5-1.5B2026.01 | 1.04 | 7.51 | |
| ReadAgentBackbone=GPT-4o-mini2026.01 | 0 | 3.69 | |
| MemoryBankBackbone=Llama3.2-3B2026.01 | 0 | 6.39 | |
| A-MemBackbone=Llama3.2-1B2026.01 | 0 | 7.26 | |
| MemoryBankBackbone=Qwen2.5-1.5B2026.01 | 0 | 7.06 |