Agentic Question Answering on AMABench
32.2A-ALF ScoreDCI-Lite
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| DCI-LiteLLM judge=true2026.06 | 32.2 | 37.2 | 25.5 | |
| DCI-Lite+SumLLM judge=true2026.06 | 30.8 | 38.1 | 27.2 | |
| Long contextLLM judge=true2026.06 | 26.9 | 43.9 | 35.4 | |
| Mem-TLLM judge=true2026.06 | 26 | 32.1 | 10 | |
| PlugMemLLM judge=true2026.06 | 25 | 16.7 | 18.1 | |
| HippoRAGLLM judge=true2026.06 | 10.9 | 28.6 | 23 | |
| AMA-AgentLLM judge=true2026.06 | 10 | 31 | 22.9 | |
| LightMemLLM judge=true2026.06 | 9.4 | 25.3 | 15.6 | |
| SimpleMemLLM judge=true2026.06 | 8.3 | 18.2 | 13.8 | |
| MemRLLLM judge=true2026.06 | 5.4 | 29.9 | 18.8 |