Personalized Citation Identification on LaMP-1
91.9AccuracyPersonaAgent
Evaluation results
| Method | Links | ||
|---|---|---|---|
| PersonaAgentLLM Backbone=Claude-3.5 Sonnet2025.06 | 91.9 | 91.8 | |
| MemBankLLM Backbone=Claude-3.5 Sonnet, Category=General Agent2025.06 | 86.2 | 86.1 | |
| PAGLLM Backbone=Claude-3.5 Sonnet, Category=Personalized LLM2025.06 | 83.7 | 83.7 | |
| ReActLLM Backbone=Claude-3.5 Sonnet, Category=General Agent2025.06 | 83.7 | 85.3 | |
| ICLLLM Backbone=Claude-3.5 Sonnet, Category=Non-Personalized2025.06 | 78 | 76.6 | |
| PromptLLM Backbone=Claude-3.5 Sonnet, Category=Non-Personalized2025.06 | 77.2 | 77.1 | |
| TAP-PERtype=Model-based, variant=+Pu§2026.06 | 73.2 | 72.8 | |
| P2Ptype=Model-based, variant=single2026.06 | 72.4 | 72 | |
| RAGLLM Backbone=Claude-3.5 Sonnet, Category=Personalized LLM2025.06 | 71.5 | 71.4 | |
| OPPUtype=Model-based, variant=+pag2026.06 | 71.5 | 71.1 | |
| TAP-PERtype=Model-based, variant=+Pq‡2026.06 | 69.9 | 69.6 | |
| PER-PCStype=Model-based, variant=+pag2026.06 | 69.1 | 68.5 | |
| PAGtype=Prompt-based, k=02026.06 | 68.3 | 68.3 | |
| PAGtype=Prompt-based, k=12026.06 | 67.5 | 67.5 | |
| OPPUtype=Model-based, variant=+rag2026.06 | 67.5 | 67.4 | |
| RAGtype=Prompt-based, k=12026.06 | 65.9 | 65.9 | |
| RAGtype=Prompt-based, k=42026.06 | 65.9 | 66.4 | |
| OPPUtype=Model-based, variant=base2026.06 | 65.4 | 65.6 | |
| RAGtype=Prompt-based, k=22026.06 | 64.2 | 64.5 | |
| TAP-PERtype=Model-based, variant=base†2026.06 | 64.2 | 64.2 | |
| PER-PCStype=Model-based, variant=base2026.06 | 62.6 | 61.9 | |
| PER-PCStype=Model-based, variant=+rag2026.06 | 62.6 | 62 |