Personalization on LaMP-3
0.241MAEPersonaAgent
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| PersonaAgentLLM Backbone=Claude-3.5 Sonnet2025.06 | 0.241 | 0.509 | |
| RPMBackbone=GPT-4o-mini, +CoT=false, Reasoning generation=enabled2025.05 | 0.259 | 0.548 | |
| ICLLLM Backbone=Claude-3.5 Sonnet, Category=Non-Personalized2025.06 | 0.277 | 0.543 | |
| PromptLLM Backbone=Claude-3.5 Sonnet, Category=Non-Personalized2025.06 | 0.295 | 0.59 | |
| RPMBackbone=GPT-4o-mini, +CoT=false, Reasoning generation=disabled2025.05 | 0.305 | 0.599 | |
| FermiBackbone=GPT-4o-mini, +CoT=true2025.05 | 0.312 | 0.635 | |
| RAGLLM Backbone=Claude-3.5 Sonnet, Category=Personalized LLM2025.06 | 0.313 | 0.713 | |
| ReActLLM Backbone=Claude-3.5 Sonnet, Category=General Agent2025.06 | 0.313 | 0.59 | |
| ICLBackbone=GPT-4o-mini, +CoT=true2025.05 | 0.317 | 0.625 | |
| MemBankLLM Backbone=Claude-3.5 Sonnet, Category=General Agent2025.06 | 0.321 | 0.582 | |
| Zero-shotBackbone=GPT-4o-mini, +CoT=true2025.05 | 0.323 | 0.63 | |
| HYDRABackbone=GPT-4o-mini, +CoT=false2025.05 | 0.324 | 0.656 | |
| FermiBackbone=GPT-4o-mini, +CoT=false2025.05 | 0.328 | 0.628 | |
| PAGBackbone=GPT-4o-mini, +CoT=false2025.05 | 0.331 | 0.662 | |
| ICLBackbone=GPT-4o-mini, +CoT=false2025.05 | 0.333 | 0.638 | |
| PAGBackbone=GPT-4o-mini, +CoT=true2025.05 | 0.339 | 0.672 | |
| PAGLLM Backbone=Claude-3.5 Sonnet, Category=Personalized LLM2025.06 | 0.339 | 0.835 | |
| HYDRABackbone=GPT-4o-mini, +CoT=true2025.05 | 0.353 | 0.672 | |
| Zero-shotBackbone=GPT-4o-mini, +CoT=false2025.05 | 0.361 | 0.68 | |
| RAGBackbone=GPT-4o-mini, +CoT=false2025.05 | 0.363 | 0.687 | |
| RAGBackbone=GPT-4o-mini, +CoT=true2025.05 | 0.366 | 0.69 |