Personalized Generation on Roleplay (test)
72.36AccuracyPOPI-Full
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| POPI-FullMode=Full Training2025.10 | 72.36 | 200.22 | 70.12 | |
| PAL-BSpecial Token=true2025.10 | 66.94 | 1 | 33.04 | |
| LoReSpecial Token=true2025.10 | 66.77 | 1 | 65.29 | |
| PAL-ASpecial Token=true2025.10 | 66.6 | 1 | 28.41 | |
| Summary-AlignedPrompt Type=Summary, Aligned=true2025.10 | 66.14 | 529.47 | 53.45 | |
| Raw-AlignedPrompt Type=Raw, Aligned=true2025.10 | 65.31 | 3,020.61 | 38.66 | |
| POPI-Inference-OnlyMode=Inference-Only2025.10 | 54.81 | 200.22 | 55.36 | |
| Summary-PromptingPrompt Type=Summary2025.10 | 53.28 | 529.47 | 39.39 | |
| Raw-PromptingPrompt Type=Raw2025.10 | 53.06 | 3,020.61 | 35.26 | |
| Base-Model2025.10 | 50 | — | 50 |