Researchers have developed Anacréon, a novel audience simulation model designed to predict individual responses within specific domains, addressing the limitations of current large language models that tend to generalize population averages. Anacréon utilizes an authorship embedding and dedicated adapters trained on a Gemma 4-12B base to capture individual heterogeneity, psychological traits, and survey responses from public text. This approach achieved a state-of-the-art ordinal alignment score of 0.775 on an external survey, marking a step towards more faithful individual-level simulation. AI
IMPACT This model's ability to simulate individuals could enhance personalized content generation and improve the accuracy of social science research.
RANK_REASON The cluster describes a research paper published on arXiv detailing a new model and its performance on a specific benchmark.
Read on Hugging Face Daily Papers →
- Anacréon
- Gemma 4-12B
- alphaXiv
- arXiv
- CatalyzeX
- Connected Papers
- CORE Recommender
- DagsHub
- Gotit.pub
- Hugging Face
- Influence Flower
- Litmaps
- ScienceCast
- scite Smart Citations
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →
