| title | Mine, Sample, Render: Test-Time Diversity Alignment Of Generative Models |
|---|---|
| authors | Kostas Triaridis, Morris E. Sharp, Nebojsa Jojic |
| date | 2026-10-15 |
| venue | arXiv (preprint, not peer reviewed) |
| id | arXiv:2610.21185 |
| upvotes | 8 (Hugging Face Daily Papers, 2026-11-01) |
| links | abs pdf hf |
| abstract | Modern generative models produce high-quality individual outputs but struggle to sample diversely, even when explicitly instructed to do so. We introduce MINE-SAMPLE-RENDER (MSR), a test-time approach that externalizes the diversity into an explicit, corpus-derived sampling procedure. Our pipeline (i) mines an interpretable categorical schema from a corpus of real artifacts and supplies statistics for sampling; (ii) samples attribute configurations from the corpus-derived statistics; (iii) uses a generative model purely as a conditional renderer of a sampled configuration. Across weather report generation, user simulation, webpage design, text-to-image generation, and agent-generated 3D assets, we evaluate the diversity and output quality of MSR-guided generation and show that MSR improves alignment with corpus distributions and diversity of generated outputs. |
cite
@misc{triaridis2026mine,
title={Mine, Sample, Render: Test-Time Diversity Alignment Of Generative Models},
author={Kostas Triaridis and Morris E. Sharp and Nebojsa Jojic},
year={2026},
eprint={2610.21185},
archivePrefix={arXiv},
url={https://arxiv.org/abs/2610.21185}
}