AI Paper Atlas

Research map / Probabilistic and scientific modeling / Molecules and proteins

Curated Synthetic Data Doesn’t Have to Collapse: A Theoretical Study of Generative Retraining with Pluralistic Preferences

Ali Falahati, Mohammad Mohammadi Amiri, Kate Larson, Lukasz Golab

ICML 2026 · Probabilistic and scientific modeling › Molecules and proteins · approach: Cold-start recommendation models

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
increase number of balanced preferences from two to five (ablation)pluralistic curation · CIFAR-10 retrainingFID: 77.6 -> 35.9Improved
increase number of balanced preferences from two to five (ablation)pluralistic curation · CIFAR-10 retrainingclass entropy: 1.098 -> 1.687Improved
increase number of balanced preferences from two to five (ablation)pluralistic curation · CIFAR-10 retrainingKL divergence: 1.204 -> 0.616Improved
use two preferences with balanced q=0.5 instead of single preference q=0.0 (ablation)pluralistic curation · CIFAR-10 retrainingFID: 100.0 -> 77.6Improved
use balanced q=0.5 instead of single preference q=0.0 (ablation)pluralistic curation · CIFAR-10 retrainingclass entropy: 0.032 -> 1.098Improved
skew q toward 0.1-0.3 (ablation)two-preference curation · CIFAR-10 retrainingclass entropy: 0.010 at q=0.1, 0.036 at q=0.3No clear effect
reduce distance between reward optima below about 2 (ablation)pluralistic curation · synthetic 2D Gaussian mixturereward varianceHurt
use pluralistic instead of single-reward curationGMM retraining · synthetic 2D Gaussian mixturereward varianceImproved
increase distance between conflicting length targets (ablation)GPT-2 pluralistic retraining · length preference textlength entropyImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy