Synthetic Eggs in Many Baskets: The Impact of Synthetic Data Diversity on LLM Fine-Tuning

发布

2026年04月28日

采集 2026年04月28日 10:31

学术前沿 6.0 分 — Investigates impact of synthetic data source diversity on model behavior

原文： arxiv.org

评分 6 · 来源： · 发布于 2026-04-28

评分依据：Investigates impact of synthetic data source diversity on model behavior

Scoring, Reasoning, and Selecting the Best! Ensembling Large Language Models via a Peer-Review Process

POPI: Personalizing LLMs via Optimized Natural Language Preference Inference