Researcher — paper.dinq.me

作者位次 #8/9NeurIPS 2026Poster

Beyond the Dirac Delta: Mitigating Diversity Collapse in Reinforcement Fine-Tuning for Image Generation

Jinmei Liu · Haoru Li · Zhenhong Sun · Chaofeng Chen · Yatao Bian · Hongdong Li · Bo Wang · Daoyi Dong · Zhi Wang

强化学习与后训练多模态OpenReview

作者位次 #5/6ACL 2026Poster· Natural Language Processing

Select Before Use: On the Importance of Reference Model Selection in Preference Alignment

Muyang Li · Runze Wu · an Jiaotong University) · Bo Han · Daoyi Dong · Tongliang Liu

摘要

The post-training stage of Large Language Models (LLMs) typically involves Supervised Fine-Tuning (SFT) followed by preference alignment to ensure LLM to generate safe, helpful, and instruction-aligned content. The SFT model critically serves as both the initialization and reference model for sub...

对齐与偏好强化学习与后训练OpenReview 项目页

Daoyi Dong

获取完整研究档案

联系方式

2026 顶会论文

Beyond the Dirac Delta: Mitigating Diversity Collapse in Reinforcement Fine-Tuning for Image Generation

Select Before Use: On the Importance of Reference Model Selection in Preference Alignment

常见合作者