Han Shen
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Han Shen has a total of 4 co-authors. Here are the top co-authors by number of shared publications:
- Tianyi Chen
Rensselaer Polytechnic Institute
2 shared publications - Payel Das
IBM Research Yorktown Heights NY United States
1 shared publication - Pin-Yu Chen
IBM Research, Yorktown Heights, New York, USA
1 shared publication - Quan Xiao
Rensselaer Polytechnic Institute
1 shared publication
Latest Publications
- Top Cited Papers
- 5 February 2025
- Published by Springer Nature in Mathematical Programming
- Vol. 214 (1-2) , 539-589
On penalty-based bilevel gradient descent method
- 5 February 2025
- Published by Springer Nature in Mathematical Programming
- Vol. 214 (1-2) , 539-589
Top Cited PapersOn penalty-based bilevel gradient descent method
- Open Access
SEAL: Safety-enhanced Aligned LLM Fine-tuning via Bilevel Data Selection
Fine-tuning on task-specific data to boost downstream performance is a crucial step for leveraging Large Language Models (LLMs). However, previous studies have demonstrated that fine-tuning the models on several adversarial samples or even benign data can greatly comprise the model's pre-equipped alignment and safety capabilities. In this work, we propose SEAL, a novel framework to enhance safety in LLM fine-tuning. SEAL learns a data ranker based on the bilevel optimization to up rank the safe and high-quality fine-tuning data and down rank the unsafe or low-quality ones. Models trained with SEAL demonstrate superior quality over multiple baselines, with 8.5% and 9.7% win rate increase compared to random selection respectively on Llama-3-8b-Instruct and Merlinite-7b models. Our code is available on github https://github.com/hanshen95/SEAL.
Open AccessSEAL: Safety-enhanced Aligned LLM Fine-tuning via Bilevel Data Selection
Fine-tuning on task-specific data to boost downstream performance is a crucial step for leveraging Large Language Models (LLMs). However, previous studies have demonstrated that fine-tuning the models on several adversarial samples or even benign data can greatly comprise the model's pre-equipped alignment and safety capabilities. In this work, we propose SEAL, a novel framework to enhance safety in LLM fine-tuning. SEAL learns a data ranker based on the bilevel optimization to up rank the safe and high-quality fine-tuning data and down rank the unsafe or low-quality ones. Models trained with SEAL demonstrate superior quality over multiple baselines, with 8.5% and 9.7% win rate increase compared to random selection respectively on Llama-3-8b-Instruct and Merlinite-7b models. Our code is available on github https://github.com/hanshen95/SEAL.