Haohan Zhao — accepted AI-conference papers

2026

Lookahead-GCG: Improving Multi-Model Gradient-Based Jailbreaking Attacks via Nesterov Momentum

Rong Feng, Haohan Zhao, Shiqin Tang, Geng Liu, Song Lai, Meng Wang +4

ICML 2026poster

Transferable jailbreaking attacks enable red-teaming of black-box large language models by optimizing adversarial prompts on open-source surrogates. A natural approach to improve transferability is multi-model training---optimizing against multiple source models simultaneously. Yet this approach has…

Cited by 0 Source ↗Scholar ↗

2026

Reinforcement Fine-Tuning Naturally Mitigates Forgetting in Continual Post-Training

Song Lai, Haohan Zhao, Rong Feng, Changyi Ma, Wenzhuo Liu, Hongbo Zhao +6

ICML 2026poster

Continual post-training (CPT) is a popular and effective technique for adapting foundation models like multimodal large language models to ever-evolving downstream tasks. While existing research primarily focus on methods like data replay, model expansion, or parameter regularization, the fundamenta…

Cited by 0 Source ↗Scholar ↗