From the 1 of 4 linked papers with an AI index.
4 papers
LoRA Scaffolded Policy Optimization (LSPO): A Sampling-Time Low-Rank Scaffold for Recovering Reinforcement-Learning Gradient on Zero-Reward Cliff Prompts
Ken Ding
The paper introduces LoRA Scaffolded Policy Optimization (LSPO), a sampling-time low-rank adapter method that recovers gradient information for reinforcement learning on difficult…
HDPO: Hybrid Distillation Policy Optimization via Privileged Self-Distillation
Ken Ding
Large language models trained with reinforcement learning (RL) for mathematical reasoning face a fundamental challenge: on problems the model cannot solve at all - "cliff" prompts…
Breaking Memorization Barriers in LLM Code Fine-Tuning via Information Bottleneck for Improved Generalization
Changsheng Wang, Xin Chen, Sijia Liu +1
Adapting pretrained large language models (LLMs) to code domains via supervised fine-tuning (FT) has been commonly used for code generation. However, we identify a previously under…
Defensive Unlearning with Adversarial Training for Robust Concept Erasure in Diffusion Models
Yimeng Zhang, Xin Chen, Jinghan Jia +6
Diffusion models (DMs) have achieved remarkable success in text-to-image generation, but they also pose safety risks, such as the potential generation of harmful content and copyri…