Jaehyung Kim

Enhancing Instruction Following of LLMs via Activation Steering with Dynamic Rejection

Align to Misalign: Automatic LLM Jailbreak with Meta-Optimized LLM Judges

TiTok: Transfer Token-level Knowledge via Contrastive Excess to Transplant LoRA

Hacking LM Arena via LLM Identification with Interpolated Preference Learning

Prior-based Noisy Text Data Filtering: Fast and Strong Alternative For Perplexity

RPM: Reasoning-Level Personalization for Black-Box Large Language Models

Training-free LLM Verification via Recycling Few-shot Examples

RoboAlign: Reinforcement Learning for Action-Aligned Multimodal Large Language Models

Efficient LLM Collaboration via Planning

Revisit What You See: Disclose Language Prior in Vision Tokens for LVLM Decoding

Spread Preference Annotation: Direct Preference Judgment for Efficient LLM Alignment

Fast and Fluent Diffusion Language Models via Convolutional Decoding and Rejective Fine-tuning

Prior-based Noisy Text Data Filtering: Fast and Strong Alternative For Perplexity

Robot-R1: Reinforcement Learning for Enhanced Embodied Reasoning in Robotics

Debiasing Online Preference Learning via Preference Feature Preservation

Chain-of-thoughts for molecular understanding

ReVISE: Learning to Refine at Test-Time via Intrinsic Self-Verification

Can LLMs Generate Diverse Molecules? Towards Alignment with Structural Diversity