Sung Ju Hwang

LS-Merge: Merging Language Models in Latent Space

Rethinking Reward Models for Multi-Domain Test-Time Scaling

Multimodal Prompt Optimization: Why Not Leverage Multiple Modalities for MLLMs

Multi-View Encoders for Performance Prediction in LLM-Based Agentic Workflows

Paper2Code: Automating Code Generation from Scientific Papers in Machine Learning

Learning from Adversity: Semantic-Aware Mask Refinement through Adversarial Perturbation

Frame Guidance: Training-Free Guidance for Frame-Level Control in Video Diffusion Models

Simple yet Effective Semi-supervised Knowledge Distillation from Vision-Language Models via Dual-Head Optimization

MA-EgoQA: Question Answering over Egocentric Videos from Multiple Embodied Agents

InfiniteHiP: Extending Language Model Context Up to 3 Million Tokens on a Single GPU

Measure Once, Mask Once: Delta Refined Block Sparse Attention

PCoreSet: Effective Active Learning through Knowledge Distillation from Vision-Langauge Models

UniversalRAG: Retrieval-Augmented Generation over Corpora of Diverse Modalities and Granularities

HoliSafe: Holistic Safety Benchmarking and Modeling for Vision-Language Model

AutoML-Agent: A Multi-Agent LLM Framework for Full-Pipeline AutoML

Distilling LLM Agent into Small Models with Retrieval and Code Tools

FedSVD: Adaptive Orthogonalization for Private Federated Learning with LoRA

Learning Diverse Attacks on Large Language Models for Robust Red-Teaming and Safety Tuning

HarmAug: Effective Data Augmentation for Knowledge Distillation of Safety Guard Models

Delta Attention: Fast and Accurate Sparse Attention Inference by Delta Correction

Cost-Sensitive Freeze-thaw Bayesian Optimization for Efficient Hyperparameter Tuning

System Prompt Optimization with Meta-Learning

Mol-LLaMA: Towards General Understanding of Molecules in Large Molecular Language Model

A Training-Free Sub-quadratic Cost Transformer Model Serving Framework with Hierarchically Pruned Attention

Bayesian Neural Scaling Law Extrapolation with Prior-Data Fitted Networks

Training Free Exponential Context Extension via Cascading KV Cache

Diffusion-based Neural Network Weights Generation

Understanding Dimensional Collapse in Cross-Modal Feature Distillation

Automatic Jailbreaking of Text-to-Image Generative AI Systems for Copyright Infringement

AutoML-Agent: A Multi-Agent LLM Framework for Full-Pipeline AutoML

Continuous Diffusion Model for Language Modeling

Cost-Sensitive Multi-Fidelity Bayesian Optimization

Unified Multi-Modal Interleaved Document Representation for Information Retrieval

Preventing Unintended Memorization by Covering with Over-Memorization

Realistic World Model for Autonomous Driving: Integrating Physical Constraints and Multi-agent Interactions