Hyun Oh Song

TWISTED: Enhancing Transformer World Models with Spatio-Temporal Encoding and Graph-Based Optimal Decoding

KVzip: Query-Agnostic KV Cache Compression with Context Reconstruction

Learning to Better Search with Language Models via Guided Reinforced Self-Training

GuidedQuant: Large Language Model Quantization via Exploiting End Loss Guidance

Q-Palette: Fractional-Bit Quantizers Toward Optimal Bit Allocation for Efficient LLM Deployment

Guided Stream of Search: Learning to Better Search with Language Models via Optimal Path Guidance