Joowan Chae

I am an undergraduate student at Sungkyunkwan University interested in language models, retrieval-augmented generation, preference optimization,autonomous data agents. I use this blog to organize paper reviews, math notes, and implementation-oriented study logs.

Research Interests

  • Retrieval-augmented generation and semantic search
  • Reasoning agents for structured data analysis
  • Embedding models and contrastive learning
  • LLM alignment and preference optimization
  • Reinforcement Learning
  • Multimodal reasoning for planning and control
  • Vision-Language-Action (VLA) models and embodied AI

Currently Studying About

  • Foundation model architectures and pre-training
  • LLM reasoning, ReAct-style workflows, and verification
  • Machine learning fundamentals and mathematical intuition
  • Course notes from operating systems, networks, and ML

Recent Posts

Text Embedding by Weakly-Supervised Contrastive Pre-training 리뷰

8 minute read

Background Text embeddings는 임의의 길이를 가진 text를 낮은 dimension으로 dense하게 표현하는 역할을 한다. 기존의 text 표현 방법인 TF-IDF은 전체 단어의 빈도수에 기반하기 때문에 dimension이 높고 sparse해서 Cosine과...