Scholar
Taehong Moon
Google Scholar ID: wBwIIYQAAAAJ
Deep Learning Researcher, KRAFTON Inc.
Citations & Impact
Citations
96
H-index
4
i10-index
3
Publications
6
Co-authors
11
list available
Publications
Resume (English only)
Academic Achievements
- - October 2025: 'Not All Bits Are Equal: Scale-Dependent Memory Optimization Strategies for Reasoning Models' accepted to NeurIPS 2025 ER Workshop
- - May 2025: 'ResGen' and 'How to Move Your Dragon' accepted to ICML 2025
- - January 2025: 'Rare-to-Frequent: Unlocking Compositional Generation Power of Diffusion Models on Rare Concepts with LLM Guidance' accepted to ICLR 2025 as a spotlight presentation
- - May 2024: 'A Simple Early Exiting Framework for Accelerated Sampling in Diffusion Models' accepted to ICML 2024
- - 2023: Contributed to 'HyperCLOVA X Technical Report'
- - 2023: 'Early Exiting for Accelerated Inference in Diffusion Models' accepted to ICML Workshop on Structured Probabilistic Inference & Generative Modeling
- - 2022: 'Fine-tuning Diffusion Models with Limited Data' accepted to NeurIPS Workshop on Score-Based Methods
Research Experience
- - November 2023 - Present: Deep Learning Researcher at KRAFTON AI
- - June 2023 - October 2023: Deep Learning Research Internship at Naver Cloud, participated in the development of HyperCLOVA-X
Background
- - A deep learning researcher specializing in generative models for gaming applications
- - Currently developing a novel type of in-game character, Co-playable Character (CPC), designed to interact with users in real-time
- - Broader research goal is to advance multi-modal generative models that bridge research and real-world deployment
- - Focusing on efficient generative modeling and synthetic data generation for speech-language models
