Junhyuck Kim

About

I research and train large language models at KRAFTON AI, led by Kangwook Lee, and collaborate closely with Dimitris Papailiopoulos. Currently, I am working on both model architectures and reinforcement learning for the Raon model family (yes, we are a small team!).

I am particularly interested in efficiency, with the long-term goal of building AI systems that are useful in practice, even under tight resource constraints. To this end, I work on reducing the memory footprint of inference and reasoning, as well as the computational cost of training. More recently, I have been exploring efficient agentic systems for automated research.

Previously, I completed my master’s at Cambridge, where I worked in the Machine Learning Group under the supervision of Richard Turner.

News

Oct. 2026 We released Raon-1 mini preview, a 30B-A4B MoE language model, to which I contributed through architecture design and reinforcement learning. 𝕏 Model Try Now
Sep. 2026 Our paper, Pruning and Distilling Mixture-of-Experts into Dense Language Models, was accepted to NeurIPS 2026 as an Oral presentation. I’ll be presenting this work in Sydney!

Publications

(* Equal contribution, † Core contribution)

Pruning and Distilling Mixture-of-Experts into Dense Language Models

Junhyuck Kim, Jihun Yun, Haechan Kim, Gyeongman Kim, Joonghyun Bae, Jaewoong Cho

NeurIPS 2026 Oral Code 𝕏

Is the Importance Ratio Necessary for Stable Reinforcement Learning in LLMs?

Shuibai Zhang*, Junhyuck Kim*, Gyeongman Kim, Jaewoong Cho

NeurIPS 2026

Raon-Speech Technical Report

KRAFTON

Technical Report, 2026 Code 𝕏 Model Demo

Not All Bits Are Equal: Scale-Dependent Memory Optimization Strategies for Reasoning Models

Junhyuck Kim, Ethan Ewer, Taehong Moon, Jongho Park, Dimitris Papailiopoulos

ICLR 2026 · NeurIPS 2025 ER Workshop Spotlight Code 𝕏

Orak: A Foundational Benchmark for Training and Evaluating LLM Agents on Diverse Video Games

Dongmin Park*†, Minkyu Kim*†, Beongjun Choi*†, Junhyuck Kim†, Keon Lee†, Jonghyun Lee†, Inkyu Park†, Byeong-Uk Lee†, Jaeyoung Hwang†, Jaewoo Ahn†, Ameya S. Mahabaleshwarkar, Bilal Kartal, Pritam Biswas, Yoshi Suhara, Kangwook Lee, Jaewoong Cho

ICLR 2026 · EMNLP 2025 Wordplay Workshop Outstanding Paper Code 𝕏

Beyond RLHF: A Unified Theoretical Framework of Alignment

Jihun Yun*, Juno Kim*, Jongho Park, Junhyuck Kim, Jongha Jon Ryu, Jaewoong Cho, Kwang-Sung Jun

ICML 2025 MoFA Workshop

Lexico: Extreme KV Cache Compression via Sparse Coding over Universal Dictionaries

Junhyuck Kim, Jongho Park, Jaewoong Cho, Dimitris Papailiopoulos

ICML 2025 · ICLR 2025 SLLM Workshop Spotlight Code 𝕏

Translation Equivariant Transformer Neural Processes

Matthew Ashman, Cristiana Diaconu, Junhyuck Kim, Lakee Sivaraya, Stratis Markou, James Requeima, Wessel P. Bruinsma, Richard E. Turner

ICML 2024