Janghoon Han

Research Scientist at LG AI Research

hanjh9439@gmail.com

About
Hello! I am a fourth-year researcher at LG AI Research, focusing on developing conversational large language models. My primary area of research is instruction tuning, and I am currently conducting preliminary research on robustness to diverse instructions. Previously, I developed efficient and effective instruction tuning methodologies for specialized language models targeting specific domains and conducted research on cross-lingual transfer for instruction alignment. I am also deeply interested in dialogue systems, with experience in retrieval-based, generative, knowledge-based, and dialogue state tracking (DST) models. I completed my bachelor’s and master’s degrees at Sogang University under Professor Jungyun Seo's guidance.
Recent News
Oct 2024     🌟 Exciting News! Instruction Matters: A Simple yet Effective Task Selection Approach in Instruction Tuning for Specific Tasks co-first authored with Changho Lee has been accepted for publication at EMNLP 2024! 🎉
August 2024     Efficient Dynamic Hard Negative Sampling for Dialogue Selection has been accepted at NLP4ConvAI @ ACL2024! 🎉
August 2024     🌟 Exciting News! Deep Exploration of Cross-Lingual Zero-Shot Generalization in Instruction Tuning co-first authored with Changho Lee has been accepted for publication at Findings of ACL 2024! 🎉

Education and Experience

LG AI ResearchMar. 2021 - Present

Research Scientist (advisors: Stanley Jungkyu Choi and Honglak Lee)

Sogang University Mar. 2019 - Feb. 2021

M.S. in Natural Language Processing (advisor: Jungyun Seo)

Sogang UniversityMar. 2013 - Feb. 2019

B.S. in Computer Science

Publications

2024

Instruction Matters: A Simple yet Effective Task Selection Approach in Instruction Tuning for Specific Tasks

Changho Lee*, Janghoon Han*, Seonghyeon Ye, Stanley Jungkyu Choi, Honglak Lee, Kyunghoon Bae

EMNLP 2024

Paper Code

Exploring the Use of Natural Language Descriptions of Intents for Large Language Models in Zero-shot Intent Classification

Taesuk Hong, Youbin Ahn, Dongkyu Lee, Joongbo Shin, Seungpil Won, Janghoon Han, Stanley Jungkyu Choi, Jungyun Seo

SIGDIAL 2024

Paper

Efficient Dynamic Hard Negative Sampling for Dialogue Selection

Janghoon Han, Dongkyu Lee, Joongbo Shin, Hyunkyung Bae, Jeesoo Bang, Seonghwan Kim, Stanley Jungkyu Choi, Honglak Lee

NLP4ConvAI @ ACL 2024

Paper Code

Deep Exploration of Cross-Lingual Zero-Shot Generalization in Instruction Tuning

Janghoon Han*, Changho Lee*, Joongbo Shin, Stanley Jungkyu Choi, Honglak Lee, Kyunghoon Bae

ACL 2024 Findings

Paper Code

2023

Leveraging Ensemble Techniques and Metadata for Subjective Knowledge-grounded Conversational Systems

Seongho Joo, Kang-il Lee, Kyungmin Min, Joongbo Shin, Janghoon Han, Seungpil Won, Kyomin Jung

DSTC11 @ SIGDIAL 2023 (Best paper)

Paper

Local Temperature Beam Search: Avoid Neural Text DeGeneration via Enhanced Calibration

Dongkyu Lee, Gyeonghun Kim, Janghoon Han, Taesuk Hong, Yi-Reun Kim, Stanley Jungkyu Choi, Nevin L. Zhang

ACL 2023 Findings

Paper

BREAK: Breaking the dialogue state tracking barrier with beam search and re-ranking

Seungpil Won, Heeyoung Kwak, Joongbo Shin, Janghoon Han, Kyomin Jung

ACL 2023

Paper

CECADA: Cause-Effect Conjunctive Adverb-based Data Augmentation Method in Low-Resource Knowledge-Grounded Dialogue

Taesuk Hong, Dongkyu Lee, Janghoon Han, Stanley Jungkyu Choi, Jungyun Seo

Information Sciences, Knowledge Augmented Methods for NLP @ AAAI 2023

Paper

2022

TemporalWiki: A Lifelong Benchmark for Training and Evaluating Ever-Evolving Language Models

Joel Jang*, Seonghyeon Ye*, Changho Lee, Sohee Yang, Joongbo Shin, Janghoon Han, Gyeonghun Kim, Minjoon Seo

EMNLP 2022

Paper Code

Towards Continual Knowledge Learning of Language Models

Joel Jang, Seonghyeon Ye, Sohee Yang, Joongbo Shin, Janghoon Han, Gyeonghun Kim, Stanley Jungkyu Choi, Minjoon Seo

ICLR 2022

Paper Code

External Knowledge Selection with Weighted Negative Sampling in Knowledge-grounded Task-oriented Dialogue Systems

Janghoon Han, Joongbo Shin, Hosung Song, Hyunjik Jo, Gyeonghun Kim, Yireun Kim, Stanley Jungkyu Choi

DSTC10 @ AAAI 2022

Paper Code

2021

External Knowledge Incorporated BERT for Retrieval-based Dialogue Systems

Janghoon Han, Youngjoong Ko, Jungyun Seo

KTCP 2021

Paper Code

Fine-grained Post-training for Improving Retrieval-based Dialogue Systems

Janghoon Han, Taesuk Hong, Byoungjae Kim, Youngjoong Ko, Jungyun Seo

NAACL 2021

Paper Code

2020

Korean Dependency Parsing Using Sequential Parsing Method Based on Pointer Network

Janghoon Han, Yeongjoon Park, Younghoon Jeong, Inkwon Lee, Jungwook Han, Seojun Park, Juae Kim, and Jeongyeon Seo

Proceedings of the 31st Annual Conference on Human and Cognitive Language Technology (Grand Prize)

Paper Code
( * indicates equal contribution )

Vitæ

Full CV in PDF.

  • LG AI Research Feb 2021 - Present
    Research Scientist (Leader: Stanley Junkyu Choi, Honglak Lee)
    Conversational Large Language Model Development, Instruction tuning
  • NLP Lab, Sogang University Mar. 2019 - Feb. 2021
    M.S. Student (Advisor: Jungyun Seo)
    Dialgue System, Dependency parsing
  • Military Service in South Korea May. 2015 - Feb. 2017
    Discharged as Sergeant
  • Sogang University Mar. 2013 - Feb. 2019
    B.S. in Computer Science
Janghoon Han
Last updated on Oct 30, 2024
-->