Research

Publications

Selected work on knowledge grounding, abstraction, commonsense reasoning, and language models.

2025
Parameter-Efficient Checkpoint Merging via Metrics-Weighted Averaging
Selected
Shi Jie Yu, Sehyun Choi
arXiv preprint
arXiv
2024
AbsInstruct: Eliciting Abstraction Ability from LLMs through Explanation Tuning with Plausibility Estimation
Zhaowei Wang, Wei Fan, Qing Zong, Hongming Zhang, Sehyun Choi, Tianqing Fang, Xin Liu, Yangqiu Song, Ginny Wong, Simon See
ACL 2024
Paper DOI
Abstract
AbsInstruct enhances language models’ abstraction ability through instruction tuning with detailed explanations and a plausibility estimator, while maintaining general instruction-following abilities.
2024
AbsPyramid: Benchmarking the Abstraction Ability of Language Models with a Unified Entailment Graph
Zhaowei Wang, Haochen Shi, Weiqi Wang, Tianqing Fang, Hongming Zhang, Sehyun Choi, Xin Liu, Yangqiu Song
NAACL 2024 Findings
Paper DOI
Abstract
AbsPyramid is a unified entailment graph of 221K textual descriptions of abstraction knowledge, designed to evaluate language models’ abstraction ability across diverse events.
2024
Cross-Architecture Transfer Learning for Linear-Cost Inference Transformers
Selected
Sehyun Choi
arXiv preprint
arXiv
2023
KCTS: Knowledge-Constrained Tree Search Decoding with Token-Level Hallucination Detection
Selected
Sehyun Choi, Tianqing Fang, Zhaowei Wang, Yangqiu Song
EMNLP 2023
Paper PDF Code Poster Slides
Abstract
KCTS guides a frozen language model to generate text aligned with reference knowledge using knowledge-classifier scores and Monte Carlo Tree Search. It also introduces RIPA, a token-level hallucination detection method.
2023
CKBP v2: An Expert-Annotated Evaluation Set for Commonsense Knowledge Base Population
Tianqing Fang, Quyet V. Do, Sehyun Choi, Weiqi Wang, Yangqiu Song
arXiv preprint
arXiv Code
2021
Benchmarking Commonsense Knowledge Base Population with an Effective Evaluation Dataset
Selected
Tianqing Fang, Weiqi Wang, Sehyun Choi, Shibo Hao, Hongming Zhang, Yangqiu Song, Bin He
EMNLP 2021
Paper Code
2021
Fine-grained Pneumonia Detection from Chest X-Rays Using Complementary Model Training
Duhyeuk Chang, Sungbin Lee, Sehyun Choi
한국정보과학회 학술발표논문집
DBpia

Parameter-Efficient Checkpoint Merging via Metrics-Weighted Averaging

AbsInstruct: Eliciting Abstraction Ability from LLMs through Explanation Tuning with Plausibility Estimation

AbsPyramid: Benchmarking the Abstraction Ability of Language Models with a Unified Entailment Graph

Cross-Architecture Transfer Learning for Linear-Cost Inference Transformers

KCTS: Knowledge-Constrained Tree Search Decoding with Token-Level Hallucination Detection

CKBP v2: An Expert-Annotated Evaluation Set for Commonsense Knowledge Base Population

Benchmarking Commonsense Knowledge Base Population with an Effective Evaluation Dataset

Fine-grained Pneumonia Detection from Chest X-Rays Using Complementary Model Training