Papers by Kangwook Lee
SLM: Learning a Discourse Language Representation with Sentence Unshuffling (2020.emnlp-main)
Copied to clipboard
| Challenge: | Recent models for learning discourse language representations focus on bottom or top-level representations, but they do not capture intermediate-size structures in natural languages such as sentences and the relationships among them. |
| Approach: | They propose a new objective for learning a discourse language representation in a self-supervised manner by shuffling the sequence of input sentences and training a hierarchical transformer model to reconstruct the original ordering. |
| Outcome: | The proposed model improves the original BERT model on downstream tasks by large margins. |
Generalizable Prompt Tuning for Audio-Language Models via Semantic Expansion (2026.findings-acl)
Copied to clipboard
| Challenge: | Prompt tuning has achieved remarkable progress in vision–language models, but its generalization ability in ALMs remains underexplored. |
| Approach: | They propose a plug-and-play framework that regularizes the prompt embedding space . they propose introducing a semantic expansion loss with margin constraints that promote compactness . |
| Outcome: | The proposed framework regularizes the prompt embedding space by incorporating semantic neighbors generated by large language models. |
Prompted LLMs as Chatbot Modules for Long Open-domain Conversation (2023.findings-acl)
Copied to clipboard
| Challenge: | Using pre-trained large language models (LLMs) as individual modules for long-term consistency and flexibility is a challenge for open-domain chatbots due to the computational burden of updating models with billions of parameters and the scarcity of data in the dialogue domain. |
| Approach: | They propose a method that uses pre-trained large language models as individual modules for long-term consistency and flexibility. |
| Outcome: | The proposed method is on par with fine-tuned chatbot models in open-domain conversations, showing it can create consistent and engaging chatbots. |
Sparse Logit Sampling: Accelerating Knowledge Distillation in LLMs (2025.acl-long)
Copied to clipboard
Anshumann Anshumann, Mohd Abbas Zaidi, Akhil Kedia, Jinwoo Ahn, Taehwak Kwon, Kangwook Lee, Haejun Lee, Joohyung Lee
| Challenge: | Knowledge distillation is a cost-effective technique to distill knowledge in Large Language Models, if the teacher output logits can be pre-computed and cached. |
| Approach: | They propose an importance-sampling-based method which provides unbiased estimates, preserves the gradient in expectation, and requires storing significantly sparser logits. |
| Outcome: | The proposed method enables faster training of student models with marginal overhead (10%) compared to cross-entropy based training, while maintaining competitive performance compared with full distillation. |
TABED: Test-Time Adaptive Ensemble Drafting for Robust Speculative Decoding in LVLMs (2026.findings-eacl)
Copied to clipboard
Minjae Lee, Wonjun Kang, Byeongkeun Ahn, Christian Classen, Kevin Galim, Seunghyuk Oh, Minghao Yan, Hyung Il Koo, Kangwook Lee
| Challenge: | Large Vision Language Models (LVLMs) are advanced models that process multiple modalities, such as images, audio, and video, alongside text. |
| Approach: | They propose to use a method to generate and verify draft tokens in parallel . they compare existing methods with small draft models and observe performance fluctuations . |
| Outcome: | The proposed method achieves an average walltime speedup of 1.74 over autoregressive decoding and a 5% improvement over single drafting methods. |
Learning to Embed Multi-Modal Contexts for Situated Conversational Agents (2022.findings-naacl)
Copied to clipboard
Haeju Lee, Oh Joon Kwon, Yunseon Choi, Minho Park, Ran Han, Yoonhyung Kim, Jinhyeon Kim, Youngjune Lee, Haebin Shin, Kangwook Lee, Kee-Eung Kim
| Challenge: | Situated Interactive Multi-Modal Conversations 2.0 aims to create virtual shopping assistants that can accept complex multi-modal inputs. |
| Approach: | They propose a joint learning approach that integrates visual inputs and performs all four subtasks at once for efficiency. |
| Outcome: | The proposed approach won the 10th Dialog Systems Technology Challenge (DSTC10) . it incorporates visual inputs and performs all four subtasks at once for efficiency . |
Utilizing Language-Image Pretraining for Efficient and Robust Bilingual Word Alignment (2022.findings-emnlp)
Copied to clipboard
Tuan Dinh, Jy-yong Sohn, Shashank Rajput, Timothy Ossowski, Yifei Ming, Junjie Hu, Dimitris Papailiopoulos, Kangwook Lee
| Challenge: | Recent studies show that unsupervised word translation is more accurate and robust without parallel corpora. |
| Approach: | They propose a method for unsupervised word translation that leverages visual observations and pretrained language-image models to align words. |
| Outcome: | The proposed method improves on the state-of-the-art language-image pretraining method for bilingual word alignment. |