Papers by Hwichan Kim

7 papers
A Single Linear Layer Yields Task-Adapted Low-Rank Matrices (2024.lrec-main)

Copied to clipboard

Challenge: Low-Rank Adaptation (LoRA) is a parameter-efficient fine-tuning method that updates initial weight matrix W0 with a delta matrix W .
Approach: They propose a method that updates initial weight matrix W0 with a delta matrix W consisting of two low-rank matrices A and B.
Outcome: The proposed method maintains a performance on par with LoRA despite the fact that the trainable parameters of CondLoRA are fewer than those of LoRA.
Zero-shot North Korean to English Neural Machine Translation by Character Tokenization and Phoneme Decomposition (2020.acl-srw)

Copied to clipboard

Challenge: a limited number of North Korean to English translation models have been developed . a zero-shot approach is proposed to train a neural machine translation model using South Korean data .
Approach: They propose a method to tokenize South Korean input sentences and decompose them into phonemes.
Outcome: The proposed method improves the BLEU scores by +1.01 points compared with the baseline . the proposed method can learn North Korean to English translation and improve the linguistic accuracy.
Learning How to Translate North Korean through South Korean (2022.lrec-1)

Copied to clipboard

Challenge: Existing NLP systems cannot properly handle North Korean inputs, despite limited data . Several NLP researchers have been working on the Korean language .
Approach: They propose to manually create evaluation data for automatic alignment and machine translation, and investigate automatic alignment methods suitable for North Korean.
Outcome: The proposed model trained by North Korean bilingual data significantly boosts translation accuracy compared to existing South Korean models in zero-shot settings.
Pruning Multilingual Large Language Models for Multilingual Inference (2024.findings-emnlp)

Copied to clipboard

Challenge: Multilingual large language models (MLLMs) demonstrate better zeroshot learning performance in non-English languages compared to large language model trained on English-dominant data.
Approach: They propose a pruning approach to prune large language models using bilingual sentence pairs from English and other languages to enhance their performance in non-English language.
Outcome: The proposed pruning strategy enhances the MLLMs’ performance in non-English language.
Does Masked Language Model Pre-training with Artificial Data Improve Low-resource Neural Machine Translation? (2023.findings-eacl)

Copied to clipboard

Challenge: Pre-training masked language models with artificial data has been proven beneficial for several natural language processing tasks, however, it has been less explored for neural machine translation (NMT).
Approach: They pre-trained masked language models with random sequences and created artificial data mimicking token frequency information from the real world.
Outcome: The results show that pre-training models with artificial data improves translation performance in low-resource situations.
Enhancing Few-shot Cross-lingual Transfer with Target Language Peculiar Examples (2023.findings-acl)

Copied to clipboard

Challenge: We use multilingual mask language models to train a model with hard examples, but if no labeled data are available, they need to be created through human annotations.
Approach: They devised a metric to select annotation candidates from an unlabeled data pool that efficiently enhance accuracy for few-shot cross-lingual transfer.
Outcome: The proposed model improves accuracy for few-shot cross-lingual transfer in 20 languages and 6 tasks with high peculiarity examples.
A Fair Comparison without Translationese: English vs. Target-language Instructions for Multilingual LLMs (2025.naacl-short)

Copied to clipboard

Challenge: Prior studies suggested that English instructions are more effective for non-English tasks . however, these studies often use datasets and instructions translated from English .
Approach: They conduct a fair comparison between English and target-language instructions by eliminating translationese effects.
Outcome: The results show that the advantage of adopting English instructions is not overwhelming . the results also show that instruction-following abilities are improved when using respective instructions.

What is GenGO?

GenGO is an NLP powered publication search system. It currenctly indexes 30k+ papers from ACL Anthology, and implements multi-aspect summarization, semantic search, and more!

Information

About
Limitations