Yu-Ting Lee
Ó +886+9+06782385 |
|
¯ LinkedIn
|
GitHub
|
Website
NLP Engineer with 5 years of experience, encompassing with 6 years of machine learning expertise.
Skills
Languages: Python, Mojo, Julia, SQL, R, C/C++, Java, MATLAB
Software & Tools: PyTorch, PyTorch Lightning, Jupyter, GCP, GNU/Linux, Pandas,
Scikit-Learn, NLTK, Hugging Face, Gradio, LangChain, Git, Docker, Tensorflow, Keras, FastAPI
NLP Techniques: RAG, LLMOps, Fine-tune LLMs, Text Generation, Multi-Document
Summarization, Recommendation System, Text Classification, Named Entity Recognition
Work Experience
Research Assistant Oct 2023 – Present
Institute of Information Science, Academia Sinica Taipei City, Taiwan
•
Developing a system for Research Methods Recommendation.
•
Applying algorithms to time-series contents generation settings and further evaluate with
state-of-the-art text generation models.
Graduate Research Assistant Oct 2021 – Oct 2023
Institute of Information Science, Academia Sinica Taipei City, Taiwan
•
Developed a state-of-the-art Knowledge Updating System in Full-Text, enabled collaborated
content generation with GPT-4, performs 30% better than existed models.
•
Fine-tuned LLaMA based models (Alpaca, Vicuna) with LoRA and Adapter.
•
Algorithm development for knowledge updating utilizing state-of-the-art NLP techniques.
Graduate Research Assistant Sep 2021 – Sep 2023
Department of Journalism, National Chengchi University Taipei City, Taiwan
•
Collaborated and developed knowledge updating system with expertise of journalism.
•
Articulated research ideas to technical/non-technical audience.
Education
National Chengchi University Taipei City, Taiwan
Department of Computer Science, Master’s Degree Sep 2021 – Oct 2023
•
Advised by Tsai-Yen Li and Hen-Hsen Huang.
•
Essay: Knowledge Update in Full Text Triggered by a News Event. Developed LLM for
knowledge updating, outperformed state-of-the-art LLMs (e.g. GPT-4, LLaMA, Vicuna, etc.).
•
Relevant coursework: Developed Wordle-based game with optimization method.
Chung Shan Medical University Taichung City, Taiwan
Department of Medical Informatics, Bachelor’s Degree Sep 2017 – June 2021
•
Honor: Supported by the Ministry of Science and Technology Research Fellowship for NLP
research.
Publications
A Multi-grained Dataset for News Event Triggered Knowledge Update
Yu-Ting Lee, Ying-Jhe Tang, Yu-Chung Cheng, Pai-Lin Chen, Tsai-Yen Li and Hen-Hsen Huang.
2022. In Proceedings of the 31st ACM International Conference on Information and Knowledge
Management (CIKM 2022). (Oral Presentation)
A Sentiment Analysis Method based on a Transformer and a Recurrent Neural Network
Yu-Ting Lee, Yen-Ching Chang. 2020. National Taiwan University Management Reviews: 1
st
Management and Medical Sciences Interdisciplinary Conference. (Poster Presentation)
Academic Services & Invited Talks
Reviewer, International Committee on Computational Linguistics (LREC-COLING), 2024
Invited Talk, Dense Information Retrieval with LLMs, Chung Shan Medical University, March 2024
Teaching and Outreach
Founder – PyTorch Taiwan, Facebook Public Group Jan 2024
•
Promoting implementations and SOTA models with PyTorch and PyTorch-based framework.
Contribute to Open-Source Project – Lit-LLaMA, Lightning AI Inc. June 2023
•
Lit-LLaMA: Independent implementation of LLaMA pretraining, finetuning, and inference code
that is fully open source. Fixed partial functions for fine-tuning LLaMA.
TA – Natural Language Processing — Dept. of CS, NCCU Feb 2023 – June 2023
Supervised by Hen-Hsen Huang Taipei City, Taiwan
•
Designed concepts explanations for coursework, assisted with homework and projects to support
70 students.
TA – Practical NLP — Taiwan AI Academy Aug 2022 – Sep 2022
Supervised by Hen-Hsen Huang New Taipei City, Taiwan
•
Designed introduction lectures and taught a two-week NLP project to support 20 intermediate
practitioners.
•
Taught Project: Summary Generation Tool for News during Important Time Durations.
Projects
Dynamic README Generation
Automatically update READMEs with each commit, highlighting new features and changes, saving
time and enhancing documentation accuracy and clarity. [GitHub]
Automated Generation and Evaluation of Research Methods in NLP
Developing an innovative system designed to automatically generate research methodologies based on
specified objectives and a review of related work in the field. [GitHub][Demo]
News Event Triggered Article Updating System
Developed a system utilizing advanced LLMs, focusing on full-text knowledge updating in response to
real-time news events, enhancing article accuracy. [GitHub][Demo]
Summary Generation Tool for News during Important Time Durations (SGIT)
Developed NLP tool for summarizing COVID-19 social media content, countering misinformation, and
improving public response through objective news analysis. [GitHub]
