Provider the idea of soft-score hierarchy for training;
KDD '23 paper: CodeGeeX: A Pre-Trained Model for Code Generation with Multilingual Benchmarking on HumanEval-X.
See CodeGeex Blog for more details.
Research in ByteDance AI Lab: The VolcTrans System for WMT22 Multilingual Machine Translation Task (WMT22)
Research in Seed Team: MARS: Unleashing the Power of Variance Reduction for Training Large Models
Project 1: A novel retrosynthesis prediction architecture integrating textual and graphic information
Project 2 (Accepted by ISMB'24): A multi-view diffusion model for enhancing Hi-C contact matrices. Both in Prof. Sheng Wang's lab.
Name Liu Yifeng(Lewis Yik-fung Lau)
From Fuling, Chongqing, China
Major Computer Science and Engineering
Phone +86 150 2393 8602
I can program with C++, Python, MatLab as well as HTML5. And I can use LaTeX for text generation.
For Deep Learning fields such as NLP and AI4Science, I am proficient in modules such as PyTorch, Numpy and Transformers.
I have learned some about linguistics. Moreover, I am good at Chinese and English, and I can recognize a little of Japanese.
I am currently a Ph.D. student at UCLA, researching on optimization and architectures of LLMs. I was previously a Yao Class Student from Tsinghua University researching on the fields of NLP following Zhilin Yang (Now CEO of Moonshot Inc.) and AI4Science following Sheng Wang as a visiting student in the University of Washington.
I can program well with C++, Python, MatLab, HTML5 as well as LaTeX. And I have participated in lots of projects of AI, especially in the field of NLP.
I has won the gold medal in the 36th CPhO and obtained the Outstanding Award of Interdisciplinary Contest In Modeling(ICM). Also, I has got the comprehensive award of Tsinghua University twice.
I am responsible, for which I was once elected as the monitor of Yao Class. And I am also sunny, optimistic as well as approachable.