Shaomu Tan

Research on Machine Translation ❙ Multilingual NLP ❙ Deep Learning.
I am currently a PhD candidate at the Language Technology Lab, University of Amsterdam. I work with Christof Monz and my amazing colleagues at LTL. Before my PhD, I obtained an M.S. degree in Artificial Intelligence at Utrecht University, Netherlands, and a bachelor’s degree in Information Science at Shandong University.
I’m on a quest to make Machine Translation and LLMs more effective, breaking down language barriers and connecting people around the world! My recent research focuses on understanding and improving Multilingual, Multi-Task Learning, Translation Evaluation, and Multilingual Reasoning. I am always open to collaborations, paper discussions, shared tasks and competitions, feel free to reach out!
news
Apr 18, 2025 | New paper on Machine Translation Evaluation: “Remedy: Learning Machine Translation Evaluation from Human Preferences with Reward Modeling”, achieving new SOTA on WMT22-24. |
---|---|
Mar 11, 2025 | We obtained Silver Medal (Ranked 19/950 teams) on the Kaggle 2025 WSDM Cup - Multilingual Chatbot Arena, a Reward Modeling Competition to predict the human preference of Multilingual LLMs. |
Sep 20, 2024 | Paper “Neuron Specialization: Leveraging Intrinsic Task Modularity for Multilingual Machine Translation” accepted to the EMNLP 2024 Main Conference. |
May 16, 2024 | Paper “How Far can 100 Samples Go? Unlocking Zero-Shot Translation with Tiny Multi-Parallel Data” accepted to the ACL Findings 2024 Conference. |
Nov 07, 2023 | Paper “Towards a Better Understanding of Variations in Zero-Shot Neural Machine Translation Performance” accepted to the EMNLP 2023 Main Conference. |
Oct 08, 2023 | Our system obtained the First Place on the WMT23 Translation Competition on the English-Hebrew and Hebrew-English Constraint Tracks. |
Sep 21, 2023 | Paper “Make Pre-trained Model Reversible: From Parameter to Memory Efficient Fine-Tuning” accepted to the NeurIPS 2023 Conference. |
Aug 25, 2023 | We obtained the Gold Medal on the Kaggle 2023 Google’s American Sign Language Fingerspelling Recognition, ranked 11/1,315 teams. |