DELORES(XIAOMAN) DING
Cambridge,MA
Overview
4
4
years of professional experience
Work History
MIT Undergraduate Research
MIT CSAIL
Cambridge, MA
09.2024 - Current
- Study the representation learned by generative models.
- Work on improving generative modeling by exploring better generation order.
MIT Undergraduate Research
MIT Tegmark AI Safety Group
Cambridge, MA
02.2024 - 07.2024
- Study the science of modular addition from the lottery ticket perspective
- Understand training dynamics from toy examples
MIT Undergraduate Research
MIT CSAIL
Cambridge, MA
09.2023 - 12.2023
- Work on finding informative trajectories within the text embedding space in Stable Diffusion models
MIT Undergraduate Research
MIT Media Lab
Cambridge, MA
09.2022 - 06.2023
- Work on developing an interactive AI-generative music system
Asia Pacific Informatics Olympiad 2021 Lecturer
APIO 2021
Beijing
05.2021 - 05.2021
- Hold a lecture on block algorithms
Education
BS Computer Science -
Massachusetts Institute of Technology
Cambridge06-2026
High school diploma -
Nanjing Foreign Language School
06.2022
Skills
- Language: Chinese, English
- Programming language: Python, C
Awards
Gold Medal, China National Olympiad in Informatics, July 2021
- Ranked 22nd among all competitors (Top 50)
Gold Medal, China National Olympiad in Informatics, August 2020
- Ranked 35th among all competitors (Top 50)
- Best Female Competitor
Gold Medal, China National Olympiad in Informatics Winter Camp, August 2020
- Ranked 1st among all competitors (Top 47)
Publication
- Ding. et al. "Survival of the Fittest Representation: A Case Study with Modular Addition." ICML Workshop on Mechanistic Interpretability, 2024.
Projects
- Evolution of color terms in language via iterated learning Model the evolution of language (color terms) using iterated learning. Project involves model construction, analysis and real data collection.
- Pricing Model of Ride-Sharing Apps in Boston Area Understand the pricing of ride-sharing apps based on user inputs (locations, type of car requested) and external factors (time, weather, etc.). Statistical methods like regressions, significance tests and centrality measures are applied.
- Segmentation ControlNet for Stable Diffusion Train a ControlNet model from scratch to guide Stable Diffusion model with segmented conditioning image.
- Transformer-friendly Languages Investigate the characteristics of languages that are easily learnable by transformers. Conclusions supported by various toy model experiments.
Timeline
MIT Undergraduate Research
MIT CSAIL
09.2024 - Current
MIT Undergraduate Research
MIT Tegmark AI Safety Group
02.2024 - 07.2024
MIT Undergraduate Research
MIT CSAIL
09.2023 - 12.2023
MIT Undergraduate Research
MIT Media Lab
09.2022 - 06.2023
Asia Pacific Informatics Olympiad 2021 Lecturer
APIO 2021
05.2021 - 05.2021
BS Computer Science -
Massachusetts Institute of Technology
High school diploma -
Nanjing Foreign Language School
Similar Profiles
Mohamed TamerMohamed Tamer
Mechatronic Trainee at MIT Technologies - MIT Innovation Sdn BhdMechatronic Trainee at MIT Technologies - MIT Innovation Sdn Bhd
Gurunath WaghaleGurunath Waghale
Assistant Professor (Economics and Finance) at MIT School Of Engineering, MIT ADT UniversityAssistant Professor (Economics and Finance) at MIT School Of Engineering, MIT ADT University
AKSHAY INDAPURKARAKSHAY INDAPURKAR
Executive Data Analyst at MIT World Peace University (MIT-WPU)Executive Data Analyst at MIT World Peace University (MIT-WPU)
Nicole MeyersonNicole Meyerson
Senior Marketing Specialist at Cognex CorporationSenior Marketing Specialist at Cognex Corporation
Tiffany BeanTiffany Bean
HR Benefits Manager at TekSynapHR Benefits Manager at TekSynap