Skip to content
View andrewzhao551's full-sized avatar

Block or report andrewzhao551

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
andrewzhao551/README.md

Zhao Chenran

Undergraduate researcher studying optimization dynamics and numerical stability in large-scale machine learning systems.

The Chinese University of Hong Kong, Shenzhen


Research Interests

  • Training dynamics in large-scale models
  • Reduced-precision optimization
  • Gradient geometry and trajectory alignment
  • Stability analysis in efficient AI systems

Current Work

I am currently investigating how reduced numerical precision influences optimization trajectories in LLM training.

Rather than focusing solely on final performance metrics, my work examines:

  • Gradient deviation under precision perturbations
  • Directional drift in optimization updates
  • Structural sensitivity across model components (MoE vs Attention)

Selected repository:

🔗 Precision Effects on Training Dynamics


Technical Background

  • DeepSeek-MoE large-scale training
  • Multi-layer activation quantization experiments
  • Gradient-level diagnostic analysis
  • Optimization and mathematical modeling background

Academic Goal

I am actively seeking summer research opportunities in the United States to further develop my research experience in ML Systems and Efficient AI.

My long-term goal is to pursue a PhD in the U.S., focusing on training dynamics, numerical stability, and optimization behavior in large-scale machine learning systems.


Contact

Email: 123090854@link.cuhk.edu.cn

Popular repositories Loading

  1. precision-effects-on-training-dynamics precision-effects-on-training-dynamics Public

    Research investigation into training stability and optimization dynamics under reduced precision.

  2. andrewzhao551 andrewzhao551 Public

    self-introduction