Skip to content
View zsyverse's full-sized avatar

Highlights

  • Pro

Block or report zsyverse

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
zsyverse/README.md

zsyverse

Undergraduate at Zhejiang University · College of Electrical Engineering

Working toward research in vision-language models and multimodal agents.

Homepage ZJU Focus


About

I am an undergraduate at Zhejiang University, studying in the College of Electrical Engineering. Most of what I build sits at the boundary between perception and reasoning — systems that take in images, video, or documents, and have to decide what to do with them rather than just describe them.

Day to day that means training and profiling multimodal models, building retrieval and agent pipelines, and writing the tooling that keeps those experiments reproducible. A fair amount of what I publish here is the tooling half: templates, generators, and study material that other people can pick up directly.

Research Interests

Direction What I care about
Vision-Language Models How visual tokens get compressed and selected under a fixed context budget — long-video understanding is where this hurts most, and where evaluation is easiest to fool.
Agents Multi-step tool use that stays grounded: planning, retrieval, and verification loops that catch their own mistakes instead of confidently continuing.
World Models Learned dynamics as a substrate for planning — what a model has to represent before "imagining the next state" becomes useful rather than decorative.
Generative Models Diffusion and autoregressive generation, especially controllability and how generated data feeds back into training.

Selected Projects

54 one-page infographic cards covering the LeetCode Hot 100 and the 代码随想录 roadmap — one card per algorithmic pattern, each with the definition, recognition signals, a worked diagram, C++ / Python templates, and a mnemonic. Ships as two ready-to-present slide decks plus the full generation scripts, so the whole set can be re-rendered or exported.

Python · Infographics · Interview prep

The Zhejiang University LaTeX Beamer template, packaged as a skill an AI agent can invoke end to end. One sentence in, and the agent copies the resources, writes the .tex, compiles with XeLaTeX, and opens the PDF — official ZJU blue, 16:9, CJK mixed typesetting, and the usual academic components already wired up.

LaTeX · Beamer · AI Agent Skill

Personal research page — interests, notes, and whatever is currently in progress.

Toolbox

Python PyTorch C++ TypeScript LaTeX Linux Git


Followers algorithm-cards zju-beamer-skill

zsyverse's contribution graph

Reach me through my homepage · Hangzhou, China

Popular repositories Loading

  1. algorithm-cards algorithm-cards Public

    🎴 54 张算法方法图解卡 · 2 套方法卡 PPT:LeetCode Hot100 与代码随想录路线,一题型一张(定义·图解·步骤·C++/Python 模板·口诀),面试复习开箱即用

    Python 2

  2. zju-beamer-skill zju-beamer-skill Public

    浙江大学 LaTeX Beamer 模板,封装为可被 AI Agent 调用的自动化 Skill

    TeX 1

  3. zsyverse zsyverse Public

    Academic profile of zsyverse