self-note.md

In general, the work is about

• process 2500+ data for recognition (labelme)
• learn cs231n (computer vision)
• read papers
• attend meetings

how I schedule the time

• work 9 hours per day, using 1/3 *3 method
• from 8-12, 1-5 in institute or in university
• also need to find time for java and reading

what i have done

• technical thing
  • setup the learning environment using google colab and google drive
  • learn kNN for image classification
  • learn svm+ gradient code in thorough
  • use backpropagation to calculate the gradient
  • learn ResNet, RNN, CNN, and MLP
  • read code and better understand the workflow of github
• awareness
  • take advantage of resourcecs
  • adopt a balanced combination of learn and do
  • feel truly respectful to people who establish, polish, and pass on the teaching materials. cs231n website
• meeting

vecFusion 
img diffusion 做生成seq difussion 矢量化
S4 block
Mamba可选择性
笔端噪声的问题 
多模态大模型