Zekai Zhang

Zekai Zhang

PhD student at University of Michigan EECS

Posts by Tags

category1

Future Note

less than 1 minute read

Published:

This post will show up by default. To disable scheduling of future posts, edit _config.yml and set future: false.

Note number 4

less than 1 minute read

Published:

This is a sample note. Lorem ipsum: I can’t remember the rest of lorem ipsum and don’t have an internet connection right now, so this placeholder text will do. Testing, testing, testing this note. Notes are cool.

Note number 3

less than 1 minute read

Published:

This is a sample note. I cannot remember the rest of lorem ipsum and do not have an internet connection right now, so this placeholder text will do. Testing, testing, testing this note. Notes are cool.

Note number 2

less than 1 minute read

Published:

This is a sample note. Lorem ipsum: I can’t remember the rest of lorem ipsum and don’t have an internet connection right now, so this placeholder text will do. Testing, testing, testing this note. Notes are cool.

Note number 1

less than 1 minute read

Published:

This is a sample note. Lorem ipsum: I can’t remember the rest of lorem ipsum and don’t have an internet connection right now, so this placeholder text will do. Testing, testing, testing this note. Notes are cool.

category2

Future Note

less than 1 minute read

Published:

This post will show up by default. To disable scheduling of future posts, edit _config.yml and set future: false.

Note number 4

less than 1 minute read

Published:

This is a sample note. Lorem ipsum: I can’t remember the rest of lorem ipsum and don’t have an internet connection right now, so this placeholder text will do. Testing, testing, testing this note. Notes are cool.

Note number 3

less than 1 minute read

Published:

This is a sample note. I cannot remember the rest of lorem ipsum and do not have an internet connection right now, so this placeholder text will do. Testing, testing, testing this note. Notes are cool.

Note number 2

less than 1 minute read

Published:

This is a sample note. Lorem ipsum: I can’t remember the rest of lorem ipsum and don’t have an internet connection right now, so this placeholder text will do. Testing, testing, testing this note. Notes are cool.

Note number 1

less than 1 minute read

Published:

This is a sample note. Lorem ipsum: I can’t remember the rest of lorem ipsum and don’t have an internet connection right now, so this placeholder text will do. Testing, testing, testing this note. Notes are cool.

cool posts

Future Note

less than 1 minute read

Published:

This post will show up by default. To disable scheduling of future posts, edit _config.yml and set future: false.

Note number 4

less than 1 minute read

Published:

This is a sample note. Lorem ipsum: I can’t remember the rest of lorem ipsum and don’t have an internet connection right now, so this placeholder text will do. Testing, testing, testing this note. Notes are cool.

Note number 3

less than 1 minute read

Published:

This is a sample note. I cannot remember the rest of lorem ipsum and do not have an internet connection right now, so this placeholder text will do. Testing, testing, testing this note. Notes are cool.

Note number 2

less than 1 minute read

Published:

This is a sample note. Lorem ipsum: I can’t remember the rest of lorem ipsum and don’t have an internet connection right now, so this placeholder text will do. Testing, testing, testing this note. Notes are cool.

Note number 1

less than 1 minute read

Published:

This is a sample note. Lorem ipsum: I can’t remember the rest of lorem ipsum and don’t have an internet connection right now, so this placeholder text will do. Testing, testing, testing this note. Notes are cool.

notes

Learning discrete diffusion – MH sampling

4 minute read

Published:

Metropolis–Hastings (MH) is a Markov chain Monte Carlo method that repeatedly draws proposals from a simple distribution and updates the current state with an accept/reject rule.

Thoughts on Generalization in Diffusion Models

less than 1 minute read

Published:

Generalization is the implicit alignment between neural networks and the underlying data distribution, shaped by both data and human perception.