Skip to content

vedprakashnautiyal/Imagica

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

7 Commits
 
 
 
 
 
 

Repository files navigation

Imagica

A image generation model from scratch (Due to inavailability of computer units, the model may not get trained efficiently, but research suggest that around 1000 epochs could generate better result)

Resources

  • This great video by Outlier gives a great indepth explaination of the math behind Diffusion Model as it went through different approaches and paper implementations.

Future Improvement Scope

  • Residual connections
  • Different activation functions like SiLU, GWLU, ...
  • BatchNormalization
  • GroupNormalization
  • Attention Layers

Releases

No releases published

Packages

No packages published