Please enable JavaScript.

Coggle requires JavaScript to display documents.

Gen AI - Coggle Diagram

- - - - transfỏm in4 into a compact form-> latent space-> rebuild
      - generate content, images, text, music
  - - - text, code, music
  - - - craff vivid images, expertise to audio, video
- - - - discerning between reality and illusion
  - - - peeling away noise from data-> bloom to realism
- - - - visual virtuoso
        
        creating images
    - - linguistic maestro
        
        no color-> vibrant picture
        
        translate languages
    - - tailor creations based on additional in4
    - - generator & discriminator
    - - human faces
    - - big images
  - - - art critics/ discriminator are too discerning
      - gradient vanishing ( discriminator too sharp/ generator overly complex
    - - Wasserstein GANs (WGANs
        
        different technique that's less prone to mode collapse
      - Unrolled GANs (UGANs
        
        smarter algorithm ->prevent gradient vanishing.
      - Spectral Normalization (SN)
        
        reventing both mode collapse & gradient vanishing.
      - Gradient Penalty
        
        encourage the artist to explore more colors.