Hugging Face's logo Hugging Face
  • Models
  • Datasets
  • Spaces
  • Buckets new
  • Docs
  • Enterprise
  • Pricing
    • Website
      • Tasks
      • HuggingChat
      • Collections
      • Languages
      • Organizations
    • Community
      • Blog
      • Posts
      • Daily Papers
      • Hardware
      • Learn
      • Discord
      • Forum
      • GitHub
    • Solutions
      • Team & Enterprise
      • Hugging Face PRO
      • Enterprise Support
      • Inference Providers
      • Inference Endpoints
      • Storage Buckets

  • Log In
  • Sign Up

Spaces:
byte-vortex
/
repro-diffusion-consistency
Running

App Files Files Community
repro-diffusion-consistency / pages /index.md
byte-vortex's picture
byte-vortex
Update logbook: Reproduction: A Random Matrix Theory Perspective on the Consistency of Diffusion Models
4e26ad2 verified 12 days ago
preview code
|
Raw
History Blame Contribute Delete
941 Bytes

Reproduction: A Random Matrix Theory Perspective on the Consistency of Diffusion Models

Pages

Page
Executive summary
Claim 1: Linear model predicts cross-split consistency
Claim 2: Self-consistency equation and renormalized noise
Claim 3: Variance factorization (anisotropy x inhomogeneity)
Claim 4: Fractional matrix power extension to sampling trajectories
Claim 5: Deep network validation (toy scale)
Claim 6: Deep network validation (scoped toy scale, CIFAR10 only)
Conclusion