Skip to content

Comments

WIP: Minimal working 1D gaussian and MNIST tutorials#7

Open
NolanSmyth wants to merge 4 commits intoAlexandreAdam:masterfrom
NolanSmyth:tutorials
Open

WIP: Minimal working 1D gaussian and MNIST tutorials#7
NolanSmyth wants to merge 4 commits intoAlexandreAdam:masterfrom
NolanSmyth:tutorials

Conversation

@NolanSmyth
Copy link

Added two simple examples/tutorials at the level of someone just learning the code

1D Gaussian with 0 mean and unit variance

  • MLP architecture
  • Can run on my CPU in ~ 1 min

MNIST

  • NCSNpp architecture
  • Can run on colab T4 GPU in ~45 minutes
  • Generates samples that are recognizably numeric, but can certainly be improved.
  • using preprocessing_fn to transform grayscale images from [0, 255] [0, 1] resulted in nonsense samples so I omitted this until I figure out what I'm doing wrong

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant