Latest commit c6db878 Nov 25, 2016 @kevinthesun kevinthesun committed with piiswrong New callback interface for training visualization. (#3849)
* Update submodule

* New callback interface for training visualization.

Neural style transfer notebook example of (refactored) live notebook charting.

Trivial notebook to run train_cifar10 working.

notebook structure in place for live charting.
Needs module cb change.

Module callbacks almost working.  Epoch numbers NaN

Module / CNN example for notebook charts.

CIFAR10 training notebook for module works again.

Matching callback code to notebook.

Add new notebook

Small fix

Change notebook name

Clean merge

Small bug fix

Import cleaning

Update with master

Point submodule to newest

Clean import statement for notebook

Fix callback function error

Seperate doc

Fix pylint

More cleaning

README.md

Neural art

This is an implementation of the paper A Neural Algorithm of Artistic Style by Leon A. Gatys, Alexander S. Ecker, and Matthias Bethge.

How to use

First use download.sh to download pre-trained model and sample inputs

Then run python nstyle.py, use -h to see more options

Sample results

It takes 30 secs for a Titan X to generate the above 600x400 image.

Note

  • The current implementation is based the torch implementation. But we may change it dramatically in the near future.

  • We will release multi-GPU version soon.