Code Monkey home page Code Monkey logo

Comments (2)

phillipi avatar phillipi commented on May 22, 2024

Yeah, setting the net to evaluate mode does two things: 1) enables dropout, 2) makes batchnorm use aggregated training set statistics. I'm not entirely sure why it results in black images, but I think it's more due to 1 than to 2 (or possibly the interaction of the two effects).

The goal of enabling dropout during testing was to make the outputs stochastic (though it only has a very minor effect). Also, our strategy was to run the net identically during training and test, to avoid any "domain gap" between the two phases. This makes a big difference for batchnorm. It could be also that the way we used dropout causes a problematic gap if you turn it off at test time.

from pix2pix.

Quasimondo avatar Quasimondo commented on May 22, 2024

Maybe worth mentioning: if you are creating animations or videos with pix2pix I recommend to un-comment that line and thus disable dropout in test mode. This will give you much better frame-to-frame consistency and reduce flickering a lot.

from pix2pix.

Related Issues (20)

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.