Published on

GPT : Beams, Sampling & more


Sam Witteveen and I started the TensorFlow and Deep Learning Singapore group on MeetUp in February 2017, and the thirty-fifth MeetUp was the second under its new name "Machine Learning Singapore" (or MLSG for short).

Our presentations "Beams and GANs" was in two parts:

  • "The State of StyleGANs" - Sam : Recently there have been multiple developments in GANs and specifically StyleGANs. In this talk Sam will go through what is new with the new StyleGAN 3 architecture and also a number of other papers and techniques that use StyleGAN models and techniques to manipulate and generate new images.

  • "GPT : Beams, Sampling & more" - Martin : If you're interested in generating text (from, for example, GPT-Neo) then just targeting 'the best' output is not as easy as you might imagine. Beam Search techniques can help, and in this talk Martin will explain the various options available, as well as other techniques for 'guided generation'.

The slides for my talk are here :

Presentation Screenshot

If there are any questions about the presentation please ask below, or contact me using the details given on the slides themselves.

Presentation Content Example