Slacker News Slacker News logo featuring a lazy sloth with a folded newspaper hat
  • top
  • new
  • show
  • ask
  • jobs
Library
← Back to context

Comment by cztomsik

2 months ago

hm, residual is what I would not expect, can you elaborate why?

1 comment

cztomsik

Reply

simsla  2 months ago

Avoids vanishing gradients in deeper networks.

Also, most blocks with a residual approximate the identity function when initialised, so tend to be well behaved.

Slacker News

Product

  • API Reference
  • Hacker News RSS
  • Source on GitHub

Community

  • Support Ukraine
  • Equal Justice Initiative
  • GiveWell Charities