Skip to content

zbyte64/pytorch-dagsearch

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

48 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

ENAS: https://arxiv.org/pdf/1802.03268.pdf PNN: https://arxiv.org/pdf/1706.03256.pdf

Graph would be a nn.Sequential model that allows for skip connections to be registered

g = Graph(
  Conv2D(...)
  MaxPool2D(...)
  Conv2D(...)
  )

#Generates a connector for different layer sizes
g.add_link(g.layers[0], g.layers[2])
#Mutes an input, useful for ENAS
g.toggle_link(g.layers[0], g.layers[2])

Graphs are also expandable, new nodes are added in front.

Node is a special layer that allows for on-the-fly hyperparam search. Unlike a layer, it specifies it's in and out dim size and initializes candidate cells to that size. Connector is responsible for reshaping and learning identity mappings across layers. World implements the game logic and actions available to the ENAS agent. DagSearchEnv is the Gym environment.

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages