WOLFRAM

Wolfram Language & System Documentation Center

WLNet (.wlnet)

Background & Context

    • Used for storing trained and untrained neural nets.
    • Binary format.
    • Developed at Wolfram Research.

Import & Export

Import Elements

  • General Import elements:
  • "Elements" list of elements and options available in this file
    "Summary"summary of the file
    "Rules"list of rules for all available elements
  • Import elements include:
  • "Net"the net, including all initialized arrays
    "UninitializedNet"the net, without any arrays
    "ArrayList"parameter arrays as a list
    "ArrayAssociation"parameter arrays as an association
    "WLVersion"version of the Wolfram Language used to create the file

Examples

Basic Examples  (2)

Define a NetChain:

Wolfram Language code: net = NetChain[{LinearLayer[100], LinearLayer[10]}]

Export this net to the WLNet format:

Wolfram Language code: Export["testnet.wlnet", net]

Import the net:

Wolfram Language code: Import["testnet.wlnet"]

Define an initialized NetGraph:

Wolfram Language code: net = NetInitialize@NetGraph[{LinearLayer[10], LinearLayer[3]}, {1 -> 2}, "Input" -> {2}]

Export the net along with its initialized parameters:

Wolfram Language code: Export["testnet.wlnet", net]

Import the network:

Wolfram Language code: net2 = Import["testnet.wlnet"]

Evaluate the net on input data:

Wolfram Language code: net2[{1, 2}]

This is equivalent to evaluating the original net:

Wolfram Language code: net[{1, 2}]

Import a version of the net without any initialized arrays:

Wolfram Language code: net2 = Import["testnet.wlnet", "UninitializedNet"]

Import an association of arrays:

Wolfram Language code: Import["testnet.wlnet", "ArrayAssociation"]