GitHub

Networks

Each network corresponds to an array of layers associated with a unique string identifier. The string identifier of the network is used to store cached models, so it's important to ensure that you don't re-use names!

Public Interface

MIPVerify.Sequential — Type
struct Sequential <: NeuralNet

Represents a sequential (feed-forward) neural net, with layers ordered from input to output.

Fields:

  • layers

  • UUID

source
MIPVerify.SkipSequential — Type
struct SkipSequential <: NeuralNet

Represents a sequential (feed-forward) neural net, with layers ordered from input to output. Unlike a regular Sequential network, this network type supports SkipBlocks, which can take input from multiple previous layers. When a SkipBlock is encountered, it receives an array of outputs from preceding layers, allowing for skip connections and residual architectures.

Fields:

  • layers

  • UUID

source

Internal