Crate candle_optimisers

source ·
Expand description

Optimisers for use with the candle framework for lightweight machine learning. Apart from LBFGS, these all implement the candle_nn::optim::Optimizer trait from candle-nn

Modules§

  • adadelta
    Adadelta optimiser
  • adagrad
    Adagrad optimiser
  • adam
    Adam optimiser (inlcuding AdamW)
  • adamax
    Adamax optimiser
  • esgd
    Stochastic Gradient Descent
  • lbfgs
    Limited memory Broyden–Fletcher–Goldfarb–Shanno algorithm
  • nadam
    NAdam optimiser: Adam with Nesterov momentum
  • radam
    RAdam optimiser
  • rmsprop
    RMS prop algorithm

Enums§

  • Decay
    Method of weight decay to use
  • ModelOutcome
    Outcomes of an optimiser step for methods such as LBFGS
  • Momentum
    Type of momentum to use

Traits§

  • LossOptimizer
    trait for optimisers like LBFGS that need the ability to calculate the loss and its gradient
  • Model
    Trait for Models: this is needed for optimisers that require the ability to calculate the loss such as LBFGS
  • OptimParams
    Trait for optimisers to expose their parameters