Module `Kaun`

module Activation : sig ... end

Activation functions for neural networks.

module Attention : sig ... end

Multi-head self-attention.

module Checkpoint : sig ... end

Model checkpointing.

module Context : sig ... end

Per-call auxiliary data for layers.

module Data : sig ... end

Lazy, composable data pipelines for training.

module Fn : sig ... end

Functional neural network operations.

module Grad : sig ... end

Automatic differentiation over parameter trees.

module Init : sig ... end

Weight initialization strategies.

module Layer : sig ... end

Composable neural network layers.

module Loss : sig ... end

Loss functions.

module Metric : sig ... end

Training metrics.

module Optim : sig ... end

Optimizers and learning-rate schedules.

module Ptree : sig ... end

Heterogeneous parameter trees.

module Train : sig ... end

High-level training loop.