API Reference

The full Fortran API documentation is generated using FORD (Fortran Automatic Documentation).

You can browse the complete API documentation here:

The API documentation includes:

  • Module Reference: Detailed documentation of all modules

  • Type Definitions: Documentation of derived types like base_layer_type

  • Procedures: All public procedures and their interfaces

  • Source Code: Annotated source code with cross-references

  • Call Graphs: Visual representation of procedure dependencies

Key Components

The following are some of the key derived types and their main methods documented in the API:

  • base_layer_type: The abstract base type for all layers

    • forward(): Method to perform forward propagation

  • base_actv_type: The abstract base type for all activation functions

    • apply(): Method to apply the activation function

  • base_init_type: The abstract base type for all initialisers

    • initialise(): Method to initialise parameters

  • base_optimiser_type: The base type for all optimisers

    • minimise(): Method to update parameters based on gradients

  • base_loss_type: The base type for all loss functions

    • compute(): Method to compute the loss value

  • network_type: The type representing a neural network

    • add(): Method to add layers to the network

    • compile(): Method to compile the network for training

    • set_training_mode(): Method to switch training-sensitive layers to training mode

    • set_inference_mode(): Method to switch training-sensitive layers to inference mode

    • train(): Method to train the network

    • test(): Method to evaluate the network on test data

    • predict(): Method to make predictions with the network

    • update(): Method to update learnable parameters

For complete details and source code, please refer to the FORD documentation linked above.