Documentation src/dl

This deep learning frontend is graph based. Build a model with

GraphModel::builder()

or load an ONNX model, then train it with

Trainer

and a

DataLoader

. Layers are created via the

FlintDL

helper functions.

Start here for model construction, ONNX load/serialize, data loading and training configuration.

Use the layer graph nodes and factory helpers to build model topology.

Implement your own optimizer and loss functions or use the provided ones.