Tutorials

One runnable script per task, in examples/.

task	what it shows
Regression	Fit a smooth function of two variables, adapt the grids to the data and refine them.
Formula discovery	Train with the sparsity penalty, prune, make every edge symbolic and read the formula (pykan's workflow).
PDEs with PINNs	Solve a Poisson equation with the PDE residual in the loss; derivative keeps the graph for second derivatives.
Operator learning	A DeepONet with KAN branch and trunk learns the antiderivative operator.
Images	Classify 8x8 digits with KAN convolutions.
Transformers	Put KAN layers into a Hugging Face GPT-2 without changing its outputs, then fine-tune only the splines.
Graphs	Node classification with KAN message passing on edge_index graphs.
Time series	One-step forecasting with an LSTM whose gates are KAN layers.
Neural ODEs	Learn the dynamics of a damped oscillator with dy/dt = KAN(y) and a differentiable RK4.
Reinforcement learning	REINFORCE with a KAN policy on a small continuous control task (no gym needed).
Deployment	Lookup tables, int8, ONNX in onnxruntime and Verilog for an FPGA from one trained model.
Backends	Convert a model between PyTorch, MLX and JAX and through one file format.