API Reference

mlx-snn's API is organized into the following modules:

Module	Description
Neurons	Spiking neuron models (LIF, IF, Izhikevich, ALIF, Synaptic, Alpha, RLeaky, RSynaptic)
Surrogate Gradients	Differentiable surrogate gradient functions for spike generation
Encoding	Spike encoding methods (rate, latency, delta, direct, repeat, EEG)
Functional	Stateless pure functions — neuron dynamics, spike ops, loss functions
Layers	Composite spiking layers (conv, pooling, dropout, flatten)
Operators	Mathematically-principled Conv SNN optimization operators
Training	BPTT helpers and training utilities
Datasets	Neuromorphic dataset loaders (DVS-Gesture, CIFAR10-DVS, N-MNIST, SHD)
Utilities	State management, visualization, metrics
NIR Interop	Cross-framework model exchange via NIR

Import Conventions

import mlxsnn                    # Main package (most classes/functions)
from mlxsnn.neurons import Leaky  # Direct submodule import
from mlxsnn.operators import TemporalAggregatedConv  # Operators

All neuron models, encoding functions, loss functions, and layers are re-exported from the top-level mlxsnn namespace.