How-to guides#
Task-oriented recipes for the BrainPy-style layer. Each guide is self-contained
and solves one concrete problem — pick the one that matches what you need right
now. The guides are organized into two tracks that mirror the dual audience of
brainpy.state.
For computational-neuroscience work: building and running biophysical networks, choosing models, shaping synaptic interactions, and reproducing published results.
For brain-inspired computing / SNN-ML work: making spiking networks differentiable, attaching readouts, and training through long rollouts without running out of memory.
The two tracks share the same models and the same transform primitives — the split is by task, not by a different library. A network you build for simulation can be trained, and a trained network can be simulated. The bridge between the two is explained in AlignPre / AlignPost — the keystone and Differentiability.
Simulation track
Training track