Core-Readout Models

The Models overview documents the public core-readout model classes and the helpers for loading pretrained checkpoints.

The components used to assemble these models are documented separately:

• Core modules extract shared spatiotemporal features.
• Readout modules map those features to neuron responses for each recording session.
• Custom layers provide convolutions, regularisers, scaling, and recurrent building blocks.
• Loss functions provide the training and evaluation objectives.