API Reference

This is the API reference for OpenRetina. It documents all public classes, functions, and modules available in the package.

How to use this reference:

• Browse by module: Navigate through the sections below to find specific components
• Search: Use the search function to find specific classes or functions
• Link to source: Most entries link directly to the source code

If you're new to OpenRetina, start with the Quick Start or Tutorials before diving into the API reference.

Modules

Models

Complete model architectures for retinal response prediction:

• BaseCoreReadout — Base class for all core-readout models
• UnifiedCoreReadout — Hydra-configurable model (recommended)
• load_core_readout_from_remote — Load pre-trained models
• Linear-Nonlinear Models — LNP cascade models
• Sparse Autoencoder — Sparse representation models

Modules

Building blocks for constructing models:

• Core Modules — Convolutional feature extractors (SimpleCoreWrapper, ConvGRUCore)
• Readout Modules — Spatial readouts (PointGaussianReadout, GaussianMaskReadout, multi-session wrappers)
• Layers — Convolutions, regularizers, scaling, GRU cells
• Loss Functions — Poisson, correlation, and MSE losses

Data I/O

Data loading and preprocessing:

• Base Data Classes — MoviesTrainTestSplit, ResponsesTrainTestSplit
• Base Dataloader — MovieDataSet, MovieSampler, dataloader factories
• Dataset implementations: Hoefling 2024, Karamanlis 2024, Goldin 2022, Maheswaranathan 2023, Sridhar 2025
• Artificial Stimuli, Cyclers

In-silico

Tools for computational experiments with trained models:

• Stimulus Optimization — MEIs, discriminatory stimuli, regularizers
• Vector Field Analysis — PCA-based response analysis
• Tuning Analyses — Gradient-based response characterization

Evaluation

Metrics and oracle computations for model evaluation:

• Correlation, Poisson loss, MSE, FEVe, variance analysis
• Oracle correlations (jackknife, global mean)

CLI

Command-line entry points:

• train_model — Full training pipeline
• evaluate_model — Full evaluation pipeline

Utilities

Helper functions for file handling, visualization, model management, and data processing.