How-to guides

Use these guides when you already know what you want to accomplish. Each route focuses on a concrete modeling, inference, analysis, or extension task.

Build and parameterize a model

• Specify priors and fix parameters.
• Use stimulus coding.
• Add smooth effects with hsgp().
• Model outliers with lapse probabilities or regress on p_outlier.

Choose and run an inference method

• Set initial values.
• Use an alternative sampler with Bayeux.
• Run variational inference.

Work with fitted models

• Plot posteriors and predictions, then consult the posterior-predictive or model-cartoon gallery.
• Compare and interpret models.
• Extract trial-wise parameters.
• Run a Bayesian t-test on posterior draws.
• Simulate interventions with the do-operator.
• Save and load fitted models.

Extend HSSM

Begin with the custom-likelihood route table. For ONNX, read the ONNX likelihood contract before using the sbi, BayesFlow, JAX-callable, or black-box walkthrough linked there.

For a contribution to HSSM itself, follow the local development setup and then the contribution guide.

For concepts rather than procedures, return to the learning index. For exact APIs, use the reference index.