Hatim El JazouliCatalog
R101

Context Tree Switching implementation

Formal methods. — 2025-10-13. — 1 tracing.

An implementation of the Context Tree Switching algorithm, written from scratch in Python for sequential-prediction research.

Context Tree Switching (CTS; Veness et al., 2012) is an algorithm for sequential prediction, and this project is an implementation of it written from scratch in Python.

It was written for sequential-prediction research, as part of work on AIXI.

Built with: Python.

Code: github.com/Enhso/cts

Tracings

Drawn from this record.

  1. F101Ensemble AI forecasting botForecasting

    Both combine many predictors instead of trusting one. CTS mixes context models as data arrives; the bot combines forecasts from different personas.

Traced from

Records that point here.

  1. N105Why I’m interested in AI safetyNotes

    AIXI is a mathematical model of an ideal agent. Implementing one of its building blocks was my way into the theory behind the systems this essay worries about.