ALPHA-FACTORY / MUZERO PLANNING LAB

Learn a model.
Inspect a decision.

Train a small neural network from experience. See where search spends its effort. Compare the result with its untrained self.

CPU learning · No API key · Reproducible experiments

Start the local lab →Run in Colab ↗

  1. Choose MiniChoice. Take 0.3 now, or discover a two-step route to 1.0.
  2. Train & compare. Watch observed rewards and three neural training losses.
  3. Inspect the search. Read action priors, visits, predicted rewards and discounted values.

This page introduces the locally executed lab. Neural training runs in Python, not on this static page. CartPole and other harder tasks remain research experiments; improvement is not guaranteed. This is not achieved AGI.

Preserved original research presentation ↗

Original sample replay

The chart below replays bundled sample data; it does not run the trained model.

🌟 **Mastery Without a Rule‑Book** — watch MuZero think in real time

🌟 Mastery Without a Rule‑Book — watch MuZero think in real time

Detailed instructions

See docs/DISCLAIMER_SNIPPET.md This repository is a conceptual research prototype. References to "AGI" and "superintelligence" describe aspirational goals and do not indicate the presence of a real general intelligence. Use at your own risk. Nothing herein constitutes financial advice. MontrealAI and the maintainers accept no liability for losses incurred from using this software.






⬅️ Back to Gallery