Simulusion documentation
Simulusion is a PyTorch library for world models, and a platform to run them on your own machines. Dreamer, PlaNet, DIAMOND, IRIS, Genie, DiT and I-JEPA share one API, and their encoders, dynamics backbones and decoders are reusable parts.
- Getting startedInstall, train a first world model and watch what it learned.
- AlgorithmsTheory and usage for each model family, with demos.
- PlatformConnect a GPU machine, run experiments and evals.
- Command linesimulusion train, play, eval and benchmark.
Install
Terminal
pip install "simulusion[gym]"Train an agent
python
import simulusion
# Dreamer from pixels on Pendulum
agent = simulusion.create_model(
"dreamer", env="Pendulum-v1", env_backend="gym", total_steps=5_000
)
agent.train()
print(simulusion.list_models()) # every model create_model() can buildContinue with Getting started for a guided first run, or pick a model under Algorithms.
Moved from GitHub Pages
These docs used to live at paramthakkar123.github.io/simulusion. Old links redirect here.