Skip to content

Latest commit

 

History

History
46 lines (36 loc) · 1.86 KB

File metadata and controls

46 lines (36 loc) · 1.86 KB

Python examples

Demo scripts for the hypercube-worldmodel API. They call only the installed package.

Script What it shows
plane_point.py A point on a plane moved by a bounded velocity: encode, fit with a validation set, a held-out score against the identity guess, rollout, a Decoder beside the model, and a save and load round trip
plan_toy.py DMC protocol adapter (encode, encode_action, predict, rollout on raw actions, cost, action_space) driving CEM to a goal on the same toy world

How to run

These files live in the GitHub repository. They are not part of pip install; use the Quick start if you only want a snippet.

From a clone of HypercubeWorldModel (repository root), after installing the package:

pip install hypercube-worldmodel
# or from this tree:  pip install ./python
python python/examples/plane_point.py
python python/examples/plan_toy.py

What these are not

  • Not the C++ programs (the smoke test, quick_start, WorldModelTest, TerrainWalkerTest) or the study write-ups. Those live under tests/ and docs/.
  • Not automated tests. Package tests are tests/test_basic.py.
  • Not hard tasks. The toy world is easy onboarding so the API is obvious; do not cite its numbers as research results. The planner adapter is the shape to copy when the environment is a real suite.

Going further

Want See
Full Python API docs/Python_SDK.md
C++ product guide docs/CPP_SDK.md
Package readme README.md
The WorldModel class and the action path docs/world_model.md
The Decoder and its file format docs/decoder.md