Quickstart
This page shows how to set up the project locally and run a complete pass from map preview to analysis. The steps below assume a Linux environment; the CI uses Python 3.11/3.12 and a nightly Rust toolchain.
Prerequisites. Install a recent Python (3.10+) and the Rust toolchain with rustup
on nightly. On Ubuntu, you may also need system libraries for font rendering, as shown in the CI workflows.
Install. From the repository root, run ./setup.sh
to create a virtual environment, build the Rust extension, and install Python dependencies. The script places the virtualenv under ./env/
.
Minimal demo. With the environment set up, the following commands exercise the main functionality:
./env/bin/python main.py preview-maps
./env/bin/python main.py mcts
./env/bin/python main.py score-prompts -i 0 -m llama3.1:8b
./env/bin/python main.py benchmark-llm -i 0 -m deepseek-r1:7b
./env/bin/python main.py analysis
Outputs. All artifacts are written to outputs/
. Map previews create per‑map folders with map.png
and abstraction.png
. Scoring and benchmarking write JSON clusterings, CSV logs, and MCTS plots to outputs/llm_scoring/
. The analysis step aggregates results under outputs/analysis/
. See Repo → Outputs & Artifacts for details.
Configuration. Experiment settings live in config.yml
, and prompt fragments live in config_prompts.yml
. See Repo → Configuration for the schema and examples.