Quickstart
This quickstart validates a fresh checkout on the smallest bundled fixture.
1. Activate the Environment
conda activate isograph
2. Run a Toy Benchmark
The default backend is vae. Run a quick smoke test on the smallest bundled fixture:
isograph benchmark -- \
fixture_filter=toy_v1 \
stage_name=quickstart_vae
To use the deterministic baseline instead:
isograph benchmark -- \
backend=baseline \
fixture_filter=toy_v1 \
stage_name=quickstart_baseline
Expected outputs (paths use the stage_name you chose):
artifacts/benchmarks/<stage_name>/toy_v1/artifacts/reports/<stage_name>-benchmark.jsonartifacts/reports/<stage_name>-runtime-memory.json
3. Inspect the Result
The benchmark report is a JSON summary with per-dataset metrics such as:
number of discovered modules
number of inferred edges
recovery score when fixture truth is available
runtime in seconds
4. Next Steps
For command-line reference, continue to CLI Reference.
For bringing your own dataset into IsoGraph, continue to Own-Data Data Model.
For tutorial-style walkthroughs, use the GitHub Wiki.