Kinetix

Normalizer

Kinetix — metrics, requirements, and run commands.

Edit this page GitHub Markdown

Last updated: Jul 22, 2026

On this page

About Metrics Environment setup Run Evaluation Requirements

About

Kinetix external benchmark manifest for JAX physics embodied action evaluation.

Metrics

Kinetix external benchmark manifest for JAX physics embodied action evaluation.

Task YAML

worldfoundry/data/benchmarks/tasks/external/kinetix.yaml

Primary

success_rate

Metric	Focus
`success_rate`	Share of episodes or tasks that complete successfully.
`task_success`	Share of episodes or tasks that complete successfully.
`episode_success`	Share of episodes or tasks that complete successfully.
`reward`	Cumulative RL reward on Kinetix procedurally generated 2D rigid-body control puzzles.

Environment setup

Runtime environment

Conda env: kinetix-jax.
Separate setup: this benchmark expects a dedicated conda env rather than the unified runtime.
Keep Kinetix in a JAX-compatible simulator environment when the unified PyTorch runtime is insufficient.

Evaluation data

Candidate model outputs: set WORLDFOUNDRY_KINETIX_RESULTS_PATH to the generated-video or rollout artifact root.
${WORLDFOUNDRY_DATA_DIR}/datasets/kinetix
${WORLDFOUNDRY_CACHE_DIR}/repos/FLAIROx--Kinetix

Checkpoints & assets

This benchmark is primarily rollout- or official-result-driven; see catalog requires and the task YAML for policy/simulator checkpoints.

Key environment variables

WORLDFOUNDRY_KINETIX_ROOT
WORLDFOUNDRY_KINETIX_RESULTS_PATH

Run Evaluation

After preparing the assets and candidate artifacts listed on this page, use the public run entry points below.

Set Candidate Artifacts

export WORLDFOUNDRY_GENERATED_ARTIFACT_DIR=/path/to/generated/artifacts

Import Official Results

worldfoundry-eval zoo benchmark-run \
  --benchmark-id kinetix \
  --mode normalizer \
  --official-results-path /path/to/official/results-or-report \
  --generated-artifact-dir "${WORLDFOUNDRY_GENERATED_ARTIFACT_DIR}" \
  --output-dir tmp/kinetix/normalizer \
  --json

Score Generated Artifacts

worldfoundry-eval zoo benchmark-run \
  --benchmark-id kinetix \
  --mode official-run \
  --generated-artifact-dir "${WORLDFOUNDRY_GENERATED_ARTIFACT_DIR}" \
  --output-dir tmp/kinetix/official-run \
  --json

Direct In-Tree Runner

worldfoundry-eval zoo benchmark-run --benchmark-id kinetix --mode official-validation --official-results-path <official_results.json> --output-dir <out> --json