Skip to content

v0.0.1

Choose a tag to compare

@dmjoy dmjoy released this 28 Oct 18:49
· 167 commits to main since this release

Version 0.0.1 -- Released 2025-10-28

  • Initial release; includes minimum working implementations for:
    • Evaluation card specification and evaluation
    • HELM benchmark output downloading and data interfaces
    • Benchmark Predictor class (with random, and perturbation based examples)
    • Utility for "offline" HELM perturbation application
    • Ad-hoc inference and direct model access through HELM
    • Command-line wrapper for helm-run supporting runs against "offline" dataset instances