Why it matters

HELM broad LLM eval. Understanding shapes benchmark choice.

Advertisement

The architecture

Multiple scenarios + metrics.

Comparable across models.

Regular releases.

HELM stackScenariosdiverse tasksMetricsaccuracy + safety + moreComparisoncross-modelHELM Safety + HELM Instruct + HELM Capabilities suites cover different aspects
HELM.
Advertisement

How it works end to end

Open source + hosted.

HELM Safety focuses safety metrics.

Multi-dimensional radar plots.