Lighteval documentation
Available tasks
Getting started
Guides
Examples using Inspect-AISave and read resultsCachingUse the Python APIAdd a custom taskAdd a custom metricEvaluate a custom modelUse HF's inference providers as backendUse litellm as backendUse vllm as backendUse SGLang as backendUse Hugging Face inference endpoints or TGI as backendContributing to multilingual evaluations
API
Reference
Available tasks
Browse and inspect tasks available in LightEval.
List all tasks:
lighteval tasks list
Inspect specific tasks
Inspect a task to view its config, metrics, and requirements:
lighteval tasks inspect <task_name>
Example:
lighteval tasks inspect truthfulqa:mc