UKGovernmentBEIS / inspect_ai

Inspect: A framework for large language model evaluations
https://UKGovernmentBEIS.github.io/inspect_ai/
MIT License
385 stars 41 forks source link

Re-scoring without re-running solver #27

Closed sohaibimran7 closed 1 month ago

sohaibimran7 commented 1 month ago

Is there a way to run a new scorer on an existing task from its logs?

aisi-inspect commented 1 month ago

Yes, you can use the inspect score command: https://ukgovernmentbeis.github.io/inspect_ai/scorers.html#sec-scorer-workflow