Skip to main content
This page describes the data model of datasets and dataset runs. For detailed reference, see the Datasets documentation.

Datasets

Datasets are a collection of inputs and, optionally, expected outputs that can be used during Dataset runs. Datasets are a collection of DatasetItems. Dataset run with traces

Dataset object

DatasetItem object

DatasetRun

Dataset runs are used to run a dataset through your LLM application and optionally apply evaluation methods to the results. Dataset run with annotations

DatasetRun object

DatasetRunItem object

Most of the time, we recommend that DatasetRunItems reference TraceIDs directly. The reference to ObservationID exists for backwards compatibility with older SDK versions.

End to end data relations

DataSetRuns can combine a few ABV objects:
  • DatasetRuns are created by looping through all or selected DatasetItems of a Dataset with your LLM application.
  • For each DatasetItem passed into the LLM application as an Input a DatasetRunItem & a Trace are created.
  • Optionally Scores can be added to the Traces to evaluate the output of the LLM application during the DatasetRun.