Experiment for a Dataset

Experiment for a Dataset

Opens the Experiments page, where you can view, filter, and launch experiments against your datasets.

The page displays a list of completed and in-progress experiment runs, along with filters for dataset, status, and target type.

This section contains the list of experiment runs and provides access to detailed results for each run.

The table shows all completed experiments, including dataset name, target, status, item count, and average scores.

Opens the detailed results page for the selected experiment run, showing scores, items, and configuration.

The page displays the experiment name, status, target, score average, and tabs for scores, items, and configuration.

The page displays key metrics including items processed, score average, and start and finish times for the experiment run.

This card summarizes the number of items processed and the average score for the experiment run.

The Scores tab displays evaluator summaries and recent score events tied to the experiment run.

This card summarizes the number of items processed and the average score for the experiment run.

The Scores tab shows evaluator summaries and recent score events for the experiment.

The Scores tab displays evaluator summaries and recent score events tied to the experiment run.

This tab displays the per-item results for each dataset entry processed in the experiment run.

The table shows the input, output, and status for each item evaluated during the experiment.

This tab displays the run configuration, including model, prompt, and evaluator settings used for the experiment.

The configuration details include model identifiers, prompt versions, endpoint IDs, and experiment group metadata.