Skip to main content
The openlayer export command downloads monitoring data from an inference pipeline for a specified time period. It supports flexible ways to specify date ranges and optional filters to refine the exported dataset.

Usage

This will fetch your monitoring data from the inference pipeline and place a dataset.json and config.json file in ./openlayer-exports/{pipeline_name}_{start}_{end}.

Date ranges

You can specify time ranges in multiple ways:

Filters

You can refine the exported dataset using --filter flags or a JSON file.

Simple syntax

JSON file (advanced)

For more complex filters, pass a JSON file:

Output format

--format (or -f) controls the file format of the exported data:

Test scores

By default the export contains no score columns. Pass --include-scores all to add each test’s score and explanation as columns:
This export is slower than the default one, because the scores are fetched per chunk of data. Score and explanation columns are named by a config hash, such as openlayer_llm_rubric_score_78812226. The hash keeps two tests that share a name in separate columns, and changing an evaluator starts a new column. With --include-scores all, the CLI adds metricColumns to config.json after it extracts the archive. Each entry maps a hashed column to its test name and test id:
scoreColumn is the column name in the exported rows. An entry can include other fields from the same registry. If the lookup fails, the CLI prints a warning and still keeps the export. The rows are usable without the mapping.
Combining --format csv with --include-scores all splits the output into several data_partNNNN.csv files instead of a single file. The CLI prints a note when it does.

Flags