MetricsRunResult

MetricsRunResult(
    output_csv,
    report_file,
    columns_to_compare,
    metrics_by_column,
    reports,
    total_inference_time,
    avg_inference_time,
    input_chars,
    output_chars,
    energy_consumed,
    emissions,
    cpu_model,
    gpu_model,
    summary_text,
    run_id=None,
    runs_csv=None,
    n_queries=None,
)

Result returned by run_metrics.

Attributes

Name	Type	Description
output_csv	Path	Filesystem path to the aggregate metrics CSV that was updated.
report_file	Path	Filesystem path to the per-column classification report text file.
columns_to_compare	list[str]	Annotation columns included in the evaluation.
metrics_by_column	dict[str, dict[str, Any]]	Nested dictionary of computed metrics keyed by column name.
reports	dict[str, str]	Human-readable report text keyed by annotation column name.
total_inference_time	float | None	Total model inference time in seconds, if available.
avg_inference_time	float | None	Mean inference time per annotation request in seconds, if available.
input_chars	int | None	Total prompt characters sent to the model, if available.
output_chars	int | None	Total response characters returned by the model, if available.
energy_consumed	float | None	Energy consumption in kilowatt-hours, if available.
emissions	float | None	Emissions estimate in kilograms of CO2 equivalent, if available.
cpu_model	str | None	CPU metadata recorded by CodeCarbon, if available.
gpu_model	str | None	GPU metadata recorded by CodeCarbon, if available.
summary_text	str	Plain-text summary of the main evaluation metrics.
run_id	str | None	Identifier linking this run’s rows across the runs and metrics tables.
runs_csv	Path | None	Filesystem path to the per-run efficiency/config table, if written.
n_queries	int | None	Number of model inference calls in the annotation run, if available.