read_mlflow_metrics

read_mlflow_metrics(
    runs
    *
    keys=None
    x='step'
    smoothing=0.0
    max_points=DEFAULT_MAX_POINTS
    tracking_uri=None
)

Read the training curves of MLflow runs.

One chart per metric, as the MLflow UI’s model metrics draw it: the step along the x axis, the value along the y axis, and one line per run.

Parameters

Name	Type	Description	Default
runs	str, mlflow.entities.Run, pandas.DataFrame, or iterable of these	The runs, as for :func:load_mlflow_metrics: a run id, a run, what mlflow.search_runs() returns, or a list of ids and runs.	required
keys	iterable of str	Only these metrics, in this order, such as ["train_loss"].	None
x	('step', 'relative_time')	What the x axis counts.	"step"
smoothing	float	A smoothing weight, from 0 up to but not including 1, applied as TensorBoard applies it. Above 0, each run is drawn twice: as logged, and smoothed, under the run’s name followed by (smoothed). 0 draws the values as logged only.	0
max_points	int or None	At most this many points per line, evenly spaced and always keeping the first and last, so a long run stays quick to walk. None keeps every point.	1000
tracking_uri	str	The tracking server to read from. By default MLflow’s own.	None

Returns

Name	Type	Description
	list of maidr.mlflow.MetricChart	One per metric. Pass one to :func:maidr.show, :func:maidr.render or :func:maidr.save_html, or store it in a run with :func:log_mlflow_chart.

Raises

Name	Type	Description
	ImportError	If mlflow is not installed.
	ValueError	If x is not one of the above, smoothing is not in [0, 1) or max_points is below 2.

Notes

Every value of a metric is read, with MlflowClient.get_metric_history, and placed in order of its step and then the time it was logged, so two values logged at one step are both kept, in the order they were logged. max_points then thins the line. A logged NaN is a gap in it.

Examples

>>> import mlflow
>>> import maidr
>>> runs = mlflow.search_runs(experiment_names=["llm-finetune"])
>>> charts = maidr.read_mlflow_metrics(runs, keys=["train_loss"])
>>> maidr.save_html(charts[0], "loss.html")