Experiments

Metric

Bases: Protocol

id property

The globally unique identifier of the metric.

Returns:

Name	Type	Description
UUID	UUID	The ID of the metric.

name property

A metric is categorized by it's name.

Returns:

Name	Type	Description
str	str	The name of the metric.

phase property

The operational stage of the model when the metric was produced. The phase helps interpret the metric in context of the model's lifecycle.

Returns:

Type	Description
Optional[str]	Optional[str]: The phase of the model in wich the metric was produced.

step property

An step is a discrete unit of time that marks a transition between successive states of a machine learning model. Each metric value is associated with some stage of the model indexed by the step.

Returns:

Name	Type	Description
int	Optional[int]	The step of the metric.

value property

A quantitative value representing the model's performance.

Returns:

Name	Type	Description
Any	Any	The value of the metric.

Metrics

Bases: ABC

clear() abstractmethod

Removes all metrics from the collection.

list(name=None) abstractmethod

Get a list of metrics in the collection.

Parameters:

Name	Type	Description	Default
name	Optional[str]	The name of the metrics to be listed.	None

Returns:

Type	Description
list[Metric]	list[Metric]: The list of metric of the model.

log(name, value, step=None, phase=None) abstractmethod

Add a metric to the collection.

Parameters:

Name	Type	Description	Default
name	str	The name of the metric.	required
value	Any	The value of the metric.	required
step	int	The step in wich the metric was produced.	None
phase	Optional[str]	The phase in wich the metric was produced.	None

remove(metric) abstractmethod

Removes the given metric from the collection.