Skip to content

Experiments

Model

Bases: Protocol

description property

A brief textual explanation that provides additional context or details about the model. It can include information such as the models purpose, configuration, architecture, or any other relevant metadata that helps describe its role within an experiment.

Returns:

Type Description
Optional[str]

Optional[str]: The description of the model if available.

hash property

A hash is a fixed length value that uniquely represents information. It acts as a locally unique identifier for a model under an experiment namespace.

Returns:

Name Type Description
str str

The hash of the model.

id property

A globally unique identifier for the models that can be used for referencing the model outside the experiment namespace.

Returns:

Name Type Description
Any UUID

The id of the model.

iterations property

Each model owns a collection of iterations that represents discrete steps in the model's training or evaluation process, typically corresponding to one step.

This accessor provides management of these iterations, allowing creation, retrieval, and listing of iterations that capture the model's progression over time.

Returns:

Name Type Description
Iterations Iterations

The collection of iterations for this model.

metrics property

Each model owns a collection of metrics.

Returns:

Name Type Description
Metrics Metrics

The collection of metrics of the model.

modules property

A model is an aggregate of modules.

Returns:

Name Type Description
Modules Modules

The modules of the model.

name property

A human redable non unique identifier for a model type.

Returns:

Type Description
Optional[str]

Optional[str]: The name of the model if any.

step property writable

An step is a unit of time that marks a transition between successive states of the model.

Returns:

Type Description
Optional[int]

Optional[int]: The steps that the model have been iterated if

Optional[int]

stored.

Models

Bases: ABC

create(hash, name=None, description=None) abstractmethod

Creates a record of a model in the database, retrieving an instance of the entity representing it.

Parameters:

Name Type Description Default
hash Any

A locally unique identifier for the model.

required
name Optional[str]

The name of the model.

None
description Optional[str]

Optional description for the model.

None

Returns:

Name Type Description
Model Model

The entity representing a model.

delete(id) abstractmethod

Deletes the experiment with the given ID.

Parameters:

Name Type Description Default
id Any

The ID of the model to delete.

required

list() abstractmethod

Retrieves a list of models.

read(hash) abstractmethod

Retrieves a model with a given hash if any.

Parameters:

Name Type Description Default
hash str

The hash of the model.

required

Returns:

Type Description
Optional[Model]

Optional[Model]: The model found if any.

update(hash, name) abstractmethod

Update the name of a model for with the given hash.

Parameters:

Name Type Description Default
hash str

The hash of the model.

required
name str

The new name of the model.

required