core.models.model

model

Classes:

Name	Description
Model	Represents a model version and manages its associated files and runtime instance.

Model(name, model_version, pretrained_weights_name=None, trained_weights_name=None, config_name=None, exported_weights_name=None, labelmap=None)

Represents a model version and manages its associated files and runtime instance.

Methods:

Name	Description
set_loaded_model	Set the runtime-loaded model instance.
download_weights	Download all configured model files (weights, config, exports) to destination.
save_artifact_to_experiment	Store an artifact file in the given experiment.

Attributes:

Name	Type	Description
name		The name of the model.
model_version		The version of the model from Picsellia.
pretrained_weights_name		The name of the pretrained weights file attached to the model version in Picsellia.
trained_weights_name		The name of the trained weights file attached to the model version in Picsellia.
config_name		The name of the configuration file attached to the model version in Picsellia.
exported_weights_name		The name of the exported weights file attached to the model version in Picsellia.
labelmap		A dictionary mapping category names to labels.
weights_dir	str | None	The directory where model weights are stored.
results_dir	str | None	The directory where model results are stored.
pretrained_weights_dir	str | None	The directory where pretrained weights are stored.
trained_weights_dir	str | None	The directory where trained weights are stored.
config_dir	str | None	The directory where model configuration files are stored.
exported_weights_dir	str | None	The directory where exported weights are stored.
pretrained_weights_path	str | None	The path to the pretrained weights file.
trained_weights_path	str | None	The path to the trained weights file.
config_path	str | None	The path to the model configuration file.
exported_weights_path	str | None	The path to the exported weights file.
loaded_model	Any	Return the loaded model instance.

name = name instance-attribute

The name of the model.

model_version = model_version instance-attribute

The version of the model from Picsellia.

pretrained_weights_name = pretrained_weights_name instance-attribute

The name of the pretrained weights file attached to the model version in Picsellia.

trained_weights_name = trained_weights_name instance-attribute

The name of the trained weights file attached to the model version in Picsellia.

config_name = config_name instance-attribute

The name of the configuration file attached to the model version in Picsellia.

exported_weights_name = exported_weights_name instance-attribute

The name of the exported weights file attached to the model version in Picsellia.

labelmap = labelmap or {} instance-attribute

A dictionary mapping category names to labels.

weights_dir = None instance-attribute

The directory where model weights are stored.

results_dir = None instance-attribute

The directory where model results are stored.

pretrained_weights_dir = None instance-attribute

The directory where pretrained weights are stored.

trained_weights_dir = None instance-attribute

The directory where trained weights are stored.

config_dir = None instance-attribute

The directory where model configuration files are stored.

exported_weights_dir = None instance-attribute

The directory where exported weights are stored.

pretrained_weights_path = None instance-attribute

The path to the pretrained weights file.

trained_weights_path = None instance-attribute

The path to the trained weights file.

config_path = None instance-attribute

The path to the model configuration file.

exported_weights_path = None instance-attribute

The path to the exported weights file.

loaded_model property

Return the loaded model instance.

set_loaded_model(model)

Set the runtime-loaded model instance.

download_weights(destination_dir)

Download all configured model files (weights, config, exports) to destination.

Parameters:

Name	Type	Description	Default
destination_dir	str	Root directory for downloaded files.	required

save_artifact_to_experiment(experiment, artifact_name, artifact_path)

Store an artifact file in the given experiment.

Parameters:

Name	Type	Description	Default
experiment	Experiment	Experiment to store into.	required
artifact_name	str	Name under which to save the artifact.	required
artifact_path	str	Path to the file.	required