HomeDocumentationPython SDKSlack CommunityContact Us
Documentation

Artifact Management

Suggest Edits

list

Lists all artifacts in the project.

Parameters

NameTypeDescription
filtersdictA dictionary containing the filters of the artifacts to be returned.
include_exportsboolThe boolean indication on whether to return the exported models. Default is False.

Return

A list of msgspec structs containing the artifact metadata with the following structure:

[Artifact(
    id='artifact_65a7678a91afa7d22455c5ba',
    project_id='proj_cd067221d5a6e4007ccbb4afb5966535',
    run_id='run_699baa54-f607-4c1b-9403-a015a0b4ff1c',
    is_training=False,
    step=3000,
    metric=ArtifactMetric(
        total_loss=0.059123,
        classification_loss=0.050808,
        localization_loss=0.0017339,
        regularization_loss=0.006581
    ),
    flow_title='Cell Detector',
    artifact_name='ckpt-12',
    create_date=1705469834692,
    is_deployed=False,
    model_name='efficientdet-d1-640x640',
    export_options=[
        ArtifactExportOptions(
            format='ONNX', 
            optimizations=ArtifactExportOptimizations(
                quantization=['float32', 'float16']
            ), 
            default_optimizations=ArtifactExportOptimizations(
                quantization='float32'
            )
        ), 
        ArtifactExportOptions(
            format='TensorFlow', 
            optimizations=ArtifactExportOptimizations(
                quantization=['float32']
            ), 
            default_optimizations=ArtifactExportOptimizations(
                quantization='float32'
            )
        ), 
    ], 
    exports=[
        ArtifactModel(
            id='model_5x6yy4q204538r3479q1351vw6w006r6',
            artifact_id='artifact_65a7678a91afa7d22455c5ba',
            status='Finished',
            format='Tensorflow',
            quantization="float32",
            create_date=1705472326144,
            download=DownloadSignedUrl(
                method='GET',
                expiry_date=1705558963838,
                url='https://storage.googleapis.com/699baa54-f607-4c1b-9403-a015a0b4ff1c-ckpt-12-tensorflow.zip'
            )
        ),
    ]
)]

Examples

  • List all artifacts in the project:
from datature.nexus import Client

project = Client("5aa41e8ba........").get_project("proj_b705a........")

project.artifacts.list(
    filters={
        "run_ids": ['run_699baa54-f607-4c1b-9403-a015a0b4ff1c']
    },
    include_exports=True
)
  • List artifact with workflow name "My Awesome Workflow":
from datature.nexus import Client

project = Client("5aa41e8ba........").get_project("proj_b705a........")

project.artifacts.list(include_exports=True)

artifact = [
    a for a in project.artifacts.list()
    if a.flow_title == "My Awesome Workflow"
][0]

get

Retrieves a specific artifact using the artifact ID.

Parameters

NameTypeDescription
artifact_idstrThe ID of the artifact as a string.
include_exportsboolThe boolean indication on whether to return the exported models. Default is False

Return

A msgspec struct containing the specific artifact metadata with the following structure:

Artifact(
    id='artifact_65a7678a91afa7d22455c5ba',
    project_id='proj_cd067221d5a6e4007ccbb4afb5966535',
    run_id='run_699baa54-f607-4c1b-9403-a015a0b4ff1c',
    is_training=False,
    step=3000,
    metric=ArtifactMetric(
        total_loss=0.059123,
        classification_loss=0.050808,
        localization_loss=0.0017339,
        regularization_loss=0.006581
    ),
    flow_title='Cell Detector',
    artifact_name='ckpt-12',
    create_date=1705469834692,
    is_deployed=False,
    model_name='efficientdet-d1-640x640',
    export_options=[
        ArtifactExportOptions(
            format='ONNX', 
            optimizations=ArtifactExportOptimizations(
                quantization=['float32', 'float16']
            ), 
            default_optimizations=ArtifactExportOptimizations(
                quantization='float32'
            )
        ), 
        ArtifactExportOptions(
            format='TensorFlow', 
            optimizations=ArtifactExportOptimizations(
                quantization=['float32']
            ), 
            default_optimizations=ArtifactExportOptimizations(
                quantization='float32'
            )
        ), 
    ], 
    exports=[
        ArtifactModel(
            id='model_5x6yy4q204538r3479q1351vw6w006r6',
            artifact_id='artifact_65a7678a91afa7d22455c5ba',
            status='Finished',
            format='Tensorflow',
            quantization="float32",
            create_date=1705472326144,
            download=DownloadSignedUrl(
                method='GET',
                expiry_date=1705558963838,
                url='https://storage.googleapis.com/699baa54-f607-4c1b-9403-a015a0b4ff1c-ckpt-12-tensorflow.zip'
            )
        ),
    ]
)

Examples

from datature.nexus import Client

project = Client("5aa41e8ba........").get_project("proj_b705a........")
project.artifacts.get("artifact_63bd140e67b42dc9f431ffe2", include_exports=True)

list_exported_models

Lists all exported models of a specific artifact.

Parameters

NameTypeDescription
artifact_idstrThe ID of the artifact as a string.

Return

A list of msgspec structs with the exported model metadata with the following structure:

[ArtifactModel(
    id='model_5x6yy4q204538r3479q1351vw6w006r6',
    artifact_id='artifact_65a7678a91afa7d22455c5ba',
    status='Finished',
    format='Tensorflow',
    quantization="float32",
    create_date=1705472326144,
    download=DownloadSignedUrl(
        method='GET',
        expiry_date=1705558963838,
        url='https://storage.googleapis.com/699baa54-f607-4c1b-9403-a015a0b4ff1c-ckpt-12-tensorflow.zip'
    )
)]

Examples

  • List all exported models by specific artifact ID:
from datature.nexus import Client

project = Client("5aa41e8ba........").get_project("proj_b705a........")
project.artifacts.list_exported_models("artifact_63bd140e67b42dc9f431ffe2")
  • List the Tensorflow model export of the artifact:
from datature.nexus import Client

project = Client("5aa41e8ba........").get_project("proj_b705a........")

artifact_id = "artifact_63bd140e67b42dc9f431ffe2"
tf_model = next((model
                 for model in project.artifacts.list_exported_models(artifact_id)
                 if model.format == "Tensorflow"), None)

create_export

Exports an artifact model in a specific model format. Model export may take up to 10 minutes to generate depending on the export format. This creates a download link that you can retrieve using artifacts.download_exported_model, together with the artifact model id in the return msgspec struct.

🚧

The process may take 3 to 10 minutes or more.

The export time depends on the size of the model, usually within 10 minutes. You can use list_exported_models to get the status, and once the status changes to Finished call download_exported_model to download the model.

Parameters

NameTypeDescription
artifact_idstrThe ID of the artifact as a string.
export_metadatadictThe export metadata for the export.

Return

A msgspec struct containing the operation metadata of the model export with the following structure:

ArtifactModel(
    id='model_15x95x513v9yrvvx2x329vvr34308w96',
    artifact_id='artifact_63bd140e67b42dc9f431ffe2',
    status='Queued',
    format='Pytorch',
    quantization="float32",
    create_date=1701927649302,
    download=None
)

Examples

  • Export the Tensorflow model of a specific artifact:
from datature.nexus import Client

project = Client("5aa41e8ba........").get_project("proj_b705a........")
project.artifacts.create_export(
    "artifact_63bd140e67b42dc9f431ffe2",
    {
    	"format": "Pytorch"
    }
)
  • Export the Tensorflow model of a specific artifact and download to local:
import wget
from datature.nexus import Client

project = Client("5aa41e8ba........").get_project("proj_b705a........")

project.artifacts.create_export(
    "artifact_63bd140e67b42dc9f431ffe2",
    "Tensorflow"
)

while True:
    model = next(
        (model for model in project.artifacts.list_exported_models("artifact_63bd140e67b42dc9f431ffe2")
         if model.format == 'Tensorflow' and model.status == 'Finished'), None)
  
    if model:
        wget(model.download.url)
        break

download_exported_model

Download and unzip an exported artifact model to the local path.

Parameters

NameTypeDescription
model_idstrThe ID of the exported artifact model as a string.
pathstrThe folder path to download the exported artifact model. Default is the current path.

Return

A msgspec struct containing the download path of the model:

LocalArtifact(
    download_path='local',
    model_filename='datature-yolov8l.pt',
    label_filename='label_map.pbtxt'
)

Examples

from datature.nexus import Client

project = Client("5aa41e8ba........").get_project("proj_b705a........")

project.artifacts.download_exported_model(
    "model_2q510q03x26882r4295x8y92yyqqqvq3"
    path="./local"
)

Updated 8 months ago