Logging Artifacts#

To log artifacts that are not of kind model or dataset, use the log_artifact() method. You can apply this method to the project object or the context object when logging an artifact in runtime, for example a job. In this section

See also

Basic logging of an artifacts file#

log_artifact can be used to log many kinds of files, for example html,pkl and more. This is the most general method of logging artifacts.

Tip

Each object type requires a different way to convert it for logging the object to a file. This is just an example of one type.

To log an artifacts file, specify the local file path to the file using the local_path, or use the body to dump the object content.

Log with local path

    with open("file.txt","w") as f:
        f.write("abc is 123")
    
    project.log_artifact(
        "file-example",
        local_path="file.txt",
        labels={"Test": "label-test"},
    )

Log with body

    project.log_artifact(
        "some-data",
        body=b"abc is 123",
        format="txt",
        labels={"Test": "label-test"},
    )

Log a Plotly object as an html file#

This example illustrates logging a Plotly figure using log_artifact as an html file:

import plotly.graph_objects as go
import numpy as np
# Create a Sin(x) Graph
x = np.linspace(0, 10, 100)
y = np.sin(x)
# Create a Plotly figure
fig = go.Figure()
# Add a line trace to the figure
fig.add_trace(go.Scatter(x=x, y=y, mode='lines', name='Sin(x)'))
# Update layout
fig.update_layout(
    title="Sin(x) Plot",
    xaxis_title="x",
    yaxis_title="sin(x)",
    template="plotly_dark" 
)
project.log_artifact("plotly-artifact",
                     body=fig.to_html(),# convert object for logging an html file
                     format="html")

Logging Plotly artifacts#

This example illustrates using MLRun to convert and handle the object:

# Use mlrun to convert the python object
plotly_artifact = mlrun.artifacts.PlotlyArtifact(figure=fig, key="sin_x") 
# Log the artifact
context.log_artifact(plotly_artifact)

Logging directory artifacts#

When using log_artifact to log a directory, by default:

  • The artifact is logged as an mlrun.artifacts.DirArtifact object.

  • The files are not uploaded. If you want to upload the files, set upload=True.

    project.log_artifact(
    "artifact-directory-testing",
    local_path="./artifact_directory/",
    labels={"Dir":"dir-example"})