2.8.0: Live.log_plot

New Features

Live.log_plot

Create DVC plots from datapoints (list of dictionaries) and plot config.

datapoints = [{"foo": 1, "bar": 2}, {"foo": 3, "bar": 4}]
with Live() as live:
    live.log_plot("foo_default", datapoints, x="foo", y="bar", title="Default Linear Plot")
    live.log_plot(
        "foo_scatter",
        datapoints,
        x="foo",
        y="bar",
        template="scatter",
        title="Plot using Scatter template",
        y_label="CUSTOM LABEL: BAR",
        x_label="CUSTOM LABEL: FOO"
    )

Example output:

• dvclive/plots/custom/foo_default.json

[
    {
        "foo": 1,
        "bar": 2
    },
    {
        "foo": 3,
        "bar": 4
    }
]                                                               

• dvclive/dvc.yaml

plots:
- plots/custom/foo_default.json:
    x: foo
    y: bar
    title: Default Linear Plot
- plots/custom/foo_scatter.json:
    template: scatter
    x: foo
    y: bar
    title: Plot using Scatter template
    x_label: 'CUSTOM LABEL: FOO'
    y_label: 'CUSTOM LABEL: BAR'

• dvc plots show

Custom plot properties for Live.log_sklearn_plot

from dvclive import Live

y_true = [0, 0, 1, 1]
y_score = [0.1, 0.4, 0.35, 0.8]

with Live() as live:
    live.log_sklearn_plot(
        "roc", y_true, y_score, 
        # These properties are new
        title="Custom Title for ROC Curve",
        x_label="False Positive Rate",
        y_label="True Positive Rate")

What's Changed

• only log from dvclive by @dberenbaum in #541
• [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci in #544
• setup: Use ranges for test dependencies. by @daavoo in #545
• live: Add log_plot. by @daavoo in #543

Full Changelog: https://github.com/iterative/dvclive/compare/2.7.0...2.8.0###