WandbChart API

Bases: AnyWidget

Live line chart that polls wandb for metric data via the GraphQL API.

Renders a Canvas-based chart that auto-updates while runs are active. Supports multiple runs for side-by-side comparison and optional rolling smoothing (rolling mean, exponential moving average, or gaussian) with raw data shown behind the smoothed line.

Examples:

import wandb
from wigglystuff import WandbChart, EnvConfig

config = EnvConfig(["WANDB_API_KEY"])
config.require_valid()

run = wandb.init(project="my-project")
chart = WandbChart(
    api_key=config["WANDB_API_KEY"],
    entity=run.entity,
    project=run.project,
    runs=[run],
    key="loss",
)
chart

Create a WandbChart widget.

Parameters:

Name	Type	Description	Default
api_key		Your wandb API key.	required
entity		The wandb entity (user or team).	required
project		The wandb project name.	required
runs	Optional[Sequence[Any]]	A list of wandb Run objects or dicts with id and label keys.	None
key		The metric key to chart (e.g. "loss").	required
poll_seconds		Seconds between polling updates, or None to disable auto-polling and show a manual refresh button instead.	required
smoothing_kind		Type of smoothing: "rolling", "exponential", or "gaussian". Defaults to "rolling".	required
smoothing_param		Smoothing parameter, or None for no smoothing. For "rolling": integer window size >= 2. For "exponential": EMA weight on previous value, 0 < alpha < 1. For "gaussian": standard deviation (sigma) > 0.	required
show_slider		Whether to show the smoothing slider in the UI.	required
width		Chart width in pixels.	required
height		Chart height in pixels.	required
**kwargs	Any	Forwarded to anywidget.AnyWidget.	{}

Source code in wigglystuff/wandb_chart.py

def __init__(
    self,
    *args: Any,
    runs: Optional[Sequence[Any]] = None,
    **kwargs: Any,
) -> None:
    """Create a WandbChart widget.

    Args:
        api_key: Your wandb API key.
        entity: The wandb entity (user or team).
        project: The wandb project name.
        runs: A list of wandb Run objects or dicts with ``id`` and ``label`` keys.
        key: The metric key to chart (e.g. ``"loss"``).
        poll_seconds: Seconds between polling updates, or ``None`` to
            disable auto-polling and show a manual refresh button instead.
        smoothing_kind: Type of smoothing: ``"rolling"``, ``"exponential"``,
            or ``"gaussian"``. Defaults to ``"rolling"``.
        smoothing_param: Smoothing parameter, or ``None`` for no smoothing.
            For ``"rolling"``: integer window size ``>= 2``.
            For ``"exponential"``: EMA weight on previous value, ``0 < alpha < 1``.
            For ``"gaussian"``: standard deviation (sigma) ``> 0``.
        show_slider: Whether to show the smoothing slider in the UI.
        width: Chart width in pixels.
        height: Chart height in pixels.
        **kwargs: Forwarded to ``anywidget.AnyWidget``.
    """
    if runs is not None:
        kwargs["_runs"] = [
            {"id": r.id, "label": r.name} if hasattr(r, "id") else r
            for r in runs
        ]
    super().__init__(*args, **kwargs)

Synced traitlets

Traitlet	Type	Notes
api_key	str	Your wandb API key.
entity	str	The wandb entity (user or team).
project	str	The wandb project name.
key	str	The metric key to chart (e.g. "loss").
poll_seconds	int \| None	Seconds between polling updates, or None for manual refresh (default: 5).
smoothing_kind	str	Type of smoothing: "rolling", "exponential", or "gaussian" (default: "gaussian").
smoothing_param	float \| None	Smoothing parameter, or None for no smoothing.
show_slider	bool	Whether to show the smoothing slider (default: True).
width	int	Chart width in pixels (default: 700).
height	int	Chart height in pixels (default: 300).