Skip to content
Murano

Plotly Utils

Stateless Plotly visualization utilities.

Replaces the legacy BaseVisualizationLens classes with pure functions that map data to plotly.graph_objects.Figure instances.

These functions do not accept Results objects; they operate on raw Python data or tensors so they can be used independently of the pipeline.

Requires the plot extra (install with pip install murano-interp[plot]).

plot_heatmap

Section titled “plot_heatmap”
def plot_heatmap(z_data: list[list[float]] | list[list[int]], x_labels: list[str] | None = None, y_labels: list[str] | None = None, title: str = '', color_scale: str = 'Viridis', hover_data: list[list[str]] | None = None, colorbar_title: str = '', square: bool = False, zmid: float | None = None) -> go.Figure:

Create a heatmap figure.

Args:

  • z_data: 2-D matrix of values (rows × columns).
  • x_labels: Labels for the x-axis (columns).
  • y_labels: Labels for the y-axis (rows). Defaults to
  • ["Layer 0", "Layer 1", …] when None.
  • title: Plot title.
  • color_scale: Plotly colorscale name (e.g. "Viridis", "RdBu_r").
  • hover_data: Optional 2-D list of custom hover text, same shape as
  • z_data.
  • colorbar_title: Label for the colorbar (the value legend); blank hides it.
  • square: If True, force square cells (equal x and y scale), for a matrix
  • whose axes share a unit such as an attention pattern.
  • zmid: Value anchored to the middle of the colorscale. Pass 0 with a
  • diverging scale so that the neutral color means zero; otherwise an
  • asymmetric range silently shifts the midpoint and shades zero as if
  • it had a sign.

Returns:

  • A plotly.graph_objects.Figure with a single heatmap trace.

plot_line_chart

Section titled “plot_line_chart”
def plot_line_chart(x_data: list[float] | list[int], y_series: dict[str, list[float] | list[int]], title: str = '', x_label: str = '', y_label: str = '') -> go.Figure:

Create a multi-line line chart figure.

Args:

  • x_data: Values for the x-axis.
  • y_series: Mapping from trace name to y-values. Each entry produces
  • one scatter trace.
  • title: Plot title.
  • x_label: Label for the x-axis.
  • y_label: Label for the y-axis.

Returns:

  • A plotly.graph_objects.Figure with one scatter trace per entry
  • in y_series.

save_figure

Section titled “save_figure”
def save_figure(fig: go.Figure, path: str | Path, width: int | None = None, height: int | None = None, scale: float = 2) -> Path:

Save a Plotly figure to disk, preferring a static image with an HTML fallback.

fig.write_image needs a Chrome/kaleido backend that is not always present (headless GPU compute nodes are a common case); when it is unavailable the figure is instead written as a self-contained HTML file (.html next to the requested path) so it is never lost.

Args:

  • fig: The figure to save.
  • path: Destination path. A static-image suffix (.png, .svg, …) is
  • honored when image export works; otherwise .html is used.
  • width: Image width in pixels (image export only).
  • height: Image height in pixels (image export only).
  • scale: Image resolution multiplier (image export only).

Returns:

  • The path actually written (the requested image path, or the .html
  • fallback).