visualization-and-plotting

Generate Plotly visualizations from ToyModel and ModelGrid objects.

13|1|Updated Feb 3, 2026
One-click install
npx skills add https://github.com/OliverSieweke/occhio --skill visualization-and-plotting
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: visualization-and-plotting
Source: https://github.com/OliverSieweke/occhio/tree/main/skills/visualization-and-plotting
Command: npx skills add https://github.com/OliverSieweke/occhio --skill visualization-and-plotting

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Researchers need quick, reusable Plotly-based visualizations for ToyModel and ModelGrid objects to inspect behavior and communicate results without writing boilerplate plots.

Core Features & Use Cases

  • Subclass SinglePlot to implement render() and configure_layout(), enabling per-model or per-grid facet plots.
  • Use CompositePlot to compose multiple plots in a grid and share the same n_render_axes.
  • Read data from ToyModel properties and leverage FigureProxy to handle axis deduplication, subplots, and interactivity safely.

Quick Start

Subclass SinglePlot, implement render() to draw traces from ToyModel data, and export an instance for discovery.

Frequently Asked Questions about visualization-and-plotting

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I create interactive Plotly visualizations for ToyModel objects?

To create interactive Plotly visualizations for ToyModel objects, subclass SinglePlot to implement the render() and configure_layout() methods, reading data directly from ToyModel properties to generate traces. The FigureProxy safely handles axis deduplication and interactivity.

Can I compose multiple plots into a single grid view for ModelGrid metrics?

Yes, you can compose multiple plots into a single grid view for ModelGrid metrics by using the CompositePlot class. It allows you to arrange multiple individual plots in a grid and share the same n_render_axes for synchronized rendering.

What is the best way to inspect model embeddings without writing boilerplate plotting code?

The best way to inspect model embeddings without boilerplate is using pre-configured SinglePlot subclasses. They read directly from ToyModel properties and leverage the FigureProxy rendering lifecycle to safely manage subplots and interactivity for feature analysis.

Does this visualization approach support model debugging and presentation workflows?

Yes, this visualization approach supports model debugging and presentation workflows by providing quick, reusable interactive plots of embeddings, representations, and grid metrics. It safely handles subplots and interactivity through the FigureProxy rendering lifecycle.

How do I configure layout and render traces for per-model facet plots?

To configure layout and render traces for per-model facet plots, subclass SinglePlot and implement the configure_layout() method for layout customization and the render() method to draw traces from ToyModel data, then export an instance for discovery.