Show HN: Visualize Model Spikiness in 3D

Hacker News Top Tools
visualization 3d ai-models spikiness model-evaluation interactive

Summary

ModelMap is a web-based 3D visualization tool that displays AI model test scores as spiky shapes, allowing users to explore and compare model performance through interactive 3D graphics.

Models are referred to as &#x27;spiky&#x27; entities - they have relative strengths and weaknesses.<p>Model map visualizes these strengths and weaknesses in 3D and give you the ability to fly around this 3D space (there is a hidden Star Wars themed mini game for pilots brave enough to try).<p>Spikiness is an intuitive mental model but the typical way of visualizing this spikiness is a generic table.<p>Is this useful? Maybe?<p>Is it fun? Yes.<p>You should try the flight controls. Click anywhere then use WASD + mouse to fly. Shoot projectiles to explore nodes.<p>Built on top of 3D Graph OSS library I made for visualizing KBs in 3D: <a href="https:&#x2F;&#x2F;github.com&#x2F;a-funk&#x2F;3dGraph" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;a-funk&#x2F;3dGraph</a>
Original Article
View Cached Full Text

Cached at: 07/06/26, 08:01 AM

# ModelMap Source: [https://www.modelmap.tech/](https://www.modelmap.tech/) One AI model's test scores as a 3D shape — longer spike = higher score\. mouselook ·Space/Cup/down ·Q/Eroll clickclick a spike to zoom to it ESCexit flight Orbit dragorbit ·scrollzoom hovertooltip ·clickdetails ⇧Mhide the interface

Similar Articles

Show HN: Spin Lab

Hacker News Top

Spin Lab is an interactive web tool that visualizes the physics of spin in table-tennis, allowing users to explore how spin affects ball trajectory.

Show HN: Are You in the Weights?

Hacker News Top

A site that checks how strongly frontier and small AI models recognize you by querying them in parallel and clustering responses.

OpenAI Microscope

OpenAI Blog

OpenAI Microscope is an open-source tool that systematically visualizes every neuron in commonly studied vision models with fast feedback loops and linkable neurons to support interpretability research. The platform reduces visualization time from minutes to seconds and aims to make neural network analysis more accessible to the research community.