Interactive visual tool built with Anthropic that shows J-space contents for a prompt and lets users edit them. The creator swapped 'tennis' for 'inference' and the output changed.
Neuronpedia J-space visualization
by Neuronpedia
Interactive Neuronpedia visualization of J-space contents for a prompt, with editable concepts, built with Anthropic.
Updated EU status checked Last shown
Neuronpedia J-space visualization at a glance
- What is Neuronpedia J-space visualization?
- Neuronpedia J-space visualization is an app from Neuronpedia. Interactive Neuronpedia visualization of J-space contents for a prompt, with editable concepts, built with Anthropic.
- Who makes Neuronpedia J-space visualization?
- Neuronpedia J-space visualization is made by Neuronpedia.
- Does Neuronpedia J-space visualization work in the EU?
- Available in the EU. It is a public web app on neuronpedia.org with no regional restrictions found, though no explicit EU statement was provided. Last verified 14 Nov 2026.
- How much does Neuronpedia J-space visualization cost?
- Free. No pricing information found; Neuronpedia is presented as an open-source platform and the tool appears freely usable in the browser.
- Where was Neuronpedia J-space visualization shown?
- In 1 creator video by Matthew Berman. Most recently by Matthew Berman on 8 Jul 2026, at 5:47 in “We just figured out how AI actually works (J-Space)”.
Sources: the vendor's own pages (linked above) and creator videos at the linked timestamps. EU status is information, not legal advice; see how we check.
The J-space visualization is a tool on Neuronpedia, an open-source interpretability platform. According to a YouTube creator's walkthrough, it was built with Anthropic and shows the J-space contents associated with a given prompt, presented as an interactive visual.
Users can edit those contents and see how the model's output changes. In the creator's demo, replacing the concept "tennis" with "inference" altered the output. This makes it useful for exploring how internal representations relate to model behaviour.
It is accessed through the Neuronpedia website in a browser. Details such as supported models, exact release date, usage limits and documentation were not available in the provided material, and the description here relies mainly on a single third-party video.
- Shows J-space contents for a given prompt
- Lets users edit the displayed contents
- Output changes reflect edits (e.g. swapping 'tennis' for 'inference')
- Built in collaboration with Anthropic
- Hosted on the Neuronpedia interpretability platform
As seen on YouTube
Watch the moment: the exact second top creators show it. Videos load only when you press play, via youtube-nocookie.com.
Similar AI tools
Closest neighbours by meaning, not just by category.
Anthropic J space paper
Research · Anthropic
Anthropic research paper on "J space", mentioned in connection with understanding the internal workings of AI models.
Weight Space Programming
Research · MIT
MIT research showing a frontier LLM can read and write another neural network's raw weights without gradient descent.
Show Harness
Library
Embodied harness from NUS Show Lab that lets vision-language models control robots via discrete semantic action units.
Embedding Gemma 2
Model · Google
Google's small open embedding model that maps text, code, images, video and audio into a shared embedding space.
JST-2
Model
Viggle's physics-native world foundation model that generates 3D scenes, rig-free characters and motion from text or images.
DeepSeek V4 Flash Vision Experimental
Model · DeepSeek
Experimental vision-enabled version of DeepSeek V4 Flash with image understanding; open weights under MIT, now retired on the API.