{
  "slug": "latent-observatory",
  "url": "/instruments/latent-observatory/",
  "title": "Latent Observatory: SAE Feature Explorer",
  "type": "instrument",
  "typeLabel": "Instrument",
  "lifecycle": "instrument",
  "epistemicStatus": "instrument",
  "statusLabel": "Live model data via the open Neuronpedia API; explanations labelled as hypotheses",
  "programme": "representations",
  "question": "What has a sparse autoencoder actually learned about a model, and does the auto-interp story survive contact with the activation evidence?",
  "date": "2026-07-11",
  "version": "v1",
  "authors": [
    "Latent Minds Institute"
  ],
  "methods": [
    "semantic search over auto-interp explanations",
    "activation-record inspection",
    "logit-effect readout",
    "decoder-space nearest neighbours"
  ],
  "models": [
    "GPT-2 small (RES-JB)",
    "Gemma 2 2B (Gemma Scope res 16k)"
  ],
  "evidence": "model data from the open Neuronpedia API",
  "code": null,
  "data": "https://www.neuronpedia.org/api-doc",
  "provenance": "Built by Latent Minds Institute on the open Neuronpedia API (MIT)",
  "related": [
    "interpretability-map",
    "evaluation-state"
  ],
  "canonicalUrl": "https://latentmindsinstitute.com/instruments/latent-observatory/",
  "apiUrl": "https://latentmindsinstitute.com/api/research-objects/latent-observatory.json"
}