© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 129326
    Prev
    Next
    INDEX
    Explanations

    The neuron is tuned to pick out isolated label‐style tokens—single words or short phrases used as field names or column headers (e.g. “Name,” “Sector,” “Accession,” “Model,” “Price,” etc.).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    category labels

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_22/width_131k
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    dicionado
    -1.05
    Spes
    -0.96
    Ilustra
    -0.94
    ﻣ
    -0.92
    Materiale
    -0.90
    ilever
    -0.89
     Außerdem
    -0.88
    icône
    -0.88
    谙
    -0.88
    johtaja
    -0.85
    POSITIVE LOGITS
     stycznia
    1.02
    perusahaan
    1.00
     июля
    0.98
    domo
    0.97
     сентября
    0.93
     июня
    0.93
     in
    0.92
     their
    0.91
    らかに
    0.89
     железнодоро
    0.89
    Activations Density 0.036%

    No Known Activations