© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 66205
    Prev
    Next
    INDEX
    Explanations

    numbers and continuing actions

    np_acts-logits-general · gemini-2.5-flash-lite

    This neuron strongly activates on tokens that appear at the beginnings of lines or blocks (e.g. the first word of a paragraph, headline, date line, or section).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    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
    cended
    -1.02
     Lovato
    -0.93
    けません
    -0.92
     however
    -0.86
     véhicules
    -0.85
     Morelos
    -0.85
     wären
    -0.85
    (%)
    -0.84
     jsonObject
    -0.84
     gás
    -0.83
    POSITIVE LOGITS
     garg
    1.00
     will
    0.96
     continue
    0.95
     continued
    0.94
     замети
    0.94
     burgeoning
    0.92
     badass
    0.92
    ährlich
    0.90
    titleMargin
    0.90
     continuing
    0.89
    Activations Density 0.004%

    No Known Activations