© 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. 10-GEMMASCOPE-RES-131K
    7. 44485
    Prev
    Next
    INDEX
    Explanations

    people count or role

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

    The neuron fires on “man” (and related count‐of‐people words like “person” or “band”) when it appears in compounds indicating group size (e.g. “one‐man,” “two‐man,” “seven‐person team”).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_10/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
     dårlig
    -1.95
     bekym
    -1.84
     dår
    -1.84
     wszyst
    -1.83
     vanskelig
    -1.78
     måte
    -1.74
     carpeta
    -1.72
    titial
    -1.68
     mør
    -1.64
     börja
    -1.60
    POSITIVE LOGITS
    n
    1.90
    h
    1.70
    na
    1.70
    "
    1.70
    j
    1.69
     weighty
    1.64
    no
    1.62
    m
    1.59
     but
    1.57
     archaic
    1.53
    Activations Density 0.006%

    No Known Activations