© 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. 26006
    Prev
    Next
    INDEX
    Explanations

    direct address followed by clause

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

    This neuron detects person names (vocative proper names) used to directly address someone.

    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
     perioden
    -2.05
     kampen
    -1.87
     programmet
    -1.69
    
    -1.64
     kalender
    -1.64
     ™
    -1.62
    HAHA
    -1.62
     deras
    -1.58
     winkel
    -1.55
     gruppen
    -1.52
    POSITIVE LOGITS
     you
    3.06
     what
    2.70
     how
    2.58
     there
    2.53
     if
    2.52
     we
    2.27
     when
    2.16
     where
    2.09
     because
    2.08
     this
    2.03
    Activations Density 0.012%

    No Known Activations