© 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. 72526
    Prev
    Next
    INDEX
    Explanations

    Dear Sir

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

    This neuron fires on vocative and address forms—greetings or exclamations used to call out to someone (e.g. “Dear,” “Oh,” “Thou,” “O,” “How”).

    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
    があるので
    -1.38
     desempen
    -1.36
    があります
    -1.34
     remé
    -1.33
    enhancing
    -1.31
     indispens
    -1.31
     GETS
    -1.31
    erer
    -1.29
    がありますが
    -1.29
    greja
    -1.28
    POSITIVE LOGITS
     your
    1.56
     as
    1.48
     surprisingly
    1.45
     old
    1.40
     unusual
    1.38
     very
    1.35
     During
    1.35
     without
    1.34
     great
    1.34
     unusually
    1.34
    Activations Density 0.009%

    No Known Activations