© 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-3-12B
    5. 31-GEMMASCOPE-2-RES-16K
    7. 8832
    Prev
    Next
    INDEX
    Explanations

    win, wind, windows

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

    This neuron fires on tokens containing the character sequence “win” (or its close variant “vin”), particularly when it appears at the start of the token.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-12b-pt/resid_post/layer_31_width_16k_l0_medium
    Prompts (Dashboard)
    392,802 prompts, 256 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    킨
    0.55
     initials
    0.55
    Personendaten
    0.54
    ジナル
    0.53
    otropin
    0.52
    чь
    0.51
     signatory
    0.51
    বৈশাখ
    0.51
     imageData
    0.50
    лены
    0.50
    POSITIVE LOGITS
    nipeg
    0.75
    ifred
    0.74
     Вин
    0.72
     Win
    0.71
    🍷
    0.71
    ograd
    0.70
     вина
    0.70
     VIN
    0.69
     vins
    0.68
     Vind
    0.68
    Activations Density 0.238%

    No Known Activations