© 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. 34331
    Prev
    Next
    INDEX
    Explanations

    What else can

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

    The neuron detects interrogative phrasing—question words (Who, What, How) often coupled with modal verbs (would, could, can) and negations—i.e. rhetorical or open-ended questions.

    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
    קב
    -1.05
    ċi
    -1.04
    Ǘ
    -1.02
     cintur
    -0.98
     RoutedEventArgs
    -0.97
    şan
    -0.96
    新款
    -0.96
     bluz
    -0.95
     curioso
    -0.94
    …).
    -0.94
    POSITIVE LOGITS
    ?
    2.84
     además
    1.28
    ?!
    1.20
    !
    1.15
    ?"
    1.13
    WithName
    1.12
     спустя
    1.09
    そもそも
    1.07
     siquiera
    1.05
     navíc
    1.05
    Activations Density 0.030%

    No Known Activations